Keywords: Vision-based control, Autonomous systems, Control applications
Abstract: In this paper the problem of performing a precision landing of an autonomous multirotor on a small barge at sea is studied. An image-based visual servoing approach, which was initially developed for landing on stationary targets, is extended to suit the shipboard landing case. The approach includes visual servoing for aligning the multirotor with a target on the vessel, and a velocity feed-forward term which is estimated online by fusing vision and GPS velocity measurements. Special considerations are made to account for the presence of wind, and the approach is validated through full-scale outdoor hardware flight tests. The hardware system is composed entirely from off-the-shelf components that are commonly used in industry.

Keywords: Vision-based control, Estimation, Autonomous systems
Abstract: Recent advances in visual-inertial state estimation have allowed quadrotor aircraft to autonomously navigate in unknown environments at operational speeds. In most cases, substantially higher speeds can be achieved by actively designing motion that reduces state estimation error. We are interested in autonomous vehicles running feature-based visual- inertial state estimation algorithms. In particular, we consider a trajectory optimization problem in which the goal is to maximize co-visibility of features, i.e. features are kept visible in the camera view from one keyframe to the next, increasing state estimation accuracy. Our algorithm is developed for autonomous quadrotor aircraft, for which position and yaw tra- jectories can be tracked separately. We assume that the desired positions of the vehicle are determined a priori, for instance, by a path planner that uses obstacles in the environment to generate a trajectory of positions with free yaw. This paper presents a novel algorithm that determines the yaw trajectory that jointly optimizes aggressiveness and feature co-visibility. The benefit of this algorithm was experimentally verified using a custom built quadrotor which uses visual inertial odometry for state estimation. The generated trajectories lead to better state estimation which contributes to improved trajectory tracking by a state-of-the-art controller under autonomous high-speed flight. Our results show that the root-mean-square error of the trajectory tracking is improved by almost 70%.

Keywords: Vision-based control, Robotics, Visual servo control
Abstract: An image-based control strategy along with estimation of target motion is developed to track dynamic targets without motion constraints. To the best of our knowledge, this is the first work that utilizes a bounding box as image features for tracking control and estimation of dynamic target without motion constraint. The features generated from a YOLO deep neural network can relax the assumption of continuous availability of the feature points in most literature and minimize the gap for applications. The challenges are that the motion pattern of the target is unknown and modeling its dynamics is infeasible. To resolve these issues, the dynamics of the target is modeled by a constant-velocity model and is employed as a process model in the UKF, but process noise is uncertain and sensitive to system instability. To ensure convergence of the estimate error, the noise covariance matrix is estimated according to history data within a moving window. The estimated motion from the UKF is implemented as a feedforward term in the developed controller, so that tracking performance is enhanced. Simulations are demonstrated to verify the efficacy of the developed estimator and controller.

Keywords: Visual servo control, Robotics, Vision-based control
Abstract: Straight lines are common features in human made environments, which makes them a frequently explored feature for control applications. Many control schemes, like Visual Servoing, require the 3D parameters of the features to be estimated. In order to obtain the 3D structure of lines, a nonlinear observer is proposed. However, to guarantee convergence, the dynamical system must be coupled with an algebraic equation. This is achieved by using spherical coordinates to represent the line's moment vector, and a change of basis, which allows to introduce the algebraic constraint directly on the system's dynamics. Finally, a control law that attempts to optimize the convergence behavior of the observer is presented. The approach is validated in simulation, and with a real robotic platform with a camera onboard.

Keywords: Visual servo control, Vision-based control, Mechanical systems/robotics
Abstract: We consider the problem of tracking a desired trajectory using bearing-only information from stationary landmarks. We propose a control law that combines a gradient descent feedback with a velocity feedforward. These two terms are calculated solely by using the bearings, their time derivatives, and the knowledge of the dynamical model. Under some technical assumptions our controller shows global convergence. We analyze the limitations associated with this control law, and provide methods to bypass most of these limitations. Finally, we verify our controller through simulation.

Keywords: Optimal control, Linear parameter-varying systems, Robust control
Abstract: Braking and steering are the typical crash avoidance maneuvers to overcome a hazardous situation. Generally, braking is a preferred measure at low speed and steering becomes an efficient alternative as speed increases. However, determining the optimum evasive maneuver in all hazardous situations is challenging since a complete evaluation of all driving complications is difficult to achieve. This paper proposes a methodology to evaluate steering evasive maneuver using mixed H_2/H_∞control in the presence of parameter uncertainties. Vehicle side slip angle and evasive trajectory tracking errors are considered for stability and tracking performance. D-stability is employed in order to consider the transient response of the controller. Then, the criticality index is developed by solving sets of linear matrix inequalities (LMIs) and bilinear matrix inequalities (BMIs) to evaluate steering evasive maneuver.

Keywords: Automotive control, Optimal control, Control applications
Abstract: This paper presents a computationally-efficient fuel-economic control strategy for connected vehicles in urban roads in the presence of uncertainty in shared information. The proposed high-level controller also focuses on reducing red-light idling, which improves traffic mobility and vehicle emissions. We model the red-light idling avoidance problem with a terminal constraint and evaluate the equivalent deterministic constraints that ensure probabilistic collision avoidance. We then employ a sampling-based approach to evaluate a feasible solution in real-time. This leads to control solutions -if a feasible solution is obtained- that can ensure avoidance of red-light idling despite the number of vehicles in front of it. We have shown that sampling from a Gaussian distribution whose mean depends on the target velocity can improve fuel economy to a good extent. This high-level controller provides a good initial solution for any deterministic low-level controller. Simulation results show the efficacy of the proposed method in terms of fuel economy and computational efficiency.

Keywords: Predictive control for nonlinear systems, Optimal control, Automotive control
Abstract: Previous work by the authors has introduced a collision imminent steering algorithm to autonomously steer a vehicle if an imminent collision cannot be avoided by braking alone. The original formulation was developed for straight roads. In this work, a new formulation is presented to account for curved roads, as well. To accommodate curved roads, a drivable tube is introduced, within which any trajectory is considered collision free. The optimal control problem is formulated to minimize the peak tire slip of the maneuver to seek a minimally aggressive solution. To ensure the safety of the vehicle in a potential maneuver, hard safety constraints on the vehicle stability and safe driving regions are enforced. Numerical simulations are presented to demonstrate the effectiveness of the proposed formulation.

Keywords: Autonomous systems, Automotive control, Automotive systems
Abstract: From the perspective of both safety and economy, autonomous trucks are considered as one of the most promising technology for future commercial vehicles. Although the local trajectory planning is one of the important techniques for autonomous trucks, most of relevant researches only considered vehicle planar dynamics/motion constraints in trajectory planning. Few research investigated the impact of rollover with respect to the planned trajectory, which is more significant for commercial trucks than passenger vehicles. In this study, a local trajectory planning for collision avoidance maneuvers of autonomous trucks is proposed. The whole collision avoidance maneuver is considered as a three-step procedure, including braking, lane changing, and acceleration. Specifically for autonomous trucks, rollover prevention is considered as an extra constraint to ensure vehicle stability and safety. The effectiveness and feasibility of the proposed trajectory planning strategy is evaluated by simulation results. The conclusion is that the integration of rollover prevention in the local trajectory planning is important to ensure the handling stability and safety as well as the efficiency of autonomous trucks.

Keywords: Automotive control
Abstract: We consider the maneuver planning problem for automated vehicles when they share the road with human-driven cars and interact with each other using a finite set of maneuvers. Each maneuver is calculated considering input constraints, actuator disturbances and sensor noise, so that we can use a maneuver automaton to perform high-level planning that is robust against low-level effects. In order to model the behavior of human-driven cars in response to the intent of the automated vehicle, we use control improvisation to build a probabilistic model. To accommodate for potential mismatches between the learned human model and human driving behaviors, we use a conditional value-at-risk objective function to obtain the optimal policy for the automated vehicle. We demonstrate through simulations that our motion planning framework allows an automated vehicle to exploit human behaviors with different levels of robustness.

Keywords: Automotive systems, Traffic control, Pattern recognition and classification
Abstract: Automated driving technology involves various modern in-vehicle systems that are designed to increase road traffic safety by helping drivers gain a better awareness of the road and its potential hazards as well as other drivers around them. Stop sign detection module is an integral part of these systems as it has great utility in automated or assisted driving applications. Although promising results have been achieved in the areas of stop sign detection and classification, these methods are heavily dependent on images and image processing algorithms and detection problem in the real world remains a challenging issue. In this paper, we propose a method to detect the stop sign, based on statistical analysis of data obtained by drive history, a project conducted by Ford Motor Company. Our detection algorithm is based on the speed profile where speed is zero or close to zero. Then, we apply clustering to our data set to extract the points with this common feature. In the end, results demonstrate that our algorithm can improve the stop sign detection efficiency and afford high precision where other algorithms are prone to failure.

Keywords: Cooperative control, Numerical algorithms, Switched systems
Abstract: Numerical simulations of physical phenomena governed by partial differential equations using parallel computing requires communication of data between processing elements. Time spent for this communication is a major part of overall simulation time on large scale systems and often becomes a bottleneck. In this paper, we propose an event-triggered communication scheme to alleviate this bottleneck. We model the system as a switched dynamical system and establish its stability. Experiments on a parallel cluster show that the method has potential to reduce simulation time while guaranteeing convergence to the correct solution.

Keywords: Cooperative control, Observers for Linear systems, Time-varying systems
Abstract: We study the problem of collaboratively estimating the state of a discrete-time LTI process by a network of sensor nodes interacting over a time-varying directed communication graph. Existing approaches to this problem either (i) make restrictive assumptions on the dynamical model, or (ii) make restrictive assumptions on the sequence of communication graphs, or (iii) require multiple consensus iterations between consecutive time-steps of the dynamics, or (iv) require higher-dimensional observers. In this paper, we develop a distributed observer that operates on a single time-scale, is of the same dimension as that of the state, and works under mild assumptions of joint observability of the sensing model, and joint strong-connectivity of the sequence of communication graphs. Our approach is based on the notion of a novel "freshness-index" that keeps track of the age-of-information being diffused across the network. In particular, such indices enable nodes to reject stale information regarding the state of the system, and in turn, help achieve stability of the estimation error dynamics. Based on the proposed approach, the estimate of each node can be made to converge to the true state exponentially fast, at any desired convergence rate. In fact, we argue that finite-time convergence can also be achieved through a suitable selection of the observer gains. Our proof of convergence is self-contained, and employs simple arguments from linear system theory and graph theory.

Keywords: Cooperative control, Robotics, Multivehicle systems
Abstract: This paper presents synchronous rendezvous conditions for minimally actuated and very short-range communicating mobile sensors in open sea environments. The working assumption is that the ocean currents of interest can be approximated by gyres or eddy flows arranged over a grid, in which each gyre is delineated by Lagrangian coherent structures. Sensor interactions can only occur between sensors in neighboring gyres, when they drift in close proximity, and over short time periods where the required distance is maintained. Within these application-dictated constraints, a cooperative synchronization controller is designed to establish and robustify periodic sensor rendezvous. Both the rendezvous conditions, as well as the rendezvous controller are tested and validated in simulation.

Keywords: Cooperative control, Time-varying systems, Lyapunov methods
Abstract: We study the synchronization problem for a network of planar harmonic oscillators with time-varying frequency. The oscillators are interconnected using a time-varying bidirectional graph. That is, the interconnections may be interrupted over some intervals of time, but a certain persistent connectivity prevails. We provide tight sufficient conditions on the graph’s connectivity to guarantee uniform exponential synchronization of the network. Our main results are based on original statements of stability for linear time-varying systems under persistency-of-excitation conditions and Lyapunov’s direct method.

Keywords: Cooperative control, Uncertain systems, Control over communications
Abstract: The interaction parameters in a system of interconnected cooperative agents (e.g. communication links) are susceptible to uncertainties. This paper provides systematic uncertainty quantification technique to study the effect of random parameters in the system of multiple agents governed by single integrator dynamics. In essence, the paper tries to answer the following question, ``Who is the weakest link?'' Sparse grid-based stochastic collocation in generalized polynomial chaos expansion framework is employed. Further, variance-based sensitivity indices are computed and analyzed to understand the significance of random variables on the probability distribution of the system responses. Numerical simulations are performed for uncertainty quantification and sensitivity analysis in the consensus and reference tracking problems as applied to multi-agent systems.

Keywords: Communication networks, Robust control, Distributed control
Abstract: In this paper, we study stabilization problems for a distributed power control algorithm (Foschini-Mijlanic algorithm) for wireless communication networks. Due to the practical importance of the algorithm, several methodologies for the stability analysis of the algorithm have been developed in the literature. However, the problem of stabilizing the dynamics of the algorithm by tuning its parameters has been left as an open problem. In this context, we consider the problem of adjusting transmission and interference gains of the channels in the communication network for ensuring the stability of the algorithm. The main results of this paper show that the following two stabilization problems, namely, the problems of input-output stabilization and stabilization subject to structural uncertainties, can be efficiently solved by geometric programming. We present numerical simulations for illustrating the effectiveness of the theoretical results obtained in the paper.

Keywords: Communication networks
Abstract: This paper considers the problem of designing the edge weights of directed graphs such that their Laplacian matrix has a prescribed spectrum. We provide a parametrization of the Laplacian matrix which is suitable for solving the problem numerically. We show how the edge weights can be further optimized to achieve secondary design goals and give an application example by designing the communication topology of a multi-agent system. Besides the general case of graphs with arbitrarily many vertices, we consider some special cases for small graphs and provide deeper insight into the solution to the problem in these cases.

Keywords: Network analysis and control, Communication networks
Abstract: In a networked system, any change in the underlying network structure, such as node and link removals due to an attack, could severely affect the overall system behavior. Typically, by adding more links and connections between nodes, networks can be made structurally robust. However, this approach is not always feasible, especially in sparse networks. In this paper, we aim to improve the structural robustness in networks using the notions of diversity and trustiness. Diversity means that nodes in a network are of different types and have many variants. Trustiness means that a small subset of nodes are immune to failures and attacks. We show that by combining diversity and trustiness within the network, we can significantly limit the attacker's ability to change the underlying network structure by strategically removing nodes. Using pairwise connectivity as a measure, we show that by appropriately distributing trusted nodes and assigning types to nodes, network robustness can be significantly improved. We analyze the complexity of diversifying and computing a set of trusted nodes, and then present heuristics to compute attacks consisting of node removals. We also present heuristics to defend networks against such attacks by distributing node types and trusted nodes. Finally, we evaluate our results on various networks to demonstrate the usefulness of our approach.

Keywords: Distributed control, Communication networks, Optimization algorithms
Abstract: We address the Desynchronization problem of achieving an equally spaced transmission schedule in a cooperative fashion. This problem arises in a shared medium communication and is of importance to achieve a fair multiple access schedule at the Medium Access Control (MAC) layer in the context of Wireless Sensor Networks (WSNs). In this paper, we investigate the convergence rate of different optimization algorithms and the potential benefits of addressing the problem as a solution of a set of linear equations. Initial results suggest that the Gauss-Seidel method can yield a faster convergence than previously proposed methods that employ a version of the Nesterov's method. Our approach also poses an interesting path for future research given the benefits of using other more advanced methods to solve systems of linear equations. Through simulations, we provide evidence to support future research on optimizing the parameter selection and also on categorizing the conditions under which one solution might be better in detriment of another.

Keywords: Networked control systems, Communication networks, Sensor networks
Abstract: Monitoring and maintaining communications in a multilayer ad hoc wireless network requires a stable communication overlay. Such monitoring should be resilient, avoiding reliance on a single (or a few) layers, and energy-aware, not being vulnerable to the exhaustion of available energy in a single network element. Thus, there is possibly a trade-off between overlay size, interlayer connectivity, and the energy distribution within the overlay. In this paper, we present three distributed energy-aware multilayer connected edge dominating set algorithms, show how they manage such trade-offs in practical scenarios, and show that CCEDS, a pruned centrality-based distributed algorithm, has the best performance.

Keywords: Control over communications, Kalman filtering, Markov processes
Abstract: In this paper, we consider a remote state estimation problem in the presence of an eavesdropper. A smart sensor takes measurement of a discrete linear time-invariant (LTI) process and sends its local state estimate through a wireless network to a remote estimator. An eavesdropper can overhear the sensor transmissions with a certain probability. To enhance the system privacy level, we propose a novel encryption strategy to minimize a linear combination of the expected error covariance at the remote estimator and the negative of the expected error covariance at the eavesdropper, taking into account the cost of the encryption process. We prove the existence of an optimal deterministic and Markovian policy for such an encryption strategy over a finite time horizon. Two situations, namely, with or without knowledge of the eavesdropper estimation error covariance are studied and the optimal schedule is shown to satisfy the threshold-like structure in both cases. Numerical examples are given to illustrate the results.

Keywords: Optimization algorithms, Statistical learning, Cooperative control
Abstract: We study a class of bilevel convex optimization problems where the goal is to find the minimizer of an objective function in the upper level, among the set of all optimal solutions of an optimization problem in the lower level. A wide range of problems in convex optimization can be formulated using this class. An important example is the case where an optimization problem is ill-posed. In this paper, our interest lies in addressing the bilevel problems, where the lower level objective is given as a finite sum of separate nondifferentiable convex component functions. This is the case in a variety of applications in distributed optimization, such as large-scale data processing in machine learning and neural networks. To the best of our knowledge, this class of bilevel problems, with a finite sum in the lower level, has not been addressed before. Motivated by this gap, we develop an iterative regularized incremental subgradient method, where the agents update their iterates in a cyclic manner using a regularized subgradient. Under a suitable choice of the regularization parameter sequence, we establish the convergence of the proposed algorithm and derive a rate of mathcal{O} left({1}/k^{0.5-epsilon}right) in terms of the lower level objective function for an arbitrary small epsilon>0. We present the performance of the algorithm on a binary text classification problem.

Keywords: Optimization algorithms
Abstract: We consider the distributed optimization problem in which a network of agents aims to minimize the average of local functions. To solve this problem, several algorithms have recently been proposed where agents perform various combinations of communication with neighbors, local gradient computations, and updates to local state variables. In this paper, we present a canonical form that characterizes any first-order distributed algorithm that can be implemented using a single round of communication and gradient computation per iteration, and where each agent stores up to two state variables. The canonical form features a minimal set of parameters that are both unique and expressive enough to capture any distributed algorithm in this class. The generic nature of our canonical form enables the systematic analysis and design of distributed optimization algorithms.

Keywords: Optimization algorithms, Uncertain systems, Stability of nonlinear systems
Abstract: In this paper, we analyze the convergence rate of the Heavy-ball algorithm applied to optimize a class of continuously differentiable functions. The analysis is performed with the Heavy-ball tuned to achieve the best convergence rate on the sub-class of quadratic functions. We review recent work to characterize convergence rate upper bounds for optimization algorithms using integral quadratic constraints (IQC). This yields a linear matrix inequality (LMI) condition which is typically solved numerically to obtain convergence rate bounds. We construct an analytical solution for this LMI condition using a specific ``weighted off-by-one'' IQC. We also construct a specific objective function such that the Heavy-ball algorithm enters a limit cycle. These results demonstrate that IQC condition is tight for the analysis of the tuned Heavy-ball, i.e. it yields the exact condition ratio that separates global convergence from non-global convergence for the algorithm.

Keywords: Optimization algorithms
Abstract: We present a computational analysis that establishes the O(1/K) convergence of the distributed gradient tracking method when the objective function is smooth and convex but not strongly convex. The analysis is inspired by recent work on applying dissipativity theory to the analysis of centralized optimization algorithms, in which convergence is proved by searching for a numerical certificate consisting of a storage function and a supply rate. We derive a base supply rate that can be used to analyze distributed optimization with non-strongly convex objective functions. The base supply rate is then used to create a class of supply rates by combining with integral quadratic constraints. Provided that the class of supply rates is rich enough, a numerical certificate of convergence can be automatically generated following a standard procedure that involves solving a linear matrix inequality. Our computational analysis is found capable of certifying convergence under a broader range of step sizes than what is given by the original analytic result.

Keywords: Estimation, Numerical algorithms, Network analysis and control
Abstract: We study the problem of finding the least-squares solution to the time-varying linear algebraic equation, zb(t)=Hb(t)yb(t), over an undirected network in a distributed manner. Each agent i has access to a time-varying row rector, hb_i^top(t) of Hb(t) as well as the corresponding entry z_i(t) in zb(t). The goal is to find the least-squares solution by communicating with neighbors over an undirected interaction graph. We propose a robust dynamic average-consensus algorithm, which allows the agents to precisely estimate the time-varying average signals needed to calculate the least-squares solution in a distributed manner. Since the estimation error corresponding to the presented dynamic average-consensus algorithm converges to zero in finite time, the proposed distributed algorithm yields the least-squares solution in finite time. Numerical simulations are provided to illustrate the effectiveness of the proposed algorithm.

Keywords: Stochastic systems, Optimization algorithms, Computational methods
Abstract: We develop a new method to solve the Fokker-Planck or Kolmogorov's forward equation that governs the time evolution of the joint probability density function of a continuous-time stochastic nonlinear system. Numerical solution of this equation is fundamental for propagating the effect of initial condition, parametric and forcing uncertainties through a nonlinear dynamical system, and has applications encompassing but not limited to forecasting, risk assessment, nonlinear filtering and stochastic control. Our methodology breaks away from the traditional approach of spatial discretization for solving this second-order partial differential equation (PDE), which in general, suffers from the "curse-of-dimensionality". Instead, we numerically solve an infinite dimensional proximal recursion in the space of probability density functions, which is theoretically equivalent to solving the Fokker-Planck-Kolmogorov PDE. We show that the dual formulation along with the introduction of an entropic regularization, leads to a smooth convex optimization problem that can be implemented via suitable block co-ordinate iteration and has fast convergence due to certain contraction property that we establish. This approach enables meshless implementation leading to remarkably fast computation.

Keywords: Nonlinear systems identification, Numerical algorithms, Metabolic systems
Abstract: Glucose-insulin metabolism models are useful tools for research on diabetes, in development of diabetes-related medical devices like artificial pancreas systems, insulin pumps and continuous glucose monitors, and may also play a role in personalized decision support tools for people with diabetes. Such models are often highly nonlinear with many parameters that are person dependent. An example is the model used in the UVa/Padova T1DM simulator, which has a large number of states and parameters. It is desirable to be able to personalize such models through parameter identification based on limited glucose, meal and insulin data obtainable from free-living settings, as opposed to clinical research settings that have traditionally been required. In this paper we use the UVa-Padova T1DM simulator model in a case study to investigate observability of the model under different measurements, and the identifiability of its parameters as a function of the model’s inputs and outputs. Structural identifiability is discussed and briefly investigated using the nonlinear Observability Rank Condition. Practical identifiability is discussed and investigated using sensitivity and Fisher information matrix analysis. We show how such analyses can be used to guide model reduction for improved identifiability, or to select the most proper subset of parameters to estimate.

Keywords: Biomedical, Nonlinear systems identification, Grey-box modeling
Abstract: In this paper, we discuss the identification of a physiological model describing the glucose-insulin dynamics in people with type 1 diabetes (T1D). The identified model has to be applied to nonlinear model predictive control (NMPC). We propose a stochastic model of the glucose-insulin dynamics in T1D. Discrete-time glucose data are provided by a continuous glucose monitor (CGM). We use maximum likelihood for parameter estimation, combined with a procedure to compute the gradient of the likelihood function. To test our identification procedure, we generate a virtual population of 10 patients using the Hovorka model and its parameter distribution. We report the estimates of the model parameters, and we use a validation dataset to evaluate the prediction errors for different prediction intervals. Whereas short-term predictions of blood glucose concentrations are consistent among patients, the accuracy of long-term predictions is more subject to inter-patient variability. The results suggest that this method has the potential to be used in NMPC algorithms.

Keywords: Human-in-the-loop control, Adaptive control, Intelligent systems
Abstract: The human insulin-glucose metabolism is a timevarying process, which is partly caused by the changing insulin sensitivity of the body. This insulin sensitivity follows a circadian rhythm and its effects should be anticipated by any automated insulin delivery system. This paper presents an extension of our previous work on automated insulin delivery by developing a controller suitable for humans with Type 1 Diabetes Mellitus. Furthermore, we enhance the controller with a new kernel function for the Gaussian Process and deal with noisy measurements, as well as, the noisy training data for the Gaussian Process, arising therefrom. This enables us to move the proposed control algorithm, a combination of Model Predictive Controller and a Gaussian Process, closer towards clinical application. Simulation results on the University of Virginia/Padova FDA-accepted metabolic simulator are presented for a meal schedule with random carbohydrate sizes and random times of carbohydrate uptake to show the performance of the proposed control scheme.

Keywords: Adaptive control, Simulation, Metabolic systems
Abstract: The insulin to carbohydrate ratio (CR) is a parameter used by patients with type 1 diabetes (T1D) to calculate the pre-meal insulin bolus and compensate postprandial glucose excursion. However, CR is known to vary over time, within and between days, hence tracking its variations is important for optimizing glucose control. Physicians periodically tune this parameter, by trial and error, based on empirical guidelines and patient's diary, but contemporary diabetes technology has the potential to move to CR adaptation. The aim of this work is to propose an algorithm to adapt patient's CR to physiological and/or behavioral changes based on minimally-invasive everyday-life technology data. We developed a run-to-run (R2R) algorithm for CR adaptation exploiting a physiology-based method for CR optimization. The algorithm retrospectively evaluates the quality of glycemic control and proposes, every 2 days, an adaptation of patient's CR by using patient's minimally-invasive data. The performance of the algorithm was assessed in silico using the single-day University of Virginia/Padova T1D simulator (Visentin et al., J Diabetes Sci Technol 2018) which incorporates a model for intra-day variability of insulin sensitivity and dawn phenomenon. The feasibility and robustness of the algorithm was tested in a 35-day scenario (7 days of run-in), with 3 meals per day, in 100 in silico subjects by including inter-day variability of insulin sensitivity (Toffanin et al., IEEE Trans Biomed Eng 2018) together with suboptimal CR or basal insulin infusion rate. Different values of the R2R gain (lambda) were tested, ranging from 0 to 1. In all simulations, CR adaptation improves glycemic control in a significant percentage of virtual subjects, within 5 weeks. Moreover, the method was safe also in case of suboptimal insulin infusion rate. Based on simulation results, a good compromise between safety and efficacy was achieved with lambda between 0.5 and 1. The proposed R2R algorithm for CR adaptation proved to be effective in silico. These results need to be confirmed clinically. The method can potentially be used in conjunction with algorithms for basal insulin adaptation and/or closed-loop control.

Keywords: Indirect adaptive control, Fault detection, Kalman filtering
Abstract: The emergence of real-time glucose sensors has prompted the development of closed-loop insulin delivery systems for type 1 diabetes patients, termed the artificial pancreas. The existing closed-loop systems rely on the user’s input to provide meal insulin boluses. However, patients, particularly adolescents, sometimes forget to announce consumed meals to the system. The performance of closed-loop systems after an unannounced meal may be improved with the addition of a meal detection module to the closed-loop system. We have developed a novel meal detection algorithm that detects unannounced meals using glucose measurements and insulin data. The model-based detection algorithm continually estimates an internal patient state using a linear Kalman filter. A generalized likelihood ratio test (GLRT) statistic is computed to evaluate the consistency of the Kalman filter under the null hypothesis that all consumed meals have been announced. A threshold criterion is applied on the GLRT to distinguish if the observed glucose increase is due to an unannounced meal. Simulation results, based on nonlinear time-varying virtual patients and noisy glucose measurements, show a sensitivity of 93.23% and a false positive rate of 4.17%. Moreover, 108 hours (4 patients x 3 visits x 9 hours) of clinical data is used to demonstrate the safety and feasibility of the meal detection module. Four patients underwent a nine-hour three-way inpatient experiment where the lunch meal was not announced to the system. The algorithm successfully detected all unannounced meals within 35 [30 - 40] minutes, without any false positives.

Keywords: Biological systems, Machine learning, Identification
Abstract: Glucose is a major source of energy for the human body and it is essential that blood glucose levels are maintained within a safe range. Type 1 Diabetes (T1D) is a metabolic disorder characterized by the deficiency of insulin, a hormone which is secreted by the pancreas and is responsible for blood glucose regulation. Thus, T1D patients need exogenous insulin injections to keep the blood glucose level within a safe range. However, the postprandial (PP) glucose regulation remains a challenging issue for diabetes treatments. In order to improve PP glucose concentrations, a data-driven modeling approach to adjust the meal-related insulin dose is proposed. Specifically, an individualized regression model able to correct the meal bolus computed with the conventional therapy is developed in order to handle the inter-patient variability characterising T1D patients that may affect PP glucose regulation. Moreover, the proposed approach exploits specific models for different day periods on the basis of the intra-day variability of insulin sensitivity. The individualized therapy is validated both on nominal and perturbed scenarios by using the UVA/PADOVA simulator, which is accepted by the FDA as a substitute for pre-clinical animal trials, and the results of a case study are presented in this work.

Keywords: Control applications, Identification for control, Model/Controller reduction
Abstract: In this paper we design and analyze a measurement feedback H2 controller to reduce the wake meandering behind a wind turbine. The control design and analysis proceeds in two steps. First, a linear reduced order model of the turbine is obtained using snapshots from a higher-order nonlinear 2D actuator disk model. The higher-order model includes a disturbance precursor generated to model realistic disturbance and turbulence scales. A measurement feedback H2 controller is then designed for the reduced order linear model assuming access to measurements at 8 downstream locations and the disturbance. The downstream measurement points are determined using insights from a static controller designed using full information of the wind field and incoming disturbances. The control performance is evaluated by simulations on the higher-order nonlinear model.

Keywords: Robust control, Uncertain systems, H-infinity control
Abstract: As the diameters of wind turbine rotors increase, the loads across the rotors are becoming more uneven due to inhomogeneous wind fields. Therefore, more advanced passive or active load reduction techniques are introduced to mitigate these uneven loads. Furthermore, measuring the disturbance can help to improve the control performance. This paper first examines how robust stability and performance are affected by uncertain sensor measurements when an integrator-based feedback is extended with an inversion-based feedforward individual pitch controller with similar bandwidth. A fixed-structured mathcal{H}_infty feedback-feedforward controller is proposed. The proposed feedback-feedforward controller ensures robust stability and performance and achieves better load reduction than a classical integrator-based feedback controller combined with inversion-based feedforward controller.

Keywords: Learning, Distributed control, Energy systems
Abstract: This paper presents a new algorithm for distributed Reinforcement Learning (RL). RL is an Artificial Intelligence (AI) control strategy such that controls for highly nonlinear systems over multi-step time horizons may be learned by experience, rather than directly computed on the fly by optimization. Here we introduce ADMM-RL, a combination of the Alternating Direction Method of Multipliers (ADMM) and reinforcement learning that allows for integrating learned controllers as subsystems in generally convergent distributed control applications. ADMM has become the workhorse algorithm for distributed control, combining the advantages of dual decomposition (namely, enabling decoupled, parallel, distributed solution) with the advantages of the method of multipliers (namely, convexification/stability). Our ADMM-RL algorithm replaces one or more of the subproblems in ADMM with several steps of RL. When the nested iterations converge, we are left with a pretrained subsolver that can potentially increase the efficiency of the deployed distributed controller by orders of magnitude. We illustrate ADMM-RL in both distributed wind farm yaw control and distributed grid-aware demand aggregation for water heaters.

Keywords: Control applications, Fluid power control, Stochastic optimal control
Abstract: Active power control for wind farms is needed to provide ancillary services. One of these services is to track a power reference signal with a wind farm by dynamically de- and uprating the turbines. Due to the stochastic nature of the wind, it is necessary to take this stochastic behavior into account when evaluating control signals. In this paper we present a closed-loop stochastic wind farm controller that evaluates thrust coefficients providing power tracking under uncertain wind speed measurements. The controller is evaluated in a high-fidelity wind farm model simulating a 9-turbine wind farm to demonstrate the stochastic controller under different uncertainty levels on the wind speed measurement and different controller settings. Results illustrate that a stochastic controller provides better tracking performance with respect to its deterministic variant.

Keywords: Distributed control, Large-scale systems, Optimization
Abstract: This paper presents a distributed approach to performing real-time optimization of large wind farms. Wind turbines in a wind farm typically operate individually to maximize their own performance regardless of the impact of aerodynamic interactions on neighboring turbines. This paper optimizes the overall power produced by a wind farm by formulating and solving a nonconvex optimization problem where the yaw angles are optimized to allow some turbines to operate in misaligned conditions and shape the aerodynamic interactions in a favorable way. The solution of the nonconvex smooth problem is tackled using a proximal primal-dual gradient method, which provably identifies a first-order stationary solution in a global sublinear manner. By adding auxiliary optimization variables for every pair of turbines that are coupled aerodynamically, and properly adding consensus constraints into the underlying problem, a distributed algorithm with turbine-to-turbine message passing is obtained; this allows for turbines to be optimized in parallel using local information rather than information from the whole wind farm. This algorithm is computationally light, as it involves closed-form updates. This approach is demonstrated on a large wind farm with 60 turbines. The results indicate that similar performance can be achieved as with finite-difference gradient-based optimization at a fraction of the computational time and thus approaching real-time control/optimization.

Keywords: Optimization algorithms, Kalman filtering, Machine learning
Abstract: Modern wind turbines require careful tuning of controller and estimator parameters. However, tuning requires expert control experience, and is therefore in practice often performed by a trial-and-error brute-force approach. The contribution of this work is twofold. Firstly, a framework for tuning the parameters for conventional control and estimator architectures with Bayesian optimization is proposed. Secondly, the proposed scheme is applied to the problem of tuning Kalman filter parameters for the estimation of the rotor effective wind speed. For accomplishing the beforementioned task, the Bayesian optimization machine learning algorithm uses entropy search as utility function. The NREL 5-MW reference wind turbine is used in high-fidelity simulation software to show the efficacy of the proposed methodology. The Bayesian optimized Kalman filter configuration, is shown to estimate the rotor effective wind speed with a root mean square error smaller than 5%, with respect to the actual effective wind speed over all load cases.

Keywords: Machine learning, Neural networks, Emerging control applications
Abstract: This paper describes a machine learning approach for controlling the flow of products from deep storage to the packing area in a retail distribution center (DC). The goal of the algorithm is to maximise earned revenue from expedited-delivery online orders, by extending the `booking deadline' for placing such orders on the retailer's website. Typically, the booking deadline is computed by summing process times backward from the promised time of delivery, including: (i) accessing the products from storage, (ii) moving the products to the packing area, (iii) packing and labelling, and (iv) shipping from warehouse to the customer location. Extension of the booking deadline by compressing the component processes has been studied extensively, with marginally reducing returns. This paper takes a different approach, by anticipating the type and quantity of products that will be ordered on a given day, and pre-emptively launching processes (i) and (ii) before actual orders are received. Anticipatory product picking is enabled by a machine learning algorithm that combines predictive analytics (for forecasting orders) and imitation learning (for computing picking decisions under capacity constraints). We show that forecasting using long short-term memory cells and decision-making using imitation learning on a post-facto optimal policy, is more successful in maximising earned revenues in a realistic test case, than several alternative approaches using heuristics and traditional time-series methods.

Keywords: Machine learning, LMIs, Stability of nonlinear systems
Abstract: Many data-driven control methodologies depend on an initial stabilizing control policy, and subsequently use operational data to refine the initial policy in order to optimize closed-loop performance. For general dynamical systems, computing such an initial policy is non-trivial, and systematic methods for this task are not available in the current literature. In this paper, we propose a systematic framework for constructing stabilizing and/or constraint-enforcing control policies for a class of nonlinear systems based on archival data. Specifically, we study partially unmodeled systems whose nonlinearities satisfy local Lipschitz conditions. We employ kernel density estimation (KDE) to learn a local Lipschitz constant from archival data, and compute control policies by solving semidefinite programs that leverage matrix multipliers informed by the Lipschitz learner. We demonstrate the potential of our proposed methodology on a nonlinear system with unmodeled dynamics.

Keywords: Machine learning, Game theory, Modeling
Abstract: This paper describes a stochastic modeling approach for predicting driver responses in highway traffic. Different from existing approaches in the literature, the proposed modeling framework allows simultaneous decision making for multiple drivers (>100), in a computationally feasible manner, instead of modeling the decisions of an ego driver and assuming a predetermined driving pattern for other drivers in a given scenario. This is achieved by a unique combination of hierarchical game theory, which is used to model strategic decision making, and stochastic reinforcement learning, which is employed to model multi-move decision making. The proposed approach can be utilized to create high fidelity traffic simulators, which can be used to facilitate the validation of autonomous driving control algorithms by providing a safe and relatively fast environment for initial assessment and tuning. What makes the proposed approach appealing especially for autonomous driving research is that the driver models are strategic, meaning that their responses are based on predicted actions of other intelligent agents in the traffic scenario, where these agents can be human drivers or autonomous vehicles. Therefore, these models can be used to create traffic models with multiple human-machine interactions. To evaluate the fidelity of the framework, created stochastic driver models are compared with real driving patterns, processed from the traffic data collected by US Federal Highway Administration on US101 (Hollywood Freeway) on June 15th, 2005.

Keywords: Machine learning, Neural networks, Intelligent systems
Abstract: Rapid freeform 3D printing of heated polymeric materials to create 3D curves in aerial space instead of 2D layers is a growing research area. We develop an arm motion planning technique that can increase printing accuracy without loss of the printing speed. The method is based on an actor-critic model-free deep reinforcement learning algorithm that operates over continuous arm action spaces. The proposed technique is able to find the trajectory for the arm that can neutralize the effect of gravity and build a filament with the desired shape. Experimental results are presented to illustrate the effectiveness of this technique in printing a straight line as well as a more difficult semicircle filament in the vertical plane.

Keywords: Machine learning, Robust control, Stochastic optimal control
Abstract: This paper motivates the need to incorporate the notion of robustness margins in the RL model free control design phase by showing an example where a naive design choice leads to an unusable real world controller. This is done by showing how a LQG control with bad robustness margins becomes a special case of RL control by design choice. Further a technique to avert this bad margins, namely control signal perturbation technique is proposed with in the model free RL framework. This techniques effectiveness is demonstrated on a real world flexible aircraft frame controller design

Keywords: Machine learning, Stability of nonlinear systems, Optimization
Abstract: Data-driven control offers an alternative to traditional model-based control. Most present data-driven control strategies either involve model identification or need to assume availability of state information. In this work, we develop an input–output data-driven control method through dissipativity learning. Specifically, the learning of the subsystems’ dissipativity property using one-class support vector machine (OC-SVM) is combined with the controller design to minimize an upper bound of the L2-gain. The data-driven controller synthesis problem is then formulated as quadratic-semidefinite programming with linear and multilinear constraints, and solved via the alternating direction method of multipliers (ADMM). The proposed method is illustrated with a polymerization reactor.

Keywords: Stochastic optimal control, Energy systems, Aerospace
Abstract: This paper presents a mission-centric strategy for control of the engine and electrical microgrid of a More Electric Aircraft (MEA). A Markov chain is used to represent the MEA mission profile with each state covering a stage in the mission and the probabilistic transitions between states representing the mission progression. Reference scenario trees are generated from potential pathways through the Markov chain and passed to a Stochastic Model Predictive Controller (S-MPC). The SMPC uses a scenario based formulation to optimize over a probability weighted average of the cost function with constraints enforced over all scenarios. Tight couplings in the dynamics of the subsystems require their careful coordination for safe and acceptable performance from the MEA. Simulations on a nonlinear model of the MEA demonstrate the ability of the SMPC to closely match the performance of an MPC with perfect preview of the future reference profiles and outperforms an MPC given no preview

Keywords: Energy systems, Modeling
Abstract: Flash boiling is a phase change cooling mechanism with the potential to meet stringent transient thermal management performance specifications across a range of applications. However, the dynamics of flash boiling are extremely fast and nonlinear and therefore create challenges in modeling and controlling a novel cooling system based on flash boiling. More specifically, the system is not only underactuated, but without the ability to measure vapor transport, sensing is limited too. Therefore, in this work we propose the use of combined plant and control design, or co-design, optimization to design both key process parameters and an optimal open loop control signal for this batch process. We use first principles to derive a nonlinear dynamic model of the system that captures both the pressure dynamics and vapor mass transport dynamics of the two-phase mixture created during the flash boiling process. We then formulate and solve a nonlinear co-design optimization problem and demonstrate efficacy of the proposed approach through simulation results.

Keywords: Hierarchical control, Predictive control for linear systems
Abstract: A multi-level hierarchical Model Predictive Control (MPC) formulation is presented for constrained linear systems with finite operation. The control formulation guarantees satisfaction of both state and input constraints during operation and terminal state constraints at the end of operation. Coordination between controllers at different levels is achieved through the use of waysets. These waysets form terminal constraints on lower-level controllers that provide flexibility in short-term operation while guaranteeing the long-term ability to satisfy constraints beyond the prediction horizon of these controllers. The waysets are represented as constrained zonotopes to enable computationally efficient on-line calculation. Constraint satisfaction is proven for a hierarchy with an arbitrary number of levels while a numerical example with a two-level hierarchy demonstrates the key features of the approach.

Keywords: Energy systems, Hierarchical control, Switched systems
Abstract: Many energy systems require control frameworks that can manage dynamics spanning multiple timescales and can make decisions for both continuous and discrete inputs. This paper meets this need for a class of switched power flow systems modeled using graphs. Conditions are provided under which each mode of these models belongs to the class of cooperative systems. Leveraging properties of cooperative systems, a two-level hierarchical control framework is constructed in which an upper level controller governs slow dynamics to plan for long-term future behavior and select modes, while a lower level controller governs fast dynamics to improve performance and reject disturbances. The control framework guarantees satisfaction of state constraints while also ensuring that a minimum bound on the rate of energy transfer to the system can always be achieved. The applicability and efficacy of the approach is demonstrated in simulation on a fluid-thermal system representative of those found in aircraft

Keywords: Energy systems, Modeling, Aerospace
Abstract: Hybrid unmanned aerial vehicles (UAVs) are gaining popularity in the defense sector. The introduction of a high-power electrical network provides new challenges in thermal management and safe vehicle operation. Existing efforts have focused on the modeling and control of thermal systems. However, the dynamic behavior of the electrical and mechanical components increases the complexity of the power management system. To enable model-based system design and real-time application tool development, this paper presents a graph-based modeling framework to represent the dynamic behavior of electrical and mechanical components onboard a UAV. An algorithm for composing a system-level graph model from component-level graph models is introduced. Cell and motor models are experimentally validated. A fault detection case is presented for a UAV model to demonstrate modeling capability for real-time applications.

Keywords: Power systems, Switched systems, Smart grid
Abstract: This paper proposes a strategy to control a group of thermostatically controlled loads (TCLs) such that the variability in their aggregate load is reduced. This strategy could be deployed in areas of a distribution network that experience voltage excursions due to net load fluctuations, such as areas with high penetrations of photovoltaic (PV) generation and/or electric vehicles (EVs). We limit variation in the power consumption of a group of TCLs using a control strategy previously developed for large aggregations of switched systems. Using this strategy, we constrain the number of TCLs that are on (i.e., actively consuming power) between upper and lower bounds. In simulations, the control strategy successfully decreases the range over which TCL power consumption varies. Percent reductions in range are greatest for medium group sizes: we find a median reduction of 82% for groups of 50 TCLs, 74% for groups of 1000 TCLs, and 59% for groups of 5 TCLs. Reducing the variability of a distribution network’s power injections helps to reduce voltage variability. In a simulation of a distribution line supplying 25 households, half with PV systems, the control strategy reduces the total range of voltage by 0.02 p.u. and prevents a violation of the 0.95 p.u. limit. Lastly, we propose a new control strategy for a more realistic TCL model that includes compressor lockout. The new strategy performs comparably to the original strategy and is demonstrated through simulation.

Keywords: Optimal control, Flight control
Abstract: There are various scenarios where Unmanned Aerial Vehicles are required to carry suspended load. The most common of these scenarios include; military deployments, tree logging, fire fighting, package delivery, sensor survey (e.g. Magnetometer survey of the earth's magnetic field) and aerial side trimming. This work is aimed at obtaining a rest to rest trajectory for a UAV suspended load such that the load vibration is minimized. The 2D-model of the Quadcopter is used. The result from differential flatness is compared with that obtained from a closed loop control with input shaping. Based on the integral of the absolute value of the ̇α swing rate of the pendulum, the three methods are compared and inferences are drawn.

Keywords: Game theory, Autonomous robots, Optimal control
Abstract: Interactions involving two friendly agents (termed prey and protector) working in the presence of a predatory (third) agent are considered in this paper. The three agents are modelled as differential drive mobile robots (DDMR). The task of the protector is to operate either in rescue mode or in interception mode to save the prey from the predator. To this end, we first perform feedback linearization on each robot model and then using a linear quadratic differential game (LQDG) approach, we design control strategies for each robot that achieve open-loop Nash equilibria. To facilitate switching (of the role of the protector) between rescue and interception while the game is in progress, we synthesize a receding horizon control policy. Simulations are presented to study the tradeoffs involved in feedback linearization of the robot models and to study the effectiveness of the synthesis strategy for switching between the modes. The simulations also illustrate a scenario where mere application of the open-loop Nash equilibrium strategy leads to capture of a prey while the protector mode switching (via the receding horizon policy) results in escape of the prey.

Keywords: Optimal control, Game theory, Learning
Abstract: In this paper, we develop a learning-based secure control framework for cyber-physical systems in the presence of sensor attacks. Specifically, we use several observer-based estimators to detect the attacks while also introducing a threat detection level function. We then solve the underlying joint state estimation and attack mitigation problems by using a reinforcement learning algorithm. Finally, an illustrative numerical example is provided to show the efficacy of the proposed framework.

Keywords: Optimal control, Iterative learning control, Autonomous systems
Abstract: An intermittent, model-free optimal control algorithm that enables an autonomous vehicle to track a non-predetermined trajectory at high speed is presented. The approach is bandwidth and energy efficient in that communication between actuators is limited to instances when it is needed rather than performing unnecessary periodic updates. We formulate the problem by properly augmenting the system and reference (trajectory) data, and then designing a triggering mechanism for the controller to work with a sampled version of the augmented states at some triggering instants. In order to obtain a model-free solution, we leverage a Q-learning framework with a zero-order hold actor network to approximate the optimal intermittent controller, and a critic network to approximate the optimal cost, resulting in appropriate tuning laws. Finally, we provide a numerical example of an ground vehicle driving autonomously at high-speed on a race track.

Keywords: Optimal control, Hybrid systems
Abstract: This paper addresses the initial condition optimal control problem in the presence of switching costs for hybrid systems which undergo autonomous switching. By rewriting the system's dynamics and cost function using the notion of occupation measures, the optimal control problem is formulated as an infinite-dimensional linear program over the space of measures. This measure-theoretic formulation is proved to be equivalent to the original problem. To solve this infinite-dimensional linear program, this paper constructs a sequence of semidefinite programs whose solutions converge from below to the true optimal as the degree of relaxation is increased. Moreover, if the solution to the relaxation has rank 1, the optimal initial condition can be recovered. The performance of the proposed method is illustrated on a pair of examples.

Keywords: Optimal control, Iterative learning control, Neural networks
Abstract: Graphical games are special classes of the standard differential games. The underlying neural network solutions are complicated and do not employ straightforward tuning laws. This issue becomes more challenging if the control strategies of the agents are constrained. An integral adaptive learning approach is developed to find an online solution for the differential graphical games with constrained control strategies. This solution employs a distributed adaptive policy iteration process in real-time. Local performance indices are utilized to assess the coupling between the agents and account for the constrained policies. Means of adaptive critics are used to develop a solution platform for each agent using single layer of neural networks, that are adapted using gradient descent tuning approach. This framework handles the main concerns related to the complexity and scalability of the distributed solution. The convergence of the adaptive learning solution is shown to hold under some graph-based conditions.

Keywords: Switched systems, Estimation, Automotive systems
Abstract: This paper investigates the design of interval observer for switched uncertain systems with bounded uncertainties that appear in both the state and output equations. The proposed observer provides an envelope covering all possible state trajectories. By means of multiple ISS-Lyapunov function as well as linear matrix inequalities formulation, the interval observer gain matrices are determined and sufficient conditions ensuring the Input-to-State Stability (ISS) of the estimation error are proposed. An application for vehicle lateral dynamics estimation is presented to evaluate the performance of the proposed algorithm. The interval observer is applied to estimate the sideslip angle in the presence of uncertain tire cornering stiffness parameters and time-varying vehicle longitudinal velocity. The obtained observer is evaluated using experimental data acquired with a prototype vehicle. The simulation results demonstrate the validity of the proposed design.

Keywords: Switched systems, Identification, Estimation
Abstract: The paper introduces a novel methodology for the identification of coefficients of switched autoregressive linear models. We consider the case when the system’s outputs are contaminated by possibly large values of measurement noise. It is assumed that only partial information on the probability distribution of the noise is available. Given input-output data, we aim at identifying switched system coefficients and parameters of the distribution of the noise which are compatible with the collected data. System dynamics are estimated through expected values computation and by exploiting the strong law of large numbers. We demonstrate the efficiency of the proposed approach with several academic examples. The method is shown to be extremely effective in the situations where a large number of measurements is available; cases in which previous approaches based on polynomial or mixed-integer optimization cannot be applied due to very large computational burden.

Keywords: Switched systems, Estimation, Control over communications
Abstract: While observers with asymptotic convergence guarantees can be used to design output feedback controllers when considering control tasks like stability, if state constraints relevant to safety exist, it is crucial to bound the estimation error at all times. In this paper, we propose an optimization-based design technique for bounded-error state estimators for affine systems that provide estimation guarantees in the presence of intermittent measurements. We treat the affine system as a switched system where the measurement equation switches between two modes based on whether a measurement exists or is missing, and model potential intermittent measurement patterns with a finite language that constrains the feasible mode sequences. By utilizing Q-parametrization, we show that an optimal estimator can be constructed that simultaneously provides an estimate of the continuous-state and implicitly estimates the specific missing data pattern (i.e., mode sequence), within the given language, according to the prefix observed so far. We illustrate with numerical examples that this approach significantly improves the achievable estimation bounds compared to earlier work.

Keywords: Switched systems, LMIs, Linear parameter-varying systems
Abstract: This document shows a dwell time based switching technique for affine linear systems. The affine system is defined as a parameter dependent homotopic combination of two base modes, where the intermediate modes are expected to create an ``artificial'' grid of subsystems with similar or closer dynamics between each other, producing lower dwell times estimates, specially when the number of modes rises, and in some practical applications, the interpolated modes may replace previously designed or existing modes over a defined variable or range. The Finsler's lemma is used to develop a relaxed Lyapunov-based condition that ensures the stability of the switched system and reduces the computational load of the developed technique. A user defined parameter allows to influence the dwell time estimation. Numerical calculations performed over a switched system derived from an adaptive vibration attenuation controller shows the effectiveness of the proposed algorithm

Keywords: Switched systems, Modeling, Manufacturing systems
Abstract: Flowrate control in Direct-Write (DW) Additive Manufacturing (AM) continues to be a challenge due to a capacitative energy storage in the system and the absence of suitable flowrate sensors at the micro-scale. Lack of precise control leads to an excess or a lack of ink while printing, resulting in manufacturing defects. The incorporation of a pressure sensor and a feedback controller is a potential approach for precisely controlling the flowrate. However, in the case where the ink loses contact with the pressure sensor, there will be an abrupt loss in feedback signal, which could potentially result in instability in the closed loop. In this paper, we present a hybrid model that represents the continuous dynamics of microextrusion with discrete switching between three different modes that captures the presence or loss of a sensed pressure. The simulation results demonstrate that the model captures the continuous dynamics of microextrusion while switching between three discrete modes: normal printing mode, retracted ink leading edge mode, and loss of pressure signal mode. The hybrid model presented here paves the way for switched controller synthesis to create stable feedback controllers.

Keywords: Switched systems, Optimization
Abstract: In this paper, we consider the problem of securely operating a cyber-physical system in an adversarial environment. The defending mechanism we introduce is proactive in nature and employs the principles of moving target defense. The defense implementation utilizes a switching structure to persistently and stochastically alter the behavior of the system with respect to both its actuators and its sensors. Thus, the ability of an adversary to successfully scan the system in preparation for the attack is decreased. The unpredictability of the system’s operation is quantified by an entropy metric which is subsequently optimized. Rigorous mathematical proofs are presented to show stability of the system under proactive switching. Simulations show the efficacy of the proposed approach on a simplified aircraft model.

Keywords: Nonlinear output feedback, Nonlinear systems identification, Automotive control
Abstract: This paper presents a practical demonstration of Nonlinear Model Predictive Control (NMPC) applied to the variable valve timing system of a 1 liter gasoline direct injection turbocharged engine over a transient maneuver. The nonlinear dynamic model utilized in the NMPC controller is a Takagi-Sugeno Neuro-Fuzzy model. The control problem has been formulated as a two-input-two-output multivariable control system with intake valve phase and exhaust valve phase angles as the two manipulated variables and engine fuel consumption and engine out NOx emissions as the two controlled variables. Two implementation methods were evaluated on the engine. In the first method, the NMPC using a sequential quadratic programming algorithm executed on a PC which was connected to the engine control unit over a controller area network. In the second method, the NMPC using a quasi-newton based optimization algorithm was deployed to and executed on the engine control unit. Results show the potential for >10% improvement in NOx emissions compared to the original engine strategy.

Keywords: Nonlinear systems identification, Estimation, Modeling
Abstract: This paper proposes a parameter estimation method for a kind of nonlinear auto-regressive (NARX) model, which is usually highly nonlinear because its parameters could vary very fast, since they are unknown nonlinear functions of past observations, called here as mapping-regressors. These parameters are poorly estimated by the standard recursive least-squares (RLS) filter since they vary much faster than standard time-varying parameters (TVP). So, our proposal reduces the fast parameters variability locally by reducing the a priori known mapping-regressors variability. This process is done by using both a reordering process according to the ascendant value of one of the mapping-regressors and the non-temporal windowing intersections of the remaining mapping-regressors. As a result, a set of local smoothed models, where a conventional recursive RLS filter works, is obtained. Experimentally, this approach works faster and simpler than alternative methods from the literature, which are discussed briefly through two simulated examples.

Keywords: Nonlinear systems identification, Estimation
Abstract: Identification of piecewise affine systems has been a major challenge in, e.g., modeling of hybrid systems. In this paper, we present an idea to exploit hidden sparsity in tight-dimensional representation spaces for piecewise affine system identification. More precisely, we propose a tight-dimensional linear transformation which reveals sparsity hidden in output signals of piecewise affine systems. This linear transformation is designed by applying a natural observation that most of output signals on neighboring data points are contained in special subspaces. Numerical examples show the effectiveness of the revealed sparsity for piecewise affine system identification.

Keywords: Nonlinear systems identification, Simulation, Estimation
Abstract: The Polynomial Nonlinear Auto-Regressive eXogenous input (P-NARX) model, a multivariable polynomial of past input and output values, is a widely used equation error nonlinear system model. The number of model parameters grows rapidly with the polynomial degree, and with the number of past inputs and outputs, but can be reduced significantly by adopting a decoupled structure, consisting of a transformation matrix followed by a bank of single-input single-output polynomials whose outputs are summed to produce the final output. Prediction Error Minimization (PEM) is a classical approach for the identification of both linear and nonlinear systems. Models trained using PEM may not be suitable for system simulation, where the model only has access to the system's inputs. In this paper, an identification method based on Simulation Error Minimization (SEM) for Decoupled P-NARX models is proposed. The proposed algorithm is applied to data from two nonlinear system identification benchmarks and the performance is compared to a previous PEM based algorithm.

Keywords: Nonlinear systems identification, Simulation, Model Validation
Abstract: Atomic force microscopes (AFMs) are used not only to image with nanometer-scale resolution, but also to nanofabricate structures on a surface using methods such as dip-pen nanolithography (DPN). DPN involves using the tip of the AFM to deposit a small amount of material on the surface. Typically, this process is done in open loop, leading to large variations in the amount of material transferred. One of the first steps to closing this loop is to be able to accurately and rapidly measure the amount of deposition. This can be done by measuring the change in the resonance frequency of the cantilever before and after a write as that shift is directly related to the change in mass on the cantilever. Currently, this is done using a thermal-based system identification, a technique which uses the natural Brownian excitation of the cantilever as a white noise excitation combined with a fast Fourier transform to extract a Bode plot. However, thermal-based techniques do not have a good signal to noise ratio at typical cantilever resonance frequencies and thus do not provide the needed resolution in the DPN application. Here we develop a scheme that iteratively uses a stepped-sine approach. At each step of the iteration, three frequencies close to the approximate location of the resonance are injected and used to fit a model of the magnitude of the transfer function. The identified peak is used to select three new frequencies in a smaller range in a binary search to reduce the uncertainty of the measured resonance peak location. The scheme is demonstrated through simulation and shown to produce an accuracy of better than 0.5 Hz on a cantilever with a 14 kHz resonance in a physically realistic noise scenario.

Keywords: Lyapunov methods, Delay systems, Stability of nonlinear systems
Abstract: This paper considers the stability and safety of nonlinear time delay systems. Utilizing a Lyapunov-Krasovskii functional we state and prove Lyapunov's theorem in its modern form and prove it with the help of the comparison lemma. Based on this we establish the notion of safety functional that allows us to ensure invariance of sets in the infinite-dimensional state space. The applicability of the results are also demonstrated using an illustrative example.

Keywords: Lyapunov methods, Mechanical systems/robotics, Identification for control
Abstract: This paper presents a nonlinear controller based on a double integrator system for trajectory tracking of a nonlinear autonomous surface vessel. Using a fixed-point on the vessel body frame as position output, we devise a simplified controller with only two error states, as opposed to the four required in a more straightforward and naive backstepping approach. The resulting controller is dynamically simple and easy to implement, and does not require higher than second-order plant dynamics and reference trajectory derivatives. The proposed approach mitigates typical problems arising during backstepping such as noise amplification and the need for a highly accurate plant model. Experimental results with an instrumented autonomous surface vessel are presented to corroborate the performance and robustness of the proposed controller.

Keywords: Lyapunov methods, Mechanical systems/robotics, Maritime control
Abstract: Dynamic surface control (DSC) methods use first order filters to compute the derivatives of stabilizing functions developed through backstepping. The performance of trajectory tracking DSC controllers can become degraded when disturbances are sudden and large. Here, we demonstrate that the robustness of DSC-based trajectory tracking controllers can be improved by augmenting the first order filters with a hyperbolic tangent vector function. Actuator constraints are included. The uniform semiglobal practical exponential stability of the augmented closed-loop control system is proved. Trajectory-tracking simulations of an autonomous underwater vehicle demonstrate the performance improvement achieved with the proposed approach.

Keywords: Lyapunov methods, Stability of hybrid systems, Biomedical
Abstract: A wide variation in muscle strength and asymmetry exists in people with movement disorders. Functional electrical stimulation (FES) can be used to induce muscle contractions to assist and a motor can be used to both assist and resist a person's volitional and/or FES-induced pedaling. On a traditional cycle with coupled pedals, people with neuromuscular asymmetries can primarily use their dominant (i.e., stronger) side to successfully pedal at a desired cadence, neglecting the side that would benefit most from rehabilitation. In this paper, a multi-level switched system is applied to a two-sided control objective to maintain a desired range of cadence using FES, an electric motor, and volitional pedaling. The non-dominant leg tracks the cadence range while the dominant leg tracks the position (offset by 180 degrees) and cadence of the first leg. Assistive, uncontrolled, and resistive modes are developed based on cadence and position for the non-dominant and dominant legs, respectively. Lyapunov-based methods for switched systems are used to prove global exponential tracking to the desired cadence range for the combined FES-motor control system.

Keywords: Control applications, Lyapunov methods, Optimal control
Abstract: A novel nonlinear control theory based feedback controller is proposed to damp the oscillations of the suspended load (pendulum) using the active modification of the length of the suspension string. This setting is a highly nonlinear one since the approximate linearization around the equilibrium working point is neither controllable, nor asymptotically stabilizable. The nonlinear design of the control law is therefore based on the conveniently selected control Lyapunov function. The resulting control law is then compared to the previously developed time-delay feedback control law, both in simulations and using the laboratory experimental realization of the suspended load system. Despite the fact that in the simulations the time-delay feedback control law suppresses the oscillations better than the nonlinear control law, in the experiments the performance of the time-delay feedback and of the nonlinear control law are rather similar. Moreover, the former keeps the pendulum string length oscillating, the latter stabilizes the nominal string length as well. Finally, the numerical optimization shows that the ideal damping would be provided by the impulsive-like control producing piece-wise constant string length dynamics.

Keywords: Stochastic systems, Delay systems, Lyapunov methods
Abstract: This paper deals with the feedback stabilization problem for a class of nonlinear neutral stochastic functional differential equations.The systems under consideration are nonlinear, nonaffine in control with discrete and distributed delays. They are described as Itô differential equation. We use a Razumikhin-type approach to establish sufficient conditions ensuring the stabilizability of the system and a class of stabilizing feedback is proposed.

Keywords: Fault detection, Distributed parameter systems, Observers for Linear systems
Abstract: This paper considers the actuator detection and estimation problem for a class of linear first-order hyperbolic partial integral differential equation (PIDE) systems. Based on the fault detectability analysis, a Luenberger-type observer is employed to achieve fault detection. However, in the case of actuator fault occurrence, modified Luenberger-type observers are developed such that actuator fault estimation is achieved in the presence of actuator fault while the system state estimation is realized. In comparison to the existing filter-based methods for distributed parameter systems, in proposed method in this manuscript, it is not necessary to transform the plant into the observer canonical form. The advantage of the proposed method is its flexible extension to other linear distributed parameter systems including all Riesz-spectral systems, as well as higher order nonspectral hyperbolic PDE systems. In particular, the proposed method is applicable to stable or unstable plants since the corresponding observation error systems are always stable, and therefore faults as well as plant state can be adequately estimated. Finally, an illustrative example is present to verify theoretical results.

Keywords: Fault detection, Fault diagnosis, Estimation
Abstract: The main objective of the paper is to provide an actuator fault estimation scheme, which takes into account various uncertainty sources. In particular, they are divided into three groups: sensor measurement noise, process external exogenous disturbances as well as unknown fault dynamics. Contrarily to the approaches presented in the literature, they are not treated in the same way but analyzed separately in a tailored fashion. Irrespective of the uncertainty source, the common approach is to minimize its effect on the fault estimation error in either H2 or Hinf sense. As a result, a mixed performance-based actuator fault estimation scheme is obtained and its convergence is carefully analyzed as well. The final part of the paper presents results obtained for the DC servo-motor and compares the proposed approach with an alternative one.

Keywords: Fault detection, Networked control systems, Linear systems
Abstract: The parity space approach is one of the most established model-based fault detection methods. The main purpose of this paper is to analyze the parity space-based fault detection monitoring system from the cyber security viewpoint considering an attack on the actuator channel. It is shown that under certain conditions a cyber attack can bypass the fault detection system based on the parity space approach without being detected. This weakness may be used by an adversary to conduct a cyber attack and remain undetected. Conditions of existence of a completely stealthy cyber attack are provided in the paper. Moreover, suggestions are given for a system developer to avoid completely stealthy attacks. An illustrative example is given to demonstrate the main results.

Keywords: Fault detection, Stochastic systems
Abstract: In this paper we study an attack design problem for interconnected systems where the attacker compromises a subsystem at each time, which is selected randomly based on a pre-computed probabilistic rule. The objective of the attacker is to degrade the system performance, which is measured based on a quadratic function of the system state, while remaining undetected from a centralized detector. First, we derive an explicit expression for the detection probability, analyze its properties, and compute an upper bound. Then, we use our upper bound to formulate and numerically solve a non-convex optimization problem for the computation of optimal attack strategies. Finally, we validate our results and show that our probabilistic attack strategy outperforms a deterministic attack strategy that compromises a fixed subsystem at each time.

Keywords: Fault detection, Variable-structure/sliding-mode control, Aerospace
Abstract: This paper describes the implementation of a sliding mode observer based fault detection and isolation (FDI) scheme, which is capable of reconstructing/estimating sensor faults, on the Japan Aerospace Exploration Agency's Multi-Purpose Aviation Laboratory (MuPAL-alpha) research aircraft. Hardware-in-the-loop (HIL) simulation results, which serve as a precursor to upcoming real test flights, are presented in this paper. The HIL simulation results show good yaw rate sensor fault reconstruction performance, and will be flight tested as the next stage of development and validation.

Keywords: Energy systems, Control applications, Fault accomodation
Abstract: Increased interest in renewable energy production has created demand for novel methods of electricity production. With a high potential for low cost power generation in locations otherwise isolated from the grid, in-stream hydrokinetic turbines could serve to help meet this growing demand. Hydrokinetic turbines possess higher operations and maintenance (O&M) costs due to their isolated nature and harsh operating environment when compared with other sources of renewable energy. As such, techniques must be developed to mitigate these costs through the application of fault-tolerant control (FTC) and machine condition monitoring (MCM) for increased reliability and maintenance forecasting. Hence, the primary objective of this paper is to address a key limitation in hydrokinetic turbine research: the lack of widely available data for use in developing models by which to conduct FTC and MCM. To this end, a 20 kW research hydrokinetic turbine implemented in Fatigue Aerodynamics Structures and Turbulence (FAST) is presented and housed within the Matlab/Simulink environment. This paper details the high-fidelity simulation platform development together with the characteristics of generated data with a focus on future FTC and MCM implementation.

Keywords: Control applications, Automotive control, Optimal control
Abstract: We present the design of a safe Adaptive Cruise Control (ACC) which uses road grade and lead vehicle motion preview. The ACC controller is designed by using a Model Predictive Control (MPC) framework to optimize comfort, safety, energy-efficiency and speed tracking accuracy. Safety is achieved by online computation of a robust invariant terminal set. The paper presents a novel approach to compute such set which is less conservative than existing methods. The proposed controller ensures safe inter-vehicle spacing at all times despite changes in the road grade and uncertainty in the predicted motion of the lead vehicle. Simulation results compare the proposed controller with a controller that does not incorporate prior grade knowledge on two scenarios including car-following and autonomous intersection crossing. The results demonstrate the effectiveness of the proposed control algorithm.

Keywords: Control applications, Constrained control
Abstract: In recent years, the research on Cyber-physical system becomes popular in the context of autonomous driving, building automation and robotics. With the implementation of digital computers and microprocessors, the controller design for such models require the consideration of several important factors, including computation resource constraints, limitation of actuator actions and safety. In this paper, we propose a control strategy that combines self-triggered control scheme with Control Barrier Function (CBF) to ensure safety, while still achieving the control objective efficiently. The control strategy works for general affine nonlinear system without the requirement to solve for a closed-form solution explicitly. Most importantly, the notion of safe period is introduced that enforces a strong safety guarantee for implementing Zeroing-Order Hold control.

Keywords: Control applications, Fault detection, Kalman filtering
Abstract: In this paper, the hydraulic cooling system of an automotive hybrid dual-clutch transmission is modeled for the purpose of observer-based fault-detection. The considered fault case is a blocking of a functionally relevant switching valve resulting in a critical lack of cooling capacity during the clutch actuation. Three different model-based fault detection schemes are presented, namely an augmented extended Kalman filter with a position-based fault detection strategy and two dual-model observers with a position-based and a gating-based fault detection strategy, respectively. The fault detection schemes are validated using data from a test bench showing promising results.

Keywords: Control applications, Genetic regulatory systems, Stochastic systems
Abstract: Sequestration of a protein by another decoy molecule, such that the protein is no longer available to perform its biological function, forms a fundamental layer of regulation in biomolecular systems. To quantify how fluctuations in protein level is controlled by decoys, we formulate a model where both proteins and decoys are stochastically expressed, with fast binding/unbinding of the protein to the decoy. Our analysis reveals that when the noise in the decoy copy number is small, the noise in the free protein numbers (as quantified by the Fano factor) monotonically decreases to the Poisson limit with the increasing average number of decoys. In contrast, for a high noise in decoys production, the response becomes nonmonotonic --- the noise level in protein counts is amplified at first with the increasing decoy numbers, before attenuating back to the Poisson limit. Motivated by recent biological examples, we next implement feedback control in the sequestration process by having the free proteins upregulate the decoy synthesis. Thus any random increase in the abundance of free proteins also results in higher decoy numbers, and hence more sequestered proteins. Intriguingly, our results show that as before, noise in free protein levels can get amplified with increasing decoys, albeit with a lesser magnitude as compared to the no feedback case. In summary, molecular decoys can play a key role in either amplifying or dampening the stochastic fluctuation of protein levels, and this study systematically quantifies this behavior across parameter regimes.

Keywords: Mechatronics, Control applications, Linear systems
Abstract: Discrete-time system inversion for perfect tracking goes at the expense of intersample behavior. The aim of this paper is the development of a discrete-time inversion approach that improves continuous-time performance by also addressing the intersample behavior. The proposed approach balances the on-sample and intersample behavior and provides a whole range of new solutions, with stable inversion and multirate inversion as special cases. The approach is successfully applied to a motion system. The proposed approach improves the intersample behavior through discrete-time system inversion.

Keywords: Biological systems, Mechatronics, Biomedical
Abstract: The motion of biological micro-robots - similar to that of swimming microorganisms such as bacteria or spermatozoa - is governed by different physical rules than what we experience in our daily life. This is particularly due to the low-Reynolds-number condition of swimmers in micron scales. The Quadroar swimmer, with three dimensional maneuverability, has been introduced for moving in these extreme cases: either as a bio-medical micro-robot swimming in biological fluids or a mm-scale robot performing inspection missions in highly viscous fluid reservoirs. Our previous studies address the theoretical modeling of this type of swimmer system. In this work, we present the mechatronic design, fabrication, and experimental study of a mm-scale Quadroar swimmer. We describe the design methodology and component selection of the system based on the required performance. A supervisory control scheme is presented to achieve an accurate trajectory tracking for all the actuators used in the swimmer. Finally, we have conducted several experiments in silicone oil (with 5000 cP viscosity) where two primary modes of swimming - forward translation and planar reorientation - have been tested and compared with the theoretical model.

Keywords: Estimation, Kalman filtering, Nonlinear systems identification
Abstract: State of charge (SOC) and state of health (SOH) estimation along with parameter identification of a Lithium-ion battery (LIB) are the primary steps towards the development of an efficient battery management system (BMS). In this paper, first, a novel nonlinear state space representation of the electric circuit model (ECM) of LIB, with SOC-varying electrical parameters, is presented as a switched system, i.e., for both charging and discharging cycles. A non-linear observer (NLO) is designed to simultaneously estimate the SOC and SOC varying internal parameters of the ECM. Second, the proposed state space representation is extended to include the change in ECM parameters with degradation due to temperature, ageing, capacity loss, and high C-rates such that the NLO can be used to estimate core temperature, surface temperature, and SOH along with SOC and time-varying parameters. The uniform ultimate boundedness (UUB) of the NLO’s state estimation error is guaranteed using Lyapunov stability analysis. Numerical simulation results are also presented to corroborate the efficacy of modeling and simultaneous estimation scheme.

Keywords: Estimation, Kalman filtering, Simulation
Abstract: The problem of sound source localization has attracted the interest of researchers from different disciplines ranging from biology to robotics and navigation. It is in essence an estimation problem trying to estimate the location of the sound source using the information available to sound receivers. It is common practice to design Bayesian estimators based on a dynamic model of the system. Nevertheless, in some practical situations, such a dynamic model may not be available in the case of a moving sound source and instead, some a priori information about the sound source may be known. This paper considers a case study of designing an estimator using available a priori information, along with measurement signals received from a bearing-only sensor, to track a moving sound source in two dimensions.

Keywords: Estimation, Lyapunov methods, Stability of nonlinear systems
Abstract: Full information estimation (FIE) is a type of optimization-based state estimation related to moving horizon estimation (MHE). In FIE, all available measurements (as opposed to those within a fixed horizon) are used to estimate the system's state. Although we expect it to produce a better state estimate than MHE because it includes more information, recent results have shown MHE to be robustly stable in the presence of bounded disturbances, while a general proof for FIE remains elusive. Lyapunov functions and ISS Lyapunov functions are invaluable tools for robustness analysis in model predictive control (MPC). Here, we present a Lyapunov-like function for FIE and use it to show that FIE is asymptotically stable in the absence of disturbances. This function is, to our knowledge, the first of its kind for nonlinear optimization-based state estimation, and it construction requires both an assumption of nonlinear stabilizability and a new type of storage function, termed an i-IOSS Lyapunov function, related to the property of incremental input/output-to-state stability (i-IOSS), a common form of nonlinear detectability.

Keywords: Estimation, Networked control systems, Communication networks
Abstract: Malicious attack against remote state estimation in cyber-physical systems has attracted considerable attention. Nevertheless, most existing works assume that the attacker is powerful and has mastered the communication information when designing the attack strategy. In this paper, we consider that the attacker is energy-limited and has no prior knowledge of transmission pattern. To solve this problem, we introduce a learning-based method for the attacker, which consists of a learning phase and an attack phase, to achieve a smart attack. We first formulate an optimal attack schedule problem, aiming to maximize the estimation error while considering the tradeoff between the learning accuracy and attack efficiency. Since it is hard to solve this problem directly, we split it into two subproblems: i) optimizing the attack pattern; ii) optimizing the eavesdropping times and attack times. Theoretically, we prove that the optimal attack pattern is that the learning phase precedes the grouped attack from the viewpoint of possibility. Furthermore, we propose an algorithm to design the rational learning times and attack times for the attacker. Numerical examples are used to demonstrate the effectiveness of the proposed method.

Keywords: Estimation, Nonlinear systems identification, Biological systems
Abstract: Models rational in the parameters arise frequently in biosystems and other applications. As with all models that are non-linear in the parameters, direct parameter estimation, using e.g. nonlinear least-squares, can become challenging due to the issues of local minima and finding good initial estimates. Here we propose a multi-step least-squares method for a class of nonlinear rational models. The proposed method is applied to an extended Monod-type model. Numerical simulations indicate that the proposed method is consistent.

Keywords: Estimation, Nonlinear systems identification, Kalman filtering
Abstract: The particle filter is a popular Bayesian filtering algorithm for use in cases where the state-space model is nonlinear and/ or the random terms (initial state or noises) are non-Gaussian distributed. We studied the behavior for the error in particle filter algorithm as the number of particles gets large. After a decomposition of the error into two terms, we showed that the difference between the estimator and the conditional mean is asymptotically normal when the resampling is done at every step in the filtering process. Two nonlinear/ non-Gaussian examples are tested to verify this conclusion.

Keywords: Linear parameter-varying systems, Estimation, LMIs
Abstract: A fixed-order set-valued observer is presented for linear parameter-varying systems with bounded-norm noise and under completely unknown attack signals, which simultaneously finds bounded sets of states and unknown inputs that include the true state and inputs. The proposed observer can be designed using semidefinite programming with LMI constraints and is optimal in the minimum H∞-norm sense. We show that the strong detectability of each constituent linear time-invariant system is a necessary condition for the existence of such an observer, as well as the boundedness of set-valued estimates. Furthermore, sufficient conditions are provided for the upper bounds of the estimation errors to converge to steady state values and finally, the results of such a set-valued observer are exhibited through an illustrative example.

Keywords: Linear parameter-varying systems, Fault detection, Optimization
Abstract: In this paper, we consider the problem of designing system parameters to improve detection of attacks in control systems. Specifically, we study control systems which are vulnerable to integrity attacks on sensors and actuators. We aim to defend against strong model aware adversaries that can read and modify all sensors and actuators. Previous work has proposed a moving target defense for detecting integrity attacks on control systems. Here, an authenticating subsystem with time-varying dynamics coupled to the original plant is introduced. Due to this coupling, an attack on the original system will affect the authenticating subsystem and in turn be revealed by a set of sensors measuring the extended plant. Moreover, the time-varying dynamics of the extended plant act as a moving target, preventing an adversary from developing an effective adaptive attack strategy. Previous work has failed to consider the design of the time-varying system matrices and as such provides little in terms of guidelines for implementation in real systems. This paper proposes two optimization problems for designing these matrices. The first designs the auxiliary actuators to maximize detection performance while the second designs the coupling matrices to maximize system estimation performance. Numerical examples are presented that validate our approach.

Keywords: Linear parameter-varying systems, Fault tolerant systems
Abstract: Considering that a faulty sensor may lead to degraded system performance, system instability, or even a fatal accident. In this work, the sensor performance degradation due to aging is modeled by the increment of sensor measurement noise covariance. The main contribution of this paper is the characterization of the OCC control synthesis conditions using linear matrix inequalities (LMI) for a gain-scheduled noisy output-feedback controller utilizes the estimated sensor noise covariance as part of the gain-scheduling parameters that minimizes the cost on control input with satisfactory system output covariance constraints (OCC control) in the presence of sensor aging. Another contribution is the use of the synthesized controller for possible detection of the effect of sensor aging. The closed-loop system performance in terms of control effort as a function of the output covariance and the sensor noise covariance is studied, and a numerical example is used to illustrate the effectiveness of the proposed control scheme. The synthesized controller guarantees the closed-loop OCC performance, and it is feasible for real-time applications.

Keywords: Linear parameter-varying systems, LMIs, Uncertain systems
Abstract: This paper is concerned with the problems of robust stability analysis and computation of H-2 or H-infinity (l2-gain) guaranteed costs for discrete-time linear systems with time-varying parameters. Two cases are investigated: i) the time-varying parameters are assumed to belong to a known interval and to have bounded rates of variation; ii) the time-varying parameters follow a known dynamics that can be represented through a discrete-time state space equation. A Lyapunov function that depends on the uncertain matrix of the system up to a certain degree k−1 provides certificates for robust stability and guaranteed costs that can be cast as linear matrix inequality optimization problems, with sharper results as k increases. Numerical examples illustrate that the proposed conditions can be more accurate than other techniques from the literature with lower complexity.

Keywords: Linear parameter-varying systems, Robust control, H-infinity control
Abstract: This paper proposes a sequential approach for designing switching LPV (Linear Parameter-Varying) H-infinity state-feedback controllers with given division of scheduling parameter subregions. Different from the traditional simultaneous design of switching LPV control that usually results in a relatively high-dimensional optimization problem, the proposed method designs the switching LPV controllers sequentially, leading to a bundle of low-dimensional optimization problems to be solved iteratively. Interpolated controller variables are utilized on the overlapped subregions and independent PLMIs (Parametric Linear Matrix Inequalities) are formulated on individual subregion to synthesize switching LPV controllers. At the same time, the formulated PLMIs are able to guarantee that the H-infinity performances on overlapped subregions are no worse than adjacent subregions. With the proposed method, the H-infinity performance over the entire scheduling parameter region can be guaranteed by sequentially designing individual controller and interpolating controller variables on overlapped subregions. Application example shows the effectiveness of the proposed method and demonstrates improved performance over the conventional simultaneous design approach.

Keywords: Automotive control, Linear parameter-varying systems, Observers for nonlinear systems
Abstract: Automatic transmission designs in automotive powertrains are expected to continue present trends towards higher numbers of transmission gear ratios with wider ratio spreads and smaller ratio steps, owing to their ability to improve fuel economy, while simultaneously satisfying stringent emissions norms. They are also expected to exhibit increased frequency of gearshifting, which requires re-evaluation of various aspects of the control of gearshifts. It is recognized here that clutch-to-clutch gearshift efficiency and quality (comfort) can be improved by minimizing the energy dissipated by the (oncoming) clutch during the speed synchronization (inertia) phase of a power-on upshift, which in turn would require control of both the engine and the (oncoming) clutch during the inertia phase. In order to achieve this control objective in a robust manner, it is proposed that both the engine and the (oncoming) clutch be under feedback control, unlike current practice in transmission control design where only the clutch is under feedback control. Gearshift control is particularly challenging for a 1-2 power-on upshift since the powertrain response is nonlinear due to the open torque converter clutch and, furthermore, a large ratio change is involved. The linear parameter varying (LPV) control technique is used here to design a gain-scheduled observer-based state feedback controller. It is shown through tests on a powertrain simulation that the resulting control solution is effective in controlling the inertia phase of gearshifts at different throttle openings, and is also robust to realistic modeling errors.

Keywords: Energy systems
Abstract: This talk presents a tutorial on the five W’s and one H of lithium-ion batteries. First, we provide an overview, from the materials to cell and pack construction up to the system level vehicle or grid-scale integration. We look at the types of models, sensors, and actuators that are required to monitor and protect the battery system. Next, we highlight the impact of load profiles and environmental factors, such as power levels and operating temperature, on performance and longevity. Finally, we close with the risks and hazards of inadequate protection and control schemes.

Keywords: Energy systems
Abstract: This talk presents a tutorial on estimation and control problems for battery electrochemistry models. Electrochemical (EChem) models present a remarkably rich set of control-theoretic questions involving model identification, state & parameter estimation, and optimal control. We discuss each of these fundamental systems and controls challenges, highlight recent results, and then present opportunities for future research.

Keywords: Energy systems
Abstract: Estimation of battery internal states and parameters, e.g. state of charge (SOC), state of health (SOH), and state of power (SOP), is one of the most important tasks of battery management. Estimation is typically performed using sensor data including current, voltage, and temperature. It is noted that different states/parameters will influence the measured output (e.g. voltage) differently, and for each variable, there will be certain input (current) patterns under which it has significant impact on the output and hence are highly observable from the data. These patterns, or data signatures, can be extracted by studying the sensitivity of the variables to the data based on the physical models governing the underlying input-output relationship. In this talk, we will discuss how to analytically derive the data signatures based on sensitivity analysis for battery phenomenological and electrochemical states/parameters and the potential use of them to optimize the accuracy of real-time estimation and offline system identification.

Keywords: Energy systems
Abstract: The complex nature of battery degradation mechanisms and their interactions has limited the application of model-based battery health prognosis approaches to real applications. On the other hand, the increase of available data from cells suggests data-driven methods could be effective. A range of approaches including Bayesian non-parametric methods such as Gaussian Process (GP) regression have been applied for modelling battery lifetime as a function of usage. This tutorial details how these types of techniques and others can be used for long-term prediction of the capacity fade of a battery. Firstly, an overview of the underlying principles is given and described in the context of the battery health problem. Secondly, a static fitting of state of health data in terms of cycles is formulated and its prediction is discussed. Lastly, a health ‘transitional’ model is proposed to capture the progressive evolution of the state of health changes, enabling their prediction under new input-varying scenarios.

Keywords: Energy systems
Abstract: This tutorial discusses the design of a nonlinear observer for the state of charge (SOC) estimation in a lithium-ion battery using measurements of both the terminal voltage and electrode expansion. The total cell expansion is the result of the simultaneous expansion of the negative electrode and contraction of the positive electrode during charging. For some chemistries, such as LiFePO4-graphite, this composite can result in an overall non-monotonic response with SOC. A nonlinear observer design based on Lyapunov analysis, that utilizes the lower and upper bounds of the Jacobian of the nonlinear output functions, is proposed. The non-monotonicity is addressed by designing a hybrid nonlinear observer that switches between several constant observer gains. The global stability of the switched system is guaranteed by ensuring overlap between regions and adequate dwell time between switches. The performance of the observer is evaluated through simulations and using experimental data. The performance of the nonlinear observer is compared with that of an extended Kalman Filter (EKF). Both perform accurately when the estimator model has very high fidelity. However, when modeling error is introduced, the EKF becomes unstable for even very small errors in the output curve. The nonlinear observer, on the other hand, continues to perform very well, providing accurate estimates and never becoming unstable.

Keywords: Energy systems
Abstract: This tutorial presents a feedback control strategy based on Reference Governors (RG) for the safe and fast charging of lithium-ion batteries and its experimental validation. Although battery rapid charging has been the subject of much research in recent years, mitigating battery aging while charging has been barely accounted for. The latter issue is overriding if degradation mechanisms like lithium plating occur, which can compromise the safe use of the battery. The proposed control scheme consists in two parts, both of them exploiting a reduced electrochemical model. The first part uses a linear quadratic regulator to push on the battery (fast) charging performance. The second part resorts to a RG to handle electrochemical constraints that enforce the proper (safe) operation of the battery. Since these constraints define a nonconvex region, a novel computationally efficient formulation of the RG was derived. The proposed feedback charge strategy was experimentally validated and contrasted with commercially available and widely used charging strategies such as constant-current/constant-voltage (CCCV). The proposed approach charges the battery faster than a high current CCCV while degrading its capacity as much as a mild current CCCV over one hundred full charge/discharge cycles. These encouraging results highlight the conservativeness of standard charging strategies obtained from empirical evidence with respect to physics-based control theoretic approaches.

Keywords: Robotics, Mechatronics, Control applications
Abstract: The Cassie bipedal robot designed by Agility Robotics is providing academics with a common platform for sharing and comparing algorithms for locomotion, perception, and navigation. This paper focuses on feedback control for standing and walking using the methods of virtual constraints and gait libraries. The designed controller was implemented six weeks after the robot arrived at the University of Michigan and allowed it to stand in place as well as walk over sidewalks, grass, snow, sand, and burning brush. The controller for standing also enables the robot to ride a Segway. A model of the Cassie robot and the controllers discussed in the paper are available on GitHub.

Keywords: Optimization algorithms, Robotics, Autonomous robots
Abstract: Practical bipedal robot locomotion needs to be both energy efficient and robust to variability and uncertainty. In this paper, we build upon recent works in trajectory optimization for robot locomotion with two primary goals. First, we wish to demonstrate the importance of (a)~considering and quantifying not only energy efficiency but also robustness of gaits, and (b)~optimization not only of nominal motion trajectories but also of robot design parameters and feedback control policies. As a second, complementary focus, we present results from optimization studies on a 5-link planar walking model, to provide preliminary data on particular trade-offs and general trends in improving efficiency versus robustness. In addressing important, open challenges, we focus in particular on discussions of the effects of choices made (a)~in formulating what is always, necessarily only an approximate optimization, in choosing metrics for performance, and (b)~in structuring and tuning feedback control.

Keywords: Robotics, Feedback linearization, Robust control
Abstract: For many planar bipedal models, each step is divided into a finite time single support period and an instantaneous double support period. During single support, the biped is typically underactuated and thus has limited ability to reject disturbances. The instantaneous nature of the double support period prevents control during this period. However, if the double support period is expanded to finite time, this introduces an overactuated period into the model which may improve disturbance rejection capabilities. This paper derives and compares the performance of two finite-time double support controllers. The first controller uses time to drive the progression of the double support period and controls the joint angles. The second controller uses a time-invariant phase variable to drive the progression of the double support period and controls the joint velocities since it is not possible to control the joint positions. The disturbance rejection capabilities of both controllers are then quantified using simulations. The instantaneous double support model is also simulated for comparison. The instantaneous double support model can recover from the largest disturbances but it requires the greatest number of steps to do. The time-based double support controller can recover from the smallest range of disturbances but requires the fewest number of steps for a given perturbation size.

Keywords: Human-in-the-loop control, Uncertain systems, Mechatronics
Abstract: In this paper we consider a class of exoskeletons designed to amplify the strength of humans through feedback of sensed human-robot interactions and actuator forces. We define an amplification error signal based on a reference amplification rate, and design a linear feedback compensator to attenuate this error. Since the human operator is an integral part of the system, we design the compensator to be robust to both a realistic variation in human impedance and a large variation in load impedance. We demonstrate our strategy on a one degree of freedom amplification exoskeleton connected to a human arm, following a three dimensional matrix of experimentation: slow or fast human motion; light or extreme exoskeleton load; and soft or clenched human arm impedances. We demonstrate that a slightly aggressive controller results in a borderline stable system—but only for soft human musculoeskeletal behavior and a heavy load. This class of exoskeleton systems is interesting because it can both amplify a human’s interaction forces—so long as the human contacts the environment through the exoskeleton— and attenuate the operator’s perception of the exoskeleton’s reflected dynamics at frequencies within the bandwidth of the control.

Keywords: Stability of hybrid systems, Stability of nonlinear systems, Robotics
Abstract: Hybrid systems theory has become a powerful approach for designing feedback controllers that achieve dynamically stable bipedal locomotion, both formally and in practice. This paper presents an analytical framework 1) to address multi-domain hybrid models of quadruped robots with high degrees of freedom, and 2) to systematically design nonlinear controllers that asymptotically stabilize periodic orbits of these sophisticated models. A family of parameterized virtual constraint controllers is proposed for continuous-time domains of quadruped locomotion to regulate holonomic and nonholonomic outputs. The properties of the Poincare return map for the full-order and closed-loop hybrid system are studied to investigate the asymptotic stabilization problem of dynamic gaits. An iterative optimization algorithm involving linear and bilinear matrix inequalities is then employed to choose stabilizing virtual constraint parameters. The paper numerically evaluates the analytical results on a simulation model of an advanced 3D quadruped robot, called GR Vision 60, with 36 state variables and 12 control inputs. An optimal amble gait of the robot is designed utilizing the FROST toolkit. The power of the analytical framework is finally illustrated through designing a set of stabilizing virtual constraint controllers with 180 controller parameters.

Keywords: Robotics, Hybrid systems, Stability of hybrid systems
Abstract: In this paper, we have theoretically synthesized a time-dependent global-position tracking controller for 3-D fully actuated bipedal robotic walking and experimentally validated it on a NAO bipedal humanoid robot. The proposed controller was theoretically established through full-order dynamic modeling, model-based state feedback control, and Lyapunov-based stability analysis, which can achieve reliable global-position tracking for enhancing walking versatility. Due to limited actuator accessibility of the NAO robot, necessary adaptation of the proposed walking controller was made for hardware implementation. Both simulation and experimental results validated the effectiveness of the proposed walking strategy. The results of this paper can be used to guide hardware implementation of global-position tracking control on legged robots with the common hardware limitation of actuator accessibility, thus helping to bridge the gap between theory and experiment.

Keywords: Optimal control, Automotive control
Abstract: This paper presents an optimal control strategy for heavy-duty trucks that minimizes fuel consumption in urban environments using infrastructure-to-vehicle communication. This strategy uses an online convex model predictive control strategy that balances a trade-off between reducing braking effort and mechanical energy expenditure and tracking an optimal desired velocity setpoint. The desired velocity profile comes from two different sources, depending on the vehicle's proximity to signalized intersections. In the presence of signalized intersections, the desired velocity profile is constructed to timely arrive at intersections when the light is green. With no intersections present, the desired velocity profile comes from a global, offline dynamic programming optimization, which is available to the vehicle through cloud connectivity. In parallel with MPC, we implement a continuous vehicle following controller to maintain safe vehicle following in the presence of traffic. Using a medium-fidelity Simulink model, based on a Volvo truck's longitudinal vehicle and engine dynamics, we characterize the control strategy's performance over a single city route with multiple signal timing and traffic patterns. Results demonstrate 10-25% reduction in fuel consumption, compared to a baseline control strategy. Furthermore, we show that this control strategy is computationally feasible for vehicle implementation. Finally, we present a detailed breakdown of where the fuel consumption reductions originate.

Keywords: Automotive control, Predictive control for nonlinear systems, Optimal control
Abstract: In recent years many works focusing on improved vehicle fuel efficiency through advanced control have been carried out, reflecting the high interest in ecodriving of vehicles. Although many studies have shown the potential that optimal control based ecodriving can offer, these solution are often difficult to be translated into online control strategies, one of the reasons being the complexity of the optimal control problem and therefore the computational burden.

To cope with this a novel online approach, based on switching Nonlinear Model Predictive Control (NMPC), is proposed. The NMPC strategy is developed for the case of conventional vehicles, where gear shifting and longitudinal dynamics are controlled. It is shown that our proposal can operate in real time, while recovering most of the performance achievable by an offline optimal solution. The development of the method is described in detail and its performance is analysed. The results show that the proposed NMPC can successfully solve the ecodriving task and seems a good compromise between computational burden and performance suitable for field implementation.

Keywords: Automotive control, Automotive systems
Abstract: The emergence of connected and automated vehicles (CAVs) provides an unprecedented opportunity to capitalize on these technologies well beyond their original designed intents. While abundant evidence has been accumulated showing substantial fuel economy improvement benefits achieved through advanced powertrain control, the implications of the CAV operation on power and thermal management have not been fully investigated. In this paper, in order to explore the opportunities for the coordination between the onboard thermal management and the power split control, we present a sequential optimization solution for eco-driving speed trajectory planning, air conditioning (A/C) thermal load planning (eco-cooling), and powertrain control in hybrid electric CAVs to evaluate the individual as well as the collective energy savings through proactive usage of traffic data for vehicle speed prediction. Simulation results over a real-world driving cycle show that compared to a baseline non-CAV, 11.9%, 14.2%, and 18.8% energy savings can be accumulated sequentially through speed, thermal load, and power split optimizations, respectively.

Keywords: Automotive systems, Optimization
Abstract: Among all advanced energy improvement technologies sought after to reduce the energy footprint, connected and automated vehicle (CAV) technologies hold a special promise of simultaneously improving energy efficiency, safety, and mobility. This work explored the significant opportunities exist in reducing the overall vehicle energy consumption through optimizing the vehicle dynamics and powertrain controls in a coordinated fashion utilizing preview information. A practical solution of Eco-Approach, -Departure, and -Cruise were proposed in this paper. These eco-applications were then integrated with autonomous driving modules. The integrated driver model can perform those eco-applications while navigating through a comprehensive traffic scenario designed on a high-fidelity virtual simulator developed for this project. Overall performance was compared with that of the baseline autonomous driver model to show the fuel economy improvement.

Keywords: Automotive control, Control applications
Abstract: This paper presents the design and implementation results of an ecological adaptive cruise controller (ECO-ACC) which exploits driving automation and connectivity. The controller avoids front collisions and traffic light violations, and is designed to reduce the energy consumption of connected automated vehicles by utilizing historical and real-time signal phase and timing data of traffic lights that adapt to the current traffic conditions. We propose an optimization-based generation of a reference velocity, and a velocity-tracking model predictive controller that avoids front collisions and violations. We present an experimental setup encompassing the real vehicle and controller in the loop, and an environment simulator in which the traffic flow and the traffic light patterns are calibrated on real-world data. We present simulation and experimental results, finding a significant potential for energy consumption reduction, even in the presence of traffic.

Keywords: Automotive control, Optimal control
Abstract: Eco-driving and hybridization both save fuel by reducing the energy loss during braking and using the engine efficiently. Considering this similarity, the hybridization requirements are different for the future connected and automated vehicles that can precisely optimize the vehicle velocity profile. This work investigates the necessary hybridization degree in an automated car following scenario. A novel hybrid system, called power split supercharger (PSS), is used in this study that is capable of working both as a parallel hybrid and a boosting device. The fuel minimization problem is divided into two independent optimizations, one for the velocity profile and the other for the energy management of the PSS. Dynamic programming is used to solve both problems. Different motor sizes are considered for the PSS system and the resulting fuel economies are compared to a turbocharged engine and a full hybrid electric (HEV) powertrain. The results show that the PSS system can provide 90% of the fuel economy gain achieved with a full HEV using a much smaller motor. For example, an optimized LA92 requires only an 18 kW motor or a 24 kW motor for the optimized US06 cycle as compared to the 60 kW motor for the full HEV.

Keywords: Aerospace, Variable-structure/sliding-mode control, Cooperative control
Abstract: er proposes sliding mode control based guidance strategies for an aircraft and defending missile team against an attacking interceptor. The proposed guidance strategies are generalizations of the line-of-sight guidance concept. By considering a nonlinear framework, problems that arise from linearization of engagement dynamics are circumvented. It is shown that knowledge of the bound on interceptor acceleration suffices to ensure its interception. Further, corresponding to different cost functions, optimal cooperative strategies are derived for the aircraft-defender team. Simulations vindicate the efficacy of the proposed strategies.

Keywords: Biologically-inspired methods, Autonomous robots, Cooperative control
Abstract: In this paper we propose a novel scheme that allows a group of mobile agents equipped with sensing capabilities to locate the unknown maximum of a scalar field. The scheme avoids the restrictions associated with gradient estimation, imposing predetermined formations on the agents or global localization. Instead the flocking approach is combined with a technique inspired by glowworm swarm optimization. Under reasonable assumptions we prove stability and boundedness of trajectories. 3-D simulation as well as 2-D experimental results illustrate that the proposed method outperform an existing technique in terms of smoothness of the trajectories.

Keywords: Biologically-inspired methods, Cooperative control, Networked control systems
Abstract: While individuals in natural swarms are collectively performing complex tasks such as foraging or synchronization, critical information such as predator warnings propagate across the swarm almost instantly and presumably without explicit communication between the individuals. In this paper, we propose a multi-layer control model composed of an interplay of decentralized algorithms for perception and swarming. Through this novel model, we demonstrate implicit information propagation and multi-tasking in swarms using only local interactions and without explicit communication or prescribed formations. For a complete graph, we prove that variations on individual speed are implicitly propagated across the swarm, which causes the swarm to turn almost instantly. Additionally, we prove that the spatial variances of the shape of the swarm are bounded, which implicitly ensures the connectivity of the graph. Finally, we provide various simulation results demonstrating the effectiveness of the model for swarms with complete and non-complete graphs performing collective synchronization and source seeking while avoiding a predator. The proposed model has the potential to be used to design various swarm algorithms such as designing tactics for a swarm of drones to avoid or chase a malicious agent.

Keywords: Nonholonomic systems, Cooperative control, Mechanical systems/robotics
Abstract: The control objective in the leaderless consensus-based formation control is to ensure that the Cartesian positions of the nonholonomic mobile robots converge to a given position in a formation pattern and the barycentre of such formation is agreed, in a decentralized manner, by all the robots, while their orientations converge to a common value. The main problem behind this stabilization objective is that for robots that exhibit nonholonomic restrictions, due to the Brocket's condition, the controller has to be designed such that it is discontinuous or it is non-autonomous (time-varying). In this work we propose a simple Proportional plus damping (P+d) smooth controller that solves the aforementioned objective and we provide a sufficient condition on the damping gain to ensure robustness with respect to variable time-delays in the interconnection. Simulations are provided to show the effectiveness of our control proposal.

Keywords: Nonholonomic systems, Cooperative control
Abstract: In this paper, the formation and trajectory tracking control problem for multi-agent systems is presented. A cascade control strategy for a group of nonholonomic mobile robots with non-negligible dynamics is proposed. An outer loop is a kinematic controller which has two different goals: virtual Leaders are defined for the purpose of trajectory tracking and pose regulation, and potential functions are defined for formation control. The cascade inner loop consists in a computed torque strategy. Assuming that the communication graph is always connected, a stability analysis ensures minimization of tracking and formation errors. Lastly, the control strategies are verified by simulation.

Keywords: Sensor networks, Optimization algorithms, Machine learning
Abstract: We study the problem of observation selection in a resource-constrained networked sensing system, where the objective is to select a small subset of observations from a large network to perform a state estimation task. When the measurements are gathered using nonlinear systems, majority of prior work resort to approximation techniques such as linearization of the measurement model to utilize the methods developed for linear models, e.g., (weak) submodular objectives and greedy selection schemes. In contrast, when the measurement model is quadratic, e.g., the range measurements in a radar system, by exploiting a connection to the classical Van Trees' inequality, we derive new optimality criteria without distorting the relational structure of the measurement model. We further show that under certain conditions these optimality criteria are monotone and (weak) submodular set functions. These results enable us to develop an efficient greedy observation selection algorithm uniquely tailored for constrained networked sensing systems following quadratic models and provide theoretical bounds on its achievable utility. Extensive numerical experiments demonstrate efficacy of the proposed framework.

Keywords: Sensor networks, Distributed control, Optimization
Abstract: Sensor network localization (SNL) is to determine physical coordinates of all sensors in a network given global coordinates of anchors and measurable distances among sensors and anchors. The SNL problem is generally NP-hard due to its nonconvex constraints. Many relaxation approaches have been proposed for solving SNL, among which semidefinite programming (SDP) relaxation is commonly viewed as an efficient method. However, since the rank constraint is ignored, the SDP relaxation requires a stringent graph condition to obtain the exact solution to SNL. In this paper, by considering the rank constraint, we solve SNL under a milder graph condition. To capture high efficiency and robustness, a distributed algorithm is also presented. We start with the centralized algorithm in which the rank constraint is converted into a linear matrix inequality and solved iteratively. Next, the SNL problem is decomposed into a group of subproblems and each subproblem is solved iteratively in a distributed manner. Furthermore, synchronous and asynchronous properties are analyzed for the proposed methods. Finally, simulation cases are presented to validate the improved localization accuracy, efficiency, and robustness by comparing to the state-of-the-art SNL method.

Keywords: Sensor fusion, Sensor networks, Linear systems
Abstract: Active-passive dynamic consensus filters consist of a group of agents, where a subset of these agents are able to observe a quantity of interest (i.e. active agents) and the rest are subject to no observations (i.e. passive agents). Specifically, the objective of these filters is that the states of all agents are required to converge to the weighted average of the set of observations sensed by the active agents. Existing active-passive dynamic consensus filters in the classical sense assume that all agents can be modeled as having single integrator dynamics, which may not always hold in practice. Motivating from this standpoint, the contribution of this paper is to introduce a new class of active-passive dynamic consensus filters, where agents have (homogeneous) linear time-invariant dynamics. We demonstrate that for output controllable agents, the output of all active and passive agents converge to a neighborhood of the weighted average of the set of applied exogenous inputs. A numerical example is also given to illustrate the efficacy of the presented theoretical results.

Keywords: Sensor networks, Adaptive systems, Observers for Linear systems
Abstract: This paper proposes a novel adaptive observer for heterogeneous sensor networks (HSNs) to estimate state vector of an unknown target or process by using the sensed output when the input to the target/process is also not known. In an HSN, nodes are considered either active or passive depending upon their ability to sense the target output. The local information exchange among the nodes is dictated by a connected graph. By using the criterion of collective observability, a novel distributed adaptive estimation is introduced where the nodes are allowed to have different sensor modalities. Stability analysis shows uniform ultimate boundedness of the state estimation and parameter estimation errors. Simulation results are included to validate the proposed approach.

Keywords: Sensor networks, Adaptive systems, Observers for Linear systems
Abstract: This paper proposes a novel event-triggered adaptive observer for each node in the heterogeneous sensor networks (HSNs) in order to estimate state vector of an unknown target or process by using the sensed output when the input to the target/ process is unknown. A subset of nodes in the HSN referred to as active nodes, can sense the target periodically, estimate the target state vector by using their adaptive observer and can communicate the estimated state vector of the target with the neighboring nodes including passive nodes only at event triggered instants. The adaptive observer parameters of active nodes are updated in a periodic fashion. A connected graph defines the local information exchange within the HSN. By using the criterion of collective observability, a novel distributed event-triggered adaptive estimation scheme is introduced where the nodes are allowed to have different sensor modalities. Using the Lyapunov analysis, uniform ultimate boundedness of the state estimation and the parameter estimation errors are demonstrated. Simulation results are included to validate the theoretical claims.

Keywords: Information theory and control, Sensor networks, Agents-based systems
Abstract: Computing communication-aware information for cooperative information gathering of multi-agent systems is important but still remains a challenge. In this paper, a mathematical formulation is presented for a communication-aware information measure for problems with non-linear dynamics and non-Gaussian distributions. This paper also presents a particle method for computation of the information measure in order to approximate target state and measurement PDFs given communication-aware measurements. The presented approach is assessed through simulations by comparing with the perfect communication scenario. The simulation results show how communication-aware information is different from the information under perfect communication and how imperfect communication affects the information measure.

Keywords: Modeling, Control education, Mechanical systems/robotics
Abstract: This paper introduces a new method for deriving equations of motion for mechanical systems subject to constraints. The method is easier and more intuitive to apply than Lagrangian mechanics in many situations. The alternative employs the principle that the net power (total derivative of energy with respect to time) is the sum of terms of generalized flow variables (velocities) multiplied by coefficients that are expressions of generalized effort (force). For conservative systems, the net power is zero. The key idea is that the flow variables can have arbitrary values, and thus their coefficients must be zero. Setting the expressions for coefficients equal to zero results in the equations of motion. The derivation for mechanical systems uses conservation of energy, making it is relatively more intuitive than the derivation of the Lagrangian. Extension to non-conservative mechanical systems with inputs and/or dissipation is straightforward by considering power gain and loss. Variations of the method are useful for other dynamic systems for which the principle of power equals effort acting over flow applies, such as electrical systems. The paper provides a motivating example, the derivation of the method for conservative and non-conservative mechanical systems, examples of its application to mechanical systems, and a variation of the method for an electrical system.

Keywords: Modeling, Estimation, Fluid flow systems
Abstract: To maintain the indoor environment quality effective estimation and control measures are required. In hazardous substances release scenario, a prompt estimation of contaminant distribution in the space is essential to enable quick control action. In this paper, we discuss the use of Perron-Frobenius (P-F) operator based approach for the design of an estimator to efficiently track the contaminant in an indoor environment. While the contaminants are evolving under the influence of nonlinear fluid flow field, the linear nature of the P-F operator is exploited for the design of estimator dynamics. In particular, the linear nature of the P-F framework is used for the design of a Kalman filtering algorithm to track the contaminants under an impulsive release scenario. Simulation results involving the International Energy Agency (IEA) two-dimensional building system model are presented to verify the main findings of the paper.

Keywords: Modeling, Energy systems, Human-in-the-loop control
Abstract: The goal of this study is to include a model of human decisions and preference in the control loop of a cyber-physical system (CPS). An electrical smart grid is selected as an example of such a CPS, and human decision and preference in utilization of thermostatically controlled loads (TCLs) are considered to be included in the control loop of TCLs. A two-stage planning and control framework is proposed. In the planning stage, a multi-objective dynamic program is developed for determining the expected optimal cumulative cost of operating a TCL over a planning horizon. The household demand is modeled with a series of Markov chains over the planning horizon and temperature comfort zone is represented by an interval as the preference. The day-ahead price of electricity is used for the price forecast, and the trade-off value between the cost of operating the TCL and the comfort obtained from its operation is user-specified. In the second stage, the expected optimal cumulative cost profile is used as the reference signal for the TCL's controller. The framework is simulated for an electric water heater unit to highlight its advantages.

Keywords: Modeling, Manufacturing systems, Stability of linear systems
Abstract: Additive manufacturing (AM) is an increasingly important enabler of smart manufacturing systems. Fused deposition modeling (FDM) is an AM technology that uses layer-wise extrusion to deposit material in 3D and most FDM machines run in open-loop. Due to the lack of feedback controllers, disturbances in the process cause failures. Although there are some spatial models for the FDM process, there are no layer-to-layer spatial dynamics models to enable control applications. This work proposes a novel modeling framework to capture the in-layer and layer-to-layer spatial dynamics of the FDM process utilizing a kernel basis approach. Individual kernels represent the deposition cross-sections and the orientation of the deposition. Layer-wise and layer-to-layer performance measures and stability definitions are proposed for the layer to- layer spatial model. A simulation example is provided to demonstrate the validity of the stability criteria.

Keywords: Modeling, Optimization, Machine learning
Abstract: This article proposes a novel data-driven adaptive robust optimization (ARO) framework for the unit commitment (UC) problem integrating wind power into smart grids. By leveraging a Dirichlet process mixture model, a data-driven uncertainty set for wind power forecast errors is constructed as a union of several basic uncertainty sets. Therefore, the proposed uncertainty set can flexibly capture a compact region of uncertainty in a nonparametric fashion. Based on this uncertainty set and wind power forecasts, a data-driven adaptive robust UC problem is then formulated as a four-level optimization problem. A decomposition-based algorithm is further developed. Compared to conventional robust UC models, the proposed approach does not presume single mode, symmetry or independence in uncertainty. Moreover, it not only substantially withstands wind power forecast errors, but also significantly mitigates the conservatism issue by reducing operational costs. We also compare the proposed approach with the state-of-the-art data-driven ARO method based on principal component analysis and kernel smoothing to assess its performance. The effectiveness of the proposed approach is demonstrated with the six-bus and IEEE 118-bus systems. Computational results show that the proposed approach scales gracefully with problem size and generates solutions that are more cost-effective than the existing data-driven ARO method.

Keywords: Biomedical, Kalman filtering
Abstract: One of the main challenges in automatic glycemic regulation in patients with type 1 diabetes (T1D) is to dispense with carbohydrate counting. In this context, we propose to equip a previously introduced switched Linear Quadratic Gaussian (LQG) controller—the so-called Automatic Regulation of Glucose (ARG) algorithm—with an automatic switching signal generator (SSG). The ARG algorithm not only regulates the basal insulin infusion rate but also generates feedback insulin spikes at meal times, i.e., no open-loop insulin boluses are needed to mitigate postprandial glucose excursions. However, in its former version, it was required to announce the meal time. In this work, the performance of the ARG algorithm combined with the proposed SSG is assessed in silico with unannounced meals. In addition, the response of the SSG is estimated using clinical data obtained with the ARG algorithm in the first-ever artificial pancreas (AP) trials carried out in Latin America.

Keywords: Biomedical, Control applications, Iterative learning control
Abstract: People with type 1 diabetes require exogenous insulin for adequate blood glucose regulation. Traditionally, the clinical therapy consists of multiple daily injections (MDIs) of insulin analogs and a finite number of self-monitoring blood glucose measurements (SMBGs) per day to achieve glycemic regulation. In this paper, we present simulation results for once-a-day dosing of long-acting insulin analog using iterative learning control (ILC) to deliver basal insulin. To facilitate validation of the control strategy for MDI, we propose modifications to a metabolic model for type 1 diabetes by adding states related to the subcutaneous insulin kinetics of long-acting insulin. Simulations on the cohort of in-silico patients demonstrate potential of the proposed strategy and its advantages over current clinical practice for basal insulin delivery. In particular, the ILC performs robustly under induced insulin resistance.

Keywords: Biomedical, Predictive control for nonlinear systems, Robust control
Abstract: This contribution presents an individualized Ensemble Model Predictive Control (EnMPC) algorithm for blood glucose (BG) stabilization and hypoglycemia prevention in people with type 1 diabetes (T1D) with detectable patterns of exercise. The EnMPC formulation can be regarded as a simplified multi-stage MPC that considers Nen future exercise patterns identified from the patient’s recent behavior. The control action is determined from a consensus across the ensemble, where each scenario is treated by a specific MPC algorithm, accounting for the underlying disturbance likelihood. The patient’s physical activity behavior is characterized by an exercise-specific input signal derived from the convolution of patients exercise records, e.g., obtained from activity monitors, and physiological impact curves from the literature. The proposed EnMPC strategy was tested on the complete in silico adult cohort of the FDA-accepted UVA/Padova metabolic simulator. Results confirm a tangible improvement in time spent below < 70 mg/dL (p<0.001) without increasing hyperglycemia after 30 min of moderate exercise in comparison to a similarly tuned MPC baseline controller (rMPC). Additionally, there was a significant reduction in the number of hypotreatments (p<0.001) that the patients received during and after exercise between the two controllers.

Keywords: Biomedical, Control applications
Abstract: A closed-loop glucose control system with automatic insulin and glucagon delivery (dual-hormone artificial pancreas) has the potential to reduce the self-management and the risk of hypo- and hyperglycemia in type 1 diabetic subjects. A novel dual-hormone closed-loop system based on a parallel control structure with intrinsic coordination among insulin and glucagon delivery is presented here, and the potential benefit of incorporating insulin-on-board limitation in such scheme is analyzed. To this end, the coordinated configuration (CC) has been extended with insulin-on-board (IOB) limitation through Sliding Mode Reference Conditioning (CC-SMRC), previously successfully tested in the context of single-hormone systems. Performance of CC and CC-SMRC has been compared through an in-silico study using the UVA-Padova simulator, extended to include various sources of variability. Three scenarios have been considered, comprising meals, snacks and exercise. The proposed coordinated strategy with the IOB limitation showed slightly lower time in hypoglycemia in meal and meals+snack scenario (0.00% vs 0.14% in meal scenario; 0.01% vs 0.11% in snack scenario), but they were not statistically significant (p=0.180 and p=0.179, respectively). However, the reduction during exercise scenario was statistically significant (1.45% vs 3.40%, p<0.001). Likewise, the time in range was similar in both configurations during meal and meals+snack scenarios (93.80% vs 94.13%, p=0.803, in meal scenario; 93.97% vs 94.32%, p= 0.356, in meals+snack; CC-SMRC vs CC), although it was greater in CC-SMRC during exercise scenario (92.98% vs 91.56%, p= 0.023; CC-SMRC vs CC). Moreover, insulin delivery was lower in CC-SMRC during the most demanding exercise scenario (45.91U/day vs 46.53U/day, p=0.001) at the expense of higher glucagon delivery to reduce hypoglycemia (1.03 ± 0.83mg/day vs 0.96 ± 0.79, p= 0.001). In conclusion, the coordinated configuration with insulin-on-board limitation demonstrated to be able to improve the performance of the coordinated configuration faced with exercise perturbation.

Keywords: Biomedical, Estimation, Computational methods
Abstract: We introduce a model-based meal detection and estimation method for the treatment of type 1 diabetes that automatically detects the occurrence and estimates the amount of carbohydrate (CHO) intake from continuous glucose monitor (CGM) data. Meal detection and estimation play a critical role in closed-loop insulin control by enabling automatic regulation of post-meal insulin dosing in artificial pancreas systems without manual meal announcements by the patient. Our approach to meal detection is based on a novel technique we call Committed Moving Horizon Estimation (CMHE), an extension of Moving Horizon Estimation (MHE). While MHE alone is not well-suited for disturbance estimation and meal detection, CMHE aggregates the meal disturbances estimated by multiple MHE instances to balance future and past information at decision time, thus providing timely detection and accurate estimation. We evaluated our CMHE-based meal detection and estimation method in-silico, using a nonlinear ODE gluco-regulatory model and random meal profiles to generate blood glucose and CGM signals. CGM data is used to detect meal occurrences and to estimate their onset, duration, and CHO amount. At the optimal operating point of the detector, we achieve an 88.5% daily detection rate and, more importantly, a 100% detection rate, with an average of 18.86 minutes onset deviation, and 70.50% CHO amount estimation accuracy for the main meals (i.e., excluding snacks).

Keywords: Optimal control, Biomedical, Optimization
Abstract: The objective of this paper is to develop an open loop insulin input profile over a span of 24 hours which makes the glucose trajectory of a Type 1 diabetic person track a target glucose trajectory. The Bergman minimal model is chosen to represent the glucose-insulin dynamics which is shown to be differentially flat. An optimal control problem is posed by parameterizing the differentially flat output of the Bergman model using Fourier series, to result in an input profile that can be repeatedly administered every day. The solution to the optimization problem is then shown to present acceptable performance in terms of tracking and adhering to imposed constraints.

Keywords: Energy systems, Robust control, Optimization
Abstract: Although spectral and pseudospectral methods have been used in a wide range of optimal control applications, to date, most of the literature uses these methods in a non-robust sense without considering possible dynamic deviation (uncertainties) from the nominal model. This study applies a recent robust approach for spectral and pseudospectral methods to a wave energy converter, considering structured uncertainty in the dynamical system. The results show that the robust approach gives better worst-case performance than an equivalent non-robust approach. Additionally, when structured uncertainty is considered in the dynamical system, the results show that the absorbed energy, obtained with the robust approach, is always positive. Finally, the advantages of this new approach are commented.

Keywords: Identification, Reduced order modeling, Optimization
Abstract: The motion of a Wave Energy Converter (WEC) can be described in terms of an integro-differential equation, which includes a convolution term accounting for the radiation forces. Since such a convolution term represents a drawback for both simulation and model-based estimation/control, it is usually approximated by a parametric form to be later embedded into the WEC dynamical equation. When an array of WECs is considered, a separate convolution term is required for each cross-coupling component (arising from device interactions), which increases the complexity of the problem. In this paper, a framework to compute a parametric model for array of WEC devices based on moment-matching is presented. The proposed method shows a significant simulation computational saving, compared to other parametric identification methods. The potential of the proposed formulation is illustrated by the means of a numerical example.

Keywords: Simulation, Maritime control, Optimal control
Abstract: During the design process and evaluation of energy maximizing control systems (EMCSs) for wave energy converters (WECs), control techniques rely heavily on numerical modeling. For fast computation, these numerical models are mostly based on low-fidelity boundary element method (BEM) codes and linear hydrodynamic models. However, to ensure optimal performance in a physical environment, more realistic, high-fidelity numerical frameworks, such as Computational Fluid Dynamics (CFD) based numerical wave tanks (CNWTs), should be considered during the evaluation of EMCSs. This paper investigates the influence of different numerical evaluation frameworks on the performance evaluation of EMCSs. The Wavestar WEC, subject to three different EMCSs with varying aggressiveness, i.e. resistive, reactive and moment-based control, is chosen as the case study. Results show that more aggressive EMCSs require high-fidelity numerical modeling to correctly evaluate their performance.

Keywords: Optimal control, Optimization, Control applications
Abstract: The roadmap to a successful commercialisation of wave energy inherently incorporates the concept of an array or farm of Wave Energy Converters (WECs). These interacting hydrodynamic structures require an optimised process that can ensure the maximum extraction of time-averaged energy from ocean waves, while respecting the physical limitations of each device and actuator characteristics. Recently, a novel optimal control framework based on the concept of moment, for a single WEC device, has been introduced in [1]. Such a strategy offers an energy-maximising computationally efficient solution that can systematically incorporate state and input constraints. This paper presents the mathematical extension of the optimal control framework of [1] to the case where an array of WECs is considered, providing an efficient solution that exploits the hydrodynamic interaction between devices to maximise the total absorbed energy.

Keywords: Observers for Linear systems, Control applications, Variable-structure/sliding-mode control
Abstract: A novel adaptive sliding mode observer (ASMO) is proposed to achieve the real-time excitation force estimation for wave energy converters in this paper. The main advantages of the proposed observer include robustness, fast convergence speed and high estimation accuracy. The proposed ASMO is proven to be finite-time convergent with a known convergence time limit, which allows one to estimate in advance when the proposed observer starts to provide accurate information. The robustness of the proposed ASMO is guaranteed by the sliding mode structure and the adaptive method. The coefficients of the proposed observer are time-varying according to the system states and a sliding mode variable is introduced to keep the estimated dynamics close to the actual dynamics. Simulation results show the effectiveness and superiority of the proposed ASMO by comparison with the Kalman Filter.

Keywords: Control applications, Optimal control
Abstract: This paper presents a case study of designing a linear non-causal optimal controller to the pitch type wave energy converter (WEC) M4. M4 is a multi-float multi-mode-motion WEC with complex dynamics and relatively high capture width compared to traditional WEC designs. There are mainly two motivations for this study. The first one is to demonstrate that, while existing model-based advance control methods are mostly investigated on simple wave energy converters (WECs) (e.g. single point absorber), they are also effective on WECs with more complex dynamics like M4 provided that a control-oriented model is properly built. The second one is to showcase that with the future incoming wave information explicitly incorporated into the controller, the energy conversion of WEC can be significantly improved.

Keywords: Machine learning, Statistical learning, Learning
Abstract: The matrix completion problem can be stated as follows: Suppose X is an unknown matrix except for an upper bound r on its rank. By measuring a small number m of elements of X, is it possible to recover X exactly, or at least, to construct a reasonable approximation of X? There are two existing approaches to choosing the elements to be measured, namely: choosing them at random or in some deterministic fashion. Practically all of the existing literature focuses on the case where the sampled locations are chosen at random. In contrast, the focus in the present paper is on deterministic methods for choosing the elements to be sampled. Specifically, we choose the measurement matrix to be the adjacency or biadjacency matrix of a so-called Ramanujan graph. Existing explicit techniques for constructing Ramanujan graphs lead to graphs whose degree d is bounded by n^1/3 where n is the number of vertices. However, for the matrix completion problem, this degree is too small. We point out that the well-known Lubotzky-Phillips-Sarnak construction of Ramanujan graphs can be used to generate graphs of very high degree, and use these for the matrix construction problem. Then we carry out numerical studies of "phase transition" in the matrix completion problem, by comparing the behavior of sampling at random with choosing the sampled locations using a Ramanujan graph. Specifically, we show that for a fixed sampling pattern, there is a maximum rank r̄ for which randomly generated matrices of rank r ≤ r̄ can be recovered with probability one, and the recovery probability drops very sharply to zero if r is increased by just two or three beyond r̄. The technique proposed here is applicable only to the recovery of square matrices. The recovery of rectangular matrices requires so-called asymmetric Ramanujan graphs, and is studied in a separate paper.

Keywords: Machine learning, Stochastic optimal control, Autonomous systems
Abstract: Fuel consumption in the vehicles are dependent on two variables of the route–deterministic and stochastic variables. The deterministic variable comprises length, elevation, curvature of road and stop signs along the route. The stochastic variable comprises variability in the velocity due to traffic and traffic lights. In this paper, we formulate the problem of traffic aware adaptive routing as a Markov decision problem (MDP) with total cost criterion featuring a continuum of state and finite actions. We show that the value iteration algorithm associated with our model would converge. Since it is impractical to implement value iteration on continuous spaces, we use Q learning with function approximation to perform numerical simulations to learn routing policies. Extensive simulations show that learned policy using Q learning outperforms the baseline algorithms that use offline and lookahead approaches when noise variances are high.

Keywords: Hierarchical control, Machine learning, Optimization algorithms
Abstract: A hierarchical control strategy is proposed to solve the optimal drainage problem in sewer systems by combining an optimization technique known as minimum scaled consensus control (MSCC) with the deep deterministic policy gradient (DDPG) algorithm. The MSCC strategy operates at the global control level, and is used to determine the flows of the hydraulic structures of the drainage system, such that the water is optimally distributed, i.e., wastewater flows are controlled to minimize saturation of water levels and/or flooding events, filling each of the drainage system components (e.g., pipes, tanks, wastewater treatment plants) proportionally to their capacity. On the other hand, the DDPG uses a model-free approach at the local control level, setting the drainage flows by operating valves and gates, without any knowledge of the inherent dynamics, so that it can be used to handle the nonlinearities of the system. Finally, a case study is presented to show the effectiveness of the proposed strategy.

Keywords: Learning, Machine learning, Biomolecular systems
Abstract: The Koopman operator has recently garnered much attention for its value in dynamical systems analysis and data-driven model discovery. However, its application has been hindered by the computational complexity of extended dynamic mode decomposition; this requires a combinatorially large basis set to adequately describe many nonlinear systems of interest, e.g. cyber-physical infrastructure systems, biological networks, social systems, and fluid dynamics. Often the dictionaries generated for these problems are manually curated, requiring domain-specific knowledge and painstaking tuning. In this paper we introduce a computational framework for learning Koopman operators of nonlinear dynamical systems using deep learning. We show that this novel method automatically selects efficient deep dictionaries, requiring much lower dimensional dictionaries while outperforming state-of-the- art methods. We benchmark this method on partially observed nonlinear systems, including the glycolytic oscillator and show it is able to predict on test data quantitatively 100 steps into the future, using only a single timepoint as an initial condition, and quantitative oscillatory behavior 400 steps into the future.

Keywords: Learning, Machine learning, Statistical learning
Abstract: In this paper we consider a problem known as meta-learning, consisting of building policies that achieve good generalization performance and adapt quickly to different tasks. In our novel formulation, which we denote by cross-learning, we introduce meta-learning through the coupled optimization of a set of rewards that are defined for different tasks. This coupling is effected by a projection step that brings the task-specific policies close to a central one which combines the information collected across tasks. Since such a projection is computationally expensive, we derive a relaxed version that can be obtained in closed-form through a geometric rule. While our initial cross-learning formulation is widely general, and connects with state-of-the art strategies, we specialize it for the case of reinforcement learning. In particular, we search for continuous policies in reproducing kernel Hilbert spaces, which are considered in order to avoid discretization and bypass parametric models. Preliminary numerical experiments performed on the classical cartpole system corroborate that the cross-learned control policy performs well in different scenarios.

Keywords: Stochastic optimal control, Machine learning, Autonomous robots
Abstract: In this paper, we address optimal control of nonlinear stochastic systems under motion and measurement uncertainty with finite control input and measurement spaces. Such problems can be formalized as partially-observable Markov decision processes where the goal is to find policies via dynamic programming that map the information available to the controller to control inputs while optimizing a performance criterion. However, they suffer from intractability in scenarios with continuous state spaces and partial observability which makes approximations necessary. Point-based value iteration methods are a class of global approximate methods that regress the value function given the values at a set of reference points. In this paper, we present a novel point-based value iteration approach for continuous state spaces that uses Gaussian processes defined over probability distribution for the regression. The main advantages of the proposed approach is that it is nonparametric and therefore approximation quality can be adjusted by choosing the number and the position of reference points in the space of probability distributions. In addition, it provides a notion of approximation quality in terms of variance.

Keywords: Delay systems, Networked control systems, LMIs
Abstract: We study a group of agents with linear time-invariant (LTI) dynamics commanded by a human operator with a general LTI model and a reaction time-delay. Under remote operation, the human receives time-delayed information from a subset of the agents, and commands, under another time-delay, the same or a different subset of the agents. Stability of this human-in-the-loop multiagent system is next studied both in "time-domain" and "frequency-domain", specifically to investigate how the system network graph affects the largest delay tolerable before losing stability. In a case study, system stability is shown to hold for larger delays in graphs with fewer edges than in a strongly-connected graph.

Keywords: Automotive control, Traffic control, Cooperative control
Abstract: Traffic waves, also known as “phantom jams”, are caused by the instability of the system describing the dynamics of traffic flow on highways: slight disturbances in the distribution of vehicles on highways are amplified when the density of the traffic is higher than a certain critical value and eventually generate traffic waves and “stop-and-go” phenomena. We propose a solution to the traffic wave damping problem via shared-control on vehicles and show that the effectiveness of the proposed controller does not depend on human drivers’ actions. In other terms, with the developed shared-controller the amplitude of the traffic wave is reduced regardless of the action of the drivers. Simulation results on a traffic control benchmark demonstrating the effectiveness of the controller are also provided.

Keywords: Adaptive control, Human-in-the-loop control
Abstract: This work proposes an Adaptive Haptic Aid system that adapts the amount of provided aid based on actual pilot performance. This is achieved by parameterizing the haptic system and adjusting the parameters on-line to match a desired closed loop performance. The parameters of the haptic aid are adjusted using a technique known as Model Reference Adaptive Control (MRAC), which has been widely studied in past years and applied to automatic control of plants with unknown dynamics. Simulations and experimental tests with naive and expert pilots show that the proposed Adaptive Haptic Aid system represents a promising solution for haptic aid design.

Keywords: Robotics, Automata, Human-in-the-loop control
Abstract: The curse of dimensionality is a challenge in many applications of Linear Temporal Logic robot motion planning, and is linked to the discretization of the robot’s workspace. The discretization aggravates the problem by introducing a multitude of atomic propositions. This paper argues that a large portion of these atomic propositions is unnecessary. It demonstrates this point by introducing local navigation functions within a temporal logic planning framework, and utilizing register automata for reactive motion planning without explicit, high-resolution workspace discretization. Motivation for this approach comes from applications in pediatric motor rehabilitation involving play-based social child-robot interactions, where the appropriate robot behavior in response to child actions is best described in temporal logic terms. A simulation example drawn from the aforementioned pediatric rehabilitation studies illustrates the advantages of the approach.

Keywords: Control applications, Human-in-the-loop control, Automotive control
Abstract: Trajectory tracking in the cut-in scenarios is challenging because the autonomous vehicles have to follow the reference trajectory and cooperate with the cut-in vehicles simultaneously. This paper proposes a trajectory tracking control method considering the cut-in vehicles with different behaviors. A model predictive control (MPC) approach incorporating driver behavior prediction is developed to track the reference trajectory and keep a safe distance with the cut-in vehicle. Moreover, the transient process of the cut-in scenario is considered for different cut-in behaviors. By synthesizing the driver behavior prediction with the trajectory tracking control, the relative distance between the autonomous vehicle and the cut-in vehicle gradually reaches the safe distance in the transient process. The designed controller is validated by CarSim® simulation. The simulation results show that the controller can not only track the reference trajectory, but also achieve a smooth transient process in different cut-in scenarios.

Keywords: Robotics, Uncertain systems
Abstract: Human motion prediction is an important component to facilitate human robot interaction. Robots need to accurately predict human's future movement in order to efficiently collaborate with humans, as well as to safely plan its own motion trajectories. Many recent approaches predict human's future movement using deep learning methods, such as recurrent neural networks. However, existing methods lack the ability to adapt to time-varying human behaviors. Moreover, many of them do not quantify uncertainties in the prediction. This paper proposes a new approach that uses a semi-adaptable neural network for human motion prediction, in order to accommodate human's time-varying behaviors and to provide uncertainty bounds of the predictions in real time. In particular, a neural network is trained offline to represent the human motion transition model. Recursive least square parameter adaptation algorithm (RLS-PAA) is adopted for online parameter adaptation of the neural network and for uncertainty estimation. Experiments on several human motion datasets verify that the proposed method outperforms the state-of-the-art approach with a significant improvement in terms of prediction accuracy and computation efficiency.

Keywords: Optimal control, Linear systems, Optimization
Abstract: A recent result on eigenstructure assignment allows for a linear feedback control to achieve desired autonomous periodic behavior for linear plants, where the oscillation pattern is characterized by a frequency and mode shape relative amplitudes and phase-offsets). We develop an optimal control theory based on the general result that parametrizes such controllers solving the eigenstructure assignment problem. The resulting controller represents an extension of the linear quadratic regulator, where the cost is a function of the transient portions of the state and control input, i.e. the portions that exponentially converge to zero as the state converges to the specified oscillation pattern. The optimal controller has a feedback structure that distinguishes it from a standard regulator about a fixed trajectory, achieving the oscillation pattern autonomously, with the absolute amplitude and phase timing adjusted by sensory feedback. A design example is presented to demonstrate the performance of the controller.

Keywords: Optimal control, Markov processes, Stochastic systems
Abstract: In this work, we present a new model-free and offpolicy reinforcement learning (RL) algorithm, that is capable of finding a near-optimal policy with state-action observations from arbitrary behavior policies. Our algorithm, called the stochastic primal-dual Q-learning (SPD Q-learning), hinges upon a new linear programming formulation and a dual perspective of the standard Q-learning. In contrast to previous primal-dual RL algorithms, SPD-Q learning includes a Q-function estimation step, thus allowing to recover an approximate policy from the primal solution as well as the dual solution. We prove a first-of-its-kind result that the SPD Qlearning guarantees a certain convergence rate, even when the state-action distribution under a given behavior policy is time-varying but sub-linearly converges to a stationary distribution.

Keywords: Optimal control, Optimization
Abstract: Consider that an agent, which moves on the two-dimensional coordinate space and is modeled as a linear time-invariant system, must be steered from a given initial condition towards an elliptical target region. This article presents a methodology to design control policies that minimize a cost subject to the terminal constraint that the target region is reached. In contrast to existing work, we consider that a time-costly elliptical area encompasses the target region. More specifically, we adopt a cost that linearly combines a quadratic term and a function of the duration of the time interval that starts when the agent first enters the time-costly area and ends when it reaches the target region. We propose a solution method that breaks up the problem in two stages (before and after entering the time-costly area), each modeled as an optimal control subproblem. The overall solution is obtained by solving an augmented problem that consists of the first stage subproblem subject to a terminal penalty determined by the optimal second-stage cost. We recast this problem, which is non-convex, as a quadratic program with two quadratic constraints. We obtain a solution by proving that strong duality holds when certain symmetry conditions are satisfied. A numerical example is provided that illustrates our method.

Keywords: Optimal control, Optimization algorithms, Feedback linearization
Abstract: In this paper, we study the optimal quadratic regulation problem for nonlinear systems. The linear operator theoretic framework involving the Koopman operator is used to lift the dynamics of nonlinear control system to an infinite dimensional bilinear system. Optimal quadratic regulation problem for nonlinear system is formulated in terms of the finite dimensional approximation of the bilinear system. A convex optimization-based approach is proposed for solving the quadratic regulator problem for bilinear system. Simulation results are presented to demonstrate the application of the developed framework.

Keywords: Optimal control, Optimization algorithms, Linear systems
Abstract: This paper presents a new pseudospectral method for solving optimal control problems with fractional orders including state and control input constraints. The proposed method employs an operational matrix of fractional-order differentiation discretizing the feasible optimal solution of the optimal control problem at Chebyshev-Gauss-Lobatto points. Besides, the Clenshaw–Curtis quadrature formula is used to discretize the performance integral. As a result, the optimization problem associated with fractional-order differential equations transforms into a nonlinear programming problem, which can be solved by means of well-developed techniques. The feasibility and effectiveness are illustrated by comparing the proposed method with other methods in a numerical example.

Keywords: Optimal control, Optimization algorithms, Variational methods
Abstract: This paper revisits the problem of synthesizing the optimal control law for linear time-varying systems by using the global optimal control framework introduced by Vadim Krotov. Krotov’s approach is based on the idea of total decomposition of the original optimal control problem (OCP) with respect to time, by an ad hoc choice of the so-called Krotov’s function or solving function, thereby providing sufficient conditions for the existence of global solution based on another optimization problem, which is completely equivalent to the original OCP. It is well known that the solution of this equivalent optimization problem is computed using an iterative method such as Krotov’s method, which may not be desirable for deploying low-cost hardware in industry. In this paper, we propose a novel method for synthesizing a global optimal control law using the sufficient conditions given by Krotov. The novelty of the proposed method lies in transforming the equivalent non-convex optimization problem into a convex problem by a proper selection of solving functions. As an immediate consequence, there is no need to compute an iterative solution.

Keywords: Switched systems, Stability of hybrid systems, Hybrid systems
Abstract: In this paper, we will present new stability conditions for a special class of linear switched systems, that evolves on non-uniform time domain. The considered systems switches between continuous-time subsystems on intervals with variable lengths and discrete-time subsystems with variable step sizes. Time scale theory is introduced to derive conditions for exponential stability of this special class of switched systems by using the dwell time approach. The conditions are based on the existence of a Lyapunov function which is non-increasing at the switching instants. This shown that this class of switched systems can be stabilized if the dwell time of each continuous-time subsystem is greater than some bound, and if the gap of the discrete-time subsystem is bounded by some specific values. Numerical examples are presented to show the effectiveness of the proposed scheme

Keywords: Switched systems, Stability of hybrid systems, Robotics
Abstract: Our objective in this paper is to establish robustness to disturbances for continuous-time switched systems with multiple equilibria (SSME) while being unaware of the disturbances. We provide an average dwell-time bound which can be computed without explicit knowledge of the disturbance. Switching signals that satisfy this bound ensure safe operation of the switched system for sufficiently small disturbances by the notion of practical stability. In essence, this paper establishes the robustness property that safe evolution of the SSME in the absence of disturbances results in the safety of the SSME under mild disturbances. Our motivation for studying SSMEs under disturbances arises from robotics, where certain motion planning problems require switching among controllers under unknown or unmodeled disturbances. However, these results are applicable to a much broader class of applications.

Keywords: Switched systems, Stability of linear systems
Abstract: The generalized Lyapunov operator spectral radius stabilizability condition, which is a sufficient condition for state-feedback exponential stabilizability in discrete-time switched linear systems, is considered and characterized in the present communication. It is shown that a switched linear system obeys the spectral radius stabilizability condition if and only if the considered switched system has a (finite length) transition matrix having spectral radius smaller than one. It is also proved that the satisfaction of the spectral radius stabilizability condition can be characterized in terms of a new (here introduced) sequence associated to the considered switched system. It is moreover shown that the solvability of a new dynamic programming equation, associated to the considered switched system, is a necessary and sufficient condition for the satisfaction of the spectral radius stabilizability condition.

Keywords: Switched systems, Variable-structure/sliding-mode control, Aerospace
Abstract: In this paper the problem of reducing stress on aeronautical electric energy generators is considered. The usage of an active Energy Storage Device (a controlled supercapacitor) is considered as a quick device able to absorb or yield quickly energy peaks caused by sudden intervention of Electro-Mechanical Actuators. The control strategy considered is a second-order sliding mode approach, able to guarantee finite-time achievement of the control goal. This characteristic is then exploited by a supervisory control to manage different objectives (including managing the State of Charge of the supercapacitor, with different levels of priority) with guaranteed stability. Detailed simulations in different situations confirm the effectiveness of the proposed strategy.

Keywords: Switched systems
Abstract: This paper considers an LQ control problem for continuous-time bimodal switched linear systems. A computation method for a continuous-valued input and a discrete switching signal that jointly minimize a finite-horizon quadratic cost is provided. Results are demonstrated by numerical examples.

Keywords: Pattern recognition and classification, Switched systems, Autonomous robots
Abstract: One approach for feedback control using high dimensional and rich sensor measurements is to classify the measurement into one out of a finite set of situations, each situation corresponding to a (known) control action. This approach computes a control action without estimating the state. Such classifiers are typically learned from a finite amount of data using supervised machine learning algorithms. We model the closed-loop system resulting from control with feedback from classifier outputs as a piece-wise affine differential inclusion. We show how to train a linear classifier based on performance measures related to learning from data and the local stability properties of the resulting closed-loop system. The training method is based on the projected gradient descent algorithm. We demonstrate the advantage of training classifiers using control-theoretic properties on a case study involving navigation using range-based sensors.

Keywords: Identification, Agents-based systems, Learning
Abstract: Learning the structure of the network from time-series data, in particular cyclostationary data, is of significant interest in many disciplines such as power grids, biology, finance. In this article, an algorithm is presented for reconstruction of the topology of a network of cyclostationary processes. To the best of our knowledge, this is the first work to guarantee exact recovery without any assumptions on the underlying structure. The method is based on a lifting technique by which cyclostationary processes are mapped to vector wide sense stationary processes and further on semi-definite properties of matrix Wiener filters for the said processes. We demonstrate the performance of the proposed algorithm on a Resistor-Capacitor network and present the accuracy of reconstruction for varying sample size.

Keywords: Identification, Energy systems, Optimization
Abstract: This paper discusses modeling and optimal experimental design for identification of a thermoacoustically unstable combustor system. We examine the impact of sensor placement, flame location, and acoustic excitation frequency on the Fisher identifiability of the parameters of a one-dimensional combustion stability model. The model uses linear delay differential equations to describe both the acoustics and heat release dynamics in a laboratory combustor often called Rijke tube. We derive analytic expressions for the frequency-domain Fisheridentifiability of the model’s underlying parameters. This leads to two key insights. First, excitation frequency, flame location, and sensor placement all have a significant impact on parameter identifiability. Second, the optimal excitation frequencies for identifiability are not a strong function of sensor placement but changing with flame location. Building on these insights, the paper concludes by using a genetic algorithm to optimize the design of a Rijke tube experiment for identifiability.

Keywords: Identification, Estimation, Linear systems
Abstract: This note extends previous results pertaining to algebraic state and parameter estimation of linear systems based on a special construction of kernel system representations that incorporate system differential invariants. Main results include explicit expressions for the kernel functions for single-input, single-output LTI systems of arbitrary order. A recursive regression type algorithm is also proposed for the purpose of joint system identification and finite interval filtering. As compared with previous results the proposed non-asymptotic estimation method proves remarkably robust to Gaussian noise in output measurements. The approach has been shown to extend to linear time-varying and linear parameter-varying systems in a multivariate setting. The idea of system-related kernels can further be employed to enhance convergence properties of moving-window and minimum energy nonlinear filtering methods.

Keywords: Closed-loop identification, Grey-box modeling, Linear systems
Abstract: The focus in this paper is identification of parameters in closed-loop systems. The system identification approach is based on a reformulation of feedback controllers in terms of its coprime factors. Based on this controller architecture, an auxiliary input is injected into the controller for the identification. The result of this setup is that the closed-loop identification problem is transformed into a standard open-loop identification problem. The models with unknown parameters are given in the form of a linear fractional transformation (LFT) where the direct term gives the nominal system and the other part gives the nonlinear maps of the variation of the unknown parameters into the system. Further, the coprime factors of this system can again be described as LFT of the nominal coprime factors and the parameter variations, i.e. an open-loop LFT identification problem.

Keywords: Identification, Estimation, Optimization
Abstract: In this paper we consider the blind identification problem of a single input single output (SISO) linear system. We exploit output oversampling to estimate the system parameter without statistical information on the unknown input signal. The identification problem is formulated as a suitable semialgebraic optimization problem that can be solved through computationally efficient convex relaxation techniques. The proposed approach provides a one-shot estimation of both the system parameters and the unmeasurable input sequence. A numerical example is given to show the effectiveness of the proposed approach.

Keywords: Identification, Estimation, Stochastic systems
Abstract: The influence model is a discrete-time stochastic model that succinctly captures the interactions of a network of Markov chains. The model produces a reduced-order representation of the stochastic network, and can be used to describe and tractably analyze probabilistic spatiotemporal spread dynamics, and hence has found broad usage in network applications such as social networks, traffic management, and failure cascades in power systems. This paper provides sufficient and necessary conditions for the identifiability of the influence model.

Keywords: Lyapunov methods, Stability of nonlinear systems, Stochastic systems
Abstract: Bell states are the maximally entangled states in two-qubit quantum systems, and play a significant role in quantum computation and quantum communication. Feedback control of stochastic quantum systems usually suffers from timedelay problems caused by the computation time for filter states and the control input. This paper addresses the preparation of Bell states in two-qubit systems with a constant delay time. A Lyapunov method is used to design a switching control law and a constant is introduced to compensate for the computation time of the estimated states and the feedback control input, thereby stabilizing the target Bell state globally. Numerical results show the effectiveness of the proposed control law.

Keywords: Lyapunov methods, Stability of nonlinear systems, Uncertain systems
Abstract: It is proved that the linear uncertain systems described by chains of integrators with unknown positive parameters can be stabilized by a kind of nested nonlinear feedback controllers with any positive gains. A Lyapunov/Chetaev function is constructed for the stability analysis of the closedloop system, and a necessary and sufficient condition for the asymptotic stability is derived by using the technique of homogeneous domination.

Keywords: Lyapunov methods, Uncertain systems, Optimization algorithms
Abstract: We present a method for synthesizing controllers to steer trajectories from an initial set to a target set on a finite time horizon. The proposed control synthesis problem is decomposed into two steps. The first step under-approximates the backward reachable set (BRS) from the target set, using level sets of storage functions. The storage function is constructed with an iterative algorithm to maximize the volume of the under-approximated BRS. The second step obtains a control law by solving a pointwise min-norm optimization problem using the pre-computed storage function. A closed-form solution of this min-norm optimization can be computed through the KKT conditions. This control synthesis framework is then extended to uncertain nonlinear systems with parametric uncertainties and L_2 disturbances. The computation algorithm for all cases is derived using sum-of-squares (SOS) programming and the S-procedure. The proposed method is applied to several robotics and aircraft examples.

Keywords: Lyapunov methods, Stability of nonlinear systems, Optimal control
Abstract: In this paper, we study the problem of constructing Lyapunov functions for nonlinear input-constrained systems with the largest possible stability regions by using a solution of the associated Hamilton-Jacobi-Bellman (H-J-B) PDE. To solve this equation, we employ a finite difference approximation and novel boundary conditions based on a recently-developed algorithmic construction of the boundary of the system's null controllable region, which efficiently determines all reachable states. Furthermore, since even smooth H-J-B PDEs are observed to contain discontinuities, the artificial viscosity perturbation method is used to improve the quality of the approximation. The sub-problem of determining the optimal constrained input at each node is reduced to finding the roots of a certain nonlinear polynomial. Lastly, we illustrate the results using simulation examples.

Keywords: Lyapunov methods
Abstract: This paper deals with the distribution of control effort in over-actuated nonlinear systems using control allocation (CA) methods. The developed CA methods rely on constrained optimization and have two key features. First, Lyapunov-based constraints and cost function are inserted into the optimization problem to improve the controller’s response to unattainable virtual inputs. Second, the CA formulation is enhanced with control barrier functions to enforce state constraints. The effectiveness of the proposed CA methods is further demonstrated through simulation tests of a motion controller for an over-actuated road vehicle.

Keywords: Lyapunov methods
Abstract: This paper investigates sufficient and necessary conditions for safety (equivalently, conditional invariance) in terms of barrier functions. Relaxed sufficient conditions concerning the sign and the regularity of the barrier function are proposed. Furthermore, via a counterexample, the lack of existence of an autonomous and continuous barrier function certifying safety in a class of autonomous systems is shown. As a consequence, guided by converse Lyapunov theorems for only stability, time-varying barrier functions are proposed and infinitesimal conditions are shown to be both necessary as well as sufficient, provided that mild regularity conditions on the system’s dynamics holds. Examples illustrate the results.

Keywords: Fault diagnosis, Differential-algebraic systems
Abstract: This paper focuses on designing a suitable input sequence for active fault diagnosis (AFD) of the discrete-time linear descriptor (DTLD) system. In order to successfully implement this objective, we propose to transform the descriptor system into a so-called equivalent slow-fast system. In this system form, the study of the slow subsystem is completely consistent with the normal linear system, while the system matrix of the fast subsystem is nilpotent. In this paper, the authors further propose to use the form of an augmented input sequence to describe the state of the fast subsystem, so that the design of an input sequence for the descriptor system can be completed by using the mixed integer quadratic programming (MIQP). Based on the established optimization objective function, this paper proposes a new optimization criterion, considering the smoothness of the input sequence and its energy, and adjusting the ratio of both of the specifications to obtain the optimal input sequence. At the end of this paper, a numerical example is used to illustrate the effectiveness of the proposed method and the new optimization criterion.

Keywords: Fault diagnosis, Fault detection, Large-scale systems
Abstract: The paper deals with a new active fault diagnosis framework for stochastic large scale systems, that are assumed to be decomposed into weakly coupled input-decentralized subsystems. The solution to the active detection problem is based on the multiple-model framework and behavior of each subsystem is represented by a set of models describing fault-free and faulty behavior of the subsystems. The centralized, decentralized, and distributed architectures of the active fault detector are introduced, the corresponding optimization problems are set up and approximate solutions to these problems are proposed. The behavior of the detector in the considered architectures is illustrated using a numerical example.

Keywords: Fault diagnosis, Observers for Linear systems, Uncertain systems
Abstract: This paper presents a sliding mode observer (SMO) scheme for robust fault reconstruction in descriptor systems that are not infinitely observable. The proposed scheme improves on existing schemes that do not consider the effect of disturbances which can corrupt the reconstruction, or require restrictive conditions. Some states are re-expressed in terms of other states, and certain other states are treated as unknown inputs, thereby forming a reduced-order system. A SMO is then implemented onto this reduced-order system to reconstruct the fault. Linear matrix inequalities (LMIs) are used to design the observer such that the L2 gain from the disturbances to the fault reconstruction is minimised. The existence conditions for the scheme are investigated. Finally, a simulation example is presented to demonstrate the efficacy of the proposed scheme.

Keywords: Fault diagnosis, Statistical learning, Time-varying systems
Abstract: The fault cases of complex industrial processes in general show typical nonstationary variations which reveal time-varying means or time-varying variances. The stationary fault information may be buried in the nonstationary fault variations and hard to be extracted. Besides, the existing fault diagnosis methods do not consider the underlying relations among different fault classes, which may lose important classification information. Here, it is recognized that different faults may not only share some common information but also have some specific characteristics. A fault diagnosis strategy with dual analysis of common and specific fault variations is proposed in this work. The nonstationary variables are first distinguished from the stationary variables by using Augmented Dickey-Fuller test. Then common and specific fault information are separated by developing two models for fault diagnosis. The fault-common model is constructed by cointegration analysis to capture the common nonstationary fault variations, and fault-specific model is built to explain the specific fault variations of each fault. With dual consideration of common and specific fault characteristics, the classification accuracy and fault diagnosis performance can be greatly improved. The performance of the proposed method is illustrated with both a real industrial process and a well-known benchmark process.

Keywords: Fault detection, Optimization algorithms, Aerospace
Abstract: This paper develops a defect diagnosis algorithm with multi-objective optimization scheme in the thermal wave image technique, in order to automatically detect defects under M/OD impact damages. In the process of defect detection, there are large amount of data and large noise interference in the thermal image sequence got by infrared imager, and feature detection can effectively avoid these disadvantages. Obtaining the fusion defect image message by selecting representative temperature points of the infrared thermal response points after clustering is an effective method in many feature detection methods. The work of our paper is proposing a new algorithm based on multi-objective optimization method to detect feature for infrared thermal image sequence(ITIS), which carries out the multi-performance analysis to more accurately choose typical temperature points(TTPs). The method takes into account the diversity of different temperature points(TPs) and the likeness of the same TPs. Using the Tchebycheff decomposition method to decompose the built multi-objective problem, and algorithm can find pareto solutions of the problem by using the competition pressure caused by Tchebycheff aggregation. Besides, to verify the feasibility and effectiveness of our algorithm, we carry out related experiments and analysis, and confirm that the algorithm has good performance in feature detection for ITIS.

Keywords: Estimation, Fault detection, Observers for Linear systems
Abstract: This paper presents a new method of disturbance estimation in discrete time invariant linear systems using game theoretic detection filters. First, the direct unknown input observer (UIO) design is reviewed, and its application in disturbance estimation is discussed. Motivated by the UIO design limitations, the game theoretic detection filter is introduced which decouples the multi-variable disturbances into a scalar target disturbance and remaining nuisance disturbances for a virtual plant model. The estimation of the target disturbance is then obtained by the UIO design for the virtual plant. This method can be used to estimate all disturbances in the system individually. Simulation and experimental results from an open loop unstable multi-input-multi-output (MIMO) Active Magnetic Bearing Spindle (AMBS) are presented to demonstrate the robustness of the proposed approach with respect to noises and unmodeled dynamics, and comparison to the direct UIO approach.

Keywords: Control applications, Optimal control, Human-in-the-loop control
Abstract: Improving a cyclist performance during a time-trial effort has been a challenge for sport scientists for several decades. There has been a lot of work on understanding the physiological concepts behind it. The concepts of Critical Power (CP) and Anaerobic Work Capacity (AWC) have been discussed often in recent cycling performance related articles. CP is a power that can be maintained by a cyclist for a long time; meaning pedaling at or below this limit, theoretically, can be continued for infinite amount of time. However, there is a limited source of energy for generating power above CP. This limited energy source is AWC. After burning energy from this tank, a cyclist can recover some by pedaling below CP. In this paper we utilize the concepts of CP and AWC to mathematically model muscle fatigue and recovery of a cyclist. Then, the models are used to formulate an optimal control problem for a time trial effort on a 10.3 km course located in Greenville SC. The course is simulated in a laboratory environment using a CompuTrainer. At the end, the optimal simulation results are compared to the performance of one subject on CompuTrainer.

Keywords: Genetic regulatory systems, Control applications, Stochastic systems
Abstract: Cellular reprogramming is traditionally accomplished through an open loop (OL) control approach, wherein key transcription factors (TFs) are injected in cells to steer the state of the pluripotency (PL) gene regulatory network (GRN), as encoded by TFs concentrations, to the pluripotent state. Due to the OL nature of this approach, the concentration of TFs cannot be accurately controlled. Recently, a closed loop (CL) feedback control strategy was proposed to overcome this problem with promising theoretical results. However, previous analyses of the controller were based on deterministic models. It is well known that cellular systems are characterized by substantial stochasticity, especially when molecules are in low copy number as it is the case in reprogramming problems wherein the gene copy number is usually one or two. Hence, in this paper, we analyze the Chemical Master Equation (CME) for the reaction model of the PL GRN with and without the feedback controller. We computationally and analytically investigate the performance of the controller in biologically relevant parameter regimes where stochastic effects dictate system dynamics. Our results indicate that the feedback control approach still ensures reprogramming even when both the PL GRN and the controller are stochastic.

Keywords: Control applications, Pattern recognition and classification, Predictive control for linear systems
Abstract: Predicting the future speed of a vehicle in driving situation is a challenging task due to the high nonlinearity of human driver’s control action in various driving environments. However, such a prediction is beneficial to a number of research areas including intelligent transportation system development, driver behavior evaluation and safety improvement. As an example, an accurate speed prediction model helps the driver prevent vehicle-to-vehicle crash by predicting crash imminent situations in advance. Most speed prediction models consider the acceleration as a control input in the longitudinal case. This paper proposes a novel method to predict the acceleration using the reference tracking model predictive control (MPC). Artificial neural networks (ANNs), a reference, are trained to design MPC to have the optimal control law that is the same as the true value of one step future acceleration. To identify various drivers’ driving abilities and to develop personalized models, personalized constraints are correspondingly proposed. By analyzing the feasibility, the proposed model is able to detect crash imminent situations based on each driver’s daily driving habits. With the personalized crash-imminence detection model, a hybrid MPC is implemented to produce an additional control input as a crash-avoidance measure.

Keywords: Control applications, Reduced order modeling, Optimization
Abstract: A major challenge of diesel engine development is the reduction of pollutant and CO2 emissions. For that reason new solutions for clean and efficient combustion are investigated. One promising approach is the so-called combustion rate shaping. In this case the entire in-cycle resolved pressure or combustion rate trace is controlled by manipulating the whole fuel injection rate profile. One substantial task within combustion rate shaping is therefore the realization of the continuously shaped fuel injection rate profile. State of the art series production injection systems do not allow for a continuously shaping of the injection rate profile. Instead multi-pulse fuel injections with multiple injection events are applied to approximate a continuous shaping. Each of these pulses is parameterized by corresponding actuation timings controlling the start and the duration of the injection event. In this contribution an optimization-based method for the digitalization of a continuous fuel injection rate profile is presented. For this purpose a reduced order injector model is presented. Additionally a suitable optimization problem is derived and presented along with numerical solution techniques.

Keywords: Control applications, Robust control, Fault tolerant systems
Abstract: This paper proposes a fault tolerant control (FTC) design for quadrotor UAVs subject to actuator faults. Its main objective is to ensures fixed time estimation and finite time accommodation of actuator faults. Timely fault estimation is achieved using a disturbance observer, and fault tolerance is attained using a Backstepping Integral Nonsingular Fast Terminal Sliding Mode Controller (BINFT-SMC). This latter is designed to ensure fast and global finite time stability of the states even when faults occur. Stability analysis of the system is carried over using the Lyapunov theory. The proposed approach was implemented to the AscTec Pelican quadrotor and assessed under partial actuator Loss of Effectiveness (LOE) faults and disturbances. The obtained results showed that the proposed approach was able to quickly accommodate faults without suffering performance degradation.

Keywords: Identification for control, Subspace methods, Process Control
Abstract: This paper presents a data-driven modeling and control formulation for achieving a desired product quality in a uni-axial rotational molding process. To this end, a data driven state-space model of the process is first identified using experimental data. For a given trajectory of input moves (heater and cold air profiles), this dynamic model is able to predict the evolution of the measured variable (internal product temperature). The dynamic model is augmented with a quality model, which, relates the terminal predictions from the dynamic model to the quality variables (sinkhole area, ultrasonic spectra amplitude, impact test metric and viscosity). The dynamic and quality model are in turn utilized within a model predictive control (MPC) framework to achieve tight quality control for new batches. Experimental results demonstrate the utility of the MPC in achieving improved and tight quality control.

Keywords: Estimation, Observers for Linear systems, Time-varying systems
Abstract: The objective of input reconstruction is to use knowledge of the output of a system to estimate the input. An estimate of the input can be used in an observer to provide more accurate state estimates. This paper presents a technique for combined state and input estimation for discrete-time, linear time-varying systems. The algorithm is based on the analysis of the rank of the time-dependent matrix that relates vectors of states and input values to a vector of outputs. The key contribution is a characterization of the time-varying delay under which the state and input can be estimated. The input and state estimate are given in terms of the generalized inverse of a partitioned matrix. The approach is demonstrated on discrete-time examples with linear periodically time-varying dynamics.

Keywords: Estimation, Observers for nonlinear systems, Autonomous robots
Abstract: This paper considers the problem of rigid-body attitude and rate-gyro bias estimation. Available measurements include a biased rate-gyro measurement and two or more linearly independent vector measurements. The gradient-based observer design methodology for Lie groups is employed to derive a provably convergent attitude observer. The observer propagates the state estimate along the gradient descent direction of a proposed attitude error function. The use of the proposed attitude error function, which is inspired by previous cost functions using the Gibbs parameterization, results in an innovation term that aggressively drives the attitude error to zero, leading to fast convergence of the observer. The innovation term can be constructed directly from the vector measurements. Numerical results are included that demonstrate the desirable convergence properties of the observer compared to previous results.

Keywords: Estimation, Observers for nonlinear systems, Vision-based control
Abstract: Estimating the pose (position and orientation) of 3D objects solely from images is a challenging task of current relevance to a wide range of robotic and artificial intelligence applications. In this paper, we employ the empirical local observability Gramian as a metric for assessing the quality of image-based pose estimation. We show how the Gramian can be used to describe the local ``estimatability'' of images of static and dynamic objects. We also show connections between the Gramian matrix and symmetries and near-symmetries of an object, and use this result to characterize the subspace of the pose vector which is unobservable using image data. Results are demonstrated using images constructed in a 3D virtual environment.

Keywords: Estimation, Optimization algorithms, Kalman filtering
Abstract: Proliferation of distributed Cyber-Physical Systems has raised the need for developing computationally efficient security solutions. Toward this objective, distributed state estimators that can withstand attack on agents (or nodes) of the system have been developed, but many of these works consider the estimation error to asymptotically converge to zero by restricting the number of agents that can be compromised. We propose resilient distributed Kalman filter (RDKF), a novel distributed algorithm that estimate states within an error bound and does not depend on the number of agents that can be compromised by an attack. Our method is based on convex optimization and perform well in practice, which we demonstrate with the help of a simulation example. We theoretically show that, in a connected network, the estimation error generated by the distributed Kalman filter and our RDKF at each agent, converges to zero in attack free and noise free scenario. Furthermore, our resiliency analysis result shows that the RDKF algorithm bounds the disturbance on the state estimate caused by an attack.

Keywords: Identification, Estimation, Modeling
Abstract: Traditional observation-to-track data association algorithms, such as Bayesian methods, suffer from exponential growth in computational complexity with increase in a number of targets, especially in dense environments with low signal-to-noise ratio. This paper utilizes tensor decomposition, a commonly used technique to tackle “curse of dimensionality” in high dimensional applications, to reduce the aforementioned complexity in data association problems. The Joint Probabilistic Data Association (JPDA) is a pseudo-Bayesian sub-optimal filter that lends itself to be modified within the framework of incremental tensor decomposition to reduce its computational burden resulting from computing the probabilities of noncompeting join association events or “feasible events”. The number of feasible events is reduced by a using a “core” tensor instead of the complete set of measurements received at a specified instance in time. Furthermore, to reduce computational overhead while performing decomposition, many such measurements are combined into “batches”. The “core” tensor is obtained from Dynamic Tensor Analysis (DTA). This reduction in computational burden from the new tensor based JPDA method when compared to the traditional JPDA method is demonstrated via a numerical example of a Space Situational Awareness Problem (SSA).

Keywords: Estimation, Robotics, Kalman filtering
Abstract: Sensor data play a significant role in the control of robotic systems. While soft robotics is promising for operation in unstructured environments, it is difficult to sensorize a soft robot because of its softness, and that the inherent softness can be disturbed by the use of sensors. One way to overcome this challenge is to use an observer/filter to estimate the variables (states) that would have been measured by those sensors. Nevertheless, applying an observer/filter scheme to a soft robot introduces challenges due to the high non-linearity in its system model. In this paper, a novel H-infinity based Extended Kalman Filter (EKF) is proposed to estimate the states of a soft continuum manipulator and its performance is compared to that of a conventional EKF. The EKFs are tested on a experimentally validated soft continuum manipulator system with highly nonlinear kinematics and dynamics. The results show that both EKFs achieves accurate estimations in pneumatic muscle actuator (pMA)'s elongation and manipulator's task space coordinates while estimation result for elongation rate is less satisfactory due to large model uncertainties, with the proposed H-infinity EKF performing better than the conventional EKF overall.

Keywords: Linear systems, Identification, Estimation
Abstract: The paper presents a relay feedback experiment for identification and model-based control of a class of time delay single-input single-output (SISO) systems. When a hysteresis relay is fed back to an unknown system, sustained oscillations are yielded at the system output around the setpoint broadly known as limit cycle. Based on limit cycle information and relay settings, a set of state space based explicit expressions is deduced for accurate identification of system dynamics in terms of first order plus time delay (FOPTD) and second order plus time delay (SOPTD) transfer function models. Following the system identification, a set of balanced tuning rules for a proportional-integral (PI) controller is suggested to achieve an enhanced closed loop transient performance. Numerical simulation of well known examples from literature and experimental results from level control system are illustrated for validation of the proposed identification and control schemes.

Keywords: Linear systems, Modeling, Behavioural systems
Abstract: This paper establishes the importance of abstractions and realizations of dynamic networks in characterizing the structure and dynamics of systems. Abstractions and realizations generate dynamically equivalent representations of systems with varying degrees of structural detail. We show that dynamic networks exist that contain the same level of detail as state space models, that other dynamic networks exist that contain the same level of detail as transfer functions, and that still other dynamic networks exist that are simultaneously abstractions of state space models and realizations of transfer functions; thus containing intermediate levels of structural detail.

Keywords: Linear systems, Networked control systems, Network analysis and control
Abstract: In this paper, we consider the structural stabilizability problem of undirected networks. More specifically, we are tasked to infer the stabilizability of an undirected network from its underlying topology, where the undirected networks are modeled as continuous-time linear time-invariant (LTI) systems involving symmetric state matrices. Firstly, we derive a graph-theoretic necessary and sufficient condition for structural stabilizability of undirected networks. Then, we propose a method to determine the maximum dimension of the stabilizable subspace solely based on the network structure. Based on these results, on one hand, we study the optimal actuator-disabling attack problem, i.e., removing a limited number of actuators to minimize the maximum dimension of the stabilizable subspace. We show this problem is NP-hard. On the other hand, we study the optimal recovery problem with respect to the same kind of attacks, i.e., adding a limited number of new actuators such that the maximum dimension of the stabilizable subspace is maximized. We prove the optimal recovery problem is also NP-hard, and we develop a (1-1/e) approximation algorithm to this problem.

Keywords: Linear systems, Networked control systems, Optimal control
Abstract: In the present paper we consider the case of fading channels with bandwidth limitation and noise coloring. We propose for the analysis an equivalent additive noise process that preserve the effect of the fading channel gain at the fading channel output in its first and second moment. We proceed with the bandwidth limitation case first and quantify the channel SNR limitation for the proposed equivalent setting, and then extend the SNR limitation result to the coloring of the channel additive noise case. We treat both cases,fading channel bandwidth limitation and channel additive noise coloring, in a regulation setting (that is with zero reference). In future research we plan to extend the scenarios studied here to account for the effect of reference and disturbance processes, either at the plant input or plant output, depending on the fading channel model location (either over the direct or feedback paths).

Keywords: Linear systems, Optimal control, LMIs
Abstract: This paper presents static and dynamic parallel feedforward controller synthesis methods that render a linear time-invariant system strictly positive real (SPR) in an H-infinity-optimal fashion. The parallel feedforward controller is designed in such a manner that when the output of the system is added to the output of the parallel feedforward controller, the transfer matrix from the system input to the new output is SPR. In order to ensure that the difference between the new output and the original system output is small, the maximum singular value of a static parallel feedforward controller or the weighted H-infinity norm of a dynamic parallel feedforward controller is minimized. The proposed synthesis methods are convex optimization problems that make use of linear matrix inequality and equality constraints. The controllers are implemented numerically on a flexible-joint robotic manipulator and compared to a parallel feedforward controller from the literature. It is shown in closed-loop simulation that a significant improvement in tracking error is achieved with one of the proposed dynamic parallel feedforward controller synthesis methods.

Keywords: Linear systems, Stability of nonlinear systems
Abstract: Conditions to determine if a linear system is externally positive are introduced. The conditions are posed in the time domain and use the properties associated with the weak majorisation of vectors to guarantee the non-negativity of the impulse response.