Keywords: Automotive applications
Abstract: Connected and Automated Vehicles (CAV) have been heralded as a transformative technology, leading to the new era of transportation with unprecedented safety and mobility benefits. They also push the energy efficiency of the transportation systems at both the macro (traffic flow) and micro (vehicle) levels to the next height with abundant new opportunities for communication and optimization. While advanced sensors and hardware, such as camera, radar, and lidar and those used in V2V and V2I communications, have been featured predominantly in CAV showcases, control again is playing the role of the “unsung” hero that enables the CAV technology in the “hidden” world with algorithms and computational intelligence.

We will discuss some fundamental technical challenges for prediction, estimation, and control at the core of the CAV technology in this talk. Using the integrated power and thermal management for CAV as an example, we will show how model-based design, complemented by data-driven approaches, can lead to control and optimization solutions with a significant impact on energy efficiency and operational reliability, in addition to safety and accessibility. Several unique problem characteristics, such as multi-timescale, the highly interactive nature of subsystems involved, and the dynamic and uncertain environment that CAVs are operating within, will be illuminated. Those features call for innovative use of existing tools and the development of new solutions and tools for prediction, estimation, and control. Finally, open challenges will be highlighted to stimulate interest from our community to sing our song loudly and collectively as the control community contributes to the connected and automated transportation systems' safety, efficiency, and accessibility.

Keywords: Automotive applications, Adaptive control
Abstract: Vehicle platooning has attracted much attention due to the significant fuel-saving potential due to the reduced aerodynamic drag force. The Battery Electric Vehicle (BEV) with One Peal Driving (OPD) feature is more suitable for platooning than traditional gasoline vehicles because of the regenerative braking. However, vehicle platooning also presents significant uncertainties in the aerodynamic drag coefficient, which is critical in vehicle spacing control and energy saving but not measurable. In this study, an adaptive controller was proposed to control the inter-vehicle space in platooning application and to estimate the aerodynamic drag coefficient simultaneously by using the embedded adaptation law. Then it was simulated on a simplified BEV with OPD model. The simulation results have demonstrated high performance of the proposed adaptive controller in spacing control, in presence of high uncertainty in drag coefficient. In addition, the real-world field tests showed promising results on spacing control.

Keywords: Automotive applications, Computational methods, Optimization
Abstract: In this paper, an eco-driving Pontryagin maximum principle (PMP) algorithm is designed for optimal deceleration and gear shifting in trucks based on switching among a finite set of driving modes. The PMP algorithm is implemented and assessed in the IPG TruckMaker traffic simulator as an eco-driving assistance system (EDAS). The developed EDAS strategy reduces fuel consumption with an optimized velocity profile and, in practice, allows contextual feedback incorporation from the driver for safety. Furthermore, the optimization over driving modes is computationally inexpensive, allowing the methodology to be used online, in real-time. Simulation results show that significant fuel savings can be achieved proportional to the number of velocity events and the difference between current velocity and final desired velocity for each event.

Keywords: Automotive applications, Control applications, Mechatronic systems
Abstract: Semi-Active suspensions can modulate the damping coefficient of the vehicle in real-time, outperforming passive systems in stability and comfort. In literature, many contributions investigate the damping of the vertical dynamics whereas the pitch motion is often disregarded. In this article, the authors present two strategies, open-loop and closed-loop, to directly tackle the pitch dynamics. Both strategies are validated experimentally on an actual vehicle in a test road which particularly excite the pitch resonance of the chassis, demonstrating how the vehicle balance, and thus the riding performance, can be drastically improved.

Keywords: Automotive applications, Distributed control, Mobile Robots
Abstract: This paper addresses the safety of two approaches to the collision avoidance of vehicle platoons with adaptive cruise control. It has been shown that L2 string stability, which is often used in the literature to achieve safety in the platoon, is not sufficient for collision avoidance. A stricter local condition, which is external positivity of the controlled vehicles, will be applied to guarantee collision avoidance by preventing overshooting responses of the vehicles. The theoretical results are verified by experiments with a set of mobile robots.

Keywords: Automotive applications, Estimation, Kalman filtering
Abstract: Accurate vehicle attitude estimation requires bias-calibrated sensors. The low-cost sensors typically found in industrial vehicles exhibit large time-varying biases, requiring periodic re-calibration. Thus far, the sensor needs to be removed from the vehicle to perform calibration, which is both costly and time-consuming.

In this work, we propose a ground vehicle attitude estimator based on the additive quaternion extended Kalman filter (EKF) with a simple in-vehicle accelerometer bias calibration procedure that can be performed by the end-user. Using a simple model based on wheel speeds and angular rates, accelerations of the vehicle can be estimated and compensated which increases estimation accuracy. The approach is validated through measurements on a real-world industrial vehicle and its performance compared to the state-of-the-art approaches with level-ground calibration and acceleration compensation.

Keywords: Automotive applications, Iterative learning control, Mechatronic systems
Abstract: In this article an iterative learning control (ILC) technique is proposed for the kiss point adaption in a dual-clutch transmission. First, the general ILC problem is introduced which is then specified as model-free phase-lead ILC. A variation of all parameters of this type of ILC is carried out in experiments with a clutch operated on a transmission test bench in order to examine performance and stability. With this, a suitable parametrization of the ILC is found that is used in combination with a Butterworth filter to improve the tracking of a prescribed clutch pressure trajectory. The kiss point adaption is done with a two-step procedure where an inner loop involves the ILC and an outer loop is used to increase the reference stepwise. A kiss point detection condition is proposed which is repeatedly checked during the inner loop. Finally, we demonstrate the performance of the adaption routine by means of experiments.

Keywords: Control applications, Automotive applications, Predictive control
Abstract: Existing Energy Harvesting Shock Absorbers (EHSAs) of vehicle suspensions are mainly designed based on the principle of linear resonance, thereby compromising suspension performance for high-efficiency energy harvesting and being only responsive to narrow-bandwidth vibrations. In this paper, we propose a new EHSA design -- inerter pendulum vibration absorber (IPVA) -- that integrates an electromagnetic rotary EHSA with a nonlinear pendulum vibration absorber. We show that this design simultaneously improves ride comfort and energy harvesting efficiency by virtue of the nonlinear effects of pendulum's inertia. To further improve the performance, model predictive control (MPC) is designed and evaluated in two cases. In the first case, we directly exploit the nonlinear dynamics of the proposed EHSA into a nonlinear MPC (NMPC) design. In the second case, we develop a novel stochastic linearization MPC (SL-MPC) in which we employ stochastic linearization to approximate the nonlinear dynamics of EHSA with superior accuracy compared to standard linearization. This leads to an MPC problem with bilinear dynamics, which is much more computationally efficient than the nonlinear MPC counterpart with no major performance degradation. Extensive simulations are performed to show the superiority of the proposed new nonlinear EHSA and to demonstrate the efficacy of the proposed SL-MPC.

Keywords: Control applications, Renewable Energy
Abstract: Gain-scheduled PI (GSPI) control has been widely acknowledged and used for a blade pitch controller for wind turbines. In this paper, we design an add-on H2 preview compensator for floating wind turbines with the GSPI-based blade pitch controller. The preview information of the incoming wind speed is assumed to be available by LIDAR and the preview compensator is added to the existing GSPI controller. Simulation results show that the preview compensator is effective in reducing the fatigue loads of the blades and tower as well as fluctuations of the generator speed and platform motions.

Keywords: Energy Storage, Observers, Distributed parameter systems
Abstract: All-solid-state batteries are one of the most promising candidates for next-generation energy storage devices capable of delivering high specific energy. Significant effort has been spent on understanding the degradation mechanisms associated with dendrite formation, while energy management and model-based estimation/control for solid-state batteries has received very limited attention. This paper examines a partial differential equation (PDE) state estimation scheme for a one-dimensional electrochemical all-solid-state battery model, using voltage and current measurements only. The state estimation framework exploits the active disturbance rejection control and PDE backstepping techniques, and we rigorously prove estimation error system stability. Electrochemical model-based estimator based on PDE models identifies physical variables for all-solid-state batteries, thus enables high-fidelity monitoring and optimal control in future battery management systems.

Keywords: Energy Storage, Reinforcement learning, Predictive control
Abstract: In this paper, we present the use of Model Predictive Control (MPC) based on Reinforcement Learning (RL) to find the optimal policy for a multi-agent battery storage system. A time-varying prediction of the power price and production-demand uncertainty are considered. We focus on optimizing an economic objective cost while avoiding very low or very high state of charge, which can damage the battery. We consider the bounded power provided by the main grid and the constraints on the power input and state of each agent. A parametrized MPC-scheme is used as a function approximator for the deterministic policy gradient method and RL optimizes the closed-loop performance by updating the parameters. Simulation results demonstrate that the proposed method is able to tackle the constraints and deliver the optimal policy.

Keywords: Energy Systems, Control applications, Optimization
Abstract: Tokamaks, which are one of the most promising approaches to energy generation from nuclear fusion, are toroidal devices confining a very hot ionized gas, i.e. plasma, where the nuclear reactions take place. Studies have shown that the shape of the safety-factor profile, which is related to the helical pitch of the magnetic fields used for plasma confinement, is a key factor towards achieving advanced operating conditions characterized by improved confinement, magnetohydrodynamic stability, and possible steady-state operation. In this work, a first-principles-driven, control-oriented model of the safety-factor profile evolution has been used to design linear-quadratic-integral (LQI) controllers for q-profile shaping in combination, in some cases, with plasma-energy regulation. Results based on nonlinear simulations are presented together with some initial experimental results from the EAST tokamak. A general framework for real-time control of both magnetic and kinetic plasma profiles and scalars has been implemented in the EAST Plasma Control System (PCS), enabling in this way the experimental testing of the proposed controllers. These experiments are among the first experiments on safety-factor profile control ever conducted on the EAST tokamak.

Keywords: Energy Systems, Stochastic/uncertain systems, Predictive control
Abstract: This paper presents two wind energy forecast methods for control of an airborne wind energy system (AWE). The primary objective is to maximize the energy production of the AWE system under a spatio-temporally varying environment with uncertainty in the future wind speed. The controller for the AWE system is formulated as a model predictive controller (MPC). We employ data-driven models to generate probabilistic forecasts of the wind using vector auto-regression (VAR) and time-of-day forecasts. Bayesian optimization is employed to find the optimum of an unknown and expensive to evaluate function. Specifically, the objective function is modelled via wind speed forecasts and then Bayesian optimization optimizes the altitude trajectory while balancing exploitation and exploration of the available altitudes. The performance of the AWE system under the VAR forecast model significantly improves energy production by incorporating wind speed correlations for nearby altitudes.

Keywords: Actuators, Optimization, Autonomous systems
Abstract: To enable autonomous vehicles to operate in cluttered and unpredictable environments with numerous obstacles, such vehicles need a collision avoidance system that can react to and handle sudden changes in the environment. In this paper, we propose an optimization-based reactive collision avoidance system that uses control barrier functions integrated into the control allocation. We demonstrate the effectiveness of our method through numerical simulations with autonomous surface vehicles. The simulated vehicles track their reference waypoints while maintaining safe distances. The proposed method can be readily implemented on vehicles that already use an optimization-based control allocation method.

Keywords: Cooperative control, Mobile Robots, Aerial robotics
Abstract: This paper presents a leader--follower navigation method for a group of unmanned aerial vehicles (UAVs) in 3D obstacle environments based on our previous method for ground robots in 2D environments. An advantage of our method is that the group's formation shape can be changed in a decentralized way so as to prevent the robots from getting stuck in narrow spaces, while preserving sensing network connectivity without data transmission between robots through a wireless communication network. Another advantage is that inequality constraints on control inputs can be explicitly considered. The effectiveness of the proposed method is evaluated in simulations and experiments using quadrotors.

Keywords: Distributed control, Linear systems, Modeling
Abstract: This paper studies the problem of decentralized continuum deformation coordination of multi-agent systems aided by cooperative localization. We treat agents as particles inside a triangular continuum (deformable body) in a 2-D motion space and let the continuum deformation coordination be defined by three leaders located at vertices of a triangle, called the leading triangle. The leaders’ desired trajectories are assigned as the solution of a constrained optimal control problem such that safety requirements are satisfied in the presence of disturbance and measurement noise. Followers distributed inside the leading triangle acquire continuum deformation in a decentralized fashion by integrating cooperative localization and local communication. Specifically, cooperative localization estimates the global positions of all agents using relative position measurements based primarily on proximity of agents. Simulation results are presented for a network of ten agents.

Keywords: Intelligent systems, Neural networks
Abstract: In this paper, a decentralized optimal tracking control problem has been studied for a large-scale multi-agent system (MAS) with heterogeneous system dynamics. Due to the agent number of large-scale MAS, the notorious “curse of dimensionality” problem has challenged the traditional MAS algorithms for decades. The emerging mean field game (MFG) theory has recently been widely adopted to generate a decentralized control method that tackles those challenges by encoding the large-scale multi-agent systems’ information into a Probability Distribution Function (PDF). However, the traditional MFG methods assume all agents are homogeneous, which is unrealistic in practical industrial applications, e.g., IoTs, etc. Therefore, a novel mean field Stackelberg game (MFSG) is formulated based on the Stackelberg game, where all the agents have been classified as two different categories where one major leader’s decision dominates the other minor agents. Moreover, a hierarchical structure that treats all minor agentsasameanfieldgroupisdevelopedtotacklehomogeneous agents’ assumptions. Then, the actor-actor-critic-critic-mass (A2C2M) algorithm with five neural networks is designed to learn the optimal policies by solving the MFSG. The Lyapunov theory is utilized to prove the convergence of A2C2M neural networks and the closed-loop system’s stability. Finally, series of numerical simulations are conducted to demonstrate the effectiveness of the developed method.

Keywords: Planning, Cooperative control, Autonomous systems
Abstract: Decentralized planning for multi-agent systems, such as fleets of robots in a search-and-rescue operation, is often constrained by limitations on how agents can communicate with each other. One such limitation is the case when agents can communicate with each other only when they are in line-of-sight (LOS). Developing decentralized planning methods that guarantee safety is difficult in this case, as agents that are occluded from each other might not be able to communicate until it's too late to avoid a safety violation. In this paper, we develop a decentralized planning method that explicitly avoids situations where lack of visibility of other agents would lead to an unsafe situation. Building on an existing Rapidly-exploring Random Tree (RRT)-based approach, our method guarantees safety at each iteration. Simulation studies show the effectiveness of our method and compare the degradation in performance with respect to a clairvoyant decentralized planning algorithm where agents can communicate despite not being in LOS of each other.

Keywords: Reinforcement learning, Autonomous systems
Abstract: In this paper, the optimal tracking control for large-scale multi-agent systems (MAS) under constraints has been investigated. The Mean Field Game (MFG) theory is an emerging technique to solve the “curse of dimensionality” problem in large-scale multi-agent decision-making problems. Specifically, the MFG theory can calculate the optimal strategy based on one unified fix-dimension probability density function (PDF) instead of the high-dimensional large-scale MAS information collected from all the individual agents. However, the MFGtheoryhasstringentlimitationsbyassumingalltheagents operate in a predefined unlimited space, which is often too ideal for practical applications due to complex environments. In this paper, the original MFG theory has been extended by considering two practical state constraints caused by the environment, i.e., boundary and density constraints. Moreover, to solve the extended MFG type control online, the actor-critic reinforcement learning mechanism is utilized and further extended to a novel actor-critic-mass (ACM) algorithm. Finally, a series of numerical simulations are conducted to demonstrate the effectiveness of the developed schemes.

Keywords: Cooperative control, Adaptive control, Control architectures
Abstract: An important practical problem in the distributed control of multiagent systems is the ability of the closed-loop system to guarantee stability and performance with respect to uncertainties. While there are a wide array of distributed adaptive control architectures that address this problem, they are developed based on a specific Laplacian matrix that has the nullspace generally spanning the vector of ones. The contribution of this paper is to make the first attempt in showing how to design and analyze distributed adaptive control architectures for uncertain multiagent systems with user-assigned Laplacian matrix nullspaces spanning any real vector. For this generalized class of multiagent systems, we first propose a distributed adaptive control architecture to guarantee the closed-loop system stability in the presence of uncertainties. We then utilize the low-frequency learning method in order to address high-frequency oscillations that can result from the fast performance recovery need that requires high-gain learning rates. Illustrative numerical examples are further provided to demonstrate the efficacy of the proposed distributed adaptive control architectures.

Keywords: Adaptive control, Autonomous systems, Cooperative control
Abstract: This paper addresses the problem of autonomously choosing navigation actions while searching for targets in littoral regions. The search amounts to performing detection and classification of measurements as they are collected by a moving sensor. Spatial grids are used to capture the detection and classification states, respectively, of the littoral region, and are used to indicate if objects exist and if so, indicate their class. The navigation actions are chosen by maximizing a recently proposed information-theoretic cost function that incorporates knowledge of the current detection and classification states captured in the spatial grids. Our prior work proposed a framework for performing this type of search with a single vehicle. Here, we propose a modified cost function appropriate for multiple cooperative vehicles capable of sharing information, and characterize the consequences as more vehicles are incorporated into the search. We examine the diminishing returns in performance as we increase the number of vehicles searching a fixed-size area.

Keywords: Distributed control, Autonomous systems, Sensor fusion
Abstract: This paper considers a multiagent, connected, robotic fleet where the primary functionality of the agents is sensing. A distributed multi-sensor control strategy maximizes the value of the collective sensing capability of the fleet, using an information-driven approach. Each agent individually performs sensor processing (Kalman Filtering and Joint Probabilistic Data Association) to identify trajectories (and associated distributions). Using communication with neighbors, the agent enhances the prediction of the trajectories by a Consensus of Information approach that iteratively calculates the Kullback-Leibler average of trajectory distributions, which enables the calculation of the collective information for the fleet. The dynamics of the agents, the evolution of the identified trajectories for each agent, and the dynamics of individual observed objects are captured as a Partially Observable Markov Decision Process (POMDP). Using this POMDP and applying rollout with receding horizon control, an optimized non-myopic control policy that maximizes the collective fleet information value is synthesized. Simulations are presented for a scenario with three heterogeneous UAVs performing coordinated target tracking that illustrate the proposed methodology and compare the centralized approach with a contemporary sequential multiagent distributed decision technique.

Keywords: Cooperative control, Mobile Robots, Robotics applications
Abstract: This paper assesses coordination strategies for heterogeneous robots teams exploring complex three-dimensional communication-limited environments. A multi-agent coordination framework is presented that enables exploration of subterranean environments by teams of aerial and ground robots, designed for and deployed in the DARPA Subterranean Challenge. Various trade-offs in team strategies are investigated including deployment order, sharing map information, and marsupial operations.

Keywords: Robotics applications, Learning, Machine learning
Abstract: Heterogeneous multi-robot sensing systems are able to characterize physical processes more comprehensively than homogeneous systems. Access to multiple modalities of sensory data allow such systems to fuse information between complementary sources and learn richer representations of a phenomenon of interest. Often, these data are correlated but vary in fidelity, i.e., accuracy (bias) and precision (noise). Low-fidelity data may be more plentiful, while high-fidelity data may be more trustworthy. In this paper, we address the problem of multi-robot online estimation and coverage control by combining low- and high-fidelity data to learn and cover a sensory function of interest. We propose two algorithms for this task of heterogeneous learning and coverage---namely Stochastic Sequencing of Multi-fidelity Learning and Coverage (SMLC) and Deterministic Sequencing of Multi-fidelity Learning and Coverage (DMLC)---and prove that they converge asymptotically. In addition, we demonstrate the empirical efficacy of SMLC and DMLC through numerical simulations.

Keywords: Biosystems, Health and medicine, Control applications
Abstract: A susceptible, infectious, removed (SIR) model for the spread of directly transmitted disease caused by pathogens such as bacteria, viruses, and fungi is considered. The nonlinear state equations are feedback linearizable resulting in second order dynamics that can be controlled to achieve constant setpoint tracking. Although the model’s transmission rate is not a control input in the traditional sense, feedback control is used to synthesize a ‘gold standard’ to assist institutions in visualizing what could be achieved via timely implementation of public health interventions, economic and other measures which are known to influence the transmission rate and curb the spread. Control goals may be an improvement in state performance, such as a reduction in the population fraction that is removed, or minimization of the peak fraction of the population that is infected, or minimization of the quarantine window weighed against the economic cost on society, or the avoidance altogether of a second peak. The examination of limiting state behaviors gives further insight into the model and what actions to take in a pandemic. Simulations illustrate several scenarios including performance impact from a time delay in the implemented actions.

Keywords: Biosystems, Nonlinear systems, Distributed parameter systems
Abstract: Microorganisms are typically cultivated in chemostat reactors. If considering self-competition, their population is described by a nonlinear hyperbolic first order integro-partial differential equation with integral boundary condition. In order to validate control algorithms for this system class a finite-time simulation model is required. To obtain such an approximated model Galerkin’s method is used. The resulting system is transformed to Byrnes-Isidori normal form to decouple the internal dynamics and the input/output dynamics. In addition to the analysis of the stability of the internal dynamics, a feedforward and a feedback controller is designed. In simulations the performance is shown and the stated theoretic stability properties are validated by means of numerical results.

Keywords: Control applications, Reinforcement learning, Neural networks
Abstract: In this work, we present a networked epidemic model comprising non-identical agents and consider the problem of learning a vaccine and antidote allocation strategy to contain an outbreak. Even though spreading processes are generally described by nonlinear dynamics, most methods for control are typically based on linear approximations of the nonlinear process and assume full knowledge of the propagation model and dynamics. We propose an alternative approach based on deep reinforcement learning. We define an environment to represent a heterogeneous nonlinear model and show that this environment can be used in conjunction with a Deep Q-Network to stabilize the spreading process. We illustrate our approach using real data from an air traffic network.

Keywords: Iterative learning control, Biotechnology, Control applications
Abstract: During the in vitro maturation process of tissue engineered heart valves in a bioreactor, the pressures ahead of and behind the heart valve as well as the transvalvular flow rate strongly influence the tissue development. In this paper, a norm-optimal iterative learning control scheme for the control of these three process input variables within a new bioreactor setup is introduced and evaluated using physiological references of an aortic heart valve. A Model-in-the-Loop (MiL) setup is established for this evaluation. The results indicate that the proposed scheme yields good tracking performance for all three controlled variables: The stationary rootmean-square errors (RMSEs) of the aortic and left ventricular pressures are, respectively, 0.9 mmHg and 4.8 mmHg (pressure range: 0 - 120 mmHg). The RMSE of the transvalvular flow rate equals to 12.3 ml/s (flow rate range: 0 - 500 ml/s). Consequently, the proposed scheme is suitable for the introduced task and can therefore be investigated further.

Keywords: Estimation, Stochastic/uncertain systems, Health and medicine
Abstract: A stochastic jump diffusion model for type 2 diabetes (T2D) patients is proposed to account for unknown meals during treatment. The model offers the chance to estimate parameters describing how often does the patient consume carbohydrates and how much is consumed. In addition, a strategy based on a Particle Markov chain Monte Carlo (PMCMC) method combined with parameter learning is proposed to estimate the stochastic parameters with continues glucose monitoring (CGM) data and injected insulin amounts only. The strategy was tested both for clinical and simulated data and was shown to be able to estimate all the stochastic parameters with various degrees of accuracy.

Keywords: Estimation, Modeling, Complex networks
Abstract: Models of spreading processes over nontrivial networks are commonly motivated by modeling and analysis of biological networks, computer networks, and human contact networks. However, learning the spread parameters of such models has not yet been explored in detail, and the models have not been validated by real data. In this paper, we present several different spread models from the literature and explore their relationships to each other; for one of these processes, we present a sufficient condition for asymptotic stability of the healthy equilibrium, show that the condition is necessary and sufficient for uniqueness of the healthy equilibrium, and present necessary and sufficient conditions for estimating the spread parameters. Finally, we employ two real data sets, one from John Snow's seminal work on cholera epidemics in London in the 1850s and the other one from the United States Department of Agriculture, to validate an approximation of a well-studied network-dependent susceptible-infected-susceptible model.

Keywords: Optimization, Distributed control
Abstract: Model predictive control (MPC) is a frequently used control technique. An extension of MPC is distributed MPC that can be used to meet restrictions in computation time and make flexible system reconfiguration possible. This contribution presents a DMPC algorithm, which uses sensitivities that contain information about how the control action of a subsystem affects the neighboring agents. Three different ways of calculation are presented. The algorithm itself performs local optimization and exchanges sensitivities on agent level until a convergence criterion is met. The method is applied to several examples to demonstrate its performance, including trajectories and time analysis. In particular, it is shown that the computation time on agent level can be kept almost constant for an increasing system size.

Keywords: Optimization, Nonlinear systems, Discrete event systems
Abstract: Finite time stability involves dynamical systems whose trajectories converge to an equilibrium state in finite time. Sufficient conditions for finite time stability have recently been developed in the literature for discrete-time dynamical systems. In this paper, we build on these results to develop a framework for addressing the problem of optimal nonlinear analysis and feedback control for finite time stability and finite time stabilization for nonlinear discrete-time controlled dynamical systems. Finite time stability of the closed-loop nonlinear system is guaranteed by means of a Lyapunov function that satisfies a difference inequality involving fractional powers and a minimum operator. This Lyapunov function can clearly be seen to be the solution to a difference equation that corresponds to a steady-state form of the Bellman equation, and hence, guaranteeing both finite time stability and optimality.

Keywords: Optimization, Process control, Machine learning
Abstract: In order to overcome plant-model mismatch in static real-time optimization (RTO) of uncertain processes, modifier adaptation (MA) schemes have gained much relevance recently. It applies first-order corrections to the model cost and constraint functions in order to reach plant optimality upon convergence. However, calculating the corrections relies on gradient information, which is challenging to obtain. A promising approach to overcome this limitation is to build Gaussian processes (GP) regression functions on steady-state data to represent the plant-model mismatch. The present paper investigates how the initial operating points and optimizer choice affect RTO performance under MA with GP regression. An experiment is designed to evaluate the system's sensitivity and convergence when initialized with random feasible operating points. Results are compared for a deterministic and evolutionary heuristic to solve the model-based optimization sub-problem: Sequential Quadratic Programming (SQP) and Differential Evolution (DE), respectively. For a semi-batch reactor system case study, we illustrate that SQP can fail to find the global optimum in RTO iterations. As a result, the system's convergence is degraded and becomes sensitive to the initialization phase. On the other hand, DE achieves a consistent convergence profile, thus being indifferent to the initial data points. For a 95% confidence interval, the results show that DE outperforms SQP for this case study.

Keywords: Optimization, Process control, Switched systems
Abstract: We address the problem of optimal operation of an induced draft cooling tower (IDCT) which can be operated in different modes (bypass, showering, ventilation) to meet the heat rejection requirement. Typically, the control design strategies focus on modulating the cooling tower fan speed to regulate the return water temperature without explicitly taking into account different operational modes. For a large scale industrial IDCT, the cooling and lubrication requirements for the mechanical assembly impede slow fan speeds, resulting in mode selection to become pivotal in the optimal operation of the IDCT. In this paper, we propose a control strategy which can account for different operational modes to ensure the optimal operation of the IDCT. A continuous time switched system representation is adopted to capture different operational modes, which is then used to formulate an optimal control problem (OCP) based on the objective of regulating the return water temperature while respecting the physical and operational constraints associated with different operational modes. The OCP is cast as a mixed-integer program (MIP) to simultaneously handle the mode selection and the optimal fan speed required to meet the heat rejection requirement. The MIP is solved in a receding horizon fashion providing robustness against disturbances and model mismatch. The efficacy of the proposed control strategy is demonstrated on an experimentally validated model of the IDCT.

Keywords: Optimization, Reinforcement learning
Abstract: The paper studies the landscape of the optimization problem behind the Model-Agnostic Meta-Learning (MAML) algorithm. The goal of the study is to determine the global convergence of MAML on sequential decision-making tasks possessing a common structure. We investigate in what scenarios the benign optimization landscape of the underlying tasks results in a benign landscape of the corresponding MAML objective. For illustration, we analyze the landscape of the MAML objective on LQR tasks to determine what types of similarities in their structures enable the algorithm to converge to the globally optimal solution.

Keywords: Planning, Optimization, Ships and offshore vessels
Abstract: To be applicable to real world scenarios trajectory planning schemes for mobile autonomous systems must be able to efficiently deal with obstacles in the area of operation. In the context of optimization based trajectory planning and control a number of different approaches to formulate collision avoidance constraints can be found in the literature. This work presents a novel formulation building on constructive solid geometry (CSG) to describe collision avoidance constraints. It is highly efficient due to a very low number of nonlinear inequality constraints required for a given number of obstacles and sample points and in contrast to the original CSG formulation allows to consider the controlled system's shape. To allow for a comparison, popular methods to represent obstacles from the literature are summarized and characterized, namely the simple ellipsoidal representation, the original CSG method as well as a direct and an indirect implementation of a signed distance based approach. A benchmark example shows the good performance of the proposed formulation. Here, optimal trajectory planning for marine surface vessels formulated as a nonlinear programming problem is used, where the scenario is designed based on the maritime test field in Kiel, Germany.

Keywords: Energy Storage, Verification and validation
Abstract: A challenge with Lithium-ion battery (LIB) cells is to study the impact of degradation parameter variations on the model outputs. These parameters not only contribute to battery aging, but also their accurate identification is crucial to enhance battery management systems design. This paper employs a global sensitivity analysis technique to analyze the impact of kinetic, design, and solid-electrolyte interphase (SEI) aging parameters on two different outputs, i.e., cell voltage and charge capacity. The cell is modeled via a coupled nonlinear partial and ordinary differential equations, and differential algebraic equations representing the electrochemical, thermal, and aging dynamics of a LIB cell via the enhanced single particle model (ESPM). To perform the analysis, we adopt different optimal currents at three ambient temperatures to achieve fast chargingminimum degradation profiles. The analysis shows that anode active phase volume fraction, anode reaction rate constant, and solvent reduction kinetic constant are the most sensitive parameters for both outputs.

Keywords: Machine learning, Data analytics, Complex systems
Abstract: Particle accelerators are complex systems composed of multiple subsystems that must work together to produce high quality beams employed for physics experiments. A fault or an anomalous behaviour in one of such subsystems can lead to expensive downtime for the whole facility. Thus, it is of paramount importance to be able to promptly detect anomalies. Given the vast amount of streaming data generated by accelerator field sensors, Machine Learning (ML)-based tools are promising candidates for efficient monitoring of such systems: an approach based on unsupervised ML techniques exploiting the data from a Radio Frequency tuning system is here proposed. Feature importance is exploited to guide the definition of the optimal windowing for feature extraction. The proposed approach is here validated on real-world data related to the ALPI accelerator at Legnaro National Laboratories in Italy.

Keywords: Stochastic/uncertain systems, Robust control, Randomized algorithms
Abstract: Impedance matching is critical to ensure the maximum power transfer on plasma etching for semiconductor manufacturing. However, it is a challenging task due to the unknown and complex plasma dynamics. In this paper, a Stochastic Perturb and Observe Fractional-Order Extremum Seeking Controller (P&O FO-SESC) is employed for plasma impedance matching. The controller uses a Fractional-Order Gaussian (dithering) Noise (fGn) as the perturbation signal and is tested for an L-type matching network with two variable capacitors. Obtained results show that the the Stochastic P&O FO-ESC controller improves the impedance matching convergence zone over the standard P&O ESC controller for different loads and initial conditions.

Keywords: Automotive applications, Machine learning, Time delays
Abstract: A neural network-based controller is trained on data collected from connected human-driven vehicles in order to steer a connected automated vehicle on multi-lane roads. The obtained controller is evaluated using model-based simulations and its performance is compared to that of a traditional nonlinear feedback controller. The comparison of the control laws obtained by the two different approaches provides information about the naturalistic nonlinearities in human steering, and this can benefit the controller development of automated vehicles. The effects of time delay emerging from vehicle-to-everything (V2X) communication, computation, and actuation are also highlighted.

Keywords: Automotive applications, Sliding mode control
Abstract: Battery electric vehicles (BEVs) with one-pedal driving (OPD) features have demonstrated significant potentials in reducing energy consumption, improving driving comfort, and enhancing driving safety. Besides, vehicle platooning can provide dramatic energy-saving benefits for BEVs and extend electric driving ranges. This paper presents a robust sliding mode control (SMC)-based inter-vehicle distance controller for an OPD-enabled BEV to achieve the desired inter-vehicle distance in vehicle platooning applications, in presence of uncertainties such as aerodynamic drag coefficient, road grade, and rolling resistance coefficient. The proposed SMC-based controller was validated during highly dynamic driving cycles in simulation and on the test track. The simulation results and experimental results demonstrated that the proposed sliding mode controller can achieve the desired inter-vehicle distance with high robustness against various uncertainties.

Keywords: Automotive applications, Transportation systems, Planning
Abstract: A new local motion planning method for automated vehicles in structured road scenarios is considered. Utilizing three consecutive clothoids, it is shown that real time generation of feasible trajectories for automated vehicles is possible. Separate path generation and consecutive velocity planning is considered. The path planning is reduced to a small set of nonlinear algebraic equations, while the velocity planning comprised of forward, backward and joint iteration processes. By varying some continuous parameters, multiple feasible plans can be generated and one of those can be chosen based on multiple criteria. A left turn at an intersection is used as an illustrative example.

Keywords: Autonomous systems, Automotive applications, Control applications
Abstract: Advanced Driver Assistance Systems (ADAS) and autonomous driving systems are relevant in the agricultural field, since they can ease personnel of demanding and repetitive tasks while increasing precision and productivity. This is particularly true in constrained environments represented by intensive and high value cultivations, like vineyards and orchards. Anyway, these contexts present numerous challenges: positioning accuracy in the range of centimeters is required in an environment with continuously-changing vegetation, reduced maneuvering space and unstable terrain.

This paper presents an ADAS of level 3 for an agricultural tractor in a vineyard, focusing on its control system. The goal of the developed controller is to bring the vehicle at a desired distance from the crop rows and keep it aligned to them, so that the operator only has to set the tractor advancement speed and can focus on the ongoing agricultural procedures. This is achieved through a Linear Quadratic Integral (LQI) controller that relies on a control-oriented model of the system describing the dynamics of the vehicle position with respect to the vines. The system proves to be effective and easily tunable in order to obtain the desired behavior. An extensive experimental campaign validates the closed-loop system performance. In particular, the controller attains a steady state error of 5 cm, using a steering angle with Root Mean Square (RMS) of 1.05 deg.

Keywords: Control applications, Control Technology, Automotive applications
Abstract: This paper focuses on optimal preview control with uncertain preview information. Accurate preview information is required in optimal preview control. However, dealing with uncertain preview information in optimal preview control is not addressed in the literature. To handle uncertain preview information, a disturbance observer is introduced to possibly recover the optimal preview control performance. The effectiveness of the proposed approach is verified by numerical simulations with an example of an active suspension system.

Keywords: Linear parameter-varying systems, Automotive applications, LMIs
Abstract: This article presents a state feedback control design strategy for the stabilization of a vehicle along a reference collision avoidance maneuver. The stabilization of the vehicle is achieved through a combination of steering, acceleration and braking. A Linear Parameter-Varying (LPV) model is obtained from the linearization of a non-linear model along the reference trajectory. A robust state feedback control law is computed for the LPV model. Finally, simulation results illustrate the stabilization of the vehicle along the reference trajectory.

Keywords: PID control, Control applications, Renewable Energy
Abstract: In this contribution, an individual pitch control system (IPC) for a large wind turbine by using a fractional order nonlinear PI approach (FO-NPI) is implemented and its performance compared with the standard PI controller analyzed.

A solution to the problem of the integrator wind-up in the case of the IPC is proposed, as well. It is considered that pitch actuators saturate not only in magnitude but also in rate. A 20 MW reference wind turbine, which is implemented in the simulation software OpenFAST, is used as virtual plant for the simulation experiments. Simulation results indicate that the IPC system based on the FO-NPI techniques improves the wind turbine control performance and reduces damage equivalent loads (DELs) by 7.5 percent on average when compared with classic PI control.

Keywords: Predictive control, Neural networks, Renewable Energy
Abstract: This paper introduces different Model Predictive Control (MPC) strategies aimed at optimizing the energy production of a Pendulum Wave Energy Converter (PeWEC). Due to MPC ability of dealing with system constraints and considering future behaviors in optimal control computation, the first proposed MPC is applied to PeWEC as a classical problem of set-point tracking using a linear model to propagate the system behaviour. Moreover, since the MPC application in wave energy conversion deviates significantly from traditional MPC ones, an economic version is also explored. The objective function of the MPC thus realized directly considers a measure of the absorbed power. However, this formulation, together with the use of a nonlinear model in predicting the system evolution, leads to an optimization problem to be solved that is neither fully quadratic nor can be guaranteed to be convex. This paper shows that this second approach brings better performances, demonstrating that it is potentially more suitable for wave energy applications. On the other side, such approach has a computational drawback from both a real time implementation and offline design perspectives. To avoid the potentially prohibitive computational costs that an online optimization would require, this work introduces a novel control based on a Deep Neural Network (DNN) able to mimic the nonlinear economic MPC. Results arising from simulations applying the proposed strategy demonstrate the effectiveness of the presented approach.

Keywords: Renewable Energy, Predictive control, Reinforcement learning
Abstract: This paper evaluates two strategies, deep reinforcement learning (DRL) and model predictive control (MPC), for maximizing harnessed power from a lifting surface controlled ocean current turbine (OCT) through depth optimization. To address spatiotemporal uncertainties in the ocean current, an online Gaussian Process (GP) is applied, where the prediction error of the ocean current speed is also modeled. We compare the performance of the MPC-based optimization with the DRL-based algorithm (i.e., deep Q-networks (DQN)) using over one week of field collected acoustic doppler current profiler (ADCP) data. The DRL-based algorithm is almost equivalent to the MPC-based algorithm in real-time optimization when the ocean current speed prediction is perfect. However, the performance of the DQN-based algorithm surpasses the MPC-based algorithm when ocean current prediction error is considered. The importance of using the DQN in improving the error-tolerance of the proposed spatiotemporal optimization is verified through the comparative results.

Keywords: Smart grid, Energy Systems, Estimation
Abstract: This paper considers the problem of estimating the dynamic and algebraic states in transmission power networks. Specifically, we leverage a linearized, differential-algebraic equation (DAE) of power networks—also called descriptor system models—consisting generator’s dynamics, stator’s algebraic constraints, generator’s complex power, and the network’s complex power balance equations. The DAE is then combined together with a phasor measurement unit-based measurement model, which assumes that bus voltages and line currents are measured, to produce a semi-explicit linear DAE which is exploited to design a linear DAE observer to estimate the dynamic and algebraic states of power networks. Numerical simulations are performed using the IEEE 9-bus transmission system to analyze the performance of the proposed observer in estimating generators’ internal states and unmeasured bus voltages subject to both Gaussian and non-Gaussian noise.

Keywords: Smart grid, Reinforcement learning, Renewable Energy
Abstract: To tackle the challenge of voltage regulation under high solar photovoltaics (PV) penetration, the slow timescale control of conventional voltage regulating devices can be combined with fast timescale control of smart inverters. In this paper, we develop a two-timescale Volt-VAR control (VVC) framework. The slow timescale control of voltage regulating devices is achieved by a model-based approach. The fast time-scale control of smart inverters is attained with a reinforcement learning-based method. The deep deterministic policy gradient (DDPG) algorithm is adopted to control the setpoints of both real and reactive power of smart inverters. The control policy of smart inverters is learned from the historical operational data without relying on accurate distribution network secondary circuit parameters. Simulation results on the IEEE 34-bus feeder show that the proposed framework can determine near optimal set points of smart inverters in real-time operations. Compared with existing reinforcement learning based smart inverter control, our approach achieves lower line losses, voltage deviations, and active power curtailment.

Keywords: Aerial robotics, Chaotic systems, Nonlinear systems
Abstract: This paper addresses the problem of surveillance of a confined area using an unmanned aerial vehicle with four rotors. The properties of chaotic systems have been exploited to design an unpredictable velocity profile for the quadrotor. To maintain the quadrotor within the area of interest, the mirror mapping approach is used. On the other hand, to guarantee velocity tracking a nonlinear velocity and attitude control laws are proposed. The effectiveness of the proposed surveillance approach is validated by numerical results. The numerical results were obtained by means of Matlab software and its Simulink tool.

Keywords: Aerial robotics, Mechatronic systems, Mechanical systems
Abstract: Conventional multirotors have coupling between position and attitude control due to underactuation in dynamics, and thus can not track six degree-of-freedom (DoF) trajectories in space. Previous works proposed fully actuated multirotors with modifications to mechanical structure to provide varying orientations of thrust forces without changing the attitude, but usually introduced additional actuators and mechanisms so that the complexity for design and construction was increased. This paper proposes a novel multirotor platform with the capability of six DoF control, which can be easily constructed with commercial quadcopters and simple 3D printed connectors, and effortlessly scaled. An easily implemented controller is designed, and the six DoF control is verified through simulation and real world experiments.

Keywords: Aerospace applications, Linear parameter-varying systems, Optimization
Abstract: When a lunar lander touches down, a large impact force may have undesirable effects such as rebounding and overturning. To ensure a safe landing, active momentum exchange impact dampers (AMEIDs) are introduced in this work to enable soft-landing control. For the purpose of designing and evaluating different control strategies applied to the AMEIDs, a two-dimensional lunar lander model is derived using Lagrangian mechanics. The model is validated via a free-fall landing experiment. Using the lander dynamical model, we design a suboptimal control law by solving the discrete-time state-dependent Riccati equation (D-SDRE). The landing response of D-SDRE based controller is compared to that of open-loop control and PID control, respectively. The simulation results show that the D-SDRE method mitigates the landing instability and reduces the side-slip phenomenon.

Keywords: Aerospace applications, Reinforcement learning, Game theory
Abstract: Predicting the outcomes of integrating Unmanned Aerial Systems (UAS) into the National Airspace System (NAS) is a complex problem that is required to be addressed by simulation studies before allowing the routine access of UAS into the NAS. This paper focuses on providing a 3-dimensional (3D) simulation framework developed using a game-theoretical methodology to evaluate integration concepts using scenarios where manned and unmanned air vehicles co-exist. In the proposed method, the human pilot interactive decision-making process is incorporated into airspace models, which can fill the gap in the literature where the pilot behavior is generally assumed to be known a priori. The proposed human pilot behavior is modeled using the dynamic level-k reasoning concept and approximate reinforcement learning. Level-k reasoning concept is a notion in game theory and is based on the assumption that humans have various decision-making levels. In this study, Neural Fitted Q Iteration, which is an approximate reinforcement learning method, is used to model time-extended decisions of pilots with 3D maneuvers. An analysis of UAS integration is conducted using an example 3D scenario in the presence of manned aircraft and fully autonomous UAS equipped with sense and avoid algorithms.

Keywords: Cooperative control, Aerial robotics, Linear robust control
Abstract: The objective of this paper is to model, simulate and analyze platooning (separation) control for a fleet of 6 quadcopter units. Control for 6 degrees of freedom (x,y,z,phi,theta,psi] is modeled for each individual quadcopter using a cascaded linear feedback control system, with the fleet modeled as leader-follower. The primary motivation of this research is to examine string instability arising from the ``accordion effect", a phenomenon observed in leader-follower systems due to which positioning or relative spacing errors arise in follower vehicles due to sudden changes in lead vehicle velocity. First, a PID separation controller is designed for a nominal case, where communication within the system is ad-hoc. Steady state separation/positioning errors for each member of the fleet are observed and documented. Second, lead vehicle acceleration is then provided to each controller (as a feed forward term), which is used to compare controller bandwidth requirements to ensure relative string stability , within acceptable error bounds. Thus the key contribution of this work is a separation controller for a fleet of quadcopters, with quantitative analysis of the string stability, using simulation data from MATLAB Simulink.

Keywords: Robotics applications, Control applications, Cyberphysical systems
Abstract: A novel control and testing platform for a smart morphing wing system is introduced to obtain optimal aerodynamic properties. This paper delves into the manufacturing process for said system, from the computer-aided modeling to the assembly, and its corresponding difficulties. The issues associated with the primary rendition of the model are addressed, as well as proposed solutions to these issues. Additionally, a hardware-in-the-loop formulation is introduced, which utilizes the 3D printed airfoil model for Computational Fluid Dynamics (CFD) analysis and reinforcement learning. In particular, image processing techniques and algorithms are employed to obtain an outline for the various morphed configurations, which are converted into airfoil coordinates and analyzed using a CFD tool, before being imported into a Q-learning algorithm.

Keywords: Autonomous systems, Game theory, Aerospace applications
Abstract: This paper considers a pursuit-evasion problem with two cooperative pursuers and one evader. The problem is posed as a zero-sum differential game where the evader aims at reaching a goal line which is protected by the pursuers. When reaching this goal is not possible, the evader strives to position itself as close as possible with respect to the goal line at the time of capture. The pursuers try to capture the evader as far as possible from the goal line. Leveraging differential game theory, state feedback strategies are both synthesized and verified in this paper. In addition, the Barrier surface that partitions the state space into two winning sets, one for the pursuer team and one for the evader, is obtained in analytical form. Under optimal play, the winning team is determined by evaluating the associated Barrier function.

Keywords: Control applications, Autonomous systems, Variational methods
Abstract: This paper considers a three agent scenario consisting of a pursuer, evader, and a defender. The pursuer's objective is to capture the non-maneuvering evader in minimum time while a defender aims at maximize contact with the pursuer by keeping the pursuer inside his circular engagement zone for as long as possible; the pursuer is considered to be faster than both the evader and the defender. Using optimal control theory, the optimal control for the defender that maximizes contact with the pursuer is posed and solved. In the event that the evader is captured by the pursuer before the pursuer escapes the engagement zone of the defender, some suboptimal strategies of the defender provide equivalent contact time. A derivation of defender's headings that maximize contact is presented along with examples that highlight the importance of the initial conditions of the engagement scenario.

Keywords: Cooperative control, Planning, Autonomous systems
Abstract: We consider a problem of cooperative evasion between a single pursuer and multiple evaders in which the evaders are constrained to move in the positive Y direction. The evaders are slower than the vehicle and can choose their speeds from a bounded interval. The pursuer aims to intercept all evaders in a given sequence by executing a Manhattan pursuit strategy of moving parallel to the X axis, followed by moving parallel to the Y axis. The aim of the evaders is to cooperatively pick their individual speeds so that the total time to intercept all evaders is maximized. We first obtain conditions under which evaders should cooperate in order to maximize the total time to intercept as opposed to each moving greedily to optimize its own intercept time. Then, we propose and analyze an algorithm that assigns evasive strategies to the evaders in two iterations as opposed to performing an exponential search over the choice of evader speeds. We also characterize a fundamental limit on the total time taken by the pursuer to capture all evaders when the number of evaders is large. Finally, we provide numerical comparisons against random sampling heuristics.

Keywords: Game theory, Planning, Mobile Robots
Abstract: With autonomy gaining popularity and acceptance in various fields, humans have begun shifting to employing autonomous vehicles to perform risky tasks in hostile environment. Therefore, it is crucial that we investigate motion planning problem with safety in mind. In this sense, reach-avoid game is a good problem to analyze due to its complex nature and the fact that worst-case disturbance is considered. In this work we present an iterative open-loop formulation for a reach-avoid game with multiple attackers and defenders. We show the proposed method can find more direct paths compared to the basic open-loop formulation, and perhaps more importantly, the algorithm can find feasible paths in situation where open-loop formulation simply cannot. By solving the game with the proposed method we can guarantee the safety of the planned path, despite the dynamic environment with the presence of unpredictable obstacles, adversarial defender, as well as other attackers. In addition, we offer a practical method that ensures agents with minimum turning radius only take paths that respect the constraints. We verify the effectiveness of our method through a series of simulations.

Keywords: Game theory, Autonomous systems, Cooperative control
Abstract: A two-player, Engage or Retreat (EoR) differential game is presented in which an attacker must choose whether to capture one of N static targets or retreat across a defined retreat boundary. Throughout the engagement, a defending player is capable of activating defensive assets located at each of the target locations. These defensive assets inflict a cost on the attacker as a function of distance and allow the defender to present a deterrent in an effort to persuade the attacker to elect retreat. The solution of the game is constructed by examining two related subproblems: the Game of Engagement (GoE) and the Optimal Constrained Retreat (OCR). Each subproblem examines a different combination of attacker termination strategies, capture vs retreat, and defender strategies, resist vs cooperate. When solving for the optimal constrained retreat strategy, a value function constraint is imposed on the retreat trajectory in order to ensure that the attacker does not maneuver into an advantageous position. Several solutions are examined to illustrate the types of behaviors found in the equilibrium game solutions.

Keywords: Game theory, Autonomous systems, Cooperative control
Abstract: This paper examines a variation of the standard one-pursuer, one-evader, pursuit-evasion problem. In this problem, the evader is capable of inflicting a cost on the attacker during the pursuit. The attacker chooses whether to continue the engagement or abandon pursuit and retreat across a defined retreat boundary. The evader strategically manipulates the attacker's cost function in order to encourage retreat. If the pursuer elects to retreat, the defender maneuvers in a way to escape to safety and prevent the pursuer from reengaging. The solution of the game is developed using analytic indirect methods and an illustrative solution is examined.

Keywords: Control applications, Modeling, Estimation
Abstract: Crowd violence and the repression of free speech have become increasingly relevant concerns in recent years. This paper considers a new application of crowd control, namely, keeping the public safe during large scale demonstrations by anticipating the evolution of crowd emotion dynamics through state estimation. The general crowd control problem is difficult for a variety of reasons, including limited access to informative sensing and effective actuation mechanisms, as well as limited understanding of crowd psychology and dynamics. This paper takes a first step towards solving this problem by formulating a crowd state prediction problem in consideration of recent work involving crowd psychology and opinion modeling. We propose a nonlinear crowd behavior model incorporating parameters of agent personality, opinion, and relative position to simulate crowd emotion dynamics. This model is then linearized and used to build a state observer whose effectiveness is then tested on system outputs from both nonlinear and linearized models. We show that knowing the value of the equilibrium point for the full nonlinear system is a necessary condition for convergence of this class of estimators, but otherwise not much information about the crowd is needed to obtain good estimates. In particular, zero-error convergence is possible even when the estimator erroneously uses nominal or average personality parameters in its model for each member of the crowd.

Keywords: Distributed parameter systems, Reduced order modeling, Process control
Abstract: Axial dispersion models are capable of delivering accurate predictions of non-ideal flow behavior, where the residence time distribution (RTD) of the flowing material is estimated. Such models are crucial for analysis, monitoring, and control purposes of processes appearing in a wide spectrum of applications, e.g. chemical, pharmaceutical. In the previous contribution [1] the problem of the variability of the residence time distribution due to its dependency on process variables is tackled using a grey-box approach. However, solving the grey-box model numerically including the varying model parameters requires a fine spatial discretization. In consequence, delivering a high-dimensional model unsuitable for control-oriented applications, e.g. model- based estimation/control, real-time monitoring. This work presents a solution to the problem by employing a projection-based Model Order Reduction (MOR) technique producing a computationally feasible Reduced Order Model (ROM) with acceptable accuracy. The exploitation of the Full Order Model (FOM) special structure along with a carefully selected input vector including the varying parameters allowed casting the model into a weak nonlinear form, i.e. bilinear. Hence permitting the investigation of two methods for parameter independent MOR, i.e. Proper Orthogonal Decomposition (POD) and bilinear mathcal{H}_2 optimal. The latter method requirements are considered and treated within the FOM mathematical formulation. The ROMs produced by the two methods are evaluated against the FOM for selected simulation test scenarios showing the superiority of the latter. That is for a practical test case of a random biased input concentration and varying flow parameters, the ROM could predict the concentration output with a Normalized Mean Square Error (NMSE) lying below 4% for all tested peclet number ranges. A computational speedup factor up to 32 is achieved using the ROM showing its potential for real-time applications even for sampling rates within milliseconds order of magnitude.

[1] A. Elkhashap, et al., “Greybox approach for the prediction of variable residence time distribution in continuous pharmaceutical manufacturing,” IFAC World Congress, 2020

Keywords: Modeling, Hybrid systems, Game theory
Abstract: In the presented work, an existing game-theoretical framework modeling intersection scenarios is extended. The framework is based on a hybrid dynamic game approach, which subdivides the motion of the vehicle into distinct maneuvers to reduce the continuous action space to a finite set of discrete actions. Therefore, the optimization problem modeling the decision process can be solved using dynamic programming. In the first step, the capability to handle situations with players driving along the same path is added. Furthermore, the approach is extended for situations with more than two players. In order to reduce the computational complexity, a method for selecting only relevant traffic participants is developed. Different simulation examples are used to verify the general functionality of the model.

Keywords: Modeling, Manufacturing systems, Nonlinear systems
Abstract: Concrete pumping is a widespread technology used in on site construction around the world. It also is a viable technology for the automized fabrication of concrete elements. This however requires precise control of the outgoing mass flow. Since measurements of the mass flow or the total conveyed mass are impractical, a feedforward control is required. This in turn requires a model of the outgoing mass flow. While the literature on concrete pumping is abundant, automation-relevant aspects, e.g. the initial time delay due to filling of the hose, have not been covered yet. This paper proposes a model to fill that gap, which shows promising results while refraining from treating the complex details of 3D-flow. This is an important aspect for both implementation on control hardware and process planning. Assuming incompressible flow, a hose can be described as a reservoir for the incoming mass flow, with an outflow function depending on the geometry. If the outlet points downward, gravitation imposes an additional dynamic behavior, for which several modeling approaches are proposed and validated by experiments.

Keywords: Modeling, Nonlinear systems, Mechatronic systems
Abstract: Controller based on ionization current measurement are state of the art in the field of wall hung gas boilers (WHB). They are used to control the gas-air ratio of the combustion. The underlying ionization mechanism is a complex three-dimensional problem. Therefore, it is a difficult task to describe this system with a white-box model for simulation and controller design. This paper investigates a new method to model the ionization current using a nonlinear Hammerstein model. The parameters of the system are identified with real measurement data. Thereby, three different approaches of Hammerstein models are trained and compared. One model is a complete black-box model, the other two are grey-box models with known static nonlinearities. These two models have an increased model accuracy in comparison to the unknown model. The results are verified with additional data sets, which are not used for training.

Keywords: Modeling, Stochastic/uncertain systems, Cyberphysical systems
Abstract: Pricing models implemented at electric vehicle (EV) charging facilities provide facility operators a means to achieve desirable system-level behavior. Furthermore, a charging facility must meet resource constraints with high levels of confidence. To achieve this, a charging facility operator can tune the pricing model parameters such that resource constraints are met with high confidence. In this paper, we propose an approximate chance-constrained optimization program that enables charging facility operators to set the pricing model parameters in an anticipatory, rather than a reactionary, manner. We present a problem formulation based on two previously developed pricing models and present results from a numerical case study for setting the respective pricing model parameters.

Keywords: Learning, Predictive control, Automotive applications
Abstract: This paper proposes an approach for the calibration of the cost function of optimization-based controllers. The approach uses a Kalman filter that estimates the cost function parameters using data of closed-loop system operation. It adapts the parameters online and robustly, provides safety guarantees, is computationally efficient, has low data storage requirements, and is easy to implement making it appealing for many real-time applications. The approach provides a data-efficient alternative to Bayesian optimization and an automated alternative to learning from demonstrations. Simulation results show that the approach is able to learn cost function parameters quickly (approximately 95% faster than Bayesian optimization), is able to adapt the parameters to compensate for disturbances (approximately 25% improvement on tracking precision), and is robust to noise.

Keywords: Observers, Linear robust control, Nonlinear systems
Abstract: A frequency domain approach is used to establish an equivalence between Reduced Order Extended State Observer (ROESO) based Active Disturbance Rejection Control (ADRC) and Disturbance Observer Based (DOB) control using nominal (estimated) state feedback. These robust control schemes are traditionally devised to solve automatic control problems for systems with unmeasured states and subject to unknown disturbances. The results are presented in the context of pure integration perturbed systems, which constitute a paradigmatic model of nonlinear, uncertain, SISO differentially flat systems. Application to the control of a non-trivial, nonlinear, mechanical example is presented along with computer simulations.

Keywords: Predictive control, Learning
Abstract: This paper proposes a receding horizon active learning and control problem for dynamical systems in which Gaussian processes (GPs) are utilized to model the system dynamics. The active learning objective in the optimization problem is presented by the exact conditional differential entropy of GP predictions at multiple steps ahead, which is equivalent to the log determinant of the GP posterior covariance matrix. The resulting non-convex and complex optimization problem is solved by the sequential convex programming algorithm that exploits the first-order approximations of non-convex functions. Simulation results of an autonomous car example verify that using the proposed method can significantly improve data quality for model learning.

Keywords: Predictive control, Linear systems
Abstract: In our previous work, we presented a constraint management technique, referred to as the Transient-Robust Reference Governor (TR-RG), to enforce state, output, and control constraints in closed-loop nonlinear systems. One of the key assumptions behind the TR-RG is that the input-output mapping of the nonlinear system is available at steady-state and, furthermore, this mapping is invertible. While this assumption holds in some situations (e.g., engine control in automotive applications), it does not hold in general (e.g., inverted pendulum control problem). To overcome this limitation, this paper extends TR-RG to a wider class of nonlinear systems, namely those with non-invertible or uncertain steady-state characterizations. Theoretical guarantees of constraint enforcement using this scheme are provided. Finally, the scheme is illustrated using two practical examples.

Keywords: Software tools, Predictive control, Simulation
Abstract: We introduce the MATLAB framework PARODIS, the Pareto optimal Model Predictive Control framework for distributed systems. It is a general-purpose, flexible and easy-to-use framework for discrete state space models. Special features are the support of distributed (hierarchical) systems, scenario-based optimization and built-in methods for determination of the Pareto front and selection of a solution. It uses the popular MATLAB framework YALMIP for the symbolic formulation of optimization problems and models.

Keywords: Predictive control, Renewable Energy, Real-time systems
Abstract: Model predictive control (MPC) is a control method that involves determining the input to a dynamical system as the solution to an optimization problem that is solved online. In the wind turbine research literature, MPC has received considerable attention for its ability to handle both actuator constraints and preview disturbance information about the oncoming wind, which can be provided by a lidar scanner. However, while many studies simulate the wind turbine response under MPC, very few physical tests have been carried out, likely due in part to the difficulties associated with solving the MPC problem in real time. In this work, we implement MPC on an experimental, scaled wind turbine operating in a wind tunnel testbed, using an active grid to create reproducible wind sequences and a hot-wire anemometer to generate upstream wind measurements. To our knowledge, this work presents the first physical test of MPC for blade pitch control of a scaled wind turbine. We compare two MPC strategies: one including preview disturbance information and one without. Our results provide further evidence that feedforward control can improve wind turbine performance in transition and above-rated conditions without increasing actuation requirements, which we hope will encourage industry experimentation and uptake of feedforward control methods. We also provide a high-level analysis and interpretation of the computational performance of the chosen approach. This work builds upon the results of an earlier study, which considered unconstrained optimal blade pitch control.

Keywords: Nonlinear robust control, Robust control, Linear parameter-varying systems
Abstract: This paper considers the attitude control problem of a generic rotating 3 degrees of freedom fully actuated rigid object. The specific studied problem is the deviation control of this object around a theoretically feasible attitude trajectory. The rotation motion has an intrinsic non linear behaviour (trigonometric, periodicity) that need to build non linear and hybrid controllers to get global stability of the closed loop system. This paper considers the opportunity to use the quaternion framework to build a continuous non linear state feedback that reaches an almost global asymptotical stability. Some perspectives to enhance this result with integrators to cancel out static and drag errors are eventually proposed.

Keywords: Nonlinear systems, Computational methods, Optimization
Abstract: The ability to quickly synthesize an optimal control signal for a nonlinear system is critical for practical implementations. In previous work, we have introduced a computational procedure to iteratively synthesize an optimal control signal for a very broad class of nonlinear control systems. This paper presents an extension of the approach to allow for different parameterizations of control inputs, resulting in a substantial reduction in the number of decision variables and thus computation time. The highlighted efficiency and effectiveness of the proposed approach are illustrated and compared against other methods using various control examples.

Keywords: Nonlinear systems, Distributed control, Autonomous systems
Abstract: Network systems involve distributed decision-making for coordination of networks of dynamic agents and address a broad area of applications including cooperative control of unmanned air vehicles, microsatellite clusters, mobile robotics, battle space management, and congestion control in communication networks. In this paper, we develop a thermodynamic-based framework for addressing consensus problems for discrete-time nonlinear multiagent dynamical systems with a fixed communication topology. The proposed controller architecture involves the exchange of state information between agents guaranteeing that the closed-loop dynamical network is semistable to an equipartitioned equilibrium representing a state of information consensus consistent with basic thermodynamic principles.

Keywords: Nonlinear systems, Mobile Robots, Control applications
Abstract: This paper proposes kinematic and dynamic tracking of mobile robot using novel fractional order control. In the proposed article two different types of control algorithms are developed. First controller is designed for kinematic model of mobile robot. The designing of kinematic control is based on Lyapunov based fractional order control. Second controller is designed for dynamic model of mobile robot. For dynamic control fractional order backstepping control has been developed. Stability analysis is given using Lyapunov stability theory for both kinematic and dynamic controllers. Simulations are done for different types of generated path and comparative study is given to show the advantage of the proposed method over existing methods.

Keywords: Nonlinear systems, Observers, Adaptive control
Abstract: This study considers the problem of adaptive feedback controller design for output stabilization of dissipative (generalized) Hamiltonian systems with unstructured dynamic. This class of models enable one to exploit the dissipative-conservative structure of generalized Hamiltonian systems for feedback control design while relaxing the burden of deriving an exact structured model representation. Assuming that the overall system is stabilizable and observable, and under mild assumptions on the unstructured part of the dynamics, a stabilizing adaptive control law is designed to stabilize systems to the desired output of the system. To fulfill the design procedure, a full order nonlinear Luenberger observer is designed for unknown states measurements. Stability of the closed-loop system is then demonstrated using Lyapunov stability arguments. A numerical illustration of the proposed approach is presented to demonstrate the potential of the design method.

Keywords: Nonlinear systems, Switched systems, Autonomous systems
Abstract: An integrated relay output feedback (IROF) system is a piecewise linear system representing many electric and/or mechanical systems. Like other piecewise linear systems IROF inherently has limit-cycle oscillation(LCO), which plays a critical role in the performance of the system, entailing the necessity of reliable analysis on the LCO. However, due to the nature of the solution trajectory of LCO, we cannot safely apply existing methods on the analysis of LCO such as describing function, Floquet theory, or piecewise linear system analysis. This research provides a new theoretical approach with which to establish the sufficient condition of an LCO in the IROF system accompanying the analysis on its stability and main parameters. We start from confirming that the switching time between each switching plane is finite. Then, by the introduction of a discrete-time nonlinear system representing the Poincar´e map and the linear phase can we establish the sufficient condition for the existence of an LCO and its stability. Finally, we demonstrate the effectiveness and efficiency of the main results with a simple numerical example.