Lunch Sessions

1. Control for climate change: empower a billion lives

Abstract: Empower a Billion Lives II is a competition to help communities in energy poverty by developing economical and scalable solutions for energy access based on nanogrids, solar home systems, improved business models, and so on. More than 200 teams participated from all over the world. IEEE CSS was one of the main sponsors, with the aim to promote solutions enabled by automation. At this lunch session, members of two winning teams will present their developments.

2. Sanjoy Mitter memorial session

Abstract: Join us for a heartfelt gathering at the CDC as we pay tribute to the cherished memory of Sanjoy Mitter, a revered mentor, a leader, and influential figure in our community, who left us this year. Several esteemed speakers, intimately familiar with Sanjoy's remarkable journey, will be sharing poignant 4-5-minute anecdotes highlighting his profound impact on our field. Time permitting, there will be an opportunity for brief remarks from attendees as well.

3. Can GPT design your controller?

Abstract: Can GPT design your controller and prove that it is safe? Can it create homework problems for your control class? Can we control GPT? What are the implications tools like GPT can have for control research and education? In this blue sky session, we will have a panel discussion with the community about the opportunities, challenges, but also the implications of foundation models (like GPT) in control systems research and education. The panel will include Magnus Egerstedt, Jonathan How, and Dorsa Sadigh.

4. MathWorks Special Session: Combining Learning and Control in Cyber-Physical Systems

Abstract: In deriving optimal control strategies for Cyber-physical systems (CPS), we typically assume an ideal model. Such model-based control approaches cannot effectively facilitate optimal solutions due to the discrepancy between the model and the actual CPS. On the other hand, in most CPS there is a large volume of data which is added to the system gradually in real time and not altogether in advance. Thus, traditional supervised learning approaches cannot always facilitate robust solutions using only data derived offline. By contrast, applying reinforcement learning approaches directly to the actual CPS might impose significant implications on the safety and robust operation of the system. In this talk, I will discuss the challenges of supervised learning and model-based control approaches in several real-world CPS applications that have been successfully commercialized using Matlab (e.g., Math and Optimization, Compiler, Database toolbox) and Simulink (real-time simulation and testing, code generation). Then, I’ll provide a theoretical framework founded at the intersection of control theory and learning that circumvents these challenges in deriving optimal strategies for CPS.

5. Future of work and workers: potential roles for automation and control

Abstract: Drawing on historical lessons from the industrial revolution, mechanization of agriculture, mass manufacturing, and other similar large-scale changes, there is an ongoing debate about the future of work and workers as ever more powerful automation, machine learning, and artificial intelligence technologies are developed and deployed commercially. This is deeply intertwined with demographic changes. It will be a major driver of our socio-economic future. In this talk, I will provide background information on these issues and likely future scenarios. Next, I will present some thoughts and ideas on the possible roles for the systems and control community in shaping the future through research and education. I will conclude with some open-ended questions. This will set the stage for the discussion session to follow.