MechE Colloquium: Towards Hands-Off Automation: Data-Driven Design, Robustness, and Real-Time Control
Event details
| Date | 25.11.2025 |
| Hour | 12:00 › 13:00 |
| Speaker | Prof. Colin Jones, Institute of Mechanical Engineering, School of Engineering, EPFL |
| Location | Online |
| Category | Conferences - Seminars |
| Event Language | English |
Abstract: Model Predictive Control (MPC) has become a cornerstone of modern automation, enabling high-performance decision making in applications ranging from energy-efficient buildings to robotics. Yet widespread deployment remains hindered by three persistent challenges: the need for accurate models, the difficulty of tuning, and the computational demands of real-time optimization.
In this talk, I will present our group's efforts to move towards "hands-off" automation — designing and deploying controllers that require minimal engineering intervention. I will highlight recent advances in data-driven predictive control methods that eliminate the need for explicit models, robust control approaches that guarantee safety even as the controller learns, and real-time optimization techniques that allow predictive control to run reliably on embedded platforms. Along the way, I will share applications in energy systems, power electronics and robotics, demonstrating how automation can be made more scalable, reliable, and cost-effective.
Biography: Colin Jones is an Associate Professor in the Automatic Control Lab at EPFL, Switzerland. Before joining EPFL, Colin held postdoctoral and senior researcher positions at the University of Cambridge and ETH Zurich respectively. He received bachelor and master degrees in Electrical Engineering from the University of British Columbia in 1999 and 2001, before completing a Ph.D. in Engineering at the University of Cambridge in 2005 for his work on polyhedral methods for constrained control. At EPFL, he is Director of the Robotics and Intelligent Systems (EDRS) doctoral program. He has also served as Associate Editor for leading journals such as the IEEE Transactions on Automatic Control, the IEEE Transactions on Control Systems Technology, Systems & Control Letters, and Optimal Control Applications and Methods. His research interests include data-driven predictive control, optimization, and embedded real-time solvers, with applications in energy-efficient buildings, smart grids, and autonomous vehicles and has published over 250 papers in these areas.
In this talk, I will present our group's efforts to move towards "hands-off" automation — designing and deploying controllers that require minimal engineering intervention. I will highlight recent advances in data-driven predictive control methods that eliminate the need for explicit models, robust control approaches that guarantee safety even as the controller learns, and real-time optimization techniques that allow predictive control to run reliably on embedded platforms. Along the way, I will share applications in energy systems, power electronics and robotics, demonstrating how automation can be made more scalable, reliable, and cost-effective.
Biography: Colin Jones is an Associate Professor in the Automatic Control Lab at EPFL, Switzerland. Before joining EPFL, Colin held postdoctoral and senior researcher positions at the University of Cambridge and ETH Zurich respectively. He received bachelor and master degrees in Electrical Engineering from the University of British Columbia in 1999 and 2001, before completing a Ph.D. in Engineering at the University of Cambridge in 2005 for his work on polyhedral methods for constrained control. At EPFL, he is Director of the Robotics and Intelligent Systems (EDRS) doctoral program. He has also served as Associate Editor for leading journals such as the IEEE Transactions on Automatic Control, the IEEE Transactions on Control Systems Technology, Systems & Control Letters, and Optimal Control Applications and Methods. His research interests include data-driven predictive control, optimization, and embedded real-time solvers, with applications in energy-efficient buildings, smart grids, and autonomous vehicles and has published over 250 papers in these areas.
Practical information
- General public
- Free