« All Events
Abstract:Recent advances in machine learning, control, and robotics show promising results towards integrating autonomous systems into society. Legged robots in particular suggest potential for the same dynamic capabilities as humans to adapt to everyday, and even challenging, environments. However, when compared with humans and animals, state-of-the-art robotic systems do not yet demonstrate the same agility nor intelligence to navigate the real world. While important contributions have been made to approach human levels in specific tasks, the desired system generalizability to interact with and adapt to new environments remains challenging. Additionally, for tasks learned with machine learning in which robotic systems do approach or surmount human-level skills, the underlying neural network function approximation lacks interpretability and performance guarantees. This is true for both Artificial Neural Networks, as well as their biological counterparts that exist in animals. In this talk, we present several methods to maximize robotic system performance and explainability by leveraging ideas from machine learning, model-based control, and neuroscience. Example applications will be shown for highly dynamic motions on systems such as quadrupeds, vehicles, and wheel-legged robots.
Bio:Guillaume Bellegarda is a postdoctoral researcher in the Institute of Mechanical Engineering at École Polytechnique Fédérale de Lausanne (EPFL). He was previously a postdoctoral researcher at University of Southern California, and received his Ph.D. and M.S. degrees in Electrical and Computer Engineering from University of California, Santa Barbara, and his B.S. degree in Electrical Engineering and Computer Science from University of California, Berkeley. His research draws inspiration from machine learning, model-based control, and neuroscience to maximize explainable performance for dynamic robotic systems, as well as to deepen understanding of their biological system counterparts to adapt to real world situations.