Powered by Indico
v3.2.8
FI Computational Methods and Data Science Journal Club
Flatiron Institute, 162 5th Avenue
Speaker: Ludovic Righetti
Title: Robots robustly interacting with their environment: algorithms and challenges
Abstract:
To walk, run, jump or manipulate objects, robots need to constantly interact with objects and the environment. Unfortunately, reasoning about physical interactions is a computationally daunting task. For this reason, robots try to avoid physical interactions at all costs and unexpected physical contacts often lead to failures. This is in stark contrast with humans or animals, that not only constantly interact with their environment but also exploit this interaction to their advantage. For example, when walking up step stairs, one would naturally grab a handrail to improve stability and reduce effort, a decision incredibly difficult to automate in robotics. In this talk, I will present our recent work to break down this complexity, leveraging both optimal control and reinforcement learning. I will argue that understanding the structure of movement optimization problems and how to regulate mechanical impedance is central to solving these challenges. I will then present model predictive control and reinforcement learning algorithms to generate dynamic behaviors for legged robots in real-time. I will also introduce our quadruped and biped open-source robots and show experiments to illustrate the performance of our methods as well as important remaining challenges.
Short bio:
Ludovic Righetti is an Associate Professor in the Electrical and Computer Engineering Department and in the Mechanical and Aerospace Engineering Department at the Tandon School of Engineering of New York University and a Senior Researcher at the Max-Planck Institute for Intelligent Systems in Germany. He holds an Engineering Diploma in Computer Science and a Doctorate in Science from the Ecole Polytechnique Fédérale de Lausanne. He was previously a postdoctoral fellow at the University of Southern California and a group leader at the Max-Planck Institute for Intelligent Systems. He has received several awards, most notably the 2010 Georges Giralt PhD Award, the 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems Best Paper Award, the 2016 IEEE Robotics and Automation Society Early Career Award and the 2016 Heinz Maier-Leibnitz Prize. His research focuses on the planning, control and learning of movements for autonomous robots, with a special emphasis on legged locomotion and manipulation. He is more broadly interested in questions at the intersection of decision making, automatic control, optimization, applied dynamical systems and machine learning and their applications to physical systems.
Please email ccaadmin@flatironinstitute.org to register.