Mon, Nov 28
Speaker: Tom Erez, Ph.D. (Post-Doctoral Candidate)
Title: Applying trajectory optimization to biomechanical simulations
Abstract: Optimal control offers an appealing framework for generating rich motor behavior (e.g., legged locomotion) from simple first principles (e.g., move forward, act efficiently). In order to apply optimal control to complex dynamical systems such as biomechanical simulations, we must venture beyond the standard linear-quadratic regulator (LQR) theory, and employ algorithms of trajectory optimization, which scale well to many degrees of freedom. In this talk I will discuss the basic principles of trajectory optimization and present my recent work on Infinite-Horizon Model-Predictive Control (see http://www.youtube.com/watch?v=kUfmnoobTHQ), a trajectory optimization paradigm that enables the control of legged locomotion in real time.
Host: Dr. Kording