CS 284: Optimization Algorithms for Robotics

Semester: 

Spring

Offered: 

2016

This course is designed to introduce students to a variety of optimization algorithms for designing and controlling dynamic motions in nonlinear dynamical systems such as walking, manipulating, and flying robots. Theoretical and algorithmic topics will include optimal control foundations, trajectory optimization, nonlinear programming, trajectory stabilization, model-predictive control, Lyapunov analysis, reinforcement learning & policy search, and applications of convex optimization to control and stability analysis. We will consider these topics in the context of a variety of classical dynamical systems as well as models of legged and flying robots. Students will gain practical experience implementing modern algorithms to control simulated systems using the Drake software toolbox.