From pixels to torques: output feedback for robotics

Russell Tedrake
Massachusetts Institute of Technology

Time and time again, I have watched clever engineers design simple controllers for complex robots which (often dramatically) outperform the best controllers I know how to design with "optimal" control. In this talk, I'll discuss a few specific examples and some recent progress towards algorithms for reliably synthesizing controllers of this form.

Back to Intersections between Control, Learning and Optimization