User Interfaces and Reinforcement Learning for Motion Control
Dept of Computer Science
Time and Place
Mauldin Auditorium (NSH 1305)
Refreshments 3:15 pm
Talk 3:30 pm
Motion control is a wonderfully rich and interesting problem in the context of both animation and robotics. We first present recent work related to user interfaces for animated motions (motion sketching, sports prototyping). We then present ongoing work on the use of policy-search reinforcement learning for automated controller design and demonstrate results on a number of interesting dynamical systems.
Michiel van de Panne
obtained his B.A.Sc. in 1987 (
His research interests are in computer graphics, physics-based animation and simulation, motion planning, robotics, and applications of machine learning. He has served on the program committees of ACM SIGGRAPH, ACM I3D, and Graphics Interface and co-chaired the 1997 Eurographics Workshop on Animation and Simulation. He recently helped found and co-chair the ACM/Eurographics Symposium on Computer Animation in 2002, and he is the co-program chair of Graphics Interface 2005.
For appointments, please contact .