RSS 2015 Workshop

Bridging the Gap between Data-driven and Analytical Physics-based Grasping and Manipulation

July 16, Rome

"Optimal and learning control for dexterous, agile and versatile service robots", Jonas Buchli, ETH Zurich

I will present a unified approach to learning and optimal control applied to robotic walking, running, flying, and manipulation. A key aspect is addressing the longstanding problem of variable impedance control as an optimal control and planning challenge. The resulting problems can be solved by a pipeline of iterative optimal and learning control techniques, based on nonlinear programming, sequential quadratic control and path integral reinforcement learning. In this way we combine the best of two worlds: model based optimal control and model free reinforcement learning. I will show the application of these methods to robotics problems stemming from walking robot research, biomechanics and medical robotics, in archaeology, and in architecture.