Integrated Dynamical Systems Lab, Kyoto University
Tuesday, November 15, 2022, 11:00 - 11:59
Room 01-012, Georges-Köhler-Allee 102, Freiburg 79110, Germany
Model predictive control (MPC) has been widely utilized in legged robotics as an online planner based on simplified dynamical models rather than a controller based on whole-body dynamical models. MPC based on the whole-body models, termed whole-body MPC, is challenging as robotic systems typically involve fast, complicated, and non-smooth dynamics. In this talk, we present our algorithm developments towards the online numerical optimization of whole-body MPC. We also show that our whole-body MPC could realize dynamic and versatile motions of legged robots both in physical simulations and hardware experiments.
Also online via Zoom:
Meeting ID: 627 9173 7415