This video shows the first application of whole-body nonlinear model predictive control (MPC) on a humanoid robot. It is the first time that MPC is applied in real time to the whole body of a humanoid robot, controlling the balance, handling different constraints (e.g. collisions, joint limits) and performing the specified reaching task at the same time. In particular, we focus on the problems raised by the delays due to computation and by the differences between the real robot and the simulated model. The implementation of a complete model-predictive controller and its application to the physical HRP-2 robot is based on the optimal-control solver MuJoCo
Социальные закладки