Fast and dynamically stable optimization-based planning for high-DOF human-like robots

Abstract

We present a novel optimization-based motion planning algorithm for high degree-of-freedom (DOF) robots. Our approach combines a multi-contact dynamics formulation with optimization-based planning to compute collision-free, smooth and dynamically stable trajectories. Our formulation is general and can handle multiple simultaneous contacts and non-planar surfaces. We highlight the performance of our algorithm in simulated environments on human-like robots with tens of DOFs.