An inverse dynamics-based multi-contact locomotion control framework without joint torque feedback

Abstract

Humanoid robots are expected to evolve in complex environments performing parallel tasks including balance in multi-contact motion. Balance requires precise tracking of the contact forces, even in the presence of external disturbances. In this paper, we propose a framework to perform stabilization, force tracking, kinematic tasks, and disturbance-rejecting compliance with robots without joint torque feedback. The solution uses a QP with concurrent tasks to produce an inverse dynamics-based feed-forward torque together with kinematic feedback to achieve feasible Lyapunov-stable motions. The framework offers a range of task formulations and parameters as tools for fine force tracking, including an admittance-like task. This framework is tested in dynamic simulations with several locomotion scenarios in complex environments with continuous non-modeled disturbances.