A time-varying-constrained motion generation scheme for humanoid robot arms

Abstract

An efficient time-varying gesture-determined dynamical (TV-GDD) scheme is proposed for motion planning of redundant dual-arms manipulation. Motion planning for such tasks on humanoid robots with a high number of degrees-of-freedom (DOF) requires computationally efficient approaches to generate the expected joint configuration when given the end-effector tasks. To do so, we investigate a time-varying joint-limits constrained quadratic-programming (QP) approach and an efficient numerical computing method. This strategy provides feasible solutions at a low computation cost within physical limits. In addition, the joint configuration can be adjusted dynamically according to the expected gestures and tasks. Comparative simulations and experimental results on a humanoid robot demonstrate the effectiveness and feasibility of the scheme.