Swing Leg Trajectory Optimization for a Humanoid Robot Locomotion

Abstract

The problem of walking trajectory optimisation for bipedal humanoid robots attracts many researchers because of excessive interest to bipedal locomotion. The main focus is usually on robot dynamics and trajectory planning for predefined walking primitives. In contrast to other works, our paper targets to obtain optimal walking primitive for swing leg trajectory of bipedal humanoid robot walking. Optimal walking primitives are obtained taking into account velocity and acceleration physical limitations of each joint and are derived for different walking parameters such as step size and hip height. To obtain a desired time-optimal trajectory dynamic programing approach is used. It is shown that a new trajectory is performed within a shorter time comparing with commonly used locomotion trajectories for bipedal robots control. The results allow us to assign walking parameters and corresponding walking primitive that maximize robot velocity for predefined environment constraints.