Motion Planning for a Humanoid Robot Based on a Biped Walking Pattern Generator

Abstract

In this chapter, we address the motion planning problem of humanoid robots dynamically walking on the ground. We first explain the analytical-solution-based online walking pattern generator. Then, we explain several motion planning problems based on the walking pattern generator. Section 7.3 explains a two-stage time-parametrized approach for the whole-body motion planning where, in the first phase, the constraint condition is generated as a function of time by using the biped walking pattern generator, and in the second phase, collision-free whole-body motion is planned by using this constraint condition. Section 7.4 explains the simultaneous planning method of both the foot-place and the upper-body motion taking the dynamics of the robot into consideration. To realize this simultaneous planning, we use the online walking pattern generator for the offline motion planning. In Section 7.5, we explain two methods for the whole-body manipulation problem. We first explain the method for modifying the motion by considering the hand reaction force, and then explain the online gait planning approach based on the hand reaction force in order to realize the stable pushing manipulation. We will confirm the effectiveness of the approach by several simulation and experimental results.