Motion control for humanoid robots based on the motion phase decision tree learning

Abstract

Humanoid robots, due to their link structure with high degree of freedom and the substitutability for human work, require a sophisticated motion control technique regardless of the type of motions or the environments. This paper gives a concept learning-based approach to this problem. We propose a motion generation method based on decision tree learning with motion phase. The system can generate a stable and anti-tumble motion which transforms the robot into a target posture. In experiment, the target motion are to stand up from a chair. Some stable and anti-tumble motions to stand up from a chair were performed by humanoid robot HOAP-1. In this paper, we discuss the validity of motion control considering motion phase.