Active acquisition of operating ranges and path planning for a humanoid robot

Abstract

This paper describes a new method of active acquisition of operating ranges for a humanoid robot by taking insights from pain perception, which is regarded as the function of self-preservation in nature. This study aims to determine the operating ranges for each joint of anthropomorphic humanoid robot by its own active behavior. Pain is regarded as an alert of unexpected events and the robot learns from pain in order to protect the joint actuators. The proposed method is described with some experimental results. The robustness of the robot system for dynamical changes of the surrounding environment will also be discussed. Moreover, a path planning in joint angle space for a 7 degrees of freedom (DOFs) arm is examined by utilizing the proposed method and Rapidly-Exploring Trees (RRTs).