Obstacle avoidance for redundant manipulator without information of the joint angles

Abstract

We propose an obstacle avoidance algorithm for redundant manipulator. In practical, we can not use distance in terms of the joint angles, only uses the distance informations between the robot and obstacles directly using range sensor like sonar and laser sensors. Therefore, we use quasi-desired point to avoid obstacles using distances between the obstacles and the manipulator's links without utilizing information of the joint angles. For each joints, the null space of Jacobian is used to avoid collision without changing the task space velocity. Also, we consider the additional subtask utilizing the null space control of Jacobian. Simulation results show the effectiveness of the proposed algorithm with 4 Degree Of Freedom (DOF) manipulator with two static obstacles.