Obstacle Avoidance Algorithm for Redundant Manipulators Based on Weighted Generalized Inverse

Abstract

In this paper, a gradient projection method for redundant robot manipulators based on weighted generalized inverse was presented. By weighting the Jacobian matrix and gradient, not only the obstacles, but also the joint limits could be avoided. In the process of obstacle avoidance, the evaluation function called danger field measuring the dangerous level between the manipulator and obstacles was used and improved. According to the relationship between the value of danger field and the preset threshold, a smooth and continuous transition between the primary and secondary tasks is allowed. This method was proved to be effective by numerical simulations in Matlab.