Obstacle avoidance for redundant robots using Jacobian transpose method

摘要

Conventional approaches for collision avoidance often use the gradient projection method on minimum distance function. This paper discusses several shortcomings to draw special attention when it is applied to avoid collision with multiple obstacles. As a remedy to the identified problems, a novel method of collision avoidance based on Jacobian transpose method is proposed, which does not require to calculate the gradient of the minimum distance function, hence resulting in a computationally simple algorithm. In addition, an efficient collision detection algorithm for all possible spatial as well as planar configurations is presented. Simulation results with a four Degrees of Freedom planar manipulator are given to illustrate the several shortcomings of conventional methods and to validate the proposed method.