Variable Sampling Area RRT Algorithm Based on Limiting Joint Angle for Robotic Arm Path Planning

Abstract

It is an important capacity to plan an optimal working path without collision for robotic arms. Rapid exploration random tree (RRT) is a popular algorithm in the field of path planning. However, the high computational cost caused by directly using RRT to plan the path of robotic arms hinders the practical application of RRT. In this paper, a variable sampling area RRT (VSA-RRT) algorithm is proposed to improve the computational efficiency of robotic arm path planning. An obstacle avoidance model based on limiting joint angle is first developed to judge the collision. Through this obstacle avoidance model, the collision judgment of robotic arm is converted into the joint angle judgment problem, and thus, the collision situation can be directly recognized with the given joint angles. Then, a sampling area dividing strategy is proposed on the basis of the limiting joint angles, where the motion space of robotic arm is divided into several subareas. Subsequently, VSA-RRT is proposed by integrating both the obstacle avoidance model and the sampling area dividing strategy. In VSA-RRT, the sampling area is automatically selected from the divided subareas, leading the sampling process to continuously move towards the target point, and the new sampling point is generated by using the obstacle avoidance model, ensuring that any new sampling point is collision-free. Finally, the effectiveness of the proposed VSA-RRT is demonstrated with three examples. The example results show that VSA-RRT decreases the sampling number by 44.98%, 19.95% and 15.00% compared with the basic RRT, RRT with SAD and APF-RRT respectively.