RRT-GD: An efficient rapidly-exploring random tree approach with goal directionality for redundant manipulator path planning

Rapidly-exploring random trees (RRT) method performs well for obstacle-avoiding motion planning. However, it is not fast enough and less goal-direction. In this paper, we propose an effective rapidly-exploring random trees method with goal directionality (RRT-GD) for redundant manipulator path planning. The presented RRT-GD model concentrates both on the status of the end actuator and on the movements of the joints that the Newton-Raphson method is employed for inverse kinematics. In order to speed up the path planning, the path searching algorithm is modified to be goal-directionality and the quintic polynomial is applied to smooth the planned path. In the experiments, it shows that RRT-GD tends to be 10 or more times faster than the usual RRT algorithm when doing the same obstacle-avoiding path planning task. Therefore, RRT-GD could be more efficient and useful in doing our daily works.