Rapidly-exploring Random Tree Based on Improved Sampling Space and Path Processing

Path planning is key to guiding the motion of robotic arms. Aiming at the randomness in sampling and the lack of smoothness in trajectories of Rapidly-exploring Random Tree (RRT) class algorithms, this paper first proposes a method that extends the line between new nodes and target points to determine a cuboid sampling space; Secondly, the obtained path is improved based on obstacle detection between odd-numbered nodes to see whether to remove intermediate branches, making the path smoother. The improved sampling and path processing methods proposed are applied to the RRT and RRT* algorithms, and respectively named as IRRT and IRRT* algorithms. Finally, the Matlab is used to simulate the proposed algorithm. The result verified the effectiveness of the proposed method.