Performance comparison of different algorithms.

In response to the widely used RRT-Connect path planning algorithm in the field of robotic arms, which has problems such as long search time, random node growth, multiple and unsmooth path turns, a path planning algorithm combining dynamic step size and artificial potential field is proposed. To solve the problem of scattered sampling points in the RRT-Connect algorithm, a goal-biased strategy is introduced. To address the problem of slow expansion caused by using fixed step sizes, a dynamic step size strategy is introduced to dynamically adjust the step size. To reduce randomness in the expansion process, the artificial potential field method is integrated to constrain the growth of new nodes by the random sampling function, the target gravitational function and the repulsion function. Finally, the planned path is pruned and smoothed using cubic B-splines to improve redundant points and turns in the path, and reduce the occurrence of shaking during the motion of the robotic arm. In the same environment, the improved algorithm reduces path length by 15.4% and planning time by 49.2%, compared with the RRT-Connect algorithm.