Research on Path Planning of Robot Arm Based on RRT-connect Algorithm

摘要

To address the shortcomings of the RRT-connect algorithm in terms of low efficiency of path planning in sampling space and high randomness of node sampling, this paper proposes a three-source fast extended random tree algorithm (GT-RRT) combined with gravitational field. The algorithm adds a third node as a new extension node between the start point and the target point, so that the algorithm can extend the random tree from the start point, the target point and the third node. At the same time, a gravitational field is superimposed on each of the three nodes to guide the generation of nodes, reducing the search range of the null space. The GT-RRT and RRT-connect algorithms are compared in 30 simulation experiments, and the results show that the improved algorithm has better path planning efficiency than the original algorithm in complex environments. Finally, this paper verifies the effectiveness of the improved algorithm in both the moveit simulation and the real environment by using the robot arm of Baxter robot as an example.