Research on Obstacle Avoidance Path Planning of Manipulator with AGS-RRT Algorithm

In order to solve the problem of improving the efficiency of robot arm path planning in the obstacle environment, this paper proposes an AGS-RRT(Rapid expanding Random Tree)algorithm, the collision detection mechanism between the manipulator and the obstacle is established by using the collision detection model based on the cylindrical and spherical bounding box, and he shortest distance between the manipulator and the obstacle is obtained instantly. In addition, the sampling process of the RRT algorithm is optimized in this paper, and the improved Gaussian constrained sampling is used instead of random sampling, which makes the sampling process closer to the channel between obstacles. This method can speed up the forward search speed. In terms of node expansion, a new node constraint strategy of artificial potential field combined with adaptive step size is adopted. This method makes full use of the environment and node information, quickly expands to the target area, and reduces the over expansion of exploration and collision areas. For path planning smoothing in complex, multi-obstacle, and dynamic environments, we use a smoothing function based on B-Spline curves, and use Cartesian space spanning trees to re-plan local paths, and the pruned and optimized paths are smoother. Through simulation experiments, the effectiveness of the AGS-RRT algorithm is verified, and it is proved that the method has good obstacle avoidance performance while improving planning efficiency.