Robotic Arm Path Planning Based on Improved BI -RRT* Algorithm

In order to solve the shortcomings of the bidirectional rapid search random tree (BI-RRT*) algorithm applied in robotic arm path planning, such as low path solving efficiency, bidirectional exploration in complex environments causing robotic arm oscillations, and complex redundancy in collision detection, etc. this paper defines an obstacle distance loss function, and quantifies the degree of risk of the sampling points by constructing an obstacle distance loss function, which realizes the dynamic adjustment of the three-stage sampling strategy to ensure path safety and improve the performance of the algorithm. In this paper, we define the obstacle distance loss function to quantify the risk level of sampling points by constructing the obstacle distance loss function, and then realize the three-stage sampling strategy, which can improve the performance of the algorithm while guaranteeing the safety of the path. In order to optimize the growth direction of the two-way search tree, this paper designs an adaptive search strategy allocation mechanism based on node evaluation. By evaluating the cost of bi-directional search tree nodes, the expansion probability of each tree is dynamically adjusted, so that the search process of the algorithm is more targeted and the path curvature is also reduced. We also use the hierarchical enclosing box method to optimize the collision detection efficiency and define dynamic safety constraints to avoid the collision risk of each linkage of the robotic arm. After the path points are generated, a greedy strategy is used to remove redundant path points to ensure that the path is more concise and meets the motion constraints of the robotic arm. Finally, the trajectory is smoothed by three times B-spline interpolation, and the MATLAB simulation results show that the improved BI-RRT* algorithm shortens the average path length by 26.2% compared with the original BI-RRT algorithm, the average path generation time reduces by 55.4%, the average number of sampling nodes reduces by 68.89%, and the positions and velocities of joints do not have large-scale abrupt changes during the movement of the robotic arm. The algorithm has been demonstrated to be effective in avoiding the problem of the robotic arm. The feasibility of this algorithm in obstacle avoidance path planning for robotic arms is confirmed.