An adaptive environment informed rapidly-exploring random tree for citrus picking manipulator path planning

In unstructured environments, citrus-picking manipulators frequently encounter various obstacles, such as leaves and branches, during motion, leading to high costs and low efficiency in path planning. To address this problem, an adaptive environment-informed rapidly-exploring random tree (AEI-RRT) path planning algorithm is proposed. First, an adaptive target bias strategy is incorporated into the informed rapidly-exploring random tree (RRT) algorithm to improve the node search's ability to adapt to the environment. Second, a hybrid sampling space is employed to reduce node randomness and avoid falling into local optima. Third, local thinking is applied to adjust the dynamic step size, reduce path cost, and improve search efficiency. Finally, the AEI-RRT algorithm is compared with three well-known RRT variant algorithms. The experimental results demonstrated that the proposed AEI-RRT algorithm offers significant advantages in terms of search time efficiency, search nodes, and path length. Furthermore, the proposed AEI-RRT algorithm is integrated into the joint space. The results indicate that the algorithm effectively plans an optimal path.