RRT-based path planning for an intelligent litchi-picking manipulator

Aiming to realize the obstacle avoidance of the litchi-picking robot in dynamic and unstructured environments, an improved rapidly exploring random tree (RRT) algorithm is presented in this study. By establishing the collision detection model of the manipulator and obstacles, the obstacle avoidance path planning is carried out. The idea of target gravity was adopted to accelerate the path search speed. To optimize the path generated by RRT, a genetic algorithm (GA) and smooth processing were proposed. Performance tests in a virtual environment showed that the optimal combination of step sizes in the path search were 1.6 degrees for step size 1 and 2 degrees for step size 2. The time to achieve the path was 0.47 s, and the success rate was 100%. The path length was shortened by 20% after optimization. Experiments in the real environment were carried out for the 6 degrees of freedom (DOF) picking manipulator to pick litchi, and the results showed that the collision-free path planned by the proposed algorithm could successfully drive the manipulator from its initial position to the goal position without any collision.