A Dynamic Obstacle Avoidance Algorithm Based on RRT for Apple Picking

This paper presents an improved dynamic obstacle avoidance path planning algorithm, named DOA-RRT* (Dynamic Obstacle Avoidance RRT*), designed to address the challenges posed by dynamic obstacles in apple picking tasks. The traditional RRT algorithm primarily focuses on spatial information and struggles with dynamic obstacle avoidance. The DOA-RRT* algorithm addresses this by introducing a temporal dimension to the RRT* structure, enabling it to consider both space and time for safer robotic arm operation in dynamic environments. This paper uses bounding spheres to simplify obstacle representation and predict their positions based on time and velocity. During path planning, DOA-RRT* initially treats dynamic obstacles as static to generate a global path, subsequently predicting their movements and monitoring for collisions. If a collision is detected, only the affected path sections are replanned. Experiments indicate that DOA-RRT* creates collision-free paths in dynamic settings. Compared to traditional static path planning algorithms, it dynamically adjusts paths in response to changing obstacles, thereby ensuring the accuracy and safety of the robotic arm during apple picking.