关键词: 双目视觉, 自动导引车, 障碍物检测, 帧差法, 模糊PID

Abstract: According to the uncertainty of materials' position and motion state in intelligent workshop,a method of obstacle detection and avoidance for Automatic Guided Vehicle (AGV) based on binocular vision was proposed.During obstacle detection process,the obstacle detection algorithm based on depth detection was used to judge the existence of obstacles.An algorithm for obtaining the information of obstacles was proposed by frame subtraction to get the information of obstacle position and velocity in the static and dynamic state.To improve the reliability of obstacle detection,the static and dynamic obstacles were detected by experiments.The result showed that the algorithm was reliable and had little error.Aiming at the different obstacles,the obstacle avoidance strategy was proposed.PID and fuzzy PID method were used to control AGV motion,and the effectiveness of obstacle avoidance strategy was proved by analyzing the angle and distance deviation of motion.A simple and efficient PC software was designed to verify the feasibility and effectiveness of the proposed algorithm.

Key words: binocular vision, automatic guided vehicle, obstacle detection, frame subtraction, fuzzy PID