Design Of AGV Obstacle Avoidance System Based On Binocular Stereo Vision

AGV(Automatic Guidance Vehicle),as a typical representative of smart factories,has been more and more practiced in the production of intelligent manufacturing.The AGV needs to monitor the surrounding obstacles during operation to achieve accurate obstacle avoidance at the complex environment of the equipment production for the purpose of meeting the safety requirements of AGV application in the production environment.This paper proposes an AGV obstacle avoidance system based on binocular stereo vision.The main contents are as follows:(1)The principle of binocular imaging and existing stereo matching algorithms are summarized.The research focuses on the realization of stereo matching.According to the characteristics of the research content,the disparity map of the edge information of the object is highlighted.And a semi-global stereo matching algorithm,based on the adaptive window of edge information,is proposed to provide a theoretical basis for the obstacle avoidance of AGV system.(2)The global path planning principles and planning steps of the traditional ant colony algorithm are analyzed.And for the unexpected static obstacles or dynamic obstacles in the environment map,the research first analyzes the movement state and orientation information of the obstacles,and then proposes obstacle avoidance strategies for local maps to achieve accurate AGV obstacle avoidance.(3)Four Mecanum wheels as the moving wheels of the AGV and fuzzy PID as the control method are selected to achieve the omnidirectional movement of the AGV.Cortex-A9 ARM was used as the core controller of the AGV.The software architecture was designed to test the AGV function based on the Ubuntu system programmed.(4)The established AGV experimental platform can realize vision-based binocular ranging and obstacle avoidance functions.It implement AGV binocular vision navigation and control systems.The algorithms and Obstacle avoidance strategies are tested and analyzed through the feasibility experiment of AGV obstacle avoidance strategy and control system,The experimental results show that the error of AGV system ranging is less than3.5% and the successful avoidance rate of the AGV system in this paper reaches 96.7%.The control accuracy and the turning ability is good.The average deviation error and the parking error are small.It can achieve accurate obstacle avoidance and meet the AGV work requirements.