Obstacle Detection And Avoidance Of Automatic Guide Vehicle Based On Stereo Vision

Accompany with the development of industrial production automation,intelligent recognition technology as well as the development of Automated Storage and Retrieval System(AS/RS),Automated Guided Vehicle(AGV)has been important tool as the link and adjustment the discrete logistics system,for the industry in the warehouse handling.In the artificial intelligence developing today,AGV will have broader development based on vision guided.Meanwhile,obstacle avoidance research of AGV is also gathering more and more attention.Machine vision image processing technology is applied to the environment identification and obstacles avoidance,to improve the recognition ability of the AGV.The study of stereo vision has always been a hot and difficult topic in machine vision.By using stereo vision system we can determine the 3D profile of arbitrary objects and get the 3D coordinates of an arbitrary point on the contour.Stereo vision system composed of dual camera from different angles and two digital images of the measured object,or by a single camera at different times from different angles for two digital images of the object to be measured.Based on the principle of parallax to recover 3D geometric information of objects,we can build 3D profile and position.This paper embarks from the reality while simulating human vision,to achieve obstacle recognition by double camera.We use camera as the image acquisition,and then image feature extraction by a pair of images.After image correspondence between images and disparity computation,according to the physical model of the camera and the image matching the parallax D,calculate the depth of field by using the triangular geometry model.The car can judge the obstacle and avoid obstacle according to the three-dimensional map.In the third chapter of this paper,we will establish a program for 3D reconstruction based on the Studio 2012 platform by using open vision library Emgucv.According to the characteristics of the visual obstacle detection,this paper put forward the artificial potential field method to match stereo vision.But the traditional artificial potential field method has defects such as local minima.This paper analyzes three kinds of local minima appear.According to the actual proposed,we insert force factor and set a virtual target to solve this problem.When AGV is facing great obstacle avoidance or has entered the obstacle danger zone,the car will activate edge walking behavior activation.Finally,we design MATLAB simulation program to prove the feasibility of the obstacle avoidance strategy.