Dynamic Obstacle Detection Of Visual Tracking Navigation AGV

As an important handling tool,AGV has high automation and high work efficiency,which can save labor costs.In recent years,with the continuous development of machine vision and automatic control technology,the visual guidance AGV has been extensively studied.In this paper,combined with the actual working environment of visual guidance AGV,the navigation marker extraction algorithm and the detection and localization algorithm of obstacles are studied.The main research contents are as follows:(1)AGV vision system design.Aiming at the single problem of the navigation mark function,the navigation mark band as the guide reference is used.Based on this reference,the camera calibration and obstacle detection are integrated into the integrated AGV vision system.The navigation marker strip not only has the function of guiding the advancement of the AGV,but also the Y-axis of the world coordinate system and the only reference for the obstacle positioning in the camera calibration.(2)Navigation marker strip extraction.By analyzing the advantages and disadvantages of edge extraction algorithm and RGB color segmentation algorithm,this paper proposes a navigation marker segmentation algorithm combining prewitt edge extraction algorithm and RGB color segmentation algorithm for the problem of incomplete extraction and low precision.Through experimental verification,the algorithm can accurately and completely segment the navigation mark.(3)Direct calibration method for calibrating block coordinate point fusion.The calibration block is laid on the navigation marker strip,and then the AGV is horizontally moved to respectively capture the calibration image,the calibration block coordinate points in all the single calibration images are extracted,the calibration block coordinate points are merged,and the calibration region is surface-fitted by the cubic polynomial.Establish a mapping relationship between camera view coordinate points and image pixel coordinate points.The experimental results show that the positioning accuracy can meet the positioning requirements of the AGV system.(4)Research on obstacle detection algorithms.For the case of camera movement,combined with the safe area of AGV travel,the three consecutive frames of images are differentiated,the threshold of the difference image is calculated separately,and the threshold is compared to determine whether there is an obstacle.If an obstacle occurs,the obstacle is differentiated by the frame difference method to achieve obstacle segmentation.The experimental results show that the algorithm can effectively judge whether there are obstacles and can accurately detect the obstacle contour information.Finally,the positioning between obstacles and navigation markers and obstacles and AGVs is achieved.