基于单目视觉的仓储物流搬运AGV累积误差检测方法研究  

Research on detection method of AGV cumulative error in warehouse logistics transportation based on monocularvision

作　　者：刘鹏 肖扬[2] 曹小华[1] Liu Peng;Xiao Yang;Cao Xiaohua

机构地区：[1]武汉理工大学物流工程学院,武汉430063 [2]天津港股份有限公司,天津300461

出　　处：《起重运输机械》2020年第9期42-46,共5页Hoisting and Conveying Machinery

基　　金：国家重点研发计划(2018YFC1407405);国网科技项目(5418-201971157A-0-0-00)。

摘　　要：双轮差速驱动搬运AGV在实际运动过程中,由于传感器读数误差和机械传动误差会产生运动累积误差,针对该问题,文中提出一种基于单目视觉的误差检测方法。首先采用基于栅格分割、聚类的方法提取二值图像中的定位块并使用平均值算法计算定位块的中心点;然后根据定位块中心点的相对位置关系建立误差检测计算模型;最后通过实验对本方法进行验证,位置和航向角检测结果误差分别小于1mm和1°,证明所述方法具有较高的检测精度。该方法对硬件要求低,具有较高的实用价值。In the actual movement process,due to sensor reading error and mechanical transmission error,double wheel dif-ferential drive AGV will produce cumulative movement error.To solve this problem,this paper proposed an error detection method based on monocular vision.Firstly,the method based on grid segmentation and clustering was adopted to extract the positioning blocks in the binary image and the average algorithm was used to calculate the center point of the positioning blocks.Then an error detection calculation model was established according to the relative position relation of the center points of the positioning blocks;finally,the method was verified by experiments.Errors of position and heading angle detec-tion results were less than 1 mm and 1°respectively,proving that the method had higher detection accuracy.The method has low requirements on hardware requirements and high values in application.

关 键 词：仓储物流 AGV 累积误差 栅格分割 聚类算法 平均值算法 误差检测 

分 类 号：TP249[自动化与计算机技术—检测技术与自动化装置]