基于机器视觉的矿用巡检机器人环境感知研究  

Study on Mine Patrol Robot Environment Perception Based on Machine Vision

作　　者：杨静宜[1] 赵莉娅 梁月肖 YANG Jingyi;ZHAO Liya;LIANG Yuexiao(Hebei Polytechnic Institute,Shijiazhuang 050091,China)

机构地区：[1]河北工程技术学院,石家庄050091

出　　处：《煤炭技术》2023年第9期227-229,共3页Coal Technology

基　　金：河北省科学技术厅科技计划项目(16210804)。

摘　　要：煤矿巡检路径环境多为边界模糊的非结构化平面,导致巡检机器人环境感知效果不理想,自主控制能力偏弱,巡检效率较低。基于机器视觉理论,运用直通滤波、B样条曲线对点云数据进行预处理,利用SEEDS分割方案对图像子区域进行划分,并以巷道地面为例,通过支持向量机(SVM)来验证拟合结果的准确性。实验结果表明,边界拟合结果与实际巷道路面的边界较为一致,巡检机器人环境感知能力有了较大提升。The coal mine patrol route environment is mostly an unstructured plane with blurry boundaries,which results in poor environmental perception,weak autonomous control ability and low patrol efficiency of the patrol robot.Based on machine vision theory,point cloud data is preprocessed with a pass filter and B-spline curve.Image subareas are divided by SEEDS segmentation scheme,and roadway ground is taken as an example to verify the accuracy of the fitting results by support vector machine(SVM).The experimental results show that the boundary fitting results agree with the actual roadway boundary,and the patrol robot′s environmental awareness has been dramatically improved.

关 键 词：机器视觉 巡检机器人 环境感知 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]