爬壁机器人焊缝跟踪过程中的焊缝检测与识别研究  

Study on Welding Seam Inspection and Identification in Wall-climbing Robot Welding Seam Tracking

作　　者：王志刚[1] 张轲[1] 罗志锋[1] 陈易新[1] 

机构地区：[1]上海交通大学上海市激光制造与材料改性重点实验室

出　　处：《热加工工艺》2018年第1期210-213,219,共5页Hot Working Technology

摘　　要：实现爬壁机器人焊缝自动跟踪的前提条件是获取实时的焊缝位置信息。首先,基于爬壁机器人系统平台,提出了一种结合被动视觉传感器和辅助光源的焊缝采集方案。采用了一套精确度与实时性较高的焊缝图像处理及特征提取算法,并针对焊缝图像细化后的轮廓线上存在毛刺的问题,提出一种有效的毛刺去除递归算法。结果表明,设计的焊缝识别算法能够提取到准确的焊缝位置信息,并且单幅焊缝图像的平均处理时间为56ms,能够满足焊缝跟踪的实时性要求。The first step for achieving the welding seam automatic tracking of wall-climbing robot is to obtain real-time weld position information. Firstly, based on the wall-climbing robot system, a welding seam acquistion system with passive vision sensor and auxiliary light source was proposed. Then, a high precision and real-time welding seam image processing and feature extraction algorithm was adopted. Aiming at the problem of burrs on the contour line after image thinning, an effective deburring algorithm was proprosed. The results show that the designed identification algorithm of welding seam can extract accurate weld position information, and the average processing time of single weld image is 56 ms, which can meet the real-time requirement of welding seam tracking.

关 键 词：焊缝图像处理 特征提取 被动视觉 爬壁机器人 焊缝跟踪 

分 类 号：TG438.2[金属学及工艺—焊接]