基于3D视觉的船体构件智能焊接研究及应用  

Research and Application of Intelligent Welding Technology for Ship Components Based on 3D VisionTitle

作　　者：许曦 张兆彪 李俊渊 高建文 闫德俊 张弓 XU Xi;ZHANG Zhaobiao;LI Junyuan;GAO Jianwen;YAN Dejun;ZHANG Gong(Shenzhen Qianhai Ruiji Technology Co.,Ltd.,Shenzhen 518000,China;Guangdong Provincial Key Laboratory of Advanced Welding Technology for Ships,Foshan University,Foshan 528051,China;Institute for Super Robotics(Huangpu),Guangzhou 510700,China;Guangzhou Institute of Advanced Technology,Guangzhou 511458,China)

机构地区：[1]深圳前海瑞集科技有限公司,广东深圳518000 [2]佛山大学广东省舰船先进焊接技术重点实验室,广东佛山528051 [3]超级机器人研究院(黄埔),广东广州510700 [4]广州先进技术研究所,广东广州511458

出　　处：《电焊机》2024年第12期35-41,共7页Electric Welding Machine

基　　金：广东省舰船先进焊接技术企业重点实验室基金(2023B1212070026)。

摘　　要：为解决船体结构件组立板人工焊接精度低、劳动强度大、效率低等问题,提出了一种基于线激光传感器与结构光传感器结合的点云采集方法。该方法通过3D视觉传感器和大范围线激光传感器采集多模态点云,提取复杂焊接-加工轨迹,开发了免示教智能焊接机器人工作站。通过3D视觉引导焊接机器人定位识别焊缝信息,生成可执行的焊接轨迹,并结合先验焊接工艺进行焊接,实现小批量多种类结构复杂船舶结构件的高效、稳定与灵活的智能化焊接及焊缝加工。试验结果表明,该系统能够有效提高焊接精度和效率,减少人工干预,满足船舶制造等领域大型复杂结构焊接的需求。研究还针对3D视觉算法在处理复杂异形曲面工件焊缝点云提取不准确、焊缝识别偏位等问题,采用了基于轮廓与焊缝点比对的新算法,优化了焊接机器人焊缝识别和特征提取算法。通过自主研发视觉硬件集成等方面的技术突破,推动船舶结构及焊接机器人产品向高端化、智能化发展。To address the issues of low precision,high labor intensity,and low efficiency in manual welding of ship hull structural components,a point cloud acquisition method based on line laser sensors combined with structured light sensors is proposed.This method collects multimodal point clouds through 3D vision sensors and large-range line laser sensors,ex‐tracts complex welding-machining trajectories,and develops an unmanned intelligent welding robot workstation.The 3D vi‐sion guides the welding robot to locate and identify weld seam information,generates executable welding trajectories,and combines prior welding processes for welding,achieving efficient,stable,and flexible intelligent welding and weld process‐ing for small batches of various types of complex ship structural components.Experimental results show that this system can effectively improve welding accuracy and efficiency,reduce manual intervention,and meet the needs of large-scale complex structure welding in fields such as ship manufacturing.The study also addresses issues such as inaccurate extraction of weld seam point clouds and offset recognition of weld seams on complex irregular curved workpieces by 3D vision algorithms,adopting a new algorithm based on contour and weld seam point comparison,optimizing the welding robot's weld seam rec‐ognition and feature extraction algorithms.Through technological breakthroughs in independently developed visual hard‐ware integration and other aspects,it promotes the development of ship structures and welding robot products towards high-end and intelligence.

关 键 词：线激光传感器 结构光传感器 3D视觉算法 特征提取 

分 类 号：TG409[金属学及工艺—焊接]