非结构化环境下的机械臂抓取位姿检测系统研究

打开文本图片集

引用格式：，，.非结构化环境下的机械臂抓取位姿检测系统研究[J].现代电子技术，2026，49（6）：133-138.

关键词：机械臂抓取；位姿检测；GraspNet；SAM；非结构化环境；Sawyer机械臂中图分类号：TN911.7-34；TP242 文献标识码：A 文章编号：1004-373X（2026）06-0133-06

Research on robotic arm grasp pose detection system in unstructured environment

SUNShijie1，KANGGaoqiang²，PANWenbo² （1.ScholofTransportationandElectricalEngineering，HunanUniversityof Technology，Zhuzhou412oo7，China; 2.CRRC Zhuzhou Electric LocomotiveResearch InstituteCo.，Ltd.，Zhuzhou 4120O1，China）

Abstract：In order to enhance theaccuracyandrobustness of robotic grasp pose detection in unstructured environments，a method of grasping pose detection integrating segment anything model （SAM） image segmentation with theGraspNet network is proposed.SAMmodelisusedtosegmenttargetmaskstofilterouttheinterferencefromnon-targetareas，andcombinedwiththe GraspNetnetwork topredict thegraspingpose.The high-robustnessschemeisscreenedbasedon forceclosure theory.The calibratedRealSensecamera isusedtoobtaintheRGB-Ddataandintrinsicparameters，soastoestablishcoordinate transformationrelationshipbetweethecameraandrobotbaseframes.Theroboticarmiscontrolledbymeansofcoordinate transformationand inversekinematicsforthegrasping.The Sawyer7-axicroboticarmand D435icameraareused toestablish theverificationplatformfortheexperiments.Theresultsshowthatthetargetdetectionrateandaveragegraspsuccessrateofthe improvedalgorithmcanreach91.7%and81.6%，respetively，whichare7.5%and8.3%higherthanthoseoftheoriginal GraspNetalgorithmandcansignificantlyreducetheincorrectposescaused byinterferencefromnon-targetobjects.Theproposed methodcanimprove thegraspingaccuracyandgeneralizationabilityinunstructuredenvironmentsbymeansofthecolaborative optimizationofGraspNetandzro-shotsegmentation，providingafeasibletechnical solutionforindustrialautomationandobot grasping tasks.

Keywords：robot arm grasping; pose detection; GraspNet; SAM;unstructured environment; Sawyer robotic arm

0 引言

机械臂自标抓取作为工业自动化与智能机器人领域的核心任务，其精度与鲁棒性直接影响生产效率与系统可靠性[1-2。（剩余9237字）