Research On Installation Method Of Cable In Constrained Space Based On Vision

In the industrial production line,the robotic arm greatly improves the assembly efficiency of rigid parts.However,linear deformable objects(DLO),such as cables,ropes,are subject to external forces due to their many degrees of freedom.The end position of the entire DLO is difficult to predict from the end position of the robot arm.Therefore,the assembly of the DLO is difficult to complete with only robot arms.Aiming at the assembly problems of cables with certain flexibility and rigidity and researches related problems of an automatic cable assembly in restricted environments,which mainly include shape tracking technology for unmarked cables under no model and cable shape servo control.Aiming at the problems that may arise in cable assembly in confined spaces and proposes related solutions.To obtain the shape information of the unmarked cable,a vision-based geometry measurement is adopted.Extract the centerline of the cable in the color image obtained by the camera.At the same time,the Convolutional Neural Network(CNN)is trained to identify the location information of the topology of the cable.According to the centerline and topology information of the cable,complete the traversal and down-sampling of DLO from end to end.In order to track changes in the shape of the cable during operation and the problem of partial occlusion of the cable during operation,point cloud registration methods including improved Coherent Point Drift(CPD)with Local Linear Embedding(LLE)are used to downsample The result is registered with the point cloud of the cable to obtain the current cable information.Finally,according to these data,the parameterization of the cable curve is completed by B-spline curve fitting.Considering that the cables are all installed in a constrained environment,the presence of other obstacles in a restricted assembly environment reduces the efficiency of path planning of the manipulator,and it is even difficult to specify an unknot strategy.A gripper with the function of pulling and feeding the cable is designed to realize the operation of the cable without the moving of the manipulator,which ensures the efficiency of the cable operation of the manipulator in a restricted environment.In order to verify the effectiveness of the above methods,a visual servo control platform based on the Robot Operating System(ROS)was built.On this platform,a vision-based,non-modeling,and automatic assembly experiment of unmarked cables was completed,and the acquisition and real-time tracking of cable shape information during the operation of the robot arm were completed.And analyzed the experimental results.