Research On Trajectory Generation And Robot Motion Planning For Workpiece Edge Machining Based On Machine Vision

Machining along the edge of workpiece is an important process in manufacturing.Typical tasks,such as arc welding,edge grinding,edge spraying,etc.,require the tool to move along the edge of workpiece.For more and more small batch and short period flexible production requirements,the production mode based on robot teaching and playback takes a lot of programming time,moreover the existing algorithms for automatically generating tracks often rely on CAD files.And for some low-cost production scenarios where the consistency of the workpiece to be processed is not high,the pre-generated fixed path is unable to meet demand.For small batch,short cycle and low cost production scenarios,by describing the object's shape with point cloud,this paper conducts in-depth research on related edge recognition and tool generation,object pose estimation,robot motion planning,etc.,and constructs a global path planning system for machining robots that can adapt to differences in workpiece position and geometry in real time.The main research contents of this article include:1.The geometric information of the workpiece surface is described based on the point cloud,the edge area is divided by the curvature information,and the edge line is accurately extracted using local features,and the position of the object corner point is extracted,and finally the trajectory template is generated through the graphical interactive interface.During the machining process,the edge area and corner points are extracted from the adjacent area of the template through the estimation of the workpiece pose,and the real-time processing path is generated by matching the template information with the corner points.2.Based on RGB-D camera to estimate the pose of the workpiece.The self-view based calibration algorithm is used to perform rapid coarse calibration of the robot vision system,which provides good initial values for fine calibration.On this basis,the geometric information is used to match the workpiece point cloud to achieve 6D pose estimation.3.Robot motion planning.In order to make full use of the feature that the processing tool can rotate around the axis,tool space discretization combined with graph search is used to generate the processing path.For the movement of the robot before and after processing to and from the photographing posture,a planning algorithm based on robot free configuration space estimation is proposed to quickly obtain the results.