Research On Robot Weld Grinding Path Planning And Control Technology Based On 3D Vision

In the manufacturing industry,the last process of some product processing often requires polishing of the workpiece.At present,some grinding operations are still mainly completed manually,and manual grinding has problems such as poor working environment and poor grinding consistency.With the development of industrial robot technology,the use of industrial robots for grinding operations is becoming the industry trend.At present,robot polishing is mainly based on teaching programming and offline programming software.For some workpieces with complex grinding paths,teaching programming needs to teach many path points,resulting in low grinding efficiency.For some non-obtainable models and workpieces with large difference between actual size and model,offline programming software also fails accordingly.In order to solve the problem of automatic planning of robot grinding path and improve the efficiency of robot grinding operation,this paper focuses on the autonomous planning and control of robot weld grinding path based on 3D vision technology,taking the grinding of weld reinforcement on curved surface as the research object.A grinding system of industrial robot combined with force control actuator is constructed.A grinding contact force control algorithm based on double-loop feedback control is developed to realize the control of contact force in the grinding process.At the same time,a grinding tool gravity compensation strategy is given to realize real-time compensation of tool gravity.The hand-eye calibration and recognition error compensation of 3D camera are studied.A hand-eye calibration method based on point set fitting is adopted,which effectively improves the accuracy of hand-eye calibration.An error compensation method based on the depth measurement of 3D camera is proposed to ensure the recognition accuracy of 3D vision system.The grinding path planning technology based on 3D point cloud of workpiece is studied.A set of point cloud preprocessing methods is given to realize the functions of region of interest selection,downsampling,smoothing and filtering of the point cloud.Aiming at the problem that the tangent plane needs to be specified manually in the point cloud slicing method,a method to determine tangent planes is proposed,and a series of tangent planes are automatically determined.This paper introduces a method of extracting weld feature points based on the idea of point cloud slicing and fitting deviation,which can accurately locate the weld center feature points.A weld feature point fitting method based on the least square method and the accumulated chord length parameter method is proposed,which realizes the fitting of 3D weld feature points and forms a continuous weld grinding path.A method of robot grinding tool posture planning and adjustment is presented,which improves the consistency and continuity of grinding tool posture between path points,reduces the vibration that may occur during grinding,and avoids potential mechanical interference.Based on the simplified material removal model and weld reinforcement,a control scheme of robot round-trip grinding times and grinding contact force is introduced.Based on the idea of "replacing curve with straight",an equal step path discretization algorithm based on chord error is presented to ensure the accuracy of robot grinding path execution.The grinding path planning and control algorithm in this paper is verified in practice,the accuracy,rationality,robustness and efficiency of the grinding path planning algorithm are verified,the contact force control response accuracy of the grinding execution system is tested,the weld grinding path planning software and the robot grinding execution system control software are developed,and the automatic programming of the weld grinding path and the control of the force control actuator in the grinding process are realized.The work of this paper lays the foundation for automatic grinding of weld reinforcement on curved surface,and helps to improve the automation and intelligence level of robot grinding operation.