Research On The Key Technology Of Robot Polishing Based On Multi-sensor

In recent years,with the rapid development of industrial robot,automatic grinding technology has gradually become a key research direction in the field of industrial processing.For parts with complex surface shapes,manual grinding is the main method at present,which is not only time-consuming and laborious,but also inefficient.Besides,the polishing quality cannot be guaranteed and the working environment is harsh,which affects the manufacturing level of the whole industry.Due to the advantages of good flexibility,high flexibility,intelligence and low cost,automatic robot grinding technology has gradually become a research hotspot.Aiming at the problem that the surface complexity of automobile stamping parts leads to low grinding efficiency and difficult to guarantee the polishing quality,this paper designs a multi-sensor robot grinding system,which is divided into two-subsystems:the stamping parts surface defect detection system based on visual sensor and the defect grinding system based on torque sensor.The specific work includes three aspects.Firstly,calibrate the grinding system.By establishing the camera perspective projection model,the transformation relations among the world coordinate system,camera coordinate system,image coordinate system and image plane coordinate system are analyzed.By analyzing the hand-eye relationship between the manipulator and the camera,the mapping relationship between the manipulator coordinate system and the camera coordinate system is established.By analyzing the position relationship between the grinding tool and the mechanical arm,the spatial transformation relationship between them is established by using the three-point method,and the calibration of the grinding tool is completed.Secondly,aiming at the problems of complex surface backgrounds and various defect shapes of automotive stamping parts,a surface defect detection method for stamping parts was proposed by using machine vision and deep learning.By collecting surface images of stamping parts and marking defects as scratches and rust,PixelNet convolutional neural network was used to segment images of surface defects and obtain defect detection results.Thirdly,aiming at the problem that it is difficult to guarantee the polishing quality of complex surface,the force/position control strategy is adopted,and the deviation of each axis of the robot is adjusted in real time based on the torque information feedback from the torque sensor,so as to maintain the constant contact force and realize constant force grinding.Experimental results show that the surface defect detection method proposed in this paper can effectively segment the surface defects of workpieces,and the accuracy(Mean IOU)of scratch and rust segmentation is 96.21%and 95.79%,respectively.The grinding force is finally controlled around 15N and the grinding effect meets the actual processing needs.