Research On Robot Grinding System Based On Machine Vision And Six-dimensional Force Sensor

With the proposal of "Made in China 2025" strategy,China is in an important transition period from a big industrial country to a powerful one.With the tide of the Internet era,the manufacturing industry has higher and higher requirements for automation and intelligence of production equipment.With the rapid development of sensor technology,artificial intelligence,motion control algorithm and other advanced technologies,robots,as an important intelligent equipment to replace manual work,are increasingly applied in advanced manufacturing industry,greatly improving the efficiency and quality of industrial production.In view of the current situation of China’s polishing industry,this paper designs an intelligent polishing system based on machine vision and six-dimensional force sensor for the polishing of small workpiece from the perspective of practical application.This system is combined with remote operation technology,deep learning,six-dimensional force sensor and motion control technology.In this case,the teaching polishing,following polishing and selfpolishing of the workpiece are realized by means of multi-sensor fusion.The main research of this paper are as follow:(1)A teleoperation teach grinding force feedback system is built,using the force feedback hand controller as the master side,teleoperation control of mechanical arm.A six-dimensional force sensor is installed at the end of the robot arm to sense the contact force with the environment and feed the contact force back to the operator through the hand controller.In this way,operator can feel the real contact force in the polishing process,realizing the force feedback remote operation teaching grinding.(2)According to the polishing requirements,a six-dimensional force sensor was designed,including mechanical structure design,hardware circuit design and acquisition system design.The collected data are filtered and gravitational compensated because of the high-frequency vibration during polishing and the influence of tool dead weight on the sensor.In this case,the force sensor can obtain the real grinding contact force.(3)The vision system is built,the real coordinates of the image are determined and the imaging distortion is corrected according to the camera calibration results.The defects of the workpiece are identified and segmented with the help of Mask R-CNN network and an algorithm is proposed based on the Halcon to measure the size of workpiece.(4)A control algorithm is designed to follow the teaching position curve.In the nonpolishing stage,PID control is adopted to follow the position.And in the polishing stage,position impedance control is adopted combined with the force data collected by the six-dimensional force sensor.(5)Experiments are designed to verify the following-polishing control algorithm and observe the actual polishing effect.On this basis,the strategy of self-polishing movement is designed to realize the self-polishing of simple workpiece without teaching position.