Compliance Control And Peg-in-Hole Assembly Of Industrial Robot

The traditional industrial robot control system has been widely used in various industrial fields.As the task requirements become more and more complex,the high-rigidity and highprecision position-based industrial robot control system can no longer meet the needs of flexible manufacturing.In this paper,aiming at the compliant assembly problem of industrial robots in the process of peg-in-hole assembly under the uncertain position and pose of peg and hole,the active compliant assembly control algorithm of 6-DOF industrial robot based on force/torque sensor is studied to achieve peg-in-hole assembly tasks in industrial manufacturing.The main research contents of this topic include the following three aspects:1)Design the compliant peg-in-hole assembly platform for the industrial manipulators..The UR5 manipulator was taken as the experimental platform and combined with the requirements of the peg-in-hole assembly scene.The robot hardware platform was built by adopting the manipulator end-actuator module Roboti Q-G85 and the force sensor measurement module ATI Mini45.Software modules such as manipulator motion planning and control,end fixture control,force sensor data acquisition,sensor gravity compensation,and compliance control are developed and integrated into the ROS system,laying a solid foundation for compliant assembly research.2)Design a random inclination base assembly algorithm based on support vector machine.In view of the deviation of the peg-in-hole assembly caused by the nonhorizontal base during the assembly process,the peg cannot enter the hole smoothly.In this paper,a contact state judgment model based on support vector machine is designed,and the attitude adjustment of guiding the peg-in-hole assembly is aligned with the shaft hole,which solves the problem that the contact state between the workpiece and the base with random inclination is difficult to determine during the assembly process,and realizes-30°~30° The target of the peg-in-hole assembly of the base at any inclination angle within the range,the assembly success rate is 98%.3)Design a method of searching holes and jacks for manipulators based on reinforcement learning.In the hole search stage,a hole search algorithm based on the combination of reinforcement learning(Actor-Critic)and force-position hybrid control is proposed to solve the peg and hole position deviation problem in the assembly process.The problem is that the realization can use fewer steps to find the hole in the real machine experiment.In the jack stage,given the problem that the fixed impedance control parameters cannot adapt to the dynamic jack task,a jack algorithm that generates dynamic impedance control parameters in real-time using reinforcement learning(Actor-Critic)is used to complete the high-precision jack task in the simulation.The process contact force is controlled below 10 N,the torque is controlled below 0.7N·m ,and the jacking process is safe and stable.