Research On Robot Load Identification And Compliance Control

The acquisition of the mass of the robot end load and the load centroid position and the active compliance control are the basis for the robot to complete high-precision tasks and ensure the safety of the environment and body in the process of human-robot interaction.At present,with the wide application of robots,the acquisition of the mass of the robot end load and the load centroid position,the stability and compliance of the system when using the drag teaching method to demonstrate the active compliance characteristics,there are very urgent research needs and profound research implications.Therefore,this thesis studies from two aspects.On the one hand,aiming at the acquisition of the mass of the end load and the load centroid position,the end payload identification based on the static teaching pose is proposed.This method complements the current method of indirectly calculating load mass and centroid position from current data.On the other hand,aiming at the stability and safety of the drag teaching system,this paper proposes a variable admittance control algorithm based on fuzzy model reference learning on the basis of using the admittance control algorithm to realize the drag teaching.Experiments show that the variable admittance control based on fuzzy model reference learning has better stability,security and compliance than admittance control.The specific work includes the following parts:(1)Establish a robot model according to the D-H parameter table of the Elite EC66 collaborative robot to solve the robot posture under different joint space positions.Establish the forward kinematics model of the robot,obtain the pose matrix of the end effector in the base coordinate system,use the pose matrix to derive the solution of the inverse kinematics equation,and verify the forward kinematics model and the solution of the inverse kinematics equation through MATLAB.Derive the robot’s velocity Jacobian matrix and force field Jacobian matrix.(2)Based on the Jacobian matrix in force domain,an end payload identification method based on static teaching pose is proposed,and a model between the load mass,the load centroid position and the joint torque is established by MATLAB.The calculation model between the joint torque and the current is calibrated by the force domain Jacobian matrix and the relevant information of the known load,and then the mass of the load and the position of the center of mass are calculated.(3)Using the ROKAE XB7 industrial robot as the controlled object,build the simulation model of the admittance controller on MATLAB/SIMULINK to realize the robot drag teaching.Through the simulation data obtained from the model,the function and control law of each control parameter in the admittance controller in the realization of drag teaching are studied.(4)On the basis of admittance control to realize robot drag teaching,a variable admittance control based on fuzzy model reference learning is proposed.The experimental platform was built using ROKAE XB7 industrial robot and On Robot HEX-E v2 six-dimensional force sensor.Offline training experiments and control experiments are designed to compare and analyze the effectiveness of the control method in terms of position and speed.Experiments show that the variable admittance control based on fuzzy model reference learning has better safety,flexibility and stability than the admittance control in the drag teaching process.