Modeling And Learning Control Research Of A Flexible Joint Robot

The harmonic gear drives are widely used in robot joints as transmission devices, which lead to the flexibility appearing in robot joints. Therefore, it is significant to have a research on flexible joint robots which have already become one of the research hotspots recently. As one of intelligent control schemes, the learning control method has received wide attention by domestic and foreign scholars. This thesis studies the flexible joint robot developed by our laboratory. A learning control method is applied to position control of the flexible joint robot for precise control of the robot link. The main research achievements of this thesis are as following:Firstly, a flexible joint robot dynamics model is established based on five assumptions. The kinematics and dynamics characteristics of the flexible joint robot are analyzed on the basis of the model. And several factors influencing the motion and positioning of the flexible joint robot are explored by simulation.Secondly, a PD controller, a PID controller, a PD plus computing angle compensation controller and a repetitive learning controller are designed based on the motor position control mode. A PD controller and a repetitive learning controller are developed based on the motor speed control mode. These controllers are proved to be feasible by simulation. Comparison of the simulation results of all the control schemes shows the superiority of repetitive learning control.Thirdly, the hardware and software of the flexible joint robot drive and control system are designed and tested, which make the robot system have the abilities of driving, sensing, controlling and data collecting. The hardware and software also promise that the position control of the flexible joint robot can be realized by the giving control schemes.Finally, the experimental platform of the flexible joint robot is established. Several factors influencing the motion and positioning of the flexible joint robot are explored by experiments. And the effectiveness of the PD control method and the repetitive learning control method which are based on motor speed control mode are proved by experiments. The differences between theory and reality of the learning control method are explored by comparing the experiment results with the simulation results.