Design Of Master-Slave Control Algorithm And Its Application In A Human-Powered Augmentation Lower Exoskeleton

Human-powered augmentation lower exoskeleton is a wearable smart device, which combines human intelligence with machine systems physical strength cleverly, and the wearer can complete the task that can’t be accomplished alone, which enhances the body’s athletic ability and reduces the work load. Nowadays, exoskeleton robot has been widely-utilized in the military, emergency relief and other fields. Moreover, it also has a broad application prospects in the field of helping the aged and the disabled. But it seems that there is no better solution for the problems of the dynamic stability, the tracking error and the interaction force between the robot and the human body in the human-powered augmentation lower exoskeleton. In terms of this issue, the thesis presented a new control method for getting command of the human-powered augmentation lower exoskeleton, based on the research of the robot’s force-position control method. The research contents of this thesis are as follows:Human-powered augmentation lower exoskeleton is worn on a person’s body, and the performance of the controler has a great influence on wearer’s experience and safety. In the human-powered augmentation lower exoskeleton control system, we need master the boundary between the force and the position. In this thesis, by studying robotic force-position control method, we determined to adopt the impedance control method, and introduced the basic principles of impedance control.In order to implement the control algorithm better, the thesis analyzed and established the inverse dynamics model of the system. Exoskeleton robot system is a wearable humanoid robot with high degree of freedom, whose kinematic model is very difficult to be established. To solve this problem, this thesis divided human walking posture into two phases, the swing phase and the stance phase, and they were respectively equivalent to a two-link mechanism and a three-link mechanism, and created the relationship between the joint torque, the joint position, the joint velocity and the joint acceleration based on the Lagrange equation. And based on the mathematical model of the system, the simulation results show the performance of the impedance control.Since application environment of human-powered augmentation lower exoskeleton is complex and the mathematical model of the system is difficult to obtain, the conventional impedance control method is not applicable. This thesis presented a view to improve the impedance control by using the reinforcement learning method to optimise the impedance control parameters online. Based on the view, the thesis presented a variable impedance control with PI2(Policy Improvement with Path Integrals) that is applicable to the high dimentional system, and applied it to the Human-powered augmentation lower exoskeleton. This method can work only by adding the reinforcement learning parameter adjustment loop in the outer loop of the controller without modifying the existed impedance controller. Finally, we verified this method’s control performance through simulation experiments.