Research On Human-robot Interaction Information Perception And Coordination Movement Control For Lower Extremity Exoskeleton

As a human-machine integrative mechanism, lower extremity exoskeleton enables the user to integrate human intelligence with the endurance and high velocity of the machine. In recent years, because of its wide application prospect, it has become one of the most attractive research focuses in robotic field. In this paper, we have de signed an innovative lower extremity exoskeleton for the purpose of enabling a heavy-loaded person walk as normal. However the high-precision measurement and prediction to human movements has long been a difficult challenge. This paper is devoted to this problem from the aspects of human-machine interactive information detection, the algorithm of movement intention estimation and the control algorithm of cooperative motion.Wearable power assistance and weight-bearing exoskeleton is required to be highly practical and can adapt to different environments. Firstly, a physical human-machine interactive method for information gathering is proposed in this paper, and then a device which combines elastomer and encoder is designed to measure the human-machine interactive force. Taking into account the requirements for the high sensitivity and stability of the device and the wearing comfort, the design of a proper elastomer is vital. We set proper parameters and selected a suitable material for the elastomer and optimized the elastomer by using the FEM. Considering the processing error, a calibration test is carried out to measure its stiffness. Finally, we can obtain humanmachine interactive information through the device designed.We have carried out some researches about different algorithms for the prediction of human motion intention according to the acquired human-machine interactive information. Considering the data delay in this measurement method, we used Kalman filter to process the signal and then the serialization process is done, after which the data lagged effect is minimized and the interactive information is acquired to a large degree. Taking into account that the mapping between the interactive information and the human motion intention is difficult to get through a mathematical method, we adopt the Gaussian Process Regressive(GPR) algorithm to get the model for human motion intention by using the off-line analysis in MATLAB. The experimental results showed that GPR has high accuracy and generalization ability.Based on the cooperative movement method using GPR intention prediction, we have proposed an innovative control strategy which combines self-adaption PID adjustment and CMAC compensation in this paper. The CMAC compensator can improve exoskeleton’s tracking accuracy for human motion intention. The results of the trajectory tracking control and simulation by using the information from the CGA show that the algorithm proposed has good tracking performance.Finally, we built the experimental platform and perform trajectory tracking experiments under multiple human body postures. The results show the feasibility of the device designed to measure human-machine interactive information and the algorithm and the control strategy are effective.