| With the rapid development of information science,mechatronics,and artificial intelligence technologies,upper-limb wearable robots have been widely used in the fields of medical rehabilitation training,military individual combat,human load operation,and teleoperation as multidisciplinary bio-mechatronics integration systems.However,on one hand,due to insufficient environmental perception and user motion intention cognition,current upper-limb wearable robots usually fail to make the timely accurate judgment and effective decision-making on the external environment and user’s motion intention,resulting in delayed motion and errors.On the other hand,for highly nonlinear human-robot strongly coupled systems including human-upper-limb wearable robots,although some human-in-the-loop closed-loop control strategies have achieved good interaction effects and performance,it is difficult to effectively realize humanrobot tightly coupled coordinated control among robot-human-environment,where the tasks are complex and dynamic and there are many unknown external disturbances and uncertainties,causing the challenge of mismatched physical human-robot interaction and unitary human-robot interaction mode.Therefore,it is of great scientific significance and application value to enhance the cognitive intelligence of the wearable robot to the motion intention of the operator,to enhance the hybrid intelligence of humanrobot cooperation under multi-task,and to improve the ability of the robot to resist uncertainty and the robustness of control.Given the above problems and challenges,an upper limb wearable robot collaborative control system framework for human-robot hybrid intelligence has been developed,which is based on the diversity of human-robot interaction and the system optimization design of robots.In this framework,the recognition methods of the operator’s motion intention are researched,broad human-robot safe interaction models are further constructed,and human-in-the-loop control strategies of human-robot strong coupling closed-loop system are designed.The main results are listed as follows:1.Aiming at the problem of continuous motion intention recognition and closedloop control of human-robot nonlinear dynamics during human-robot interaction,a muscle-synergy-based adaptive active control scheme is proposed,which realizes continuous recognition of human motion intention and cooperative motion of upper-limb prosthetic robot.Firstly,a muscle-synergy model based on surface electromyography(sEMG)is proposed.The complex limb movement information is modeled as the motion primitive information of low-dimensional muscle group,and the human intention is decoded continuously by using sEMG signals that are extracted muscle activation and constructed the synergy basis.A novel activation to force mapping(A2F)model is developed to map intention to force information driving the affected side(prosthesis robot)through sEMG signals from the healthy side of the limb.Then an adaptive controller based on neural network approximation is proposed,where the stability of the system is proved theoretically.Finally,four healthy subjects and one trans-humeral amputee of the upper extremity are tested to demonstrate that the proposed control scheme can effectively realize human-robot cooperative motion based on human motion intention.2.Aiming at the safety problems of multi-human-robot cooperative interaction and the target transfer of human-robot,an asymmetric cooperative control scheme based on the human collaborative manipulation models is proposed,which extends the cooperative skills between human beings to the human-robot cooperative tasks,realizing the asymmetric cooperation of two human-robot systems in associated tasks.Firstly,two kinds of human-human collaborative-based moving target tracking models are proposed,including the following the better model and the interpersonal goal integration model.A region function is designed to divide the human-robot operation space into the human-leading-based human region and the robot-leading-based robot region.Then,an adaptive controller based on the region-based barrier Lyapunov function is designed to handle the change of leadership roles between human and robot,ensuring that the exoskeleton can drive the human partner in the restricted region.The convergence of tracking errors during the region switch is proved by theoretical analysis.Finally,independent task experiments and associative task experiments are designed,and the trajectory tracking tasks are performed on three groups of subjects to verify the effectiveness of the controller and the cooperative manipulation models.3.Aiming at the problem of role assignment in human-robot interaction and the dependence of human motion intention estimation on complex model parameters or multiple sensors,a continuous control scheme of intention assimilation based on human virtual target is proposed,which realizes the imprecise estimation of human intention target based on the assumed human control gain using interaction force to adapt to human-robot dynamic and safe interaction.Firstly,an approach obtaining the virtual target is proposed by using human control gain parameters and interactive force,which predicts human future behavior by integrating human motion intention target.The intention assimilation strategy based on virtual target is constructed to provide continuous interaction behavior of human-exoskeleton system from assisted movement to confrontation.Then,an adaptive controller based on linear regression of dynamics parameters is designed,which enables the robot to directly deal with the problems including nonlinear system uncertainty and the unknown perturbation,and the stability of the system through theoretical analysis is guaranteed.Finally,human-robot interactive experiments are performed to verify the effectiveness of the controller and illustrate that the assimilation control method can enable the robot to assist the user,and resist the user’s original behavior at the presence of the obstacle for the safe interaction. |
PDF Full Text Request