Tracking control of knee exoskeleton with time-varying model coefficients under compliant interaction

Abstract

Knee exoskeletons can help injured subjects regain locomotion ability by providing external movement compensation as a robot-assisted rehabilitation technique. Tracking control of joint angle of knee exoskeletons often encounters time-dependent (time-varying) model parameter issues, which would dramatically put an influence up on dynamic behaviors. In a number of applications, inertial and viscous parameters of knee exoskeletons are estimated to be time-dependent (time-varying) due to unexpected mechanical vibrations and contact interactions. Moreover, to achieve adaptively compliant interaction torque between the human and the robot will contribute to have an positive effect on comfortable experience on wearers. However, All of these may increase difficultly of accurate control of knee exoskeleton to follow desired joint angle trajectories. This paper proposes a novel control strategy for controlling knee exoskeleton with time-dependent (time-varying) inertial and viscous coefficients, with disturbance interaction torque from pilots considered as well. Such controller is designed based on Zhang dynamics (ZD) method and utilizes twice Zhang function (ZF) so as to make the tracking error of joint angle of knee exoskeletons exponentially converge to zero. Meanwhile, such ZD based method is robust to guarantee the tracking error bounded when the gap between the disturbance torque and the exoskeleton torque always exists. Illustrative examples are presented to show efficiency of this type of controller based on ZD method. Comparisons with gradient dynamic (GD) approach are also presented to demonstrate efficiency and superiority of ZD-type control strategy for tracking joint angle of knee exoskeleton.