A Modified Approach to Sensitivity Amplification Control to Handle Uncertainties

Abstract

Recently, different exoskeleton devices have been developed to provide required mechanical power in augmentation and rehabilitation applications. The design of an appropriate control is one of the challenges in the field of exoskeletons which have attracted extensive attention in the past years. One of the noticeable approaches proposed for the control of augmentation exoskeletons is sensitivity amplification control (SAC), which presents remarkable merits such as no need for interaction force measurement. However, its performance is highly susceptible to the presence of an uncertainty, due to its high model dependency. In this paper, an intermediary control approach as a potential solution is presented to preserve the control performance considerably in the presence of an uncertainty and to take the advantage of the SAC controller. The control scheme is composed of two phases: training and utilization. In the training phase, it is assumed that the measurement of the interaction forces between the user and the robot is temporarily available by using additional force sensors. Therefore, the controller is designed in the form of a robust admittance controller accompanied by an LWPR network as a universal approximator to estimate the uncertainty. In the next phase, the controller is modified to solve the constraint of measuring the interaction forces and simulate the SAC control method in the presence of the uncertainty. The convergence of the closed-loop system is evaluated by the Lyapunov theorem in both phases. Moreover, based on a simulation on a swing leg, the effectiveness of the proposed control strategy is investigated. The results show significant improvement in exhibiting the desired dynamics in the presence of an uncertainty, compared to the classical SAC controller.