Control of knee rehabilitation robot based on combination of backstepping and admittance algorithms

In this study, the control problem of a knee rehabilitation robot is examined. The main drawback of rehabilitation facilities, such as continuous passive motion, is the lack of feedback from the interaction force between the robot and patient leg. This means that if during the exercises, an unvoluntary motion by patient is generated, the increased interaction force can then damage the patient leg. The interaction force is increased because the robot tries to hold the patient leg along the prescribed reference path. In the current paper, to realize the compliant behavior of the robot, the concept of admittance along with two control methods including adaptive model reference and integral backstepping will be utilized. Adopting admittance control method, the robot will deviate the prescribed path so that the interaction force can be decreased. The obtained simulation results reveal the good performance of the robot even in the presence of noisy sensory data. Additionally, it has been shown that the proposed combined admittance and backstepping controller has better performance, in terms of tracking error and decrease of interaction force, as compared with the model adaptive reference model.