Multi-modal control scheme for rehabilitation robotic exoskeletons

Abstract

In the past few decades, a variety of rehabilitation robotic exoskeletons have been developed for patients with stroke and traumatic brain injury, which can assist therapists and potentially improve the functional outcomes. Many robotic exoskeleton systems adopted series elastic actuators, which are known to offer a number of advantages over rigid actuators, such as high force/torque fidelity, low output impedance, intrinsic compliance, and tolerance to shocks. While several control schemes have been proposed for robotic exoskeletons driven by series elastic actuators, existing methods are limited at high level, by assuming a perfect inner control loop. Due to the nonlinearity in robot dynamics, changing interaction forces, etc., the stability of the overall system consisting of both the robot dynamics and the actuator dynamics may not be guaranteed. This paper presents a multi-modal control scheme for rehabilitation robotic exoskeletons. Three control modes are smoothly integrated into the controller, where the robot-assisted mode allows the human to exert voluntary efforts within a desired region, the robot-dominant mode is activated to correct the movement of the human when the robot is outside the region, and the safety-stop mode is to stop the robot whenever the interaction force is too large which may result in injuries. By using the regional position and force feedback, the proposed controller achieves the paradigm of “assist-as-needed” and also guarantees the safety of the human. The development of the proposed controller follows the singular perturbation approach, and the stability of the overall system is rigorously proved by using Tikhonov’s theorem. Experimental results in both upper-limb and lower-limb exoskeleton systems are presented to demonstrate the effectiveness of the proposed control method.