Electromechanical Design and Control of an Upper-Limb Rehabilitation Exoskeleton

Abstract

Motor disabilities caused by neurological disorders such as stroke are a prevailing health issue in modern societies. To expedite the rehabilitation of patients with such disabilities, the scientific community has been exploring the possibility of utilizing technologically advanced methods to enhance the neuroplasticity of the brain after injury. Robotic systems have been the cornerstone of a category of such efforts, focused on increasing the intensity and diversifying the sensory feedback experienced by the patient. In particular, exoskeletons are found to be very promising for assistive and therapeutic applications due to their ability in producing controlled torque in individual joints of the paretic limb. Despite the continuous efforts of the research community, there are still major issues for large-scale adoption of exoskeleton-based rehabilitation in clinical settings.

This dissertation seeks solutions to some of the short-comings of available upper-limb rehabilitation exoskeletons, both in design level and control algorithm development. The focus of the design efforts outlined in this work is optimization and electro-mechanical embodiment of a kinematic structure, developed in our research group. Two upper-limb exoskeletons, CLEVERarm, and CURE are developed based on the same kinematic architecture, but with different design philosophies. While compactness and reduction of weight are the main design criteria of CLEVERarm, the design of CURE is focused on achieving high mechanical compliance. In addition to the innovative designs, the development of therapeutic control strategies tailored for rehabilitation applications is the second major contribution of this research work. The novel control strategy proposed in this dissertation is aimed at achieving arm posture control, without imposing explicit timing laws. The proposed method uses the nonlinear control framework to robustly impose a set of holonomic constraints on the system which describe the correct posture of the arm during a motion. The proposed controller was implemented on CLEVERarm and tested on healthy subjects. The results of the experiments confirmed the expected functionality of the controller.