Design of a novel Assist-As-Needed controller for gait rehabilitation using a cable-driven robot

Abstract

In this paper design of a novel assist-as-needed controller for a fully-constrained cable driven robot is addressed. The robot is going to be used as a part of a gait training system. Robotic training systems have numerous advantageous over traditional physical therapy, but using common controllers (such as CTM and SMC) cannot fulfil all of expectations of an effective training system. Precise gait pattern tracking may cause unnatural feeling for the patient, while a careless control action may lead an ineffective training. Therefore, the ability to assist a patient to follow a desired movement, and the ability to provide only the minimum necessary assistance seems to be crucial for this type of systems. In this paper a robust design of an adaptive Assist-As-Needed controller for a cable robot is studied. Robustifying terms are added to the controller to deal with structured and unstructured uncertainties in dynamics of the robot. At the end, the controller's performance is validated with simulation results.