Learning-based upper limb robotic rehabilitation

Abstract

In this doctoral thesis, several advancements are presented to enhance the capabilities of a seven DoF robotic exoskeleton designed for upper-limb rehabilitation. First and foremost, a human-like inverse kinematic solution, underpinned by a machine-learning technique, is introduced. This solution paves the way for generating natural and comfortable upper-limb postures. Unlike conventional methods, this approach provides a practical and efficient means of finding appropriate inverse kinematic solutions for upper-limb rehabilitation via redundant robotic exoskeletons. The core advantage lies in its real-time capabilities. Additionally, a robust learning-based controller is developed to tackle uncertainties and disturbances that are inherent in the patient-robot interaction. This includes addressing unknown forces and ensuring compliance with predefined input and state constraints. By doing so, the controller offers functional safety and optimal performance during rehabilitation exercises. The last development in this thesis is the design of a mirror rehabilitation system. This system is devised with the aim of enhancing the motor skills of individuals with hemiplegia through brain neuroplasticity stimulation. The approach involves providing visual feedback and proprioceptive stimulation while both arms move symmetrically and simultaneously. This methodology represents a significant leap forward in rehabilitation, harnessing the power of neuroplasticity to drive recovery. Furthermore, these advancements cater to both passive and active rehabilitation modes, allowing for a flexible and tailored approach to recovery. In the passive mode, the patient can relax while the exoskeleton guides the arm through a predefined trajectory. In the active mode, the wearer gains the independence to initiate movements and complete desired tasks without external assistance, empowering them to take charge of their rehabilitation journey. To validate the efficacy and real-world applicability of these advancements, a series of real-time experiments have been conducted. The results of these experiments were documented, submitted and/or published in several journals, contributing to the scientific community and the field of rehabilitation robotics. The advancements showcased herein open new horizons for enhancing the quality of life and recovery prospects for individuals in need of rehabilitation, setting the stage for a promising future in the field of robotic exoskeleton-assisted rehabilitation.

Date	14 Mar 2024
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Maarouf Saad (Supervisor) & Cristobal Ochoa Luna (Co-supervisor)