Task-based exoskeleton synthesis and patient-centered assist-as-needed control strategy for an effective upper-arm rehabilitation
The purpose of this study is to enhance the quality of robotic-assisted physical therapy for patients with stroke and age-related neuromuscular conditions by creating a reliable and measurable therapeutic method that considers the variations in biomechanical and bio-signal parameters. The proposed approach includes using bio-inspired upper-arm rehabilitation exoskeletons equipped with force and inertial sensors, as well as patient-centered machine learning algorithms that provide an assist-as-needed support and recovery tracking system. The research has four significant contributions: (1) developing a methodology to capture and represent the effective workspace of an upper limb for rehabilitation: one of the research challenges in human-exoskeleton interaction alignment is modeling the shoulder joint and quantifying its axes moments. Utilizing motion capture data and Principal Component Analysis on the glenohumeral joint workspace, the finding indicates that up to 95% of the data set remains on a planar surface; (2) robot synthesis techniques to address alignment and fitting of an exoskeleton: New task-based mechanism was synthesized based on motion capture data while subjects perform elbow flexion-extension task. The mechanism was tested and compared with a joint-based exoskeleton, and the results suggest that a task-based exoskeleton has better fitting and alignment; (3) subject-centered assist-as-needed (AAN) support strategy: unlike other existing AAN approaches, the proposed method does not require any muscle activity or active range of motions (ROM) to initiate the customized model for each person; it also does not need a predefined model based on other subjects data. The feasibility of the technology has been tested with patients with upper-arm disabilities with an accuracy rate of 91.22%; and (4) provide real-time visual feedback to patients: a graphical user interface was developed to furnish the user with real-time feedback in the form of muscle effort and range of motion to keep them motivated while maximizing rehabilitative progress. Overall, this study is expected to make a significant contribution to the mobility, health, and well-being of individuals with disabilities while establishing a new set of fundamental knowledge on human limb motion and effort allocation.