Design, biomechanical modelling, and assist-as-needed control of a shoulder rehabilitation robot

Abstract

Neurological disorders such as stroke is one of the leading causes of upper limb impairment in humans. Task-oriented and repetitive movements can strengthen muscles and aid in improving the functional range of motion of the patients. Traditional physical therapy is labour intensive and lacks the objective assessment of the patient’s improvement. Over the years the field of rehabilitation has undergone substantial transformation. Rehabilitation robots have been proposed by the researchers as a replacement for manual physical therapy. These robots have not only shown promising results in improving the functional range of motion of the upper limb of the patients but also provide objective assessment of the improvement while easing the physical workload of the therapists.
The upper limb rehabilitation robots, by definition, work in proximity with the human subject thus the physical interaction is certain. So, despite their enormous advantages it is challenging to model and control these devices due to high nonlinear and uncertain dynamics. The information regarding the musculoskeletal parameters of the joints is essential for efficiently controlling this physical interaction between subject and robot. In the human upper limb, shoulder joint is the most complex and provides a broad range of motions. Analysing the biomechanics of the human shoulder is a paramount in efficiently controlling the upper limb rehabilitation robots. Existing in-silico musculoskeletal models can be used to study the associated kinematic parameters. However, to save the time consumed in the control related computations, normally these musculoskeletal models are required to be simplified compromising their versatility. In this thesis presents a parallel mechanism based Virtual Biomechanical Shoulder Robot Model (VBSRM) is presented. The kinematic model of the proposed VBSRM is developed and validated by a case study. A singularity analysis of the robot’s Jacobian matrix was carried out to avoid a possible singular VBSRM design. Wrench analysis and structural design analysis were conducted to obtain actuator forces and dynamic stiffness behaviour of the VBSRM respectively. Initially, only condition number of the robot mechanism is optimized using Genetic Algorithm and performance objectives from the optimal design are analysed. Later, the three objectives are grouped to form a single function, and a single objective-based optimization is also conducted. However, further investigation revealed the conflicting nature of the objectives and hence these were simultaneously optimized using the Non-dominated Sorting Genetic Algorithm (NSGA II). The results obtained from various optimization routines are compared and it is found that the results from the NSGA II provide a better trade-off between the performance objectives. This optimal design is than used to evaluate the workspace of the proposed VBSRM. The motion trajectories from the optimal design of the VBSRM are later analysed vis-à-vis human shoulder motions for its intended use as a robotic model of the human shoulder joint in various applications.
An upper limb rehabilitation robot is than designed and developed to aid in the rehabilitation of upper limb disabilities. The proposed upper limb rehabilitation robot (ULRR) is designed to improve the transparency, incorporating shoulder, elbow, and wrist joints. The proposed ULRR provides six actuated degrees of freedom (DOF) out of which three are associated with the shoulder joint, one with the elbow joint, and two DOF for the wrist joint. The telescopic features at the upper and lower arm of the ULRR allow a close alignment between the elbow joint of the robot and the user. To evaluate the transparency of the ULRR, two comparative studies were conducted with ten healthy subjects. Firstly, the tests were conducted for the motions of the shoulder joint only in all three DOF individually. Later, a trajectory from the active daily life (ADL) was given as a reference and the ULRR was operated in active and passive modes. The force data collected during these experiments was analysed to assess the transparency of the ULRR for human-robot interaction.
The inverse kinematics problem in serially manipulated upper limb rehabilitation robots implies the usage of the end-effector position to obtain the joint rotation angles. In contrast to the forward kinematics, there are no systematic approaches for solving the inverse kinematics problem. Furthermore, for some morphologies of the upper limb rehabilitation robots, the inverse kinematics problem is particularly challenging to solve. Conventional methods to solve the inverse kinematics problem reported in the literature are computationally expensive. In the present work, a deep learning-based model is proposed to acquire the joint angles for a given end-effector position. The proposed approach exhibits high efficacy in determining the joint angles for various target positions and can accurately predict the end-effector positions once trained, improving the ability of the upper limb rehabilitation robot to adapt to varying patient needs. Due to its improved capability and effectiveness to track positions, the proposed algorithm lays the foundation for the development of efficient controllers for ULRR. Human-robot interaction during rehabilitation demands precise control over movement dynamics to ensure smooth, adaptive, and effective assistance. This study investigates the coupled dynamics model of the human-robot by computing joint stiffness, damping, and interaction force. In comparison to the conventional mathematical models that often assume these dynamic parameters to be constant, this study proposed a Reinforcement Learning (RL) based framework for estimating stiffness, damping, and interaction force. Using dataset comprising joint angles and torques, RL models learn the viscoelastic properties of human-robot interaction by iteratively adjusting stiffness and damping based on movement variations. Results indicate that the model successfully adapts these dynamic parameters to achieve stable and coordinated movement while ensuring patient safety. This adaptation is particularly beneficial in rehabilitation applications, where patient effort and movement variability require continuous modulation of robot-assisted forces.
Further, given the varied level of disability among the patients, the rehabilitation robot understudy incorporates a patient cooperative control strategy. A novel assist-as-needed control is designed for this purpose that integrates the VBSRM designed in this study with the ULRR. This novel approach enables the robot to dynamically adjust the assistive force during rehabilitation training. The proposed control scheme was tested with ten healthy subjects in two distinct control modes “active” and “passive”. In active mode the robot actively follows the human movement while providing minimal assistance while in passive mode the robot assists the movement on the predefined trajectories. These results indicate that individually tailored robotic assistance according to individual patient-specific needs can enhance their participation, which is essential to optimize treatment outcomes.
The essential contribution of this research are design and development of a Virtual Biomechanical Shoulder Robot Model (VBSRM). It advances the field of musculoskeletal modelling for human shoulder joint. The deep learning model presented for estimating the joint angles for the given end effector positions advances the field of kinematic analysis of the upper limb rehabilitation robots. An in-depth transparency evaluation of the proposed device contributes significantly towards the design and development of an adaptable, and efficient upper limb robotic rehabilitation system. In addition, the use of assist-as-needed control based on the VBSRM promotes active patient cooperation by modulating the assistive force as a function of the degree of the patient's disability. These advances contribute considerably to the design and development of robotic systems for upper limb rehabilitation.

Date of Award	2025
Original language	English
Supervisor	Shahid HUSSAIN (Supervisor) & Girija CHETTY (Supervisor)