The doctoral dissertation addresses the development of the ReGrip robotic device for the rehabilitation of finger and wrist movement, its further upgrade into two versions with more advanced control modes, and the evaluation of their performance. A distinctive feature of the developed robotic device is its parallel-serial mechanism, which, with a single actuated degree of freedom, enables simultaneous movement of the fingers and wrist in the sagittal plane. The target motion of the mechanism mimics the opening and closing of the hand, imitating the action of grasping and releasing objects. The position-controlled ReGrip device was, through several modifications, developed into two versions: the ReGrip ACA and the ReGrip VIA. Each of these, through its own approach, enables adjustment of the level of mechanical impedance. The main research objective was to develop a device for hand movement rehabilitation intended for safe and effective independent use, and to determine how the principle of varying mechanical impedance affects the impedance parameters and response of the developed devices, the modes of robot-assisted exercise and the user experience.
The first chapter provides an overview of the rehabilitation process of hand movement after stroke and presents the role of robotic rehabilitation and its advantages over conventional rehabilitation approaches. A classification of upper-limb rehabilitation devices is presented based on the type of mechanical interface with the user, type of actuation and training modalities they enable, as well as how these modalities enable progression in exercise intensity. Furthermore, commonly used control strategies are described, including impedance and admittance control, as well as the concept of series elastic actuators and variable impedance actuators. The chapter concludes with the definition of the dissertation’s objectives.
The second chapter describes the initial version of the position-controlled ReGrip device. It presents the target motion, the developed mechanism, the actuation system, and the basic control architecture that enables a passive training mode. Based on this prototype, an experimental validation was conducted with healthy volunteers, comparing hand movement with and without the device. The results showed that the ReGrip prototype satisfactorily replicates wrist and finger movement from simultaneous flexion to extension.
The third chapter describes the further upgrade of the basic version of the ReGrip device, which enabled the development and comparison of two mechanically identical versions of the ReGrip device with different principles of changing the mechanical impedance. The ReGrip ACA device actively changes the mechanical impedance of the mechanism by adjusting the gain parameters of the admittance controller, whereas in the ReGrip VIA device changing the mechanical impedance with passive elastic elements is enabled through antagonistic preloading of nonlinear springs.
The fourth chapter presents the experimental evaluation and comparison of the impedance response of the developed ReGrip devices in interaction with a dynamic environment. A collaborative robot UR5e was used to simulate the dynamic environment by imposing motion at the top of the distal finger support through a custom mechanical interface, while simultaneously measuring interaction forces. Based on the measured motion and interaction force data, we evaluated the impedance and frequency response of the devices. Both ReGrip devices exhibited similar and stable responses within the same training modalities, with comparable values of impedance parameters. At higher disturbance excitation frequencies, gradual changes in the frequency response were observed, with the ReGrip VIA device proving to be more robust.
The final chapter presents a technology acceptance study conducted with ten healthy participants and five post-stroke patients. Participants tested all four developed training modalities on both ReGrip devices in a trajectory-tracking task. User experience was evaluated using the SUS and a user satisfaction questionnaire, while tracking performance was assessed through the analysis of tracking error. The results indicate differences in tracking performance between modalities, but not between devices within the same modality, except in the passive mode. The tracking error increased in line with the difficulty of the task for both devices. Questionnaire results showed comparable and high user satisfaction, with both developed ReGrip devices rated as excellent according to the SUS questionnaire.
|
