| Popis: |
The human wrist is to the hand, as what the pen is to the poet. The latter cannot work without the former. The wrist not only serves to orientate the hand before a grasp, but also to stabilize it during a grasp. These two functions of the upper limb are inseparable, and a disorder of the wrist joint negatively affects dexterity and grip strength of the hand. Upper limb paresis is the most common impairment following neurological disorders such as stroke, and affects more than 3'700 individuals in Switzerland each year. Stroke survivors often suffer from abnormal muscle tone such as spasticity, tremors, and pain, which affects wrist function and negatively impacts independence and quality of life. The rehabilitation of these functions is possible during conventional therapy and can be enhanced through high dose and intensive movement-based training delivered by robotic systems. Robot-assisted therapy promotes active participation combined with proprioceptive feedback that reinforces motor learning and somatosensory recovery. By quantitatively assessing the recovery and providing a motivating environment, robot-assisted therapy stands as an adequate candidate to supplement conventional therapy. Nevertheless, therapy administered via robotic devices remains in the minority of treatments and many patients after discharge from the hospital suffer from persistent wrist and hand impairments. Therefore, novel and accessible technologies that empower the patient to self-initiate and continue rehabilitation training must be developed and made commonplace. Home-based rehabilitation using robotic technologies is a promising and growing field that has triggered the development of many devices. In addition to promoting independent rehabilitation training, powered wearable devices have the potential to provide assistance during functional everyday tasks. However, besides meeting the requirements for supporting a given motor function, the development of such solutions must strike a balance between functionality, usability and wearability. In particular, the ease to mount and unmount (don and doff) is an essential aspect that has so far been rarely addressed in many projects targeting home-based therapy. The aim of this thesis was to develop, characterize and evaluate a fully wearable wrist exoskeleton - the eWrist - that actively supports extension and flexion movements. Envisioned as a tool for assistance during daily tasks, the development focused on usability and wearability of the device. Furthermore, this thesis aimed at implementing a robust and intuitive control scheme on a wearable exoskeleton that promotes voluntary effort using physiological signals. To achieve this goal, existing technologies targeting the upper limb were reviewed and requirements for a wearable wrist exoskeleton were determined. Weight, size, actuation torque, angular velocity, range of motion, and most importantly ease of implementation were aspects considered when choosing an appropriate transmission type that meets the requirements. With the goal of building a device for independent use, a mechanism to don the eWrist with a single hand was implemented. Moreover, the development of the first prototype and subsequent iterations prioritized the selection of widespread and affordable components, and the use of 3D printing techniques and open-source software that would facilitate potential integration into maker communities. The non-backdrivability of the transmission imposed the implementation of an admittance control scheme that allowed smooth and stable interactions between the user and the robot. To investigate the feasibility of intuitive control promoting voluntary effort, an sEMG-based controller was implemented and evaluated on a single healthy subject. The results showed that the fastening system enabled quick and easy donning and doffing, and a firm attachment to the forearm and hand. Moreover, the sEMG controller proved to drive the assistance support in accordance with the intention of the user. To further improve functionality and wearability, a new iteration of the eWrist was characterized and evaluated in fifteen healthy participants and two stroke survivors. Shortcomings of the previous iteration were addressed by: reducing weight and physical profile, increasing durability, improving interaction with the device, and further improving the donning procedure. A novel fastening system including electronics and battery was developed that enabled donning of the entire exoskeleton using one hand. Standardized human-robot interaction metrics and impedance planes were used to characterize and evaluate the various behaviours that can render the device. Based on the established requirements, the developed solution fulfilled or even outperformed expectations. The time required to mount the eWrist revealed that after a few practice trials participants could don it independently in about 1 min. In addition, standardized usability questionnaires completed by the participants showed that they all embraced the device and found its attachment system efficient and simple to use. The non-backdrivability of the transmission combined with a stiffening of the wrist joint generates instabilities in the physical human-robot interaction (pHRI) that were assessed in a goal-directed visuomotor task. A variable admittance control scheme was implemented to detect and dampen these disturbances, and was evaluated in ten healthy participants and six stroke survivors performing the task. In addition, an improved sEMG-based controller, together with a gravity compensation controller were implemented to promote voluntary effort and support wrist weakness. The stability and transparency of the pHRI, characterized by metrics such as jerk, interaction force, and angular velocity/acceleration, was used to assess the effectiveness of the variable admittance scheme. In the context of the visuomotor task, the variable admittance controller proved to significantly reduce instabilities in the human-robot interaction with healthy participants. Additionally, both controllers could enhance wrist functionality of stroke survivors, especially in the most extreme angular positions and more impaired patients. After many iterations, the latest version of the eWrist exoskeleton has resulted in a solution that combines lightweight, low physical profile, ease of donning, and intuitive control to support extension and flexion wrist function in patients with neuromotor impairment. Furthermore, the portability of the eWrist makes it suitable for deployment in various environments whether in a clinic or at an individual's home. Finally, thanks to a focus on accessibility and simplicity throughout the design process, the eWrist meets an optimal trade-off between complexity and functionality to increase access to affordable orthoses for stroke rehabilitation. |
