Published Paper Details:

Stroke ranks as the second leading cause of death in India, with approximately 1,85,000 cases occurring annually. A common consequence of stroke is hand paralysis, necessitating effective rehabilitation strategies such as Functional Electrical Stimulation (FES) for hand prehension. Existing commercial FES devices often pose usability challenges due to manual operation via touchpad or buttons. In this study, we propose a novel deep learning-based approach using the YOLOv8 algorithm for object detection to facilitate multiple grasp types and control an electrical stimulator for assisting grasping. Our algorithm tracks hand and object positions, selecting appropriate grasp types based on proximity to the hand. Upon selection, an ESP32 BLE device, which imitates specific behaviors of a custom FES stimulator, executed predefined stimulation sequences to enable the user to grasp nearby objects. This contactless, vision-based solution holds promise for home-based stroke rehabilitation by eliminating setup complexities and reducing dependence on caregivers. Future research will integrate this model with FES devices, evaluating its efficacy in restoring hand movements among stroke patients.