Design, modeling, and validation of a computer-vision-enabled 3D-printed soft robotic glove for stroke telerehabilitation

Post-stroke rehabilitation for upper-limb mobility faces challenges such as limited availability, high costs, and insufficient insurance coverage. This work addresses these challenges by integrating soft robotics with computer vision. A lightweight soft robotic glove (159 g), featuring five three-dimensional (3D)-printed soft pneumatic actuators that function as fingers, is designed and optimized using finite element modeling (FEM) in terms of bending behavior and force output. The fingers demonstrated durability, bending deformation repeatability, force repeatability, and reliable holding force capability. The proof-of-concept system comprises a wearable robotic glove for the patient and a computer vision interface for the therapist. The therapist performs hand gestures captured by a camera to control the glove through the computer vision interface. The soft robotic glove and integrated computer vision interface were experimentally validated on healthy participants. This validation demonstrates the feasibility of controlling the glove using computer vision and its potential for future telerehabilitation applications.