Optimized Kalman filtering in dynamical environments for thumb robot motion estimation

Abstract

Stroke, a prevalent nerve disorder in Indonesia, necessitates post-stroke rehabilitation like physical and occupational therapy. Hand and finger muscle training, crucial for restoring movement, often involves innovative solutions like finger prosthetic robotics arms. In particular, the advancement in thumb robotics emphasizes the estimation of thumb motion, where the ensemble Kalman filter square root (EnKF-SR) and H-infinity methods are deemed dependable for both linear and nonlinear models. Simulation results, using 400 ensembles, demonstrated nearly identical accuracy between the methods, exceeding 99%, with a 6-7% increase in accuracy compared to 200 ensembles. These advancements offer promising prospects for effective post-stroke rehabilitation and improved thumb movement restoration.