Real-time thumb-tip force predictions from noninvasive biosignals and biomechanical models

Abstract

The opposable thumb has allowed humans to develop accurate fine motor skills and has provided the same level of functionality in artificial prostheses or similar robotic hands. This study describes a real-time isometric thumb-tip force prediction that uses a biomechanical muscle model and surface electromyography signals under four different angle configurations. Of the nine muscles that contribute to the thumb-tip force, the activities of five muscles were measured, and the activities of four muscles were inferred based on measured muscle data. The force exerted by each individual muscle was computed using a Hill-based muscle model. The thumb-tip force in the palmar direction was then estimated based on the contributing ratio of each muscle. The results indicated a high correlation between the thumb-tip force predictions from the model and the measured data. The possible applications of this research include the control of finger-tip forces from noninvasive neurosignals in hand prostheses.