Force Estimation of Five Fingers Using Infrared Optical Sensors and an IMU and Its Application to Analysis of Sports Motion

Abstract

Research on scientifically analyzing human movements for applications in sports and skill inheritance is actively conducted. Particularly, methods for estimating human output through the analysis of muscle deformation can be applied to static forces like grip strength. However, when focusing on finger force analysis through muscle deformation, traditional studies have only targeted a single finger, and there are no examples of simultaneously estimating the force of all five fingers. Furthermore, the most common method for acquiring muscle deformation, using electromyography (EMG), suffers from reduced accuracy due to the influence of sweat and sebum on the skin, making it challenging to analyze intense movements. In addition, as the posture of the arm changes, the muscle configuration within the arm also changes, making finger force estimation difficult with conventional methods when there is significant arm posture variation. Therefore, this study aims to simultaneously estimate the force of all five fingers under varying arm postures by using optical sensors, which are less affected by changes in skin condition, to measure muscle deformation, and a six-axis inertial sensor (IMU) to measure the posture of the upper arm. By using the IMU to detect the posture of the upper arm, it is possible to indirectly estimate the changes in muscle configuration within the arm. In the experiments, the accuracy of finger force estimation was compared with and without the use of the IMU, focusing on sports movements, to discuss its effectiveness. In addition, the study demonstrated how the accuracy of force estimation decreases by applying saline solution to the skin to simulate sweat. The results showed that the use of the IMU improved the accuracy of finger force estimation, and although the accuracy decreased due to sweat, force estimation remained possible. This method, which involves attaching sensors only to the upper arm, does not interfere with hand operations, suggesting its potential application for analyzing fingertip forces in various scenarios that involve the use of hands.