A Decoupling Method Based on Equivalent Mechanism Analysis for Motion Measurements of Human Multi-DOF Joints by Using Wearable Strain Sensors

Abstract

Wearable soft strain sensors are widely used in the measurement of human joint motion in the fields of medical, animation, sports, and exoskeleton. However, for the multi-degree-of-freedom (multi-DOF) joints of the human body, the measurement precision is still low due to the coupled effect that the signal of wearable sensors is from all the joint sub-motions. Here, we proposed a decoupling method to improve the precision of the wearable sensor information on multi-DOF joints. The main idea of the method is to transform the measurement into the forward kinematic analysis of an equivalent mechanism, where the deformation of the strain sensor is equivalent to a sliding pair of the mechanism. Experiments demonstrate that the proposed method has strong adaptability to different human motions: in walking, the dynamic measurement errors of ankle dorsiflexion/plantarflexion, adduction/abduction, and inversion/eversion are 0.95°, 1.10°, and 0.50° (5.28%, 6.89%, and 5.78%) respectively; in a mixed pattern of walking, stepping and squatting, the errors are 0.85°, 1.48°, and 0.57° (2.70%, 8.87%, and 6.55%) respectively. The proposed method has the advantage in measurements of multi-DOF joint motion, especially in the motion coexisting with various patterns.