Wearable-based pedestrian inertial navigation with constraints based on biomechanical models

Abstract

Our aim in this paper is to analyze inertial navigation systems (INSs) from the biomechanical point of view. We wanted to improve the performance of a thigh INS by applying biomechanical constraints. To that end, we propose a biomechanical model of the leg. The latter establishes a relationship between the orientation of the thigh INS and the kinematic motion of the leg. This relationship allows to observe the effect that the orientation errors have in the expected motion of the leg. We observe that the errors in the orientation estimation of an INS translate into incoherent human motion. Based on this analysis, we proposed a modified thigh INS to integrate biomechanical constraints. The results show that the proposed system outperforms the thigh INS in 50% regarding distance error and 32% regarding orientation error. Our approach allows us to improve the performance of INS using solely inertial measurements and biomechanical constraints. Additionally, it allows us to state that the performance of an INS can be improved by correcting the incoherent orientations.