Prior Kinematic Information Fusion for Pedestrian Localization With Toe-Heel-Shank MIMUs

Abstract

With the development of the micro-electronic mechanical system (MEMS) inertial measurement units (IMU), the foot-mounted pedestrian inertial navigation system (PINS) based on the zero velocity update (ZUPT) is widely used. The systematic errors of MEMS IMU cause low precision of PINS at swing phase. However, the bio-mechanical constraints, which can be used to suppress the divergence of PINS, exist in lower limbs. In this study, three magnetometer and IMUs are fixed on the toe, heel and shank, respectively. The attitude of each segment is estimated by a MAHONY filter. The gait is defined as four successive phases. For each phase, the velocity and position constraints are constructed based on dynamic geometric constraint and Coriolis theory. Based on the perfect measurement method, the kinematic constraints, which are calculated by inertial data and attitude, are added into the observed equations. The noises are considered as additive Gaussian white noise. The linear Kalman filter is set to iteratively estimate the optimal positioning result. Several experiments with different walking velocities and trajectories are performed. The results show that the algorithm which is proposed in this study, greatly suppresses the velocity and positioning error.