Pedestrian Dead-Reckoning Algorithms For Dual Foot-Mounted Inertial Sensors

TL;DR

This paper introduces algorithms for reconstructing pedestrian trajectories using dual foot-mounted IMUs, employing Kalman filters and ground contact assumptions, with methods for comparison and trajectory optimization.

Contribution

It presents novel algorithms for pedestrian dead-reckoning with single and dual IMUs, including trajectory comparison and a generalized human motion trajectory construction method.

Findings

Algorithms successfully reconstruct pedestrian trajectories.

Comparison methods reveal advantages and disadvantages.

Optimized computation time demonstrated.

Abstract

This work proposes algorithms for reconstruction of closed-loop pedestrian trajectories based on two foot-mounted inertial measurement units (IMU). The first proposed algorithm allows calculation of a trajectory using measurements from only one IMU. The second algorithm uses data from both foot-mounted IMUs simultaneously. Both algorithms are based on the Kalman filter and the assumption that while a foot is on the ground its velocity is supposed to be zero. Two methods for comparing the obtained trajectories are proposed, advantages and disadvantages of each method are indicated and a way to optimize the computation time is presented. In addition, a method is proposed for constructing one generalized trajectory of human motion based on the trajectories of each leg.