Model fusion for inertial-based personal dead reckoning systems

This paper introduces a model fusion approach that improves the effectiveness of Personal Dead Reckoning Systems that exploits foot-mounted Inertial Measurement Units. Our solution estimates a sensor orientation by exploiting the Madgwick's algorithm integrated with popular Kalman-based solution. This way, attitude and heading correction is not based on the Zero-Velocity phase assumption which introduces significant error. The experiments conducted on ground-truth data shows, that the proposed approach outperforms state-of-the-art solution by reducing systematic and modelling errors and also provides better heading estimation.