Indoor geolocation based on earth magnetic field

—One of the most recent and popular tourism application
is the virtual visits (virtual tour guide) which are based on
Augmented/Virtual Reality technology (AR/VR). Such systems
suffer from the lack of precise indoor geolocation of visitors
within the cultural heritage site. This issue can be resolved by
using smartphone (and tablets) embedded sensors, which can
catch a huge number of data on visitor behaviour (acceleration,
orientation, etc.). This data can be fused using dedicated methods
(Extended Kalman Filter (EKF), Particle Filter, etc.) in order to
estimate visitor position. However, such algorithms are highly
dependent on: sampling time, sensors technologie, number of
used sensors, etc. With this kind of solution the error in position
estimation growth with the covered distance. In this paper, we
present an hybrid solution for reducing error in time without
additional infrastructure such Wifi, etc. For this, we present an
indoor geolocation solution based on smartphone inertial sensors
and earth magnetic field. The proposed method is divided into
three phases: heading estimation using 3 sensors (accelerometer,
compass and gyroscope) with an Extended Kalman Filter, steps
detection and step length estimation, error correction using
fingerprinting. The proposed solution is implemented on a
smartphone (Samsung-Galaxy S7 based on Android OS) and
maintain the error under of 1.5m.