Integration of IMU in indoor positioning systems with non-Gaussian ranging error distributions

Abstract

The performance of a wireless positioning system can be improved through the integration with an inertial measurement unit (IMU). Conventional sensor fusion algorithms based on a Kalman filter (KF) are not accurate for indoor positioning systems since the ranging errors in indoor environments are non-Gaussian distributed due to non-line-of-sight (NLOS) and multipath propagation. In this paper, we propose a novel sensor fusion scheme for integrating range-based indoor positioning systems with IMUs, which supports non-Gaussian ranging error distributions at a low computational cost by using a pseudo position measurement for the measurement update step in the KF. The performance of the proposed fusion scheme is evaluated experimentally using an indoor positioning platform in an office environment, and is shown to be significantly better than conventional approaches. Particularly, the median positioning error is reduced by 52% using the proposed scheme, and the maximum positioning error is reduced to 0.36 m using an offline tracking algorithm extended from the proposed approach.

Effective Landmark Placement for Robot Indoor Localization With Position Uncertainty Constraints

“…Nevertheless, these values are well below ξ(p k ) = 0.8 m. Hence, in light of the discussion in Section II-C, the target uncertainty constraint is met for any |ρ| ≤ 1, thus confirming Scatter diagrams of the position estimation errors ex and ey associated with the NBE (empty circles) and the EKF (filled circles). In the same graph, the ellipses corresponding to covariance matrices P x,y k given by (7) for the NBE (dashed line) and for the EKF (solid line) are also shown for the sake of comparison with the circle of radius ξ(p k ) delimiting the target uncertainty region (dash-dotted line).…”

“…Indoor localization and position tracking systems rely on a variety of sensing technologies including (but not limited to) fingerprinting-based techniques based on radio signal strength intensity (RSSI) measurement [1], [2], electronic circuits measuring the Time-of-Flight of wireless signals [3], [4], inertial platforms [5], [6], [7], calibrated vision systems [8], [9], or a combination thereof [10], [11], [12]. While sensing technologies and accuracy specifications depend on the target application or the type of agents to be tracked (e.g., pedestrians or robots) [13], [14], common general requirements for indoor localization are:…”