Data Fusion Architectures for Orthogonal Redundant Inertial Measurement Units


  1. Gagnon, E.
  2. Vachon, A.
  3. Beaudoin, Y.
Corporate Authors
Defence Research and Development Canada, Valcartier Research Centre, Quebec QC (CAN);Numerica Technologies Inc, Quebec City (QUE);Laval Univ, Quebec QUE (CAN)
This work looks at the exploitation of large numbers of orthogonal redundant inertial measurement units. Specifically, the paper analyses centralized and distributed architectures in the context of data fusion algorithms for those sensors. For both architectures, data fusion algorithms based on Kalman filter are developed. Some of those algorithms consider sensors location, whereas the others do not, but all estimate the sensors bias. A fault detection algorithm, based on residual analysis, is also proposed. Monte-Carlo simulations show better performance for the centralized architecture with an algorithm considering sensors location. Due to a better estimation of the sensors bias, the latter provides the most precise and accurate estimates and the best fault detection. However, it requires a much longer computational time. An analysis of the sensors bias correlation is also done. Based on the simulations, the biases correlation has a small effect on the attitude rate estimation, but a very significant one on the acceleration estimation.
orthogonal redundant Inertial Measurement Units;data fusion architectures;sensors bias
Report Number
DRDC-RDDC-2018-P083 — External Literature
Date of publication
01 Jul 2018
Number of Pages
Reprinted from
Sensors, MDPI, Sensors 2018, 18(6), 1910
Electronic Document(PDF)

Permanent link

Document 1 of 1

Date modified: