Decentralized Architecture for Asynchronous Sensors

  • Authors:

  • Eduardo M. Nebot;Mohammad Bozorg;Hugh F. Durrant-Whyte

  • Affiliations:

  • Australian Centre for Field Robotics, Department of Mechanical and Mechatronic Engineering, The University of Sydney, NSW 2006, Australia. nebot@mech.eng.usyd.edu.au;Department of Mechanical Engineering, University of Yazd, Yazd, Iran;Australian Centre for Field Robotics, Department of Mechanical and Mechatronic Engineering, The University of Sydney, NSW 2006, Australia

  • Venue:
  • Autonomous Robots
  • Year:
  • 1999

Quantified Score

Hi-index 0.00

Visualization

Abstract

This paper presents an efficient method of multi-sensorestimation that can be used with asynchronous and synchronoussensors. A decentralized architecture is used for the fusion ofinformation obtained from several asynchronous measurements. Theissue of the synchronization of the information, which is critical inthe proposed method, is addressed. The information form of the Kalmanfilter (information filter) is used as the main algorithm forestimation. The method is demonstrated with the implementation of anavigation system for an autonomous land vehicle. The integrity issueis also addressed with the implementation of multiple independentestimation loops. The proposed method allows for efficient fusion ofinformation obtained from different measurements for covariancereduction, while providing the benefits of decentralized estimationarchitecture for integrity purposes. The resulting estimates areequivalent to an optimal centralized filter when the loopsincorporate all the information available in the system. Theinformation obtained from each measurement is then broadcast to theother loops after being synchronized. This information is used in anassimilation stage to achieve more accurate estimates. Theassimilation frequency is also discussed considering the trade off offault detectability and estimation covariance reduction. Theperformance of the navigation method is examined by comparing the resulting position estimates to those of independent navigationloops.