A RULE BASED DATA FUSION SCHEME FOR WHEELED MOBILE ROBOT LOCALIZATION

• Published in: IFAC Proceedings Volumes Vol. 39; no. 16; pp. 867 - 872
• Main Authors: Blume, Holger, Abelbeck, Frank, Heimann, Bodo
• Format: Journal Article
• Language: English
• Published: 2006
• Subjects: data fusion; localization; recursive least squares; wheeled mobile robot; localization; data fusion; wheeled mobile robot; recursive least squares
• ISSN: 1474-6670
• DOI: 10.3182/20060912-3-DE-2911.00149

This paper deals with a data fusion algorithm for wheeled mobile robot localization using data from multiple localization systems. One is dead reckoning, which is subject to different systematic and random errors, summing up over time. Still, it offers high resolution in the close-up range and by data fusion with a landmark-based low resolution localization system qualifies for high accuracy path control. Here a data fusion algorithm is developed, which chooses the more reliable data source providing smooth transitions. Furthermore the system makes use of time spans when the availability of both systems is high to improve the quality of the position estimation. The parameters of the error correction algorithm are continuously estimated by a recursive least squares algorithm. The methodology is applied to a 4-wheel omnidirectional wmr. A significant improvement of precision in position estimation against simple dead reckoning can be shown.