24th April 2007 — Data Fusion in the Presence of Unknown Correlations

Simon Julier will be the speaker. PLEASE NOTICE THIS WILL BE AT 3PM.


Data fusion, or multi-sensor estimation, is one of the most important capabilities of modern computing systems. However, many data fusion algorithms are based on Bayes' Rule or assumptions of conditional independence. Although these assumptions are convenient to make in theory, they are rarely satisfied in practice and can sometimes lead to catastrophic estimator failure.

In this talk I shall discuss how unknown correlations arise in two contexts: distributed data fusion and simultaneous and localisation and map building algorithms. I shall describe some of the techniques which have been developed to overcome these difficulties and, in particular, discuss the use of the Covariance Intersection (CI) algorithm for fusing state estimates when only mean and covariance information is known.

Some references:

1. S. Grime and H. F. Durrant-Whyte, Data fusion in decentralized sensor fusion networks,Control Engineering Practice, vol. 2, no. 5, pp. 849863, 1994. [note: apparently, no free access from UCL]

2. S. J. Julier and J. K. Uhlmann, A Non-divergent Estimation Algorithm in the Presence of Unknown Correlations,in Proceedings of the IEEE American Control Conference, volume 4, pages 2369-2373, Albuquerque NM, USA, June 1997.

3. C. Chong and S. Mori, Graphical models for nonlinear distributed estimation,in Proceedings of FUSION 04, Stockholm, Sweden, 28th June 1st July 2004.

4. S. J. Julier, An Empirical Study into the Use of Chernoff Information for Robust, Distributed Fusion of Gaussian Mixture Models,in Proceedings of FUSION 06, Florence, Italy, 10th 13th July 2006.


Simon J. Julier is a Senior Lecturer at the Vision, Imaging and Virtual Environments Group, in the Computer Science Department at UCL. Before joining UCL, Dr. Julier worked for nine years at the 3D Mixed and Virtual Environments Laboratory at the Naval Research Laboratory in Washington DC. There he was PI of the Battlefield Augmented Reality System (BARS), a research effort to develop man-wearable systems for providing situation awareness information. He served as the Associate Director of the 3DMVEL from 2005-2006. His research interests include user interfaces, distributed data fusion, nonlinear estimation, and simultaneous localisation and mapping.