In order to deal with the real world, automated systems must deal with data which is noisy, incomplete, and not always accurate. Multi-sensor fusion has emerged as the best method for maximizing the reliability of data inputs to these systems. This book represents the first broad, practical text on the subject - covering all the technologies and data structures associated with multi-sensor fusion.

The book explains the underlying concepts of multi-sensor fusion, which allows data to be collected from several sources and combined, delivering the most accurate overall picture for use by automated systems. It includes over 100 illustrations, especially helpful in understanding data structures needed to model multi-dimensional data. Part 3 of the book builds on six years of sensor fusion research for the Office of Naval Research, presenting novel approaches to fusion related problems such as image registration, distributed agreement, and sensor selection, as well as efficient algorithms for fusing sensor data. The book comes with a set of functioning C programs that implement the applications discussed, including neural networks, genetic algorithms, tabu search and simulated annealing. The Sensor Fusion Toolkit includes both an example Kalman filter and the authors' own Distributed Dynamic Sensor Fusion algorithm, which is easier to use and solves a wider variety of problems.