Output details
12 - Aeronautical, Mechanical, Chemical and Manufacturing Engineering
University of Sunderland
Environment classification using Kohonen self-organizing maps
This paper describes a novel implementation of Kohonen self-organising maps (SOMs) to automatically classify environments from sensor, without the need for human intervention. It was primarily aimed at enhancing the navigational capabilities of autonomous underwater vehicles (AUVs), but the method described would be equally applicable to unmanned autonomous aircraft. The work described in the paper was the result of experiments conducted during a successful PhD project and involved training and testing SOMs using realistic three dimensional environmental data obtained from digital elevation maps. Expert Systems is a respected international journal that publishes papers dealing with all aspects of knowledge engineering, for a wide range of applications.