On constructing an expert system for contact localization and tracking

Abstract

This paper presents a growth layered approach to construct an expert system for Contact Localization and Motion Analysis (CLMA). It delineates present difficulties in CLMA approaches and provides a conceptual approach for remedy. Of importance is the inclusion of knowledge of structure and mathematical relationships to the classic If … Then production rules. Also incorporated are the time relationships to depict trends in the growth of the system states. Existing CLMA approaches are severely limited by being unable to utilize various sources of knowledge and by relying almost entirely on human experts to reduce many and diverse uncertainties. Expert system approaches are much needed to provide not only a significant improvement in performance but also to execute expert level tasks quickly and reliably. The growth layered expert system considered in this paper evolves from existing CLMA approaches and as such may not initially follow the precise construction of current expert systems. As the expert system grows from the entry level, it can better meet the demands for multisensor integration, multicontact tracking, and other complex situations in difficult ocean environments. The system configuration integrates the tools of signal/channel analysis, pattern recognition, and artificial intelligence. An example of an entry level multipath ranging expert system is presented.