Sensor Data Fusion for Target Classification and Identification

The proposed effort addresses the problem of Non-Cooperative Target Recognition/Identification, NCTR, for airborne targets that are beyond visual range. To address the problem, The proposed research is focused on novel artificial intelligence techniques,such as fuzzy logic, neural nets, and wavelets to make NCTR techniques perform better. Although artificial intelligence techniques have been investigated for single sensors, they have not been explored extensively when there is data available from avariety of sensors. Topics to be investigated also involve the application of such techniques as Pattern Recognition, Decision Theory, and Optimization Theory. Current NCTR techniques work quite well but only under constrained battlefield conditions.The intent is to use a data fused algorithm using some of these techniques as a basis from which to begin the development of an algorithm that is much less susceptible to battlefield conditions. Artificial Intelligence techniques have had success in thecontrol field. The approach is to build upon the application of those techniques and to adapt to the extent possible the manner in which those techniques were applied. Once the criteria and logic flow have been determined, the design of the algorithmwill occur.If the proposed research is successful, it can have tremendous impact on numerous weapon systems that have a need for Non-Cooperative Target Recognition/Identification. The ability to achieve NCTR has been sorely need by all DOD services. Ithas the potential to give the battlefield commander better control over his resources and could reduce logistics considerably. With this technique, the battlefield commander will have a capability to determine which of his resources is the most applicablefor any threat engagement scenario. This can reduce operating and support costs significantly. Not only can costs be reduced, but better weapon system performance can be realized.