SBIR-STTR Award

The use of computer processing to detect and identify targets is becoming critically important in several military applications. We present candidate algorithms that implement multisource information integration with truth maintenance and uncertainty representation. These algorithms will allow non-monotonic logic and probabilistic reasoning. Graphical software tools are developed to demonstrate the efficacy and logic of the algorithms. The team assembled for this task has practical experience with the solution of similar problems in smart weapon applications as well as theoretical and practical knowledge of the mathematics and algorithms of sensor fusion and evidence accumulation.

Assumption analysis and tracking plays a fundamental role inmodel-based Automatic Target Recognition (ATR) design and evaluation. Simply put, an Assumption Truth Maintenance System (ATMS) provides a mechanism for tracking assumptions and logical propositions, and recognizing contradictions. We propose to implement an ATMS variant, a Belief Maintenance System (BMS) as an evidence accumulation module available to model-based pattern recognition algorithms. The BMS will allow evidence accumulation within a single algorithm as well as evidence combination for complementary algorithms operating in parallel. The BMS will take the form of a software library to be used in the Center for Night Vision's MAXIMIZE system. The proposed Phase II system will expand the capabilities of the Phase I demonstrator and be integrated into MAXIMIZE.