Genetically-Evolved Digital Networks for Target Planning

Target planning is one of the most significant technical challenges to developing a tactical or strategic war plan. The requirement is to analyze considerable amounts of data concerning the enemy, common and allied force situation and provide results in a form useful for quality decision making in a short period of time. Due to the complexities of mission planning, these systems will becom more autonomous, with a man-in-the-loop to override, correct errors and to monitor performance. Short timelines and semi-autonomous operations pose a classic challenge to the artificial intellegence community to explore new and innovative adaptive information processing approaches that minimize reliance on heuristic or ad hoc techniques. Networks of dynamically programmable logic modules (DPLM networds) appear promising in such areas as pattern recognition and classification, signal processing, and robotics. One specific requirement is to develop new parallel processing techniques to support target planning applications. The objective fo the proposed Phase I project is to demonstrate that a genetic algorithm can be used to evolve DPLM network software to support complex target planning applications.