| Objective: ||Develop methods and tools required to allow human interaction with automated systems performing data association and motion pattern analysis.
|| Description: ||Much of the research and literature in the areas of tracking, as well as situation refinement, and threat assessment (JDL Fusion Levels 1-3) have focused on the application of automated reasoning techniques to derive their results. These results may be based on continuous multi-intelligence tracking algorithms, models-based situation and threat assessment algorithms or applications of various calculi to achieve the optimal understanding of the battlespace. One key shortcoming of this approach in the real world is the lack of total understanding of the dynamic adversary, coupled with incomplete features externalization by sensors, precluding continuous and fully automated tracking and reasoning. At the same time, most of these designs ignore the hugely powerful inputs available from human analysts into the fusion process. Most of these algorithmic approaches are based upon an a-priori model of sensed environment and the adversary in question. In most cases uncertainty caused by algorithmic inability to associate data based on a lack of features or priors leads to a product that is not usable by the warfighter due the uncertain nature of the fused data.
This effort focuses on two main aspects: Smart Data Association and Motion Pattern Analysis. This effort should consider the entire range of typically available sensor assets, including but not limited to MTI, fixed imagery, motion imagery, MASINT, and SIGINT. While focused on sensor based data, other ancillary information or data sources (e.g., terrain, cultural information, road networks, etc.) where useful can/should be applied to deduce and exploit the underlying behavior. Smart Data Association requires the development of processes, algorithms and interfaces to optimally utilize smart association algorithms to combine elements that may be separated by time and space. The design should include the means for a human to dynamically assign their own criterion for association of observed data in the battlespace. This design should allow the analyst to define how many sources of a particular type or specific sources they will accept to call sufficient for association and creation of an “associated” entity in the battlespace. Addition association criteria to be defined by the operator include geolocational and temporal accuracy. Another aspect of this SBIR, is the development of techniques that automatically detect and recognize operationally significant motion patterns once the data has been fused using the Smart Data Association.
|| ||PHASE I: Explore smart information association and motion pattern analysis techniques in the fusion decision space. Develop a human friendly approach to data association and pattern analysis and demonstrate feasibility of the proposed approach.
|| || ||PHASE II: Implement a prototype of the Phase I approach. Integrate prototype into a government supplied tracking architecture. Demonstrate integrated prototype using a variety of Multi-Intelligence sensor track data. Evaluate results against a set of government supplied measures of performance.
|| ||DUAL USE COMMERCIALIZATION: Military application: Tools and component technology developed in this SBIR are applicable to a broad range of sectors, including commercial logistics, homeland security, and transportation. Commercial application: Technology developed in this area can be applied to tracking of commercial vehicles, as well as used by the FAA for tracking of commercial and private aircraft.
|| References: ||1. D. L. Hall, Mathematical Techniques in Multi-sensor Data Fusion, Artech House, Norward, Ma, 1992.
2. Y. Bar-Shalom, and T.E. Fortmann, Tracking and Data Association, Academic Press, New York, 1998.
3. J.W. Guan, and D.A. Bell, Evidence Theory and It’s Applications, vol 1. Studies in Computer Science and Artificial Intelligence|
|Keywords: ||Smart Association, Motion Pattern Analysis, Human on the loop|