Advanced Tracking through Remote Supervision

"Target recognition and tracking requires the solution of several hard problems. A target signature must be identified in sensor data, this data must be associated with prior information on the target, and the association must be used to gain understandingof the target. In the presence of countermeasures and clutter, each of these components is computationally onerous even when using suboptimal algorithms. To improve tracking, our approach offloads computationally expensive decisions to remotesupervisors. These supervisors could be large terrestrial computers or humans. Our primary focus is on the latter. The goal is to compensate for the shortcomings in fielded hardware by incorporating human intelligence into the tracking process. We willuse a new certainty-based scoring mechanism, model-based tracking, and a new method to propagate supervisory information through the track tree. To support these algorithmic techniques, we will also create a software architecture for data selection, datatransfer, data presentation, and feedback capture. For fail safety, the system we propose will always be able to continue operation without remote assistance. Energid will extend the software components developed under this effort into a commercial software toolkit. Though the newly developed source code will be provided to the DoD, other potential customers will purchase the toolkit as software libraries andheader files. By linking these