Personal Recovery Improvement with Multifactor Models and Algorithms for Learning (PRIMMAL)

Personnel recovery missions are a race against time for rapidly recovering personnel that are reported lost or missing to ensure their survival and recovery without being captured.   Using a last known location or area of interest, a search and rescue (SAR) mission must locate the target by reducing the search space of the last known location area.  Currently, the Joint Personnel Recovery Agency utilizes LandSAR, OSPRE, and FINDER to narrow the search space.  Although these tools exist for SAR, the baseline data needed for creating an initial region of interest of search, a bounding box, are determined by pen and paper.  An improvement for the state of the art will be using artificial intelligence and machine learning (AI/ML) to create motion models conferred through more accurate and timely output of a bounding box given a starting point, possible target destinations, areas to avoid, and parameters for integrating speed of advancement.  Input parameters such as land terrain, elevation, landcover, weather, and age, health, or injury status of the targeted personnel will improve the input models integrated with LandSAR, OSPRE, or FINDER.  To advance the state of the art for personnel recovery using AI/ML tools, we propose to create a new plug-in capability using both heuristic approaches and closed-form models to answer the last known location of personnel, understand personnel decision making, environmental attributes for personnel motion, physical traits of the personnel, and the accuracy of using an AI/ML approach.