RAISR (Robust Artificial Intelligence for Search and Rescue)

Efficient Search and Rescue (S&R) of personnel is critical for providing lifesaving care during combat and natural disasters (e.g. floods, hurricanes, earthquakes, etc.). Current S&R practices have significant technical challenges including: (1) information used for determining subject locations and safety risks is often incomplete, error prone, contradicting, and/or out of date, (2) search procedures are sub-optimal due to vast search areas and/or unique domain constraints/obstructions, with high dependence on the search domain (i.e. combat, lowland, mountain, urban, cave, or maritime), and (3) dynamic environments such as threat movements, weather, and flood water runoff, significantly impact the performance of sensors and scene understanding algorithms. The RAISR technical solution interfaces with the next generation C2 (Command and Control) systems and UxV (Unmanned Vehicle) platforms to address these technical challenges, by: (1) integrating automated scene understanding and low-level planning cues onboard manned/unmanned assets, allowing for undisrupted operations in dynamic, low-communication environments, (2) performing data fusion and high-level planning at mobile command stations for further determining of safety/threat risks, optimizing overall search plans, and identifying safe rescue paths, and (3) incorporating human-in-the-loop features for adding human knowledge to the system, providing human explanations, and options for human adjudication of any automated decisions. RAISR follow-on work aims to enable lightly trained teams to deploy domain-configured toolboxes with pre-packaged UxV platforms, sensors, and algorithms, while interfacing with mobile command station C2 systems and humans-in-the-loop. The future deployed RAISR solution is expected to have an order of magnitude improvement in the following four areas: reduction of search/rescue times, increase in percentage of successfully rescued personnel, improvement of S&R worker safety, and reduction of S&R equipment cost to maximize scalability.