FA864922P0875

Small Business Innovation Research Program (SBIR) Phase II

AIMBOT: AI Maneuver and Battlespace Optimized Targeting

The F-22 program requires rapid integration of cutting-edge AI-powered tools to advance Pilot-Assisted Autonomy, Optimized Intercept Guidance, and BFM (dogfight) instruction. AIMBOT will adapt and advance key features of our existing Tactical AI system to meet F-22 program needs in these key focus areas. First, the threat analysis, path planning, target assignment, and WEZ management features of our Golden Horde efforts will be adapted to F-22 and F-35 weapons, kinematics, and signatures, increasing situational awareness, recommending courses of action, and generating targeting and intercept guidance for the desired transition platforms. DARPA ACE provides the background for collaborative control in MvN engagements, coordinating similar or dissimilar platforms to effective force employment. Finally, the resulting guidance commands will be presented to the pilot through a variety of display modalities tactical overview via FOX tablet, HMD guidance through the Scorpion or other helmet, and potential data injection into platform host avionics. As the mission progresses to BFM, EpiSci's best-in-field BFM agents will be retrained for F-22 characteristics and weapons, delivering a comprehensive BFM coaching, guidance, and debriefing system. EpiSci's agents operate by choosing policies which determine the optimal macro-level BFM maneuver and desired energy state, including turn circle entry & alignment, vertical maneuvering, defensive responses, and weapon employment for both engaged and supporting fighters. These macro-actions are executed by a traditional control system, allowing for portability between different aircraft types and providing a natural interface for bandit performance replication. Our patent-pending AI-BFM Flight Director, flown on L-29 and L-39 aircraft in support of ACE and Test Pilot School programs, provides real-time guidance via HUD/HMD while capturing quantitative debrief information on differences between pilot and agent actions alongside a real-time risk and success estimate. Finally, in debrief the recorded data may be used to generate a what-if scenario, demonstrating agent choice vs. pilot decisions at any moment in the flight and playing each scenario to a logical conclusion for lessons learned. Using agile development practices, open architecture interfaces, and frequent operator feedback, AIMBOT will demonstrate advanced capability delivery at the speed-of-need, equipping the F-22 fleet for continued Air Dominance.

The goal of phase II is to continue the R&D efforts initiated in Phase I. Funding is based on the results achieved in Phase I and the scientific and technical merit and commercial potential of the project proposed in Phase II.

AF221-DCSO1

F2D-4019

X22.1

Epiphany Ryu