This post is also available in: עברית (Hebrew)
Artificial intelligence ‘at-the-edge’ can be used to dramatically shorten the OODA (Observe, Orient, Decide, Act) loop and provide military forces with a distinct advantage in the fight. This has been demonstrated in the flight of a new AI-enabled Unmanned Aerial Vehicle (UAV) for the US Army.
The RAPTOR – Robotic Autonomous Platform for Tactical Operations and Reconnaissance – developed by SSCI integrates the company’s Finding Objects via Closed-loop Understanding of the Scene (FOCUS) autonomy software with a commercial off-the-shelf UAV airframe and EO/IR sensor.
FOCUS provides a number of innovative machine intelligence capabilities that, with only high-level direction from a supervisor, also known as “commander’s intent,” provides full autonomous control of the UAV and its sensor to find, fix, track and identify targets of interest in complex environments.
The last step, identification, is performed onboard “at-the-edge” using machine-learning based automatic target recognition software running on a high-performance graphics processing unit onboard the UAV.
The autonomy software was developed under various DARPA programs. It was specifically designed to be agnostic to the UAV and sensor it controls so that only small modifications are needed to control other, potentially larger UAVs, and given the demonstration approach, shows the ability to evolve into multi domain operations.
During the flight, the RAPTOR test vehicle was given the task to find and localize an “enemy” target military vehicle more than 1 km from the UAV’s launch location. Once tasked, the autonomy software successfully navigated the UAV to the search area and, with knowledge of the terrain below, controlled the aircraft flight path and camera direction. Identification of the military vehicle was performed using computer vision. An immediate alert was provided to the supervisor of the potential threat with a precise location and image for positive identification. RAPTOR returned home without requiring any additional direction from the supervisor.
The UAV has been developed under contract from Army Futures Command DEVCOM C5ISR Night Vision Electronic Sensors Directorate.
The flight took place within the framework of Project Convergence, the Army’s largest test event that showcases how cutting-edge AI capabilities can be used to speed up the sense-to-effects kill chain, according to the company’s website.