Air-to-air intercept now seems more like a software problem than a stick-and-rudder problem.
During the recent autonomy flight, the General Atomics Aeronautical Systems’ MQ-20 Avenger took off under conventional control but switched to the autonomy stack while in the air, after which the approved trajectory path has been developed through the human-machine interface and uploaded to the vehicle. Once the transfer is accomplished, the aircraft showed the capability to dynamically honor the “keep-in” and “keep-out” geofences while maintaining the mission.
The main event was an independent live engagement sequence where the aircraft remained quiet in the radio frequency domain. With the infrared search and track sensor provided by Anduril, the Avenger passively detected and tracked a manned target aircraft. Based on these sensor returns, the autonomy on board calculated an intercept geometry and simulated a solution against the live target. There was no ambiguity in the description of the event as provided by GA-ASI: The simulated shot, if real, would have destroyed the target. This was done “without human intervention” as part of the engagement sequence.
This combination—passive sensing combined with decision-making on board—is significant in that it negates two traditional weaknesses of Remotely Operated Aircraft: reliance on continuous human input and the need to turn on a radar to create an air picture. Current IRST designs are specifically trying to move in exactly the opposite direction. Anduril has explained its Iris series as being based on a “Computational Pixel Imager (CPI) technology” which “uses processors in every pixel to reduce noise,” along with real-time AI-based detection and classification to detect a large number of objects. In terms of functionality, passive IR search makes it harder to detect by an adversary waiting for a signal to turn on its own radar.
The airframe is not a paper airplane either. The Avenger has been flying since 2009 and is marketed as a faster and less observable variant of the Reaper family: about 13 meters long with a 20-meter wingspan and one Pratt & Whitney PW545B turbofan engine. General Atomics has specified cruise speeds of 740 km/h at altitudes above 15,000 meters and more than 20 hours of flight time. There is also an internal weapons bay for low-observable sensor/store carrying and externally-mounted hardpoints for a more lenient signature environment when speed is less important.
The reason Avenger has remained relevant has been its use as a proxy development environment for the Collaborative Combat Aircraft strategy of the Air Force, where aircraft are expected to be less “hand-flown drone” and more autonomous team member. The recent Avenger activities have focused on software portability and open interfaces such as flight activities that handed over control of the aircraft to third-party autonomous software stacks and teaming activities that involved virtual and real-world team members. This fits well with the overall effort of the United States government to have standardized autonomy via government reference architectures such as the Autonomy-Government Reference Architecture (A-GRA) effort of the Air Force.
Details such as geofences and passive tracks are the foundation that supports scaling. When autonomy is able to comply with airspace restrictions, accept new mission tasks, and perform instrument procedures, developing an intercept solution based on sensor data, it changes from being a pilot’s responsibility to managing multiple platforms and monitoring intentions. This is where the key performance curve is no longer speed or turn capability but rather sensor data processing, safety boundary definition, and development of an actionable engagement solution within machine speed.
