Northrop Grumman’s Scaled Composites Model 437 Vanguard jet, an AI testbed aircraft, has returned to the skies after a year-long hiatus, ready to host autonomy software from at least six partner companies. These partners are bringing a diverse range of capabilities to the platform, including augmented reality training systems, tactical combat algorithms, and mission planning and analysis tools. The announcement was made at the Air Force Association (AFA) conference last week.
Prism Autonomy Package: The Key to Integration
The Vanguard jet took off on September 20 for the first time in roughly a year. During the down time Northrop Grumman installed its Prism autonomy package. Prism is designed not only to fly the aircraft but also to accept specialized software modules from any company that conforms to certain common technical standards.
So far, six companies — Red 6, Autonodyne, EpiSci, Merlin Labs, Soar Technology and Shield AI — have publicly announced their participation in Northrop’s flight test program, known overall as Beacon. Beacon includes the Vanguard aircraft, its Prism software, and assorted support systems on the ground.
Augmented Reality Training with Red 6
Red 6 specializes in augmented-reality training for military pilots. Their software superimposes virtual aircraft on the pilot’s field of vision — and feeds realistic data about these unreal planes to the real aircraft’s sensors. Chief Strategy Officer Kevin Fesler, a former Air Force fighter group commander, explained that their technology allows a single pilot in a single aircraft to train in scenarios too complex, expensive, or dangerous for real-life execution.
Autonodyne: Coordinating Virtual and Physical Formations
Autonodyne’s software focuses on coordinating the maneuvers of entire formations, blending physical and digital elements in a tactically optimal way. CEO Steve Jacobson described the initial focus as coordinating the maneuvers of that entire formation, both its physical and digital elements, “in a tactically optimal way.”
EpiSci’s Human-Machine Teaming Expertise
EpiSci, recently acquired by Applied Intuition, is contributing autonomy software focused on effective human-machine teaming. President Dan Javorsek, a former Air Force squadron commander, emphasized the importance of „getting the reps and sets with human pilots.”
Merlin Labs: Dynamic Decision-Making in Flight
Merlin Labs is testing „a package of mission autonomy behaviors,” said General Manager for Tactical Autonomy Chris Gentile. Their tactical algorithms will analyze sensor data, augmented by Red 6’s synthetic data, and instruct the aircraft on how to react. Gentile claims that their technology will enable the aircraft to make “dynamic, highly effective human or superhuman decisions.”
Soar Technology: Collaborative Autonomy for Combat Air Patrol
Soar Technology is providing a software module that enables the aircraft to conduct a Combat Air Patrol (CAP), a complex set of coordinated actions involving multiple aircraft. Vice President for Autonomy Jack Zaientz said that this enables the aircraft to conduct what’s called a Combat Air Patrol, which is a standardized but complex set of actions that requires multiple aircraft to coordinate their movements.
Shield AI: Learning and Adapting Through Mission Lifecycles
Shield AI’s software learns and adapts over multiple autonomous missions, supporting mission planning, execution, and analysis. Chief Strategy Officer Ryan Tseng said that their algorithms help the aircraft carry out its mission in flight and that they also help plan the mission before takeoff and analyze the outcome afterwards, potentially even updating the software and tactics before taking off on the next mission.
Testing and Integration Ahead
Northrop President of Aeronautics Tom Jones cautioned that the Vanguard aircraft requires further test flights before integrating partner software. The company has five more such test flights before it flips the switch and lets Prism take full control, which should happen by the end of the year. Once Prism is thoroughly proven, the partners can start plugging their software into it for flight tests. The plan is for multiple partners’ modules to be running at the same time during the same flight, executives said, allowing for complex interactions that are more than the sum of the individual parts.
