Johns Hopkins APL Hosting Final Installment of DARPA Virtual Combat Competition

The Johns Hopkins Applied Physics Laboratory (APL) in Laurel, Maryland, will virtually host the third and final installment of the Defense Advanced Research Projects Agency’s (DARPA) AlphaDogfight Trials (ADT) from Aug. 18 to 20. Audience members can watch online as participating teams pit artificial intelligence (AI) algorithms against one another in simulated air combat, known as a dogfight. The winning AI will go on to compete against an experienced Air Force fighter pilot flying a virtual reality F-16 simulator.

Originally slated to take place at the U.S. Air Force’s AFWERX innovation hub and then at Nellis Air Force Base (both in Nevada), but postponed due to the COVID-19 pandemic, DARPA opted to instead return to APL for ADT’s final installment.

The competition will be the final event in the three-part trials, serving as the springboard for DARPA’s Air Combat Evolution (ACE) program. A broader effort aimed at automating certain air-to-air combat operations, ACE is also focused on improving human-machine teaming. These trials provide the participating organizations a chance to develop and demonstrate their algorithms for the Air Force and receive feedback from actual fighter pilots.

Streaming live from APL via a ZoomGov webinar, the virtual program event — AlphaDogfight Trials TV (ADT TV) — will offer viewers several perspectives of the dogfights in real time. The program will also feature experts and guests from the “Control Zone” commentary desk. Col. Dan “Animal” Javorsek, program manager in DARPA’s Strategic Technology Office, and other DARPA officials will recap activities and scores on Aug. 20, the final day of the competition.

APL hosted the first two ADT competitions in its Intelligent Systems Center in November 2019 and January 2020. The events brought together teams spanning technical and defense industries, including Aurora Flight Sciences, EpiSci, Georgia Tech Research Institute, Heron Systems, Lockheed Martin, Perspecta Labs, physicsAI and SoarTech.

DARPA tapped APL’s expertise to develop the simulation environment and adversary agents for the trials. The Lab leveraged its capabilities in software development, AI development, modeling and simulation, and aircraft dynamics and controls to create the simulation environment and AI for the three ADT competition events. APL has worked closely with DARPA on previous projects involving modeling and simulating autonomous systems.

“DARPA chose APL for this role due to the diversity of technical skills required to execute the trials, and their confidence that we could meet an incredibly ambitious schedule,” said Christopher DeMay, program manager for the DARPA ACE program at APL.

After ADT, APL will continue working with DARPA on the ACE program, developing additional AI in phases: modeling and simulation, subscale aircraft and full-scale aircraft testing.

“We’re bringing together the AI research and operator communities,” DeMay said. “Our ultimate goal is demonstrating trusted, scalable, human-level autonomy for air combat.”

Register now to watch the AlphaDogfight Trials at https://events.jhuapl.edu/ADT3AUGUST2020 . Registration closes Aug. 17.

Watch team videos and highlights from the first two ADT competitions here: https://www.youtube.com/playlist?list=PL6wMum5UsYvaTmVj10WZo5-1vVH-pdLR4.