Follow Datanami:
May 13, 2019

AI Enlisted for Air Combat Training

illustrate dogfight for DARPA AI program

Source: Lockheed Martin

Despite the billions of dollars invested in front-line fighter aircraft, there has been little actual close-range air combat—so-called dogfights—since the Gulf War of the early 1990s.

Nevertheless, the Defense Advanced Research Projects Agency is using the aerial dogfight scenario to leverage AI as a way of improving the interface between pilot and extremely fast-moving machines. Earlier this month, DARPA rolled out its latest AI effort dubbed ACE, as in Air Combat Evolution, to develop trusted AI to assist fighter pilots in making split-second decisions while flying at supersonic speeds. The idea, program officials said, is to automate air-to-air combat, freeing fighter pilots to focus on the overall air battle.

While dogfights between front-line fighter jets like those between American and Russian pilots during the Korean War are now exceedingly rare, U.S. fighter pilots are nevertheless trained to perfect dogfighting maneuvers. The ACE program will use the aerial combat scenario as a way to train “semi-autonomous airborne assets” on the basics of dogfighting. The goal is to build trust between pilot and machine to “accelerate the transformation of pilots from aircraft operators to mission battle commanders,” DARPA said.

ACE will train AI models in much the same way new fighter pilots are taught the fundamentals of aerial combat, including fighter maneuvers designed to get behind an enemy aircraft. Program officials noted that the highly “nonlinear” characteristics of aerial combat make it a “good test case for advanced tactical automation.”

Another consideration is the physical limitations placed on fighter jocks enduring huge gravitational loads in front-line aircraft. Hence, the DARPA program seeks to “co-evolve tactics” for automating aerial combat.

“Only after human pilots are confident that the AI algorithms are trustworthy in handling bounded, transparent and predictable behaviors will the aerial engagement scenarios increase in difficulty and realism,” said U.S. Air Force Lt. Col. Dan Javorsek, ACE program manager in DARPA’s Strategic Technology Office.

“Following virtual testing, we plan to demonstrate the dogfighting algorithms on sub-scale aircraft leading ultimately to live, full-scale manned-unmanned team dogfighting with operationally representative aircraft,” Javorsek added.

ACE is among several DARPA programs under the rubric of “mosaic warfare” designed to shift the U.S. military away from costly manned systems that can take a decade or more to develop. The AI effort represents a shift toward less expensive unmanned systems that can be developed and deployed more rapidly, then be upgraded to address emerging threats.

The resulting “mosaic” of manned and unmanned systems is promoted as a set of weapons that can be “recomposed” as threats evolve or can be replaced if destroyed, creating what the defense agency asserts would be more resilient fighting force.

The agency is holding a “Proposers’ Day” this week to gather ideas for implementing the ACE program. In the initial phase, DARPA said it expects to work with algorithm developers with experience spanning air combat simulations to gaming.

Recent items:

DARPA Embraces ‘Common Sense’ Approach to AI

Algorithms Go to War

Datanami