The upcoming test of the U.S. Air Force Advanced Battle Management System (ABMS) in the next few weeks is to feature all the military services and dozens of platforms, as the Air Force moves toward fielding rapidly upgradable systems and clean sheet designs featuring artificial intelligence/machine learning (AI/ML) that may one day see use on autonomous drones as the leading edge of U.S. military force, Air Force Acquisition Chief Will Roper said on Aug. 25 in a YouTube “Ask Me Anything” discussion moderated by Preston Dunlap, the Air Force’s chief architect.
The ABMS “on ramp” tests are to be held every four months to highlight how new technologies perform, including Cloud One, Platform One, and software defined radios, and move the promising ones toward fielding within weeks through the employment of agile software. That would mark a marked change from the legacy, years-long acquisition cycles of the Cold War.
“This [ABMS] on ramp is massive,” he said. “There are going to be 70 different industry teams participating, 65 different government teams, 33 platforms, two combatant commanders, all acting as one team. Why do we do these four month cycles…and see what we’re able to successfully connect and not? It’s precisely to force the type of rapid learning so that we can move, and build, and operate at Internet speeds.”
The U.S. Air Force is planning to integrate Project Maven into the test (Defense Daily, Aug. 12). Cloud One/Platform One will be a hosting environment for Project Maven to turn “a developmental system into a warfighting system” during next month’s on ramp, Roper has said. Kicked off in 2017 with the oversight of the office of the undersecretary of defense for intelligence, Project Maven has looked to develop an AI tool to process data from full-motion video collected by unmanned aircraft and decrease the workload of intelligence analysts.
The Air Force has requested $3.3 billion for ABMS over five years, including $302.3 million in fiscal 2021.
ABMS, which the Air Force describes as the air and space “military Internet of Things,” is part of Joint All-Domain Command-and Control (JADC2), an effort to build a cross-service digital architecture for multi-domain operations–in effect, a military Internet of Things with machine-to-machine interfaces to ensure that “our personal lives are at least moderately reflected in the military that goes to war,” Roper said on Aug. 25.
“I think we’ll eventually rename Advanced Battle Management System because it really isn’t battle anymore,” he said. “It started as the recap of JSTARS, as airborne battle management, and expanded into advanced battle management when we realized we’d probably have to have air and space and maybe ground components to replace that mission. But then we realized, if you actually want to be able to distribute something and have those distributed platforms be able to work together seamlessly, you’ve signed yourself up to build an Internet of Things. You’re going to need software-defined systems, cloud, containerized software so you can move it from the cloud to the edge.”
The Air Force’s intent with ABMS is to tailor data to user needs, but also to make data widely available.
“What we want to do is make sure that machine-to-machine data exchanges occur everywhere, that if any sensor sees something, that data is available to a shooter anywhere without impediments [like] human driven processes, phone calls, work chats,” Roper said.
AI-enabled drones may one day be the leading edge of U.S. military force.
“If we’re going to take a big airplane with people on it into harm’s way in a future fight, we’re going to have to explain why we couldn’t do that mission differently,” Roper said. “I’ll be the first to tell you we’re not ready to pull people out of the fight, to AI everything up. R2D2 is great in the movies, but R2D2 in the real world gets really confused when an adversary is trying to mess with the data they’re ingesting to make decisions. Adversarial tactics are the death of current machine learning.”
“I think the first thing we’re going to have to do is pull people back from that leading edge of warfare where things are so lethal and uncertain, especially early in a conflict,” he said. “I see a huge potential to automate that, to have drones and unmanned systems take on that dangerous job. One of the most important things they’re going to do is produce data that we can use to get smarter about what that contact point of war looks like. I think that those leading edge assets, whether in the air, on the sea, or on the ground, will have to be quarterbacked by people in platforms that are standing back, ready to make the calls.”