Air Force Collaborative Pushes AI Frontiers — Fast

 In U.S. Air Force, Air, Forces & Capabilities, U.S. SOCOM

WASHINGTON:  Air Force Research Lab  (AFRL) has pio­neered a unique indus­try col­lab­o­ra­tion to speed solu­tions to the tough­est AI tech­ni­cal chal­lenges. Its most recent suc­cess is the just-fin­ished DARPA AlphaDogfight con­test.

The Autonomy Research Collaborative Network, or ARCNet, is a part­ner­ship between AFRL and an indus­try con­sor­tium man­aged by the non-profit SP Global Institute (SPGI). The con­sor­tium boasts some 140 mem­bers, Mark Pohl, ARCNet oper­a­tions direc­tor for SPGI said in an inter­view today. About half of the mem­bers are small, non-tra­di­tion­al firms, he said, and the other half are estab­lished defense con­trac­tors.

As DoD focus­es more heav­i­ly on autonomous sys­tems for every­thing from new weapons to logis­tic net­works, almost all devel­op­ment pro­grams keep run­ning into the same fun­da­men­tal prob­lems, explained Corey Schumacher, AFRL’s ARCNet pro­gram man­ag­er.

“There are sev­er­al under­ly­ing tech­nol­o­gy chal­lenges for almost any­thing we’re doing on auton­o­my,” he said.  “Cybersecurity is absolute­ly one of those. Verification, val­i­da­tion and test is anoth­er. Artificial intel­li­gence and learn­ing —  how we how we deal with those in autonomous envi­ron­ments? … How to inte­grate var­i­ous mul­ti­ple autonomous sys­tems? And how to make the human inter­act with them and make them inter­act well with the human? There are some major under­ly­ing threads that recur over and over again in this space.”

DARPA used ARCNet to find and con­tract the indus­try com­peti­tors in the AlphaDogfight, designed to explore the via­bil­i­ty of machine algo­rithms to fly a fight­er jet in combat con­di­tions. As Breaking D read­ers know, the AlphaDogfight result­ed in an AI ‘pilot’ built by Heron Systems beat a top Air Force F-16 pilot in a sim­u­lat­ed WWII-style aerial battle.

The trials were designed as a risk-reduc­tion effort for DARPA’s Air Combat Evolution (ACE) pro­gram to flesh out how human and machine pilots share oper­a­tional con­trol of a fight­er jet to max­i­mize mis­sion suc­cess. The over­ar­ch­ing ACE con­cept is aimed at allow­ing pilots to shift “from single plat­form oper­a­tor to mis­sion com­man­der” in charge not just of flying their own air­craft but man­ag­ing teams of drones slaved to their fight­er jets, the ACE program website explains. And as Sydney reported last week, Defense Secretary Mark Esper says the effort will now advance to test­ing in real air­craft — with DARPA eyeing a goal of hand­ing over the pro­gram to the Air Force in 2024.

The goal of AFRL in set­ting up ARCNet a year and a half ago was not just to tackle those fun­da­men­tal prob­lems plagu­ing autonomous system devel­op­ment, but also speed actual solu­tions. Thus, the con­sor­tium works with AFRL under a “col­lab­o­ra­tive agree­ment” that fos­ters both rapid con­tract­ing and a com­mu­ni­ca­tions-inten­sive envi­ron­ment so that DoD and indus­try researchers can more easily over­come obsta­cles and reach con­clu­sions, Pohl and Schumacher explained.

“Our tagline, and our goal both as we were start­ing off and as we go for­ward, is ‘we award con­tracts with 60 days.’ And we’ve done it con­sis­tent­ly,” he told me. Indeed, Schumacher said with no small amount of pride, the “median time to award” at this point is 53 days. This, of course, is light speed for DoD con­tracts, even for most other con­tract vehi­cles that DoD uses out­side the Federal Acquisition Regulation (FAR) process.

ARCNet award­ed nine projects to 17 com­pa­nies last year worth $23 mil­lion, and cur­rent­ly has $37 mil­lion worth of projects under­way, accord­ing to a set of back­ground slides pro­vid­ed to Breaking D. Customers so far include not only AFRL and DARPA, but also DoD’s Research and Engineering Office (OUSDR&E) and its Test Resource Management Center; Special Operations Command (SOCOM); and the Air Force Life Cycle Management Center (AFLMC).

Schumacher said sev­er­al other DoD offices are now in the process of work­ing with the con­sor­tium, but demurred on pro­vid­ing specifics. He noted, how­ev­er, that one vector for rais­ing inter­est in the con­sor­tium has been the the DoD Autonomy Communities of Interest (ACoI) within the Defense Innovation Marketplace.

According to the ACoI web­site: “The Autonomy CoIs will close­ly exam­ine the DoD’s S&T invest­ments in the enabling of autonomous sys­tems, to include the strate­gic assess­ment of the chal­lenges, gaps, and oppor­tu­ni­ties to the devel­op­ment and advance­ment of autonomous sys­tems, and iden­ti­fi­ca­tion of poten­tial invest­ments to advance or ini­ti­ate crit­i­cal enabling tech­nol­o­gy devel­op­ment.”

While the funds involved may not be big bucks, the lure for indus­try is two-fold, Pohl and Schumacher explained.

First, win­ning firms receive gov­ern­ment fund­ing to do basic research on projects they needed or wanted to do for their own efforts. The gov­ern­ment gets “gov­ern­ment pur­pose rights” to use the research and its results; the com­pa­ny gets to use it in its own pro­grams includ­ing those aimed at the com­mer­cial market. For this reason, all ARCNet research is unclas­si­fied.

Second, many of the projects being fun­neled through ARCNet are “Phase 0” efforts for bigger projects that have already agreed tran­si­tion plans — mean­ing that a win­ning firm can get car­ried along the acqui­si­tion cycle into a pro­gram of record.

Breaking Defense source|articles

Recommended Posts

Start typing and press Enter to search