WASHINGTON — As a U.S. Air Force fighter pilot closes in on a target, she fires four GPS-guided bombs. As the weapons swarm toward their destination, sharing information about their surroundings, one of the munitions sights a higher-priority target nearby. Instantly, the game plan changes — the bombs’ programming now directs two of the weapons toward the high-priority target, while the rest carry out the original strike.

That scenario is similar to one the Air Force Research Laboratory plans to test this year using a Collaborative Small Diameter Bomb, one of the two munitions the Air Force is developing in the Golden Horde program.

With Golden Horde, the Air Force aims to find out whether munitions can be networked together and operate autonomously after launch according to a set of predetermined rules. Col. Garry Hasse, director of AFRL’s Munitions Directorate, emphasized that this capability is different than a weapon that can independently make decisions based on artificial intelligence.

“We get that question a lot,” he told Defense News in a June 18 interview. “With all the talk about artificial intelligence and other things in ‘Terminator’ movies … [there’s] certainly some trepidation of what independent capability a weapon may have.”

Golden Horde is one of the three technology development efforts chosen by the Air Force in 2019 as a “Vanguard program” — a high-priority prototyping and experimentation initiative that the service earmarked as potentially groundbreaking. Along with its fellow Vanguards — the loyal wingman drone known as Skyborg and Navigation Technology Satellite-3, an experimental satellite that would augment GPS — the Air Force hopes to speed Golden Horde toward either fielding or failure.

AFRL is working on two networked munitions for Golden Horde: the Collaborative Small Diameter Bomb I (CSDB-1) and the Collaborative Miniature Air-Launched Decoy (CMALD). Both involve taking munitions currently in production — the laser-guided version of Boeing’s Small Diameter Bomb I and Raytheon’s Miniature Air-Launched Decoy — and then outfitting them with new radios that allow the weapons to exchange information, and equipping them with new processors for additional computing power, said Norma Taylor, the program manager for Golden Horde.

That in turn enables a massive software upgrade known as the “autonomy module,” a playbook of algorithms that tell the weapon how to respond to specific changes on the battlefield, whether that means the sighting of a new threat or the destruction of some of the collaborative weapons.

So far, AFRL has completed the hardware design for CSDB-1 and is working on software and algorithm development, Hasse said, “and then also doing simulations and running through different scenarios that is part of defining those playbooks.”

AFRL is on track to begin F-16 fighter jet flight tests with the CSDB-1 this fall and winter, with similar tests of the B-52 bomber carrying the CMALD planned for summer 2021, Hasse said.

The first flight test scenarios will be simple, helping the Air Force gauge whether the weapons are properly communicating across the network and acting in accordance with the mission playbook. For example, a team of CSDB-1s could come across a threat while en route to attack a target and would have to change trajectory to avoid it.

A pilot’s experience launching the CSDB-1 or CMALD should be about the same as the baseline version of the weapon, with mission-planning teams working to preprogram the networked munitions with information about its rules of engagement ahead of takeoff, Taylor said.

“[Mission planners] would give information to the weapon on an appropriate engagement zone, where it would be considered proper for the weapon to engage targets … and they would give the weapons information on known targets in that zone,” she said. “But if they have some idea that there might be other targets out there that they don’t know, they will give the weapon some information in terms of priorities, so that if you come across a higher-priority target that’s in the authorized engagement zone, then you have permission to change your assignment.”

Eventually, AFRL will use the CSDB-1 and CMALD together in an integrated swarm in a more complex scenario, currently scheduled around 2022, Taylor said.

“One of the important aspects of the networking aspect of the technology is the ability to send information to the weapon while it’s in flight and give it a new mission download, so to speak,” she said. “It might have been loaded with ‘mission A’ on as its primary mission, and if things change, then there is an ability through that networked radio to provide mission updates.

“But the important thing is to realize that whether it’s preprogrammed in mission planning or whether it’s a mission update to your flight, the operators are still required to provide the weapon those rules of engagement are on, where they can and can’t engage, and what their mission priorities are.”

In the future, Hasse noted, the Air Force could eventually use the series of platforms and sensors linked through the Advanced Battle Management System to cue collaborative munitions on changes in the rules of engagement.

The integration of hardware and software on the Golden Horde program is being performed by Scientific Applications and Research Associates Inc. — which won a $100 million contract in 2019 for CSDB-1 — and Georgia Tech Applied Research Corporation, which won $85 million for CMALD, Air Force Magazine reported.

AFRL plans to present an acquisition plan for collaborative weapons “soon,” Hasse said, but the weapons developed during Golden Horde might not be the munitions that become acquisition programs of record.

“I don’t know [that it’s] specifically a collaborative [small diameter bomb] that we’ll transition into the inventory. … I think more it’s a matter of looking at some of the plays and the networking capabilities and the collaboration capabilities that could be put into other systems.”