Soldiers and Commanders to Assess FIRESTORM AI Technology
The system will face tougher challenges in Project Convergence 2021.
Both soldiers and combat commanders likely will get hands-on experience in the coming months with one of the Army’s hottest new artificial intelligence systems known as FIRESTORM.
The artificial intelligence (AI)-enabled system, formally named FIRES Synchronization to Optimize Responses in Multi-Domain Operations, still is in the science and technology phase and is not yet a formal program of record. It ingests data from sensors and other systems, uses One World Terrain to map the battlefield and recommends the best weapon system to engage specific targets, saving commanders precious time for making decisions. Prior technologies took almost 20 minutes to relay data back to warfighters. FIRESTORM takes 32 seconds.
FIRESTORM was the megastar of last year’s Project Convergence and has been highly praised by Army officials helping determine the service’s future, including Gen. John Murray, USA, commanding general, Army Futures Command, and Brig. Gen. Ross Coffman, USA, director of the Next Generation Combat Vehicle Cross Functional Team. Gen. Coffman has suggested the system could fundamentally change the way the Army fights in the future.
The technology will face more of a challenge at this year’s Project Convergence, which will be conducted at White Sands Missile Range, New Mexico, and Yuma Proving Ground in Arizona from October 12 to November 9, than last year’s. For example, soldiers will get their hands on the system and provide feedback for the first time. Also, the system will consider more factors before making decisions.
Last year, FIRESTORM had only one so-called decision node. “This year, we’re going to have a lot more decision-aid nodes … to validate the decision tree and whether it is making the decision based on a decision tree, which has many, many factors. You need to look at all of those factors to see if it provided the right recommendation,” says Ketula Patel, FIRESTORM program manager and Intelligence Systems branch chief with the U.S. Army Combat Capabilities Development Command Armament Center, Picatinny Arsenal.
Beyond Project Convergence, the Picatinny researchers intend to improve FIRESTORM in other ways as well. Future enhancements likely will include even more advanced AI and automation algorithms. Some eventual enhancements also will be tailored to the needs of commanders. Feedback from commanders will help researchers refine some capabilities, such as targeting, predicting air clearance processes and deconflicting air space.
“Project Convergence gets it ready for a lot of different kinds of use cases, but it doesn’t consider everything. Going beyond, we would be working with a lot of the [combat command] partners and active duty, like III Corps, or even divisions, to make sure we’re incorporating a lot of that feedback into FIRESTORM,” Patel says.
Patel suggests her team could work directly with combatant commands, or COCOMS, in the Indo-Pacific region or Europe to better meet their needs. “We’re trying to connect with some of the COCOMs. As the software matures, I think we’ll be able to support a lot more COCOM events, probably in the later part of 2021 or 2022 for sure.”
For this year’s Project Convergence, the system also will be integrated with about 20 other systems and will support joint missions. That includes the Air Battle Management System (ABMS), an Air Force solution for the Joint All-Domain Command and Control Concept. ABMS allows a joint force to use cutting-edge methods and technologies to rapidly collect, analyze and share information and make decisions in real time, according to an Air Force press release.
It also includes the Army’s Air and Missile Defense Workstation, a staff planning and battlespace situational awareness tool. It provides the user with an air defense picture and supports the Surface-Launched Advanced Medium Range Air-to-Air Missile air defense system by providing an automated defense planning capability for deployed units.
“There’s a lot of integration work that’s happening to make sure we can receive data from all kinds of systems that are fielded—or emerging technologies—and have interoperability with them,” Patel reports. “We’re probably integrating to, I’d say, about 20 technologies for Project Convergence. We’re extending our interfaces to work with a lot of those different, newer platforms.”
She notes that FIRESTORM can work for commanders at higher echelons, including joint task forces, or at the tactical level for individual tank or helicopter crews. “The platforms, such as the ground tank commander, could utilize the system all the way to the joint task force. So, we’re also integrating with the Abrams since Abrams is at the tactical edge.”
The system will benefit overwhelmed tank commanders. “It’s not going to get involved with a direct fire mission. If a commander sees a target, he’ll continue to engage as he sees is the best for him to minimize any kind of fratricide and also a self-defense kind of scenario,” Patel explains. “Where it helps is if that specific tank platoon is getting an overwhelming number of targets. FIRESTORM will continue to communicate with all the other inorganic assets that could support those fire missions.”
For example, FIRESTORM could alert the commander of unseen dangers. “Let’s say a target that came from a higher echelon, or an intel system saw a target that was beyond line of sight of that platform, that could be alerted to that commander,” Patel adds.
FIRESTORM already has been partially integrated with the Army’s Tactical Assault Kit and Nett Warrior and also is available on Linux-based laptops. “The Android capability, I would say, is not as mature from a decision-aiding and algorithm perspective. It’s really mainly on the Linux laptop,” Patel says. “We will continue to develop other technologies, such as Microservices, to have it working with cloud and all the other newer, modernized architectures that the joint partners, as well as the Army, are developing.”
In addition to the technology piece, researchers must consider differing policies and doctrine. “Once we establish what we’re sharing between the systems, and we all are on the same page between us and all the developers, then it becomes a little bit easier. I think we are still figuring out what it is that makes sense from a commander’s perspective that they would want to see … across different warfighting functions,” Patel explains.
Other enhancements are in the works as well. “We’re improving everything from working at scale to being able to handle a lot of the data that we’re getting … given that we will be at the different echelons, there is a lot of that work getting implemented to scale it up,” she adds.
While reducing the decision-making time from 20 minutes to 32 seconds is impressive, Patel stresses optimization as another crucial benefit. “We’re optimizing the target assets that we want to utilize and not just using any available asset that could be the shooter. We’re looking for the best shooter and the best effects for any target,” she says.
That benefit cannot yet be quantified. “Right now, we don’t have quantified data. We will as we build out use cases and run out simulations in fiscal year 2022. We will scale it up where we will ingest a lot of targeting data and utilize what might be all the effects and see which ones were selected and come up with a cost metric. I don’t mean cost from a money perspective but from an optimization perspective,” Patel explains.