AI in Defense - Enabling Decision Dominance

There’s an arms race going on, and all the big players—the U.S., Russia, China, Europe—are involved.

But it’s not like previous arms races where militaries were only concerned with faster, more maneuverable, stealthier, harder-hitting platforms and effectors. This time around, it’s being played out in the world of software development, computer processing power, and the ability to disseminate information in real time across multiple domains.

This is the artificial intelligence, or AI, arms race.

And let’s be clear, we’re not talking about automated killer robots here, military AI technology is principally being rapidly developed and deployed in response to the dramatic increase in the amount of raw data that is being generated in the battlespace.

Vast numbers of sensors of all kinds are creating mountains of data that require analysis and distribution to those that need the information, in the form they want it, and in real-time.

At the Association of the United States Army’s 2021 annual meeting and exposition, Army Secretary Christine Wormuth said, “Make no mistake – data and software will be as important as ammunition on the future battlefield.”

The problem is that human operators no longer have the capacity to sift through all the raw data to extract the actionable intelligence required to conduct operations. Enter AI teammates, in the form of data-crunching engines.

How is AI used in defense?

By introducing AI into the process, the cognitive burden is greatly reduced on the warfighter, freeing them to concentrate on making informed decisions based on the data that is critical to their mission. AI does the hard work in extracting and analyzing that data so the warfighter doesn’t have to.

Consequently, the ability to make decisions is enhanced by providing all necessary information, while also flagging up elements that might otherwise be missed by humans.

More importantly, the required information is delivered and available with much-reduced delay, significantly shortening the critical sensor-to-shooter (S2S) time.

Acting first—whether it’s a commander ordering a critical force maneuver, or the crew of an individual weapon system taking a first shot—has always been a decisive ability in warfighting, and AI has revolutionized the speed at which those decisions can be taken.

Achieving “decision dominance” is the guarantee to success in the battlespace.

Bringing AI to the battlefield

The power of AI is being brought to bear at all levels of the military machine, from the highest levels of command to edge-deployed forces. Together with advances in computer-to-computer communications, AI is changing the nature of warfare, but it can only be a true game-changer if it is implemented across the whole multi-domain force.

At command levels there are few restrictions on the employment of powerful computing systems. Large computers can run without issue in cooled environments unrestricted in size or power, and linked to large data server farms without bandwidth or latency concerns. The same cannot be said for computers operating at the tactical edge, such as those in manned and unmanned aircraft and ground vehicles.

For a start, the combat vehicle cannot reliably link to large data servers, and all its processing must be done “on premise”, within the vehicle itself. While the establishment of local networks with accompanying platforms is possible, there is no significant cloud-computing capability available.

Edge-deployed platforms are subject to incredibly harsh operating environments including extreme temperature ranges, shock and vibration, EMI/EMC, and dust and water ingress. There are also severe SWaP – or size, weight and power – constraints as computers vie with a host of other systems, all hungry for space and power.

With increasing amounts of data being processed continually by higher performance electronics in smaller enclosures, heat generation becomes a major issue, requiring innovative solutions to conduct and dissipate thermal energy in an ambient temperature within the vehicle that can extend to 75C or higher.

Artificial intelligence in military applications

AI is being applied to many facets of modern warfare. The list of applications is near infinite across all battlespace domains and service branches. A few examples are:

• Aided Target Recognition (AiTR)
• Electronic Warfare/Cyber Warfare (EW/CW)
• Counter-UAS (C-UAS)
• Predictive Maintenance
• Vehicle and Force Protection

Again, the overarching goal in support of the Joint All Domain Command and Control (JADC2) doctrine is to harness AI to reduce S2S time, improve accuracy, and reduce risk to soldiers. This is being supported by the discipline of AI known as computer vision, which effectively enables computers to “see” and more accurately process and analyze imagery and video, which then supports both human and machine decision making.

Drawing on AI also opens up possibilities that were hitherto unavailable. An example is facial recognition technology, in which AI can be “trained” to search through huge mounts of imagery gathered by vehicles on patrol to spot and flag up individuals “of interest” in crowds, or to note and record recurring faces at incident scenes.

As AI is implemented at an increasing rate, and with the ongoing increase in the size of data sets, so the technology itself and its capabilities are improving.

Algorithms can be continually improved through constant feedback from exercises and real-world operations, and deep-learning algorithms allow computers to train themselves through ongoing data generation.

While AI is being rolled out at all levels of the military chain of command, and across many sectors of commercial industry and public security, it is its adoption within the platforms operated by edge-deployed forces that are the key to warfighter success.

What is Systel doing?

We aim to be at the leading edge of making reliable, AI-enhanced computer performance a reality in the hostile operational environments encountered by those who need it most: our soldiers.

We offer a variety of rugged embedded and rackmount server solutions purpose-built for mission-critical edge AI applications in demanding environments.

Systel’s Kite-Strike SFF embedded edge computer products are highly configurable and purpose-built for demanding edge-AI workloads in austere environments in any domain.

We utilize a single line replaceable unit (LRU) system product architecture and hyperconvergence design philosophy, minimizing system SWaP while maximizing performance and capabilities, with a laser-focus on thermal engineering to ensure reliable operation across wide temperature ranges.

Our solutions are powerful enough to host complex AI algorithms and provide the connectivity required to interact with sensors, displays and communications devices, all while offering maximum reliability for any mission in any environment.