Envision high-speed, multi-domain, massive mechanized warfare in 2030 -- -- a fast-moving, lightweight armored vehicle detects an approaching column of enemy tanks with long-range infrared targeting sensors. The targets are dispersed across expansive, mountainous terrain, yet moving in coordination for attack. The armored vehicle cannot fire upon the enemy tanks and give away its position, so it “networks” the targeting specifics to an armed overhead drone which then attacks the enemy tanks -- exploding them with Hellfire missiles, all without putting soldiers at risk.
In similar fashion - perhaps a forward operating unmanned ground vehicle receives the targeting information and, controlled by a human operator, fires on the enemy tanks without exposing the location of a manned crew. Survival and combat success in these kinds of anticipated future warfare scenarios hinge not only upon accurate detection “data,” but entirely upon the “speed” of processing and instantly networking data.
The successes of these kinds of hypothetical, high-intensity warfare scenarios, it is safe to say, describes and inspires some of the core mission parameters of the Army’s emerging AI Task Force, Army Futures Command.
“One of our challenges is how do we get more data off of combat systems and curate that data,” Brig. Gen. Matt Easley, director of the Army Artificial Intelligence Task Force, told reporters at the Association of the United States Army Annual Symposium.
Simply put, the faster target information is found, processed and shared, creating greatly reduced sensor-to-shooter time -- the better the chance of victory or survival in war.
“It is about enabling a speed-up capability, so you can instantly recognize targets for soldiers who would normally have to go through intelligence data manually to look for cues to get the right targeting information. We are getting the right kind of data to enable this speed up. We are working on how to get the right data to enable maneuver with minimal input from soldiers,” Lt. Col. Christopher Lowrance, Autonomous Systems Lead, AI Task Force.
Perhaps a drone, for instance, spends hours patrolling above high-interest enemy territories, yet only detects a minute or two of enemy activity; AI-compatible computer processing can identify those few minutes crucial to combat success -- instantly -- without there needing to be a human to sift through hours of video to find those combat crucial minutes of ISR data.
Newer algorithms are now enabling engineers to not only lower the hardware footprint by consolidating multiple sensor nodes or “boxes” into one but also use AI-enabled computer processing speed to organize otherwise disparate sets of data.
Easley mentioned the current practice of accumulating data from drones such as the Army’s Gray Eagle or modernized Bradley Fighting Vehicles; many of these platforms, he explained, have distinct or “stovepiped” sensor systems which need to aggregate, pool and organizing vast amounts of incoming data...such as navigational information and specifics related to weapons, electronics, sensor cameras and on-board computers. Integrating AI is intended to streamline and "fuse" otherwise separate sources, bringing an F-35-like "sensor fusion" both within and across combat platforms.
Emerging combat vehicles, for instance, such as the Army’s Bradley replacement Optionally Manned Fighting vehicle and the newest v4 variant of the Abrams tank, are slated to receive a 3rd Gen FLIR sensor. This is targeting technology with greater resolution, image fidelity, range and information processing technology. It is something that is expected to change armored vehicle combat techniques by increasing range and target accuracy. Easley cited 3rd Gen FLIR as an example of the kind of information-gathering technology which can be massively improved through the use of AI.
“We are at a critical juncture of getting requirements into the system now … so that 10 years from now we are not where we are today, where it can be difficult to organize some data,” Easley said.
Making a point to emphasize the importance of organizing “wartime information first,” Easley stressed the need to analyze pressing information such as navigational details, fuel data and other key items such as weapons performance.
Alongside infrared sensors, Easley also mentioned both Multi-Spectral sensors and Synthetic Aperture Radar (SAR) technologies as among those slated to benefit from AI-empowered data processing. AI-enabled advanced algorithms, for instance, can monitor or organize hours of drone video feeds or seemingly limitless amounts of sensor-acquired targeting data in a nearly instantaneous fashion. Incoming information is instantly bounced of a vast database with compiled information about terrain, prior targeting, shapes and contours of relevant objects such as enemy drones or combat vehicles, navigational information, distances, atmospheric conditions and countless other pertinent pieces of data - in order to perform real-time analytics, solve problems and organize data for human decision-makers. SAR, for instance, sends electromagnetic “pings” onto surrounding terrain before analyzing the return signal to create a rendering or image for commanders. Naturally, using AI to gather, analyze and process return signals can create much faster and more accurate results. AI-enabled algorithms will instantly compare radar returns against its existing database to generate faster results.
An AI-informed system, for instance, would instantly be able to recognize the form of an existing enemy tank, bounce it off a known database, and immediately determine the identity and location of the target. All of this, in light of fast-evolving processing speeds, takes place in a matter of milliseconds.
The newest experiments with AI, interestingly, are speeding this up even further. DARPA and others are working on AI applications that perform near real-time machine learning and analyze context. In effect, analysis and comparison of new information is performed on data ….as it is arriving… almost allowing for a two-way data flow in real-time.
Using AI-empowered sensor consolidation and data networking introduces a range of new war possibilities and brings an expansive set of new combat tactics. This is quite significant to Army Futures Command, as it works with Army acquisition to both harness new technologies and explore how they will change formations, combat maneuver and key combat tactics. With this kind of fast-paced, AI-centric networking, combat platforms can operate in a more dispersed fashion, yet retain connected interoperability. This opens up new multi-domain options as well, should intelligence data from aircraft be processed and shared with ground commanders.
One example of this cited by Easley was the importance of “creating interoperability” between the Patriot and Terminal High Altitude Air Defenses (THAAD). While similar in mission concept and application, THAAD and Patriot interceptor missiles have different ranges, radar systems and flight speeds. THAAD, for instance, can destroy approaching enemy missiles as they re-enter the earth’s atmosphere during the “terminal” phase of flight. Patriot missiles, while similar, are designed for closer in approaching enemy fire.
Should these air defense systems be better networked, they could expand the defensive envelope and possibly hand-off targeting data as incoming attacks pass into the earth’s atmosphere. Ballistic missiles, for instance, follow a certain flight trajectory, so information on an approaching missile could be handed from one radar system the other to coordinate and expand defenses. This increases the prospects for a successful intercept or “kill” of an incoming enemy missile. By increasing the data consolidation, and therefore speed, of information sharing, AI can improve air defenses and enable nodes to defend larger areas and operating farther apart.