New Air Force AI will address data influx problem
The Air Force is planning to design next-generation drones based on the data being sought and implement artificial intelligence solutions to process that data.
The drone warfare of tomorrow will be different than today’s, according to Kenneth Bray, the Air Force's acting associate deputy chief of staff for intelligence, surveillance, and reconnaissance (ISR). The next generation of drones will have to be able to overcome physical and electromagnetic defenses and the Air Force will have to be capable of exploiting and analyzing the data they collect, he said.
For this reason, the Air Force is initiating a “paradigm shift” in its ISR capabilities and analysis. As described by Air Force officials at the 2017 Defense News conference this month, the shift will involve two key changes. For one, drone platforms will now be designed based on the specific data desired, rather than around their sensor payloads. Secondly, artificial intelligence will play a greater role in drone data exploitation.
“I need to design outward from the data now,” explained Bray. “To ask if it’s even collecting the right size data or do I need to have different sensors on those platforms? Are those platforms even relevant anymore or do I need a different platform because what I need is this type of data, and only this type platform can get me that type of data?”
Part of this new design methodology will include developing drones specifically for highly contested environments. Technological research along these lines includes non-GPS navigation and completely autonomous operations. Another aspect is the need for rapid development of new technology for collecting a specific type of data.
“[The Air Force] must stop trying to make the next [drone] be the best on the planet,” said Bray. The focus should be on developing only a specific capability per platform to be the best it can be so that it can reach the field as fast as possible. “Because when we face an adversary of like capabilities…we need to have the answer before they have the answer so that we can shape how the outcome occurs,” he explained.
In addition to re-conceptualizing drone development, the Air Force is also looking for solutions to the problem of data influx. Although the service has collected the largest image repository in the world, Bray said, there is currently no comprehensive search mechanism for that repository. This is where artificial intelligence and algorithmic warfare comes in.
AI won’t replace human analysts, said Bray. However, software algorithms can “learn” to recognize and categorize key objects in an image, for example, so that human analysts can focus on analyzing the objects’ ISR significance.
The Algorithmic Warfare Cross-Functional Team (AWCFT), also known as Project Maven, has been charged with exploring algorithms as a way to turn the influx of input data into manageable intelligence, according to a memorandum from the deputy secretary of Defense.
“Algorithms will be an important element of our weapons systems…we triage lots of imagery, and then it’s just a matter of how big our labelled data sets can get. We expect to see some phenomenal results just in this calendar year,” said Marine Corps Col. Drew Cukor, director of the ISR processing, exploitation, and dissemination for the DoD.
The first next generation data exploitation algorithms are expected to enter into use this year, and will be compact at just 75 lines of code each, according to Cukor.
Ultimately the idea behind the Air Force’s two-fold shift is that “you’ve got to understand your problem,” said Bray. “Like using the right tool in a tool shop, when you need a hammer, you don’t use a screw driver, and sometimes you have to create a new tool…in order to do the job.”