As we look at how we close kill chains, there’s very much a man in the loop at every single point. That process today takes minutes — sometimes even hours — and when we look at conflicts of the future, we have to be operating in the seconds and milliseconds.
Today the showcase was a simulated threat, represented by a T-6 flying at around 200 or so knots. We use a network of sensor towers that autonomously detecting, tracking, classifying, and then identifying this as potentially a cruise missile threat. We then upped the game and brought in F-16s started at 350 knots, 400 knots, 500 knots, and we had a solid detection and ID within about one second on an F-16 at 500 feet going 550 knots.
The way they want to build that is through the use of a low cost maritime surveillance radar on top of each one of these sensor towers that got the initial detect, that then queued in an AESA [active electronically scanned array] radar, and that AESA radar queued in an electo-optical and IR [infrared] sensor that allowed the computer vision — or machine learning — to start to do the identification of that threat. That whole process happened in about one second.
That was a great video from Anduril [UPDATE: video was removed bc it contained sensitive but unclassified data, there will be future videos that I’ll link to] showing how their technology is contributing to the Advanced Battle Management System (ABMS). HT: Richard W.
Once the system identifies the threat, a human can choose to disregard it or, if action is required, the system can pull up actions based on which available assets are closest. ABMS, of course, is a much larger and more ambitious program with several contractors participating. Watch this discussion with Chief Architect Preston Dunlap to learn more.
Here’s another good part from the Anduril demo:
So today what we showcased was the ability to have an off-grid sensor that can automatically do detection, classification, identification of a class 1 UAS threat, but then automatically on its own send out an interceptor to go out and track that treat, monitor that threat, or if required to actually kill that threat as well — all without having a human in the loop.
The reaction we saw from those in the room who haven’t been familiar with this tech is really astonishment, that you could get detects down to an ID in a matter of seconds.
Here’s another good video from Anduril [UPDATE: also removed] showing some similar capabilities from a year ago. It is interesting to see some significant changes and improvements between the videos which were about a half-year apart.
The Lt. Colonel narrating the video remarked how military operators in the room reacted with “astonishment” about the technology. It is truly unfortunate that the GAO, in their performance of program oversight on behalf of Congress, didn’t even show up to the operational demonstrations happening in December 2019. Just this four minute snippet shows significant capability has been achieved on short timelines and with a small budget — and that’s just one aspect of the 28 subsystems comprising ABMS! Yet the GAO bashed the ABMS program for not having sufficiently detailed multi-year plans, cost estimates, and so forth. Along with many others I think the judgment is unfair.
Surely one or many of the 28 ABMS subsystems (e.g., radioONE, fuseONE, dataONE, spectrumONE) will have significant pivots or will be totally cancelled. And the GAO can say, “see, you didn’t plan well enough!” But that doesn’t mean ABMS was a failure. The ability to update plans based on real world experiments ultimately leads to better products and value in the end.
As long as ABMS is continually delivering capabilities that military users approve of, it should receive support in OSD and Congress. ABMS is gearing up to do experiments with Space Force and Indo-Pacom. They rolled out devices to support the Covid-19 fight. It seems unlikely that ABMS will escape the regular justifications and processes required of major defense acquisition programs. The bigger question is whether ABMS can maintain a decentralized structure with rapid iterations, pivots, and emergence (as Roper phrased it) while staying compliant with the bureaucracy.
Leave a Reply