The Air Force Will Embed Airmen at Carnegie Mellon University
Credit to Author: Nicholas Thompson| Date: Wed, 01 May 2019 21:59:17 +0000
Earlier this week, WIRED spoke with Heather Wilson, the secretary of the Air Force. She was just finishing a visit to Carnegie Mellon University, which has developed a special relationship with the service. We spoke about the Air Force's new Science and Technology Strategy, which was announced earlier this month, as well as their new initiatives in artificial intelligence, surveillance, and space. We unfortunately did not have the opportunity to discuss whether the allied forces of the living had misused their tactical air power advantage in the battle with the Night King.
Nicholas Thompson: Secretary Wilson, hello. Thank you for talking with WIRED.
Heather Wilson: Nick, thanks for doing this. Just as a little bit of background, Carnegie Mellon is one of our strongest research partners. In 2016 the Air Force established a six-year agreement with them on trusted swarming and autonomy. Then in the fall of 2018, we established a center of excellence here in human-machine teaming and trust. Now we think the next step in implementing the Science and Technology Strategy is to place airmen in the hubs of activity. And in the case of Pittsburgh, that is autonomy and robotics. So Pittsburgh will be one of the places where the Air Force puts a cell of airmen, based on our long partnership with Carnegie Mellon, focused on autonomy and robotics.
NT: And those airmen will be at Carnegie Mellon or elsewhere in Pittsburgh?
HW: They will be at Carnegie Mellon.
NT: What exactly will they do?
HW: Well, there’s a variety of things. One is to get the operator’s perspective to the researchers. Carnegie Mellon has about 250 researchers on faculty, doing some research on robotics and associated technologies or machine learning. So it’s a massive faculty. They also have probably three dozen robotics companies already located in the city of Pittsburgh, who also have relationships with Carnegie Mellon. It's becoming a hub for robotics and autonomy. And so our airmen will be involved in research and they will be engaged with researchers. The goal is to more closely connect the Air Force and the airmen with America's best universities at the centers of innovation.
NT: And the specific technologies that they'll be working on include developing drone swarms?
HW: Autonomy and robotics, generally. To give an example, how can an unmanned air vehicle also connect with robots on the ground? A single human doesn't want to, and can’t, control a huge number of robots, whether they're flying or on the ground. So how do you get robots to talk to other robots? You want a general command that says: We're going to move to the right, and the robots talk to each other about how to do that without stumbling over each other.
"Anytime war extends into space, everybody loses."
Air Force Secretary Heather Wilson
Another issue involves sustaining aircraft. Seventy percent of the cost of aircraft is in the maintenance and upgrading them after we buy them. When you look at where we spend the man hours on aircraft maintenance, the number one demand for an aircraft in depot is painting.
NT: Really?
HW: So we have been driving forward the technology for laser paint removal. Carnegie Mellon is one of the universities doing that research. An aircraft can only have the paint stripped a certain number of times before it affects the metal underneath it. We're now advancing the technology to do laser paint removal that doesn't touch the underlying metal, but takes off paint and then re-paints at microscopic kind of levels. On a C-5, by using this method, the weight that they save is equal to the entire weight of an F-16 aircraft. Paint is heavy!
NT: So the reason paint is important is because of its mass. Is it also due to radar avoidance?
HW: No, not in the case of the C-5. A C-5 is not a low observable aircraft. We put paint on to avoid corrosion. But we have to repaint, and when we strip it, traditional methods of stripping are both labor intensive and do some damage to the metal underneath.
NT: What are the other technological advances that you've been most excited about in your tenure as Secretary of the Air Force?
HW: Oh my gosh, there’s a whole lot of them. I'm not even sure I could start to list them. To me, probably the most important advances were having are not in platforms, but in the ability to connect platforms.
NT: In the Science and Technology report, you talked about rapid effective decision making. What did you mean?
HW: Well, you collect information from multiple sensors. If you fuse the data, you can give somebody more time to make decisions. If we are monitoring patterns of light—near-global persistent surveillance—then you try to see things that are different. We don't want people staring at screens looking for the different thing. If you can get machines to figure out there's something different here, and then zoom in and tell you what it is, then you can help somebody in a cockpit make a decision about where the bad guy is.
One of the other things that Carnegie Mellon is doing and helping with is machine-based object identification. You take a huge set of pictures of the earth and can you train machines to say, “This is a pickup, this is a pickup, this is a pickup with trailer attached.” We probably have 80,000 people in the Air Force who are looking at computer screens trying to identify and discriminate pictures. It’s useful, but not very efficient.
NT: What is the current status of Project Maven?
HW: That’s a project continuing to go forward.
NT: And how effective has it been? It's obviously something of great interest to our readers.
HW: I probably have not been updated on it in four or five months. But the last demonstration I saw is it is continuing to advance and be useful.
NT: Back to image recognition: What have been the most important leaps to improve your image recognition capabilities? Is it software and artificial intelligence? Is it more highly refined sensors? Is it better communication between the sensors?
HW: It’s the actual software in the learning algorithms to discriminate different things. The other with this connection of near-global persistent awareness, which is going to happen—even outside of the military, it’s just going to happen—so how do you manage and develop the data science to be able to make that data useful? That is going to be a huge challenge for everyone.
NT: And then surveillance, obviously lots of the near persistent, near universal surveillance will come from satellites. Tell me your feelings right now about the biggest risks in space conflict.
HW: You're right that some of the data will come from satellites, but the whole point is to connect all sources of data and be able to connect things to each other. So they have an X-37 that's connected to an unmanned aerial vehicle that passes data to a submarine, and that same information is available to a special operations team. We call that multi-domain operations or multi-domain command and control.
As for our ops in space, we are the best in the world at space, and our adversaries know it. But they can deny us the capability to use space in the crisis of war. The technologies they're using include directed energy, jamming, and kinetic operations.
NT: And do you think we've sufficiently hardened our assets in space, or is that a top priority right now?
HW: In the fiscal year 2019 budget, the president proposed and the Congress approved a significant increase in space and an adjustment to our strategy in the program supporting it. He has also proposed in the FY 2020 budget another double-digit percentage increase in space. So in the last three years, we have seen consistent double digit increases in space. And anytime war extends into space, everybody loses.
NT: What are you most excited about as you go back into academia?
HW: You know, I have so much to do before I leave the Air Force, and I’m not going to Texas until August. So I'm going to focus on all the things I’ve gotta do by the end of May.
NT: We wrote about the pitch day that you recently hosted in New York. Was there anything particularly noteworthy that came out of it?
HW: There were a couple of things that I just thought were great. With one guy, we signed a contract with him in three minutes, and he said, “You know, it’s faster doing business with the Air Force then it is to order a beer in a bar in New York City.”
The other thing: By doing the first progress payment that day, these small businesses didn’t have to take out bridge loans or find some other source of financing to contract with the Air Force in order to get going. So they didn't have to do work for 30 days and then do billing. We were able to work with companies that would normally not be able to work with us.