Flight Tests Validate Collision - Avoidance system

(high-pitched warning sound) Many pilots today don't go out with an immediate concern on their flight that they're going to run into the ground. There is a problem in aviation of controlled flight into the terrain. Controlled flight into the terrain is a case where you have a perfectly healthy aircraft that is, what's called "under controlled flight." In other words, it hasn't stalled. The pilot has full control ability of that aircraft. Yet, the aircraft ends up running into the ground in some way. We're developing an automatic system to avoid a ground collision in a case of a fully functional aircraft. Which will automatically take control of the aircraft and automatically fly that aircraft away from the ground to save the pilot and the vehicle. You don't need a multi-million dollar fighter with digital-fly-by-wire flight controls and very expensive hardware to implement a system like this. We can put these algorithms that determine if there's a imminent ground collision, we can put them on a phone, which we've demonstrated on the DROID, and give out that application. When you go from an F-16 to something like a small UAV, we unplug the F-16 model, we plug in the small UAV model and the rest of the system remains fairly much the same. The technology that we're developing is important. We all have friends and colleagues who aren't with us anymore because they've run into the ground. (Sound of jet flying.) (Sound of ejection seat exploding.) (music plays) (chatter) In the fighter area, there were about one or two people dying a year in the Air Force alone due to this controlled flight into terrain problem. And that is when an airplane in perfectly good health crashes into the ground due to either um, pilot error or possibly the pilot is unconscious as in fighter jets pulling high G's, they may lose consciousness. This is something that's not just confined to military aircraft. And it's not confined to UAVs. General aviation aircraft have a big problem with running into the ground. Pilots do for whatever reason, they're distracted, they can't see, they're in a fog bank. Another case is where they are spatially disoriented. A good example of this is the J.F.K., Jr. mishap that happened out on the East Coast a number of years ago. They fly in, uh, sometimes extreme landing and take off in a locations, up on mountains and things where their aircrafts don't have the aerodynamic ability to climb like they think they should. A number of years ago, the Defense Safety Oversight Council out of the Undersecretary of Defense for Personnel and Readiness saw this problem and asked us to begin addressing it for fighters. In conjunction with the Air Force, the Air Force Research Lab and Lockheed, we went out and did an initial program for, on the F-16 that would help address the problem on all Air Force fighters. The ground collision avoidance maneuver chosen for the F-16 is a wings level climb. It climbs really well and it doesn't turn very well. So we don't try to turn in front of an obstacle, we try to climb up over the top of the obstacles. So regardless of what position you are in, in the F-16, if there's an impending collision, the airplane will roll to wings level and begin a five to six G wings level pull to clear the local terrain. Okay. Gonna roll inverted. Pull the nose down. It's going to roll itself upright. And pull. Bingo. That's all automatic. When you have that kind of automatic computer actuation of a decision, you have to take into consideration a few more things than a warning system would. Don't make things worse is the first requirement and the second one is don't, um get in the way of normal piloting activities. And then the third is, actually avoid collisions. You've got an auto pilot that has the ability to really move this airplane around. We have the ability to put six Gs on this airplane. Six Gs is not an easy thing to withstand. Fighter pilots get used to it but still, six G's when you didn't ask for it is a surprise and it's really gonna get somebody upset if it didn't have to happen. Pull up. Pull up. Pull up. This is just downright annoying. Altitude. So what ends up happening is after a few false warnings in the cockpit, a pilot either tunes out the warnings or turns the system off and now that warning system provides no benefit. In a figher airplane, these collision avoidance maneuvers are very aggressive so when they happen when they shouldn't it's going to get in the way of the mission and it's going to upset the pilot, severely. So avoiding nuisances in a fighter airplane is extremely important. Their concern was that they weren't going to be able to carry out their mission and that the system was going to interfere. When they were shown that they could still do their mission and the system would actually wait longer than they would before initiating a recovery they realized that was not going to be a problem. We believe that the system we developed in the ACAT program that's being fielded in the F-16 now, will prevent above 90%, maybe as much as 98% of the mishaps. We went out and tested the system against every historical category of controlled flight into terrain mishap that the F-16 has seen and based them on actual mishap cases. And we showed that across the envelope of operation of the system, it prevented every single mishap. (music plays) We wanted to set up a system that would be easy to adapt to any different aircraft. At the heart of that is a modular software architecture. These pieces could be replaced like an F-16 piece could be removed from the system and then a UAV piece inserted in and then the rest of the algorithms could remain the same. The F-16 represents the high-performance top-end of the performance spectrum. Airplanes that go real fast, climb real well, have great climb performance, but because they're going fast, their lateral turn performance is not as good. In the case of a small UAV aircraft, like the DROID, we have a limited climb performance, something very typical of a general aviation aircraft. That performance model is very different. And hopefully, by anchoring both ends of that spectrum, we cover everything in between. I think it makes it a lot more plausible that we could move to a different aircraft all across the spectrum of aviation. One of the main costs in bringing any kind of a capability to an aircraft is the actual hardware. A very small difference in where the airplane actually goes. We had an eye on the future and we recognized that eventually we wanted to get this into the general aviation community. And, the best way to get something into the general aviation community would be to put something on tools that would be readily accessible to them. So we moved the software from high-dollar fighter hardware processors to cell phone. And we've now migrated it onto a cell phone app. At bare minimum, an app that would warn you. The more advanced versions would tie into an auto pilot and actually take control and recover the aircraft. Not all general aviation airplanes have the auto pilot that would be necessary, but at the bare minimum it's an app that you can put on your cell phone when you would get a warning that you are about to run into something. (sound of model airplane engine) And that has enabled this technology to be potentially brought to the rest of the public for a much broader application and life savings. I couldn't tell on that very first run... Basic pieces that are needed to make this system work are an understanding of where the aircraft is. This is provided throughy GPS. You also need to know what the terrain is around you. So we have a computer model of the world that is a digital database of elevations. We then, inside the software, have a model of how well the aircraft can perform. And that model is very precisely tuned to that aircraft's performance capability. Even since they first started developing this capability, back in the 90s, capabilities of computers have gotten much, much better um, the capabilities of GPS navigation systems have become better, the uh, the terrain database that allows the aircraft to know where, where the ground is where the mountains and the valleys are, um, has gotten a lot better. What we did was, how we kind of overcame that is we had to go to different sources, and we kind of have this mish-mash of sources that were at different resolutions and different accuracies, and we put those together into our, you know, a custom database that really I think is tailored more to what we are doing with Auto GCAS than the original databases were. We've built a utility that compresses this terrain into a rather small file size, such that we can carry the entire globe on the phone. So we got a homogeneous, much more accurate digital terrain database to use now. So by using improved digital terrain, we get improved performance, and then by using, um improved navigation system, we know where we are over that terrain much better. So we can predict the airplane to terrain collision geometry much better than we could ten years ago. The system monitors the approach of the ground and the aircrafts ability to maneuver to avoid it. It predicts an escape trajectory that keeps the aircraft on a forward trajectory but pulls it up away from the ground. But it also computes a trajectory that moves the aircraft to the right and one that moves the aircraft to the left. It compares these two things, the, the terrain, and the aircraft's ability to avoid it. If the system thinks that the aircraft is trouble, it will select the last possible option of those three trajectories. It isn't forced to try and climb over it. It can turn to the side of it. That's unique to the DROID system we've developed, and that is a feature that will likely have, uh, a great advantage when we take it and apply it to uh, general aviation aircraft. (music plays) On a typical day of flight, the team shows up here, and at early in the morning, we gather all our gear, we load it into our command-and-control van, head out into the field, assemble the aircraft, set up our, our systems, and then go through a series of tests. Our first flight is going to be against that, that hill...pilot flown, through the gap. And we're using two VHF frequencies for... What I've been doing is operating our user interface which is the link really between us on the ground, the test team, and the phone. We do a run, we actually set up the auto pilot through the ground and control station. We give it a point that that it's gonna fly to that's actually under the ground. We're telling it to, you know, to really run into the ground. The system is monitoring this and at the last instance disengages that auto pilot system and uses the auto collision avoidance system to take control of the aircraft and avoid hitting the ground. One...Zero...fly up (music plays) Fly up right. (music plays) OK...Run complete GCAS control (music plays) Just over the course of our testing, and we went through our testing as rapidly as we could, there were no fewer than five fatalities and seven mishaps that we could have prevented with that software on the very fighter aircraft that we were trying to design it for. I think the system we're fielding now with the F-16 isn't perfect, uh, but I think everybody in the field is going to love it and I think it's going to save lives. Do we have the perfect answer? Of course we don't have the perfect answer and that's something that's going to evolve over time. But we're trying to save lives. We're trying to save pilots and passengers lives from controlled flight into terrain. NASA's taking the risk, and the Air Force is taking the risk to put this system on, and I, I think, I really do, that it'll, it'll be game changing and save a lot of lives and really be a benefit to society and really show what's possible. I think it's a system whose time has come. There's no major obstacles from a technology standpoint. Um, it does take, it does take work, it does take development effort, it takes flight testing to prove it. The technology is out there to do this. It's a matter of bringing all the pieces together and sewing it together with software to make the capability happen. (sound of jet taxiing) (music plays)