Aerospace

AFRL’s Autonomous Aircraft Experimentation

Artificial intelligence (AI) is a hot topic, a powerful tool, and here to stay. Watch this video to see how the Air Force Research Lab (AFRL) is working to integrate artificially trained neural networks into the real world — namely into controlling an airplane. See some of the AFRL’s experimentation and why “collaborative combat aircraft in that type of autonomy is revolutionary and will be the future battlespace.”



Transcript

00:00:23 thank you our senior leaders have been clear and Direct in saying we're dealing with new technology and we're dealing with a new threat we need to go fast in determining the competitive advantage of autonomy and how to ultimately operationalize autonomy for the Warfighter we are we are trying to figure out General networks training a simulation

00:01:06 how to integrate those into the real world in this case integrate that into controlling an airplane we need to recognize that AI is here it's here to stay it's a powerful tool collaborative combat aircraft and that type of autonomy is revolutionary and will be the future battle space in order to deliver operational

00:01:39 capability we need to know initially what the user wants and so there's an element of concept refinement that needs to occur with our warfighters and with the technologists and the industry expertise that are available one of the things that I love about our country is the amount of effort that we put into our people and the value that our people have working on the automatic

00:02:00 systems we counted every life that had been lost over the decades to mishaps controlled flight in a Terrain collisions with other airplanes and each one of those lives I mean that was a person that was loved by many people so looking at Future capabilities where the human doesn't have to be at risk that is a huge benefit for this community oftentimes we'll talk about the

00:02:25 lieutenant taking these aircraft to war and we want to make sure that they are safe have a capability that's controllable and usable and reliable most importantly adding collaborative aircraft of any kind and autonomy capabilities ultimately are going to increase The lethality and survivability of their human operators in a high-end fight

00:02:42 so we have basically our expert Pilots expert Engineers who are effectively the teachers for this autonomy if you're staying on a football team they're the ones who are designing the drills giving you scrimmages and then putting you through a bunch of different repetitions of a slightly different scenarios so that you can handle that situation and then when you get to game day you're

00:03:00 going to be able to handle any situation that gets thrown at you printing these these neural networks takes millions and millions of runs of training runs you can't do that in a real airplane you know a flight hour 16 is tens of thousands of dollars but you can do it in a simulator a key to the whole non-linear approach

00:03:22 if you will is being able to take a AI powered system and in a model or in a simulated environment iterate on that way faster than real time like at the speed of light in the span of 24 hours we may have trained this thing many millions of times to do something that we've only seen once or twice in reality technology matures in the Sim we bring

00:03:46 it into flight onboard actual crude tactical platforms where the pilots can effectively go hands off we're teaching them in a simulator and then we're putting them in the real world and then seeing how that goes one of the advantages to test is our ability to experiment and demonstrate some of these capabilities in a controlled environment both so that we can get

00:04:06 important data and Lessons Learned out of it so we can develop our systems and risk reduce for future development as well as making sure that we're walking down the right Road one of the things that sandboxes like the Vista aircraft can do is allow us to demonstrate some of these Concepts or experiment with these Concepts show them working alongside fifth generation aircraft and

00:04:24 then we can develop our systems in the right direction with the test data to help the developers make the capability that we want it's engaging the simulation first we make sure all the Vista systems are engaged and healthy then we turn on the AI agent and let it have control of the aircraft good control request test point is loaded copy and clear and maneuver

00:04:48 initially it just flies straight in level and then it's three two one fights on you are clear to maneuver copy agent is the aircraft drop your foot's on you have a set of computers inside Vista and those computers can make the airplane fly like other airplanes you're flying this F-16 but it feels like you're flying an F-35 or a B-52 or a 707 plus we have multiple

00:05:13 data Lakes we can actually change the simulation while we're flying the airplane engine free test Point setup is complete you are clear to continue the emphasis there is simulation and Vista are not in States right they do not exist in a vacuum by themselves if they were this would be a science project right but instead they're Stepping Stones to a

00:05:36 future capability and that future capability will reside on these future combat collaborative aircraft you can use that to train artificial intelligence agents on board the X Cube 58. the xu58 it's a low-cost Runway independent air vehicle that is rocket launched and parachute recovered and initially controlled by a ground control

00:06:06 station two one launch launch launch however we're going to be able to switch over to artificial intelligence driven flight and in that case it's all up to the archers to fly it from there we're attempting to build trust in these AI agents we're attempting to ensure that its tactics and techniques and procedures follow certain guidelines and

00:06:32 Rules of Engagement and ethical considerations as well before we deploy it to be used in that combat collaborative aircraft maneuver think of this as the road to Future capabilities both on how we interface with these systems in our human machine interface onboard F22 and F-35 with our highly instrumented systems we can pull that data out and use it to further

00:06:53 mature and develop these systems we need industry working alongside Academia working alongside DOD in order to get us to the Future State allowing us to protect our national security interests against an adversary that does not share our values with the experimentation campaign we're taking the autonomous research Concepts that exists in a lab and we're turning

00:07:24 them into operational realities for the Warfighter we're taking that senior leader Direction and we're moving out