Air Force Tests Robot Pilot


The Air Force has flown a Cessna 206 with a robotic pilot it says can essentially turn an aircraft into a drone and then get turned back into a manned aircraft. The ROBOpilot flew a two-hour autonomous test flight at Utah’s Dugway Proving Ground Aug. 9 handling the takeoff, flight maneuvers and landing without human intervention. “Imagine being able to rapidly and affordably convert a general aviation aircraft, like a Cessna or Piper, into an unmanned aerial vehicle, having it fly a mission autonomously, and then returning it back to its original manned configuration,” said Dr. Alok Das, senior scientist with the Air Force Research Lab’s, or AFRL’s, Center for Rapid Innovation, in a statement. “All of this is achieved without making permanent modifications to the aircraft.” DZYNE Technologies is partnering with the Air Force in developing the kit.

The machine mimics human interaction with the yoke and pedals, along with all the knobs and switches on the panel, even watching the gauges. “At the same time, the system uses sensors, like GPS and an Inertial Measurement Unit [essentially a way for a machine to locate itself in space without GPS] for situational awareness and information gathering. A computer analyzes these details to make decisions on how to best control the flight,” AFRL said in a statement. The accompanying video shows the robot making the corrections necessary to stay on the centerline during takeoff and maintaining runway heading after takeoff. The robot bounced the landing but recovered.

Russ Niles
Russ Niles is Editor-in-Chief of AVweb. He has been a pilot for 30 years and joined AVweb 22 years ago. He and his wife Marni live in southern British Columbia where they also operate a small winery.


  1. This yet another step along the route safer travel, if used for good purposes, but it could as well become a poor man’s flying bomb, by cramming it with explosives.

    As it is it seems to be very heavy, as most of the interior is missing in the plane.

    Not that long ago the driver of a car had to control the ignition timing manually, and adjust the fuel mixture, just as many pilots flying older aircraft do to this day. No modern car allows you to do that!

    Many modern cars have anti-lock brakes and some have collision-avoiding systems, and the gearshifting has been taken over by electronics in almost every car, which some claim takes the thrill of driving out of the driving, so some can even imitate manual gear-shifting.

    Even a simple autopilot takes over many of the pilot’s chores, is that a bad thing?! Been in use since WWII.

    Personally, I am much more worried about all these autonomous cars that will tour our cities in a vain hope to find a free parking spot, cramming the traffic further and cruising around hour after hour, spreading toxic fumes and noise.

  2. Looks like it needs a lot of refinement – which I’m sure it’ll get. It’ll get smaller, lighter, cheaper, and more capable. This kind of thing is inevitable – the alternative is to try and end progress, which ends up being worse. There will be challenges to our current way of doing things, which will have to change. The end of the human race as we know it? Of course. Just like agriculture and the industrial revolution, this kind of automation changes everything. Doesn’t mean we have to crawl into a hole somewhere and die.

  3. When I read this it makes me think of the movie Big, when Tom Hanks character Josh says “I don’t get it.”.

    Hopefully, there’s a follow-up piece on why turning an old Cessna/Piper into a UAV, and back again, is a good, practical idea, from an applications standpoint.

    Captain Obvious here: If there is greater merit to it – why not start by using existing technology, installing a Garmin G3, three-axis autopilot, autothrottle, plus some software and you’re 85% there?

    If not, then perhaps the follow-up can be on who sold this idea to the Air Force – they should be hired by one of the 2020 presidential candidates; or maybe on who received the country club membership in return for this contract.

    Now, if the ‘robot’ walked up to it, hopped in, could operate the primer, and get her started in less than three tries…. then it would seem like 2019 technology.

  4. This is old technology. I have seen many demonstrations with a machine that flies the airplane much smoother. The reaction times are faster, with a computer that calculates and processes the different variables at a much higher rate. It is even equipped with an audible sensor to detect engine parameters that might not show up on the gauges along with a device to detect fuel leaks and exhaust gas emissions. We call it a pilot.
    Why all the smiles on the teams faces. Is it because they came so close to crashing, but didn’t. Why is everyone trying to develop robotic cars and now airplanes? Is it because they themselves can’t drive a car or fly an airplane? We have the resources already in place for those that can’t do these things. They are called airlines, buses, and trains. You would think one of the easiest things to automate would be a cargo ship, but to my knowledge there aren’t any transoceanic ships plying the seas without a crew.

  5. “All of this is achieved without making permanent modifications to the aircraft.”

    So, no stinkin’ STC or even nary a logbook entry; just throw it in and you’re good with the FSDO!