Toward The Optionally Piloted Aircraft

Aviation Week and Space Technology, the industry’s leading magazine, has been publishing “pilot reports,” on new aircraft for decades. Its pilot report on an aircraft called Centaur ⁠was the first in which the pilot doing the test never touched the controls. Centaur is an optionally-piloted aircraft, or OPA.

The reporter conducted the test while sitting in the back seat of the small, twin engine aircraft. Up front sat a person acting as the safety pilot, his arms calmly resting on his lap. Sitting beside him, in what is ordinarily the co-pilot’s seat, was an engineered series of linkages, actuators, and servos. The safety pilot pulled a lever to engage the mechanisms, and they began moving the pilot’s control stick and pressing the rudder pedals. The actuators are double and redundant; if one set fails another will immediately take over. The safety pilot can disengage the mechanism with a single pull of the lever if something goes wrong; unless something goes wrong he does not touch the controls.

In the back seat, the “operator,” commanded the plane through a laptop, using an interface identical to that of the ground control station for an unmanned vehicle. Through the screen, he could change altitude, fly to waypoints, takeoff or land. Pushing the “launch” button began an autonomous takeoff. The computer held the brakes, pushed the throttles forward, checked the engines and instruments, and released the brakes for the takeoff roll. The plane accelerated, took to the air, and began to climb out on a semi-autonomous flight.

As an OPA, this aircraft can fly in a normal mode under pilot control, as a certified aircraft as though it was fresh from the factory. But it can be flown in this semi-automated mode, as the reporter issued high-level commands through the laptop in the back seat. Or it can be flown unmanned, with the same computer interface operating on the ground through a radio link. It is called “Centaur” after the mythical half-human half-horse creature.

Diamond DA-42 [Image Source]
Engineers at Aurora made Centaur by converting a commercially-available aircraft, a Diamond DA-42 to this OPA mode. They added the mechanism in the front seat, a host of independent electronics, and digitized the flight manuals and emergency procedures into computer algorithms.

Unmanned aircraft are currently mostly illegal in the US national airspace, but Centaur is certified by the FAA to fly in its unmanned mode if a safety pilot is aboard. Hence the aircraft can help develop sensors, algorithms, and procedures for unmanned systems. It has been flown with a pilot aboard to Alaska, where it has permits to operate in an unmanned mode to collect data for climate research.

Centaur is a transitional technology, suitable for flight testing and engineering development while regulations and techniques are worked out. It contains within it, however, the seeds of a new way of piloting, for eventually, even in US airspace, the front-seat safety pilot’s job can be transformed to one operating through the laptop interface.

Will the Future Include Human Pilots Inside the Cockpit?

DARPA ALIAS artist’s rendering shows tablet and computer, controlling a robot, controlling the aircraft.

Are we moving into a future of airliners without pilots? Probably not for the foreseeable future, but Centaur shows us how much of the technology exists today. A DARPA program called ALIAS, seeks to extend the OPA idea to any aircraft by building a general-purpose robot to sit in the right seat. In a technical sense, the automation to taxi an airliner, command a takeoff, follow a route and autoland is all well proven. The unmanned aircraft problem appears to have been solved — but only when artificially divorced from its human context.

But many examples from automation and robotics have taught us that fully autonomous operation is the lesser problem. We can say that the technology exists today, but for the innovation to have a social importance and contribute to human welfare requires not only the machines but the social, policy, and economic systems to situate these machines into human life and to enhance our practices and our experiences, a much more open problem.

Airliners need to be safety certified so they don’t fall on people. They need to provide not only statistical safety for passengers, but also the experience of safety. They need to operate not only in every emergency we can think of, but in nearly all of the emergencies we can’t think of. This is why, when placed within human settings of reliability, risk, liability and trust, the unmanned aircraft problem, like the driverless car problem, has not been solved. Building the trust in such systems will require years of demonstrations, operations, and smart engineering to prove reliability and work within human environments.

David A. Mindell is a professor at MIT and founder and CEO of Humatics, which is developing technologies for trusted, transparent autonomy situated within human systems. He adapted this article from his book, Our Robots, Ourselves: Robotics and the Myths of Autonomy (Viking/Penguin 2015).

25 thoughts on “Toward The Optionally Piloted Aircraft

  1. There has been frequent discussion regarding the use of a standardised unmanned “pilot” that could be certified and then used over and over again in different aircraft to enable them to be optionally manned and also to split the certification of a computerised pilot from the certification of the airframe. Unfortunately this almost never works out as the integration cost of all the sensors and processing required adds complexity that is not predictable in its impact, scope, and cost.

    I think the approach has merit. It would help if there was any agreement within regulatory bodies regarding what sense and avoid standards will need to be met and in what manner the computerised “pilot” would need to demonstrate equivalent or better functionality to a manned pilot. The sad fact manned pilots, their unpredictability, and their ability to fail are assumed in the system. The same will not be allowed for computerised “pilots”.

    1. I was in a plane that landed flying at around 15′ yaw into a strong wind. Having built AI based platforms, I know there is no current safe way to integrate this kind of counter intuitive behaviour where it doesn’t pose a risk to passengers.

      People forget that Google buried the story of how the autonomous cars almost killed passengers numerous times.
      I’d rather pay an extra $5 and have a well rested human heuristic doing its job.

        1. It was an interview with a Google Street View car driver a few years ago, and it is no longer easily searchable on YouTube… The incident occurred travelling on a highway through a canyon (note GPS is often unreliable under such circumstances), where the guidance system interpreted a passing container truck as a canyon wall. Accordingly, the panicking driver was fighting the autopilot for directional control to avoid the incorrect course.
          There is no way current generation AI could cover such an improbable edge-case scenario.

          “Citation needed” tends to apply to published assertions where a chain of accountability is necessitated to reduce baseless rhetoric. Yet “close calls” did not seem to encourage rational risk assessment with Google’s management or the general public ( ).

  2. I’ve seen previous designs where you actually need to inflate the pilot using your mouth. Having to instead pull the pilot’s lever with your hand seems to be a bit of an improvement.

  3. Pilots train in full motion simulators do they can feel and see what the airplane is doing in bad situations. To have a pilot at a desk “fly” an aircraft I in a bad state and with no luck is a recipe for a lost aircraft.

    How many failures happen? Not many, one or two every couple thousand flights maybe. (Every couple weeks)

    Imagine an unresponsive pilot and a bad communications link.

    1. You’re confusing remote control with autonomous piloting.

      It’s not a pilot at a desk, where he/she pushes the stick and the robot arm follows in the plane. They program or command an end point and the machine goes through the steps in between.

      When someone gives directions to the store, they won’t tell you to lift your right leg, move it in front a little bit, then make it touch the ground again.

  4. My understanding is that in Australia the train have controls that address the conductor becoming unalert on long runs. The conduct is required to hit a button in certain time person. Failure to do so results on the train automatic all stopping. Not practical in an aircraft, but perhaos in can be set so the robot responding to air traffic control for landing at the closest available airport. INMO just another attack on labor for the profit for the few. The most important component in business, the customer isn’t going to see any benefit. This isn’t using robotics to construct consumer items that can’t be practically done with human labor. Like the Camera-PDA- photocopier-telephone that’s in my front pants pocket. Robotically controlled Private motor vehcle have the potential to mange traffic congestion, and make surface travel safer. This could possibly increase air traffic congestion, because it could increase the number fro private air craft in the sky, for those who could afford to own one. This is backwards, the robot should be monitoring the aircraft and the pilot, and alerting the pilot, as many now do. BTW any reason while the depicted relaxing pilot appears to be female? AFAIK male pilots are still predominate in both commercial and private aviation. Perhaps it’s because I live where the owls date chickens because there aren’t that many owls to date, I never seen a male where his hair in a bun.

    1. Robotizing the aircraft could also safely decrease minimum separation, which is what it is because humans simply aren’t fast enough to fly the planes like that and ground control doesn’t have data accurate enough to guide them with closer separation…
      Smaller minimum separation = more planes can “fit” into the airspace.
      Also, machines would probably be superior at maintaining optimum flight parameters (like a quick and yet safe approach), leading to better fuel efficiency = cheaper and faster air travel, if that doesn’t benefit the customer, I don’t know what will :P

      1. The required separation between planes due to vortices and turbulence means that you can’t really cram them in close like cars on a highway. They always have to be far enough apart unless they’re going in a planned formation.

        As long as this is the limiting factor, human reaction time is plenty enough.

        1. Not quite. At least in the U.S., separation minimums are based partly on the distance needed to safely route aircraft by completely manual means in the event of radar failure, which happens sometimes because enroute radar is often of Kennedy-era vintage. Another issue is the resolution of said radar and the accuracy of altitude transponders.

          A decade or so ago they introduced Reduced Vertical Separation Minima for the flight levels, which required re-equipping and certifying aircraft with more accurate altimeters so they could reduce vertical separation above IIRC 21,000 feet. In the en route phase, turbine aircraft in the flight levels are almost always on autopilot anyway, in large part due to the fact you’re working your way towards the “coffin corner” where stall speed and critical Mach number converge.

  5. It’s like the old joke – in the future, the flight crew will consist of a pilot and a dog. The dog is there to bite the pilot if he tries to mess with the automatic controls.

  6. People will fly in a plane without a pilot, but not one without a captain.

    There will probably always be one person ‘at the helm’ even if they’re not actually doing anything because people need to know that someone else is in charge even if they’re not in control.

    1. Airliners now spend most of the route on autopilot, although I believe takeoff and landing are manual, even though they don’t need to be. AFAIK airliners have that automated, but nobody tends to use it.

      I think a human does always need to be ultimately in control. That’s a pilot, who can fly the plane if he decides he ought to. Sure, people are fallible, and pilot error is often the cause of a crash. But I think I’d still rather have a human go wrong and kill me than a machine. Something to do with the inevitability of it, at least a man would have tried. I dunno, it’s one for the sociologists.

  7. Wouldn’t it be much more sensible to build a sort-of OBD-II into the planes, rather than have mechanisms pulling the levers? Seems stupid to go computer -> mechanical arm -> mechanical lever -> actual control. I know in smaller planes “fly-by-wire” means having cables attached from the control surfaces to the joystick. But automatic controls should be as close to their effect as possible, no need to adapt them to controls adapted to humans.

    I suppose this experiment was done this way because it didn’t require the plane to be modified. Because it’d need expensively re-certifying. The experiment is just a proof of concept. But most people fly by giant computerised airliner anyway, that’s where the difference would be made. It’s pointless developing a physical robot pilot when you could just plug the laptop into the plane’s computer.

    In fact you don’t need the laptop either. This is just a matter of letting the plane’s autopilot do more than it currently does. A software matter. And more importantly, regulatory. I suppose that doing testing on an Airbus would be more expensive than this method, hence all the clutter, but it’s really the silly way round.

  8. I am afraid we don’t have the required technology yet. For a pilotless aircraft to be as secure as I imagine today aircrafts are they need to be fully autonomous. There MUST NOT be any way to control it remotely (other than providing some information like weather or which runway is free to land). Today, crashing a plane requires a “volunteer” on board. It happens, sometimes even pilots go mad, however the cost is very hight compared to opportunity of doing it remotely with feet on the ground. And I do not belive we’ve got communication protocols secure enough (including a human factor) not to be broken.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.