Poll: How Do We Train Pilots For Autonomous Aircraft?

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. Wait and see if any “autonomous aircraft” are ever sold–and if any ARE sold, what the FAA will require for training and certification. So far, it’s mostly talk and “vaporware.” So far, “other has over 50% of the responses, with 940 selecting that option.

  2. At the end of my tenure in the training department I noticed some very disturbing trends in the younger generation of new hire pilots. Very generally speaking:
    1. They are addicted to social media and their devices.
    2. They are distracted from a task very easily unless it is their device and media.
    3. They are lacking in communication skills and would rather text than talk. They are either very hesitant to ask for help or they totally rely on having someone else tell them step by step.
    4. Most disturbing is that they don’t want to know what is happening when they push “this button” only that pushing the button will, on a normal flight, give them the desired outcome.

    The problem is that when “the button” does not give them the desired result they don’t understand why. They applied rote memorization to the systems guide enough to pass the required test, when in initial training. They can repeat back to an examiner or on an examination the correct answer but they truly don’t understand why or how. Most of the time they know the right answer because someone in the class ahead of them told them the answers. With the very high volume of training occurring this is overlooked in an effort to get them in the seat.

    This is culturally different from previous generations where pilots were curious to learn more after they were online. In the past when an occasional pilot struggled with initial training we saw marked improvement when they returned for recurrent. With this generation, learning stops when they leave the center. When they return for that first recurrent they still have issues. I did seat support once for a retraining validation. This retest was after 3 days of additional simulator training beyond the recurrent footprint. The Pilot passed. During the debrief however, I became alarmed. I was told by this pilot that our training was bull…. because at the Regional, when there was an engine failure, they could always feel the Captain pushing the correct rudder pedal for them. This pilot claimed that this was more realistic because “that’s what would happen in real life”.

    In a normal world, pilots who have not learned to understand systems and procedures would just be stuck in the right seat for many years at the Regionals. However, because of the pressure to fill seats, the Majors have begun hiring pilots from the Regionals who have never been Captains. This cycle then repeats itself at the Major. Fortunately for us all there are two pilots in the Aircraft, and Airlines have not relaxed the standards required to be a Captain.

    So you might think, so what? This new generation will eventually absorb enough over time and grow to be good Captains. Not quite. Europe and Asia went to the philosophy of use the auto pilot and just operate by pushing buttons. This failure to understand was shown in the crash of Asiana Airlines flight 247 that hit the sea wall at SFO. This was a crew of “Veteran” Pilots who did not know enough about the aircraft systems to be able to visually land the aircraft without the use of the auto land feature. Yikes!

    You may say, “Well Airlines in North America have Captains and the standard has not been lowered.

    The best example I can think of is Air France 447 that crashed in the Atlantic. Captain in the back of aircraft on a break. Two younger generation pilots flying. Pitot static system failed due to icing. Aircraft gave erroneous commands. Pilots could not figure out that they were stalling the aircraft. Pilots did not communicate and work as a crew. Pilots did not understand the systems enough to simply figure out which set of instruments was correct. They had minutes to interpret and mitigate the problems, which in an aircraft is a really long time.

    The Captain was finally able to gain access to the flight deck and immediately knew from the instruments that the aircraft was stalled. However it was too late and they pancaked into the ocean killing all 228 people onboard.

    There are proponents of autonomous aviation wanting to move in that direction. The big question is the one asked here. Who will control the aircraft when things fail and how were they trained?

  3. I concur with ttuite767, the thought of autonomous aircraft carrying passengers is terrifying. FAR 121 requires two pilots for a very good reason: redundancy. Transport Category certification demands redundancy of all critical functions as a bedrock principle; two engines, two pilots, two ADCs, etc. Everything can (and eventually will) fail. True safety requires maintaining this philosophy of redundancy. Safety is forward looking, not backward looking. Safety is not measured by how many accidents you have or haven’t suffered in the past, it’s all about how much safety margin you have for things to go wrong and you still have enough slack to trap errors before they metastasize into accidents. All the proposals I have heard for “autonomous airliners” ignore this principle and assume that the engineers are smart enough to foresee and plan around all possible threats. Hogwash! There is NO safe substitute for well trained humans (plural) in control on board. Never buy a ticket to ride on a drone.