Autonomous driving is currently one of the hottest topics in the field of artificial intelligence. However, a much less talked-about mode of transportation that could also become autonomous in the future is flying. The airline industry is known for its low profit margins, and in 2012 airlines made an average of $4 per passenger carried. It thus makes sense that airlines always try to find ways to cut costs and one of the ways to do that is by cutting the number of crew members per flight. The number of crew members in the cockpit has already declined over the years, but will we soon see commercial airplanes without a cockpit at all?
During the 1950s, when commercial flying began to grow rapidly, there were five crew members in the cockpit: a pilot, a co-pilot, a flight engineer, a navigator and a radio operator. As technology improved, the tasks of the navigator, the radio operator and the flight engineer were replaced by computers. And with the introduction of the autopilot, constant hands-on control by a human was not necessary anymore. It is however important to note that the autopilot cannot replace human operators, as it merely assists them in controlling the aircraft. The pilot and co-pilot have to constantly monitor and adjust the aircraft’s navigation and systems, communicate with air traffic control and prepare for the next phase of the flight.
Tests have been done to reduce the number of pilots from two to one, but this reduction is not expected to come in the very near future. If the reduction from two to one is already such a difficult task, how likely would it be to go from two pilots to no pilots at all? Last June, Mike Sinnett, Boeing’s vice president of product development, announced that Boeing wants to test aspects of self-flying passenger aircrafts next year. He also stated that the basic building blocks of the technology are already available. These are however only tests on the aspects of self-flying aircrafts and an aeronautics professor at the Embry-Riddle Aeronautical University in Florida says that even optimists predict that pilotless flying is at least 50 years away from now.
However, even if autonomous flying would be possible, it would be difficult to convince people to actually fly with an autonomous aircraft. A survey of 8.000 people done by UBS found that only 17% of respondents would buy a ticket for a pilotless flight. Nevertheless, I think that this percentage could rise with the commercialization of autonomous transportation for other modes of transport. For many people, autonomous transportation still sounds like something from movies, but if there are more and more autonomous cars on the road, the consensus on autonomous transportation including autonomous flying could drastically change. Even though it seems likely that we will not see any pilotless commercial flights in the next 50 years, I am wondering what your current thoughts are on getting on a pilotless flight. Do you feel safer on a flight knowing that there are human operators on board or do you think taking out the human error factor makes flying safer?
[1] – IATA. (2013, March). Strong Passenger Growth Trend Continues. Opgehaald van IATA: http://www.iata.org/pressroom/pr/Pages/2013-07-03-01.aspx
[2] – Kottasová, I. (2017, August). Pilotless planes could save airlines billions. But would anyone fly? Opgehaald van CNN: http://money.cnn.com/2017/08/07/technology/business/pilotless-planes-passengers/index.html
[3] – Lerner, P. (2017, August). Would You Fly on an Airliner Without a Pilot? Opgehaald van Air & Space Magazine: http://www.airspacemag.com/flight-today/02_aug2017-airplanes-without-pilots-180963931/
[4] – Scott, A. (2017, June). Boeing wants to build a futuristic plane that can fly without a pilot. Opgehaald van Business Insider: http://www.businessinsider.com/r-boeing-studies-pilotless-planes-as-it-ponders-next-jetliner-2017-6?international=true&r=US&IR=T
Hi Oscar,
Thanks for your informative blog post. For my bachelor’s thesis, I did some research at KLM’s Flight Safety department and I wanted to share some things I learned over there in this comment.
You say that “the number of crew members in the cockpit has already declined over the years”. While that’s partially true (for example, as you stated, the Flight Engineer isn’t needed anymore in modern planes), it’s good to know that the number of pilots is mostly determined by the duration of the flight. If crew members experience fatigue, they have to report this by the means of an ASR (Air Safety Report), and the airlines have to reassess the number of cockpit crew for each flight periodically, inter alia based on the number of fatigue reports.
What’s more, some ATC clearings are already processed digitally! The route where this is used the most, is the transatlantic cross. KLM actually explained this system really well in a video: you can find it at this link –> https://www.youtube.com/watch?v=fwqEe0XkggI#t=1m20s . Computers would be perfectly able to take over entering and receiving this data, as it is done almost automatically already – the pilot just has to push the enter button.
However, the biggest role a pilot can play is in case of emergencies. People can show “good airmanship” through extraordinary maneuvers (I’ll list two at the end of this comment for the enthousiasts!), and computers either have to be
a) programmed to do so, or
b) have learned this through e.g. reinforcement machine learning
Luckily, not a lot of data is available to learn from, so you’ll have to stick with programming the computer to handle every single scenario. This actually is not an option. Airmanship can’t be taught to computers, it just arises from experience. If things were to go horibbly wrong, there’s always a *small* chance that the pilot might save the day (and the plane, and the passengers). I think computers would not have a clue on what to do.
So, for the time being, I think pilots will really be needed in case of emergencies. And you’ll want two of them, since the workload in the cockpit can be quite high in times of emergency. One of them will still have to do the radios, and the other one will have to fly the plane.
Now, as promised, I’ll link to some videos of extraordinary feats that pilots managed to achieve – and what computers probably could not have done.
– US Airways Flight 1549: after losing all engine power, the pilots were forced to land the plane in New York’s Hudson river: https://www.youtube.com/watch?v=4v7kASXPQMc
– A DHL Cargo Flight: after being shot by a missile, the pilots managed to land the plane using only the engines as controls: https://www.youtube.com/watch?v=GQbsceR1DiE
Hi Laurens,
Thanks for your great comment, it’s definitely a good addition to my blog post. I agree with your examples of why we still need pilots in the cockpit and at first I thought of including the example of the Hudson River incident. However, I wanted to keep the discussion of pilotless flying open which is why I didn’t include it.
Anyways, there’s a big chance that you’ve seen the movie Sully already, but if not you should definitely watch it. I think it’s a great example of why we are still far away from having autonomous commercial flying. After the engines got hit by birds, the pilot decided to land the plane in the Hudson river. And even though the first simulations were stating that the plane could have made it back to a nearby airport, further analysis showed that it would have crashed on the way back to the airport. The reason that everybody survived was because of the judgement of an experienced pilot and this is something that is extremely hard to replace by A.I.
Hi Oscar,
Thanks for your interesting article about pilotless planes. I agree with you and Laurens that the problem is two-sided. First, for now it still is a technical limitation, which is so far not solved. However, I disagree with Laurens that airmanship cannot be taught. With the current developments of AI, we are getting close to this. I believe that within ten years AI software will have a least the same ability to make decisions as pilots have (researchers believe it will be possible between nine and 45 years [Gent, 2017]). Image a tool that could absorb all the recorded data in airline history. This would make a pilot not standing a chance. Things like emotion, state of mind, response speed and alertness all suddenly would become irrelevant.
The second problem, like you already stated, is obvious the willingness of people. If we assume that the technological advancements will be there, then the people still need to be convinced. This is something I am neither concerned about. First, the price. Like you already mentioned, pilotless flying will be cheaper, since labour costs will decrease (Gibbs, 2017). This probably will result in lower ticket prices and will trigger people, since people are to a certain extend price sensitive.
Besides that, like will all new things, pilotless planes will also encounter the innovation adaption lifecycle (Comindwork, 2016). Which means that in the beginning, most of the people are not interested or even refusing it, but there already is a small group of early adopters which will start to adapt it. When the majority will see that pilotless planes are as safe as staffed planes, they will adapt it as well in time. This is a very natural process.
Concluding, I would say that people should never underestimate the pace in which technological developments take place. We might be sitting in a pilotless plane sooner than expected!
References:
Comindwork (2016) Innovation Adoption Lifecycle from Innovators to Laggards, http://www.comindwork.com/weekly/2016-02-29/productivity/innovation-adoption-lifecycle-from-innovators-to-laggards , 28th of September 2017.
Gent, E. (2017) When Will AI Be Better Than Humans at Everything? 352 AI Experts Answer, Singularity Hub, https://singularityhub.com/2017/07/25/when-will-ai-be-better-than-humans-at-everything-352-ai-experts-answer/ , 28th of September 2017.
Gibs, A (2017) Pilotless planes could save airlines $35 billion, UBS says, CNBC, https://www.cnbc.com/2017/08/07/pilotless-planes-could-save-airline-industry-billions-of-us-dollars-ubs-note.html, 28th of September.