Breakfast Bytes Blogs

The final keynote at this year's DATE was by Pascal Traverse of Airbus, titled Autonomy: One Step Beyond on Commercial Aviation. This was part of this year's DATE's two-day special initiative on Autonomous System Design (ASD)...not to be confused with Aero-Space and Defense despite the Airbus name. Pascal was talking about increasing autonomy on future Airbus aircraft.

Today's post will cover his presentation. There was then a lengthy and perhaps even more insightful Q&A which will be tomorrow's post.

Pascal started with an overview of Airbus. It is, of course, most well-known for its commercial aircraft. Some, like the medium-haul Airbus 320 are extremely commercially successful; some, like the enormous double-decker Airbus 380, not so much. However, it also works on satellites and air-taxis (giant drones).

Airbus's forecast is that, driven by the growing global middle-class, there will be a doubling every 15 years in air traffic (after the pandemic is over). This will also result in a pilot shortage although, as Pascal admitted, "it seems a bit absurd to discuss this when so many are furloughed right now". Obviously, one way to address this is to train more pilots. But there is another trend he pointed out:

On the left is a Caravelle, built in the 1950s by Sud Aviation, a forerunner of Airbus. It has a pilot, a co-pilot, a navigator, and an engineer. With increased automation of the navigation function by the commercial aviation ecosystem, the navigator went. With increasing automation of the flying with improved autopilots and fly-by-wire, the engineer went, and then there was just the pilot and co-pilot (officially the first officer). All modern planes are fly-by-wire, not just the Airbus since the A320, but also the Boeing 777 and 787. One difference is that Boeing has kept the yoke, the steering-wheel-like control in front of each pilot, whereas Airbus switched to something more like a video-game joystick to the side of the pilot (you can see one in the top right-hand picture above).

I remember experiencing the combination of automation and fly-by-wire myself back in the late 1980s when I was flying from Nice to London on an A320. It was extremely foggy (yeah...London in winter) and we were the first plane to land at Heathrow Airport that day. After we'd touched down, the captain told us that it had been a fully automatic landing with the pilots doing nothing beyond paying attention in case something broke. Unfortunately, my colleague flying in from  Munich was not on an aircraft equipped with automatic landing so I had to sit in the airport drinking coffee while he sat in a holding pattern drinking coffee, and we both waited for the fog to clear. On a different flight on an A320 from Nice to London, I think the crew was bored. In those innocent pre-9/11 days, they invited anyone who wanted to come and chat to them in the cockpit, so I did. They turned the autopilot off to show off their video-game controllers and did a few turns. 

The obvious next trend is to increase the level of automation. Pascal was already at Airbus in 1984 when the A320 was launched, and was already fly-by-wire (as was Concorde, by the way). He reminded us that the processors of the era were the Intel 8080 and the Motorola 6800. So even within existing models, the levels of automation have improved as the electronics have been upgraded.

Single Pilot Operation

The next step would be single pilot operation, or SPO. Apart from the potential of alleviating the anticipated cockpit crew shortage by increasing the number of available qualified pilots, this would save the airline industry about $15B per year. However, Airbus is not planning full autonomy. As Pascal put it:

We do not foresee flying commercial aircraft without a pilot on board. We keep a human in the loop and all the systems we are working on are focused on the crew.

He also had charts of aircraft showing that the more automation the safer they have become. You won't be surprised that he didn't mention the embarrassing incident in 1988 when a demonstration flight of the Airbus A320 crashed in a forest just outside Mulhouse airport doing a demonstration fly-by with landing gear down for an airshow taking place there. It was the A320's first passenger flight, although most of the people on-board were journalists and raffle winners (they all survived the crash, but a couple of people died from smoke inhalation). The reason for that crash remains controversial, but my memory at the time—I was living in France, so it was big news—was that the pilots were doing the planned fly-over, but the automated fly-by-wire systems figured that with the gear down and the flaps down, they were trying to land. There is a detailed Wikipedia page about it, Air France Flight 296.

This is a mockup of an SPO cockpit, what he calls a "minimum viable simulator". On the right, a wooden mockup to show to test pilots and other "customers". On the left, an actual SPO simulator.

One feature that needs to be addressed is that in an emergency, it cannot guarantee to rely on satellite navigation, nor every airport (and runway) having an instrument landing system (ILS). So they are also working on image processing. For this, Airbus is relying on image processing and it is borrowing what it can from other industries: lidar, radar, and cameras from the automotive industry. This is also required for automating taxiing, where, as he put it, a "passer-by is more likely wander across". He showed us a video of the first time an aircraft took off without human intervention. I was surprised that, having experienced automatic landing 30 years ago, that takeoff had never been automated before.

He wrapped up the pre-recorded video part of his presentation with the conclusion that:

We are likely to get to single pilot operation (SPO).

Tomorrow

The live Q&A that followed this pre-recorded conversation.

Sign up for Sunday Brunch, the weekly Breakfast Bytes email.