Breakfast Bytes Blogs

Yesterday's post DATE: What Is Single Pilot Operation? Airbus Explains was the first of two about the Airbus keynote at DATE. After a pre-recorded presentation there was a live Q&A. The questions covered a lot of ground and I thought it was worth trying to reproduce in full.

Q&A

After the pre-recorded presentation, there was a live Q&A with Pascal. I've written this section as if the questions and answers are verbatim, but I'm working off my notes and so these are not perfect quotes.

Q: It seems that the environment for flying is less complex than the environment for a car in a crowded city. Or maybe in the future with air-taxis, too, it will be different. Is that a fair assumption?

In the air, apart from birds, we don't have any objects to worry about and we have air-traffic-control (ATC) to guide us away from objects. Also we will always have a pilot on board to make an avoidance maneuver. We are working on adding a third barrier of safety to detect objects on the ground for the taxi phase where there is a higher probability that two planes come close or a vehicle crosses the path. We are looking at using something similar to cars with lidar, radar, and cameras. But yes, the environment is simpler. The aerodynamics is more complex than a car, with a third dimension, but the road can be trickier for a car. In terms of air mobility (taxis), we need to look at the numbers. If there is no pilot on board, that can reduce cost by 25% (not the salary of the pilot but reduced capacity of the vehicle). Likely we will have some corridors for air taxis from one heliport to another. It will be unlikely we will have an air taxi coming to your house to pick you up, instead it will take you from one regulated place to another. One idea is to equip railway stations, there are several in Paris, and create a predictable environment. But yes, I think the environment is simpler in the air than on the ground.

Q: Is a plane with a single pilot really safer than a plane with no pilot?

We are still convinced a human can invent a solution better than something a machine will invent by itself. In the literature, it is 2045 when maybe computers are better than humans. After that perhaps we will see if we keep a pilot on board but it is not in our plan. We don’t plan that far ahead when we don’t know if the technology will come or not. Our ambition is for one pilot to be as safe as two, and significantly better than today.

Q: In SPO, you mentioned the importance of human-machine interface. Can you give more details on how humans and autonomous systems will be synchronized?

We have to give time to the crew. If most of the handling of the aircraft is done by the system, then we can use the display to show the most critical information for operation. Whereas today we show altitude, velocity,and so on all the time. If they are not relevant to the crew there is a shift in priority. When the pilot is in cruise they have nothing to do for 1, 2, or 3 hours and it is a challenge to get them engaged and not completely disconnected mentally. We need to engage them with the interface. When the pilot is acting on a parameter we ensure that the parameter is on the display. This kind of principle. There the workload for the crew, it needs to be not too high, but not too low.

Q: What about the certification process, getting to SPO, and then no-pilot?

Again we are not targeting zero pilots. But 2 to 1 is already a certification issue since the certification is already written as “you shall have 2 pilots” and even when not explicity stated, it is implicit. We will need to address regulations, but before that, we have to be confident. Once we are confident, we believe we will be able to share that confidence with the authorities. The aircraft has to be safe. Regulation is an issue but just one of the multiple issues we have…but we are engineers, we solve issues.

Q: What about reliability?

Planes are fly-by-wire. If that is lost, the airplane is lost. So we are used to having a system with five computers and just one is enough for flight and landing. We have two types of computers, so even if one type is failing the other will be there. Making a system that is flight-operational we are used to doing. What is more tricky is to handle all single points of failure for not just electronics but all the sensors outside the plane. Also, we need to look at security issues like spoofing or jamming of GPS. Cameras and image processing to detect where the runway is so we will normally have two different systems to navigate (ALS or GPS, and vision), and also we have the pilot.

Q: In SPO, what happens if you are mid-Pacific and the pilot passes out? What is the backup? Tethered operation?

If literally in the mid-Pacific, we will have to pilot for a long flight, so we think SPO will come first for shorter flights. But yes, the plane will be autonomous. We don’t intend to control the aircraft from the ground, just be able to upload a flight plan from the ground to the plane. Waypoints including the airport itself. But not actual control from the ground, since first there is a latency, and second it is hard to detect from the ground if, say, “turn to the left” is safe or not.

Q: Could the increase of autonomy and reduced need for human action result in less experienced pilots?

Right now,  a pilot is flying perhaps two minutes per flight so already very little of the flight time. Their experience comes in two forms. One is handling the aircraft, feeling the acceleration and so on, and this today not coming from flying a large plane like a Boeing or Airbus, it is on the simulator they keep their competence. Then there is basic airmanship, to understand the situation: storm coming, sick passenger, and this will be pretty much the same. In terms of training, they may lose a little bit, but airmanship will be the same basically.

Q: Is there a fundamental difference between military and civil aviation? The military controls drones controlled from ground.

Military drones, if they crash, it is not normally not a big parable, just the cost. It is not the same with a civil plane. On the military side, for communication, there is a very secure network and it is a very limited zone of operation, but civil aviation has to go everywhere. We cannot bank on always having communication, so we want to extend our fly-by-wire inside the plane, not to the ground.

Q: What similarities and differences do you see between aviation and automotive?

One similarity is that we will use automotive components and we will be happy since they will be cheaper than a pure aviation system. Everything needed for the taxi phase we will likely use lidar and radar and the range is similar for a car. Image processing algorithms, classical image processing, and machine learning use a similar framework as in an automobile. After that, we have the third dimension so that is obviously different. In terms of human monitoring, to detect if a pilot is incapacitated, we also have in some cars already.

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