Conventional pilotless airliners could have ‘too easy’ mistakes made with humans, warns expert
A high-tech flying machine that adapts to the people in its path and could make “too easy” mistakes when flying without a pilot would revolutionise flying, a leading air safety expert has said.
Regulatory bodies should give commercial flight plans permission to include a computerised “no pilot” mode to make it easier for small planes and drones to fly with fewer safeguards, according to David Learmount, the editor of Flightglobal’s Fleets Analyzer.
Learmount, who has more than 40 years’ experience as an aeronautical accident investigator, said: “Autonomous systems, when they are properly implemented, do not change our view of pilots being competent.
“However, once they’re incorporated into aviation systems, autopilot and autonomous systems interact with the pilots and have the potential to make errors we see when there’s a lack of self-awareness from pilots and pilots make errors.”
Learmount called for computers to get permission to fly planes with a “no pilot” mode so crews could focus on safety rather than worrying about whether they were flying in their “main” cockpit.
His comments were made in a blogpost on airnews.co.uk.
Learmount said: “Let’s put safety first and amend the legislation to allow the automation of pilotless flights – not for the benefit of robot pilots, but for the benefit of the pilots, whose workload will be lower.”
The idea of letting autopilot take over flying could increase air safety by making it easier for small aircraft to get airborne.
About 250 aircraft are capable of functioning without a pilot, according to Professor Tony Long, head of civil aviation research at Imperial College London.
Some of these include Civil Air Patrol planes, like the 662E, a Viking S-bit composite radar-equipped patrol aircraft designed for quick responses to radar operations and other missions, which have already been deployed in the Caribbean and parts of East Africa, he said.
However, Long told the BBC: “Flying planes with no human element, with autopilot, sounds very intelligent but it doesn’t sound as self-aware as having a human in the cockpit. What the airlines are going to have to decide about is whether they want to go through that regulatory process, which will be very, very complicated, to allow the automation of operating a very small aircraft.”
Learmount said the best case scenario was a study from the UK regulator, the Civil Aviation Authority, which concluded that autopilot was “almost invisible” to the pilot, adding: “It is almost like there is a robot pilot, but you don’t know it is a robot pilot.”
He added: “Autonomous systems, with their sensors and their systems, can make a very, very precise decision about what to do next and what to do as the situation develops … and it is very difficult for pilots to figure out how to do that.”
But he acknowledged that if air traffic controllers were to lose their ability to control aircraft, mistakes could result.
“There is a risk that if we fall short in design for safety by a step or two, accidents could occur because the system is designed to try to prevent an accident,” he said.
Learmount criticised some pilot training regimes and the low use of simulator training, saying: “In all areas of air safety, training has just gone down, I think, over the last few decades.”
Aircraft simulation training had traditionally been “laser focused” on an aircraft going straight ahead or at a high angle, he said.
Learmount said: “I just wonder when we reach 100% reliance on machines and decide to stop using those machines is that going to be the best policy, because these machines have abilities we don’t yet understand?”