April 22, 2015:   Would you fly an airliner lacking a flight crew and operated by software and a few back-up pilots on the ground? Actually many people would, just as many people are keen to try out a driverless car (or use existing self-parking software). The fact of the matter is that all this automation has been sneaking up on us, just has it has been doing for over half a century, until it is fully developed and accepted. This is all despite the fear of new technology, often for good and practical reasons.

Let’s go back a long time to the first people who learned how to control (start when needed) fire. There were probably warnings that this would lead to more burn injuries and even then it was probably realized that inhaling too much smoke was bad for you. Today there are still hundreds of millions of people depending on open fire for cooking and heating and there are medical statistics to show that this population suffers many more burn injuries and ill effects (on lungs and other organs) of all that smoke. Yet no one has seriously advocated outlawing controlled fire, just making it safer. Same with more recent tech, like steam engines (the early ones, be they for boats, railroads or factories, were quite dangerous) and automobiles (which are still dangerous, but much less so that a few decades ago, thanks to a lot more software and microprocessors). The same thing has been happening with automated flight controls (for engines, other equipment and piloting as well). If all this seems inevitable well, that’s because it has been.

The current enthusiasm for robotic flight crews was sparked by a recent (March 24) incident where a German A320 airliner was deliberately crashed by a suicidal pilot (killing himself and 149 others). This again raised questions about human error in aircraft accidents. Commercial aviation accident rates have declined 90 percent since World War II, mainly through the introduction of more safety devices and more reliable aircraft. Much of this included software and robotic functions. This has made pilot error the major cause of military and civil air accidents.

With military aircraft, especially fighter pilots practicing combat maneuvers and collide in the air, it is all about pilot error. But this is difficult to avoid, as the high speed maneuvering, in close proximity to other aircraft, is unavoidable, when you are training pilots for combat. While some of the most dangerous such training has been shifted to flight simulators, you still have to spend time practicing in the air to obtain useful combat skills. For commercial pilots mental problems and bad behavior (coming to work drunk or too tired) have become too common to ignore as “too rare to bother with”.  But an increasingly common cause of human error is pilots who have forgotten how to quickly employ basic techniques to deal with inflight emergencies.

This situation is more acute in civil aviation because pilots of modern aircraft spend 5-10 minutes per flight actually controlling the aircraft. The rest of the time automated flight controls do all the work. Some pilots spend more time actually flying an aircraft in a simulator than they do at work. While combat pilots spend nearly all their training time (between 100-200 hours a year) actually flying the aircraft. Commercial pilots can spend up to a thousand hours a year in the air (but no more than 30 hours a week and 100 hours a month) and get less than half as many hours actually flying the aircraft than a peacetime combat pilot who flies only training missions. When in combat military pilots will fly two or three times as many hours (over a thousand a year). Since September 11, 2001 American combat pilots have spent a lot more hours in the air.

The growing use of flight control software, which can now handle landings and most of the cruising time (flying from one point to another) tasks is moving relentlessly towards “piloting” being reduced to switching the autopilot on and then, after the aircraft has reached its destination and parked at the gate, turning it off. Software now can handle all flight operations for UAVs, including takeoffs and the patterns to be flown for surveillance or reconnaissance as well as in-flight refueling. This has led to growing calls for the use of fully automated commercial flights. There is a lot of public (and pilots) opposition to that but the idea makes sense. There would still be human pilots available on the ground for emergencies. This was always the justification for having humans in the cockpits. But pilots who entered the profession since 2000 are getting a lot less time actually flying the aircraft and there have been growing incidents of such pilots facing emergencies, or even routine operations, lacking basic flight skills or enough experience to act immediately. One solution is to let commercial pilots spend more time in flight simulators, practicing emergency procedures. In addition to being a lot (80 percent or more) cheaper than doing it in the air, you can practice very dangerous emergencies in the simulator that are simply too risky to try in an actual aircraft. Even with this approach, there will always be the temptation (because of lower cost and fewer accidents) to use software to replace pilots on every aircraft.

Combat pilots are also being faced with replacement by software. New flight control software for UAVs is going to replace bomber aircraft soon and eventually fighters as well. For example the U.S. Navy’s X-47B UCAV (unmanned combat air vehicle) recently performed its first in-flight refueling under software control. During the last few years this unmanned light bomber and recon aircraft has successfully carried out numerous tests aboard aircraft carriers. The X-47B made its first catapult launch from an aircraft carrier on May 14th 2013. That was followed by several touch and go landings on a carrier. The first carrier landing, as expected, followed soon after. X-47B made two out of three carrier landings for the first time on July 10th 2013. Later in the year more flight tests further stressed the capabilities of the automatic landing system, especially in high speed and complex (different directions) winds. The autolanding systems passed all these tests. In the X-47B demonstrated its ability to land and be off the carrier deck in less than 90 seconds, just like manned aircraft. There are more tests scheduled to see how effectively and reliably the X-47B can operate on the carrier and hanger deck and do it alongside manned aircraft. There will be more tests in 2015 and after that the slightly larger X-47C will take over. Long-term the navy currently has a plan to introduce a UCAS replacement for the F-35 (which is soon to replace the F-18s) in the 2030s. But if the UCAS progress continues to be swift and the costs low (compared to manned aircraft), the F-35 could find its production run much reduced to make room for a UCAS.

While software controlled landing systems have been around for decades, landing on a moving air field (an aircraft carrier) is considerably more complex than the usual situation (landing on a stationary airfield). Dealing with carrier landings requires more powerful hardware and software aboard the aircraft. The navy expected some glitches and bugs and overcame these and is now rapidly catching up to the reliability of commercial landing software (which has been used very successfully on UAVs).