Summary – as this technology advances and nominally improves, even where man remains ‘in-the-loop’, he is less and less in control.
It’s no good winging about it. In all forms of transportation, natural intelligence (manual skills) will soon be ‘out’ and artificial intelligence (nerdie skills) with full automation will be ‘in’. There is no question of ‘if’ – just when (the Author’s guess, in 10-15 years). How often in a day does one feel frustration and stress at the failure of digital systems that impact our lives? Now lift them 15 km into the air and the impact is exponentially greater. This is an opinionated review on the need for concern in this regard (Note – the views are personal to the author alone).
Actually, in aviation (and astronautics) automation is a very mature technology. The author first flew in a fully automated helicopter some 50 years ago. Space ships have always been almost fully automated – Yuri Gagarin in the Sputnik in 1959, did not have any controls to touch ! In a general context there are three levels of automation.
The simplest are Pilot-Assist technologies, otherwise referred to as auto-pilots. All commercial aircraft have this aid. It has been around for longer than the biblical three score years and ten. The automation is limited to the maintenance of height, heading and speed, and more recently, aircraft attitude. In addition, the engine responses to Pilot power demands are now also fully automated. But, on airliners (sic Fixed Wing) thus fitted, using hydraulic assisted controls (power steering if you will) the pilot remains in full control of the aircraft and can over-ride the automated elements at the flick of a switch and fly it manually. For Helicopters (sic Rotary Wing), the same is essentially true but limited to the aircraft’s guidance – it’s stabilization (whereby any control movement in one dimension impacts control in the other two) is a more complicated issue as discussed in the previous feature under the heading of ‘Safe but not Sure’. This latter can be switched-off but few of the current generation of pilots have flown unstabilized aircraft and would find themselves juggling to maintain level flight, so it is generally left on. (The author’s generation learned their trade on unassisted aircraft and so instinctively had to develop the necessary coordination and so have no problem in this regard).
Then there are automatic flight control systems (afcs) with man-in-loop. These aircraft are flown by computers with pilot inputs. There are separate programs for take-off, climb, cruise, descent and landing – in truth to fly from A to B, highly paid pilots need do nothing more than enter a flash-drive with the flight plan into a computer port in the cockpit and select the 5 buttons that run these programs. All these aircraft use Fly-by-Wire (FBW) control input technologies – and here is the scary part. The so-called ‘joy-stick’ that pilots have traditionally used to fly aircraft, is replaced in FBW aircraft with a dinky side-arm controller (see right-hand photo). This is actually not a control at all – it is a computer mouse. As such, any control movement (say to turn left or right) is no more than a mouse telling the computer what the pilot wants the aircraft to do. Unlike with the above hydraulic control systems (left-hand photo), he has no direct access to the control surfaces. Indeed, neither does the aircraft computer. It just sends a digital signal to an analogue sub-system at each control surface which then drives an electric motor to move it as required. So, a pilot may be ‘in-the-loop’ but he is no longer ‘in control’ and can do no more than monitor the computer performance. Were it to fail, there is nothing he can do except to scream “Mayday, Mayday” on the radio and into the aircraft intercom to “brace-brace-brace”! He cannot switch off the afcs and fly it himself (as in the above autopilot) because he has no access to the control surfaces. (Incidentally, as an amusing digression, why scream “Mayday”? It’s those frogs again who just will not accept that international speak is now in English. In the 1920s when aviation was maturing, the language of international diplomacy was still French. It took the late arrival of the Yanks in two World Wars, for it to be changed to English, so that they would better understand what was going on…..! But with international civil aviation being based in Montreal (a francophone town) that linguistic aberration lingered on: hence m’aidez, m’aidez – help me!! (Zat accen’ has always been a challenge for we anglophones…………).
The one exception to this is the modern airship (known as a Hybrid Air Vehicle – HAVs) which, using a combination of gas-lift and aerodynamic lift, are very large (some 100m in length and almost 20 stories high). They also use the same afcs principal except using fibre-optics (fly-by-light – FBL) instead of electrical wiring (FBW). Essentially they are very similar but FBL allows some 10 times the data rate. Because of its huge size, in the HAV there is the space (and lift) to put in a duplicate system allowing the pilot mouse (side-arm controller) to directly access the analogue actuator and messily get the gargantuan balloon back to base in the event of complete afcs computer failure. Airliners instead have duplex critical sub-systems, triplex safety-critical systems and five computers. It is thus very unlikely to go wrong, but that is not to say it ‘never’ fails. It has so on several, but not frequent, occasions and each time everyone died. However with a frequency of such event being very small (less than the regulatory standard of 0.002%), it is correctly perceived as an acceptable hazard – after all, every day folk use their cars where there is almost a 0.1% chance of a major incident.
Nonetheless, the thought that the so-called man-in-loop cannot actually assume direct control when computers cease to properly compute, is scary. A nice anecdote demonstrates this lack of pilot control. In a demonstration to client airlines of a new type of FBW airliner by a senior test pilot (ie a guy with vast experience), he decided to end the demo with a very low, high speed pass (as one does on such occasions). It was impressive not least because he over-cooked it and realised he was going to have a problem clearing the trees at the airport perimeter. So he slammed the power controls fully forward and heaved up the nose – except the aircraft responded to neither. His dinky side-arm controller could only tell the computer what he was seeking to do to not hit the trees at the end of the runway but the computer (which could not ‘see’ the looming disaster) knew better. Realising that to fulfil the (Chief) pilot’s demand would overstress (and damage) the engine and airframe, the compute opted instead to increase engine power and use control movements that would stress neither. The result was that the aircraft clipped the tree-tops, filling the engines with scrub (thus writing them off) and seriously damaging leading edges of the wings to say nothing of the hull the paintwork. Fortunately the only casualty was a bruised pilot ego and a few millions of dollars of repairs (which the subsequent sale of many aircraft to the very impressed client airlines, fully compensated!!).
From this it can be seen that the next step, to fully automated passenger aircraft with no man-in-loop (ie unmanned aircraft), technically speaking, is incremental – the main issue preventing this change is not one of technologies but simply a matter of perception. Even if pilots can do very little to control a FBW aircraft, not having one (indeed two) up-front would not be good for ticket sales. In the military, the remote piloting of long range killer drones all over the world controlled from a hangar in California is now well established. Airliners may eventually follow, ultimately with a single pilot sitting at an airline HQ, from where he could control multiple aircraft. After all, in the cruise, automated vigilance will suffice – only for brief periods at take-off and landing will dedicated pilot surveillance be required. What will he do if something goes badly awry? Probably exactly the same as those today in any FBW aircraft cockpit – issue a mayday and politely invite the unfortunate passengers to brace themselves for the final curtain – then, no doubt, go to a bar to console himself.!
The driving force for this change will be the new generation of E-Vtol Tupperware, so-called, air taxis (in the AMS view a technology bubble which will burst in a year or two and about which we will also write soon). Notwithstanding their small size and limited Pax. payloads, their automated flight profiles are similar to any airliner, but in a much more challenging flight environment. Bubble or not, these platforms will be required to achieve the same level of certification as a 400-seat airliner (called ‘Transport Category’). Once they have done this, the door will be opened to the full automation of the airliners themselves. Hence our guess that such is not more than some 15 years away……
Space ships have almost always been thus full automated – although the Apollo-series of moon landing fame maintained a man-in-loop capability (which was lucky as, on the big day in 1966, it soon became apparent that NASA has got it wrong and Neil Armstrong had to land the Eagle module manually, which he miraculously did with just some 10 seconds of available fuel remaining !!). Today we watch in awe as space ships do increasingly amazing things both in earth orbit and to the very edges of the solar system. But the fact is that, being subject only to the simple principals of Newtonian laws of motion established some 400 years ago, space ships are far easier to control than aircraft operating in the unpredictable atmospheric conditions on Planet earth. In space, there are just two main forces – one generated by man-made power plants and the other gravitational pull. These can be calculated to the micro-Newton and never change thus making precision control, almost literally, childs’ play. (Helicopters, as you may have seen in the last dissertation, are anything but……!).
Conclusion – based on media prognostication, full automation is typically presented as a futuristic projection, but in aviation (rather than the automotive milieu) it has effectively been there for a generation or more. The technologies are fully understood as is the reliability of computers – we each have our own experience in that regard !! But, as stated in earlier essays, notwithstanding all the negatives highlighted above, the chances of a flight accident are minimal. Travel on roads is significantly more dangerous. So take a deep breath, double your beverage of choice and sit back, (try to) relax and enjoy your automated flight………!!