| BlueSky Business Aviation News | ||||||||||||||
| . | ||||||||||||||
The USAF didn't want to do pilotless aircraft because their entire organizational hierarchy was built around pilots (with fighter pilots at the top of the heap), so they created this name with the word "unmanned" in it, to imply that those who "flew" them were somehow less manly than those who piloted actual aircraft. A couple weeks ago, we also saw a news piece on Google's autonomous car being involved in a fender-bender accident. Turns out the car was under human control at the time of the accident, so no great public relations problem for "unmanned" automobiles. This is an area of research that is surprisingly complex. We think it's complex to fly airplanes automatically, but that is easy compared to letting a computer drive your car. The advantage cars have is that if the the automation gets confused, the car can just coast to a stop on the side of the road. It's not quite so easy with airplanes. Problems associated with automated flight of jet aircraft has also been in the news recently as the final reports on the crash of Air France Flight 447 were released. The problem in this case is not that the automation failed, or did something unexpected. Quite the opposite. The automation shut down as it was supposed to in response to a loss of reliable air data, and the pilots hand-flew the aircraft into a high speed stall, where it stayed all the way to impact on the ocean's surface. This opens up an entirely different area for discussion. Where is my computer pilot? Looking back at the origins of automated flight, what we see are various systems designed to make the pilots' jobs easier. We got gyro-compasses, artificial horizons, radio navigation aids, autopilots, and flight management systems. All designed with the intent of making it easier for the pilot. Unfortunately, by the very nature of those systems, the pilot is forced further and further away from the primal reality of the airplane. The automation comes between the pilot and the reality of the aircraft. The information is mediated by the computers and the pilot is left in an awkward position if the automation fails in various ways, both mild and catastrophic. There is a very good case to be made that pilots simply monitoring the automation are a bit behind the curve when the automation has problems, their awareness lags the event curve. The goal is for the pilots to be right on the event curve, to be completely aware of what is happening and what the downstream impacts of the failure will be. If the flight crew ride the event curve, they can safely cope with what went wrong. Things get ugly when the crew are lagging events and the situation unfolds faster than they can assess and act. The current paradigm There is a paradigm at work here. Our model of automation is to take what the pilot would normally do and make a computer do it instead. Often, there is a rational grouping of tasks like navigation or managing the engines, but in all cases, the computer receives the raw data and reports anomalous behavior in tersely worded failure notices posted on the monitors mounted on the panel, the EFIS. By its very nature, with each technical enhancement this design paradigm pushes the pilot a bit further away from the reality that is the airplane, providing an unparalleled opportunity for the crew to get behind the airplane. Mediation is normal Mediation of reality by the instruments has been part of aviation since the very beginning, of course. None of us are born with a magnetic compass in our head (though apparently carrier pigeons are), nor an altimeter nor an airspeed sensor, though initially we did what we could with the wind on our cheek and a careful eye on the ground below us. But to fly at the speeds and altitudes that are common today, an open cockpit and VFR-only isn't really an option. There are limits to what human beings can know from direct sensory inputs; we did not evolve to fly high above the earth at speeds approaching Mach 1. To do this safely requires we use some instruments, some technical mediation, and that forces us away from primal reality. A good design is one that balances our need for mediation with our need to be directly involved in the process. To redesign the aircraft's required suite of automated systems to become more pilot-centric is huge, expensive deal. And yet, I think it really is necessary. Our airspace in a very few places has gotten crowded with aircraft, all headed to one or two or possibly three airports near one city. So crowded that it is now a serious problem to divert an aircraft that is having some sort of difficulty. To add to the pressure, airplanes have gotten so large, and cities have encroached to such an extent on the airports--some, in fact, designed that way--that in a few cases it has become a potential problem to hand fly the airplane all the way down because of the required precision of the approach. Keep in mind that the aircraft's avionics are simply the reverse side of the air transportation coin. The obverse is the air navigation system and air traffic management system. Back in the mid '90s, when I was working on the New Large Airplane for Boeing (the competing program for what became the A380) we were designing airplanes that were so big that to bring them safely into existing airports, we had to keep the navigational error smaller than the space inside a pilot's chest. In order to drive up passenger throughput for a given airport, we needed bigger airplanes because half a dozen or so airports were already slot limited. Slot-limited means that the airport was at maximum capacity in terms of the number of aircraft per runway per day (slot-limited airports are where huge delays ripple through the system when bad weather hits because they have no surge capacity); i.e., no more airplanes could land because of the aluminum overcast. But, to bring in those bigger airplanes safely, we had to have highly precise automated systems, which meant no more hand flying the approach and landing. There are, of course, concrete-centric means of fixing this. To bring more airplanes and more people through, an airport can add more runways and more air traffic controllers. Denver, Colorado recently (in airport terms) shut down their original commercial airport, Stapleton, and opened up a new one way outside of town because there was no more room at the airport to pour more runways. Boston's Logan is already too small for the city's needs, but there is no room for expansion and no political will to build a larger airport farther out. The skies are crowded The skies are crowded, but only at a few cities. That is, yes, we have a problem, but so far, the problem is limited in scope. Currently, the problems with huge congestion are limited to a dozen or so cities around the world, depending on how we play with the definitions. One can also make a perfectly valid argument that as markets get overcrowded, they fragment; that is, we get flights between outlying city pairs. Not everyone flying into a London airport is actually on their way to a destination in London. In these cases, airport improvements in smaller cities can help resolve the congestion on the larger, denser cities. But, this does not always work. Destinations are driven by economics. People fly somewhere for a reason. If the company you need to meet with does not have a headquarters in a smaller city, then you have to fly to London (or New York, or Chicago, or Singapore, or wherever). So, even as traffic spills to the smaller markets, the large markets get larger, at least in part because they were large to begin with and that size is typically expressed as economic diversity and complexity; Detroit is an obvious exception because it hasn't been economically diverse since the 1930s. Unfortunately, significant air congestion will not stay limited to a dozen or so cities around the world; we can expect air traffic to grow with the global GDP. As the economic centroid shifts from the US and Europe towards Brazil, Russia, India, and China, we can expect their big cities to experience the same difficulties as the current economic superstars. Also, it makes sense to have a uniform, global system of air navigation and terminal area traffic management. At the very minimum we in aviation need a consistent methodology. Business aviation in particular is vulnerable to odd inconsistencies and changes in the air traffic management methodologies since we do not have as many dedicated employees sorting out the issues, and we don't just run between the same two city pairs, day after day. Coping with multiple kinds of air navigation and air traffic management systems adds cost, and that is the bottom line ugliness. But, we already have a single system! Some will argue that we already have a single-paradigm global air navigation and air traffic management system. To an extent, that's true, depending on how you want to talk about it. Unfortunately, the best examples of that same paradigm are barely able to cope safely with the existing traffic levels in a crappy economy. Should a miracle occur and we fight past the banksters to create significant economic growth again, we will exceed the capacity of the systems in a few key cities, which will inhibit economic growth. To support continued safe business aviation travel, it makes sense to prepare for global economic growth. At some point, the banksters will have siphoned off enough money to feel good about themselves again (think of their desire for more cash as an additional tax upon the system--that's what it is) and the global financial system will spool up again enabling business expansion, which will drive the need for more flights, and those flights will tend to be to or from the centers of global economic expansion, which are exclusively cities. Task saturation Even in cities with large, modern, multiple runway airports, we are up against the limits of the current human-airplane-air traffic control system interface. Possibly worse, with the current system, even the present level of automation, the pilots are often at task-saturation within the terminal control area. Task saturation means they are already doing as many things as they can possibly do. Their plates are full. Tasks that cannot be completed during this difficult time are dropped, deferred, or delayed, the pilots have no choice. There has to be a better way. The alternative I think we need a new paradigm. Parts of the system that are not currently automated need to be. Task saturation of pilots during approach and during climb out is a very bad idea, we need to fix that. Moreover, having to monitor the automation as it flies a highly precise approach pushes the flight crew away from the airplane; again, this is bad idea. At the same time, the air traffic control system cannot continue functioning this way. It needs to be better integrated and automated, with attention paid to graceful failure modes and how to cope with aircraft emergencies as well as more mundane problems like loss of power or bad weather (think tornadoes). The current system is at its limits in certain areas. We need to admit that this is so, and move to the phase where we find a solution. We must also keep in mind that a new paradigm of avionics and air traffic control systems has to be designed to work well even in the smaller markets; there is no point in having two different paradigms at work (I can make an argument that it is unsafe). My airport manager should be able to buy a box or set of services that will allow a one-runway regional airport to be a fully integrated part of the global air traffic system, no different than a large, multiple-runway international airport. Moreover, I should be able to retrofit my classic Lear Jet 28 with an avionics suite that will make my airplane a fully-functional part of that same system and enhance the flight crew's awareness of what is happening (and why, and what the likely implications are). And I should be able to retrofit for less than a prince's ransom. (While I'm in here making demands, I want increased reliability along with reduced weight and power consumption. Thanks.) I don't know how to do all of that stuff, but I have a couple of clues where to look. One oft-repeated part of the discussion about Air France Flight 447 is that “it's like a video game!” Sorry, no, it isn't. Any of you, gentle readers, happen to play video games? How about those massive multi-player on-line role-playing games? Anyone ever get a new video game that sucked you into it and next time you looked up it was 3 AM? I have. There are people who study this stuff and we should talk to them. This is what driving the airplane should be like. It should be sticky for your attention. I don't know how to do that, but clearly, there are people who do. The resulting new design may not look a lot like what we have now, but that is fine. It doesn't have to. Next Gen Some of us recall several attempts to upgrade the existing air traffic management system here with just the US. All of them, to my knowledge, have turned out poorly. It may have been too soon (i.e., the technology wasn't ready). The limits of the current paradigm may have been too difficult to expand. Or, the companies doing the work may simply not have had the right people. The next one out of the box is called the Next Generation Air Transportation System (come to the BA-Meetup (http://www.ba-meetup.com/index.htm ) this November and sit in on Matt Glasser's presentation). The descriptions of the system are very encouraging. In fact, my only issue with it is that it does little, if anything, to address task saturation in the cockpit. Nevertheless, if it is well designed, Next Gen should be able to add that capability down the line. I hope so. We need it. Terry Drinkard is currently consulting on an aviation start-up. His interests and desire are being involved in cool developments around airplanes and in the aviation industry. Usually working as a contract heavy structures engineer, he has held positions with Boeing and Gulfstream Aerospace and has years of experience in the MRO world. Terry’s areas of specialty are aircraft design, development, manufacturing, maintenance, and modification; lean manufacturing; Six-sigma; worker-directed teams; project management; organization development and start-ups. Terry welcomes your comments, questions or feedback. You may contact him via terry.drinkard@blueskynews.aero Other recent articles by Terry Drinkard:
|
||||||||||||||
