Flight deck automation
Monitoring automated systems in the cockpit.
By Shannon Forrest
President, Turbine Mentor
ATP/CFII. Challenger 604/605, Gulfstream IV, MU2B
Some say it’s the greatest time in the history of aviation to become a professional pilot. Just about every employer running a flight department is looking for employees to join their ranks.
Salaries have gone up considerably in concert with the high demand and low supply economics exhibited by the current pilot job market. A person who obtains an Airline Transport Pilot (ATP) certificate in the coming decade will have endless career possibilities – or at least it seems so right now.
The other side of the argument is that piloting as a career will cease to exist at some point in the future. Proponents of automation argue vehemently that autoflight technology has reached a point that renders the pilot an unnecessary component of a transportation system.
Pilots vs computers
Those who make a living flying an aircraft should at least be aware of the 3 arguments working against manned aircraft. It’s been said that removing a sentient being from the pilot seat and replacing him or her with a computer has 3 advantages – cost reduction, efficiency, and increased safety margins. Pilots are expensive to the bottom line. Unless employers can find mortgage-free applicants who don’t need to put kids through college, eat daily, or operate motorized vehicles, they’ll continue to be so.
Imagine the joy on the face of the chief financial officer of an airline or a midsize corporate flight department if he/she could eliminate the 2 lines in the annual budget labelled “salary” and “benefits,” or if he/she could eliminate the cost of attending CAE or FlightSafety International for initial or recurrent training.
Remember the time the operational flight plan said that peak efficiency would be achieved with a cruise speed of Mach 0.77 but somehow the thrust got bumped up to fly at 0.82 just to get to the destination 7 minutes faster so as to beat the traffic on the commute home? Computers don’t need to get to a soccer game or birthday party on time.
Or the time the separation with the preceding aircraft looked a little too close when landing behind that Boeing 757 in ATL (Hartsfield–Atlanta GA), so airspeed was reduced below what the air traffic controller had assigned? The resulting loss of separation with the aircraft in trail meant that the airport arrival capacity was reduced by 2 aircraft per hour.
Computers do what they’re programmed to do. They don’t slow early. They don’t need to keep their skills fresh by flying the River Visual approach to Rwy 19 at DCA (National, Washington DC) when the autopilot flown RNP-19 is available. Computers don’t take off without a clearance in a Learjet 60 on Rwy 9 at BOS (Logan, Boston MA), necessitating that an Embraer 190 landing on intersecting Rwy 4R go around to avert a collision.
The incident at BOS, which occurred in February 2023, is currently under investigation, and it’s unknown why the pilot began a takeoff roll despite reading back the “line up and wait” instruction provided by the aircraft controller. If you’re detecting a sardonic tone here, you’re correct. But unless some things change regarding how pilots interact with automation, 10 years from now the title of this magazine might be Former Professional Pilot.
The history of the autopilot
The autopilot dates back to the early 1900s, and the original intent stemmed from nothing more than the precocious curiosity of a young inventor named Lawrence Sperry. The fact that a bicycle or motorcycle could remain upright as long as it was in motion intrigued him. Sperry, who was already a pilot, surmised that the same principles of the gyroscopic effect that keep 2-wheeled vehicles upright could be applied to aircraft.
Sperry was the first to link the flight controls directly to a gyroscope to keep an airplane oriented without pilot input. He was so confident in his invention that at times he would get out of the pilot seat and onto the wing to prove that the aircraft could maintain level flight even when acted upon by external forces.
He even installed an anemometer to detect an incipient stall as a function of decreasing airspeed. If the speed got too slow, the rudimentary autopilot would pitch the aircraft down to avert the stall.
Fast forward 100 years, and the same basic principles are still in effect. At the highest level of technology, laser accelerometers and fly-by-wire have replaced spinning gyroscopes and mechanical linkages so that aircraft control is even more precise. Over time, auto flight technology has gone from “can it do it?” to “how can it improve it?”
Each advance in technology is directed at mitigating some sort of inherent human limitation. Glass cockpits, for example, exist because scanning a hodgepodge array of tiny round dials for extended periods of time is physically and mentally fatiguing. Autobrakes were conceived because pilots can’t apply consistent pressure on their own to avoid skidding, or manage to stop in the available runway.
Autothrottles/thrust are here because a pilot needs help managing power for hours on end. Enhanced vision systems were developed because a pilot can’t see through the clouds. And we have flight envelope protection because maintaining the structural integrity of the aircraft is a prerequisite for a good landing.
Pilots and automation
In an article titled “The Human Factor,” published in Vanity Fair in October 2014, author and pilot William Langewiesche describes what happens when you put pilots on automation. “Their manual abilities degrade and their flightpath awareness is dulled: flying becomes a monitoring task, an abstraction on a screen, a mind-numbing wait for the next hotel,” he says.
It happens that William is the son of Wolfgang Langewiesche, the author of the aviation classic Stick and Rudder. William wrote the article largely in response to the crash of Air France Flight 447.
He goes on to say, “This is another consequence of designing airplanes anyone can fly: anyone can take you up on the offer.
Beyond the degradation of basic skills of people who may once have been competent pilots, the 4th-generation jets have enabled people who probably never had the skills to begin with and should not have been in the cockpit. As a result, the mental makeup of airline pilots has changed – a different crowd is flying now, and although excellent pilots still work on the job, on average, the knowledge base has become very thin.”
There’s a mantra in aviation that says that the purpose of automation is to make a good pilot bad and a bad pilot good. Automation is the great equalizer that masks deficiencies in proficiency. Langewiesche’s remarks are controversial, but 2 things are certain. Automation is here to stay, and flying with automation has become a monitoring task (at least on the professional level of flying passengers).
Monitoring skills
Unfortunately, it turns out humans aren’t so good at monitoring. According to NTSB, 84% of major airline accidents over a 10-year period were attributable to a failure to monitor. Similarly, ICAO concluded that inadequate monitoring was responsible for 50% of controlled flight into terrain (CFIT) accidents. The Flight Safety Foundation identified inadequate monitoring as responsible for 63% of approach and landing accidents.
In the early 2000s, FAA published AC-120-71A, Standard Operating Procedures for Flight Deck Crewmembers. The AC focused heavily on the concept of monitoring. Around that time, the term pilot monitoring (PM) began to be used in lieu of the previous nomenclature, pilot not flying (PNF).
The switch is derived from a psychological concept called the framing effect. This technique is used quite often in marketing products and political campaigns, and states that people are biased and make decisions based on how something is presented or phrased. It’s better to tell someone what to do – monitor – than what not to do – not fly the plane. PNF connotes a passive role, whereas PM suggests a more active role. The goal is that both crew members stay continuously engaged.
In June 2007, research scientist Chris Wickens and his colleagues conducted a study that looked at monitoring strategies and performance on automated flight decks. They outfitted pilots with eye tracking devices to quantify what they were looking at during critical phases of flight. The subjects flew normal flight profiles and were instructed to identify any anomalies.
Unbeknownst to the subjects, the researchers had programmed the primary flight display to indicate an erroneous pitch mode during the climb. One third of the pilots never looked at the erroneous indication, and those who did never reported it. In addition, 30% of pilots missed a failed glideslope indication, even after beginning an ILS approach. Failing to look at autopilot mode change indications was also prevalent. We’ve got to do better.
Improving active monitoring
Poor monitoring has become such a concern that in 2014 the Flight Safety Foundation published the findings of a working group that focused on active monitoring. The guide is 65 pages long, and identifies key incidents and accidents that were used to establish conclusions and recommendations.
It’s important to note that the concept of PM – or monitoring in general – is to avoid an undesired aircraft state. The goal is to identify small deviations or anomalies, whether human or automation, before they manifest as a bigger problem. The first step is to develop, publish, and enforce standard operating procedures (SOPs). If everyone is doing things the same way, it’s easier to detect error. It’s akin to reading a movie script. Everyone knows what the next line is. If the actor flubs the line, it’s obvious.
SOP compliance is fairly consistent in airline operations, largely because pilots might be flying together for the first time. In the business aviation world, 2 pilots might be paired together flying the same trips year over year, and as a result, they tend to learn (and sometimes ignore) each other’s habits.
Sometimes, the failure to monitor is attributed to trust in the other pilot. We all have a bad day now and then that can result in degraded performance, so it’s best to continue monitoring irrespective of who’s sitting next to you and how long you have been flying together.
Things that should be monitored include the flight path (whether in the air or on the ground), aircraft systems, operational factors (weather, weight and balance), and the situational awareness level of the crew. Monitoring is accomplished through attention management, deliberate checking, and cross-checking and verifying. It’s very difficult to identify failure on things that function normally almost all the time, and that’s a pitfall of automation. Expectation bias is “looking without seeing” and thinking something is the way we believe it to be, not the way it really is.
Recommendations
Slow, methodical attention to detail is the solution. Areas of vulnerability appear any time the aircraft is in a dynamic state, like climbing or descending. These conditions are times to avoid distraction and focus on the pilot who’s hand flying, or watch the autopilot to ensure it’s doing what it’s supposed to. Communication can also be helpful. A pilot flying (PF) who annunciates deviations (eg, “I’m fast … correcting) is a sign that active monitoring is taking place, and alerts the PM that the PF is taking corrective action.
Lastly, when automation fails and the aircraft is headed toward an undesired state (often associated with pilot confusion), intervene by taking manual control. The acronym CAMI is a useful tool when addressing automation: Confirm, Activate, Monitor, Intervene.
On February 14, 2023, FAA issued a safety memorandum that stated, in part, “We are experiencing the safest period in aviation history, but we cannot take this for granted. Recent events remind us that we must not become complacent. Now is the time to stare into the data and ask hard questions…”
Let’s not allow one of those hard questions to be whether the pilot is the weak link in the chain. Automation is a wonderful tool at the hands of a competent pilot. With any luck, we will all be reading Professional Pilot magazine well into the future.
Shannon Forrest is a current line pilot, CRM facilitator, and aviation safety consultant. He has more than 10,000 hrs TT and holds a degree in behavioral psychology.