Estimated reading time 13 minutes, 33 seconds.
Few mantras are as deeply ingrained in most pilots as “Aviate. Navigate. Communicate.” From our first day of flight school, pilots are taught that our most important priority is always to “aviate” — maintain positive control of the aircraft — followed by “navigate,” or attend to the extended flight path. Only once both are assured should we divide our attention to the extent required to communicate with other people onboard the aircraft or air traffic control (ATC).
Most of us who have been flying for a while believe we do a pretty good job of this. Even routine flights can involve unexpected events and distractions: a surprise request from ATC, a minor fault that requires some in-flight troubleshooting, or a queasy passenger in the back. Every successful landing is proof that we aviated, navigated, and communicated in the correct order. Or is it?
Ongoing research from HeliOffshore suggests that we’re not always as attuned to aviating and navigating as we’d like to think, especially when relying on automation. Using eye-tracking devices to understand pilot attention patterns, researcher Dr. Steve Jarvis of Jarvis Bagshaw Ltd. has subjected numerous two-pilot helicopter crews to unexpected events in a simulator. He has consistently found that even very experienced crews show an immediate and sustained loss of focus on the flight path following an unexpected event, substituting their expectation for what the aircraft should be doing for awareness of where it actually is in space.
Most of the time, expectation and reality coincide. When they don’t, the results can be catastrophic. Armchair accident investigators (and sometimes actual accident investigators) then condemn the unfortunate pilots’ elementary monitoring errors — failing to realize that this is “how most pilots fly aircraft some of the time,” Jarvis said. Although monitoring dropouts are common, he explained, they only cause serious problems when they combine with other events or hazards, like the autopilot uncoupling or an obstacle in the flight path. “Because of this, the problem is almost invisible despite being ever present,” as he put it.
Fortunately, we pilots are not helpless in the face of our cognitive shortcomings. Jarvis’s latest study for HeliOffshore, summarized in a public report and video that were published online in September, illustrates how a simple mitigation strategy can be highly effective in reducing monitoring dropouts following unexpected events. It all starts with understanding the nature of the problem.
Jarvis, an expert in human factors, has been using eye trackers to study flight crew behavior since the mid-2000s, which is when he felt the technology had advanced enough to be practically useful in a flight simulator. Around that time, he told Vertical, “there were several airline incidents where it was deemed that pilots weren’t monitoring effectively or they were looking at the wrong things or they didn’t spot something. And out of that several companies began to say, ‘Well, we want to know what our pilots are doing.’ So I was going in there with the eye trackers and looking at pilots in the simulators to start to understand how pilots were monitoring.”
Those initial studies of airline crews revealed some striking behavioral patterns. One was a noticeable difference in pilots’ scan patterns when automation was engaged versus when they were flying manually, with much less attention paid to primary flight references under automation. Another was a phenomenon Jarvis named “parallel monitoring,” in which both the pilot flying (PF) and pilot monitoring (PM) tended to replicate each other’s scan patterns, despite having different cockpit responsibilities. When unexpected events were injected into research scenarios, most pilots exhibited large dropouts in monitoring of flight and navigational references, with the “navigate” dropouts being especially long-lasting.
Meanwhile, monitoring issues were being implicated in helicopter accidents as well, including the August 2013 crash of an Airbus AS332 L2 Super Puma operated by CHC. Four people were killed when the aircraft lost airspeed and likely entered vortex ring state while on an instrument approach to Sumburgh Airport in Scotland; investigators with the U.K.’s Air Accidents Investigation Branch ultimately concluded that “the helicopter’s instruments were not monitored effectively.” That accident contributed to the creation of the Joint Operators’ Review Board, a collaborative safety effort between offshore helicopter operators that became HeliOffshore in 2014.
HeliOffshore commissioned Jarvis to conduct eye-tracking studies of helicopter crews in flight simulators, leveraging a Joint Industry Project (JIP) fund endorsed by the Safety and Aviation Committee of the International Association of Oil and Gas Producers (IOGP). The fund supports key programs designed by IOGP and HeliOffshore to deliver the greatest safety value, while also providing a platform for further strategic investment in subsequent years.
According to Tony Cramp, chairman of the IOGP Aviation Committee, “the unique collaboration demonstrated by the JIP participants offers a means of accelerating vital offshore helicopter safety programs, combining with the expertise of HeliOffshore and helicopter partners, to deliver the zero fatality operational performance we all strive for.” Bristow, Cougar, Babcock, Airbus, Leonardo, and CHC all provided support for the research.
Jarvis started with an initial phase of observational research that was conducted with Super Puma crews. After the 2016 crash of another CHC Super Puma in Norway effectively ended the model’s offshore career, Jarvis had to shift to Sikorsky S-92 crews for the second phase of research, which studied pilots’ reactions to unexpected events such as a master caution light for a non-emergency technical failure. Although the change in helicopter model hindered direct comparisons between the two phases of research, Jarvis noted consistent patterns that also mirrored what he was seeing in his airline studies.
“Basically, both sets of research — heli and airline —were coming up with the same thing,” he said. That suggested that there are “general homogeneous vulnerabilities within pilot scan practices,” he said, not limited to specific individuals or aircraft models.
As laid out in the recent HeliOffshore paper, flight crews typically devote the bulk of their monitoring to “aviate” references such as attitude and airspeed, with a good chunk of their attention also paid to “navigate” references that show the aircraft’s location and path over the ground. Only a small amount of attention is directed toward instruments that show the status of various aircraft systems. When a caution light appears in the cockpit, however, both pilots exhibit sharp drop-offs in aviate and navigate tasks, instead directing almost all their attention to systems monitoring.
In Jarvis’s research for HeliOffshore, the PF would generally return their attention to aviate tasks within five to 20 seconds, which is still plenty of time for something to go wrong in a helicopter. But navigate tasks suffered even more, with neither pilot consulting navigational references for 60 seconds or longer. Just a minute after the unexpected event, pilots’ situational awareness had degraded significantly, with their perception of the aircraft’s heading off by an average of 37 degrees (and in several cases by over 100 degrees). In one experiment, nearly half of pilots flying in simulated instrument conditions incorrectly thought the aircraft was over sea instead of land, or vice versa.
Jarvis emphasized that the crews who took part in the study were highly trained and experienced professionals. In other words, these weren’t just rookie mistakes.
“Pilots might well be tempted to say, ‘No, it won’t happen to me,’ ” Jarvis said. “But I’ve seen the full spectrum, and I’m telling you, I’ve seen really experienced, really good pilots get distracted away from primary flying displays . . . for a long period of time. And sometimes when things change, they just don’t notice it. And it can happen to anyone.”
Less talking, more aviating
Alarmed by the results of Jarvis’s research, one airline commissioned him to lead the design of an intervention to minimize monitoring dropouts following unexpected events. That procedure, implemented by the airline in 2019, provided the basis for a similar mitigation tool that was designed and tested in the third phase of HeliOffshore’s eye-tracking research, again with S-92 helicopter crews.
A key finding of Jarvis’s earlier research was that when a PM talks through their evaluation of a technical issue — something they are often encouraged to do from the perspective of crew resource management — that verbalization can actually exacerbate monitoring dropouts by drawing the attention of the PF away from flight and navigational references. That can lead to parallel monitoring, in which both pilots are looking at the same references and neglecting others.
The mitigation tool thus begins with a short statement by the first pilot to notice a problem: an acknowledgement of the issue (e.g. “master caution light”) followed by “I have control” (if that pilot is flying) or “you have control” (if the pilot is monitoring). As Jarvis explains in his HeliOffshore video, “in this context, the main aim of the statement I or you have control is not actually about establishing which pilot has control. It’s used because it’s a very powerful and ingrained cue that creates an automatic response in the flying pilot to return their attention to the flight path.”
From that point, the PF attends to aviate and navigate tasks while the PM evaluates the issue, with no other talking unless urgent. When ready, the PM clearly states what they see and the need for any immediate actions/memory items, then temporarily attends to the aviate task while the PF confirms. After performing the necessary actions, the crew update their situation awareness and continue their assessment of the issue, with no unnecessary verbalization and the PF always returning to aviate and navigate tasks.
In the latest HeliOffshore study, crews first received extensive briefings to understand the nature of monitoring dropouts, the fact that they’re common to all pilots, and why the mitigation tool was designed as it was. They then rehearsed the procedure in a static simulator condition before practicing it repeatedly in a simulated flight scenario. Finally, they were tested with eye trackers in a number of simulated events.
The results were encouraging, with crews exhibiting much less dropout for aviate and navigate tasks. Their situation awareness was also much stronger than it was for crews who were not trained to use the mitigation tool. While the trained crews took longer to action emergency checklists (41 versus 28 seconds), pilots commented favorably on this aspect, with several mentioning that the situation felt more under control.
In terms of what pilots and operators can take away from the research, Jarvis believes there’s value in simply being made aware of the likelihood of aviate and navigate dropouts following unexpected events. This is also true for single pilots, who must maintain control of the aircraft while assessing an emergency situation. “That’s a real challenge for a single pilot, it’s absolutely true,” Jarvis acknowledged. “But I would say that in most events . . . you will have some time to at least make a check of that flight path during the assessment of the emergency, if not before the assessment of the emergency.”
While a little awareness can go a long way, implementing an effective mitigation tool like the one trialed in the research requires more effort — and care.
“There’s a lot of thought that goes into the process and the order and the sequence,” Jarvis cautioned. “What we wouldn’t want to do, obviously, is put in place some kind of a new intervention which solves 90 percent of the accidents here, but causes a different one over there that we’ve never heard of before. That’s very important with this stuff, because we’re dealing with very complex interactions and complex systems, which you can’t always easily predict.”