Working Group Outlines Recommendations for Human-automation Interaction

 - January 3, 2014, 12:15 AM
Pilots have come to rely on technology in the cockpit, but they must know how to make it work for them under difficult situations, says a working group.

Humans’ attempts to interact with cockpit automation have provided fodder for pilot anecdotes for years, and the recently released Operational Use of Flight Path Management Systems offers a precise roadmap for where the industry needs to focus. The 34-member research team responsible for compiling the report comprises members of the Performance-based Aviation Rulemaking Committee’s (PARC) flight-deck automation working group, which evolved not long after the 1995 American Airlines accident in Cali, Colombia.

The flight-deck automation working group identified 29 separate issues affecting automation management on a modern transport flight deck and generated 18 distinct recommendations. The report should be considered a must read for anyone involved in aircraft operations, automation design or pilot training.

As the Asiana Flight 214 accident at SFO demonstrated recently, cockpit automation is merely a tool to help achieve a safe flight, not the ultimate solution. Even the most advanced computers still need pilots not only to watch over them but also to fully understand precisely what the automation is actually doing at any given juncture. In fact, the 2005 Circuit City Citation loss-of-control accident at Pueblo, Colo. taught a similar lesson in a much less automated aircraft.

Despite commercial aviation’s stellar safety record of late, the working group remains relevant for its efforts to study both the safety and efficiency of flight-path management on modern glass-cockpit aircraft, including energy-state management, for both current and future operational use. The FAA last addressed this issue in 1996, the early days of automated airplanes, when it published Human Factors for Flight Deck Design. The authors of the new document reference the 17-year-old document a number of times as both a vehicle for checking off items rectified since 1996, and as a means of flagging issues unresolved as yet.

The working group’s research efforts gathered and reviewed information from the Aviation Safety Reporting System (ASRS), Line Oriented Safety Audits (LOSA) and accident and incident reports from around the world. The group interviewed 11 operators, six aircraft manufacturers and one training organization. The research items were coded for review using a number of broad categories covering the information reporting methodology, automated system design, operational and training issues, a variety of threats and errors, and undesired aircraft states. While the working group believed it was important to look beyond pilot errors and dig into the good behavior for the 99 times out of a hundred when pilots did everything correctly, it noted that that kind of data collection system does not exist yet.

Before the Boeing 757, the first glass-cockpit aircraft, was certified in 1982, errors resulting from poor manipulation of the flight controls, or from poor communications among crewmembers (often three pilots) or ATC, suggested that the people in the cockpit were not acting as a team. When trouble appeared, the training fallback was “aviate, navigate, communicate.” This report makes it clear that now, with usually no more than two humans in the cockpit at any one time, it’s time to acknowledge the idiosyncrasies of the third pilot, the one that happens not to be human. Despite the sophistication of current-generation automation, people and machines still do not always speak a common language. Today when trouble appears, pilots think “aviate, navigate, communicate”; they need to add “automate” in there somewhere, the working group emphasizes.

Manual Flying Skills Still Needed

The 29 key findings were broken down into operational experience categories such as risk mitigation, manual flight operations, pilot use of automation, autoflight mode confusion, communication between crewmembers and ATC, flight-crew procedures, task/workload management, pilot knowledge and skills, equipment design, air traffic and airspace considerations, regulator’s knowledge and data collection issues.

A look at just one category–manual flying skills–highlights the methodology the group used to dig deeply for insights. They first identified the insufficiency of pilot knowledge and skill at operating the aircraft when the technology becomes unavailable for any reason. At the back of most safety experts’ minds is the recent Boeing study conclusion that between 2003-2012, more aircraft fatalities worldwide were attributable to flight crews’ losing control of their aircraft than to any other cause. Loss of control is still listed as the most critical threat to aviation safety today.

The working group found that “more than 60 percent of the accident reports reviewed identified a manual handling error as a factor in the accident.” The group qualified a number of handling errors into topic areas, including incorrect upset recovery, inappropriate control inputs and a lack of correct manual handling after autopilot or autothrottle/autothrust disconnection. Also identified was a failure to recognize those disconnects that resulted in poor monitoring of energy and speeds and included the mismanagement of autothrottle/autothrust.

The group also learned that a number of factors regularly coexist with manual flight input errors, such as difficulties in transitioning between autoflight and manual control, crew coordination, poor cross-checking and verification, as well as inadequate training in fully understanding the automation itself. Interestingly, although almost every person interviewed admitted to concerns about the decline of manual flying skills, few were able to agree on precisely which flying skills would be the most needed to keep a pilot’s skill level high. They also couldn’t agree on how these skills might decay and what could be done to retain them.

Consider crew interaction with the automation during an all-engine go-around, especially (but not only) when initiated before decision altitude. Pilots expressed concern that high power, low weight, low-altitude level-off and autoflight logic combine to create a challenging maneuver to be performed while reconfiguring the aircraft. LOSA data showed that “87 percent of unstabilized approaches resulted in safe landings within all parameters, while 10 percent of those approaches resulted in injury-free landings that exceeded some parameter, such as going off the end of the runway.” The other 3 percent conducted a go-around, but in 98 percent of those go-arounds the pilots exceeded some parameter such as maximum flap speed.

Another unexpected insight the group uncovered related to training. Some training professionals noted that pilots sometimes learn to use the automated systems by “watching things happen” in fixed-base trainers. So it should be no surprise that when pilots have to hand fly, they are accustomed to watching things happen and essentially forget to shift gears, reacting instead of flying proactively.

These manual flying issues were a concern in the 1996 FAA human factors report. The situation could well be getting worse since accidents that involve these skill vulnerabilities continue to happen. Future flight operations based on more precise navigation functions such as performance-based navigation (PBN) are expected to be even more demanding for humans, because the need to control the aircraft manually is not going to disappear anytime soon. “Specifically addressing the development and maintenance of pilots’ manual flight operations knowledge and skills is critically important,” according to the working group.

While the group developed a considerable number of valuable insights, some of the recommendations lacked specificity. The report’s recommendation related to the issue of manual flying skills, for example, was to “Develop and implement standards and guidance for maintaining and improving knowledge and skills for manual flight operations that include providing pilots with opportunities to refine knowledge and practice skills,” adding that “training and checking should directly correlate.”

The report did not include a timeline for implementation of its recommendations. The future value of the study itself then seems to depend upon someone–a team at the FAA working with industry, most likely–translating the elements of this document into action. This is a significant challenge, since the FAA is already struggling to figure out which portions of the decade-old NextGen program will be sacrificed on the altar of sequestration.

The working group noted the current high level of safety in commercial aviation and attributed much of that record to technology already in use. Most important in the brief conclusion was the mention of “vulnerabilities within the aviation system relating to the operational use, equipment design and management and training of flight path management systems …” Both the short- and long-term recommendations in the report address the [system’s] immediate vulnerabilities, and the working group “believes implementing these recommendations is necessary to make improvements in safety and operational effectiveness.”

The 1998 Commercial Aviation Safety Team (Cast) made a bold commitment to the industry to reduce commercial aviation’s accident rate 80 percent by 2007. The group actually achieved an 83-percent reduction. Today, the question is whether or not this current flight-path management study will spur the FAA and the industry to seek the greatness in aviation safety that the Cast team began in 1998.