Pilots Rely Too Much on Automation

The Wall Street Journal by Andy Pasztor 17 November 2013

Many Aviators Have Difficulty Manually Flying Planes, Study Commissioned by FAA Finds

Commercial airline pilots have become so dependent on automation that poor manual flying skills and failure to master the latest changes in cockpit technology pose the greatest hazards to passengers, an international panel of air-safety experts warns

.

A soon-to-be-released study commissioned by the Federal Aviation Administration determined, among other things, that "pilots sometimes rely too much on automated systems and may be reluctant to intervene" or switch them off in unusual or risky circumstances, 

The study found that some pilots "lack sufficient or in-depth knowledge and skills" to properly control their plane's trajectory, partly because "current training methods, training devices and the time allotted for training" may be inadequate to fully master advanced automated systems.

Among the accidents and certain categories of incidents that were examined, roughly two-thirds of the pilots either had difficulty manually flying planes or made mistakes using flight computers.

Relying too heavily on computer-driven flight decks—and problems that result when crews fail to properly keep up with changes in levels of automation—now pose the biggest threats to airliner safety world-wide, the study concluded. The results can range from degraded manual-flying skills to poor decision-making to possible erosion of confidence among some aviators when automation abruptly malfunctions or disconnects during an emergency.

The observers found that in most instances, pilots were able to detect and correct automation slip-ups before they could cascade into more serious errors. But when pilots "have to actually hand fly" aircraft, according to one section of the narrative describing interviews with trainers, "they are accustomed to watching things happen…instead of being proactive."

Pilot lapses and automation were implicated in the high-profile 2009 crash of an Air France Airbus A330 that stalled and went down in the Atlantic Ocean, killing all 228 aboard, just as they are suspected of causing last July's crash of an Asiana Airlines Inc.Boeing 777 during a botched landing in San Francisco.

With the reliability of engines and flight controls continuing to improve, airline pilots spend the vast majority of their time programming and monitoring automated systems—typically relegating manual flying to barely a few minutes during takeoffs and right before touchdowns.

Overreliance on automation, however, has been recognized for years as an industrywide problem, with numerous earlier studies delving into the consequences.

The 34-member committee, for example, agreed that "pilots must be provided with opportunities to refine" manual flying skills, while receiving enhanced training in computer complexities and automation modes. In addition, the draft recommended training for rare but potentially catastrophic malfunctions "for which there is no specific procedure" or readily available checklist.

The panel also called on manufacturers to develop cockpit designs that are "more understandable from the flightcrew's perspective" and specifically guard against technology failures resulting from integration of various onboard systems.

Problems with training

A couple of links picked up from Twitter today about the limitations of training.

The surprising science of workplace training

http://us2.campaign-archive1.com/?u=bc04df008d4705e4e77c2eb35&id=12a0719723 by Annie Murphy Paul

Training is not as intuitive as it may seem.  Eduardo Salas, professor of psychology at the University of Central Florida states that “There is a science of training that shows that there is a right way and a wrong way to design, deliver, and implement a training program.”

Here are some pointers

1. It is not just what happens during training that matters. Actually, what happens before and after training can be just as important.
2. Trainers need to distinguish between content that is “need-to-know,” and that is “need-to-access.” For the latter training should teach people where and how to find that information rather so they don’t have to rely on memory
3. Testing may not be an effective part of training.  Research shows that training is more effective when it’s presented as an opportunity, rather than as a test.
4. Skill decay is a major problem in training.  Trainees lose up to 90 percent of what they learned within a year.  This can be prevented by giving workers frequent opportunities to practice their new skills, and by scheduling “refresher” training.
5. Employers can act to increase workers’ motivation: by being clear about the link between what’s being taught and how it will be used on the job, and by making sure employees feel supported in their efforts to learn.
6. Workbooks, lectures, and videos are usually poor for learning when compared to practice and feedback components.  Training can be improved by making it more active and engaging for participants.
7. Workers will get the most out of practice when they are provided with constructive and timely feedback that identifies what they may be doing wrong and how to fix it.
8. Research shows that conditions that maximise performance during training are often different from those that maximise the transfer and retention of those skills.  
9. Because errors often occur on the job, there is value in training people to cope with errors both strategically and on an emotional level.  Guiding workers to make errors, and then providing them with strategies to correct their mistakes, will lead them to understand the task in greater depth and will help them deal with errors on the job.
10. Technology must be implemented in a thoughtful way, in accordance with scientific findings, in order to add to the effectiveness of training.
11. Left to their own devices, workers may not be knowledgeable or motivated enough to make wise decisions about how and what to learn.
12. Training using simulations does not require the virtual setting to be precisely the same as the one the worker will encounter on the job. What matters is not the “physical fidelity” of the simulation, but its psychological fidelity—how accurately it evokes the feelings and the responses the worker will have on the job.

The Trouble with Training

Centre for Leaning and Performance Technologies by Jane Hart 7 June 2013

In most cases this training requires participants to take time out of their daily jobs – often going to a separate place or room. Although more recently learners have been able to sit at their own desks and complete online courses, they have still had to stop what they were working on in order to study the course. And more and more people are now beginning to question the validity of this model to address all learning problems, citing its ineffectiveness, the fact that it relies on study and memorisation, as well as the cost and time requirements of developing instructional solutions.  

“A recently-published report by the UK’s National Audit Office (NAO) estimates that the Civil Service wastes hundreds of millions of pounds every year – some £275m in the last year alone – putting staff through training courses that “do not work”. Less than half of the staff questioned by the NAO felt the training they received in the past 12 months had helped them to do their job better.” (Bob Little, Checkpoint eLearning, November 2011)

“Companies’ spending on training and development accounts for hundreds of billion pounds globally each year. But every year, according to successive empirical studies, only 5 to 20 per cent of what is learnt finds its way back into the workplace. While this failure to transfer and apply new learning in the workplace has long attracted academic interest, practitioners have been slow to change their ways. Despite the imperative that things cannot be managed without being measured, training has been getting off lightly. Surely a training industry that delivers less than 20 per cent cannot be fit for purpose?” Accountability needed for workplace training, Robert Terry, FT, 12 December 2011

10 reasons NOT to create a course and 10 other options

Learning in the Social Workplace by Jane Hart (Blog) 23 March 2013

Here are the 10 reasons why you should not produce a elearning course:

    You don’t want to take your people out of the workflow unnecessarily.

    You don’t want to bore your people to tears with page-turner/click-next solutions.

    You don’t want to treat your people like idiots making them click on every link or action button in a course – because their manager thinks that’s proof they’ve read something and hence learned it!

    You don’t want to dumb down the learning process and make your people have to  work through trivial interactions – in a desperate attempt to engage them.

    You don’t want to force your people to stay on a course for a prescribed amount of time – just to prove they’ve had the required length of training.

    You don’t want to require your people to communicate with one another in a course – because that’s what others think “social learning” is all about.

Tributes paid to Trevor Kletz

From IChemE 1 November 2013

IChemE president Judith Hackitt has paid tribute to safety pioneer Trevor Kletz who passed away yesterday at the age of 91.


Kletz was one of industry’s most respected figures following a celebrated career as an industry safety advisor, lecturer and writer.


Hackitt, who is also Chair of the Great Britain Health and Safety Executive said: “Trevor’s impact on industry was striking. His ability to convey safety information succinctly, and effectively, was central to his success. On behalf of IChemE, I extend our sincere sympathies to his family and friends. We will ensure the memory and work of Trevor lives on within the chemical engineering community.”


IChemE chief executive David Brown says that the impact of Kletz’s work will be felt for many years: “Trevor unquestionably saved lives. There are people working in the process industries today who will go home safely to their families and loved ones, thanks to Trevor. He had a profound impact on industrial safety.”


Kletz worked for ICI from 1944 to 1982. In 1968 he was appointed as one of the process industry’s first technical safety advisors with a broad remit which included advising designers and operators about how to avoid accidents, specifically with regard to process accidents. On leaving ICI, Trevor built a second career as a process safety consultant, writer and lecturer. He was elected a Fellow of IChemE in 1978, a Fellow of the Royal Academy of Engineering in 1984 and awarded an OBE for services to process safety in 1997. He authored fourteen books and more than one hundred peer-reviewed papers on process safety.


Kletz remained active professionally until earlier this year where a formal retirement reception was staged at IChemE’s Hazards 23 conference in Southport, UK.


A full obituary will be published in the December issue of tce.