A team headed by an award-winning professor at the Massachusetts Institute of Technology, John Hansman, a professor of aeronautics and astronautics and engineering systems at the school as well as a 5,600-hour pilot, has come up with a really cool idea: mining data from flight data recorders to identify anomalies. Check out our news story on the subject here. The researchers use algorithms to group together similar datasets from different flights and then spot anything that seems out of the ordinary. For instance, the group's initial research showed a pilot on one flight using an unusual flap setting — that one must have stuck out like a sore thumb — and descending below glideslope. Once the data analytics tool identifies such an out-of-the-ordinary event, analysts can then take a closer human look to see if the event represented a normal operation or if it was an indicator of a problem that might someday cause an accident. If the latter seems to be the case, the airline can take measures to remedy the situation.
Exactly what kind of measures those might be is the subject that worries pilots unions. It's the old concern that the airline is going to be looking over their shoulder at every move, making every flight, in essence, another checkride. The subject is far from academic. With careers, seniority and dollars on the line, most pilots are loath to let the airlines or the feds have any more data on them than is already in the contract. There is already a program in place to let airlines take a limited snapshot of pilot performance. Flight Operational Quality Assurance (FOQA) is a voluntary program intended to do much of what the MIT research project is looking to do, but in a limited way and with protections and perks for pilots written into the agreements.
Data reporting has never been a touchy subject for pilots of light airplanes because there was no way to generate or share any useful kind of data, but that might be changing. The airplane I regularly fly, the Cirrus SR22, has built in engine data recording, which the operator of the airplane — in my case, this was shared ownership provider PlaneSmart — can use to monitor engine parameters in flight. This can be very useful to monitor trends and spot hot spots before they become emergencies, which is how PlaneSmart uses the data.
It's not hard to imagine this kind of data monitoring going to the next level. With the current sophistication level of light airplanes, with GPS, AHRS and FMS common, a limited set of flight parameters, including altitude, airspeed, attitude parameters, density altitude, and bank angle, among many others, could easily be made available to the FAA (or some other agency or entity) for monitoring.
One thing the analysts could do with these morsels of information is try to spot, as Hansman's team is working to do, anomalies. What are some pilots doing that doesn’t jibe with what the data analysis would expect to see under those circumstances? Granted, this would be a lot more difficult in the GA world, as there are so many different kinds of airplanes flying so many different kinds of missions that finding that common data set from which to compare could be a challenge.
But in theory the benefits of such a data-monitoring program could be tremendous. By spotting the kinds of pilots who have accidents — and I am of the opinion that many GA accidents are caused not so much by "pilot error" as by "pilot predisposition" — the feds could, again in theory, remediate and thereby cut the accident rate.
Could this ever happen? Maybe. But one big obstacle would be the reluctance of many pilots to submit to this kind of monitoring. Some of the most reluctant would doubtless be those most worried about having their flying behaviors put under the data-scope becaue the like to push limits. Others would feel the monitoring was a violation of their right to privacy — if they did nothing wrong, why should their flight data be subject to government review? Yet others would simply fear an agency, the FAA, that they have come to distrust, and for good reasons.
The fear of detection argument holds no sway with me. If you're flying like an idiot, you deserve to be held accountable. The privacy argument doesn't move me either. We're monitored on radar most of our flying lives and are required to transmit a squawk code most of the time when we're in controlled airspace. At least to a point, the FAA has a compelling interest in using surveillance technology to safeguard the airspace.
That last objection — fear of the FAA — is clearly the hardest to dismiss. I've seen enough abuse of power on the part of the FAA to last me a lifetime. That fear would be the deal breaker for me.
If the National Transportation Safety Board or NASA were to be charged with mining and administering the data, that would a different story. Those folks seem to understand the sensitive nature of the convergence of safety and privacy and they tread lightly upon it.
If data monitoring is to ever take off in light airplanes — and I hope that it does some day — it will take a light touch indeed to make it work and make it work solely for good.
All Comments
Robert, I'd have more faith in a private organization collecting and analyzing the data, for a very simple reason: private organizations can make enforceable commitments, but the government can't. If the government (FAA, NTSB, NASA) collects the data with a promise that it will be used in a particular way, it can later simply change its mind, and there is no recourse. A corporation or non-profit that makes a similar commitment, can't (without getting sued).
I'd be relatively comfortable with, say, AOPA or EAA (or Google or Yahoo or Garmin) providing a portal where I could anonymously submit my data and get comparative statistics with other pilots, for me and my flight instructor. Heck, I might even want to get a rating I could share with my insurance company. But I know that giving the data to the FAA is just the first step toward an enforcement action, sooner or later, probably for some obscure violation.
Good topic!
Interesting idea, Thomas. The NASA self reporting program that's been in effect for around two decades now is an example, and I should have pointed this out, of a government entity keeping its word. There has never been an enforcement action to our knowledge against a pilot who used the self-reporting system, even though the FAA and NASA know most reports to done to immunize the reporter and not in the altrusitic furtherance of safety.
I believe this would be a step in the wrong direction for GA. One of the greatest feelings as a new pilot is the overwhelming feeling of Freedom when you first get your license. The pilots I know would not appreciate "Big Brother" evaulating every second of every flight. Police officers are struggling with a similar battle. People are video taping their actions with personal cameras and it is having an affect on their real-time decisions. You shouldn’t question split-second decisions made in critical circumstances. If a mistake is made, a proper investigation will give you the whole picture and hold the appropriate people accountable.
Thomas has the right idea. Have a 3rd party that you can voluntarily submit your flight data to. The anonymous stats should be used in the interest of improving safety for the whole system. Making this mandatory and assigning a government agency to manage it will be a financial and Personal Privacy disaster. It will only further reduce the already shrinking number of people supporting General Aviation.
FOQA is a monitored program between the company, the union and the FAA, one in which anybody can pull the plug if they don't like the way the others are playing.
Giving the FAA access to anything like this for GA is a HUGE mistake.
I would like to agree with the commentator that the the idea of analysing such information is excellent but certainly not by a government agency. In Australia we have confidential (?) reporting, but any inference to neglect or culpability by our CASA and they will take all action necessary to defend themselves.
Safety reporting is the backbone of any system provided the information is used for the common good. In our case with an overly complex and outdated regulatory system which can be interpretted in a multitude of ways the CASA will go to any lengths to prove themselves right, especially in cases where the opposite is true. Vexatious actions are common.
NOT A CHANCE. I HAVE SO LITTLE FAITH IN OUR CURRENT FEDERAL GOVERNMENT, THAT I WOULD NOT WANT TO PROVIDE THEM WITH ANYTHING NOT ABSOLUTELY DEMANDED BY LAW. I HAVE NO IDEA WHAT THEY WILL DO WITH THE DATA And how it will be used for or against the aviation community.
We cannot get rid of the third-class medical! We certainly don't want to give the FAA any more power than it already has.
I've read a number of accident reports where there simply wasn't enough information to figure out what happened. And then there is the student pilot / instructor loop - analyzing the student's performance during solo time. And finally, just the ability to better and proactively manage my aircraft's maintenance. Oh, don't forget managing the cost of flying.
The concept of a "black box" which could provide the basis for all of these is a "good idea" - yet I agree with the respondents who are saying "What, let the government have access? What brownies have you been eating for lunch???!!!"
Technology has reached the point where we can collect data from our various boxes (e.g. my Garmin GNS-530W / JPI EDM-800 / Shaddin ADC-200) and stream the data in near real time to someone. Depending on the service agreement, this might be nothing more then a data warehouse, flight modeling service, or an aircraft maintenance analysis service. It might even be a service endorsed by my insurance underwriter (use our flight analytics service and receive a 5% policy discount).
My point simply being: there are a number of great business opportunities which benefit the GA flying community - and they don't involve plugging Uncle Sam into the loop to do it. While the NTSB and insurance industry most likely would demand access to data in the case of an accident, the rest of the time the operator should have control of the data and its use.
So.... Mr. Goyer, you fly a SR22 with some data link capabilities tied to "PlaneSmart". What if, let's say, next month you were to receive a letter from the company that provides insurance for your rented aircraft (or put in place, an aircraft owner) that indicated, based on information downloaded by "PlaneSmart" and shared, at a price, to other data centers, that based on your flying skills, they were cancelling your insurance coverage for ownership of, or rental of, any aircraft.
Basically the "NEW" downloaded data showed that your ILS approaches were out of limits, your approach speeds were to high for conditions, you used way to much braking in your landing rollouts, and you like to do a lot of slow flight (which we all know leads to stall spin incidents). For the lousy $1200 a year in renters insurance you pay, (for the insurance company to cover a 300K asset) you're to high a risk.
Insurance companies conclussion .... your airmanship sucks, and now we have proof, so don't even bother to schedule your next medical. Your flying days as PIC are over.
p.s. Hope you have enough personal assets to cover any losses if you're planning on flying anyhow.
Welcome to the future - where flying an airplane just became a liability, we can no longer afford.
Oh!
Remember that WE are the GOVERNMENT of this country! It's private industry that scares the hell out of me.
An observation, based upon what I've seen with Transport Canada and the FAA: TC's objective is to improve aviation safety, while the FAA's appears to be to punish transgressions.
In Canada, a minor violation will typically be dealt with by having the pilot write an explanation of what happened, and how s/he will avoid a recurrence of the situation. Unless it was an egregious violation, or the pilot is a repeat offender, that will typically be the end of it. The FAA on the other hand...
I hold both Canadian and US licenses, and I must confess, I fear the consequences of a transgression in the US far more than in Canada!
It may be time for the FAA to reconsider its stance, and try to determine exactly what it is trying to achieve, and whether its current approach achieves those goals.
Now, I'm going to generalize beyond the FAA: the US does seem to have this rather puritanical fascination with punishment. Consider the following signs found in the corridor from Portland to Blaine, WA:
"Click-it or TICKET!"
"Litter and it WILL HURT!"
"Picking these flowers will incur a $2000 penalty"
... and so forth.
(By contrast, in Ireland I encountered the following sign: "Ladies and Gentlemen will not, and others must not, pick the flowers." Rather more genteel, don't you think?)
I would consent to a joint partnership between AOPA, EAA, Air Safety Foundation collecting the data and making aggregate statistics (not individual data) available for free to anyone who wants it. In addition, I would consent to my individual data being used between myself and that joint partnership to, for example, compare myself with the aggregate averages, make personalized recommendations, even make personalized recommendations of products/services/training options geared towards my types of operations (i.e. Amazon.com's personal recommendations). However, I would not be comfortable with a 3rd party (i.e. a government or insurance entity) having access to my individual data. There is too much room for abuse and corruption.
Robert,
The FAA has not misbehaved with the ASRS system, because reporting is voluntary and there is no practical way to require reporting - it can't be automated. So, to get the reports, NASA and the FAA have to behave.
On the other hand, look what FAA has decided to do with transponder returns and the associated personally identifying information! It can require the data collection, and therefore feels free to act badly.
As you can see from the comments, government has got itself into a position where it has a lot of work to do to regain trust. I realize that many people - including you, I think - do trust the government, perhaps based on personal contact with some of the many well-intentioned people who work for it; but others have a well-honed mistrust of any institution with arbitrary power, and the FAA's sporadic bad behavior doesn't help matters.
As the aviation safety officer of a busy company - I am absolutely for coming up with ways to collect more data points which would given us the insight to develop better behaviors, training, and procedures. Data that would help us locate habits which are a link in the chain to an accident or incident would be welcomed.
However, given my experience with some - not all - of the folks at our friendly aviation administration - sending them full, real-time, unfiltered access to every moment of all flights is ludicrous. And pilots have good reason to fear it.
Can you imagine, for example, if we utilized the same technology in your car to broadcast to the police every time you exceeded the posted speed limit by more than a 1/2 mile, every time you failed to use your blinker to change lanes or make a turn, every time you drove through a yellow light, every time you got too close to the car in front of you, etc.? Following which you automatically get a citation in the mail? You'd better just keep your credit card on file! For most of us it would result in a monthly bill thicker than that of my telephone! Is that going to make everyone safer and drop the accident rate to zero?
Or would you expect, after not flying IFR for some months and legally going up with a safety pilot to get current, that you should have every deviation from PTS standards you make while scraping the rust off, sent to the FAA for analysis - for which a violation could follow? Are you really saying that no one could ever again make an error of any sort in an airplane without immediate government response? That does frighten me. It's not possible to be error free. Can't be done. What we need is the data to help us manage errors to a tolerable level. I'm for that.
All of that said, I do believe a trusted company with the skill to appropriately parse the information and keep it anonymous from the government and the general public might be a really good idea. Especially if individual pilots could log in and get feedback on their behavior, or companies could see the anonymous error trends in their fleet operations. Or even log in to get performance or equipment checks on their aircraft.
A non-government run program with this technology could go a long ways towards helping us improve the safety levels of aviation as a whole.
As a captain on an A330Airbus, my company can tell, “through the automatic data reporting system,”
what buttons I have pushed and the time I pushed them. What this doesn’t explain is the reason or reasons I might have pushed them. The information given the company was supposed to be used to enhance safety and was not supposed to encourage retribution on the pilot or pilots sitting in the cockpits. What I have experienced has been quite different than the utopia of perspectives offered when the systems and programs came on line. I have seen pilots pulled in for fuel enhancement training because of the information gathered. Your fuel burn is to high, guess what, you might just get a call for driving up the operating cost for that leg and be given additional training to help correct your attitude. Like alcohol and drugs used responsibly and properly, the reporting system is fantastic. Its trend monitoring capabilities of systems have certainly provided useful information in preventing major systems’ malfunctions. The US Navy’s Submarine program has been using such programs since the late 60’s to early 70’s. However, give that information to the wrong regulator, pilot or corporate manager and you will have dug a hole so deep even the best of pilots will not be able to climb out. It is easy to say I have nothing to hide, but those that have been falsely accused will probably say it might be best to not create a scenario that could put you in that “guilty till you prove yourself innocent situation.” Like the slogan, “Zero Defects,” we would all like to see, ”Zero Accidents.” I’m sorry to say automatic data reporting, great as it might be, will not accomplish that end goal. We must always keep in mind that “Safety” is not a procedure, a checklist a technique, or a data collection and reporting system. Safety is first and foremost a mind set coupled with knowledge and good training. The procedures, checklists, techniques as well as data collection and reporting systems will be of no help if not used and viewed though the safety first attitude window.
The Air France accident over the Atlantic is a prime example. The data was being reported faster than one could decipher the information. What is not as easily explained is the confusion that existed in the cockpit. What we are being told is that the pilots kept the aircraft in a stalled condition all the way to the ocean’s surface. They say that it all started with blocked pitot tubes due to icing. However, at some point on the way down to the ocean’s surface, the ice should have melted and cleared the obstruction on those tubes . The aircraft, with all of it’s reporting and automated systems, did not save the crew or its passengers. I would say that all those automated systems were the major contributor to the accident through the confusion caused by those automated sensing and reporting systems in the cockpit. The data reported only shows the flawed pilot inputs not the confused reasoning for those flawed pilot inputs. In other words, “flawed inputs to the pilots resulted in flawed outputs by the pilots. ”
The other aspect we in general aviation need to consider is the cost factor of installing and maintaining the auto reporting systems. Under the current economic conditions we are experiencing, I believe safety can best be served through quality instructors and training. The sense of freedom we find in aviation needs to be shared with the 911 generation. We must be careful that, in our desire to obtain zero accidents, that regulations and intrusive monitoring do not kill the economic ability to obtain the sense of freedom uniquely found in aviation and the need for that freedom to be shared with our future generations.
One of the best comments I've ever read on our site. Thanks!
Being Canadian I don't have a personal stake in this issue. But I have seen instances where regulatory bodies did a 180 on policies they originally claimed were to be permanent.
I think the short answer to this idea is: Given your FAA's infamous culture of violating minor and major errors by pilots, why should they get access to data describing specific pilots' mistakes? FAA already has incident and accident reporting systems in place. I agree with Tom Boyle that as a general rule we cannot trust regulatory bodies to agree to use this data solely to enhance safety.
I shudder at what might arise from providing this type of specific data to insurance companies. Given their industry's infamous abuses in health insurance, denying coverage to people with in many cases irrelevant preexisting conditions, what is to stop America's aviation insurers from denying policies to pilots who don't fly exactly s the companies prefer?
We are all human and thus are all fallible—even politicians and regulators—capable of making mistakes or having a bad day in the air. Not all of our landings are greasers and occasionally we might dip a tad below the middle of the glideslope while still thousands of feet above the ground. Does anyone really think punishing our minor mistakes is a good way to improve saferty? If the FAA does adopt such a system and uses it to find violations, will American pilots have an organization that works to point out mistakes and/or abuses by the same FAA employee(s) who wrote them up?
Such streaming data might best be used by collecting and correlating it so that it may be employed to enhance aviation safety in an anonymous way. For example, it can spot trends/tendencies/anomalies which have the potential to cause and/or exacerbate unsafe conditions. Since most accidents are caused by a cascading chain of mistakes or misjudgements, perhaps such data can better identify sequences of decision making that tend to cause safety concerns. Then those "error chains" could be identified and stressed during ground schools, flight training and biennial flight reviews.
As long as this is done to improve safety, rather than employ FAA employees who write-up violators, perhaps this data mining could lead to preventing more accidents every year. .
Douglas M
Surrey, British Columbia
@afello - The "nice" thing about commercial vs. governmental use of the data is that commercial use can be constrained via contract while governmental use has been shown to be non-constrainable (regulators will do whatever they deem appropriate regardless of public statements or even published regulation). So, given commercial reporting services which I can voluntarily subscribe to, I would have significantly fewer concerns (given a contract I can understand, humm) then anything offered / mandated by Big Bro'.
Besides - if my flying behavior is that poor, then I am a poor risk and my fellow insureds' should demand that I pay an appropriate premium - if for no other reason than to justify they pay one appropriate for their lower risk profile.
