“Crew Forgets To Raise Gear” (Avweb) describes an Airbus A320 flight where a robot could have helped:
An Air India flight was diverted last week because the crew didn’t raise the landing gear. There were apparently lots of clues that something was terribly wrong but the A320 crew pressed on, climbing to 24,000 feet instead of the normal 35,000 to 37,000 feet and reaching only 230 knots as the plane gobbled huge amounts of fuel to beat the drag of the wheels. About 90 minutes into the flight from Kolkata to Mumbai, the fuel state demanded a diversion to Nagpur. They reportedly didn’t realize their error until they went to drop the gear for landing.
Imagine a camera up in the dome light watching the same instruments and indicators as the pilots. The robot could use the camera image to ask “The altimeter shows us at FL240; the FMS shows that we’re still 400 nm from the destination; are you sure that you need the landing gear down?”
Obviously this would be more useful for single pilots and non-professional crews, but even the best pilots can make mistakes after a few days of airline flying.
Related:
- post from 2010 wondering why the fancy computers on Colgan 3407 didn’t ask “why don’t you push the power levers forward?” (50 people died for want of a few IF statements)
Would such a system need to through the same certification process as flight hardware and software?
40 years ago, the same alarm which went off based on minimum altitude would go off based on maximum altitude & the state of the landing gear. Nowadays, it has to be a camera with image recognition driven by a convolutional neural network, written in functional style React JS(TM), on a 50 core CPU, with 1 terrabyte of RAM.
Great jobs program suggestion, Jack. Who needs Star Track? We are trying to figure out how to manage landing gears notifications.
Neal: portable electronics that are brought into an aircraft are not subject to the same process as stuff that is hardwired into the aircraft. That’s why you have ForeFlight and other phone/tablet apps that are vastly more capable than anything built into an aircraft.
Jack: I agree that you could get this by tapping into signals within the aircraft’s wiring harness. However, that would likely drive up the cost and time to deliver the product to the point where it could never be economic to build.
Forwarded your post to my CFI, who responded:
“One fact which is left out of this article was that it was an all female crew…. not that that means anything, but it is a fact. …”
For some strange reason, the story completely omits the fact that both pilots were female. Here’s a more complete report, from a British tabloid that doesn’t censor such details: https://www.thesun.co.uk/news/4127604/air-india-flight-nearly-out-of-fuel-pilots-retract-landing-gear/
Four months ago, when an all-female crew did something or other for the same airline, it was big news: http://www.independent.co.uk/news/world/asia/air-india-female-crew-first-round-world-flight-pilots-flying-gender-equality-a7611591.html
And here’s another female-pilot-boosting story from the same airline, the day after two female pilots forgot to retract the landing gear, and failed to notice the large drag that would have caused: http://www.cntraveler.com/story/air-india-pilot-becomes-first-female-commander-of-a-boeing-777
An A320 IS a robot, but somebody does have to supervise it. This is a situation no designer could imagine, kinda like trying to take off in reverse.
What was the impetus for pushing through incompetent pilots? I could see a western company doing this, but Indian culture doesn’t strike me as insanely progressive just yet?
Maybe the CEO has a bet to win at Davos, “whoever manages to push through the dumbest possible affirmative action hire wins $1.00” ala Randolph and Mortimer Duke. Surely sending a plane load of people to die has to win top marks…
If you are going to technologically solve the AA pilot issue, you couldn’t rely on machine vision. The ditzy broads could be busy taking the perfect selfie and block the cameras view of the warning lights. A better solution is to tie in to the avionics network on the back end and have the plane flown drone-like by competent white male american baby boomers.
The airline gets all of the credit for having the right mix of underrepresented minorities or womyn or mental-disorder du jour, yet keeps its customers alive.
Probably doesn’t need a certificated robot in the dome light, may only need a hacker listening to the bus back in coach.
philg: “portable electronics that are brought into an aircraft are not subject to the same process as stuff that is hardwired into the aircraft”
Do you know how common ForeFlight (or similar) usage is on commercial flights?
In addition, do we really need a camera for this case? An app that monitors coordinates, height, speed and know the type of the plane it is, could say “You fly too low and too slow for too long. Let’s check for common problems together. First, what about landing gear?”
the other Donald: “An A320 IS a robot, but somebody does have to supervise it. This is a situation no designer could imagine, kinda like trying to take off in reverse.”
It would make sense if this particular human error never happened. But it looks like pilots do make this mistake. Just look at the avweb article comments:
“OK, I feel just slightly better about my (more than once) failures to raise the gear. Each time it involved some distraction at liftoff time, wonder what this one was?”
If this problem is indeed happening, of course the airplane/app designer could imagine it and try to prevent it.
Aurora’s ALIAS [1] is creating a fully robotic pilot that looks at the gauges and moves the controls like a human. I suppose that just the visuals and proposed resulting actions without the robot movement could be available. Though it would decades before this would be a standard feature.
[1] http://www.unmannedsystemstechnology.com/2017/05/autonomous-co-pilot-flies-lands-simulated-boeing-737/
last paragraph of previous link: “Aurora is also working on a version of the system without robotic actuation that instead aims to support the pilot by tracking aircraft physical, procedural, and mission states, increasing safety by actively updating pilot situational awareness.”
Okay, the fact that they missed the gear up call during the checklists is mind-boggling. But it can happen, I guess.
But the fact that they didn’t recognize what had happened after a few minutes is crazy. These two definitely don’t have the right stuff.
That same “Dome Light Assistant” would probably have told Captain Sullenberger and First Officer Skiles NOT to land in the river. Had the listened to logic, there’d probably have been tremendous loss of life.
Stuart: if that Airbus had been connected to the Xavion app on an iPad it would have landed itself at LGA. See http://www.popsci.com/xavion-ipad-app-can-make-emergency-airplane-landing-autopilot
(Separately, in the absence of a reachable runway, a rule that said “don’t hit anything if you don’t have to” would naturally guide an aircraft to the flattest nearby area and therefore a computer also might have chosen the river. Remember that none of this is anywhere near as tough for a computer as driving a car. That’s why the first aircraft autopilot was demonstrated in 1914, nearly 100 years before the first self-driving car experiments. See http://jalopnik.com/lawrence-sperry-inventor-of-the-autopilot-and-the-m-1592623110 )