Last month, New Yorkers were stunned when a helicopter crashed into a building on a miserable cloudy day. The NTSB report describes the machine as an Agusta A109E, the “Power” edition of the twin-engine helicopter that came standard with an autopilot.
Thus we have a machine with autopilot servos that can manipulate cyclic and collective. The machine came with a glass cockpit so it also should have at least two digital attitude sources (whether the helicopter is pitched up, banked left, etc.). Finally, it almost surely had a GPS receiver and a digital terrain database, which would have included the obstacles of Manhattan.
Media coverage centered on the pilot’s lack of an instrument rating (example: CNN). (In fact, being capable of instrument flight does not help that much unless one is actually planning an IFR flight from airport to airport with established procedures for departure and approach/landing.)
Nobody seems to have asked “If it had autopilot servos, attitude sources, and a GPS, why couldn’t a $10 million helicopter fly itself through the low clouds, away from the buildings, and to the destination? A DJI drone would have been able to do that.”
We expect so much of our phones and so little from our aircraft!
9 thoughts on “New York helicopter crash: why not robot intelligence?”
Because FAA. Certifying any new capability is a truly daunting task, given lack of criteria and usual bureaucracy bias towards denying permission resulting from responsibility avoidance.
If you’ve seen the video of that helicopter (just minutes before the crash) panic-diving out of the clouds then climbing right back up into the same clouds, you’ve got to wonder about the pilot’s thought process…
Seemed to have lost sight of the horizon & couldn’t fly by instruments. He didn’t crash into a building Bin Laden style but landed hard on a roof. The way autopilots are implemented, at least in the 737 videos, makes it look extremely involved. There’s no panic button to just level off. They have to dial in tons of stuff.
Not sure about heli autopilots, but in small a/c just turning autopilot on puts it into “Roll” mode where it will try to maintain wings level (i.e. flying straight). Another button push (in 2-axis APs) turns on altitude hold. The recovery from inadverent flight into IMC as taught during initial VFR flight training involves using attitude indicator to manualky keep a/c aporoximately level, but really more emphasis should be on “turn the autopilot on, then stop panicing”.
Actually some current autopilots for small aircraft do have “level off” buttons. They are not super expensive and just newer versions of autopilots that have been around for years.
What @averros said. Phones don’t require a years-long review process before they’re approved for sale, although they can get stalled for a few months at the FCC.
A friend who worked on private plane avionics described the job as “1/3 hardware, 1/3 software and 1/3 FAA paperwork.”
Not sure about the 109 auto-pilot but the ships I fly allow you to pretty much set things up ahead of time so that you can get on the auto-pilot right after take-off. They also have a go-around feature in the auto-pilot that puts you “wings level” and gets you into a 500 ft/min climb… but you have to prepare because doing it AFTER you get into trouble doesn’t work well.
Why don’t these kind of helicopters come with a “panic button” autopilot that returns you to a stable hover when you get into trouble?
The new DJI drone is pretty impressive at obstacle avoidance, it has stereo pairs of cameras all over it, above below in back, sides , front, specifically for ranging, and also has sonar and IR sensors for landing. If it sees something too close, it stops, but you can override by switching to a manual control mode. In low light it turns on bright landing lights to see the landing spot. It records the image of the spot it took off from, so it can land to within centimeters of where it took off it there are any visually distinguishing features.
Comments are closed.