APM response to catastrophic air frame/wing failure?


All -- what do you believe is an appropriate response of APM (ArduPlane) to a complete air frame failure?

Over Thanksgiving break, I test flew an Experimental Airlines AXON build (in a remote, rural area). The plane was flying fine with default settings, and I flew from FBWA to RTL to Auto. During the Auto mission the main spar (an oak dowel) failed and the wings collapsed at around 300' AGL. The plane, as you might expect, took a nosedive to the ground. If my recollection is correct, I could hear the motor running at full throttle as the plane headed to the ground -- my guess in an attempt to pull itself out of the dive. I didn't have the time (or the reflexes) to switch to manual and cut the throttle myself.

Is there ArduPlane code that could somehow detect such a failure and cut off the throttle?

I certainly don't plan on this happening again -- carbon spar from here on out -- but you never know...



E-mail me when people leave their comments –

You need to be a member of diydrones to add comments!

Join diydrones


  • So you want to ride your tricycle blindfolded into freeway traffic and then expect that some iPhone app should sense a bus bearing down on you and whisk you up into Winnie the Pooh's tree?  Would you like fries with that?  

    1. Learn how to fly. Difficulty: none, just some time and fun. 

    2. Don't fly junk. Difficulty: none, just abandon false economy and spend a very few extra dollars on a  simple but solid, proven airframe.  Stay away from designs of the "Look, Mom, I made an airplane out of flattened beer cans and hot glue" variety. 

    3. Ignore bad, stupid, dangerous, uneconomical advice to keep building and crashing cheap-ass weak airframes that are big enough to haul an APM etc. Crashing is in fact a big deal, and cost is not main the reason why. Sure it happens to everyone, but so does dying. Crash once into the side of your new car (I did that...) and you'll never forget it, but that doesn't make it useful. You "learn" from something like that the same way you "learn" from being bitten by a bad dog. Not what you want. 

  • Implementing autopilot behavior switch on an already dead airframe - would that not introduce more risk than it would mitigate? Like if it falsely triggers etc.

    But it would be nice to have a working post crash detection and shutdown, which might save the battery and/or motor.

  • Moderator

    Eric maybe you should pick up a few Bixlers and learn to fly on them first, if I were running a student program I would want my students to be able to fly a manual circuit before I let them loose on a platform with an autopilot in it. That is in fact what I do if anyone comes to me for advice, I make them fly in manual. That way you can recognize when things are going wrong and have a realistic flight profile in your thumbs.  

  • Doug, it doesn't have to be very complicated to limit airframe load. It doesn't even have to be close to the ultimate load capability of the airframe. Just a programmable z acceleration limit would be sufficient to avoid cratering due to excessive control inputs, either through manual or automatic flight control.
    Most manned aircraft have specified g limits and limit testing isn't beyond the scope of most DIYers
  • Thanks all for the wonderful responses. I've learned how to use the log browser to peruse the log file (available here: http://goo.gl/oaIsPd) and saw that I did switch to FBWA (although I don't remember doing this) as the airframe failed. With my current flying skills, manual has largely been off limits, so there are no quick reflexes yet.

    The gyros did go a little crazy at the end:


    Here's the graph of the throttle (ThrOut) and pitch:


    Given other comments about saving a disabled plane, I don't feel qualified to comment on the proper behavior of ArduPlane. Looking at the pitch log after the wing failure, my plane nosed up before it headed down. But, it wobbled more than I had perceived on the way down -- the pitch was a high as -20 for a moment. So, if code existed to cut off the throttle after a consistent nose-down position, it may not have triggered in my case.

    As for the AXON, I believe I only lost the replaceable nose, one servo (HXT900), and perhaps one lipo (hadn't put the $400 camera in it yet). The Pixhawk and other electronics appear to be okay. I'm rebuilding the plane at the moment, so the crash hasn't dulled my interests.

    As Nathaniel guessed, I'm interested in using this platform for aerial mapping, and the AXON appears to fit the bill well for a low-cost, modifiable, replicable, and replaceable plane. I am learning along with my students (I teach GIS at a Wisconsin university) -- this won't be the last airframe that we destroy.



  • Developer

    ArduCopter has a crash detector (see video on this wiki page).  The detector compares the vehicle's roll and pitch angles to the desired roll and pitch angles (either from the pilot or autopilot) and if they're off by more than 20degrees for a full 2 seconds and the pilot's throttle is at zero it cuts the motors.

    Maybe something similar could be done for Plane.

  • I agree with Andrew Rabbit. You can make some assumptions on how your aircraft should fly if everything is working correctly. If the accel/gyro absolute angle says that the nose is pointing down past 75 degrees, then you can assume it's not functioning correctly and shut down the motor. Put a delay in there if you think you will hit rough seas while flying. Same type of thing could be done to detect a flat spin. For autonomous flight I would want the motors to shut down if something went out of reasonable parameters during flight.

  • Moderator


    The Axon is a great plane to use as a test bed for the APM IMHO. Inexpensive, quick building, flies well, and can be easily modified to suit your needs. If you had this disaster with an airframe in which you had invested a lot of time and money, you may very well have been set back and given up. However that's one of the great things about this building method, you crash, big deal! Build another one LEARN from the crash and move on. Total loss in time and money.....minimal. Crank out a dozen of them if you need to, and still have money in the bank.

    I understand the attraction to the Experimental Airlines series, that said, once you do get comfortable with the APM and it's capabilities, how to tune it, you will undoubtedly want to move on to a platform that has better aerodynamic properties :)

    Build another one and keep at it!

    Please post the log from your last flight and we can help you with any changes that might make your next flight less exciting!

    BTW you can buy carbon fiber arrow shafts at Walmart for about $3. They make good spars for the Axon and similar aircraft.

    Good Luck,

    Nathaniel ~KD2DEY

  • Andrew, we are talking about a $3 airframe here... To have APM 'know' the mechanical limits of all possible airframes (fixed wing only in this discussion) would require an agreed upon standard in destructive (or near destructive) testing of the airframes to determine where the limits were.

    That's how airframes are certified by government agencies and the firms that build them.

    (There are many, many images of wing load testing on the web)

    Phoenix+wing+load+test+from+tip.jpgOf course the engineering and math was done before the loading tests were performed. Still the stresses have to be induced, the data analyzed and then agreed upon limits set in place for the software. Which takes me to my original point of...

    The APM would have to have an airframe supervisory sub-system to alert it to this kind of failure.

    That is waay too much to ask for APM at this point. Better to use a known reliable airframe, quality parts, and as Gary mentioned:

    Keep trying to fly is what I would want it to do...

    One thing we are missing is how much flying experience Eric has and the budgetary constraints he is operating under. If new to the RC thing, build another Axon and fly the butt off it before putting his APM (if it survived) back in it.

    I'd build him a decent airframe if he lived close to me. I'd love to see his work with GIS as acquired with RC airplanes and APM.

    Eric, Do not give up!


  • It would be easy enough to put in a throttle limit as a function of pitch angle.
    I think it'd be better if APM didn't let the airframe be overstressed in the first place
This reply was deleted.