This was a crash reported on Iris Owners facebook page. This is not my log but I had some questions about it. The video that was included with this log shows the pilot taking off and then hovering for a while. After a few seconds the quad just shoots up and becomes unresponsive. It eventually starts rolling right and crashes into the ground. After looking at the logs, it looks like one of the IMUs failed. I believe the deviation between the 2 IMUs (especially in the z axis) is what caused it to shoot off. You can see the baro recognizes that it is gaining altitude but the FC thinks it is losing altitude. So here are my questions. Am I correct in saying that one of the IMUs failed and that’s what caused the crash? Also, if that is the case, would switching to Stabilize have stopped the rapid climb? Even if Stabilize mode stopped the climb, would it have given the pilot full control back and prevented the crash? I’m not sure how the 2 IMUs are being used in 3.2 and if one fails does that guarantee a crash? Last question, I don’t believe the pilot had EKF enabled, but had he, would it had caught the failed IMU and been able to stabilize out and not crash?
Thanks for this report
I have been following this on the Facebook page, it was great to see the diagnostics going on there!
Thanks for posting here and brining this to the attention of the team
Yes, the MPU accelerometer failed. Just after line 24,000. Definitely caused the crash.
Yes, switching it Stabilize probably would have stopped the climb. The EKF, had a small error due to the failure, but also would have been under much better control. It only had about a 20m error.
It’s hard to say if this would have been totally controllable in Stabilize, as this failure will also mess up the attitude estimate. I have no way of knowing if the attitude estimate was right or wrong. Hoever, the DCM estimate is close to the EKF so chances are both would have been OK at least in the short term. You would definitely want to land ASAP.
Unfortunately the pilot did the worst possible thing, switching into a high-auto mode, RTL. In this mode, the bad accelerometers became critical and it forced the crash.
Whenever my copter does something unexpected, I immediately switch to Stabilize.
Sorry about your crash
Please ask the original poster to contact help@3DRobotics so we can get this vehicle RMA’d and examined.
Thanks for the help guys. Philip Rowse validated my theory as well on the Facebook page. I’ve been trying to act in a support role on the Facebook page and have reiterated multiple times to all the new pilots that if their bird is not flying right, flipping it to RTL is the worst thing they can do. I’ve been recommending to everyone to make Stabilize mode one of their flight modes and if they start to see erratic flight, stabilize is their best chance at regaining control.
Are the different EKF parameters documented somewhere? I would like to understand how to determine the differences in how EKF would respond versus DCM. Thanks again for the help.
It’s a bit of a complicated topic, but this is a good page to read if you want to get into it:
plane.ardupilot.com/wiki/common- … r-overview
This concerns me as the current IRIS+ is being shipped with Stabilize as not one of the selectable modes. The only option would be to select LAND as a mode but that may not help with system issues like this
I’ve had 2 fail the same way, 1 causing a crash.