3.6.7 Take off without controller (Dronekit only) crash

Hi everyone!

I’m writing applications using dronekit to control arducopter based drones. So far it is going successfully. I gained lots of experience on dronekit. Today I was doing more test flights to gain more confidence with my setup and look for errors. I have done two test flights:

  1. With remote plugged in to arducopter for emergencies (radio failsafe on, radio on, waiting on my lap)
  2. No radio plugged in (PPM cable plugged out)

My first flight was a total success. No problems, no anamolies. My second flight was a crash which I believe was not due to a remote being absent. After I unplugged the remote I rebooted the autopilot, waited for it to get gps lock, all ekf and gps was fine.

I issued a take of command also with channel 3 override (so it would arm) After taking off I experienced wild yaw spins, quite fast and I immediately issued a land command. While It was spinning wildly it start to to descent and before reaching the ground it also lost pitch and roll controls and just hit the ground nose first (two props and the front of the quad are obliterated but the back is all fine)

I’m suspicious on the following:
One motor was loose and started making dangerous vibrations that messed up the ekf
I did not re calibrate remote after unplugging it, arducopter default rc in values made it spin??? (I don’t think so)

I tried to analyze the logs a little bit by myself I believe it is a hardware failure.

Here are the logs:
Successful flight with the remote: https://drive.google.com/open?id=1Ro6sQnIKnX6JjflNeSgqi6_1D6t10ich
Crash: https://drive.google.com/open?id=14lKCUoRvrDWzpOqKAon_X2hGFFzGliJf

I’m attaching the previous successful flight because it was couple of minutes before the crashed one, so if a motor was getting loose there might be some evidence there as well.


I’m using pixhack 2, the imu is on another board with vibration dampers, i’ve discovered that it was a little free, but it can be after the crash as well, also I have one loose motor as well, but again that can be caused by the crash as well


Just took a quick look but to me the issues are due to the RCIN being ansent in the crash flight. Basicly all rc channels being at zero was like all sticks were deflected to the angles. I think that is why it took off spinning and having pitch and roll all over the place.

I hope thats the problem, for my next flight I will limit the yaw speed so if it is from the RCIN being zero It will spin slowly… I’m 3d printing the drone at this very moment.

There is new parameter in 3.7-dev called RC_OPTIONS you can disable rc receiver with it.

You have same prob on pitch and roll

I just simulated this on dronekit-sitl (gave all the channels their minimum value). The quadcopter lost control even in simulation, had EKF problems and gps glitches. I’m currently developing a code to override rc channels on boot to use without rc controller.