Drone crash fail safe settings , offboard mode

PX4 Autopilot
1

https://review.px4.io/plot_app?log=452d7328-c961-49d4-b608-33ee82d99133

drone is crashing , i dont know the reason ,can anyone help me out ?

Screenshot from 2026-09-15 13-11-46
Screenshot from 2026-09-15 13-11-461657×553 47.4 KB

2

I would not change the failsafe parameters yet. There are already a few important events in the log that should be separated before assuming this is simply an Offboard failsafe configuration problem.

From the event sequence shown here:

  • Preflight Fail: Flight termination active
  • primary EKF changed 0 (unhealthy) -> 1
  • repeated Switching to mode 'Land' is currently not possible

The EKF instance becoming unhealthy is particularly worth correlating with the actual loss of control/crash timeline.

I would first reconstruct:

  1. the exact point where the vehicle starts deviating from the commanded trajectory,
  2. the EKF health/selector state immediately before that point,
  3. the Offboard heartbeat/setpoint stream and whether Offboard was actually lost,
  4. the vehicle/failsafe state transitions,
  5. and actuator outputs versus attitude/rate response.

That should tell us whether the crash was initiated by an estimator problem, loss of Offboard control, a failsafe transition, or something downstream such as control/actuation.

I would avoid changing the failsafe settings until that sequence is clear, otherwise the configuration change could mask the original failure.

The Flight Review log you linked should be enough to investigate this systematically. I can take a closer look at the event timeline and try to identify the first abnormal state rather than the final crash condition.

3
  1. vehicle was following the trajectory suddenly , the drone is flipping and activating he failsafe settings, and disarrming the drone. 2. ekf is looking okay at that point 3. offboard signal was neveer lost and after the drone crashed ,i clicked on land to be sure, it was safe to touch. 4.vehicle failsafe was triggered when the drone is reaching 80 degree, which is okay but my question is why drone is recahing 80 when i have set the maxtilt angle to 45 degree
4

Thanks, that clarification helps a lot.

If Offboard was never lost and the EKF was healthy when the flip started, then I would treat the failsafe at ~80° as a consequence of the loss of attitude control, rather than the event that initiated it.

The important point about the 45° limit is that it limits the attitude/tilt commanded through the relevant position-control path; it does not physically guarantee that the vehicle’s measured attitude can never exceed 45°.

So the next thing I would check is the exact moment the flip begins and compare:

  1. vehicle_attitude_setpoint vs vehicle_attitude
  2. angular-rate setpoints vs measured rates
  3. actuator/motor outputs immediately before the divergence
  4. which Offboard setpoint/control mode was active at that moment

If the commanded attitude stays within the expected limit while the actual attitude rapidly goes toward 80°, that would point us away from the tilt-limit parameter itself and toward a control-tracking / actuator / saturation problem.

If the commanded attitude itself exceeds the expected limit, then we need to trace the Offboard command path and determine whether your commands are actually going through the controller where that tilt limit applies.

So I would not change the 45° limit or the failsafe parameters yet.

Could you also confirm what your Offboard node is publishing position/velocity setpoints, attitude setpoints, or body-rate/thrust setpoints? That will narrow this down significantly.

5

I completed the ULog timeline check. One correction/point to verify first: the parameter snapshot stored in this log shows MPC_TILTMAX_AIR = 30° and MPC_MAN_TILT_MAX = 35°, not 45°. Please check whether the 45° value was changed after this flight or belongs to a different parameter/configuration.

The recorded vehicle_attitude_setpoint never exceeds 30°. During the final event, actual tilt begins diverging from the setpoint at 462.123 s, crosses 45° at 462.532 s (46.3° actual vs 4.8° setpoint), and crosses 80° at 462.862 s (82.6° actual vs 5.3° setpoint).

This is followed by actuator outputs reaching their recorded extremes: [1999, 109, 1999, 109] at 462.586 s. fd_roll asserts later at 467.388 s, and failsafe/flight termination occurs at 468.015 s. So the 80° failsafe is downstream of the attitude-control loss, not the initiating event.

The log also records Offboard velocity and attitude control mode as active, while the commanded attitude remains small. This points away from a tilt command above the configured limit. However, the ULog alone cannot identify a specific motor, ESC, propeller, or wiring cause: fd_motor, fd_imbalanced_prop, and motor_failure_mask remain zero.

6

It looks like you had zero current from the battery source when the incident began to happen… …typical case of loose connections or heavy vibrations …rerun the flight without offboard mode and in Altitude mode to verify…What is the type of fc you have used here…

7

i think i have current related problem , i have a very similar drone and it is flying very good , so i loked at battery current from start to end and saw that battery current is almost zero where as in the working drone it is never zero , anyways thanks bro it really helped me