Quadcopter flies in desired path but changes heading abruptly during AUTO mission

Hello community,

We are facing an intermittent navigation/heading issue with an ArduPilot-based quadcopter during autonomous missions and would appreciate your guidance. We have been using the same quadcopter for the past 3 years and have taken almost 400 flights without any issues.

It’s a T-Drones M1000 drone with two MicaSense MX cameras as payload

Problem

During an AUTO mission, the quadcopter sometimes behaves as follows:

  • The vehicle flies along the desired mission flight path, but its heading/yaw is oriented in a different direction.
  • The yaw offset remains almost constant throughout a particular flight leg. In other words, the vehicle may be travelling north along the planned path while the nose is pointing, for example, 20–40° away from the direction of travel.
  • The vehicle does not necessarily exhibit this behaviour on every leg or every flight. It’s kind of random. Sometimes the same/similar mission executes normally, while on other flights this issue occurs.

Logs are attached: https://drive.google.com/file/d/1_4SOTLZ8W-0t8f3YjSH60NEuBhz-2kN4/view?usp=drivesdk

Troubleshooting already performed

  1. We performed magnetic/compass calibration, but the problem persisted.
  2. We updated the ArduPilot firmware and also reverted to the previously working firmware version, but the issue was still present.
  3. The mission itself appears to be correctly generated, and the same mission has worked properly previously.

Could you please help us find the problem and resolve this weird issue? What is the recommended systematic approach to troubleshooting this kind of intermittent behaviour?

Have you used MagFit Ardupilot web tool? I would recommend using it to configure current compensation.

But how can that suddenly become a problem after so many flights. It was flying perfect for past 3 years and for more than 400 flights and never showed such a problem!!

There must be some other problem !

If interference got stronger I would suspect lose power wire somewhere.

Updating from 4.4.4 to latest 4.6 or 4.7 might help, there was some bug fixes regarding waypoint yaw handling. WP_YAW_BEHAVIOR looks correct, but you can try 3 instead of 2.
Recalibrate your remote, it might override autopilot yaw, you have RC4_TRIM set to 1529 with 20 as DZ, while getting ~1509 from RC, that can be a reason too.
You also have the filtering misconfigured, it can fly better.

RC transmitter trim would clearly cause the issues you are seeing

Per Amilcar’s suggestion, RC issue