My rover has some weird EKF3 issues. As I drove it would “EKF3: Yaw Inconsistent by # degrees” every few second. I have attached some logs.
I was sending a velocity vector while in guided mode, this vector was 180 degrees local (south) bearing. So the expected outcome was the rover driving south. However as it would move in that direction it would only go for a second or so, then the EKF would fail putting it in HOLD mode. Then a second or so later it would be OK and then it would start driving again (after we switched it back to guided).
I am wondering if the rover is actually using the built in compass. I have the compass setup screen selecting the external as primary and used. Whereas the internal compass is not selected for use. We have the compass mounted as high as the cable allows it. We have the autopilot mounted on the lower plate of the rover, the jetson xavier mounted above that, then some space and the compass above it all. I am worried that if the autopilot is erroneously using the internal compass, there may be a lot of interference with it being only a few inches from the power distribution board and an inch or so from the bottom of the xavier.
We are using wheel encoders and no GPS. We will be integrating T265 once we get through all these issues.
- can anyone say that the compass is the issue?
- If the compass is the issue, its the primary/external one right?
- Best practice for shielding for this type of issue?
My current setup is this:
Hardware: Pixhawk4 from HolyBro, holybro GPS/compass module.
Firmware: Rover - 4ac6a647d7b305861e927506beeaf5e9ca554c8b - off master a few weeks back.