Correcting Yaw w/ Litchi ?

[RoboticTundra]

Hopefully I'm asking the right question...

I have a simple route flying in a straight line due north from A to B over a garden with some paths and fencing that are laid out N<>S and W<>E (checked w/ compass). Heading in mission hub indicates expected 0°. I've 3 versions of this mission; 1) Both waypoints set to POI-1 which is on the path, 2) interpolate (-50° gimbal), and 3) interpolate (-50° gimbal) w/ POI set to 'none'. Overall results are as expected with the first video staying close to POI-1 and the other two identical with a fixed down gimbal.

In all 3 the Mavic, though flying and following the 0° path correctly, is aimed or pointed about -10° so while flying to 0° it is pointed at about 350° and so images are cocked a bit.



Airdata indicates that it is flying correctly:



Screen Grabs:




You can see how the fence lines are not straight up and down as expected. I've had this same issue at other locations. I've done both Compass and IMU calibrations. Should I do them again even though it seems to be flying the route quite accurately? Is there another adjustment that I should look at?

Also:

- When I take off it doesn't go straight up and one time today it rotated about -90° as it rose to 10m.

- I cannot get precision RTH to work. On take-off I go up slowly to about 11m and hover for a while. If I press RTH on Litchi it comes back close and then starts circling rather fast. If I press RTH on the mavic RC it comes back and lands but about 20' east of the landing pad (orange circle in grass).

Thanks all,

 

[Aerial-Pixel]

Does it do the same thing when if you remove the POI alltogether? It then should just be aligned to the path.

 

[RoboticTundra]

Does it do the same thing when if you remove the POI alltogether? It then should just be aligned to the path.

Yes. I flew three similar routes, first with POI, second changed gimbal to interpolate w/ fixed -50° and 3rd w/ POI set to 'none'. Same result in all (except for gimbal adjustment to stay on the POI in the first).

 

[RoboticTundra]

Did some more runs. I modified 01c (POI=none + Gimbal=Interpolate @ -50°) to have a heading of +10° and saved as 01d. It ran the path perfectly 0° north and seemed to have a heading of about +10°. Then ran 01a (POI=1 + Gimbal=POI) and it seemed to now have a heading of 0°. Then ran 01c (POI=none + Gimbal=Interpolate @ -50°) which now exhibited a heading of about -10° and then ran 01a again and this time it seemed to have a heading of about -10°. Landed, shut down, restarted and ran 01a again and this time it had a heading of about +5° (but still seemed perfectly centered on the path.

Overall it seems to be having difficulty keeping a consistent heading. Is the heading compass based and the waypoints and path GPS based? So compass could be off and need to be calibrated again?

BTW, it twice did a proper RTH and landed back on the orange landing pad. No idea why it's suddenly working now when it didn't earlier. Only difference is that the gimbal was @ -90° (straight down) on take off and landing the first time, but I thought it wasn't supposed to use the main camera for precision RTH at all so this shouldn't make a difference?

Thanks,

 

[RoboticTundra]

If a compass calibration issue then should it consistently be off by the same amount (E.G., heading always -12° for instance) or could calibration cause it to be inconsistent. I'd think the former?

Should I be able to expect it to maintain a heading within x°x'? Max gusts are 4.5 m/s but mostly around 1.5 m/s.

Thanks,

 

[RoboticTundra]

Did approx the same run w Go4 and got somewhat similar results. First pass heading was off about +10° and subsequent 3 passes off by about -10° One note, on the first pass it turned CCW to begin the run and on the subsequent runs it turned CW so seemingly it's not doing a full 180°. I'll maybe test the affect of turn direction when I get another break.

Run 1:


Run 2:

Run 3: