Drone Heading Alignment

My comment was only confirming the cameras are the same but that mechanical shutter is selectable. That’s it.

1 Like

I see no setting for this nor anything in the manual?

I can’t look now but I think it’s under the gear icon in camera settings?

Yeah, that looks nothing like my screen. Is that DJI Go?

It’s a screen grab from a 3 year old video on youtube. I would guess it’s GO on an android phone. But I really don’t know.

Okay, I’ll look more into it. I did figure out one thing though that may give us a clue. I tried to calibrate the RC this morning and it would not start telling me that my sticks weren’t at the center point. Just another thing to Google.

1 Like

Just checked my P4P and the manual shutter setting is still there in the same place in GO.

I feel silly. I was looking in the regular menu.

Mechanical shutter was on. In continued searching I found an interesting thread on a very similar topic where the aircraft seemed to be slightly strafing left and the camera was not exactly centered. The only semi-solution was to do a gimbal calibration with the drone pointing vertically resting on the back arms and motors, not the arms and legs. I did notice I had a slightly left-facing camera and it is drifting slightly left in manual flight. This calibration method appears to have straightened the camera, but I just flew another manual mission and it still drifts slightly left. It’s so subtle that I cannot 100% tell if it is strafing or rotating.

The reason I questioned DD is because it is happening with two drones, but after reading some articles I see several reports of P4P’s not flying straight. Possibly it is a result of the drone drifting, but being controlled to go in a straight line to a waypoint which causes it to yaw?

I’ve only noticed drift while the drone is going to starting points or coming home (aside from the occasional grid drift in windy conditions). Sometimes the drift is quite bad, but it always makes it to the proper start position.

The drift problem seems to be more of a problem with the P4P v2.0 than my older P4P. Could it be the “low noise” props?

With my two drones (P4P and Mavic 2) I often see up to +/-20 degrees heading error. I think DJI does a good job of flying the desired transact lines, but it’s heading estimation can be pretty rough at times. I’ve had other flights that are better overall, occasional (rarely) I’ve seen > 25 degree heading errors. Most mapping/stitching tools handle this just fine. The reality is that magnetometers are notoriously unreliable sensors. The DJI calibration method is a tradeoff between user ease and accuracy (so they make it quicker/easier for us to calibrate the mags at the cost of not as great of a calibration.)

FWIW, our research lab (U of MN AEM dept) has an alternative EKF formulation that uses intertials (change in speed/direction) to accurately estimate true heading. However, we have always used it for fixed wing flight where we do get plenty of changes. This system doesn’t work well if you are mostly hovering, or mostly flying in a perfectly straight line. We have developed a hybrid that uses mags a little at the start, but trusts the inertials more as the flight proceeds, but this can take 5’s of minutes to squeeze that initial heading error out.

I have some visual techniques to improve the heading estimate, but I can only do those in post process. Stitching the images with DD or Pix4d, agisoft, etc. is a sophisticated (and highly accurate) visual method to correct the yaw error … but it’s obviously a tedious and slow process done in post, not during flight.

My take is that with commodity hardware and sensors, this is an issue without a good solution. You need to get up into the range of tactical grade IMU’s to really get good accuracy. The gap is narrowing though, and every few weeks some new amazing sounding attitude determination system or strategy is announced, so …

1 Like

Thanks Curtis. Mine is being really strange and the worst photos I have analyzed using CAD alignement have been 14-15 degrees pretty consistently. The odd thing is that it seems to be pretty straight coming out of corners, transitions to 15 degrees left and then almost returns to straight on long straight legs. Shorter legs do the same thing, but to a lesser degree.

For me, the results are all over the map (so to speak) from one flight to the next. It seems to be relatively stable at whatever the error is along the straight stretches, but stops and turns at the end of each row can cause sharp changes in the heading error. I always wrote it off as it is what it is, but if there is some way to do a better calibration or make that error smaller, I’d love to know more about it. DJI’s internals are mostly a black box, so I can make guesses based on my small slice of perspective, but I don’t really know what’s going on under the hood.

1 Like

Dude all of mine due it, mostly in NADIR flight. I have a theory why but I’m not going to go that in depth.

I’m seeing the same issue with the yaw not being maintained consistently to provide a consistent heading that pointing straight along the path of the flight lines. It seems to gradually shift as it goes along the flight path from the correct straight heading to about 20 degrees skewed and/or vice versa. As noted it doesn’t (and shouldn’t) effect the stitching (unless maybe you are running low overlap about borderline high enough to stitch) but in many cases my clients like to look at some of the individual images so sort of makes me look like I’m not a very good pilot. LOL.

This was also of course after compass calibration so appears to either being done intentionally by DD or its a bug.

Any final conclusions, solutions, confirmation from DD?

Update. I ran some other missions with UGCS and I’m seeing the same issue it looks like. So it appears to possibly be an issue with the drone and not DD. That would make sense as DD shouldn’t be gradually changing the yaw during a flight line. I’m seeing it happen at multiple altitudes as I thought maybe it was because I was flying low (60 feet) and possibly metal object were interfering with the compass. Seems a bit of a mystery. Probably need to take this to the Phantom Forum now.

1 Like

Flew a brand new P4P V2.0 today and it appears to be flying at a skew as well. There was a 10pmh wind so I will monitor, but when I paused the mission and took manual flight it flew pretty straight. I will continue to monitor and update. @Jamespipe

Wasn’t there a big deal about the P4P V2 not flying straight when it first came out, firmware issues or some such thing. In fact when I got mine I ended up using my older P4P more than the V2. It has over time seemed to get better and now they both fly similarly but do have the occasional hiccups.

1 Like

That sure may be. I do have it completely updated, but it has been happening with 2 different V2’s and 2 V1’s for at least 4 or 5 months. It doesn’t do it in manual flight or Litchi.