Strange auto yaw behavior rc5/6

When following straight line waypoints the vehicle heads toward the next waypoint for a couple of meters before changing yaw angle toward next waypoint. Low angle turns are not as pronounced but 90 degree corners it heads toward next waypoint 90 degrees off and then yaws into a straight line to target point. I have tried several different mission layouts with both my quad and hex (pix 2.1) with the same results. Spline missions are executed perfectly. I attached a short log from the hex.

Cheers RB

00000039.BIN (1.4 MB)

RandyB,

This is almost certainly because of the changes we made to try and resolve an older issue in which the vehicle turned too soon towards the next waypoint. Is there a typo in your comment above (“next” vs “current” maybe?). In any case, I think what you maybe don’t like is how the vehicle may start flying towards the next waypoint but there’s a small delay before the vehicle begins to turn towards the next waypoint? It’s kind of the opposite of the older problem.

I’ve created a video to explain the change. I’ve left it unlisted for now although I may make it public later if it’s useful. For now, could you confirm that it’s the delay in the turn towards the next waypoint that you don’t like? It’s possible we can resolve this. It was a conscious choice by Leonard (our attitude and controls expert) but we could make changes if people don’t like the behaviour.

Thanks Randy for the explanation and quick reply. I read the rc5 release notes and expected that the yaw behavior was due to the change. I hesitated to even say anything about a purely visual condition but after several flights and experimentation I thought it was worth at least a comment. I have never noticed the previous pre-yaw at waypoint condition and the post-yaw in rc5 is very noticeable, two to three meters sideways before the yaw correction. Certainly a non-safety or stability issue and you guys have done a remarkable job in that regard. 3.5 is rock solid and predictable, a huge improvement over the last year or so. Thanks for all the work into a truly great product.

Cheers RB

Hi rbachtell,

Would it be possible for you to send us a video of what you are seeing?

Thanks for giving us the feedback!!! Don’t ever hesitate, it is the feedback that drives this project forward!! :slight_smile:

Leonard, I put together a short auto course with multiple right angle turns and altitude changes. I think I have a better idea now as to when and why the yaw on turns changes. If the vehicle approaches the waypoint at altitude with lateral momentum, it hits the waypoint and brakes coasting past the waypoint. It then comes to a stop and yaws into the line toward the next waypoint as it moves in that direction giving the appearance of a coordinated turn. If however it approaches the waypoint with a significant altitude change, it doesn’t have momentum or vertical angle to coast past the waypoint and stops very close to the waypoint. It then heads sideways in a line toward the next waypoint and yaws into the line after a second or two. It looks like a time delay from waypoint target to yaw command. Let me know if you need more information, gives me an excuse to fly.

Cheers, RB

I have similiar situation in new 3.5-rc6 firmware.
I uploaded the problem on the github with logs.
(https://github.com/ArduPilot/ardupilot/issues/6259).
but couldn’t find someone experienced same problem until now.

I have real frame having issue with heading controller and simulations also.
which is heading controller keep rotates vehicle.
regardless of flight modes, (I have tested auto, poshold, and althold mode)
looks like heading rate command diverges for some time.

I suspect new function here, but I couldn’t understand this function yet.

https://github.com/ArduPilot/ardupilot/blob/master/libraries/AC_AttitudeControl/AC_AttitudeControl.cpp#L429

Problem video with SITL + Gazebo is here.

https://drive.google.com/open?id=0BytnOIWJ3_zJSHQ1UXlxUmpzYVE
https://drive.google.com/open?id=0BytnOIWJ3_zJSHQ1UXlxUmpzYVE
Thank you.

Randy and I have discussed a fix to this and will let you know when we have something to look at.

Thanks for your video’s and help!!!

Thank you for all the work on APM. I have run through everything I have hardware for with rc7 and it looks very solid. I have two vehicles that I use for testing that have been in service for two years and I have been able to track the progress of the firmware with only minor changes to hardware. Very stable and predictable firmware at this point. Thanks again

Cheers RB

I am having a similar issue with my yaw behavior. I am trying to map a couple housing lots in auto mode by flying a grid of parallel lines. My copter used to fly this perfectly, with yaw always pointing along the direction of flight (along GPS course). Now the yaw takes over half the length of the flight path (one of the parallel lines) to adjust towards the next waypoint. I thought this may be the new change Randy mentions effecting me because my auto mission does not cover a lot of area, but I tried increasing the length of the flight path (each parallel line) to accommodate and it still takes over half the distance to fully adjust the yaw.