This is a discussion re the ArduCopter-3.1 AutoTune feature. A place for people to report successes and failures, ask questions and get answers.
The AutoTune wiki page is here.
You need to be a member of diydrones to add comments!
This reply was deleted.
Replies
I also just realised that your GPS also has the compass in it so that will probably be messing with a few things too. So when you flip the GPS make sure it is correctly orientated for the compass and recalibrate the compass.
I'm not an expert but your logs show a gps problem Hdoop too high, but Altitude hold not use gps, your vibs looks near limit too, I'm working on triyng solve that problem here http://diydrones.com/forum/topics/vibration-isolation-and-dampening... sometimes vibs gives a poor althold performance but I'm not shure if it is your problem.
Hello everyone. This is my first post. I've built an hexacopter with a 960mm frame, DYS 40A SimonK ESCs, Foxtech KV288 motors and Pixhawk for FC. I would like to try autotune.
The wiki pages are amazing, so far I've managed to do everything without having to ask a single question here. But while I was reading the wiki for the autotune feature I read that warning about low kv motors and simonk escs... Exactly as what I have.
My question is, is it safe to try it? Is there anything I can do to minimize any risks?
Unfortunately in the wiki there are no answers for that, and I couldn't find much after googling all afternoon today.
Maybe you can help? I would really appreciate it.
Thanks in advance!
Cheers
Lucas
Hi Leonardo Bassin,
It looks like you have to recalibrate your radio. You are constantly asking for a 2 degree pitch. You have not changed your dead bands on the apm so it isn't the jitter you were concerned about.
So recalibrate your radio in mission planner and you should be fine.
Sorry for the slow reply!!
Hi Developers and my fellow users,
I would like to share with you my great joy and amazement with the success of an autotune I performed on my SteadiDrone QU4D X. The platform uses 100kv motors and had an all up weight of over 9kg. I was very hesitant to do it because of the cost of the propellers and motors as well as the warnings on the AutoTune page about low kv motors but after having talked with Brandon Basso from 3DR, decided to give it a go. This platform is now incredibly stable and responsive. After the Autotune, I flew two ~40min automatic mapping missions in 5-6m/s winds which went flawlessly. If the log file from the autotune and subsequent flights would be helpful to either developers or other users, please do not hesitate to let me know and I will make them available.Here is the video of the AutoTune:
As I mentioned in another post, I used a slightly modified version of the 3.2 branch to perform autotune via ch5 on my spider style quad build (1.7Kg AUW, 12x4.5 inch props) with forward camera balanced by the battery in the back. The resulting parameters are asymmetric as expected and it seems to fly well with nice snappy roll and especially crisp pitch changes.
However, descending at 1-1.5m/sec often causes uncontrolled oscillation on the roll axis. When the oscillation gets to about 45 degrees I have to stop descending to avoid a possible crash. It smooths right out when I stop or climb a little. I have not yet tried performing autotune a second time to see if the result is the same.
Also, the yaw seems a little less stable than I expected. I'm not even sure if this is touched by autotune. Any suggestions here?
I tried increasing my Rate Roll P setting from .055 to .09 and it seems to cure the bad descent behavior at the cost of less crisp roll response in constant altitude flight.
Was this the correct adjustment to make? Is there a way to have better roll response and still have relatively stable descent? Is there any way autotune could detect and avoid this behavior?
Thanks!
28.BIN
I will look at your logs a little later today to confirm.
Leonard,
Thanks for the response.
The mass distribution is pretty spread out along the roll axis and the front two motors are slightly farther apart than the back two on this style of frame. But I believe the CG is in the correct location for this frame (within half a cm or less based on battery placement anyway). The only really odd thing is that because it worked out best for cable length reasons, the controller and compass are mounted backwards (arrows point to the rear). But that is dealt with by setting AHRS_ORIENTATION and COMPASS_ORIENTATION both to 4. Oh, and there is a home built servo based gimbal that was active. Would it help to just turn that off? I suppose it might be adding some small torque impulse in delayed form and the part that moves is not exactly on the axis of rotation. But the camera is very light compared to the rest of the drone so I wouldn't think it would have a large effect.
Thanks!
I tried again this morning after disabling gimbal stabilization so the servos won't move. The result was similar (worse, actually).
One last thought. I am not starting autotune via sw7/8 but I think I have arranged via a simple code change to make direct swiching to autotune mode include the proper initialization. Here is the pull request I just created for this change if you want to take a look. Did I miss something?
https://github.com/diydrones/ardupilot/pull/1520
If I have time this weekend I may try without the orientation=4 settings just in case that is confusing autotune.
Any clues in the log?
Thanks!
-
24
-
25
-
26
-
27
-
28
of 55 Next