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.

 

Views: 74319

Reply to This

Replies to This Discussion

Cmon,its not that hard...but i agree we should always have printed check-up list befeor take off.

 

Leonard,

I assumed that it was related to the Arris CM200 GoPro gimbal that is mounted using rubber vibration dampers.  The copter is a 3DR Hexa that I converted to a Y6 so the arms & frame are pretty tight.  I am certain the CG was OK.

Two questions:

- What log sets should I be certain to enable for proper analysis?

- If it is the gimbal, can I remove it then auto-tune and then replace...

Karl

Emin,

3DR Hexa converted to a Y6 with 850KV motors and a 5000mAh 3S.  An Arris CM2000 2-axis gimbal mounted underneath (using vibration isolation rubber mounts) with a Hero 3 on it.

Ideas?

Karl 

Hi Euan,

First up Autotune is not something you have to do it is not a "requirement". It is something you can do to get the most out of your frame. However, the default parameters (or close to them) should be fine for most people.

For those people that want to get the most out of their frame, AutoTune makes it easy. Approximately 95% of people are successful when using AutoTune without any changes to their frame. However, if you do wish to use AutoTune then you do need to achieve a basic level of construction and rigidity of your setup.

This isn't possible when carrying a large camera gimbal on vibration mounts. In this case the camera should be removed and the copter tuned on it's own. This should result in a safe tune for the copter as additional weight is added even if it is able to move around.

The remainder of the reasons Autotune won't work all fit under the "you shouldn't be flying that" category or at least the "fly at your own risk" category. If your arms are too soft for the power of your motors (I have been guilty of that one), your CG is way out, or the vibration dampening on your APM is so soft it wobbles around in flight. Then sorry, you are on your own. The Autotune feature will detect that your frame isn't up to scratch and will tell you to go fix your problems or do the tuning yourself :)

An from a personal note, as a developer I am not trying to attract people that have never flown models before (others might be, but I am talking personally). I am not trying to attract people that want to fly a 5kg Parrot AR Drone because they got the little one and wanted to go outside without getting blown away by the slightest breeze.

I want to attract experienced pilots from DJI or OpenPilot because Arducopter control is THAT good (we are there or close and I still have a few polishing ideas left), because the control system design is based on a complete physics model of the multirotor, and Autotune generates the control gains that bring them to the edge of what the copter is capable of in a single 6 to 8 minute flight. I want to make sure that the copter is easy to set up (and that setup makes sense) for an experienced pilot. This in turn makes it relatively easy for a beginner to get up to speed and have a top performing machine.

For the pilot that doesn't want to keep the CG at least close to where it is supposed to be, or make sure that all the screws on the frame are tight, or straps a 2 kW system to a $20 flimsy plastic frame, then expects the Autopilot to make up the difference. Sorry, you will be disappointed. 

Hi Karl,

Yeh, I have seen this before. Based on experience so far, and some analysis, the copter is most responsive when it is unloaded so doing an auto tune without the gimbal is the way to go.

As a camera ship you may find that the Stab_P is too high for your tastes. You can back it off to make corrections softer. But you should leave the rate values alone. 

Having said this, be careful as it is always possible there will be some combination of parameters that will make this different to most. Generally though, Autotune finds the tune that give the minimum response time in rate PID and Stab P. This also results in safer tunes but also means the copter can handle Stab_P values 2 to 3 times higher than we have been using.

So, remove the camera and gimbal along with any wobbly mass hanging from your copter. Secure your battery solidly to the frame. Do your autotune. Then reattach your gimbal.

And it would be great if you could post the autotune log here. Autotune automatically logs all relevant data, you don't need to turn anything on.

Leonard,

When time and weather permit, I will try as you suggest - thanks.

I don't recall anything regarding gimbals in the wiki, perhaps your suggestions should be included with the other notes on auto-tune.

Thanks again,

Karl

Hey Leonard,

You you just reminded me of something with the answer to the logs having any additional metrics enabled.. have you published a post/blog/wiki article or anything that we can read to try and go through the autotune bits of a df log ourselves?  I tried looking the other day to determine why AT keeps stalling and couldn't find anything.  This is on the same frame we've discussed in the past even offline with Craig a bit as well (solid arms and now solid foam mount), and AT has succeeded in the past.

Also to verify, having MP connected via telemetry is NOT a requirement for AT to complete, right?

-josh

Here is a pic of what I can find about ATUN, you can see three attempts.  Is the D term too high for it to get started with?

Leonard,

OK, so there was a bit of a breeze but I gave it a go and you will see multiple corrections to avoid trees and such. Log attached.

Much better results this time.

Thanks,

Karl

Attachments:

Euan, it's important to understand that the autotune feature is an optional extra feature, not something that is mandatory to do.  So if you can't get an autotune to complete, it's not going to prevent you from using the system.  Again, this is an optional extra, a feature that nobody else has been able to get working (as far as I've seen).

It's entirely possible to take a quad that can't be autotuned, and still use it without issue.  I had a Tarot F450 frame that was flexible, and would not autotune.  But, I tuned it myself, and flew it tons.  Does it perform as good as a stronger frame?  No.  But it doesn't mean I can't fly it.  I have flown it in 60+ km/h winds.  I've flown it in acro with Acro_RP_P set at 10, doing backflips at 60 km/h.

There really is nothing wrong with the frame.  All the screws are tight, and the damping isn't too soft, and the CG is spot on.  It may be overpowered though I'm not sure that should cause an auto-tune failure.  As far as we can tell, the arms are just too soft.  But it still flies well with a manual tune.  And the upshot is it bounced well, which is a great thing for some uses.

The purpose of auto-tune is to make it easier for your average pilots to achieve a decent PID tune. But it requires certain physical requirements to be met.  If you can't use it, it's not the end of the world.

As for the vibration complaints, I have no sympathy for that.  It is now SO EASY to vibration damp your APM, so many people have shared so many good ideas, you have absolutely no excuse for having vibration still.  My Tarot 450 is a real paint shaker, and I have no vibration problems at all.  It has very flexible arms, crappy Turnigy motors, and I don't even bother to balance the props.  Yet it's no problem to meet the minimum vibration targets.

The vibration sensitivity was a necessary fallout from the excellent inertial control that Randy and Leonard developed.  Arducopter would not be what it is today without that.  So complaining about that isn't going to get you very far.

Pixhawk will absolutely have the same requirements.  Because the problem is not with APM.  The problem is we must obey some basic physical laws.

I am an un-experienced multicopter pilot. Just started and got my information from RCGroups and this forum. I decided to build my own quad from scratch. I have flown fixed wing models for many years (one with APM using APM:Plane), this is my first quad.

After assembly and testing, decided to do autotune. Not a single problem, it flies like a dream. Thanks for this feature.

Hi Karl,

Yeh, that is what I like to see!

As I said, you may want to back the Stab_P off if you find it too sharp.

Let us know how you go once you put the camera back on.

Thanks!

Reply to Discussion

RSS

Groups

Season Two of the Trust Time Trial (T3) Contest 
A list of all T3 contests is here. The current round, the Vertical Horizontal one, is here

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service