The first I took my quad out to test 2.5.3 loiter was a bit sketchy. The quad screamed from one edge of the field to the other, overshooting by a greater amount each time. (See below, it started somewhere in the middle. The pattern is about 35m across).

I dropped the Loiter P parameter to 0.2 and took it out today. It performed a bit better. The overshoots were more controlled, but it still covered and area of about 25m across.

Logs showed GPS was picking up about 10 satellites each time. Should I keep reducing P, or is something else at play here?

Thanks,

Chris

Views: 3443

Reply to This

Replies to This Discussion

Hi Chris, 

These are the parameters that influence loiter in 2.5.3:

 

HLD_LAT_I
HLD_LAT_I
HLD_LAT_P
HLD_LON_I
HLD_LON_IMAX
HLD_LON_P
LOITER_LAT_D
LOITER_LAT_I
LOITER_LAT_IMAX
LOITER_LAT_P
LOITER_LON_D
LOITER_LON_I
LOITER_LON_IMAX
LOITER_LON_P

My understanding is, that there are two separate processes involved in LOITER. There is the HLD process (hold) and the LOITER process, each with their respective set of parameters.

The purpose of the HLD process seems to be the holding of a successful loitering point and the LOITER process actually attempts to return the copter to the correct position when it has drifted away.  At least that is how I understand the workings of the current version.

TO GET A SUCCESSFUL LOITER, YOU MUST TUNE BOTH SETS OF PARAMETERS. 

 

In short, start with P parameters around 0.1 and I set at 0. Start raising the all the P values involved in loiter by the same increments until you get an oscillation. Then Dial back until the oscillation disappears. At this point you will have a workable loiter with no wind resistance. To get wind resistance start dialling I terms up in small increments (of 0.005) until you get oscillation, then dial back. Don't ask about the D terms, as I haven't got that far.

Nice helpful information!
Thanks guys, I'll give that a try once the wind lets up here in the northern Chicago area. 
Keith

Start with the break of dawn (usually no wind), by the time the wind picks up as the day goes on, you should be done with the P terms and you can move onto I...

Thanks Hans. I'm curious if the default loiter values work for anyone at all? All the other default control values seem to work fine for my quad without any extra tuning. It's a home made frame but similar weight/dimensions to the Arducopter.

Regarding the values, this is what I see under the configuration tab for loiter:

Which of the parameters that you mentioned do these relate to?

Looking at these I think I must have inadvertently changed "Loiter Speed" instead of "P". Odd, I don't remember loiter speed being there before. Anyway, I'll try reducing P further for tomorrow's flight.

Hi Chris, I strongly suggest you forget the ''loiter tuning box on the right''. The values it contains change from one apmplanner version to the next. The proper - and consistent - way is to the change the appropriate values in the left scroll tab. That way you know what you are changing i.e. the actual name of the parameter, rather than a potentially confusing description of what it does - ''loiter speed''. If you want to know what the values in the box actually are, just change them and you will see changes in the corresponding parameter in the left scroll box.
And no, the default loiter parameters don't work for anyone that I am aware of. For starters the P values are 5 to 10 times as high as they should be.

Thanks, that's good to know. I just wanted to get a feel of what some reasonable parameters might be. At the moment, I don't have the capability to change them when I'm in the field. I can only fly maybe one weekend fortnight (lucky this weekend, I can go twice!), so tuning the way you outlined will likely take me many months.

Try P parameters of 0.2 all around and an I of 0.010 to 0.015. That should give you a passable loiter in no wind or very light wind conditions. Be aware that you also need the correct magnetic declination setting for your area and that this does not always exactly equal the declination given on the web. If you get circling then you will have to find a compensation factor... I'll write how to do this when I get to my desktop.

Are you sure that it's "5 to 10 time as high as they should be" .  Most of the successful loiter parameters I have seen posted have Loiter_P around 2.0.   This is with the latest version of the code.

Indeed. The highest I have seen is Marcos 0.600. You can find it in the 2.5 thread.

Hmm .. I will look later, but those are the param files I am referring to.  Generally, the HLD_Ps are around 0.2 while the LOITER_Ps are 2.0.  This also what I am currently flying.  There are some oscillations but it's not terrible.

Reply to Discussion

RSS

© 2019   Created by Chris Anderson.   Powered by

Badges  |  Report an Issue  |  Terms of Service