ArduCopter 2.4 released

Randy March 17, 2012 at 6:32pm

Ok, 2.5 has just been released so please join us on the new thread. It took a little longer than expected as we hit some last minute questions over alt hold and loiter quality. We believe that's been resolved and certainly the code has taken a step forward with this new release.

Dave C March 17, 2012 at 2:15pm

More latest GIT (0d4ce7129b0f - March15th) (2.5.0) testing on my APM1, DJI flamewheel 450, stock 850 ardu gear, 10*4.5 really crappy bendy props – my only niggle with this set up. Other than that prepare for a glowing report!

First thing to mention is that I decided to push all parameters as far as I could go to see where my tuning limits lay.

I noticed something fantastic immediately. I was able to vary most of my tuning parameters greatly (more than ever before ) whilst still maintaining a flyable machine. Everything just did what I expected it to, no drift, no surprises, just different flight ‘feel’ for different settings.

In stab mode, more rate_p gave a faster responding super stable craft, more stab_P made it even more solid. I could add in rate_i and stab_i and the copter would hold better against the wind. (playing with rate_I and stab_I at high levels a bit today I could get the quad to almost loiter in stab mode in 8mph winds – at times it wasn’t getting blown away, it just kept leaning back into the wind!!!)

I could also for the first time add in rate_d and stab_d together, and achieve ‘ninja settings’ (previously only possible with stab_d at zero and loads of rate_D), but now with a handful of stab_d too to handle the bigger wobbles in transitions.

So when I dialled all settings right up it flew rock solid. But then I could go the other way and reduce the settings – remove all rate_I and stab_I and rate_D and stab_d (yes put them all to zero) and the bird still flew well, just much softer.

I didn’t really have time to do any proper tuning as I was just so excited by the increased range of settings at my disposal I was stabbing around all over the place with all sorts of crazy variations. I am planning a full morning of more scientific testing tomorrow, results to follow.

But anyway, the best news so far for me has to be LOITER mode!!!

If I want a soft stay-in-one-place loiter in low winds (like the video below), I back off on my settings a bit ( although in the vid I'm running rate_D at 0.08 which is fairly ninja ;)), if I m in gusty wind like at the field and want ‘skylab’ style loiter, I dial in more rate_D, more rate_P and the whole process becomes more ‘determined’ for want of a better description, but nowhere near as affected by overshoots and circling as before.

I should have had much better videos from the field, but discovered I’d taken a full SD card with me and no USB lead to empty it on to the laptop. So I just had to have another little fly when I got home in the ‘garden’ to ensure I had something for your viewing pleasure ;)

It’s a shame though cause at the field, for the first time, I grabbed it by one leg and threw it away, high in the air with the TX on the ground and watched it sort itself out and come home to Daddy! AWESOME!

Anyway, attached are the params that I used for this video, using baro only, they aren’t tuned and aren’t the best I’d had that day, but well, just watch....

Testing Pre-release Arducopter 2.5 on APM 1 Quad. from Kamakaze UK on Vimeo.

Can I have a prize for best loiter yet please? ;)

So, for me, it looks like it’s tuning guide and optical flow time :D

Oh and a bit of info for those using sonar and/or baro. I’ve found that the throttle and altitude params have to be very different depending on whether or not I’m using sonar. Using baro-only the throttle rate_p and altitude hold _P need to be about half what they are compared to when I’m using sonar.

I’ll post my comprehensive settings and explanations after some more vigourous testing.

So, in summary, I’m really very happy indeed, it seems my quad is a mild tune away from exactly what I’m looking for:

A quad with more functionality and precision that any other out there, for a few hundred quid.

Time to order some more APM’s I think :D

Wookong?

Meh ;)

2.5 garden loiter params, baro only.param

Gustav Kuhn March 17, 2012 at 12:41pm

Hi, is there a problem with 2.4.1 when using a quad in x-mode?

My + quad is very stable, part of my setup test is to advance the throttle to hover speed, and then holding it by one leg, the quad remains level.

When doing the same on a x-mode quad, it seems to first "lean" INTO the error, before trying to correct itself.

Both using APM 1, my quad is the old standard quad, the other one a make I don't know, but the motors, props, and physical layout/spacing the same.

Same code in both.

Thanks,

Gustav

Todd Hill March 16, 2012 at 5:06pm

Just came back from flying Git version 2.5. Quad did well, Y6 was worse than v2.4.1:(

Calibrated mag, ESCs', usual setup. Used the same PIDs' as previous version with no luck. Tweaked them some, and still no better. Yawing bad left and right. Try to give yaw input and it would tilt sharply in direction of input. Adjusted top bottom ratio, but no good. Sometimes it would not stop yawing after giving it yaw input. It was as if it was drunk off of its stool.

Version 2.5 was much better than v2.4.1 on my quad. Y6 actually flew better on 2.4.1. Will reload and run the setup again. Attaching a log if anyone wants to take a gander. Y6, 650kV motors, APM1 2560.

2012-03-16 06-08 76.log

Y6_Param_v2.5.param

https://storage.ning.com/topology/rest/1.0/file/get/3692374454?profile=original

glenn nafin March 16, 2012 at 3:29pm

how do you pick what to use for alt hold? sonar or baro??

Jonathan M March 16, 2012 at 2:34pm

Just came back from the field after testing with the latest (as of yesterday evening) GIT v2.5. My setup: 3DR frame, DT750 motors, 11x4.7 SF props, APM 2.0 mounted on o-ring suspension, GW: 1300-1400 grams depending on battery.

Wow! By far the most stable and least drift I have ever flown. Very impressive dev team! I tested stable mode, alt hold, loiter, and RTL.

Stable: As stated, little to none of the drift I had become accustomed to since I started on my project. I haven't tuned much yet, but it's very stable with the param file attached. Aggressive flying doesn't seem to get it out of sorts. There is still some minor drifting every now and again, but it's much improved. This platform seems stable enough to finally do some aerial photography/film!

Alt hold: Best yet, and i only use the baro. It does seem to sink a bit when aggressively flying in this mode ( 3 to 5 ft), but I may have turned the Throttle P down too much.

Loiter: Not so good ,but I haven't tried tuning yet. I get oscillations and over shoots.

RTL: didn't work well, and seemed to mimic loiter in the over shoots.

One issue that I did notice was this strange uncommanded yaw of about 10-20 degrees when flying away from me. Reading up on this thread makes me think that this is a known issue. Since, I have the latest GIT, I thought this was resolved. Maybe not?

All in all, great effort and results! I am going to tune it up a bit and test again. Of course this weekend has to be the biggest storm of the season in San Diego :)

v2.5_good1.param

Developer

Jean-Louis Naudin March 16, 2012 at 2:15pm

Here a video of a test flight of the ArduCopter v2.5 firmware (offical GIT version Revision:1ed94f0064f5) done with a F450 DJI Flame Wheel quadcopter and the ArduPilot Mega v2 board.

There was a medium wind of 21 Km/h blowing and gusting from the left. You will see that the Loiter mode works well and that the quadcopter is able to fight itself against the strong gusts. In some others tests flights, I have also tested the RTL, super simple mode and off course the stabilize mode without any problem.

Regards, Jean-Louis

Vincent Mees March 15, 2012 at 12:06pm

Just flew the latest GIT and can confirm it flies smooth.

First flight I noticed that yaw was going from left to right about 15°. I then recalibrated the compass through the MP and it worked great. I'm using the sparkfun magn.

ALthold reacts slow. Wlst te other mention quick reactions?

Hopefully tomorrow some more tests.

Using X650 value frame with apm1, MT2812 710Kv motors with 11x4.7 APC SF and 3000 up to 5000 mAh batts.

meroicon March 15, 2012 at 10:28am

Hi,

I want to use the latest GIT v2.5.0 using arduino 0100 relax with APM2 . What else should i define in APM_config.h in addition to FRAME_CONFIG QUAD_FRAME

Dave C March 15, 2012 at 9:59am

Hey all,

At last RL has eased its grip on me, I had an hour to play today. So i got the latest GIT (0d4ce7129b0f - March15th) (2.5.0) on my APM1, dji, flamewheel, stock 850 ardu gear and did 3 batteries in the garden.

Muahahahahaha!

Didn't have time to tune anything, so all stock params for now, and tested stab, alt-hold with sonar and loiter (yes, in my 4*4 garden!!!)

Quick examination of logs show no evidence of the leans what-so-ever, pitch-in and pitch, roll-in and roll, were very tight indeed, despite a new VERY soft experimental APM mount)

Althold just perfect (throttle slightly too aggressive but easily tuned out on my next outing, may well be fine in higher winds)

Loiter - hilarious!!! First flick into loiter the quad went for a visit next door (about 5 m away) with an unexpected gust, (wind 4mph - gust 8mph) then tipped over gave a nice blip on the throttle to clear the fence (thanks sonar!) and came back to me and stayed there! Had a small circle initially but calmed down and stayed well inside 2m*2m box. YES!

Can't wait for the 2.5 thread to start, but so far I've have no concerns whatsoever, at all. My quad seems to loiter OK with fairly high rate_p (stock) . Can't wait to get to the field over the weekend for more testing and tuning. I've got all my fpv gear working sweetly so I'm gonna let APM take me for a ride :D

Fantastic developments again chaps, great work.

And since it looks like this release has dealt with most of the big issues I'll dust the cobwebs off the half-complete tuning guide and get something posted for version 2.5 - for those who were wondering ;)

Discussions

ArduCopter 2.4 released

Replies

Activity