ArduCopter-3.1 released!

Randy January 14, 2014 at 5:04am

ArduCopter-3.1.1-rc1 is now available through the Mission Planner's Beta firmwares link. It's really a patch release mostly for Pixhawk users who had run into some SD card reliability issues.

There are very few changes for APM2 users but includes fixes for Warning #5 (Loiter init bug) mentioned above and fixes an annoying issue when loading ROI commands from the APM.

If there aren't any significant issues we will make this the official version in about a week.

Thanks and all feedback welcome!

Changes from 3.1
1) Pixhawk improvements:
    a) Telemetry port 2 enabled (for MinimOSD)
    b) SD card reliability improvements
    c) parameters moved to FRAM
    d) faster parameter loading via USB
    e) Futaba SBUS receiver support
2) Bug fixes:
    a) Loiter initialisation fix (Loiter would act like AltHold until flight mode switch changed position)
    b) ROI commands were not returning Lat, Lon, Alt to mission planner when mission was loaded from APM
3) TradHeli only fixes:
    a) Drift now uses same (reduced) collective range as stabilize mode
    b) AutoTune disabled (for tradheli only)
    c) Landing collective (smaller than normal collective) used whenever copter is not moving

Lars Karlsson January 13, 2014 at 9:04am

Hi All,

I have run into problems and need some help to understand. I have two copters, a Hexa and an Octo. Both upgraded to the official 3.1 and I have done the Autotune without problems. I have no cameras onboard until I feel safe with the behavior. The copters became very crisp in their behavior and they have lots of power. I flew a simple Auto mode with two waypoints and automatic landing with the Hexa. Just perfect. Then I added a circle command after the first waypoint. The copter got wild just before the first waypoint followed by a crash. Low vibration level and no wind.

A couple of days later I tried my Octo. After Autotune, a perfect two waypoint flight. I inserted a circle after the first waypoint, the same thing happened just before reaching the first waypoint. No wind and good enough vibration level. This time I was more alert and switched over to loiter, alt hold and stabilized and succeeded to save the copter after some wild maneuvers.

Some thoughts about what could have caused the problem.

The PID loops are too close to self oscillation after the AutoTune. They were set much more aggressive than what I used earlier.
The APM is too slow in calculating the control loop, as I am collecting too much data.
Transition problem in the Circle command
I have seen similar behavior when hitting the fence, very aggressive

After the incidents I have reduced the roll/pitch (50%) Rate P and I. Reduced Yaw P from 4,5 to 2 and the data collection to default.

Now the problem is confirmed with several flights with and without Circle command. It seems that the Circle command takes over before slowing down and reaching waypoint #1 and pull full speed towards Circle waypoint and at the same time full yaw. See the added video. Still very aggressive although lowered gains. I will skip the Circle command and wait for a solution.

Circle command problem

Please let me know if someone else has experienced the same behavior and thanks for all work you all have done to make it possible for me to fly.

Hexa:
Frame Foxtech foldable FH H800
Prop       RCtimer carbon 12x3.8
Motor    Tmotor 3506-25
ESC         RCtimer 20A     High timing
APM 2.5                 3.1 official firmware
Battery                   4s 9 Ah Lipo
Weight                   2630 g
Radio     Spectrum DX8

Octo:
Frame Foxtech Octo FH-H800
Prop       APC 12x3.8
Motor    Tmotor 3506-25
ESC         Hobbywing Flyfun 30A     High timing
APM        2.0   3.1 official firmware
Battery                   4s 9 Ah Lipo
Weight                   2930 g
Radio     Spectrum DX8

2013-12-20 09-55.log.gpx

2013-12-20 09-55.kmz

2013-12-20 09-55 hexa crash.log

Moderator

Graham Dyer January 9, 2014 at 11:58am

Re Arducopter, from what version will it be for Pixhawk only and no longer be for APM? i.e. Is there still a version or two planned or how close is APM to the end of the road?

Ray Gamero January 9, 2014 at 8:23am

This has Pixhawk support.

I guess you can load AC3.1 on the pixhawk using MP.

Can you load AC3.1 using QGCS?

Will QGCS work with trad helis?

Ray Gamero January 8, 2014 at 8:00am

I like the new version. On my Trex 450 I can take off, go to another point, land, take off again, RTL and land automatically.

It's a little scary to watch because on the first land there was a good crosswind and the heli sat on the upwwind skid for a two second eternity until it decided to take off again. It did it over and over and I think it's just being smart.

If I put a loiter delay between the land and Take off waypoints,will it sit there nicely until the time is up (maybe even on one skid)? I'm reluctant to try.

Andrew Chapman January 7, 2014 at 10:49pm

Hi all, I finally got around to giving AC3.1 a good thrashing on the weekend. I put it through about 2 hours of flight time, here's some observations, mostly general feedback rather than support so I'll post it here rather than the forum, though I might follow up on some things there.

1) Two thumbs up for 'Spin when armed'. I was skeptical but after one or two flights it feels perfectly natural now and it definitely does the job of commanding more respect for an armed copter. For example I forgot to hit record on the GoPro before arming one time, usually I wouldn't bother disarming and just reach through the prop area and hit record, but with the slow spinning props it was clear I should shut down properly before going near it

2) And three thumbs up for Auto Tune, wow! I knew I still had some tuning to do on my quad, but never bothered to spend the time learning the arcane art to get the last 10%. After auto-tune it is a whole new beast, *super* tight and responsive and much less wobble on descent. It was a touch too aggressive in pitch so I de-tuned that just a little then pumped up the expo on my transmitter so that my control inputs were smooth again with the much tighter response. It is amazing to be able to click the pitch trim up and down a notch and see the copter flick to a slightly new angle with each click, so tight!

Two little things I noticed though:
2a) I initially had some pitch trim on the transmitter which prevented it starting the auto tune until I reduced the trim to get back into the deadband. We should add a note about this to the auto tune wiki page. I imagine the auto-tune process could be adversely affected by an out of trim copter too?

2b) The autotune procedure tweaked my gains accurate to 4 decimal places but I noticed when saving out the config from Mission Planner that it only retains 3 decimal places. I've added an issue to the MP issues list on github about this (https://github.com/diydrones/MissionPlanner/issues/295).

3) Tested DO_LOITER_TURNS and ROI as part of a complex auto mission, they work as advertised, though I was initially caught out by needing a separate ROI command between each waypoint. Although the command is documented that way, I think it makes a lot more sense for it to be persistent as mentioned by a few people here: https://github.com/diydrones/ardupilot/issues/569. Though if it were persistent you'd need a way of cancelling ROI to revert to normal look ahead yaw.

4) Tested geofence, it works as advertised, though on one fence break the copter flipped over almost backward and lost a lot of height, almost crashed. Log attached, you can see at line 18122 the pitch goes to 83 degrees and it drops. The geofence ERR message is just before that on line 18100. I tried reproducing that both with geofence as well as by manually switching from STAB to RTL in aggressive forward flight, but didn't see it happen again.

5) A couple of disarm issues/inconsistencies:

5a) after letting a geofence initiated RTL proceed through to an automated landing and automated disarming, I later couldn't re-arm, and it took me a minute to figure out that I was still in RTL mode. I think it would be better if copter always switched back to STAB as part of the automated disarm routine, to return to a known state where you can re-arm again. Anyone agree?

5b) when you land in an AUTO mission it winds down the props to 'spin when armed' speed but doesn't auto-disarm after 15 seconds. The LAND command is documented that way, but it is a different post-landing behaviour to other modes, and that is more apparent now with spin when armed. I would hope to be able to do a one-way mission where it lands land and disarm safely at the end. Is there a parameter to configure this, or some way of invoking a disarm at the end of an auto mission?

6) New Acro mode is pretty awesome, but I think the default ACRO_RP_P gain should be a lot higher. My first flips were pretty scary as I watched it slowly rolling through inverted even though the stick was jammed all the way over. After increasing ACRO_RP_P to 7 it felt much better and safer to flip. Also, I noticed that when using CH7/8 for the Acro trainer switch it doesn't seem to actually update the ACRO_TRAINER parameter, so you can't refresh in MP and verify you've got the trainer mode set up correctly before using it. It would likely end badly if someone thought they had it in acro trainer mode 2 (limited and self-levelling) when it was actually set to 0.

7) I thought Drift mode would be another simple/trainer mode I wouldn't be interested in, but it was great fun rocketing around without having to worry about keeping your turns coordinated. This is a pretty big deal for newcomers I think - in no time at all they can be ripping around the sky like a pro in this mode. Nice work Jason!

2014-01-05 12-30 - Capilano and geofence and acro.log

Distributor

Guto Santaella January 5, 2014 at 5:00pm

Hi Randy, happy 2014!!!

I just updated my PX4 to AC3.1 and after that I have no data from my spektrum satellite receiver when I try to do the radio calibration (the biding led on receiver is orange that means found TX)

Could be any incompatibility with the new firmware? Do you have noticed the PX4 satellite receiver port was tested on AC3.1 for anyone?

See you,

Guto

TheMightyThor January 4, 2014 at 7:18am

What Parameter tells you that the aircraft hit a failsafe in the Logs? Thanks

gsenroc January 3, 2014 at 3:59am

Randy, are you using s.bus with your futaba? After upgrade to 3.1 my PX4 doesn't reconize s.bus signal anymore, If I flash 3.01 back it will work again. Is there anything I need to do to make it work?

Euan Ramsay January 2, 2014 at 11:52pm

Ahah! So it does. Cool - looks like it's a straight swap in for me. Cheers Randy.

Another query - The safety switch - is it really that big? I've no idea how I'm going to mount that...

Discussions

ArduCopter-3.1 released!

Replies

Activity