After months of testing AC3.1 is finally available in the Mission Planner and GitHub.
Warning #1: If the vehicle is left landed (but armed) while in AltHold or Loiter mode for more than a few seconds, the vehicle may suddenly jump into the air the moment the pilot raises the throttle above zero. The reason is the baro altitude may fall while the vehicle is sitting on the ground. In AC3.2 this bug is fixed because we reset the target before take-off.
Warning #2: Motors will spin (slowly) by default as soon as the copter is armed! You can switch this feature off by setting the MOT_SPIN_ARMED parameter to zero.
Warning #3: if you see "Bad Gyro" on the HUD then you may have a blown 3.3V regulator. See this thread for details.
Warning #4: if you use a sonar set the SONAR_GAIN parameter to 0.8 or lower
Warning #5: Trad Heli users should now connect their main rotor ESC to APM's RC8 output
Warning #6: Trad Heli BUG causes loss of collective control in ACRO mode if swash type set to flybarred (i.e. not flybarless). Will be fixed in AC3.2.
The major improvements over AC3.0.1 are (full list here):
1. Pixhawk support (Tridge & PX4 development team)
2. Autotune of Roll and Pitch gains (Leonard/Randy)
3. Drift mode (Jason)
4. Improved Acro (Rob/Leonard) and new Sport mode (Leonard)
5. Arming, Take-off, Land in Loiter or AltHold
6. TradHeli improvements (Rob) including:
a) support for direct drive tail rotors
b) smoother ramp-up of main rotor
c) reduced collective for better control in stabilize mode (STAB_COL_MIN, STAB_COL_MAX params)
7. Support for SingleCopter (Bill King)
8. Performance improvements resolve AltHold troubles for Hexa & Octacopters
9. Safety Improvements:
a) GPS Glitch detection (Randy)
b) Motors spin when armed (Jonathan Challinger)
c) crash detector shuts off motors if copter flipped for 2secs
d) batt failsafe option to RTL instead of LAND, gps failsafe option to trigger AltHold instead of Land
e) more pre-arm checks of inertial nav speed,
10. Bug fixes:
a) optical flow working again (although performance is still not great because not integrated with inertial navigation)
b) ROI working (use DO_SET_ROI command)
How to upgrade:
Click on the MissionPlanner's Initial Setup >> Install Firmware screen. The version numbers should appear as "ArduCopter-3.1", then click the appropriate frame icon and it should upgrade as per usual.
Special thanks to Marco and the many beta testers on the AC3.0.1 thread who put their copters at risk during the testing of this new version and uncovered many problems so that you don't have to! Here are some of their videos: loiter in wind, patio sonar, autotune, autotune2, roi, beach, acro, vrbrain, tricopter, train spotting, radio failsafe, the tree, double loiter
If you haven't seen it recently, please check out the ArduCopter wiki.
Feel free to post comments below and we're going to use the new APM Forum for support issues so please post your support requests there.
Replies
AC3.1.5 has just been released as the new official version through the Mission Planner. This is exactly the same code as the final release candidate (AC3.1.5-rc2) with the exception of the firmware version number change. It's changes over AC3.1.4 are in the ReleaseNotes.txt and below:
1) Bug Fix to ignore roll and pitch inputs to loiter controller when in radio failsafe
2) Workaround to read from FRAM in 128byte chunks to resolve a few users boot issues (Pixhawk only)
3) Bug Fix to allow compassmot to work on Pixhawk
Thanks very much to those who tested this release including Artur, Dom, Kyle and flip-master Christian!
Quentin,
The accelerometers (used for inertial nav) can't provide heading. They can tell you how much the vehicle has moved but only from the vehicle's point of view. i.e. "I've moved right, I've moved left". They can't tell you which way the vehicle has moved on the earth (i.e. they can't tell you "i've moved North, I've moved South"). For that the vehicle needs an external guide which is where the compass comes in. For rotation we can use the gyros but they drift so they'll be ok for 30 seconds perhaps but certainly after a couple of minutes with no correction from the compass they will be incorrect.
If loiter is working then I'm sure your compass is still enabled somehow.
It was disabled before uploading 3.1.4. I can't check for a week as I am away from base. Unless the tlog gives that information?
2014-05-26_20-07-51.tlog
Quentin,
Yes, it's visible in the tlogs and indeed you're right that your compass is off and you're in Loiter mode but it doesn't seem to be taking off in random directions. So either the copter was armed facing directly north and it wasn't flying long enough for the gyro drift to build up too much or our COMPASS_USE flag doesn't actually disable the compass somehow even though it's set to zero!
Interesting, I am a flight instructor on the EC225 and was talking to the Airbus Helicopters engineers the other day. They said the EC225 is designed to use only true information, but they had to add magnetic pickups as EASA wouldn't certify it without magnetic heading information!
Do I assume that the gyros are a low grade quality on the board if they drift so quickly?
That only works in manual flight modes. If you try to use loiter or any auto modes you'll need the compass.
"why do we need the compass?" is a common question. The short answer is that the gps tells you the direction of travel, not the direction the vehicle is facing so the moment the vehicle stops moving you'll start getting drift in the heading and copters spend a lot of time hovering. In fact though, the EKF (Pixhawk only) can figure out the heading as long as the vehicle is moving. The EKF is enabled in AC3.2 but it won't be the main attitude & position management system until AC3.3 most likely.
Should CRate look like first derivative from RelAlt on dataflash logs?
They both should be based on Acc + Baro, but it looks like Acc component for CRate has higher weight in 3.1.4 At least in my case.
As vibrations increase, CRate drops down into negative area, while RelAlt increases (see image). Is that normal?
(I know, I should eliminate vibrations first, but anyways, can CRate be negative, while RelAlt increases?)
Screenshot 2014-05-27 10.10.09.jpg
2014-05-26 22-36-48_14_fix.log
-
17
-
18
-
19
-
20
-
21
of 106 Next