Compiling SITL testing on Mac OS X?

[Niels Joubert]

Hi guys!

Has anyone managed to get the SITL environment running on Mac OS X 10.9 (Mavericks)? I'm working through trying to get this up and running, but no luck so far.

After fixing the OS inconsistencies in ./sim_arducopter.sh, I'm getting a large amount of errors in compiling the C++ code that appear to be clang-related problems, such as:

/Users/njoubert/Code/THIRD_PARTY/UAVs/ardupilot/libraries/AP_InertialSensor/AP_InertialSensor.cpp:259:18: error: variable length array of non-POD element type 'Vector3f' (aka 'Vector3<float>'); Vector3f prev[num_accels]; 

I'm also not able to run the sim_multicopter.py file:

gnome-terminal -e "../Tools/autotest/pysim/sim_multicopter.py --frame=$frame --home=-35.362938,149.165085,584,180"

Dynamic session lookup supported but failed: launchd did not provide a socket path, verify that org.freedesktop.dbus-session.plist is loaded!

Error constructing proxy for org.gnome.Terminal:/org/gnome/Terminal/Factory0: Session D-Bus not running. Try running `launchctl load -w /Library/LaunchAgents/org.freedesktop.dbus-session.plist'.

I'm curious whether anyone else has gotten this to run on OS X?

-Niels 

[Jason Short]

Yes, I have it working.

Bill made a few changes to the Inertial and optflow classes to make it complile

https://github.com/billbonney/ardupilot/tree/bills-wip

Look at his commits and cherrypick the 94cb6fd and 24bf59

Jason

Here is my cheat sheet:

Install just the command line tools for OSX including GCC:

https://developer.apple.com/downloads/index.action?=Command%20Line%20Tools%20%28OS%20X%20Mountain%20Lion%29

Get modules installed to Python 

sudo easy_install pexpect

sudo easy_install pyserial

sudo easy_install pip

grab Mavlink:

https://github.com/tridge/MAVProxy

https://github.com/tridge/mavlink

run this setup command in both directories

sudo python setup.py build install

or???

sudo pip install pymavlink MAVProxy

Add this to .profile:

export PATH="$PATH:/Users/jasonshort/Projects/ArduPilot/GIT/MAVProxy"

Setting TMPDIR for Make

cd /Users/jasonshort/Projects/ArduPilot/GIT/Ardupilot/ArduCopter

TMPDIR=/Users/jasonshort/Projects/ArduPilot/GIT/Ardupilot/tmp

make sitl

-------------------------

go to the main Arducopter directory

cd /Users/jasonshort/Projects/ArduPilot/GIT/Ardupilot/ArduCopter/

Get the elf running:

cd /Users/jasonshort/Projects/ArduPilot/GIT/Ardupilot/tmp/ArduCopter.build

./ArduCopter.elf

Connect via Telnet

telnet localhost 5760

Connect via mavproxy

cd /Users/jasonshort/Projects/ArduPilot/GIT/Ardupilot/ArduCopter/

mavproxy.py --master tcp:127.0.0.1:5760

or

mavproxy.py --master tcp:127.0.0.1:5760 --sitl 127.0.0.1:5501 --out 127.0.0.1:14550 --out 127.0.0.1:14551 --quadcopter --out 10.25.200.15:14550

Mavproxy : Loading Params into elf - > eeprom.bin

param load ../Tools/autotest/ArduCopter.parm

Mavlink arm check

param set ARMING_CHECK 0

param set FLTMODE1 11

Running Pysim

cd /Users/jasonshort/Projects/ArduPilot/GIT/Ardupilot/Tools/autotest/pysim

./sim_multicopter.py --frame=+ --home=-35.362938,149.165085,584,180

./sim_multicopter.py --fgout --frame=+ --home=-35.362938,149.165085,584,180

./sim_multicopter.py --frame=+ --home=-35.362938,149.165085,0,0 --fgout=127.0.0.1:5503

Custom autotest command:

cd /Users/jasonshort/Projects/ArduPilot/GIT/Ardupilot/Tools/autotest

./autotest.py build.ArduCopter fly.ArduCopter logs.ArduCopter convertgpx

FAIL GPS

param set SIM_GPS_DISABLE 1

[Niels Joubert]

Can we get those commits pull-requested into master?

-Niels

[Randy Mackay]

Niels,

 

     The optical flow one is in master now.

 

-Randy

--
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.
For more options, visit https://groups.google.com/groups/opt_out.

[Andrew Tridgell]

> Can we get those commits pull-requested into master?

The one from Bill for AP_InertialSensor is not correct - it doesn't
handle num_gyros being less than INS_MAX_INSTANCES.

I will push a different form of it to master.

Cheers, Tridge

[Andrew Tridgell]

btw, regarding clang support, I think that we should not in general make
changes to ardupilot to support compilers other than gcc. Once we get
into that game then we will quickly end up spending lots of time on
compiler issues and make compromises in the code because some compiler
doesn't support a feature we're using (in this case variable sized stack
arrays).

Making our code portable to multiple boards is worthwhile I think, but
making it portable to multiple compilers for those boards isn't. There
are just too many bad compilers out there and too many ways in which
compiler writers disagree about what is acceptable in C++.

For MacOS SITL what I think should be done is that someone put together
a true gcc/g++ package for recent MacOS versions, or a wiki page
describing how to install it if it already exists. Oh, and if you find
out who decided that installing clang should install an executable
called 'gcc' that calls clang without first making clang support _all_
gcc extensions then please tell them I hate them :-)

Cheers, Tridge

[Geoff Stearns]

I'm trying to set this up now, and came across the following problem when running 'make sitl' (after installing all the other stuff listed above):

HAL_BOARD=HAL_BOARD_AVR_SITL TOOLCHAIN=NATIVE EXTRAFLAGS=-DGIT_VERSION="55c60b8f"

building /var/folders/00/0d8k0000h01000cxqpysvccm001n28/T/ArduCopter.build/ArduCopter.cpp

%% ArduCopter.cpp

make: *** No rule to make target `/Users/user/Documents/git_public/ardupilot/libraries/AP_AHRS/AP_AHRS_HIL.h', needed by `/var/folders/00/0d8k0000h01000cxqpysvccm001n28/T/ArduCopter.build/ArduCopter.o'.  Stop.

Looks like a file was removed, but make (or some other file) expects it to be there. Maybe an easy fix? I'm not familiar enough with Make to debug it.

[Jason Short]

Have you successfully compiled before? Make clean forces make to rebuild all the objects which could help.

I don't think anyone knows how to compile for SITL without Clang yet.

Jason

-- 
You received this message because you are subscribed to the Google Groups "drones-discuss" group.
To unsubscribe from this group and stop receiving emails from it, send an email to drones-discus...@googlegroups.com.

For more options, visit https://groups.google.com/d/optout.

[Geoff Stearns]

I had not, but a make clean and then make sitl fixed it. Thanks!

I had an older (only a few weeks) checkout and had tried to run make, then realized there were changes merged in since the post above, so updated my copy and try to run it again without make clean. Anyway, looks like it's working using the above instructions just without having to cherrypick the two commits.

[Geoff Stearns]

Well it builds and I can run things, but I can't figure out how to do a full simulation.

I'd like to be able to connect via Droidplanner so I can test changes I make to that.

Using the above instructions, I've built the .elf using make sitl, all goes well.

Here's what I'm trying (all on a mac running 10.9.2):

From the ArduCopter directory in one terminal window, I start the elf:

ArduCopter$ ./../tmp/ArduCopter.build/ArduCopter.elf

I get: 

Starting sketch 'ArduCopter'

Starting SITL input

bind port 5760 for 0

Serial port 0 on TCP port 5760

Waiting for connection ....

Then I start up the copter simulator (also tried + frame, no difference):

autotest/pysim$ ./sim_multicopter.py --frame=X --home=-35.362938,149.165085,584,180

I get:

Simulating 4 motors for frame X

Starting at lat=-35.362938 lon=149.165085 alt=584.0 heading=180.0

Then, I start up MAVProxy:

autotest$ mavproxy.py --master tcp:127.0.0.1:5760 --sitl 127.0.0.1:5501 --out 127.0.0.1:14550 --out 127.0.0.1:14551 --quadcopter --out 10.25.200.15:14550

I get:

Logging to mav.tlog

Loaded module log

Loaded module rally

Loaded module fence

MAV>

Init ArduCopter V3.2-dev (9c5f564d)

Free RAM: 4096

FW Ver: 120

----------------------------------------

Firmware change: erasing EEPROM...

done.

online system 0 component 1

STABILIZE> Mode STABILIZE

APM: Initialising APM...

Entering ESC Calibration: please restart APM.

APM: Initialising APM...

APM: Initialising APM...

[repeats this every second or so forever]

If I try sending a command like 'auto', I get:

STABILIZE> auto

STABILIZE> Got MAVLink msg: COMMAND_ACK {command : 300, result : 0}

AUTO> Mode AUTO

APM: Initialising APM...

[repeats again forever]

Sometimes I see it say something about needing to restart APM 

-----

I thought maybe I had to load the copter params, but when I try to run:

> param load ArduCopter.parm

I get:

Unknown parameter FRAME

Unknown parameter MAG_ENABLE

Unknown parameter FS_THR_ENABLE

Unknown parameter BATT_MONITOR

Unknown parameter CH7_OPT

Unknown parameter COMPASS_LEARN

[ ... repeated for every param in that file ... ]

Loaded 0 parameters from ArduCopter.parm (changed 0)

So, questions:

Do I need to get the params loaded? Any idea why it's not working when I try via MAVProxy?

See anything else I might be missing or doing wrong here?

Thanks for any help!

[Jason Short]

You can't run EKF on OSX SITL. There is an issue with the compiler.

Turn off EKF by adding #define AP_AHRS_NAVEKF_AVAILABLE 0 in Arducopter.pde near the top.

Jason

[Geoff Stearns]

Added that line and did a make clean; make sitl and tried it all again, seems about the same:

$ mavproxy.py --master tcp:127.0.0.1:5760 --sitl 127.0.0.1:5501 --out 127.0.0.1:14550 --out 127.0.0.1:14551 --quadcopter --out 10.25.200.15:14550

Logging to mav.tlog

Loaded module log

Loaded module rally

Loaded module fence

MAV>

Init ArduCopter V3.2-dev (9c5f564d)

Free RAM: 4096

FW Ver: 120

----------------------------------------

Firmware change: erasing EEPROM...

done.

online system 0 component 1

STABILIZE> Mode STABILIZE

APM: Initialising APM...

Entering ESC Calibration: please restart APM.

APM: Initialising APM...

APM: Initialising APM...

[continues forever]

Any other ideas? The ESC calibration message showed up before as well, but not every time I run it.

--
You received this message because you are subscribed to a topic in the Google Groups "drones-discuss" group.
To unsubscribe from this topic, visit https://groups.google.com/d/topic/drones-discuss/kLx9kJAT9As/unsubscribe.
To unsubscribe from this group and all its topics, send an email to drones-discus...@googlegroups.com.

[Jason Short]

The EKF work is a work in progress. Maybe my hack doesn't shut it off properly so it falls back on the DCM.  Look around for how it is disabled on APM and try to force that. 

Jason

Sent from my iPhone

[Geoff Stearns]

Looks fine to me, everywhere that calls it is either controlled by that flag, or is disabled entirely.

How do you know it's the EKF stuff that's messing it up? Is that a hunch or is there some other troubleshooting I could try to figure out the problem?

[Ben Nizette]

I think it's actually something to do with parameters rather than (or in addition to?) EKF, specifically a GCC-ism that we use that isn't correctly implemented by Clang.  The only workaround I've heard of that works at this stage is to install /real/ GCC on your Mac and build with that, rather than Clang pretending to be GCC (use "gcc -v" to tell the difference).

HTH.

--Ben.

[Jason Short]

That was my problem two weeks ago. It looked almost identical, always initializing. When I disabled EKF it worked right away.

Jason

[Jason Short]

If you could explain how to do that I would be grateful.

Jason

[Vũ Nguyễn]

Hi Jason,

I'm new to Ardupilot. Could you please instruct how to disable the EKF?

Many thanks.

Vu

[Jason Short]

There is a define in the code you can use to disable EKF. Just put it in Arducopter.pde 

I don't have a computer right in front of me so you'll have to poke around. 

Jason

Sent from my iPhone

[Víctor MV]

In case anyone is interested in developing in Mac OS i needed to make these changes to be able to compile SITL.

Cheers