ICub machines configuration: Difference between revisions

From ISRWiki
Jump to navigation Jump to search
m (→‎YARP and iCub: CREATE_GUIS -> YARP_COMPILE_GUIS, CREATE_LIB_MATH -> YARP_COMPILE_libYARP_MATH)
 
(138 intermediate revisions by the same user not shown)
Line 1: Line 1:
In this page we will list configurations that are commonplace to machines used at [[Vislab]], particularly Linux machines that work with the iCub robot.
In this page we describe the setup of the computers connected to the iCub robot, which all share a common network disk and configuration.


''See [[VisLab machines configuration/Archive]] for obsolete information and environment variables.''
''See [[iCub machines configuration/Archive]] for obsolete information.''


= Operating system installation =
= Operating system installation =


Nothing fancy here - follow Ubuntu defaults and partitioning. For servers, use LTS releases. Machines involved in demos must have a user named 'icub', to make the distributed setup easier (so that the graphical application manager, gyarpmanager, can manage applications on remote machines).
Ubuntu LTS, default settings and partitioning. The first user to be created must be called '''icub''', to make the distributed setup possible: for NFS network mount, this user has to have uid 1000 and guid 1000.


== Operations after first boot ==
In order to add a user:
 
* either use the Ubuntu graphical frontends
* update packages:
* or use a Terminal:
** either with Ubuntu graphical frontends
sudo adduser icub
** or from a Terminal: <code>sudo apt-get update</code>; <code>sudo apt-get upgrade</code>
sudo usermod -aG sudo icub # gives sudo privileges
* add 'icub' user if not done already:
** either with Ubuntu graphical frontends
** or from a Terminal: <code>sudo adduser icub</code> followed by <code>sudo usermod -aG admin icub</code> to give it sudo privileges


= Other operations =
= Other operations =
Line 22: Line 19:
''See also: [[VisLab network]], [[ISR computing resources]]''.
''See also: [[VisLab network]], [[ISR computing resources]]''.


Manually configure your internet connection as detailed in one of the following subsections, depending if the machine is a desktop or a server one.
Configure a static IP as explained in one of the following subsections, depending if the machine is a desktop or a server one.


Optionally, you may also customize the <code>/etc/hosts</code> file [[Chico3 laptop machine configuration#Network|like we did on Chico3]]; this would allow us to quickly access other machines, as in: <code>ping cortex1</code>. Alternatively, we can do nothing and just use <code>ping cortex1.visnet</code> (i.e., attach the <code>.visnet</code> part after a machine name, see [[VisLab network]] for details).
Also, it is recommended to set up the <code>/etc/hosts</code> file as follows:
 
10.10.1.41 icubbrain1
10.10.1.42 icubbrain2
10.10.1.50 pc104
10.10.1.51 icub-cuda
10.10.1.53 icub-laptop
 
You should be able to do <code>ping icubbrain1</code>


=== Desktop machines ===
=== Desktop machines ===


With the graphical Network Manager (https://help.ubuntu.com/12.04/ubuntu-help/net-fixed-ip-address.html), configure the connection "Auto eth0" IPv4 as follows:
With the graphical Network Manager (https://help.ubuntu.com/16.04/ubuntu-help/net-fixed-ip-address.html), configure the connection "Auto eth0" IPv4 as follows:
{| class="wikitable" border="1"
{| class="wikitable" border="1"
|+  
|+  
Line 42: Line 47:


Edit <code>/etc/network/interfaces</code> like this:
Edit <code>/etc/network/interfaces</code> like this:
  auto lo
auto lo
  iface lo inet loopback
iface lo inet loopback
    
    
  auto eth0
auto eth0
  iface eth0 inet static
iface eth0 inet static
  address 10.x.y.z # put your IP here, see above table
address 10.x.y.z # put your IP here, see above table
  netmask 255.255.x.y # see above table
netmask 255.255.x.y # see above table
  network 10.10.1.0
network 10.10.1.0
  broadcast 10.10.1.255
broadcast 10.10.1.255
  gateway 10.10.1.254
gateway 10.10.1.254
dns-nameservers 10.0.0.1 10.0.0.2
 
In some versions of Ubuntu, to configure DNS you also need to edit <code>/etc/resolvconf/resolv.conf.d/head</code> like this:
nameserver 10.0.0.1
nameserver 10.0.0.2
 
then run:
sudo resolvconf -u
 
== Dependencies ==


Edit <code>/etc/resolv.conf</code> like this:
Installing the '''icub-common''' metapackage is sufficient. It is a bundle of the following packages (for more details see [http://wiki.icub.org/wiki/Linux:Installation_from_sources here], [http://wiki.icub.org/wiki/Linux:_dependencies here] and [https://github.com/robotology/yarp/blob/master/.travis.yml here]):
  nameserver 10.0.0.1
sudo apt install build-essential libace-dev libedit-dev libeigen3-dev libncurses5-dev gfortran libtinyxml-dev libgraphviz-dev
  nameserver 10.0.0.2
sudo apt install git-core ssh gcc g++ make cmake-curses-gui freeglut3-dev libxmu-dev libswscale-dev libavformat-dev


== Essential packages ==
Qt5 graphics dependencies in Ubuntu 16.04 Xenial:


  sudo apt-get install ssh git-core libace-dev libgsl0-dev libncurses5-dev gfortran libtinyxml-dev
  sudo apt install qttools5-dev qtdeclarative5-dev qtdeclarative5-controls-plugin qtdeclarative5-dialogs-plugin qtmultimedia5-dev qtdeclarative5-qtmultimedia-plugin qtquick1-5-dev libqt5opengl5-dev
  sudo apt-get install gcc g++ make cmake-curses-gui qttools5-dev qtdeclarative5-dev qtmultimedia5-dev
 
sudo apt-get install subversion libgtkmm-2.4-dev
Qt5 graphics dependencies in Debian Stretch:
  sudo apt install qml-module-qt-labs-folderlistmodel qml-module-qt-labs-settings
 
iCub Simulator dependencies: SDL and ODE.
 
GTK graphical programs are obsolete and replaced by their Qt equivalents. If you still want the old GTK programs, install <code>libgtkmm-2.4-dev</code>


== Environment variables ==
== Environment variables ==


* Create a file called ~/.bash_env like this one:
* Create a file called ~/.bashrc_iCub like this one. Usually you do not need all of the following variables and settings, just a subset.


  # /usr/local/src/robot directory can be mounted from NFS, or created manually with permissions:  sudo chown icub.icub /usr/local/src/robot -R
   export ROBOT_CODE=/usr/local/src/robot
   export code=/usr/local/src/robot
 
   export YARP_ROOT=$code/yarp
   export YARP_ROOT=$ROBOT_CODE/yarp
   export YARP_DIR=$YARP_ROOT/build
   export YARP_DIR=$YARP_ROOT/build
   export ICUB_ROOT=$code/icub-main
   export ICUB_ROOT=$ROBOT_CODE/icub-main
   export ICUB_DIR=$ICUB_ROOT/build
   export ICUB_DIR=$ICUB_ROOT/build
   export YARP_ROBOT_NAME=iCubLisboa01 # only for machines that connect to the real robot
 
   export IPOPT_DIR=/home/icub/Ipopt-3.10.2/build # only for servers (IK solver) - see http://wiki.icub.org/wiki/Installing_IPOPT
  export ICUBcontrib_DIR=$ROBOT_CODE/icub-contrib-common/build
   export OpenCV_DIR=$code/OpenCV-2.4.8/build # manually installed, with TBB support - see below
   export YARP_DATA_DIRS=$YARP_DIR/share/yarp:$ICUB_DIR/share/iCub:$ICUBcontrib_DIR/share/ICUBcontrib:$ROBOT_CODE/speech/svox-speech/build/share/speech
   export PATH=$PATH:$YARP_DIR/bin:$ICUB_DIR/bin # if you only need main iCub binaries; for contrib, see http://mediawiki.isr.ist.utl.pt/wiki/RobotCub_software#Getting_the_iCub_repository_in_Linux
   export PATH=$PATH:$YARP_DIR/bin:$ICUB_DIR/bin:$ICUBcontrib_DIR/bin
   source $YARP_ROOT/scripts/yarp_completion 
 
  export ODE_DIR=$ROBOT_CODE/ode-0.13/build
 
  # OpenCV: select one of the following
  #export OpenCV_DIR=$ROBOT_CODE/opencv2/build
  #export OpenCV_DIR=$ROBOT_CODE/opencv2/build-cuda
  export OpenCV_DIR=$ROBOT_CODE/opencv3/build
   #export OpenCV_DIR=$ROBOT_CODE/opencv3/build-cuda
 
  # To enable tab completion on yarp port names
  if [ -f $YARP_ROOT/scripts/yarp_completion ]; then
    source $YARP_ROOT/scripts/yarp_completion
  fi
 
  # Set the name of your robot here.
  export YARP_ROBOT_NAME=iCubLisboa01
 
  # Set-up optimizations
  export CMAKE_BUILD_TYPE=Release
 
  # DebugStream customization
  export YARP_VERBOSE_OUTPUT=0
  export YARP_COLORED_OUTPUT=1
  export YARP_TRACE_ENABLE=0
  export YARP_FORWARD_LOG_ENABLE=0
 
  # Lua
  ### DO NOT REMOVE ';;;' ###
  export LUA_PATH=";;;$ROBOT_CODE/rFSM/?.lua;$ICUBcontrib_DIR/share/ICUBcontrib/contexts/interactiveObjectsLearning/LUA/?.lua"
  export LUA_CPATH=";;;$YARP_ROOT/bindings/build-lua/?.so"
   export PATH=$PATH:$ROBOT_CODE/rFSM/tools:$ICUBcontrib_DIR/share/ICUBcontrib/contexts/interactiveObjectsLearning/LUA
  export PATH=$PATH:$YARP_ROOT/bindings/build-lua
   export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$YARP_ROOT/bindings/build-lua


* Then, before the following line of /etc/bash.bashrc
* Then, before the following line of /etc/bash.bashrc
  [ -z "$PS1" ] && return
[ -z "$PS1" ] && return
add this:
add this:
  # per-user environment variables (non-interactive and interactive mode)
# per-user environment variables (non-interactive and interactive mode)
  source $HOME/.bash_env
source $HOME/.bashrc_iCub


The reason why we use the above custom file (as opposed to the standard ~/.bashrc) is that we want to enforce the variables both during interactive and non-interactive sessions, such as commands launched via <code>yarprun</code> from another machine.
The reason why we use the above custom file (as opposed to the standard ~/.bashrc) is that we want to enforce the variables both during interactive and non-interactive sessions, such as commands launched via <code>yarprun</code> from another machine.


== Subversion ==
See also https://git.robotology.eu/mbrunettini/icub-environment for the variables and scripts employed at IIT.
 
For security, uncomment (thus enabling) the following parameters in <code>/etc/subversion/servers</code>, in the [global] section:
  store-passwords = no
  store-plaintext-passwords = no
This implies that SVN will ask you for your password every time you do a commit, as opposed to storing it in plain text on the system. (Don't worry about changing your personal <code>~/.subversion/config</code> file: the parameter is not actually set there, so the global <code>/etc</code> setting is used.)


= Additional software =
= Additional software =
Line 100: Line 147:
=== Ubuntu packages ===
=== Ubuntu packages ===


  sudo apt-get install libcv-dev libhighgui-dev libcvaux-dev libopencv-gpu-dev
* this is the easiest way to install OpenCV, however the Ubuntu packages might be too old and/or missing some specific features (in that case proceed with a manual installation, see below)


This is the easiest way to install OpenCV, however some machines may require a custom manual compilation instead (see below).
  sudo apt install libcv-dev libhighgui-dev libcvaux-dev libopencv-gpu-dev


=== Manual compilation ===
=== Manual compilation ===


This is needed for some iCub vision modules (e.g. motionCUT, stereoDisparity). Instructions:
* on most machines: download OpenCV, create a build directory, CMake, set WITH_CUDA=OFF, compile, set OpenCV_DIR to the path of OpenCV-x.y.z/build, for example:
export OpenCV_DIR=$code/OpenCV-x.y.z/build
 
* on CUDA machines, in order to compile CUDA-enabled modules: create a build-cuda directory, CMake, set WITH_CUDA=ON, compile, set OpenCV_DIR to the path of OpenCV-x.y.z/build-cuda. You may need to disable <code>cudacodec</code>, see also https://github.com/opencv/opencv_contrib/issues/1786
 
== YARP and iCub ==
 
* In general, when compiling this software, do '''not''' use <code>sudo make install</code> but simply <code>make</code> (and configure the PATH variable in such a way that it finds the binaries from the build directory).
 
* If you work with the robot, use the volume shares exported from the NFS server.
 
* In other cases, follow the instructions on the [[iCub software]] article.  


* download [http://threadingbuildingblocks.org/download TBB Source], make
* yarp CMake configuration
* download OpenCV 2.4.3 or higher, CMake, set WITH_TBB=ON and insert the paths obtained with the TBB compilation, such as
CMAKE_BUILD_TYPE Release
  TBB_INCLUDE_DIR=/usr/local/src/robot/tbb41_20130314oss/include
YARP_COMPILE_GUIS
  TBB_LIB_DIR=/usr/local/src/robot/tbb41_20130314oss/build/linux_intel64_gcc_cc4.6_libc2.15_kernel3.2.0_release
YARP_COMPILE_libYARP_MATH


* compile OpenCV
// to enable 640x480@30Hz images with Bayer encoding
// install libraw1394-dev libdc1394-22-dev then enable
CREATE_OPTIONAL_CARRIERS
ENABLE_yarpcar_bayer ON
ENABLE_yarpcar_mjpeg ON


* set OpenCV_DIR to the path of OpenCV-x.y.z/build, for example:
* icub-main CMake configuration
export OpenCV_DIR=$code/OpenCV-2.4.3/build
CMAKE_BUILD_TYPE Release
// on servers, do http://wiki.icub.org/wiki/Installing_IPOPT then enable
ENABLE_icubmod_cartesiancontrollerclient ON
ENABLE_icubmod_cartesiancontrollerserver ON
ENABLE_icubmod_gazecontrollerclient ON


If the above fails, try BUILD_TBB=ON (i.e. automatically "Download and build TBB from source"). More information about TBB support in the latest OpenCV [http://answers.opencv.org/questions/query:tbb here].
* final configuration
# <code>yarp namespace /icub</code>
# <code>yarp conf 10.10.1.53 10000</code> (yarpserver runs on [[iCub laptop]])


=== Possible problems with manual compilation ===
* special machines such as [[pc104]] need different flags


'''error: ‘ptrdiff_t’ does not name a type'''
== CUDA ==


Solution: edit OpenCV-x.y.z/modules/core/include/opencv2/core/core.hpp, add #include <stddef.h>
=== Prerequisites ===


'''fatal error: linux/videodev.h: No such file or directory'''
sudo apt install freeglut3-dev libdevil-dev libglew-dev
sudo apt purge libcudart*  // because we will manually install it


Solution:  
Troubleshooting: http://askubuntu.com/questions/410604/installing-nvidia-drivers-with-pkg1-run-ends-with-no-version-h-found
sudo apt-get install libv4l-dev
cd /usr/include/linux
sudo ln -s ../libv4l1-videodev.h videodev.h


'''undefined reference to `cvCreateCameraCapture_V4L(int)''''
=== CUDA Toolkit, SDK and Examples ===


Solution: see [https://code.ros.org/trac/opencv/ticket/324 here] (including the change suggested in the comments)
* stop X servers: <code>sudo service lightdm stop</code> (or <code>gdm stop</code> depending on configuration)
* download and install NVIDIA CUDA Toolkit from https://developer.nvidia.com/cuda-downloads
* preferably, use Installer Type: runfile (local). Alternatively, deb (local)
* if you obtain the error "Toolkit:  Installation Failed. Using unsupported Compiler.", use the override option, e.g., <code>./cuda_6.0.37_linux_64.run --override</code>
* if you obtain the error "The driver installation is unable to locate the kernel source. Please make sure that the kernel source packages are installed and set up correctly":
** read the CUDA log in /tmp and verify that the graphics card is currently supported -- if not, you might need to install a legacy NVIDIA driver. For example, the Quadro FX 580 card needs NVIDIA legacy drivers 340.xx: install them and then answer '''no''' when CUDA Toolkit installer asks "Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 346.46?"
** <code>sudo apt install linux-generic linux-headers-$(uname -r) linux-headers-generic-lts-trusty</code> (or other Ubuntu version codename)
** call the installer specifying the kernel source path, e.g., <code>./cuda_7.0.28_linux.run --kernel-source-path=/usr/src/linux-headers-3.13.0-52-generic/</code>
* output of successful installation:
Driver:  Installed
Toolkit:  Installed in /usr/local/cuda-7.0
Samples:  Not Selected
Please make sure that
-  PATH includes /usr/local/cuda-7.0/bin
-  LD_LIBRARY_PATH includes /usr/local/cuda-7.0/lib64, or, add /usr/local/cuda-7.0/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-7.0/bin
To uninstall the NVIDIA Driver, run nvidia-uninstall


'''error: ‘AVERROR_NUMEXPECTED’ was not declared in this scope'''
* troubleshooting:
** http://askubuntu.com/questions/451672/installing-and-testing-cuda-in-ubuntu-14-04
** http://troylee2008.blogspot.pt/2012/05/linking-error-while-compiling-cuda-sdk.html
** https://devtalk.nvidia.com/default/topic/617414/-solved-cuda-driver-version-is-insufficient-for-cuda-runtime-version-fedora-18-rpmfusion-driver/


Solution: apply the two patches located [https://code.ros.org/trac/opencv/ticket/1020 here] with the following commands.
=== SiftGPU ===
wget https://code.ros.org/trac/opencv/raw-attachment/ticket/1020/ffmpeg_build.patch https://code.ros.org/trac/opencv/raw-attachment/ticket/1020/ffmpeg_build_2.patch
patch -p1 < ffmpeg_build.patch
# when asked for which file to patch, give the full path to OpenCV-x.y.z/modules/highgui/src/cap_ffmpeg.cpp
patch -p1 < ffmpeg_build_2.patch
# ditto


'''error: ‘av_rescale_q’ was not declared in this scope'''
* download it from http://cs.unc.edu/~ccwu/siftgpu/
* unzip it - typically in your home directory, not in the NFS shared folder
* compile with <code>make</code>
* if you obtain "unspecified launch failure" and 0 sift features/matches, check that the X server is running: <code>sudo service lightdm start</code>
* if you obtain the error "/usr/local/cuda/bin/nvcc: Command not found", this can help: <code>sudo ln -s /usr/lib/nvidia-cuda-toolkit /usr/local/cuda</code>. See also http://askubuntu.com/questions/231503/nvcc-compiler-setup-ubuntu-12-04
* run the test program <code>SimpleSIFT</code> - it should work via ssh, as well as in a local session
* example of successful execution:
$ ./SimpleSIFT
Unable to create OpenGL Context!
For nVidia cards, you can try change to CUDA mode in this case
NOTE: changing maximum texture dimension to 32768
[SiftGPU Language]: CUDA
Image size : 800x600
Image loaded : ../data/800-1.jpg
#Features: 3347
#Features MO: 3910
[RUN SIFT]: 0.339
Image size : 640x480
Image loaded : ../data/640-1.jpg
#Features: 2372
#Features MO: 2767
[RUN SIFT]: 0.208
NOTE: changing maximum texture dimension to 32768
[SiftMatchGPU]: CUDA
2247 sift matches were found;


Solution: edit OpenCV-x.y.z/modules/highgui/src/cap_ffmpeg.cpp, add #include <libavutil/mathematics.h> before #include <ffmpeg/avcodec.h> ([http://eternalwandering777.wordpress.com/2012/05/18/opencv-2-3-1-build-on-ubuntu-12-04/ source])
* define <code>export SIFTGPU_DIR=~/SiftGPU</code> or similar in your iCub bashrc file, so that libsiftgpu.so is found by GPU accelerated modules


'''error: #error The Eigen/Array header does no longer exist in Eigen3. All that functionality has moved to Eigen/Core.'''
== himrep ==


Solution: apply the patch located [https://code.ros.org/trac/opencv/ticket/805 here] with the following commands.
Clone https://github.com/robotology/himrep, follow the instructions to compile <code>liblinear</code>, CMake, make install
wget https://code.ros.org/trac/opencv/raw-attachment/ticket/805/diff.txt
patch -p1 < diff.txt
# when asked for which file to patch, give the full path to OpenCV-x.y.z/modules/features2d/src/matchers.cpp


'''No rule to make target `/usr/lib/python3.0/config/libpython3.0.so', needed by `lib/cv.so'.'''
To use the deep neural network object recognition, based on Caffe, follow the instructions at README_Caffe.md.
If you get "error: kernel launches from templates are not allowed in system files", use an older GCC version like 4.6 (see also https://github.com/BVLC/caffe/issues/337). If you get "ImportError: No module named 'yaml'", do <code>sudo apt install python-yaml</code>.


Solution: ccmake ., toggle advanced options (press 't'), set
== IOL ==
PYTHON_LIBRARY /usr/lib/python2.7/config/libpython2.7.so


'''error: ‘avformat_free_context’ was not declared in this scope'''
* <code>sudo apt install lua5.1 liblua5.1-dev</code>
* clone [https://github.com/kmarkus/rFSM rFSM] (no need to compile anything here)


Solution: remove any manually installed ffmpeg version, then
Clone https://github.com/robotology/iol, CMake, make install
sudo apt-get install libavcodec-dev libavformat-dev libswscale-dev


== YARP and iCub ==
== stereo-vision ==
 
Clone https://github.com/robotology/stereo-vision, CMake with USE_SIFT_GPU=ON, make install
 
= Best practices =
 
Below are some tips and tricks taken from:
 
* http://wiki.icub.org/wiki/Yarp-config
* http://www.yarp.it/yarp-config.html
* [https://docs.google.com/document/d/1S9m4DbrV1AJWo_E1r3F8k3ZzmhaTOWRJ9WFjXPFqVos IIT - How to setup and start the iCub from scratch] (last updated December 2016)
 
== XML files ==
 
Edit XML files locally, in /home/icub/.local/share/yarp
 
To install robot-specific XML files, compile icub-main (just <code>make</code>) then use commands like <code>yarp-config robot --import-all</code> (installs all files) or <code>yarp-config robot --import iCubLisboa01 affordancesExploration.xml</code> (installs specific files)
 
== ini files ==
 
To install application conf/*.ini files, compile a project (e.g., icub-main, poeticon, iol) then use commands like <code>yarp-config context --import-all</code> (installs all files) or <code>yarp-config context --import actionsRenderingEngine</code> (installs the files of specific applications)
 
== Disabling some robot parts ==


Follow the instructions on the [[RobotCub software]] article. For compilation on servers, do '''not''' use <code>sudo make install</code> but simply <code>make</code> (we have configured the PATH variable to find the latest compiled binaries, and we do not want two copies of the same thing on the system).
If the robot is not complete (or some parts need to be disabled):


* usual yarp CMake flags
* In a pc104 shell, type <code>yarp-config robot --list</code> to look for the .ini files from where the configuration values are launched: these are local copies of the robots-configuration repository files
  CMAKE_BUILD_TYPE Release
* Go to the <code>INSTALLED DATA</code> directory path and then, within the corresponding robot folder (e.g., iCubLisboa01), look for the file yarprobotinterface.ini, which points to a .xml file which contains the configuration paths for all the robot parts (e.g., icub_all.xml)
  CREATE_GUIS
* If the local config file does not exist, there is only the canonical file in the build path: in that case, create a local one using <code>yarp-config --import</code>
  CREATE_LIB_MATH
* Make a copy of the .xml file giving it a descriptive name (e.g., icub_no_legs.xml) and, in the copied file, comment or remove all lines that refer to .ini files of the part(s) that you want to disable
* Edit the contents of yarprobotinterface.ini so that it points to the new .xml file where the parts have been commented
* In the iCubStartup application GUI or xml, modify the way that gravityCompensator and wholeBodyDynamics are launched, so that they don't look for the configuration files of the parts that have been disabled: in the legs example, just add <code>--no_legs</code> to the argument list


* usual iCub CMake flags
= See also =
  CMAKE_BUILD_TYPE Release


Special machines such as [[pc104]] or robot computation servers need additional flags. In addition, machines that work with the physical robot need the following setup:
* http://wiki.icub.org/wiki/Generic_iCub_machine_installation_instructions
# <code>yarp namespace /icub</code>
# in order to find the yarpserver running on [[Chico3 laptop machine configuration | chico3]], write <code>10.10.1.53 10000</code> into the file specified by <code>yarp conf</code>


[[Category:Vislab]]
[[Category:Vislab]]

Latest revision as of 16:06, 28 July 2019

In this page we describe the setup of the computers connected to the iCub robot, which all share a common network disk and configuration.

See iCub machines configuration/Archive for obsolete information.

Operating system installation

Ubuntu LTS, default settings and partitioning. The first user to be created must be called icub, to make the distributed setup possible: for NFS network mount, this user has to have uid 1000 and guid 1000.

In order to add a user:

  • either use the Ubuntu graphical frontends
  • or use a Terminal:
sudo adduser icub
sudo usermod -aG sudo icub  # gives sudo privileges

Other operations

Network configuration

See also: VisLab network, ISR computing resources.

Configure a static IP as explained in one of the following subsections, depending if the machine is a desktop or a server one.

Also, it is recommended to set up the /etc/hosts file as follows: 

10.10.1.41 icubbrain1
10.10.1.42 icubbrain2
10.10.1.50 pc104
10.10.1.51 icub-cuda
10.10.1.53 icub-laptop

You should be able to do ping icubbrain1

Desktop machines

With the graphical Network Manager (https://help.ubuntu.com/16.04/ubuntu-help/net-fixed-ip-address.html), configure the connection "Auto eth0" IPv4 as follows:

Address Netmask Gateway DNS Servers notes
10.10.1.x 255.255.255.0 10.10.1.254 10.0.0.1, 10.0.0.2 visnet (iCub machines)
10.0.x.y 255.255.0.0 10.0.0.254 10.0.0.1, 10.0.0.2 isrnet (rest of ISR)

Servers

Edit /etc/network/interfaces like this: 

auto lo
iface lo inet loopback
  
auto eth0
iface eth0 inet static
address 10.x.y.z # put your IP here, see above table
netmask 255.255.x.y # see above table
network 10.10.1.0
broadcast 10.10.1.255
gateway 10.10.1.254
dns-nameservers 10.0.0.1 10.0.0.2

In some versions of Ubuntu, to configure DNS you also need to edit /etc/resolvconf/resolv.conf.d/head like this: 

nameserver 10.0.0.1
nameserver 10.0.0.2

then run: 

sudo resolvconf -u

Dependencies

Installing the icub-common metapackage is sufficient. It is a bundle of the following packages (for more details see here, here and here): 

sudo apt install build-essential libace-dev libedit-dev libeigen3-dev libncurses5-dev gfortran libtinyxml-dev libgraphviz-dev
sudo apt install git-core ssh gcc g++ make cmake-curses-gui freeglut3-dev libxmu-dev libswscale-dev libavformat-dev

Qt5 graphics dependencies in Ubuntu 16.04 Xenial: 

sudo apt install qttools5-dev qtdeclarative5-dev qtdeclarative5-controls-plugin qtdeclarative5-dialogs-plugin qtmultimedia5-dev qtdeclarative5-qtmultimedia-plugin qtquick1-5-dev libqt5opengl5-dev

Qt5 graphics dependencies in Debian Stretch: 

sudo apt install qml-module-qt-labs-folderlistmodel qml-module-qt-labs-settings

iCub Simulator dependencies: SDL and ODE.

GTK graphical programs are obsolete and replaced by their Qt equivalents. If you still want the old GTK programs, install libgtkmm-2.4-dev

Environment variables

  • Create a file called ~/.bashrc_iCub like this one. Usually you do not need all of the following variables and settings, just a subset.
 export ROBOT_CODE=/usr/local/src/robot

 export YARP_ROOT=$ROBOT_CODE/yarp
 export YARP_DIR=$YARP_ROOT/build
 export ICUB_ROOT=$ROBOT_CODE/icub-main
 export ICUB_DIR=$ICUB_ROOT/build

 export ICUBcontrib_DIR=$ROBOT_CODE/icub-contrib-common/build
 export YARP_DATA_DIRS=$YARP_DIR/share/yarp:$ICUB_DIR/share/iCub:$ICUBcontrib_DIR/share/ICUBcontrib:$ROBOT_CODE/speech/svox-speech/build/share/speech
 export PATH=$PATH:$YARP_DIR/bin:$ICUB_DIR/bin:$ICUBcontrib_DIR/bin

 export ODE_DIR=$ROBOT_CODE/ode-0.13/build

 # OpenCV: select one of the following
 #export OpenCV_DIR=$ROBOT_CODE/opencv2/build
 #export OpenCV_DIR=$ROBOT_CODE/opencv2/build-cuda
 export OpenCV_DIR=$ROBOT_CODE/opencv3/build
 #export OpenCV_DIR=$ROBOT_CODE/opencv3/build-cuda

 # To enable tab completion on yarp port names
 if [ -f $YARP_ROOT/scripts/yarp_completion ]; then
   source $YARP_ROOT/scripts/yarp_completion
 fi

 # Set the name of your robot here.
 export YARP_ROBOT_NAME=iCubLisboa01

 # Set-up optimizations
 export CMAKE_BUILD_TYPE=Release

 # DebugStream customization
 export YARP_VERBOSE_OUTPUT=0
 export YARP_COLORED_OUTPUT=1
 export YARP_TRACE_ENABLE=0
 export YARP_FORWARD_LOG_ENABLE=0

 # Lua
 ### DO NOT REMOVE ';;;' ###
 export LUA_PATH=";;;$ROBOT_CODE/rFSM/?.lua;$ICUBcontrib_DIR/share/ICUBcontrib/contexts/interactiveObjectsLearning/LUA/?.lua"
 export LUA_CPATH=";;;$YARP_ROOT/bindings/build-lua/?.so"
 export PATH=$PATH:$ROBOT_CODE/rFSM/tools:$ICUBcontrib_DIR/share/ICUBcontrib/contexts/interactiveObjectsLearning/LUA
 export PATH=$PATH:$YARP_ROOT/bindings/build-lua
 export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:$YARP_ROOT/bindings/build-lua
  • Then, before the following line of /etc/bash.bashrc
[ -z "$PS1" ] && return

add this: 

# per-user environment variables (non-interactive and interactive mode)
source $HOME/.bashrc_iCub

The reason why we use the above custom file (as opposed to the standard ~/.bashrc) is that we want to enforce the variables both during interactive and non-interactive sessions, such as commands launched via yarprun from another machine.

See also https://git.robotology.eu/mbrunettini/icub-environment for the variables and scripts employed at IIT.

Additional software

OpenCV

Ubuntu packages

  • this is the easiest way to install OpenCV, however the Ubuntu packages might be too old and/or missing some specific features (in that case proceed with a manual installation, see below)
 sudo apt install libcv-dev libhighgui-dev libcvaux-dev libopencv-gpu-dev

Manual compilation

  • on most machines: download OpenCV, create a build directory, CMake, set WITH_CUDA=OFF, compile, set OpenCV_DIR to the path of OpenCV-x.y.z/build, for example:
export OpenCV_DIR=$code/OpenCV-x.y.z/build
  • on CUDA machines, in order to compile CUDA-enabled modules: create a build-cuda directory, CMake, set WITH_CUDA=ON, compile, set OpenCV_DIR to the path of OpenCV-x.y.z/build-cuda. You may need to disable cudacodec, see also https://github.com/opencv/opencv_contrib/issues/1786

YARP and iCub

  • In general, when compiling this software, do not use sudo make install but simply make (and configure the PATH variable in such a way that it finds the binaries from the build directory).
  • If you work with the robot, use the volume shares exported from the NFS server.
  • In other cases, follow the instructions on the iCub software article.
  • yarp CMake configuration
CMAKE_BUILD_TYPE Release
YARP_COMPILE_GUIS
YARP_COMPILE_libYARP_MATH

// to enable 640x480@30Hz images with Bayer encoding
// install libraw1394-dev libdc1394-22-dev then enable
CREATE_OPTIONAL_CARRIERS
ENABLE_yarpcar_bayer ON
ENABLE_yarpcar_mjpeg ON
  • icub-main CMake configuration
CMAKE_BUILD_TYPE Release
// on servers, do http://wiki.icub.org/wiki/Installing_IPOPT then enable
ENABLE_icubmod_cartesiancontrollerclient ON
ENABLE_icubmod_cartesiancontrollerserver ON
ENABLE_icubmod_gazecontrollerclient ON
  • final configuration
  1. yarp namespace /icub
  2. yarp conf 10.10.1.53 10000 (yarpserver runs on iCub laptop)
  • special machines such as pc104 need different flags

CUDA

Prerequisites

sudo apt install freeglut3-dev libdevil-dev libglew-dev
sudo apt purge libcudart*  // because we will manually install it

Troubleshooting: http://askubuntu.com/questions/410604/installing-nvidia-drivers-with-pkg1-run-ends-with-no-version-h-found

CUDA Toolkit, SDK and Examples

  • stop X servers: sudo service lightdm stop (or gdm stop depending on configuration)
  • download and install NVIDIA CUDA Toolkit from https://developer.nvidia.com/cuda-downloads
  • preferably, use Installer Type: runfile (local). Alternatively, deb (local)
  • if you obtain the error "Toolkit: Installation Failed. Using unsupported Compiler.", use the override option, e.g., ./cuda_6.0.37_linux_64.run --override
  • if you obtain the error "The driver installation is unable to locate the kernel source. Please make sure that the kernel source packages are installed and set up correctly":
    • read the CUDA log in /tmp and verify that the graphics card is currently supported -- if not, you might need to install a legacy NVIDIA driver. For example, the Quadro FX 580 card needs NVIDIA legacy drivers 340.xx: install them and then answer no when CUDA Toolkit installer asks "Install NVIDIA Accelerated Graphics Driver for Linux-x86_64 346.46?"
    • sudo apt install linux-generic linux-headers-$(uname -r) linux-headers-generic-lts-trusty (or other Ubuntu version codename)
    • call the installer specifying the kernel source path, e.g., ./cuda_7.0.28_linux.run --kernel-source-path=/usr/src/linux-headers-3.13.0-52-generic/
  • output of successful installation:
Driver:   Installed
Toolkit:  Installed in /usr/local/cuda-7.0
Samples:  Not Selected
Please make sure that
-   PATH includes /usr/local/cuda-7.0/bin
-   LD_LIBRARY_PATH includes /usr/local/cuda-7.0/lib64, or, add /usr/local/cuda-7.0/lib64 to /etc/ld.so.conf and run ldconfig as root
To uninstall the CUDA Toolkit, run the uninstall script in /usr/local/cuda-7.0/bin
To uninstall the NVIDIA Driver, run nvidia-uninstall

SiftGPU

  • download it from http://cs.unc.edu/~ccwu/siftgpu/
  • unzip it - typically in your home directory, not in the NFS shared folder
  • compile with make
  • if you obtain "unspecified launch failure" and 0 sift features/matches, check that the X server is running: sudo service lightdm start
  • if you obtain the error "/usr/local/cuda/bin/nvcc: Command not found", this can help: sudo ln -s /usr/lib/nvidia-cuda-toolkit /usr/local/cuda. See also http://askubuntu.com/questions/231503/nvcc-compiler-setup-ubuntu-12-04
  • run the test program SimpleSIFT - it should work via ssh, as well as in a local session
  • example of successful execution:
$ ./SimpleSIFT 
Unable to create OpenGL Context!
For nVidia cards, you can try change to CUDA mode in this case
NOTE: changing maximum texture dimension to 32768
[SiftGPU Language]:	CUDA
Image size :	800x600
Image loaded :	../data/800-1.jpg
#Features:	3347
#Features MO:	3910
[RUN SIFT]:	0.339
Image size :	640x480
Image loaded :	../data/640-1.jpg
#Features:	2372
#Features MO:	2767
[RUN SIFT]:	0.208
NOTE: changing maximum texture dimension to 32768
[SiftMatchGPU]: CUDA
2247 sift matches were found;
  • define export SIFTGPU_DIR=~/SiftGPU or similar in your iCub bashrc file, so that libsiftgpu.so is found by GPU accelerated modules

himrep

Clone https://github.com/robotology/himrep, follow the instructions to compile liblinear, CMake, make install

To use the deep neural network object recognition, based on Caffe, follow the instructions at README_Caffe.md. If you get "error: kernel launches from templates are not allowed in system files", use an older GCC version like 4.6 (see also https://github.com/BVLC/caffe/issues/337). If you get "ImportError: No module named 'yaml'", do sudo apt install python-yaml.

IOL

  • sudo apt install lua5.1 liblua5.1-dev
  • clone rFSM (no need to compile anything here)

Clone https://github.com/robotology/iol, CMake, make install

stereo-vision

Clone https://github.com/robotology/stereo-vision, CMake with USE_SIFT_GPU=ON, make install

Best practices

Below are some tips and tricks taken from:

XML files

Edit XML files locally, in /home/icub/.local/share/yarp

To install robot-specific XML files, compile icub-main (just make) then use commands like yarp-config robot --import-all (installs all files) or yarp-config robot --import iCubLisboa01 affordancesExploration.xml (installs specific files)

ini files

To install application conf/*.ini files, compile a project (e.g., icub-main, poeticon, iol) then use commands like yarp-config context --import-all (installs all files) or yarp-config context --import actionsRenderingEngine (installs the files of specific applications)

Disabling some robot parts

If the robot is not complete (or some parts need to be disabled):

  • In a pc104 shell, type yarp-config robot --list to look for the .ini files from where the configuration values are launched: these are local copies of the robots-configuration repository files
  • Go to the INSTALLED DATA directory path and then, within the corresponding robot folder (e.g., iCubLisboa01), look for the file yarprobotinterface.ini, which points to a .xml file which contains the configuration paths for all the robot parts (e.g., icub_all.xml)
  • If the local config file does not exist, there is only the canonical file in the build path: in that case, create a local one using yarp-config --import
  • Make a copy of the .xml file giving it a descriptive name (e.g., icub_no_legs.xml) and, in the copied file, comment or remove all lines that refer to .ini files of the part(s) that you want to disable
  • Edit the contents of yarprobotinterface.ini so that it points to the new .xml file where the parts have been commented
  • In the iCubStartup application GUI or xml, modify the way that gravityCompensator and wholeBodyDynamics are launched, so that they don't look for the configuration files of the parts that have been disabled: in the legs example, just add --no_legs to the argument list

See also

Retrieved from "http://mediawiki.isr.tecnico.ulisboa.pt/index.php?title=ICub_machines_configuration&oldid=7093"