Skip to content

Installation Guide

Jump to bottom
Cameron Finucane edited this page Sep 19, 2013 · 35 revisions

LTLMoP is supported on Linux (developed mainly on Ubuntu), Mac OS X (developed mainly on 10.8.x), and Windows (developed mainly on Win7). If you have any LTLMoP-specific trouble with installation, please feel free to post an issue.

Step 1: Install Dependencies

Quick Cheat Sheet

Listed here is the bare-minimum software that must be installed for LTLMoP to run.

For a step-by-step guide on how to install these dependencies, please see the next sections.

  • Programming language environments:
    • Java Runtime Environment (JRE) and Java Development Kit (JDK)
    • Python 2.7.x
  • Python libraries:
    • wxPython (A cross-platform GUI library)
    • NumPy and SciPy (Libraries that provide matrix operations and optimization routines)
    • Polygon (For decomposition and other polygon operations)

Step-by-Step Guide (Windows)

  1. Java may already be installed on your computer; otherwise, see the Java SE Download Page. Be sure you have both the JRE and the JDK (you should be able to successfully run both java and javac from the command prompt).

    Note: You may need to manually add the JDK bin/ folder to your system path. On Windows, this can be done from the command-line using the command setx PATH "%PATH%;C:\Program Files\Java\jdkXXXX\bin" where XXXX is the JDK version number. Otherwise, see this page for more information.

  2. For Python, you can use the official Python installer and official wxPython installer along with this repository of library binaries (be sure to choose the packages for your version and architecture of Python).

    Alternatively, for 32-bit Windows, the Python(x,y) package is a fairly convenient way to install everything at once (though you may want to disable installation of large unrelated packages like Eclipse).

    Note: You may also need to manually add Python to your system path. The procedure is almost identical to the instructions above for adding the Java path, except you will save the path to the Python folder. Python usually places its folder in C:\PythonXX where XX is the version number.

  3. The Polygon library for Python can be installed using the installer from the official website.

Step-by-Step Guide (Mac OS X)

  1. To install dependencies on OS X, you will need to install a package manager.

    Homebrew is recommended, and these instructions will assume you are okay with that choice.

    • If you already have a working Homebrew installation, you should skip to Step 4.
    • If you already have MacPorts installed, please see the appendix.

    To install Homebrew, open up a Terminal window and run: ruby -e "$(curl -fsSL https://raw.github.com/mxcl/homebrew/go)"

  2. If you do not have them already, install the Command Line Tools for Xcode from Apple's developer downloads website (free but unfortunately requires Apple registration).

  3. Run brew doctor. If you see a message like "Consider setting your PATH so that /usr/local/bin occurs before /usr/bin", run:

echo export PATH="/usr/local/bin:$PATH" >> ~/.bash_profile
source ~/.bash_profile
  1. Ensure you have a java compiler. Run javac; if it asks if you want to install, say yes.
  2. brew install git
  3. brew install python. If it complains about having trouble linking, run brew link --overwrite python.
  4. Install Numpy and Scipy:
brew tap samueljohn/python
brew tap homebrew/science
pip-2.7 install nose
brew install numpy scipy

This will take about ten minutes.

  1. brew install wxmac. This will take a while (10-20 minutes).
  2. pip install Polygon2

Step-by-Step Guide (Ubuntu)

  1. Java may already be installed on your computer; otherwise, see your package manager (e.g. Synaptic on Ubuntu). You should be able to successfully run both java and javac from a terminal.

  2. Most versions of Linux come with Python pre-installed. The necessary Python packages can be installed using sudo apt-get install python-numpy python-scipy wx-python, or via Synaptic Package Manager (which itself can be installed with Ubuntu Software Center).

  3. The Polygon library for Python can be installed using pip install Polygon2.

Other Optional Dependencies

This section lists other packages that can be used for extended functionality.

  • If you will want to be able to view automata as PDFs: install Graphviz.
  • To perform unrealizable core-finding, LTLMoP currently requires PicoSAT to be installed into an appropriate subdirectory of src/lib/cores, e.g. src/lib/cores/picosat-954. Be sure to enable trace support (i.e. ./configure --trace). See also Tips for compiling PicoSAT on Windows.
  • To use OMPL for motion planning, you will need to install it.
  • If you want to use the ODE simulator ( tested only on Windows and Ubuntu ):
    • Install pygame, PyOpenGL, and PyODE (see the previous section for library sources)
    • Install the Open Dynamics Engine (ODE):
      • Get version 0.8 ( no later versions! ).
      • If you are installing ODE on Windows, you should download the source package, then navigate to the build directory which is populated with different versions of Visual Studio solution folders named vs200X. Open one of these solutions in Visual Studio and build the solution there. This will install ODE on your system.

Step 2: Download LTLMoP

There are three ways to download a copy of LTLMoP:

  • The first option is to download an up-to-date copy of the LTLMoP development branch (.zip) from GitHub. After downloading, extract the contents of the LTLMoP zip file to the desired install directory.

  • The second is for Git users to clone the repository by running git clone https://github.com/LTLMoP/LTLMoP.git).

  • Lastly, those that wish to automatically set up a development environment for LTLMoP can download and run this install script (Note: this is the recommended method for ASL members). After downloading, simply run the script in Python. This script will help you get set up with a fork of LTLMoP on GitHub and check it out onto your computer. LTLMoP uses Git for version tracking. Make sure it is installed on your system (Windows: get Git for Windows, OS X: run sudo port install git-core, Ubuntu: install the git Synaptic package).

(Note: On Windows, due to write permissions, LTLMoP does not behave well when installed to the Program Files directory. The recommended install directory is C:\LTLMoP.) 

Step 3: Prepare LTLMoP for first use

Compiling the Synthesis Subsystem (JTLV/GROne)

This only needs to be done once (unless you upgrade your copy of LTLMoP, in which case you should be sure to do this again!)

  • Navigate to the directory (LTLMoP install dir)/src/etc/jtlv/
  • Run the build.sh script, either by double-clicking it or typing sh build.sh at the command prompt.
  • If the script succeeds, it will exit with no output and the GROne/ directory will contain several .class files.

(Note: If you have trouble running the build script, see the instructions in (LTLMoP install dir)/src/etc/jtlv/JTLV_INSTRUCTIONS. Don't forget that the JDK needs to be in your path (see the Dependencies section); if you can't run javac successfully from the command-line, the build process will not succeed.)

Now go check out the tutorial.

Appendix

For LTLMoP Developers

You may want to install wxGlade, which is used for developing the LTLMoP GUIs.

Platform-specific Notes

  • To get crisp text rendering on Retina Macs, you need to use wxPython >= 2.9 (on MacPorts this may mean you need to use the py27-wxpython-devel package) and you need to add the following lines to Info.plist inside the Python.app package (e.g. /opt/local/Library/Frameworks/Python.framework/Versions/2.7/Resources/Python.app/Contents/Info.plist):
  <key>NSPrincipalClass</key>
  <string>NSApplication</string>

Note: This trick is from this page on the wxWidgets Wiki.
Note: If this doesn't work, try modifying /System/Library/Frameworks/Python.framework/Versions/2.7/Resources/Info.plist instead

Installing dependencies with MacPorts

You should be able to just run: sudo port install python27 py27-numpy py27-polygon py27-scipy py27-wxpython-devel.

Special warning for users of the Aldebaran Nao

  • As of Oct. 2012, the Windows Naoqi SDK only supports 32-bit Python installations.
  • As of Oct. 2012, the OS X Naoqi SDK seems to only work with the version of Python shipped with OS X (10.8). Copying the libraries necessary for LTLMoP from your MacPorts installation to the OS X Python installation directory seems to work okay, despite being kind of scary.
Clone this wiki locally