Simulation Lab Core

Simulation Lab (SL) is a robotics simulator and a real time control engine that is used by the IAS lab. It was originally developed by Stefan Schaal and colleagues at USC, and it is used at various institutions (e.g. Max Planck Institute in Tuebingen, CMU in the USA, ATR in Japan, etc.), allowing for easy cooperation and joint projects between those labs.

SL can be used in two different modes: simulation and real time control. In simulation, new tasks and methods can be tried out on models of the available robots before testing on the actual real system, thereby making the setup of potentially dangerous (to the robot or its environment) or time consuming experiments more comfortable. However, SL and its concept may take some time getting used to in the beginning. SL runs on Linux and MacOS, and it is currently distributed internally.

Quick Links

Installation Guide

Mac OS

We recommend to use Homebrew to install packages /usr/bin/ruby -e "$(curl -fsSL https://raw.githubusercontent.com/Homebrew/install/master/install)"

All packages installed using brew install go into /usr/local. You should never need to use sudo with brew.

Install CMake
brew update brew install cmake
Install XQuartz (required to show graphics)
Install freeglut brew install homebrew/X11/freeglut
In order for SL to find X11 and freeglut, you need to create a symlink to the folder with these libraries
mkdir /opt/local
ln -s /opt/X11/lib /opt/local/lib

Make sure that /opt/X11/bin is in your PATH by running echo $PATH. If it is not, add line export PATH=$PATH:/opt/X11/bin to ~/.profile and restart Terminal.

Install readline
brew install readline
Change the shared memory size by editing /etc/sysctl.conf and adding
sudo nano /etc/sysctl.conf
kern.sysv.shmmax=2147483648
kern.sysv.shmall=524288
kern.sysv.shmmni=128
kern.sysv.shmseg=32
For older version of OSX (prior to 10.4)
sudo nano /etc/sysctl.conf
kernel.shmmax=2147483648
kernel.shmall=524288
kern.sysv.shmmni=128
kern.sysv.shmseg=32

If you want to know more about shared memory on Mac OS X, read

Configuring Shared Memory on Mac OS X

Adjusting Shared memory segment values.

Restart the computer. Check if you configured shared memory correctly by running sysctl kern.sysv.shmmax. Similarly, you can check other variables.

Ubuntu

Install the following dependencies
sudo apt-get install libncurses5-dev libreadline6-dev freeglut3-dev libxmu-dev cmake cmake-curses-gui libedit-dev libstd++-6-dev clang xterm
Change the shared memory size
sudo nano /etc/sysctl.conf
kernel.shmmax=2147483648
kernel.shmall=2147483648
Restart the computer

Compiling and Running SL

Go to your sl folder and run
mkdir build
cd build
ccmake ..
The CMake interface up will open and you can edit compiling preferences. You can choose which Matlab version to use or which task to compile for each robot. The first time do
- c (configure)
- select the robots to build (eg, BUILD_barrett)
- c
- select the tasks to build (eg, barrett_gen_matlab)
- g (generate)

It can happen that the Matlab root is not automatically recognized and you receive an error at the second c. If that happens, press e to exit the message and manually set the Matlab root. Also, if you want to compile a robot task which needs Matlab, you have to enable the Matlab interface (eg, BUILD_MATLAB_INTERFACE_barrett) or the task will not be compiled (you will not receive any compilation from CMake, though).

It is recommended to change CMAKE_C_COMPILER and CMAKE_CXX_COMPILER to clang and clang++ by toggling the advanced preferences (t)
You are then ready to build
make install
You can also build a single robot from its folder
cd sl/build/barrett
make install
Finally, to run a robot
cd sl/build/barrett
./xbarrett

Optional Compillation and Running Steps

MATLAB 2013a on OSX

Do ccmake press t and add to C_FLAGS
-Dchar16_t=uint16_T

Remove Real Time Limits

sudo nano /etc/security/limits.conf
Add anywhere the following two lines
* - rtprio 99
* - memlock unlimited

stulp/sl