Demos for the ORPHEUS library

Each demo has its own folder, but the Demos repository is now treated as a whole. Please find instructions below on how to compile all demos. The build system does not use special environment variables or anything that might interfere with other software, other than installed dependencies. More demos/tutorials will be added in due time.

Setup instructions for Ubuntu (tested with 12.04 and 14.04).

1) Installing dependencies

sudo add-apt-repository "deb $(lsb_release -sc) universe"
sudo apt-get update
sudo apt-get install git g++ cmake automake libtool libeigen3-dev libboost-system-dev libboost-thread-dev freeglut3-dev

Note: these commands rely on apt's automatic detection of other dependencies while trying to install the mentioned libraries. Therefore, the dependency installation step might vary for other distributions.

2) Downloading the repository

mkdir ORPHEUS.Demos
cd ORPHEUS.Demos
git clone .

3) Preparing for the build

The following script will automatically download and compile: 1) the ORPHEUS library and 2) a custom version of Bullet Physics which is tweaked to compile with the -fPIC flag (it is an archive uploaded on dropbox)


4) Building the demos

The following script will build the demos, assuming that previous steps have succeeded:


5) Running the demos

Demos are built in the bin folder. To run one of them, simply navigate to the folder corresponding to each demo, in the bin folder, and launch the executable.

Orphy the Cat -> OrphyTheCat

Orphy is the first main test of the ORPHEUS library. It focuses on the functionality of mental simulations.

Results of this demo will appear in:

M. Polceanu, M. Parenthoen and C. Buche, "ORPHEUS: Mental Simulation as Support for Decision-Making in a Virtual Agent"
In the 28th International Florida Artificial Intelligence Research Society Conference (FLAIRS'15), AAAI Press, 2015.

KRLS demo -> DemoKRLS

DemoKRLS uses a simplified environment that resembles that in the OrphyTheCat demo. The goal for the cat is to learn to reenact the bird's trajectory, in its mental simulations.

The algorithm used is the Kernel Recursive Least Squares by Yaakov Engel, using the dlib ( implementation. The aim of this example is to predict cyclical behaviour.

Prediction error is outputted at the console. Best used with mental simulation visualisation interface (will be made public soon).