• Florent Latombe <florent@bokbox.com>
• Antoine Pierlot-Garcin <antoine@bokbox.com>

We're both students at INP-ENSEEIHT, Toulouse, France.

We decided to release these sources as an example of code in the Artificial Intelligence field which uses the Pyke knowledge engine feature. Hopefully, this can help other people working with Pyke and who lack some example code.

This project was done as an assignment for computer science engineering students in last year of the standard 5-year French engineering cycle. It is NOT fully functionnal or working. The basics work, but there is room for improvement and we haven't had it run on "big" situations.

For this project, we had to build a multi-agents system. Agents communicate and each have their own knowledge base. They evolve in a universe, and can execute the following actions:

• Move from one place to another
• Take an object on the floor
• Put an object down
• Put an object in a box (which is also an object)
• Fetch an object from a box
• Give an object or a box to another agent
• Ask another agent for information
• Give some information to another agent

Official site : http://pyke.sourceforge.net

This project is released under the DWTFYW Public License.

• Required packages :

  • Python 2.6 or 2.7
  • Pyke 1.1 (probably works with newer versions)

• In the Makefile, replace PYTHON=python with PYTHON=python2 if

  python 3 is installed (in Arch Linux for example).

• Pyke environment :

  Pyke must be installed. You can find it on the sourceforge page, at the following address : http://sourceforge.net/projects/pyke/

• Run the tests :

  A series of basic tests is available in src/tests and is called through the make command. It is made of some basic situations to test the basic functionalities of the program.

• Run a simulation :

  Launching a given simulation is done in simulation.py file. The situation has to be entered in it and then executed with python, for example python simulation.py.