MadKit Quick start

Jacques Ferber, Olivier Gutkecht, Fabien Michel

Version 4.1

Abstract

This document describes how to quicky develop agents in the MadKit platform.

Note: This document is not finished. Comments and unfinished parts are surrounded by double brackets. For instance {{link}} means that there is a missing link here.

First step

Download MadKit at www.madkit.org following the instructions on the site. We will call the 'madkit directory', the directory (or folder) in which you have installed MadKit. The 'madkit/bin' (or 'madkit\bin') directory is the 'bin' directory located inside madkit.

Start MadKit, using one of the following solutions:

  • Windows user: in the madkit\bin directory, double click on the madkit icon (madkit.exe program), or on the desktop.bat dos program.
  • Unix/Linux user: from the shell go to the madkit/bin directory, type chmod +x *.sh (to make all .sh executable) and ./desktop.sh
  • Mac OS X user: go to the madkit/bin directory and click on the madkit icon (madkit.app program), or follow the steps of the Unix user.

Note: it is possible to start madkit using different techniques, see the userguide for more information.

The MadKit desktop will appear. The first time may take some time because it needs to cache some information. You will see a few lines printed in the output area of MadKit. To know how to use the desktop, please refer to the desktop section of the userguide.

To create your first program in MadKit, double click on the designer agent, which is represented by the following icon on the desktop.

MadKit (Multi-Agent Development Kit) is a versatile agent platform. Among its features:

  • Cross-platform. MadKit works on Win xx, Linux, Mac Os, Other Unix based OS, etc..
  • There is no prerequisite on agent model. Users may choose or develop the agent architecture of their choice.
  • Can support many simultaneous communication models : communication using message passing, broadcast, etc...
  • Transparent distributed mode, using the Communicator or the new NetCom agent
  • Can host multiple, heterogeneous agent applications (from BDI-like agents to artificial life applications).
  • Componential and flexible graphical agent interfaces
  • May be used for designing any multiagent applications, from distributed applications to multiagent simulation.

Motivation

A major characteristic in agent research and applications is the high heterogeneity of the field: both agent model heterogeneity, characterizing agents described and built with different models and formalisms; language heterogeneity, with agents using different communication and interaction schemes, and finally applicative heterogeneity, as multi-agent systems are used with various goals and in many applicative domains.