/commuters

A reinforcement learning, agent based simulation

Primary LanguageJavaGNU General Public License v3.0GPL-3.0

Commuters

A reinforcement learning, agent based simulation

This is an application I realized in 2010 to simulate a group of commuters, which learn their way through Reinforcement Learning.

To try it, you will find ready-made, stand-alone Mac, Linux and Windows app in the Release section.

Ignorant agents

Agents at work

Agents going home

You can define your own environments (the agents will adapt) by changing the file commuters/map.txt. You can find more examples in the maps/ directory.

A different map

If you instead wish to build it from source, you can do so with ivy and ant, like this:

cd commuters
ant ivy-setupjars
ant
. ./setcp.sh
java visualizzazione.Pendolari