animatai

Animats

Reference code for:

A General Model for Learning and Decision-Making in Artificial Animals by Claes Strannegård, Nils Svangård, David Lindström, Joscha Bach and Bas Steunebrink

Submitted to IJCAI-17 AGA workshop, Melbourne, Australia

This repo is work in progress and will contain a completely refactored version of the original code. The repo ecosystem contains the old version of the framework (which isn't maintained anymore).

Setup

The easiest way to install animatai is with pip install animatai

At least Python 3.5 is needed since async is used in the web sockets server (wsserver.py).

Run the program

Examples using the ecosystem classes are available in the repo examples

A little setup is needed first:

• Create config.py. Start with copying config.py.template and try some of the examples from the repo mentioned above.

Start a web server and a browser:

• Run the server: python wsserver.py
• Run index.html in a browser and follow the instructions.

Development

First setup a development environment. I'm using virtualenv:

• First init virtualenv for Python3: virtualenv -p python3.6 venv3 (virutalenv needs to be installed)
• Activate virtualenv: source venv3/bin/activate
• Install the necessary Python packages: pip install -r requirements.txt. Add --no-compile when running on ubuntu.

Use Google Style Guide and make sure that the unit tests are maintained.

Build (lint and run unit tests) with: ./build.sh Building will also Create a source distribution in the dist folder.

Upload the build to the public package repo for installation with pip: twine upload dist/animatai-X.Y.Z.tar.gz

Credits

Using some classes from the AIMA book