Skip to content

Latest commit

 

History

History
79 lines (54 loc) · 2.52 KB

README.md

File metadata and controls

79 lines (54 loc) · 2.52 KB

Haxball ai

This project is a fork of the https://github.com/vkurenkov/haxball-chameleon project.

Decision strategy

Link to the decision strategy png

Contribution guidelines

Git commit summary template:

[cs7632] SUMMARY (Max 60 characters)

Discription (Max 90 characters in a line)
Testcase and results (if applicable)

Licenses

This work is distributed under dual - MIT and Apache 2.0 license. The original work from haxball-chameleon is licensed under MIT license. The rest of the source code is licensed under the Apache 2.0 license. Both the licenses are available in the repository.

Please continue reading for the original Repo's Readme

Haxball Chameleon

This repository contains an interface for interacting with the game and an implementation of behavioral clonning agent.

What is HaxBall anyway?

HaxBall is a realtime multiplayer game that plays like a mix between football and air-hockey, and it's a real blast (official about).

How to run my agents?

LINUX ONLY.

Make the server running.

  1. Add the following line to your hosts file (c:\windows\system32\drivers\etc\hosts on windows, /etc/hosts on linux)
127.0.0.1 inter.haxball.com
  1. Start the server
python run_server.py

Call up your agent.

  1. Edit run_bot.py script to instantiate your agent (an example is provided).

  2. Start interaction between the agent and the server. The server allows multiple connections, thus you should provide your connection channel (any integer you like, defaults to 0).

python run_bot.py --channel-id=0
  1. Open inter.haxball.com:8080/#channel_id in your browser. This is the original game with slight modifications to allow sending states and accepting the inputs from the bot.

How to parse replays?

  1. Run raw replays converter. It will preprocess raw replays, put them into a separate folder, and create a file with nickname mapping (nickname -> all replays).
python run_converter.py --path=replays/
  1. Use utilities provided in replay.py to parse converted replays.