Skip to content

Recreating Bill Seiler's 1985 version of Space War and training RL agents with Self-Play

License

Notifications You must be signed in to change notification settings

e-dong/space-war-rl

Repository files navigation

Space War RL

I will be remaking the 1985 version of Space War using Pygame and train deep reinforcement learning agents to battle it out. I will try to do everything from scratch as much as possible as a learning exercise!

Quickstart Demo

Run make prod game to startup the game in "production" mode.

If switching between dev/game, use the clean target to be safe.
e.g. make clean dev game if switching from prod to dev.

Development

Note: I only tested this in x86_64 linux, so there might be linux specific dependencies in the requirements

First time setup

  1. Install python on your system. At the time of writing this, python version 3.11.5 was used.
  2. Setup virtual env and dependencies with make dev

Starting the game

Start the game with make, game is the default target, so you dont need to specify make game every time

You can also run make dev game everytime, the requirements will installed when there is a change either requirement files.