GitXplorerGitXplorer
M

game-ai-lib-test-app

public
0 stars
0 forks
0 issues

Commits

List of commits on branch main.
Unverified
b11c1e8fdc352f71cbf921b30d0547b0e9173895

Init

MMatPoliquin committed 6 months ago
Verified
3a532b2552f561a261c8839df99fefbc9b8447a2

Initial commit

MMatPoliquin committed 6 months ago

README

The README file for this repository.

stable-retro play

Emulator frontend that supports Machine Learning models to control players. The source code is based on OpenAI's gym-retro-integration tool.

This is a prototype version so only NHL94 (1 on 1 Sega Genesis version) is supported for now.

Building the app

sudo apt update
sudo apt install git cmake unzip libqt5opengl5-dev qtbase5-dev zlib1g-dev python3 python3-pip build-essential

git clone https://github.com/MatPoliquin/stable-retro-scripts.git

Generate makefiles and compile

cd app
cmake . -DBUILD_UI=ON -UPYLIB_DIRECTORY ..
make

Example

Launch the app

./sr-play

Load NHL94 1on1

  • On the top right menu, click: Game->Load Game
  • Open your rom file in app/retro/data/NHL941on1-Genesis

Load 2 player state

  • On the top right menu, click: Game->Load State
  • Open the following state: app/retro/data/NHL941on1-Genesis/PenguinsVsSenators.2P.state

You should now be able to play againsts the custom AI. You control player 2 and the AI controls player 1. You might need to config the inputs using Windows->Controls menu