nullvoxpopuli/doctor-who-thirteen-game-ai

JavaScript AI player for the original 2048 game (http://gabrielecirulli.github.io/2048/) that runs from the address bar

A.I. player for the Doctor Who "Thirteen" game.

This is JavaScript AI player for a variant of the game "2048".

Play the game "Thirteen" here: https://doctorwho.tv/games/thirteen/

How to use the A.I.

1. In your browser's bookmark bar, right click

2. Click "Add Page"

3. Paste the following content into the URL field

   javascript:(function(){let src='https://raw.githubusercontent.com/NullVoxPopuli/doctor-who-thirteen-game-ai/master/ai.js';async function fetchAndInsertScript(){let response=await fetch(src);let script=await response.text();let element=document.createElement('script');element.innerHTML=script;document.body.appendChild(element);}fetchAndInsertScript();})();

4. Click "Save"

5. Visit https://asset-manager.bbcchannels.com/m/2fzi3/

6. Click the bookmark created in step 4

7. Watch how for it gets

Development

A.I.

1. clone the repo
2. cd docker-who-ai
3. yarn && yarn start
4. This starts a development environment with a local copy of the game

Bookmarklet

1. clone the repo
2. change the bookmarklet URL to your fork/branch
3. copy contents of bookmarklet.js to https://ted.mielczarek.org/code/mozilla/bookmarklet.html
4. copy "Your Bookmarklet" url to your bookmark

Resources

• 2048 NN A.I. - source
• Browser A.I. Library Comparisons
• CovNetJS - not used here, but has good explanations and demos

doctor-who-thirteen-game-ai's Issues

Resources to read through

https://medium.com/@pierrerouhard/reinforcement-learning-in-the-browser-an-introduction-to-tensorflow-js-9a02b143c099

https://github.com/nrox/q-learning.js/blob/master/q-learning.js

Add to readme
https://towardsdatascience.com/qrash-course-deep-q-networks-from-the-ground-up-1bbda41d3677

Use simpler UI

Like here: https://aj-r.github.io/2048-AI/

Penalize branches for the number of moves they take to achieve a score

Prioritizes merging bigger tiles together

Publish to GH Pages

Initially start with A* on every empty space to account for random tile spawn