futurium-suitcase-reinforcement-learning-2

Reinforcement Learning 2 as a suitcase exhibit for Futurium

This variant of the original exhibit splits the functionality over two screens. It's designed for two iPads with 2360-by-1640-pixel resolution at 264 ppi (1180x820 logical resolution). One in landscape orientation, the other in portrait.

Building

Requires Node.js (v18.19 or greater) and npm.

Run the following from the root directory of the project:

npm install
npm run build

This will create a dist directory with the compiled exhibit. This directory can be served by any web server.

Development

Run both

npm run watch:copy

and

npm run watch:compile

The first will copy any files changed in extras to dist (without requiring a full build), and the second will recompile the JavaScript and SASS files (it actually runs watch on the inner project, inside of dist).

Configuration

The config directory has several data definitions.

You can override any of them in the settings-exhbit.yml file in the root of the built project.

Running

Open station-1.html and station-2.html in two separate iPads.

Sentry

The app supports Sentry.

The index.html page can take the DSN from the sentry-dsn query string parameter.

It can also get the DSN from the app.sentry.dsn configuration key in the settings.yml file.

Credits

This adaptation was developed by Eric Londaits for Imaginary gGmbH.

Supported by Futurium.

License

Code licensed under the MIT License. See LICENSE for details.

Copyright 2025 Imaginary gGmbH.