Amorphous Fortress Online: Collaboratively Designing Open-Ended Multi-Agent AI and Game Environments
M Charity, Mayu Wilson, Steven Lee, Dipika Rajesh, Sam Earle, Julian Togelius
TL;DR
Amorphous Fortress Online addresses the challenge of open-ended, generalizable AI in multi-agent environments by delivering a web-based platform where users design, remix, and visualize tiny FSM-driven agents within ASCII fortresses. The approach combines a JavaScript-engine port, modular editors (entity, fortress, text), an interactive play view, and an X-Ray FSM visualization to support emergent behavior exploration and debugging, augmented by a backpack system for cross-fortress collaboration. Key contributions include the FSM-based fortress engine, browser-based tooling, and a roadmap toward mixed-initiative design, entity recommendation, and sprite-generation to cultivate a large, user-generated dataset for future AI research. The work offers practical impact by enabling community-driven AI environment creation, facilitating reinforcement learning training data, and supporting generative modeling research through open-access fortress definitions and interactions.
Abstract
This work introduces Amorphous Fortress Online -- a web-based platform where users can design petri-dish-like environments and games consisting of multi-agent AI characters. Users can play, create, and share artificial life and game environments made up of microscopic but transparent finite-state machine agents that interact with each other. The website features multiple interactive editors and accessible settings to view the multi-agent interactions directly from the browser. This system serves to provide a database of thematically diverse AI and game environments that use the emergent behaviors of simple AI agents.
