Show HN: Watch a neural net learn Clash Royale defense in the browser

Category: ai-ml

Tags: reinforcement-learning, game-engine, cpp, python-bindings, webassembly

Score: 7.3/10 (Innovation: 7, Technical: 9, Documentation: 8, Utility: 5)

A full reimplementation of Clash Royale's battle mechanics as a deterministic C++ engine, paired with a recurrent PPO agent, search-based teacher, and screen-scraping perception pipeline. The project is remarkable for its engine fidelity, speed (20,000× real time), and the inclusion of a browser-based WebAssembly demo that teaches a small neural net to defend in real time.

Target audience: AI researchers, reinforcement learning practitioners, game developers

Repository: https://itzik123.github.io/ClashRoyaleAi/lab/ · Python · MIT · 24 stars

View on Hacker News