Dino Reinforcement Learning

This project explores using reinforcement learning to train an agent to play the chrome Dino Game

Single-Threaded Version

The single-threaded implementation achieved the following results:

• At 100k iterations: It was capable of recognizing the cactuses and jumping. Highscore: 450
• At 250k iterations: It was capable of going until stopped. Highscore: 2025

Performance and Runtime

Training was conducted on my desktop with an RTX 3070 Ti. The training time was approximately 10 hours per 100k iterations.

How to Run This Yourself

Frankly, I doubt anyone cares enough to do so, but I will include this nonetheless.

Requirements

• Python: 3.6.0 - 3.9.0
• Stable-Baselines3: 1.5.0
• Pip: 21.0 (or newer)
• Gym: 0.21.0
• Wheel: 0.38.0
• Pillow: 10.3.0
• Setuptools: 65.5.0
• opencv: 4.10.84
• pytesseract: 0.3.10

(All packages can be installed via pip)