Smarty Ratty
An AI-powered simulation where a virtual mouse learns to solve mazes through trial-and-error and environment feedback.
Python
Pygame
NumPy
How it was implemented
Built in Python using Pygame for rendering. The agent learns via simple reward signals while exploring the maze; states are encoded as grid positions with collision checks and step penalties.
What I learned
- Building frame-based game loops and rendering with Pygame.
- Using NumPy to manage state arrays and efficient numeric operations.
- Applying Q-learning concepts to iteratively improve agent behavior.