Evolving Zoo is an AI-powered project where I trained popular games like Flappy Bird and the Chrome Dinosaur game to play on their own using the NEAT (NeuroEvolution of Augmenting Topologies) algorithm. The goal was to evolve the game agents through neural networks, enabling them to improve autonomously through gameplay.
Key Features
- Flappy Bird: The bird learns to navigate pipes on its own, improving over time through evolutionary strategies.
- Chrome Dinosaur: The dinosaur evolves to dodge obstacles and survive for longer durations in the offline Chrome game.
Tech Stack
Frontend:
- Pygame: Used for rendering the game visuals and simulating game environments.
Backend:
- Python: Core programming language for implementing the NEAT algorithm.
- NEAT-Python: Library to implement neuroevolution, training the game agents through evolutionary algorithms.
Algorithm Details
- NEAT Algorithm:
- Uses evolutionary techniques to evolve neural networks over generations.
- Adjusts network topology dynamically, enhancing performance through crossover and mutation.
- Helps the agents learn complex behavior (e.g., jumping at the right time in Flappy Bird or avoiding cacti in the Dinosaur game).
Demo
Ahh and one more thing!! I replaced the flappy bird with an INDIAN SUPERMAN..
Let me show you the superman first…
I just hope you’re not one of those super serious HR folks who’d reject me because of these things. But anyways I would also not like to work for you if your company’s environment doesn’t allow a little humor to flourish.
Video Demo: Flappy Bird AI
Click here to watch the Flappy Bird AI demo on Google Drive
But I didn’t stop there! This idea was initially inspired by a YouTube video, and I wanted to see if I could apply the same concept to other projects. So, I took on the Chrome Dino game and gave it a twist.
Introducing: Dino Venkateshwar
This time, I decided to add some custom game art to make things more interesting. So, meet Dino Venkateshwar!
Video Demo: Dino AI
Watch Dino Venkateshwar play the game all by himself!
This project showcases how neuroevolution can be applied to teach AI agents to play simple games autonomously, demonstrating the power of algorithms like NEAT in evolving intelligent behavior.