CyberX: Race

CyberX-Race

What happens when you teach an AI how to race — and then just let it go?
Welcome to CyberX-Race, where an autonomous driver learns the track all on its own, one crash (and recovery) at a time.

What It’s All About

CyberX-Race is a racing experiment built in Unity using machine learning.
Instead of you taking the wheel, an AI agent trains itself to drive around a custom-made track — no scripts, no cheats, just pure learning.

Over time, it figures out how to:

• Stay on track

• Avoid smashing into walls

• Find the best racing lines

• And eventually… complete full laps without help

It’s like watching a baby driver evolve into a pro racer — powered by code and patience.

Features

• Custom racing track built from scratch

• Realistic car physics (accelerate, brake, drift!)

• AI trained with Unity’s ML-Agents (PPO algorithm)

• Reward system that teaches the AI how to race smarter

• On-screen visual tools that show how the AI learns

The Techy Bits

If you’re into the behind-the-scenes stuff:

• Made in Unity 2021

• Uses C# for game logic and Python for training

• Trained for about 2 months to go from chaos to consistency

What’s Next

• AI vs AI races

• Smarter behavior for overtaking and cornering

• Dynamic tracks and changing environments

Why It’s Cool

CyberX-Race isn’t about who’s the fastest — it’s about watching intelligence evolve.
Every lap is a little smarter than the last. Every crash teaches something new.
It’s racing… but with a mind of its own.