Self Driving Car & Object Detection

Merging image segmentation with Unity ML-Agents for autonomous driving in a 3D simulated environment

An experiment in merging image segmentation with Unity ML-Agents to create a self-driving car that understands obstacles in real time inside a 3D simulated environment.

This project consists of two components:

  1. Udacity Self Driving Car in Unity — 3D simulation environment for training the autonomous agent
  2. CamVid Image Segmentation — Object detection for identifying obstacles and road boundaries

The goal is to replicate a real driver experience where the car perceives its environment via segmented camera input and navigates autonomously.

Tech Stack

  • Unity + ML-Agents
  • PyTorch
  • Python
  • Image Segmentation