JetRacer Road Following Project

This project, completed at Deggendorf Institute of Technology in SS24, showcases a high-performance AI-based racing car developed with the JetRacer Pro AI Kit. The goal is to use computer vision and deep learning to train a model that autonomously navigates a track as quickly as possible. This project involves data collection, dataset labeling, model training, hardware setup, and testing on a real-world racetrack.

I received a top grade (1.0) for this project and successfully passed the course.

Project Overview

This project includes:

1. Dataset Collection and Labeling:

   • Recording video from the car's camera in different positions to recreate different situations on the road that the car can see.
   • Splitting the video into frames & labeling the frames.

2. Model Training (road_following_learning.ipynb):

   • Training a road-following model in Google Colab.
   • This notebook includes data preprocessing, model architecture setup, training, and evaluation.

3. JetRacer Hardware Setup:

   • Composing the Car(pretty much like Lego)
   • Installing Linux system & setting up the environment

4. Testing on Track (racing.ipynb):

   • Running the trained model on the JetRacer car.
   • Optimizing performance for smooth and fast navigation around a circular track.

Files Included

• road_following_learning.ipynb - This notebook covers the training process in Google Colab.
• racing.ipynb - This notebook runs the trained model on the JetRacer car for testing and racing.

Results

After training and testing, the JetRacer car successfully navigated the track autonomously, demonstrating reliable and fast performance.

Usage

Running the Training Notebook

1. Clone the repository:

   git clone https://github.com/LittleErny/JetRacer
   cd JetRacer

2. Open road_following_learning.ipynb in Google Colab (or another Jupyter environment).

   • Follow the steps in the notebook to set up the environment, load data, and train the model.

3. Export the Trained Model:

   • Save the trained model and transfer it to the JetRacer for real-world testing.

Running the JetRacer

1. Transfer racing.ipynb to the JetRacer device.
2. Open racing.ipynb on the JetRacer and run.

Demo

To add the GIF in your README:

Acknowledgments

Special thanks to Deggendorf Institute of Technology for support and resources<3