Senior design project in which we created autonomous vehicles using machine learning
The core of this project uses a powerful device called the NVIDIA Jetson Nano. The Jetson Nano is an incredible microcontroller that is capable of creating multiple neural networks using Nvidia's proprietary SDKs. With such large and passionate developer community, we were able to create a variation of the popular JetBot and modify it to our specific needs. Using Jupyter Notebooks, we were able to using machine learning in order to train various models for vechicle self autonomy. In time we were able to create a pair of 2 self autonomous vehicles capable of road following, collision detection, traffic light detection, and vehicle detection.
Please feel free to read our Project Report about how the models were implemented, how the data was compiled, and much more!
Due to the document sizing limitations of GitHub, the models can only be shared on a Google Drive, however the datasets for the final models are also shared in addition to the methodology for regression training and transfer learning.
In spirit of the amazing community, feel free to clone our repo and use our data to train your bots! All our data and notebooks are free to use.