tawnkramer/sdsandbox
A Unity-based simulator for training self-driving cars using end-to-end neural network learning.
Not currently ranked — collecting fresh signals.
star history
This project provides a sandbox environment for training self-driving car agents using the Nvidia PilotNet neural network architecture. It combines Unity for realistic physics simulation with Python, Keras, and TensorFlow for model training. Users generate image-steering training pairs by driving in the simulator, then train a convolutional neural network to predict steering angles from camera input. The trained model can be loaded back into the simulator to autonomously control the vehicle.
Frequently asked
- What is tawnkramer/sdsandbox?
- A Unity-based simulator for training self-driving cars using end-to-end neural network learning.
- Is sdsandbox open source?
- Yes — tawnkramer/sdsandbox is open source, released under the BSD-3-Clause license.
- What language is sdsandbox written in?
- tawnkramer/sdsandbox is primarily written in C#.
- How popular is sdsandbox?
- tawnkramer/sdsandbox has 432 stars on GitHub.
- Where can I find sdsandbox?
- tawnkramer/sdsandbox is on GitHub at https://github.com/tawnkramer/sdsandbox.