Autonomous Driving Simulation Dataset

This repository contains the dataset, simulation builds, and supporting scripts used for reproducibility of the Autonomous Driving Simulation (ADS) project. The dataset was collected in a Unity-based WebGL simulation environment and includes both raw image data and pre-processed training/test splits. In addition, we provide Jupyter notebooks for model training (vanilla and adversarial PGD), benchmark generation in VNN-LIB format, and a WebGL build of the simulation platform for interactive evaluation. All materials are provided anonymously for the purpose of peer review and are intended to support formal verification and reproducibility studies in autonomous driving research.