Abstract
In this poster, we present the construction of HGDAVE (Heterogeneous Generative Dataset for Unmanned Autonomous Systems), a new dataset for Connected and Autonomous Vehicles (CAVs) and Unmanned Aerial Vehicles (UAVs), namely Unmanned Autonomous Systems (UASes). The dataset will be used to train artificial intelligence (AI) models to detect cybersecurity and safety-related risks, malfunctions, and crashes. The dataset was collected from three sources: 1) script-generated flying or driving missions, 2) software fuzzer-generated crashes instances, and 3) cybersecurity exploits generated by ethical hackers. To collect the data, we utilized the Digital Twin (DT) to replicate the behavior of UASes, which provides data that can be used to analyze, develop, and detect new anomaly detection algorithms.