Enhancing Road Safety by Anomalous Driving Behavior Detection Based on V2X Messages

Recent advancements in smart road technologies are addressing the global issue of road safety to reach the "vision zero" long-term goal of the European Union, moving as close as possible to zero fatalities in road transport by 2050. This research aims to detect anomalous driving behaviors by extracting vehicle data from cooperative awareness messages (CAMs) of the vehicle-to-everything (V2X) communication protocols currently adopted in a smart road segment of an Italian highway. Our approach captures various types of driver-vehicle interactions by specialized models that assess whether a vehicle exhibits anomalous dynamics during a specific time window. The proposed anomaly detection framework is designed to be both robust and interpretable, and its effectiveness is demonstrated through evaluations on both simulated and real-world datasets.