Real-Time Freeway Crash Occurrence and Type Prediction Using Connected Vehicle Data via a Spatiotemporal BiLSTM-Transformer

Contribution

Uses connected-vehicle data and spatiotemporal sequence learning to study both freeway crash occurrence and crash type.

This study connects connected-vehicle observations with spatiotemporal sequence modeling, examining both whether a freeway crash occurs and the type of crash. It extends proactive safety assessment beyond a single occurrence outcome.