Model of collision avoidance with lane departure warning in real-world departure collisions with fixed roadside objects

This study developed and presented a computational model of road departure collisions which used data collected from real-world crashes as the basis for simulations of collision avoidance for vehicles equipped with Lane Departure Warning (LDW) systems. Real-world collisions were extracted from a database of 890 serious road departure crashes with detailed reconstructions of the vehicle trajectory and impact locations. The model was developed using the simulation software PreScan (TASS Americas) and simulated driver steering maneuvers following a road departure and LDW. We used the model to simulate a subset of road departure collisions with objects close to the road and with little driver steering after the road departure. In our dataset, these crashes corresponded to approximately 26% of departures with available data and lane markings (47 cases). Between 3% and 5% of hazards were avoided with a LDW delivered at lane crossing. If the LDW was delivered 1 s prior to departure, between 19% and 34% of crashes were avoided. This model is a first step in a larger LDW benefits estimates study. Future models should include simulations of drivers who were steering prior to the crash without LDW, which were excluded from the current study.