Evaluating Lane Departure Warning Systems Considering Characteristics of Truck Drivers’ Response Behaviors

Abstract

According to statistics from the National Highway Traffic Safety Administration (NHTSA) in the U.S., lane departures contribute to more than one-third (37.4%) of fatal crashes. Lane departure warning (LDW) systems have been widely installed and used, including in trucks. However, truck LDW false alarm rates have reached 33%–55%. Improper warning time is a key factor that affects LDW false alarms. Truck drivers’ response behavior affected the warning time of LDWs. Proposing an evaluation method that considers drivers’ response behavior is helpful to reduce false alarms for LDWs. This study extracted 777 lane departure events from ICARVISIONS’s truck LDW records of the China Pacific Property Insurance Company’s driving data. In-vehicle data collected during LDW activation were used to classify response behaviors by k-Shape clustering, which considered the changes over time in truck drivers’ response behaviors for six scenarios. Using the Responsibility-Sensitive Safety model to evaluate the LDWs, the parameters of safe lateral distance and safe longitudinal distance were calibrated for each scenario by the Non-Dominated Sorting Genetic Algorithm-II. The key findings include: 1) The difference in time-to-lane crossing under various scenarios needs to be considered in the truck LDW algorithm; 2) The safe longitudinal distance should not be ignored for truck LDWs. Analyzing driver response characteristics guides improvements in the adaptability of LDW algorithms. Evaluating the impact of safe longitudinal distance on false alarms confirms critical safety parameters in testing standards.