Analysis of cut-in behavior based on naturalistic driving data

Abstract

Cut-in maneuvers, when vehicles change lane and move closely in front of a vehicle in the adjacent lane, are very common but adversely affect roadway capacity and traffic safety. Yet little research has comprehensively explored cut-in behavior, particularly in China, which has a challenging driving environment and is often used for connected and autonomous vehicle testing. This study developed an extraction algorithm to retrieve 5608 cut-in events from the Shanghai Naturalistic Driving Study. The data were used to identify cut-in characteristics, including motivation, turn signal usage, duration, urgency, and impact. Results showed that almost half of drivers did not use a turn signal when cutting in, and that cut-ins had a shorter time to collision (TTC) than other lane changing. A lognormal distribution was found to produce the best fit for cut-in duration, which varied from 0.7 s to 12.4 s. As characteristics were found to vary by roadway type and motivation, multilevel mixed-effects linear models were developed to examine the influencing factors of cut-in gap acceptance. Acceptance of lead and lag gaps was significantly affected by environmental variables, vehicle type, and kinematic parameters, which has important implications for microsimulation, as does the large variance in duration that makes specifying duration essential when setting scenarios. Improvement in safety education is warranted by the high degrees of risk and aggression shown by TTC and turn signal usage; but the ability of drivers, who needed to yield to the cut-in, to predict danger and adopt safe, suitable, and timely strategies suggests that advanced driver assistance systems and connected and autonomous vehicles can learn similar responses.