Examining lane change gap acceptance, duration and impact using naturalistic driving data

Analysis of lane change is important for microsimulation and safety improvement, and can also provide reference for advanced driver assistance systems (ADAS) and connected and autonomous vehicles (CAVs). Yet little research has comprehensively explored lane changing, particularly in China, a site of current CAV testing. This study developed an automatic extraction algorithm to retrieve 5,339 lane change events from the Shanghai Naturalistic Driving Study, and used the data to examine the core lane change components: gap acceptance, duration, and impact on the following vehicle (FV). Multilevel mixed-effects linear models were employed to develop relationships between gap acceptance and duration and the influencing factors; impact was then assessed using speed change rate, brake timestamping, and time-to-collision (TTC). Key results showed that 1) gap acceptance varied by roadway type and motivation, and lead and lag gaps were significantly affected by environmental variables, vehicle type, and kinematic parameters; 2) duration varied from 0.7 s to 16.1 s, significantly affected by variables similar to gap acceptance, but notably, not by motivation; 3) as many as 1 in 5 Chinese FV drivers responded to lane changes with acceleration exceeding 10%; 4) nearly half of FVs braked when they perceived a vehicle’s lane-change intention, and 90% braked before TTC reached 4.7 s; 5) in over 70% of lane changes, the minimum TTC occurred between the initiation and cross-lane points. In addition to advancing the international development of lane-change theory, one of this study’s important applications is that CAVs can be designed to brake during a safer TTC phase.