Modeling and Analysis of Lateral Driver Behavior in Lane-Changing Execution

Existing studies regarding lane-changing behavior mainly focus on modeling lane-changing decisions. Lane-changing execution (LCE), which happens just after the adjacent gap has been accepted by the driver who wishes to change lanes, has not been much studied. This is especially true of the lateral movement aspect of a lane change, even though the LCE has considerable influence on driving safety and traffic simulation results and is also an indispensable part of the control design of an automatic vehicle. This paper proposes the first model for LCE in the lateral direction on the basis of empirical analysis using NGSIM data. After analysis of the data, two types of LCE, the continuous LCE (CLCE) and the discontinuous LCE (DLCE), were identified. A CLCE happens when the vehicles surrounding the lane-changing vehicle do not constrain the LCE behavior, and the entire LCE process is continuous. In contrast, a DLCE occurs when the lane-changing vehicle has to defer the lateral movement for a short time to make sure that changing lanes is safe. A model is proposed for each of the two types of LCE and then calibrated and validated. The results show that the proposed models can replicate real lane-changing execution behavior in the lateral direction with small errors.