Investigating gap acceptance behavior based on correlated random parameter survival model with heterogeneity in means

This paper aims to investigate gap acceptance behavior during discretionary lane-changing maneuvers considering the heterogeneity among drivers and lane-changing urgency on a freeway segment from a microscopic perspective. It was found that there were statistically significant differences in gap acceptance behavior between changing lane to the left/right. The accelerated failure time(AFT) model, random parameter AFT model, and correlated random parameters with heterogeneity in means AFT model were established. The results show that the third model have the best goodness of fit. There is obvious mean heterogeneity in the gap acceptance models of different lane change directions, and the correlation between random parameters significantly affects the gap acceptance behavior. The urgency of lane change has a significant heterogeneous effect on the lagging gap acceptance behavior of lane-changing drivers and is significantly correlated with other random parameters. The study results can help improve the safety lane-changing modules for connected and autonomous vehicles.