Enhancing social acceptance of autonomous vehicles : a data-driven approach for modeling lane change response

Abstract

This study aims to improve the social acceptance of Autonomous Vehicles (AVs) by enabling them to respond to lane-changing events in a manner that mimics human behavior, making their actions more predictable to human drivers. Using the Waymo Open Dataset, the research focuses on the impact of initial conditions—such as the relative gap and velocity at the time of lane intrusion—on the velocity response of an approaching vehicle.

Despite extensive development of data-driven models for AVs, the influence of these initial conditions on maneuver execution has been underexplored. This study applies a time series analysis approach using ALKS guidelines to examine the correlation between initial conditions and the intensity of velocity changes in approaching vehicles. The findings reveal no clear trend between the initial conditions and the maneuvering behavior of the approaching vehicle, suggesting that human responses are influenced by factors beyond initial lane change conditions. This study highlights the complexity of human driving behavior and the need for comprehensive models to ultimately increase social acceptance of AVs.