loading page

Analysis of Time-to-Lane-Change-Initiation Using Realistic Driving Data
  • +2
  • Sarang Jokhio ,
  • Pierluigi Olleja ,
  • Jonas Bärgman ,
  • Fei Yan ,
  • Martin Baumann
Sarang Jokhio
Department of Human Factors

Corresponding Author:[email protected]

Author Profile
Pierluigi Olleja
Author Profile
Jonas Bärgman
Author Profile
Martin Baumann
Author Profile

Abstract

Lane changing is complex, yet it is one of the most common manoeuvres a driver performs in everyday driving. Before initiating a lane change, drivers should indicate their intention using a turn signal. When drivers are changing lanes, their use of a turn signal could provide meaningful insights about their behaviour. Therefore, we investigated the behaviour of drivers initiating a lane change using realistic data. In particular, time-to-lane-change-initiation (TTLCI) was investigated. TTCLI is the time from when a driver signals to when the driver initiates the lane change. We used the Kaplan-Meier method and mixed effect Cox Proportional Hazard (CPH) model to analyze the lane change data obtained from a large dataset collected as part of the L3Pilot project. The results indicate that most lane changes were initiated within 2 seconds of signalling. Furthermore, the results of the mixed-effect CPH model revealed that the speed of the lane-changing vehicle, the type and the direction of the lane change, and the presence of either a lead or lag vehicle and the lag gap significantly affected the TTLCI. The findings of this paper provide detailed insights into pre-lane change behaviour and pave the way for future studies. The findings of this study have important  implications for the development of regulations concerning the usage of turn signals by human drivers. Additionally, they have implications for the development of advanced algorithms for autonomous vehicles, particularly in improving the detection of lane changes of surrounding vehicles and estimation of the remaining time to it.