The time course of a lane change: Driver control and eye-movement behavior

https://doi.org/10.1016/S1369-8478(02)00011-6Get rights and content

Abstract

In this paper we explore the time course of a lane change in terms of the driver's control and eye-movement behavior. We conducted an experiment in which drivers navigated a simulated multi-lane highway environment in a fixed-base, medium-fidelity driving simulator. We then segmented the driver data into standardized units of time to facilitate an analysis of behavior before, during, and after a lane change. Results of this analysis showed that (1) drivers produced the expected sine-wave steering pattern except for a longer and flatter second peak as they straightened the vehicle; (2) drivers decelerated slightly before a pass lane change, accelerated soon after the lane change, and maintained the higher speed up until the onset of the return lane change; (3) drivers had their turn signals on only 50% of the time at lane-change onset, reaching a 90% rate only 1.5–2 s after onset; (4) drivers shifted their primary visual focus from the start lane to the destination lane immediately after the onset of the lane change. These results will serve as the basis for future development of a new integrated model of driver behavior.

Introduction

Driving is a highly complex task that requires continual integration of perception, cognition, and motor response. Of the various subtasks that comprise the entire driving task, lane changing is one subtask that incorporates many of the critical aspects of driving, such as lower-level control (e.g., steering, acceleration), monitoring (i.e., maintaining situation awareness), and decision making (e.g., when to change lanes). Despite the vast attention given to the driving task in general, much less attention has been directed to lane changing, despite its ubiquity in common driving environments––for instance, highway driving, which accounts for roughly 70% of vehicle miles on American roadways (Federal Highway Administration, 1998). Existing work on lane-changing behavior in large part emphasizes the decision-making aspects of the task, particularly gap acceptance and the decision of when to change lanes (e.g., Ahmed, Ben-Akiva, Koutsopoulos, & Mishalani, 1996; Gipps, 1986). While other studies have addressed different aspects of lane changing, from behavioral aspects such as typical durations (e.g., Finnegan & Green, 1990) to practical development of lane-change collision warning systems (Talmadge, Chu, & Riney, 2000), we have yet to form a complete picture of when and how drivers change lanes including all the various perceptual and motor processes involved.

This paper attempts to flesh out this picture of lane-changing behavior by analyzing the time course of a lane change across several modalities. The paper expands on our recent study of eye movements during lane changing (Salvucci, Liu, & Boer, in press) and examines the integration of steering, throttle, turn signals, and eye movements for left-to-right and right-to-left lane changes. By aggregating and displaying these multi-modal together as they occur before, during, and after a lane change, the analysis elucidates several interesting and surprising properties of drivers' lane-changing behavior. These results form a solid foundation on which to develop and validate a rigorous integrated model of driver behavior.

Section snippets

Method

This study of lane-changing behavior is based on data collected for a previous study (Salvucci et al., in press) that focused exclusively on eye-movement data. The experimental task involved driving on a multi-lane highway with traffic in a fixed-base driving simulator. For clarity, we briefly review specific details of the data collection below. We also describe the data processing into time course graphs as needed for the multi-modal time-course analysis in the next section.

Results

Our analysis of the lane-changing data centers around a time course analysis in which we visualize aggregated data sequences from different modalities before, during, and after a lane change. Fig. 1, Fig. 2 show the time-course graphs for right-to-left and left-to-right lane changes, respectively. The remainder of this section explains how these graphs were created and what they indicate about the nature of drivers' lane-changing behavior.

The time-course graphs were created in several stages.

Summary

To summarize the results above, we noticed several aspects of the time course of driver lane changes that stood out in our analysis:

  • 1.

    Drivers exhibited the expected sine-wave steering pattern except for a longer and flatter second peak as they straightened the vehicle. In addition, drivers exhibited a slight tendency to steer away from the destination lane 2–3 s before the onset of the lane change.

  • 2.

    Drivers decelerated slightly before making a pass lane change, accelerated soon after the lane

Discussion

In addition to augmenting our general knowledge about lane-changing behavior, we are in the process of using these data to develop and validate an integrated model of driver behavior (Salvucci, Boer, & Liu, 2001; for similar models, see also Aasman, 1995; Levison & Cramer, 1995). Our initial efforts have attempted to formalize a simple control mechanism that enables basic steering, lane changing, and curve negotiation in as straightforward a way as possible. To this end, we have constructed an

Acknowledgements

We thank Geoff Underwood and two anonymous reviewers for helpful comments on earlier drafts of this paper. Portions of this work were done at Nissan Cambridge Basic Research in Cambridge, MA.

References (16)

There are more references available in the full text version of this article.

Cited by (0)

View full text