Model-Based Prediction of Operation Consequences When Driving a Car to Compensate for a Partially Restricted Visual Field by A-Pillars.

Sayako Ueda,Takatsune Kumada,Toshihisa Sato

doi:10.3389/fnhum.2021.697295

Abstract

The partial restriction of a driver’s visual field by the physical structure of the car (e.g., the A-pillar) can lead to unsafe situations where steering performance is degraded. Drivers require both environmental information and visual feedback regarding operation consequences. When driving with a partially restricted visual field, and thus restricted visual feedback, drivers may predict operation consequences using a previously acquired internal model of a car. To investigate this hypothesis, we conducted a tracking and driving task in which visual information was restricted to varying degrees. In the tracking task, participants tracked a moving target on a computer screen with visible and invisible cursors. In the driving task, they drove a real car with or without the ability to see the distant parts of a visual field. Consequently, we found that the decrease in tracking performance induced by visual feedback restriction predicted the decrease in steering smoothness induced by visual field restriction, suggesting that model-based prediction was used in both tasks. These findings indicate that laboratory-based task performance can be used to identify drivers with low model-based prediction ability whose driving behavior is less optimal in restricted vision scenarios, even before they obtain a driver’s license. However, further studies are required to examine the underlying neural mechanisms and to establish the generalizability of these findings to more realistic settings.

Highlights

IntroductionDriving a car is primarily a visual task (Sivak, 1996), the physical structure of the car, such as the A-pillar between the front windscreen and the front side windows, can cause visual restrictions that have been reported to potentially contribute to road accidents, where a driver fails to notice other road users before it is too late to take appropriate action (Road Research Laboratory, 1963; Porter and Stern, 1986; Chong and Triggs, 1989; Quigley et al, 2001; Wade and Hammond, 2002; Millington et al, 2006; Reed, 2008; Marshall et al, 2012; Ekroll et al, 2021)
These findings indicate that laboratory-based task performance can be used to identify drivers with low model-based prediction ability whose driving behavior is less optimal in restricted vision scenarios, even before they obtain a driver’s license
We investigated the relationship between the decrease in task performance induced by the restriction of visual feedback regarding tool operation consequences and the decrease in steering performance induced by the partial restriction of the driver’s forward field of view

Summary

Introduction

Driving a car is primarily a visual task (Sivak, 1996), the physical structure of the car, such as the A-pillar between the front windscreen and the front side windows, can cause visual restrictions that have been reported to potentially contribute to road accidents, where a driver fails to notice other road users before it is too late to take appropriate action (Road Research Laboratory, 1963; Porter and Stern, 1986; Chong and Triggs, 1989; Quigley et al, 2001; Wade and Hammond, 2002; Millington et al, 2006; Reed, 2008; Marshall et al, 2012; Ekroll et al, 2021). Model-Based Prediction When Driving of literature shows that the partial obscuration of a driver’s forward field of view degrades their steering performance by preventing them from acquiring essential visual information during steering (Godthelp, 1985; Land and Horwood, 1995; Hildreth et al, 2000; Wallis et al, 2002, 2007; Frissen and Mars, 2014). Many efforts to reduce road accidents have focused on developing technologies such as ultrasonic sensors and cameras to detect visually occluded information, and present corresponding visual, auditory, or tactile signals to the driver (e.g., Advanced Driver Assistance Systems). Identification of the cognitive processes involved in visually restricted driving is a central requirement for the effective design of driver assistance systems

Objectives

Methods

Results

Discussion

Conclusion