Remerciez-le!

Remerciez @Admin pour avoir partagé cet document gratuitement, de la manière la plus simple, en partageant sur les réseaux sociaux.

Evaluating Head-Up Displays across Windshield Locations Bethan Hannah Topliss U

Evaluating Head-Up Displays across Windshield Locations Bethan Hannah Topliss University of Nottingham Nottingham, UK bethan.topliss@not tingham.ac.uk Sanna M Pampel University of Nottingham Nottingham, UK sanna.pampel@not tingham.ac.uk Gary Burnett University of Nottingham Nottingham, UK gary.burnett@not tingham.ac.uk Joseph L Gabbard Virginia Tech, Blacksburg, Virginia, USA jgabbard@vt.edu ABSTRACT Full windshield displays (WSDs) have the potential to present imagery across the windshield. Current knowledge on display location has not investigated translucent displays at high eccentricities from the driver’s forward view. A simulator study (n=26) was conducted aiming to, (a) investigate the effects of Head-Up Display (HUD) location across the entire windshield on driving performance, and (b) better understand how the visual demand for a complex HUD imagery differs from that for a Head-Down Display (HDD). Lane-keeping was poorer when HUD imagery was furthest from the driver (and for the HDD compared to the HUD). Equally, counts of “unacceptable” driving behaviour were greater for displays furthest from the driver’s forward view. Furthermore, drivers preferred HUD imagery that was closer to them. The results indicate that HUD evaluations should account for image location, because of how driver gaze location can impact lateral driving performance. Author Keywords head-up display, evaluation, distraction CSC Concepts Human-centered computing~Displays and imagers INTRODUCTION Driving is a predominantly visual-manual task, such that the appropriate allocation of visual attention is fundamental to effective driving performance [5]. Thus, it is vital to assess how in-vehicle visual displays may impact the primary driving task, to ensure the display is not too visually distracting to the driver. Head-Up Displays (HUDs) present the driver with visual information on a translucent screen over the driver’s forward view of the road environment. Despite known issues, they offer the opportunity to improve driver interaction with visual information over Head-Down Displays (HDDS) such as those placed on the centre console [6]. Comparisons of HUDs and HDDs have largely demonstrated HUDs to be beneficial. There is commonly a preference for HUDs over HDDs [13] and faster response times to the tasks they present [24]. Importantly, HUDs regularly result in drivers responding faster to hazardous or urgent events [11] [12]. Similarly, Liu and Wen [20] found that commercial lorry drivers rated their mental stress as lower and responded faster to urgent events when using a HUD rather than HDD. Thus, HUDs are likely to continue increasing in popularity, since they have the potential to greatly reduce the limitations of HDD displays while maintain the flexibility of visual displays [9]. The benefits of HUDs are largely attributed to the eye-movement behaviours they encourage [9]. First, as Ablaßmeier et al. [1] demonstrated, HUDs result in a lower gaze retention period across all age groups. Furthermore, being presented over the forward road environment means that HUDs have the potential to reduce the transition time between the driver taking-in information from the display and then looking back to the road. Finally, the positioning of the driver’s focal attention (usually towards the forward road environment) enables drivers to better detect hazards compared to HDDs [12]. However, the potential for a full Windshield Displays (WSDs) is increasingly being investigated (e.g. [8] [3]). These displays enable HUD imagery to be presented anywhere across the full windshield of the vehicle. As a result, secondary task interfaces (which do not aid the primary task of driving) may be placed at increasing eccentricities away from the driver’s forward road view. Previous work on opaque displays, displaying secondary tasks, has clearly demonstrated screen location and eccentricity can have a marked impact on driving performance and situational awareness (e.g. [31] [27] [19]) with displays at greater eccentricities largely resulting in poorer performance. However, HUD imagery in future WSDs is likely to be positioned at different and more extreme eccentricities compared to these displays (e.g. high on the windscreen, to the left/right of the driver’s natural gaze) making this previous work often not applicable to Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page. Copyrights for components of this work owned by others than ACM must be honored. Abstracting with credit is permitted. To copy otherwise, or republish, to post on servers or to redistribute to lists, requires prior specific permission and/or a fee. Request permissions from Permissions@acm.org. AutomotiveUI '19, September 21–25, 2019, Utrecht, Netherlands © 2019 Association for Computing Machinery. ACM ISBN 978-1-4503-6884-1/19/09…$15.00 https://doi.org/10.1145/3342197.3344524 244 AutomotiveUI ’19, September 21–25, 2019, Utrecht, Netherlands Topliss et al. future display designs. Furthermore, HUD imagery varies greatly from the opaque displays used in these studies: it can be translucent and vary in luminance/contrast, which has the potential to delay reaction time [30] and thereby recognition. As a result, the findings of these studies are not readily applicable to future WSDs. This study aims to fulfil the need for this knowledge. Recent work has begun to address some of these points. Smith [23] investigated three potential positions of a HUD in a vertical arrangement in front of the driver. The author found higher locations resulted in better longitudinal vehicle control whilst the lowest HUD position resulted in less lane position deviation. However, the locations inspected only encompassed one axis (and a total of three positions) which were not at large eccentricities from the driver’s forward view. Equally, the tasks used were visual search tasks which allowed participants to glance between the task and the forward road view. Therefore, more work is needed to inspect how highly complex HUD tasks (which require long glances) impact drivers in order to appreciate a “worst-case scenario”. Thus, the tasks used in this work required long continuous glances to complete successfully and the locations used span the whole windshield to large eccentricities. Consequently, this work primarily aimed to address: (a) How the location of HUD imagery across the windshield impacts on visual demand. Furthermore, the research aimed to extend previous work comparing HUDs and HDDs by: (b) Examining how the visual demand between the recommended HUD position and HDD differs when using a time-consuming, complex task. To investigate these aims, a simulator study compared normal driving to instances where participants were required to look at HUD imagery across nine different locations of a windshield (and a HDD). The display presented a secondary task, rather than one supporting the primary task of driving. So far previous studies concerning HUD location (e.g. [29]) have aimed to identify the ideal location for short, non-visually demanding messages in relatively close proximity to the driver’s forward road view. In contrast, the current work examines an extreme scenario where a complex task requires uninterrupted focal attention for a comparatively long time period (~15 seconds), in order to further explore how the visual demand of HUDs vary with location and compare to a HDDs. Equally, the study progresses the area by examining extreme, windscreen fixed locations using two laser-based HUDs. This work is required to further explore differing visual demand due to HUD imagery location, in order to better inform how HUDs should be evaluated. METHODS Design The study was conducted in a medium fidelity driving simulator (Figure 1) at The University of Nottingham and a within-subject design was employed. The study involved a secondary non-augmented reality task, which was highly visually demanding. The independent variable was the location of the display imagery (see Figure 2). Each participant experienced ten experimental drives (a different display location was active during each drive). The order of display location was counterbalanced across participants. Dependent measures included driving performance (lateral and longitudinal) collected from driving simulation software (Carnet Soft) and subjective data from questionnaires on demographics and display preferences. Figure 1. An external view of the driving simulator used for study. Figure 2. A representation of all the display locations. Participants Twenty-six drivers were recruited for the study (mean age=36.5, standard deviation=13.04 years) via a University mailing list. In total, there were 16 male participants, 9 female and 1 other. The participants had held a driving license for an average of 15.9 years and self-reported that they drove regularly. On average the participant eye-line was 14.6cm down from the car’s ceiling. 245 Evaluating Head-Up Displays across Windshield Locations AutomotiveUI ’19, September 21–25, 2019, Utrecht, Netherlands Materials The car simulator was composed of the front half of a 2001 Honda Civic (right-hand drive) and Carnet Soft simulation software was used to portray the road environment. Three projectors displayed the driving simulation software on to screens placed around the car. No rear-view, side mirrors or speedometer were incorporated for this study. During the drives in the simulator, participants completed a secondary task using a display in one of ten positions. The task was designed to ensure continuous visual attention over an extended time period and involved 60 rapidly alternating letters in the centre of the screen. The letters appeared for a total of 0.2 seconds each. Three times per task a letter would be delayed and remain on uploads/Litterature/ topliss-et-al-2019.pdf