Quantifying Distraction Potential of Augmented Reality Head-Up Displays for Vehicle Drivers

摘要

Background: A recent National Highway Traffic & SafetyAdministration (NHTSA) report states that 10% of fatal crashesand 18% of injury crashes were reported as distraction-affectedcrashes. In that same year, 3,179 people were killed and anestimated 431,000 injured in motor vehicle crashes involvingdistracted drivers, many of which involved secondary visualdisplays (NHTSA, 2016). Augmented reality (AR) head-updisplays (HUD) promise to be less distractive than traditionalin-vehicle displays since they do not take drivers’ eyes off theroad (Gabbard, Fitch, & Kim, 2014). However, empiricalstudies have reported possible negative consequences of ARHUDs, in part, due to AR graphics’ salience (Sharfi & Shinar,2014), frequent changes (Wolffsohn, McBrien, Edgar, & Stout,1998), and visual clutter (Burnett & Donkor, 2012). Moreover,current in-vehicle display assessment methods which are basedon eye-off-road time measures (NHTSA, 2012), cannot capturethis unique challenge.Objective: This work aims to propose a new method for theassessment of AR HUDs by quantifying both positive(informing drivers) and negative (distracting drivers)consequences of AR HUDs which might not be captured bycurrent in-vehicle display assessment methods.Method: We proposed a new way of quantifying the distractionpotential of AR HUDs by measuring driver situation awarenesswith operational improvements on the situation awarenessglobal assessment technique (Endsley, 2012) to suit ARusability evaluations. A human-subject experiment wasconducted in a driving simulator to apply the proposed methodand to evaluate two AR HUD interfaces for pedestrian collisionwarning. The AR warning interfaces were prototyped by theaugmented video technique (Soro, Rakotonirainy, Schroeter, &Wollstdter, 2014). Twenty-four participants drove whileinteracting with different types of AR pedestrian collisionwarning interfaces (no warning, bounding box, and virtualshadow). Drivers’ situation awareness, confidence, andworkload were measured and compared to the no warningcondition.Results: Only one of the warning interface designs, the virtualshadow (Kim, Isleib, & Gabbard, 2016), improved driversituation awareness about pedestrians which were cued by theAR HUD, not affecting situation awareness about otherenvironmental elements which were not augmented by theHUD. The experiment also showed drivers’ overconfidencebias while interacting with the bounding box which is anotherwarning interface design. The empirical user study did notprovide any evidence for reduced driver workload when ARwarnings were given.Conclusion: Our initial human-subject study demonstrated apotential of the proposed method in quantifying both positiveand negative consequences of AR HUDs on driver cognitiveprocesses. More importantly, the experiment showed that ARinterfaces can have both positive and negative consequences ondriver situation awareness depending upon how we designperceptual forms of graphical elements.Application: The proposed assessment methods for AR HUDscan inform not only comparative evaluation among designalternatives but also assist in incrementally improving designiterations to better support drivers’ information needs, situationawareness, and in turn, performance, and safety.