Analysis of Naturalistic Driving Study Data: Roadway Departures on Rural Two-Lane Curves (2014)

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

69 A p p e n d i x B Additional coding of the video data for each of the rural road curve segments was completed by the University of Iowa. This was done to collect the following information: • Passenger presence and environmental conditions; • Eyeglance location, frequency, and duration; and • Driver distraction. Because of the identifiable nature of the video data, all cod- ing was completed at Virginia Tech Transportation Institute’s (VTTI) secure data enclave. VTTI has many procedures in place to ensure that the SHRP 2 data are protected and only used for the purpose that was specified in the data plan. Spe- cifically, all researchers who enter the enclave are required to use a passcode and leave all electronics outside, all materials are examined when researchers leave the enclave to ensure that data are not removed, and a proctor is scheduled to be in the room whenever a researcher is present. Initial video coding was done by examining the occupancy snapshot available for a particular event. From that frame, the research team coded front and rear passenger presence as well as environmental conditions (e.g., light, weather, road sur- face). For some events, the quality of the video or the size and position of the passenger made this difficult. With only a sin- gle frame, the reviewers cannot see movement that could be attributed to passengers. The lack of audio excludes another way of identifying passenger presence. On several occasions it was necessary to view additional snapshots that might have occurred outside the event but during the same drive in order to see if different lighting conditions or passenger position within the vehicle would make that information obtainable. eyeglance data Driver eyeglance data were coded beginning at the straight- away leading up to the curve and ending at a certain point beyond the curve. These data were coded frame by frame at approximately 10 Hz using Virginia Tech’s Hawk-I software. Table B.1 shows the eyeglance locations and the rules associ- ated with how the glances were defined and applied. Eyeglance data coding under naturalistic driving condi- tions was challenging for a number of reasons: • Bright sunlight caused the camera to “wash out” the entire face, especially at certain times of the day when the sun- light was more direct. External light sources at night, such as street lights, created the same effect. • Night videos had a grainy quality, making it difficult to dis- cern the driver’s eye from the rest of the view. It was thought that this might have been caused by the artificial light created by the infrared light on the camera. • Many drivers wear sunglasses, as well as prescription glasses, both of which create problems associated with glare. These problems were not unexpected and do not make the videos as a whole uncodable. In many cases when the driver’s eye is not visible, a dark spot indicating the pupil can be seen and used for coding. Head movements were also used to aid in the coding of some eyeglances. And glare and other problems associated with sunlight change as the direction of the vehicle changes. One challenge the research team encountered that was unexpected was the discontinuous camera view. As research- ers coded, they found that the camera zoomed in and out to get the best possible view of the driver’s eyes. Therefore, the distance of the eyes from the camera varied among drivers and even within events. While this function was intended to be helpful, in many cases, it did not improve the view and in fact made it harder to remain consistent in the coding. driver distraction Visual distractions are those that cause the driver to take their eyes off the roadway. Table B.2 lists the distractions coded for this study and some of the eyeglance locations that might have been associated with that distraction. Pairing this with Data Reduction Method for Coding Driver Glance Location and Distraction

70 Table B.2. Potential Distractions Associated with Eyeglances Distraction Probable Glance Locations Situation Passenger Right (front-seated passenger), rearview mirror, or over the shoulder (rear-seated passenger) A glance associated with a front- or rear-seated passenger with indica- tion of a conversation or other distracting activity. The glance location depends on the seating position of the passenger. Route planning (locating, viewing, or operating) Steering wheel, down, center console A glance associated with the actions performed during the use of a paper map or in-vehicle navigation system. The glance location depends on where the driver holds the instrument while looking at it. Moving or dropped object in vehicle Down A glance associated with the driver reaching for something in the vehi- cle. The glance location depends on the location of the object. Animal/insect in vehicle All locations are possible A glance associated with the driver being preoccupied by the presence of an animal/insect and taking action to remedy the distraction. The mere presence is not to be coded as a distraction. The glance loca- tion depends on where the animal/insect is located in the vehicle. Cell phone (locating, viewing, operating) Steering wheel, down, center console A glance associated with the actions performed during cell phone use. The glance location depends on where the driver holds the phone while looking at it. IPod/MP3 (locating, viewing, operating) Steering wheel, down, center console A glance associated with the actions performed during the use of an in-vehicle entertainment system. The glance location depends on the location of the device. In-vehicle controls Center console, steering wheel, down A glance associated with the actions performed using the in-vehicle controls (e.g., HVAC, radio, CD player, wipers, windows, door locks). The glance location depends on the control being activated. Drinking/eating Steering wheel, down A glance associated with locating/adjusting food item or drink con- tainer. The glance location depends on where the driver is holding the food/drink. Table B.1. Eyeglance Coding Rules Location of Eyeglance Coding Rule Forward Gazes to the center, left, or right that involve little or no head movement and appear to be mostly directed to the left or right portions of the windshield should be coded as “Forward.” Center console Eyes move slightly down and to the right. There is little or no head movement (e.g., HVAC, radio). Steering wheel Eyes move down slightly. There is little or no head movement (e.g., speedometer, fuel gauge, cruise control). Down Draw an imaginary horizontal line in the middle of the steering wheel. If a gaze is directed above the line it should be coded as “Steering wheel” or “Center console.” If it is below that line, it should be coded as “Down.” Some head movement is associated with a “Down” glance (e.g., looking at something in lap or on floor). Up Eye movement to the upper-left or upper-central portion of the windshield should be coded as “Up.” This glance is rare and is usually associated with the visor or sunroof, if present. Left Any gazes to the left of the A-pillar should be coded as “Left” whether the driver is looking at the left mirror or out the driver’s side window. Right Any gazes that involve both eye and head movement to the right should be coded as “Right” whether the driver is looking at the right mirror, glove box, front-seated passenger, or out the passenger’s side window. Rearview mirror Eye movements up and to the right with a slight head movement should be coded as “Rearview mirror.” These include scanning the roadway behind the vehicle, as well as glances to the rear-seated passengers. Over the shoulder Any glance over the left or right shoulder of the driver. This movement requires the driver’s eyes to pass the B-pillar. Other Blinks, squints, or closed eyes that last more than 10 frames. Any blinks, squints, or closed eyes less than that should be disregarded. Missing Code as “Missing” if • The eyes are obscured or obstructed for more than 10 frames. • The video freezes or video signal is dropped. • The locus of gaze cannot be inferred because of glare, excessive head movement, or camera location. (continued on next page)

71 the frame-by-frame eyeglance data allowed researchers to determine not only the duration of the glance but the cause. Manual distractions are those that require the driver to take a hand off the steering wheel to perform a task unrelated to driving the vehicle. In most cases, these distractions were identifiable and codable using the video data. These distrac- tions include drivers rubbing their nose, twirling their hair, or holding a phone to their ear. In most cases, drivers performed these actions without ever taking their eyes off the road. Time constraints did not allow researchers to code these data frame by frame. However, they were noted in the data file. Cognitive distractions, in which drivers’ attention shifts away from the task of driving, are not easily coded using naturalistic driving data. For this particular study, some of the difficulties stemmed from not having sound or a view of the passengers. Without these, it was not possible to determine whether conversations (phone or personal) were occurring in the vehicle. Even if the researcher could see that the driver’s mouth was moving, it was not possible to discern whether he or she was singing, talking to himself or herself, conversing with a passenger, or using a hands-free phone. Smoking Steering wheel, down, center con- sole, left A glance associated with locating, lighting, smoking, or disposing of ashes. The glance location depends on where the driver holds the cigarette and where he or she discards the ashes. Personal hygiene Up, rearview mirror, steering wheel, down A glance associated with the driver performing an action related to personal hygiene (e.g., fixing hair, applying makeup, blowing nose). The glance location depends on the activity the driver is performing. Other task Any are possible A glance not fitting another category (make a note if used). Table B.2. Potential Distractions Associated with Eyeglances (continued) Distraction Probable Glance Locations Situation