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.
8C h a p t e r 4 One of the tasks in Phase II of the S02 project was to assess the viability of answering the global research questions pri- oritized in Phase I and presenting them in a manner that was understandable to prospective S08 proposers. To accomplish this, the global research questions were refined into global topics areas. The topic areas reflect the scope of the research questions presented in the Phase I report, but they were revised to better convey the intent of the research questions and to clarify their wording for a broader audience. Of the eight high-priority topics listed above, six were con- sidered to be the highest-priority topic areas for Project S08 within the time frame and constraints of SHRP 2. These six top- ics and the factors leading to their selection are presented next. The first high-priority topic, the influence of driver inter- actions with roadway features on lane-departure crashes, is timely because the evaluation of current roadway designs may offer recommendations to reduce the number of lane- departure crashes and subsequent fatalities. Crashes of this type represent a significant percentage of all crashes, and FHWA estimates that 39% of roadway fatalities are single- vehicle roadway departures. Naturalistic driving data provide a good way to examine such crashes, especially because expo- sure to different roadway features may not be accessible in other data sources. Countermeasures or safety interventions for this type of crash are highly feasible because roadway fea- tures (e.g., variable signs, lighting conditions, and pavement markings) can be modified. The second high-priority topic concerns the influence of driver interactions with intersection features (configuration and operations) on crash likelihood. In particular, red light running (RLR) contributes to more than 100,000 crashes and over 1,000 fatalities each year (NHTSA 2007); crash rates at nonsignalized intersections are also high and cause a significant portion of annual crash fatalities (Burgess 2005). Naturalistic driving data provide a way to quantify the rela- tionship of driver factors with intersection features that may not be observed by using crash data only. Given the crash rate and fatalities associated with intersection navigation, interventions and crash countermeasures are both feasible and necessary. Driver impairment from the use of alcohol, prescribed medication, or illicit drugs is a major contributor to crash risk (NHTSA 2001). Although the influence of driver impairment on crash likelihood, the third high-priority topic, has been extensively studied, naturalistic driving data provide a means to quantify a more appropriate estimate of the frequency of alcohol exposure while driving. Safety countermeasures and interventions are entirely plausible and can include enhanced alcohol detection systems, policy changes, and better driver education. The fourth high-priority topic examines the influence of driver distraction on crash likelihood. Driver distraction is a major issue in terms of traffic safety and was identified as con- tributing to 5,870 fatalities and over 515,000 injuries in 2008 (NHTSA 2009). Both the public and state and federal agen- cies are strongly interested in reducing driver distractions. The diverse array of potential sources of distraction from within and outside the vehicle makes this area of research timely and significant for public safety. Because NDS data are particularly suited for capturing the prevalence of distract- ing activities and identifying more accurate rates of engage- ment, feasible safety interventions and mitigation strategies for driver distraction can be better realized with information gained from this naturalistic driving data set. The fifth high-priority topic, the influence of driver fatigue on crash likelihood, is less well known by the public but is esti- mated to have involved over 1.35 million drivers in fatigue- related driving crashes over a 5-year period (Royal 2003). The NDS offers a means to quantify exposure to fatigued driving that is not possible using other study designs. This area repre- sents a major source of research and offers opportunities for highly feasible and potentially implementable interventions that can substantially reduce the number of driver fatigueâ related crashes. Overview of Phase II: Formulating the High-Priority Research Topics
9 problems with other variables that may be of interest, such as drowsiness, distraction, or alcohol-related impairment. However, while driversâ presumed mental states may be inaccessible or somewhat subjective, their behaviors are not. Aggressive driving behaviors, for example, can be quantified using data collected by SHRP 2 through the characterization of behavior as aggressive that is abrupt, impetuous, or risk taking by reducing the safety margin for the driver in a given driving scenario. Examples of aggressive driving behaviors include close headway distance, sudden or excessive accel- eration or braking, speed exceedance, and frequency of lane departures (Fancher et al. 1998). These variables are typically included as indicators of unsafe driving and are of great inter- est to the transportation community. The SHRP 2 NDS will have data related to this construct so that it can be studied in greater detail. Because the sample size associated with driver support systems in the first data collection year might not be suffi- cient to merit appropriate statistical analyses, this topic was also recommended for exclusion. A preliminary review of the types of vehicles to be included in Safety Project S07, In- Vehicle Driving Behavior Field Study (data collection in 2010 and 2011), suggested to the study team that the number of vehicles with advanced systems such as adaptive cruise con- trol and lane-departure systems will not be large enough (at least not for the first set of S08 RFPs) to provide sufficient power in the statistical comparisons. In addition, because the characteristics and algorithms of driver support systems vary greatly among different makes and models, the team thought that comparisons would be challenging. The team notes that these excluded topic areas are impor- tant, and outcomes from examining these issues could support interventions with large safety benefits. However, these topics would be better addressed in a simulator, on a test track, or in a separate NDS. Finally, the team deemed a third question (âHow do dynamic driver characteristics, as observed through driver performance measures, influence crash likelihood?â) to be redundant with other questions that relate to distraction and fatigue, and this question was also removed. Effective crash surrogate measures and the analytical mod- els necessary to predict suitable crash surrogates form the sixth high-priority topic. While each of the preceding areas of research is critically important in terms of its potential to reduce the number of fatalities, injuries, and crashes, crash events are anticipated to be rare in the NDS data. It is there- fore important to identify and determine events that can be used as precursors or surrogates for severe crash events. NDS will provide a means of evaluating the relationship between surrogates and collisions. This topic is timely, as a large por- tion of the other analyses depends on appropriate crash surrogate measures. Moreover, without such surrogate mea- sures, crash countermeasures and safety interventions may actually increase crash rates and decrease safety. Two of the eight high-priority topics were excluded from consideration because it was unlikely that a specific out- come would be produced or that data to address the topic area would be available for the first S08 request for propos- als (RFPs). The first of these relates to aggressive driving behavior and its influence on crash likelihood. This topic was excluded because the authors believed there may be dif- ficulties in developing an operational definition of aggression as the affective state motivating this behavior. For example, one definition of aggressive driving behavior is to presume an emotional state (e.g., road rage) with the driving behavior being goal directed toward manifesting that state and harm- ing the target of the emotion (Ward et al. 1998). This per- spective is not amenable to naturalistic investigation because of the difficulties in measuring the emotional state of the driver independent of the observed behavior. Indeed, spe- cific measures related to this construct are not well defined in the literature (Dula and Geller 2003; Shinar 1998; Smith et al. 2006). In the absence of valid measures of presumed states that motivate behavior, proposers must take care to relate what can actually be measured to hypothetical influ- ences on behavior. For this reason, other recommended high-priority topic areas might be more appropriate to study in the first round of proposals. The challenge in operational- izing the concept of aggressive driving is illustrative of similar
