National Academies Press: OpenBook

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

Chapter: Chapter 5 - Selection of Crash Surrogates

« Previous: Chapter 4 - Data Reduction
Page 26
Suggested Citation:"Chapter 5 - Selection of Crash Surrogates." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Roadway Departures on Rural Two-Lane Curves. Washington, DC: The National Academies Press. doi: 10.17226/22317.
×
Page 26
Page 27
Suggested Citation:"Chapter 5 - Selection of Crash Surrogates." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Roadway Departures on Rural Two-Lane Curves. Washington, DC: The National Academies Press. doi: 10.17226/22317.
×
Page 27
Page 28
Suggested Citation:"Chapter 5 - Selection of Crash Surrogates." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Roadway Departures on Rural Two-Lane Curves. Washington, DC: The National Academies Press. doi: 10.17226/22317.
×
Page 28

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.

26 C h a p t e r 5 The goal of SHRP 2 and stakeholders interested in the out- comes of Project S08 research is understanding crashes, par- ticularly severe and fatal crashes. For this reason, the best measure of analysis would be to study crash causes. However, even with the significant amount of data collected in the SHRP 2 NDS, crashes are rare events. At the time data requests were made, only one roadway departure crash and three near crashes had been identified. As a result, it was necessary to use crash surrogates for this project. In addition, the use of crash data to address safety problems is a reactive approach, which is not able to take into account events that lead to successful outcomes. Consequently, the use of surrogates provides an opportunity to study what happens preceding and following an incident or event. For studying crash surrogates, the most significant advantage of naturalistic driving studies is that they provide a firsthand record of the events that precede crashes and incidents. Roadway, environ- mental, vehicle, and human factors can be extracted directly rather than from secondhand information, such as police records and crash databases, to identify relationships among factors that influence roadway departure crash risk. This first- hand information can also be used to determine the factors that lead to a positive outcome using crash surrogates. The following sections discuss the rationale for selection of the identified crash surrogates. Identification of possible roadway Departure Crash Surrogates The team surveyed the literature on crash surrogates in gen- eral and crash surrogates that have been used specifically for roadway departures. Time to collision (TTC), also referred to as time to accident or time to conflict, is one of the most com- mon crash surrogates (Burgett and Gunderson 2001; Gettman and Head 2003; Chin et al. 1992). The concept is logical and provides a repeatable and easily understood metric to assess level of crash risk. Risk can be measured as a function of TTC, where, at TTC = 0, the subject vehicle and another vehicle/ object collide. This makes setting boundaries relatively straightforward. However, to apply the concept of time to collision, the safety-critical event that results in a crash needs to be defined. For an intersection crash, this is a simple process because the safety-critical event is usually collision with another vehicle or pedestrian. The safety-critical event is not so easily defined for roadway departures because multiple crash outcomes could occur for a given roadway departure. For instance, the same roadway departure could result in a rollover or fixed object crash or, if the driver overcorrects, in the vehicle returning to the roadway and colliding with another vehicle. Because TTC depends on knowing the likely outcome, it is difficult to use TTC as a crash surrogate for roadway departures. Use of TTC is also difficult because GPS data from the DAS are not accurate enough to locate the subject vehicle at a given point with sufficient precision to determine distances between objects. Initially, it was thought that calculation of TTC might be possible using distance from forward radar and the nearest strikable object. However, an initial review of the forward radar for several traces suggested that the radar output did not have sufficient detail to determine TTC with another vehicle or object. At the point data used in this study were obtained, the radar had not been processed. Once radar data have been processed, researchers may explore the possibility of using it for estimating time to collision. Time to leaving the shoulder or distance intruded on the shoulder has been suggested as a measure of TTC (Dingus et al. 2008). Unpaved shoulder width is not collected with mobile data collection, and other methods to measure unpaved shoul- der width are not sufficiently accurate to estimate time to leav- ing the shoulder. Distance intruded on the shoulder is related to lane deviation and will be included in this analysis, as described below. Selection of Crash Surrogates

27 Lane deviation is another measure used as a crash surro- gate for both ROR crashes and crashes due to distraction (Donmez et al. 2006). Porter et al. (2004) used lateral place- ment and speed to evaluate the effectiveness of centerline rumble strips. Miaou (2001) developed a method to estimate roadside encroachment frequency and the probability distri- bution for the lateral extent of encroachments using an accident-based prediction model. Taylor et al. (2002) observed vehicle placement relative to the edge line using single versus double paint lines to delineate the presence of shoulder rumble strips. Hallmark et al. (2011) used lateral position to evaluate the effectiveness of edge line rumble stripes. Description of Selected Crash Surrogates The data necessary to identify each of the potential crash sur- rogates mentioned above was considered against what was available in the actual NDS. As already described, TTC was not feasible because the available data were not of sufficient accuracy to determine the distance of the vehicle from the nearest hazard. Lane position or amount of encroachment was another surrogate measure used by researchers and ideally would have been used to address the research questions. Lane deviation is provided as “offset” in the DAS data. A number of other lane position variables are reported by the DAS that can be used to calculate other metrics, such as distance from the left or right lane line. These variables include the following (shown in Figure 5.1): • O = offset (distance from the vehicle centerline to the lane center, in cm). • W = lane width (distance between the inside edge of the innermost lane marking to the left and right of the vehicle centerline, in cm). • LCL = distance from vehicle centerline to the inside of the left lane marking, in cm. • RCL = distance from vehicle centerline to the inside of the right lane marking, in cm. • LPR = probability that the lane marking evaluation is cor- rect for the left-side lane line. • RPR = probability that the lane marking evaluation is correct for the right-side lane line. Offset from lane center and distance from the right lane (RD) or left lane (LD) line are the metrics currently being used as crash surrogates. RD and LD are calculated as shown in the following equations (in meters). 2 2 CL CL L L T R R T D w D w = − − = − where LD = distance from left edge of vehicle to left edge of lane line; RD = distance from right edge of vehicle to right edge of lane line; and Tw = vehicle track width. Use of offset or lane position was explored as the main crash surrogate of interest for Research Questions 2, 3, and 4. At the time data were reduced for this research project, the accuracy of the DAS lane-tracking system had not yet been established. There was also no method by which the CTRE/ ISU team could verify the accuracy of the reported offset and lane position values. As a result, these variables were examined for a number of traces, and several observations were made. First, there is a certain amount of noise in the various vari- ables, as is to be expected from a large-scale data collection of this nature. As an example, lane position offset is shown in Figure 5.2 for one vehicle trace for a distance 300 m upstream and then through the curve. As noted, there is a significant amount of variation and several spikes that do not represent actual erratic changes in lane position. This was resolved in many cases by use of smoothing algorithms, as discussed in Chapter 4. Second, the machine visioning algorithm depends on lane lines or differences in contrast between the roadway edge and shoulder to establish position. When discontinuities in lane lines occur, offset is reported with less accuracy (indicated as lane marking probability, which varies from 0 to 1,024, with higher values indicating better probability). Discontinuities occur for several reasons, such as lane lines being obscured, Figure 5.1. Description of lane position variables.

28 natural breaks being present in lane lines (e.g., turn lanes, intersections), or visibility being compromised in the for- ward roadway view. An attempt was made to set a threshold to indicate the probability of reliable versus unreliable data. A threshold of 512 was selected based on review of the data and consultations with VTTI regarding their assessment of the lane-tracking system. Third, in a number of cases the lane tracker did not appear to be working sufficiently to be considered reliable, but other indicators, such as side acceleration, suggested that a roadway departure may have occurred. Consequently, it was not real- istic to exclude traces in which the lane-tracking system was not working. As a result of these issues, lane offset or position could not be reliably used as a crash surrogate for a large portion of the data. Research Question 3 required the largest sample size in order to include a large number of driver, roadway, and envi- ronmental variables in the analyses. Logistic regression could also use binary dependent variables. As a result, it was deter- mined that encroachments would be the best crash surrogate for Research Question 3. A right-side encroachment is defined as the right vehicle edge crossing the right edge line (when present) or the estimated boundary between the lane and shoulder (when lane lines were not present). A left-side encroachment is defined as the left vehi- cle crossing the centerline. In all cases, the centerline was visible. Figure 5.2. Offset in lane position. An encroachment was determined to have occurred when two of the following criteria were present: • Vehicle edge is 0.2 m beyond edge line/centerline/lane– shoulder boundary. • 0.2 g lateral acceleration is present. • Edge line/centerline/lane–shoulder boundary crossing is visually confirmed using the forward view. It should also be noted that left-side roadway departures may be drivers intentionally crossing the centerline (i.e., “cut- ting the curve”). However, it was not possible to identify when this occurred versus an inadvertent encroachment. The amount of speed over the advisory or posted speed limit at curve entry was also used as a crash surrogate for Research Question 3. Although the correlation between speed and crashes on curves has not been established, speeding has been identified as a major crash contributor. Curve advisory and posted speed limit were known in all cases, and speed appeared to be universally present and reasonably accurate. Research Questions 2 and 4 required a smaller sample size than Research Question 3. Additionally, offset was the only crash surrogate that made sense for the analyses selected. As a result, Research Questions 2 and 4 used offset or lane posi- tion as a crash surrogate and only included traces when offset or position were of sufficient reliability and continuity.

Next: Chapter 6 - Analysis for Research Question 1 »
Analysis of Naturalistic Driving Study Data: Roadway Departures on Rural Two-Lane Curves Get This Book
×
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

TRB’s second Strategic Highway Research Program (SHRP 2) Report S2-S08D-RW-1: Analysis of Naturalistic Driving Study Data: Roadway Departures on Rural Two-Lane Curves analyzes data from the SHRP 2 Naturalistic Driving Study (NDS) and Roadway Information Database (RID) to develop relationships between driver, roadway, and environmental characteristics and risk of a roadway departure on curves.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!