National Academies Press: OpenBook
« Previous: Front Matter
Page 1
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 1
Page 2
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 2
Page 3
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 3
Page 4
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 4
Page 5
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 5
Page 6
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 6
Page 7
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 7
Page 8
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 8
Page 9
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 9
Page 10
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 10
Page 11
Suggested Citation:"Executive Summary." National Academies of Sciences, Engineering, and Medicine. 2014. Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk. Washington, DC: The National Academies Press. doi: 10.17226/22297.
×
Page 11

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.

1Background Communication technology pervades our daily living and is increasingly integrated into the car, where it has the potential to distract drivers. Consequently, there is a critical need to better understand distraction and the limits of attention while driving. Distracted driving, which has long been a contributor to motor vehicle crashes, is flourishing in the fertile environment of communication, information, and entertainment technology that is transforming the car. The term distraction refers to both instances when drivers take their eyes off the road—visual distraction— and instances when drivers take their mind off the road—cognitive distraction. According to the United States–European Union (EU) Driver Distraction and Human Machine Interaction (HMI) Working Group, driver inattention is defined as a mismatch between the current attention allocation (distribution) and that demanded by activities critical for safe driving, whereas driver distraction is defined as diversion of attention away from activities critical for safe driving to one or more activities that are not critical for safe driving (Engström, Monk et al. 2013). In the current context, the activity critical for safe driving is attention to and control of headway to the lead vehicle. The specific mechanisms and indicators of the risk of inattention are unfortunately not definitively quantified. Initial analyses of the Virginia Tech Transportation Institute’s 100-car naturalistic driving study focused on general relationships, such as the proportion of crashes involving inattention as a contributing factor (Dingus et al. 2006), or the relative and population- attributable risk associated with different inattention-related activities (Klauer et al. 2006). Subsequent analyses have examined the influences of various characteristics, such as total eyes- off-roadway time (glance history), single glance duration, and glance location. Previous work has also focused on calculating the risk associated with (human-identified) classifications of distracting tasks, such as talking, dialing, eating, and texting (e.g., Fitch et al. 2013; Klauer et al. 2006, 2010, 2014; Olson et al. 2009). Although this task risk approach has merit, especially for policy decisions and education on what tasks should or shouldn’t be done while driving, it does not explain why the tasks are dangerous—nor does it provide the inattention performance risk information needed for many countermeasures. It is important to be able to determine whether the particular way a driver is doing a task (e.g., radio tuning) is dangerous, rather than simply detecting what task is being done. The radio can be tuned in a safe or unsafe way, and the inattention performance quantification approach presented here focuses on being able to measure this and, in various ways, provide countermeasures based on this. Compared with data from driving simulator and field experiments, naturalistic driving data are valuable because they are able to quantify real crash risk (e.g., NHTSA 2013). But naturalistic driving data, until now with the SHRP 2 data set, have had a limited number of crashes. Risk has generally been calculated for safety-critical events, which group crashes, near crashes, and Executive Summary

2incidents together. Detailed driving behavior data recorded in the seconds leading up to crashes and near crashes cannot be obtained from test tracks, simulators, or observational data (e.g., crash databases). The SHRP 2 Naturalistic Driving Study can provide the data that are needed to provide inattention performance measures associated with precrash situations. The data are essential to improve the understanding of driver inattention, for guidelines to reduce distraction from electronics devices, for countermeasures that detect and act to reduce distraction while driving, and for regulation and education. Research Questions The current research aims to determine the relationship between driver inattention and crash risk in Lead-vehicle precrash scenarios (corresponding to rear-end crashes). It aims to develop inattention-risk relationships, describing how an increase in inattention performance variables combines with context in Lead-vehicle precrash scenarios to increase risk. The inattention-risk relationships are intended to show which glance behaviors are safer than others and pinpoint the most dangerous glances away from the road. A glance is the time from the moment the eyes move toward an area of interest (such as the radio, rearview mirror, or forward path) to the moment they move away from it. The results aim to (1) support distraction policy, regulation, and guidelines; (2) improve intelligent vehicle safety systems; and (3) teach safe glance behaviors. The main research question is this: What is the relationship between driver inattention and crash risk in Lead-Vehicle Precrash Scenarios? Rear-end crashes corresponding to Lead-Vehicle Precrash Scenarios are as follows (Najm and Smith 2007): • Scenario 22: Following vehicle (SHRP 2 driver) making a maneuver and approaching lead vehicle. • Scenario 23: Following vehicle approaching an accelerating lead vehicle. • Scenario 24: Following vehicle approaching lead vehicle moving at lower constant speed. • Scenario 25: Following vehicle approaching decelerating lead vehicle. • Scenario 26: Following vehicle approaching a stopped lead vehicle. These Lead-Vehicle Precrash Scenarios constitute about 29%, or 1.7 million, of the crashes that occurred in the United States in 2004 (Najm and Smith 2007). The specific research questions needed to answer the main question are these: • What are the most dangerous glances away from the road, and what are safer glances? • Can risk from distracting activities (secondary tasks) be explained by glance behavior? • How does the timing of lead-vehicle closing kinematics in relation to off-road glances influence crash risk? • What crash severity scale is best suited for analysis of risk? • How can we change glance behavior to be safer, and how do the results of this research translate into countermeasures? Data Formation and Methods The SHRP 2 Naturalistic Driving Study (NDS) is the largest and most comprehensive ever undertaken—and the largest coordinated safety program ever undertaken in the United States. The study collected data from 3,147 volunteer drivers of all age and gender groups, during a 3-year data collection period (most drivers participated for 1 to 2 years), amounting to about 49.7 million vehicle miles, 5 million trip files, 3,958 data years, more than 1 million hours of video, and 2 petabytes of data. Data were collected across six sites in Florida, Indiana, New York, North Carolina, Pennsylvania, and Washington. An onboard data acquisition system (DAS) was

3 designed, manufactured, and installed in each volunteer’s own vehicle. Data were recorded continuously while the participant’s vehicle was operating and sampled at the original resolu- tion of the sensors. Recorded data included vehicle speed, acceleration, and braking; all vehicle controls; lane position; forward radar; and video views forward, to the rear, and on the driver’s face and hands. According to Najm and Smith (2007), five Lead-vehicle precrash scenarios were targeted because they were (a) highly ranked in crash frequency, functional years lost, and economic cost; (b) proven to be of particular relevance for inattention; and (c) suitable for planned analyses. These Lead-Vehicle Precrash Scenarios correspond approximately to rear-end crashes in National Automotive Sampling System (NASS) crash databases. The final data set that was used for analyses contained 46 rear-end crash events, 211 near-crash events, 257 matched baseline events, and 260 random baseline events. Because this project was one of the first to have access to the SHRP 2 data set in fall 2013 and spring 2014, while data collection was ongoing, the final data set was not yet ready. All crashes and near crashes that were available at the time of data extraction were used. It was estimated that about 20% to 30% of the expected final data set was fully surveyed through kinematic triggers but that the full data set was “surveyed” by automatic notification processes (such as onboard Automatic Crash Notification algorithms, incident button presses, and site reports). One random baseline (RBL) per crash or near-crash event was extracted completely at random from all trips in the available data and across all drivers and locations. The random baselines are used as controls in a case-control approach. One matched baseline (MBL) per crash or near-crash event was selected to match each crash or near-crash event. The matched baselines are used as controls in a case-crossover approach. Analysis using the matching baseline was expected to be more robust to possible confounding contextual factors such as traffic density, weather, and road type. The matching criteria (e.g., driver, trip, no standstill, traffic flow, intersections, speed, weather, day/night) were intended to control for contextual factors that could influence glance behavior and thus create controls that provide a more similar context for comparison between baseline (control) events and crash or near-crash events than the random baselines. Event data, time-series data, and video were delivered for each event. Event data variables describe each event as a whole (in a single value such as precrash scenario type or driver age). Time-series data describe the event over time at a sampling frequency that is specific to each variable. The primary common time period for analyses was 12 seconds before the crash point (in crash events), 12 seconds before minimum time to collision (in near-crash events), and 12 seconds before a reference point in the matched and random baselines. As a complement to the variables defined in the SHRP 2 data dictionaries, a number of other variables were defined. A method was developed to derive kinematic and optical vari- ables related to the lead vehicle (LV) by manually annotating lead-vehicle width in forward video. From this, many lead-vehicle-related variables were derived, such as optically defined range and range rate variables. Manual video annotation of eyeglance location variables and a number of other variables were also coded from video. The main glance variable, Eyes off Path, was defined as glances away from the vehicle’s path, the direction of the vehicle’s travel; a number of glance metrics were derived from this variable. Many other variables were defined and used in the detailed analyses of driver behavior (such as reaction points or start of evasive maneuvers), vehicle kinematics (such as time to collision), and driving context (such as brake light onsets). Several methods were used estimate the risk of having a critical event (crash, near crash, or both, depending on the analysis) as a function of various predictors. The primary method used to estimate risk was odds ratios, as calculated by conditional logistic regression: the higher the odds ratio, the higher the likelihood of being involved in a crash or near crash. The purpose of logistic regression is to develop models of crash or near-crash risk as a function of various pre- dictors associated with driver behavior and environment.

4Key Results Risk from Distracting Activities (Secondary Tasks) The analysis started by replicating previous findings. The analysis shows generally similar results that are consistent with previous findings regarding distracting activities and glance metrics. In general, distracting activities occurred frequently—much more frequently in crashes, near crashes, and baselines than impairments such as drowsiness. In line with previous naturalistic driving studies (e.g., Fitch et al. 2013; Klauer et al. 2006, 2010, 2014; Olson et al. 2009), visually demanding tasks were associated with the highest risk. When considering crash and near-crash situations combined (CNC), the results showed that the aggregate category of Portable Electron- ics Visual-Manual [odds ratio (OR) 2.7, confidence interval (CI) 1.4–5.2] and, in particular, one individual activity in that category, Texting (OR 5.6, CI 2.2–14.5), had the highest odds ratios, suggesting a substantial risk. Talking/Listening on Cell Phone (Figure ES.1) was found to decrease crash/near-crash risk sig- nificantly compared with not engaging in a phone conversation (OR 0.1, CI 0.01–0.7), representing an estimated 10-fold reduction in risk compared with the baseline (OR 10 if the sign of the coeffi- cient is reversed). There were no crashes when drivers were Talking/Listening on Cell Phone. Odds ratios for more than 50 distracting activities were examined. However, many of the activities did not occur frequently enough to achieve statistical significance. Distracting activities do not occur as frequently as glances and thus need larger sample sizes. Other individual catego- ries, such as Locating/Reaching/Answering a Cell Phone or Adjusting/Monitoring the Radio, or other aggregate categories, such as Original Equipment or Vehicle External Distraction, were not significantly risky. Figure ES.1. Odds ratios (numbers inside circles) and confidence intervals (horizontal lines) for specific distracting activities in crashes (C), near crashes (NC), and crashes and near crashes combined (CNC). An odds ratio is significant only when the confidence interval is fully above or below 1 (does not cross the vertical line at 1).

5 Figure ES.1 shows the ORs associated with specific distracting activities. The precise OR is shown in the center of each dot, and the lines surrounding the dots indicate the 95th percentile confidence interval. Odds ratios are significant only when the confidence interval does not cross the vertical line at 1. Figure ES.1 indicates that texting is significantly risky for crashes and near crashes combined and for near crashes alone. The figure also indicates that Talking/Listening on a Cell Phone shows a significantly reduced risk for crashes and near crashes combined. Glance Location and Eyes-off-Path Timelines Before Crash or Minimum Time to Collision Figure ES.2 indicates that the glance locations in the crash events are predominantly toward the cell phone and interior objects, followed by left and right windows/mirrors. Noticeably, there is a reduction in forward path location viewing until about 1.5 seconds before the crash. The eyes return quickly to the forward path location after the 1.5-second mark. Figure ES.3 provides a concise summary, showing only the percentage of Eyes off Path for crashes, near crashes, matched baselines, and random baselines. Figure ES.3 plots Eyes off Path over the 12-second period preceding the crash point in crash events, the minimum time to collision (TTC) in near-crash events, and the reference points for the baselines. It seems that there is generally more Eyes off Path in crashes than in other events and that Eyes off Path is increasingly off the road until 1.5 seconds before the crash. In near crashes, a similar but less pronounced effect is shown. Figure ES.2. Glance locations over time in crash events for the 12 seconds before and 1 second after the crash point (at 0 seconds).

6Most Sensitive Glance Risk Metrics To determine whether risk from distracting activities (secondary tasks) can be explained by glance behavior, it was necessary to first find the most predictive glance metrics. We found that many Eyes-off-Path glance behavior metrics were powerful drivers of risk, much more so than the type of distracting activity (secondary task). The finding that glance behavior has a key contributing role in crashes and near crashes is in line with existing research (e.g., Klauer et al. 2006, 2010, 2014). However, our analyses of single glance metrics quantified this risk more strongly. In general, the greatest risk estimates were shown when crashes were analyzed separately from near crashes. Although very strong Eyes-off-Path–risk relationships were shown in separate glance metrics, the relationship between glance behavior and risk cannot be reduced to a single metric, as there is no separate metric that fully accounts for risk on its own. The relationship is analogous to accounting for discomfort associated with heat. Temperature is a good metric that accounts for much of the variance, but including humidity would result in better predictions, as would including wind speed. Each glance metric helps inform the risk estimates. The most sensitive glance metric model was a linear combination of three-glance metrics because it was most predictive of crashes and near crashes. The first glance metric, Off3to1, denotes the proportion of time the eyes were off path from 3 seconds until 1 second before the crash or minimum time to collision. The second glance metric is the mean duration of off-path glances, mean.off. The third metric, mean uncertainty, is the mean value of a composite measure based on the Senders et al. (1967) uncertainty model of the driving situation. Of the three metrics, the Off3to1 metric is the strongest individual risk-predicting metric. Clearly, factors other than Eyes off Path contribute to rear-end crashes. For example, factors such as age (younger and older) and visibility problems (visual obstructions or rain) were significantly different in crashes compared with near crashes. Figure ES.3. Percentage of Eyes off Path (for each event type at each time point) in relation to minimum TTC or crash point (zero point), and a histogram of the time of the precipitating events associated with each crash and near crash. Precipitating events correspond to the lead-vehicle brake light onsets.

7 Timing of Eyes off Path Relative to Situation Kinematics and Visual Cues In Figure ES.4, it can be observed that the crashes, near crashes, and matched baselines are rela- tively well separated in this state space (i.e., Glance Length and the rate of change of inverse Time to Collision). In particular, the majority of the crashes—and a subset of the near crashes—are organized along a line with a negative slope. This analysis revealed a distinct mechanism for many of the crashes. In line with Tijerina et al. (2004), we found that drivers in most cases did not start to look away when lead vehicle was closing. Rather, drivers who crashed typically looked away just before the lead vehicle started closing and did not look back until collision was unavoidable. The criticality when looking back, and hence the crash risk, is largely determined by an interaction between last glance duration and the rate at which the situation changed during the glance (operationalized here in terms of inverse TTC change rate). The event outcome is also determined by the vehicle’s braking capacity and the driver’s time to react. Thus, the key mechanism behind these types of rear-end crashes (grouped as Category 1, Inopportune glance, in Figure ES.4) can be understood as a “perfect mismatch” between last glance duration and situation change rate (in line with the general mismatch conceptualization of inattention suggested in Engström, Monk et al. 2013). The crashes grouped under a different category (Category 2, Looking away in an already critical situation) followed a similar pattern, but here the driver looked away when the vehicles were already closing, often because of visibility problems that presumably impaired looming detection. The probability of such a mismatch depends on the joint probability distributions of glance durations and situation kinematics. Since long glances are very rare and short glances are very common, many crashes occur due to the combination of a relatively short glance and a high Figure ES.4. Last glance duration versus inverse TTC change rate (the change rate of lead-vehicle looming). Ovals mark the three main categories of crashes identified through video inspection. Cases marked by squares are included as examples in text. The dashed line represents a hypothetical boundary for safe glances.

8change rate. Thus, an important finding of the present analysis is that glances that lead to crashes may not necessarily have to be long. In fact, the majority of the crashes in the present sample were associated with glances shorter than 2 seconds. A prototypical case for the general mechanism in Category 1 can be seen in Figure ES.5. Here, the driver does not expect the lead vehicle to brake and thus looks away from the road while the situation is not yet critical (when inverse TTC is close to zero). During the glance away, the lead vehicle initiates braking. The level of criticality the driver faces when looking back depends on the interaction between the duration of the glance and the rate at which the situation develops, yielding the linear organization with a negative slope in Figure ES.4. Category 2 cases (Figure ES.4) represent situations in which the driver looks away at a point when the situation is already critical (i.e., the vehicles are already closing and the inverse TTC has already risen significantly above zero). This typically involves a very brief glance (around 0.5 second) before the gaze is presumably redirected to the road by the strong loom- ing cues. In Category 3 cases (Figure ES.4), the driver looks away and back again before the situation turns critical, leading to a small change rate during the last glance and varying glance durations. Here the off-path glance most likely did not interfere with the reaction to the event (although it may possibly have affected the detection of available predictive cues). Thus, these events can be regarded as functionally similar to crashes in which the driver did not look away at all within 8 seconds. Category 3 events typically involve a strong violation of expectation [e.g., the lead vehi- cle stops late at yellow light or a traffic queue builds in front of the principal other vehicle (POV) in an unexpected location]. Analyses of driver reactions in the crashes and near crashes showed that driver reactions were not notably affected by the lead-vehicle brake lights but were instead strongly coupled to situa- tion kinematics. Brake light onsets occurring while the driver looked forward were generally ignored and do not seem to have influenced the willingness of the driver to take the eyes off the road. Instead, the data indicate that drivers did not seem to react below an inverse TTC threshold of 0.2 and had progressively faster reactions for larger looming values above this 0.2 threshold. This behavior was successfully predicted by an accumulator model of reaction timing. Another key finding was that reactions in eyes-off-threat situations were generally faster for near crashes compared with crashes. Figure ES.5. Example of Category 1 crash (event ID 19147492). Y-axis units in legend.

9 Discussion and Conclusions What are the most dangerous glances away from the road, and what are safer glances? The team’s initial answer to this question was that the most dangerous and safest glances are quantified by a three-metric glance model. The model combines a metric of inopportune glance, mean glance duration, and a composite measure estimating the driver’s uncertainty of the driving situation. Can risk from distracting activities (secondary tasks) be explained by glance behavior? The three- metric glance model and many of the individual glance behavior metrics were substantially more predictive than the models based on distracting activities. Portable Electronics Visual-Manual interactions were explained by the proportion of Eyes off Path in the 2 seconds overlapping the precipitating event, but Texting and Talking/Listening on Cell Phone were not. However, the three-metric model could not be compared with the distracting activities because the distracting activities were only coded in the 5 seconds preceding the precipitating event until 1 second after. This comparison with the three-metric model should be made in future research. How does the timing of lead-vehicle closing kinematics in relation to off-road glances influence crash risk? A key finding in this report was a distinct mechanism for many of the crashes. The mismatch depends on the joint probability distributions of glance durations and situation kine- matics. Thus, an important finding of the present analysis is that glances that lead to crashes may not necessarily have to be long. This key finding motivates reconsideration of the first question: What are the most dangerous glances away from the road, and what are safer glances? One way to think about which glances are safer than others is in terms of the boundary drawn in Figure ES.4. Under normal conditions (e.g., a dry road surface, normal braking capacity, normal visibility conditions), glances can be regarded as safe as long as they appear under this line, which is determined by the interaction of glance duration and kinematics change rate. Thus, the answer to the first part of the question can be reformulated like this: Dangerous glances are those during which the driver gets exposed to the risk of a rapidly changing situation. This answer is naturally partly related to the glance dura- tion: the longer the glance, the greater the probability that the kinematics will develop in such a way that the perfect mismatch occurs. However, the second part of the equation is the natural variability in vehicle-following situation kinematics. Drivers are normally successful in control- ling this variability by means of anticipation. However, as shown in the present analysis, the safety margins adopted by drivers when looking away are often insufficient to protect them from rapid changes in situation kinematics. For a given glance duration, a certain minimum change rate is needed to produce a crash. Conversely, for a given change rate, the glance has to be suffi- ciently long for a crash to happen. A reformulated answer to the second part of Question 1 is thus as follows: An off-road glance is only perfectly safe when the safety margins adopted are sufficient to protect the driver if the situation changes rapidly during the glance. Further, driver reactions were found to be strongly coupled to situation kinematics and not notably affected by lead-vehicle brake lights. Driver reactions do not occur below a 0.2 inverse TTC threshold (i.e., while there is no perceived looming), and driver reactions are progressively faster with larger inverse TTC values (a looming lead vehicle) once the eyes return to the road. Many researchers have assumed a constant, situation-independent distribution of driver reaction times to the situation itself (Sugimoto and Sauer 2005; Kusano and Gabler 2012). The present results indicate that such an assumption is inadequate. What crash severity scale is best suited for analysis of risk? Our analyses resulted in the formulation and proposal of two potential severity scales: Model-estimated Injury Risk (MIR) index and Model- estimated Crash Risk (MCR) index. These scales are created using mathematical simulations and applying a model of driver glance behavior to kinematics based on actual crashes and near crashes; they represent what may have happened had the event played out according to a specific driver model. The two scales (MIR and MCR) provide continuous values and can be calculated for actual crash and near-crash events. However, further work is necessary to validate these scales. It should be noted that the severity scales are simulated. Actual severity scales—Delta Velocity (DeltaV) for

10 crashes and minimum time to collision (minTTC) for near crashes—are still the most relevant metrics when analyzing actual severity (what actually happened in the event). The main drawback with the actual severity scales is that they cannot be used to compare both crashes and near crashes. This property—the ability to compare potential severity across crashes and near crashes—is enabled by the proposed MIR and MCR scales. It is important to note that these scales are also enabled by naturalistic data. Without the detailed time-series data leading up to crashes and near crashes, it would not be possible to compute the MIR and MCR scales. The answer to the main research question—What is the relationship between driver inattention and crash risk in Lead-Vehicle Precrash Scenarios?—can be found in the general pattern of our results. In line with previous naturalistic driving studies, the results show that some activity types significantly increase risk (such as Texting and the aggregate category of Portable Electronics Visual-Manual). However, for Talking/Listening on Cell Phone, a strong significant decrease in risk was found. Notably, there were no crashes while talking/listening on the phone. Three types of glance metrics showed the largest odds ratios: (1) the proportion of time the eyes were off path between 3 seconds and 1 second before the crash or minimum time to collision, (2) mean dura- tion of off-path glances, and (3) the mean value of a composite measure estimating the driver’s uncertainty of the driving situation. However, it was when these three-glance metrics were com- bined in a model that they were most predictive of crashes and near crashes. Analyses of the timing of off-path glances with lead-vehicle closing kinematics and visual cues revealed a distinct mechanism behind most of the crashes that can be understood in terms of a “perfect mismatch” between last glance duration and the change rate of the lead vehicle closing. Crashes occur with short glances and high closure rates, just as crashes occur with long glances with slow closure rates. These mismatches can be understood in terms of a joint prob- ability distribution for glance durations and closure rates in which the most likely combinations will show up in a crash sample like the present one. Since long glances are rare, many crashes occur due to the combination of a relatively short glance and a high change rate. Another group of crashes followed a similar pattern in which the driver looked away when the vehicles were already closing, often due to visibility problems that presumably impaired looming detection. This pattern of results, or mechanism, was further confirmed in what-if simulation and modeling of reaction time. The main pattern is that lead-vehicle crashes can be understood as the mismatch between glance duration and the lead-vehicle closure rate. Timing matters greatly, and taken together, the analyses strongly reflect this mechanism. How can we change glance behavior to be safer, and how do the results of this research translate into countermeasures? The findings from this project have clear implications for countermeasures, as summarized below. Regarding human-machine interaction design, distraction guidelines, and other regulatory agency countermeasures, the results emphasize the need to tackle the distraction problem as a joint probability problem. Risk can most effectively be reduced by removing the timing mismatch of eyes off road and lead-vehicle closure rates (inverse TTC change rate). A reduction of both sides of the equation—reducing eyes-off-road occurrence and reducing closure rates—is recom- mended. The results point to the importance of designing interfaces that minimize the need for visual interaction, particularly in portable electronic devices. They also indicate that eliminating long glances (e.g., glances above a limit of 2 seconds) will not eliminate the distraction problem, because inopportune glances of normal short duration with the wrong timing relative to high lead-vehicle closure rates often produce rear-end crashes. Further, the results support the poten- tial for nonvisual interfaces because Talking/Listening on a Cell Phone significantly reduced risk. In other words, reduction of off-road glances alone will not solve the problem; a reduction of lead-vehicle closure rates is needed. The results provide strong support for vehicle design and driving support countermeasures, in particular active safety systems such as autonomous emergency braking (AEB) systems, forward collision warning (FCW), and autonomous cruise control (ACC). Active safety systems provide

11 the safety margins needed to protect the driver if the situation changes rapidly during an off-path glance by creating more time headway, issuing warnings to alert the driver to rapid closure rates, and actively braking. For education and behavioral change, it is recommended that the public be made aware of the inopportune glance mismatch mechanism, that the importance of adopting safe headways be emphasized (particularly for ages 16–17 and 76+), and that usage-based insurance be encouraged (e.g., rewarding longer time headways). Regarding road and infrastructure design, emphasis should be placed on creating smooth flowing traffic, reducing the occurrence of sudden, unexpected kinematic changes. Further, improving road surfaces to decrease stopping distances and developing self-explaining roads to reduce unexpected situations are also needed.

Next: Chapter 1 - Background »
Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk 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-S08A-RW-1: Analysis of Naturalistic Driving Study Data: Safer Glances, Driver Inattention, and Crash Risk explores the relationship between driver inattention and crash risk in lead-vehicle precrash scenarios (corresponding to rear-end crashes).

  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!