100-Car Naturalistic Driving Study Research Articles

New Traffic Conflict Measure Based on a Potential Outcome Model

Abstract A key issue in the analysis of traffic accidents is to quantify the effectiveness of a given evasive action taken by a driver to avoid crashing. Since 1977, the widely accepted definition for this effectiveness measure, which is called traffic conflict, has been “the risk of a collision if the driver movement remains unchanged.” Although the definition is expressed counterfactually, the full power of counterfactual analysis was not utilized. In this paper, we propose a counterfactual measure of traffic conflict called Counterfactual Based Conflict (CBC). The CBC is interpreted as the probability that a driver avoided a crash actually by taking an evasive action in the counterfactual situation in which the crash would have occurred if he/she had not taken an evasive action and the crash would not have occurred if he/she had taken an evasive action. The CBC captures realistic aspects of the traffic situation, and lends itself to modern causal analysis. In addition, we provide some of identification conditions for the CBC. Furthermore, we formulate bounds on the CBC when the proposed identification conditions are violated. Finally, through an application of the CBC to the 100-Car Naturalistic Driving Study, we discuss the usefulness and limitations of the proposed measure.

Glass half-full: On-road glance metrics differentiate crashes from near-crashes in the 100-Car data

BackgroundMuch of the driver distraction and inattention work to date has focused on concerns over drivers removing their eyes from the forward roadway to perform non-driving-related tasks, and its demonstrable link to safety consequences when these glances are timed at inopportune moments. This extensive literature has established, through the analyses of glance from naturalistic datasets, a clear relationship between eyes-off-road, lead vehicle closing kinematics, and near-crash/crash involvement. ObjectiveThis paper looks at the role of driver expectation in influencing drivers’ decisions about when and for how long to remove their eyes from the forward roadway in an analysis that consider the combined role of on- and off-road glances. MethodUsing glance data collected in the 100-Car Naturalistic Driving Study (NDS), near-crashes were examined separately from crashes to examine how momentary differences in glance allocation over the 25-s prior to a precipitating event can differentiate between these two distinct outcomes. Individual glance metrics of mean single glance duration (MSGD), total glance time (TGT), and glance count for off-road and on-road glance locations were analyzed. Output from the AttenD algorithm (Kircher and Ahlström, 2009) was also analyzed as a hybrid measure; in threading together on- and off-road glances over time, its output produces a pattern of glance behavior meaningful for examining attentional effects. ResultsIndividual glance metrics calculated at the epoch-level and binned by 10-s units of time across the available epoch lengths revealed that drivers in near-crashes have significantly longer on-road glances, and look less frequently between on- and off- road locations in the moments preceding a precipitating event as compared to crashes. During on-road glances, drivers in near-crashes were found to more frequently sample peripheral regions of the roadway than drivers in crashes. Output from the AttenD algorithm affirmed the cumulative net benefit of longer on-road glances and of improved attention management between on- and off-road locations. ConclusionThe finding of longer on-road glances differentiating between safety-critical outcomes in the 100-Car NDS data underscores the importance of attention management in how drivers look both on and off the road. It is in the pattern of glances to and from the forward roadway that drivers obtained critical information necessary to inform their expectation of hazard potential to avoid a crash. ApplicationThis work may have important implications for attention management in the context of the increasing prevalence of in-vehicle demands as well as of vehicle automation.

Predicting crash-relevant violations at stop sign–controlled intersections for the development of an intersection driver assistance system

ABSTRACTObjective: Intersection crashes resulted in over 5,000 fatalities in the United States in 2014. Intersection Advanced Driver Assistance Systems (I-ADAS) are active safety systems that seek to help drivers safely traverse intersections. I-ADAS uses onboard sensors to detect oncoming vehicles and, in the event of an imminent crash, can either alert the driver or take autonomous evasive action. The objective of this study was to develop and evaluate a predictive model for detecting whether a stop sign violation was imminent.Methods: Passenger vehicle intersection approaches were extracted from a data set of typical driver behavior (100-Car Naturalistic Driving Study) and violations (event data recorders downloaded from real-world crashes) and were assigned weighting factors based on real-world frequency. A k-fold cross-validation procedure was then used to develop and evaluate 3 hypothetical stop sign warning algorithms (i.e., early, intermediate, and delayed) for detecting an impending violation during the intersection approach. Violation detection models were developed using logistic regression models that evaluate likelihood of a violation at various locations along the intersection approach. Two potential indicators of driver intent to stop—that is, required deceleration parameter (RDP) and brake application—were used to develop the predictive models. The earliest violation detection opportunity was then evaluated for each detection algorithm in order to (1) evaluate the violation detection accuracy and (2) compare braking demand versus maximum braking capabilities.Results: A total of 38 violating and 658 nonviolating approaches were used in the analysis. All 3 algorithms were able to detect a violation at some point during the intersection approach. The early detection algorithm, as designed, was able to detect violations earlier than all other algorithms during the intersection approach but gave false alarms for 22.3% of approaches. In contrast, the delayed detection algorithm sacrificed some time for detecting violations but was able to substantially reduce false alarms to only 3.3% of all nonviolating approaches. Given good surface conditions (maximum braking capabilities = 0.8 g) and maximum effort, most drivers (55.3–71.1%) would be able to stop the vehicle regardless of the detection algorithm. However, given poor surface conditions (maximum braking capabilities = 0.4 g), few drivers (10.5–26.3%) would be able to stop the vehicle. Automatic emergency braking (AEB) would allow for early braking prior to driver reaction. If equipped with an AEB system, the results suggest that, even for the poor surface conditions scenario, over one half (55.3–65.8%) of the vehicles could have been stopped.Conclusions: This study demonstrates the potential of I-ADAS to incorporate a stop sign violation detection algorithm. Repeating the analysis on a larger, more extensive data set will allow for the development of a more comprehensive algorithm to further validate the findings.

Evaluating the influence of crashes on driving risk using recurrent event models and Naturalistic Driving Study data

ABSTRACTDramatic events such as crashes could alter driver behavior and change driving risk during post-event period. This study investigated the influence of crashes on driving risk using the 100-Car Naturalistic Driving Study data. The analysis is based on 51 crashes from primary drivers. Driving risk is measured by the intensity of safety-critical incidents (SCI) and near-crashes (NC), which typically occur at a high frequency both before and after a crash. We applied four alternative recurrent event models to evaluate the influence of crashes based on actual driving time. The driving period was divided into several phases based on the relationship to crashes, and the event intensities of these periods were compared. Results show a reduction in SCI intensity after the first crash (; CI ) and the second crash (; CI ) for male drivers. No significant response to the first crash was observed for females, but SCI intensity decreased after the second crash (; CI ). The findings of this study provide crucial information for understanding driver behavior and for developing effective safety education programs as well as safety counter measures.

Driver Behavior During Overtaking Maneuvers from the 100-Car Naturalistic Driving Study

Objective: Lane changes with the intention to overtake the vehicle in front are especially challenging scenarios for forward collision warning (FCW) designs. These overtaking maneuvers can occur at high relative vehicle speeds and often involve no brake and/or turn signal application. Therefore, overtaking presents the potential of erroneously triggering the FCW. A better understanding of driver behavior during lane change events can improve designs of this human–machine interface and increase driver acceptance of FCW. The objective of this study was to aid FCW design by characterizing driver behavior during lane change events using naturalistic driving study data.Methods: The analysis was based on data from the 100-Car Naturalistic Driving Study, collected by the Virginia Tech Transportation Institute. The 100-Car study contains approximately 1.2 million vehicle miles of driving and 43,000 h of data collected from 108 primary drivers. In order to identify overtaking maneuvers from a large sample of driving data, an algorithm to automatically identify overtaking events was developed. The lead vehicle and minimum time to collision (TTC) at the start of lane change events was identified using radar processing techniques developed in a previous study. The lane change identification algorithm was validated against video analysis, which manually identified 1,425 lane change events from approximately 126 full trips.Results: Forty-five drivers with valid time series data were selected from the 100-Car study. From the sample of drivers, our algorithm identified 326,238 lane change events. A total of 90,639 lane change events were found to involve a closing lead vehicle. Lane change events were evenly distributed between left side and right side lane changes. The characterization of lane change frequency and minimum TTC was divided into 10 mph speed bins for vehicle travel speeds between 10 and 90 mph. For all lane change events with a closing lead vehicle, the results showed that drivers change lanes most frequently in the 40–50 mph speed range. Minimum TTC was found to increase with travel speed. The variability in minimum TTC between drivers also increased with travel speed.Conclusions: This study developed and validated an algorithm to detect lane change events in the 100-Car Naturalistic Driving Study and characterized lane change events in the database. The characterization of driver behavior in lane change events showed that driver lane change frequency and minimum TTC vary with travel speed. The characterization of overtaking maneuvers from this study will aid in improving the overall effectiveness of FCW systems by providing active safety system designers with further understanding of driver action in overtaking maneuvers, thereby increasing system warning accuracy, reducing erroneous warnings, and improving driver acceptance.

Population distributions of time to collision at brake application during car following from naturalistic driving data

ProblemForward collision warning (FCW) systems are designed to mitigate the effects of rear-end collisions. Driver acceptance of these systems is crucial to their success, as perceived “nuisance” alarms may cause drivers to disable the systems. In order to make customizable FCW thresholds, system designers need to quantify the variation in braking behavior in the driving population. The objective of this study was to quantify the time to collision (TTC) that drivers applied the brakes during car following scenarios from a large scale naturalistic driving study (NDS). MethodsBecause of the large amount of data generated by NDS, an automated algorithm was developed to identify lead vehicles using radar data recorded as part of the study. Using the search algorithm, all trips from 64 drivers from the 100-Car NDS were analyzed. A comparison of the algorithm to 7135 brake applications where the presence of a lead vehicle was manually identified found that the algorithm agreed with the human review 90.6% of the time. ResultsThis study examined 72,123 trips that resulted in 2.6 million brake applications. Population distributions of the minimum, 1st, and 10th percentiles were computed for each driver in speed ranges between 3 and 60mph in 10mph increments. As speed increased, so did the minimum TTC experience by drivers as well as variance in TTC. Younger drivers (18–30) had lower TTC at brake application compared to older drivers (30–51+), especially at speeds between 40mph and 60mph. DiscussionThis is one of the first studies to use large scale NDS data to quantify braking behavior during car following. The results of this study can be used to design and evaluate FCW systems and calibrate traffic simulation models.

Age and Gender Differences in Time to Collision at Braking From the 100-Car Naturalistic Driving Study

Objective: Forward collision warning (FCW) is an active safety system that aims to mitigate the effect of forward collisions by warning the driver of objects in front of the vehicle. Success of FCW relies on how drivers react to the alerts. Drivers who receive too many warnings that they deem as unnecessary—that is, nuisance alarms—may grow to distrust and turn the system off. To reduce the perception of nuisance alarms, FCW systems can be tailored to individual driving styles, but these driving styles must first be characterized. The objective of this study was to characterize differences in braking behavior between age and gender groups in car-following scenarios using data from the 100-Car Naturalistic Driving Study.Methods: The data source for this study was the 100-Car Naturalistic Driving Study, which recorded the driving of 108 primary drivers for approximately a year. Braking behavior was characterized in terms of time to collision (TTC) at brake application, a common metric used in the design of warning thresholds of FCW. Because of the large volume of data analyzed, the TTC at which drivers braked during car-following situations was collected via an automated search algorithm. The minimum TTC for each vehicle speed 10 mph increment from 10 mph to 80 mph was recorded for each driver. Mixed model analysis of variance was used to examine the differences between age and gender groups.Results: In total, 527,861 brake applications contained in 11,503 trips were analyzed. Differences in TTC at braking were statistically significant for age and gender (P <.01 for both cases). Males age 18–20 (n = 7) had the lowest average minimum TTC at braking of 2.5 ± 0.8 s, and females age 31–50 (n = 6) had the highest average minimum TTC at braking of 6.4 ± 0.9 s. On average, women (n = 32) braked at a TTC 1.3 s higher than men (n = 52). Age was a statistically significant factor for TTC at braking between participants under 30 (n = 42) and participants over 30 (n = 42), with the latter braking 1.7 s on average before the former. No statistical significance was found between ages 18–20 (n = 15) and 21–30 (n = 27) or between ages 31–50 (n = 23) and 50 + (n = 19).Conclusions: There are clear statistical differences in TTC at braking for both gender and those over 30 vs. those under 30. Designers of FCW systems can use the data found in this study to tailor alert timings to the target demographic of a vehicle when designing forward collision warning systems. Appropriate alert timings for FCW systems will maximize effectiveness in collision reduction and mitigation.

A Looming Crisis

Long glances away from the road present a significant risk to driving safety. The tail of the distribution for in-vehicle glance duration has been proposed to be more relevant to crash risk than the mean of the distribution. Using data collected in the 100-Car Naturalistic Driving Study (Dingus et al., 2006), this study examined the changes in the distribution of driver off-road glance duration, as well as in the frequency of such glances, at different time points before crash/near-crash events and compared with baseline (i.e., normal) driving. The results showed that the shape of the distribution of off-road glance duration at the onset of the precipitating factor of crashes/near-crashes was similar to the distribution in epochs during the preceding 25 seconds; nonetheless, the frequency of off-road glances increased in the approach to crashes/near-crash events. Compared with baseline epochs, drivers in crashes/near-crash events tended to look away from the forward road more often and with longer duration (i.e., with thicker tail end of the distributions; exceeding 1.7 seconds as observed in the distributions). It suggests that frequent off-road glances longer than 1.7 seconds present a high-risk glance pattern in the seconds preceding a safety- critical event and that the 2.0 second-threshold that is frequently cited in defining dangerously long off-road glances might be a liberal estimation.

Using naturalistic driving data to explore the association between traffic safety-related events and crash risk at driver level

There has been considerable research conducted over the last 40 years using traffic safety-related events to support road safety analyses. Dating back to traffic conflict studies from the 1960s these observational studies of driver behavior have been criticized due to: poor quality data; lack of available and useful exposure measures linked to the observations; the incomparability of self-reported safety-related events; and, the difficulty in assessing culpability for safety-related events. This study seeks to explore the relationships between driver characteristics and traffic safety-related events, and between traffic safety-related events and crash involvement while mitigating some of those limitations. The Virginia Tech Transportation Institute 100-Car Naturalistic Driving Study dataset, in which the participants’ vehicles were instrumented with various cameras and sensors during the study period, was used for this study. The study data set includes 90 drivers observed for 12–13 months driving. This study focuses on single vehicle run-off-road safety-related events only, including 14 crashes and 182 safety-related events (30 near crashes, and 152 crash-relevant incidents). Among the findings are: (1) drivers under age 25 are significantly more likely to be involved in safety-related events and crashes; and (2) significantly positive correlations exist between crashes, near crashes, and crash-relevant incidents. Although there is still much to learn about the factors affecting the positive correlation between safety-related events and crashes, a Bayesian multivariate Poisson log-normal model is shown to be useful to quantify the associations between safety-related events and crash risk while controlling for driver characteristics.

Cell Phone Resting Locations

The use of cellular phones while driving has been a growing topic of concern in the field of transportation safety research. Traditionally, a vehicle may have been equipped with multiple fixed displays that provided the driver various information, such as vehicle-related information and entertainment options (for example, AM-FM radio). The introduction of a mobile display has brought dramatic changes to the driving environment and has created a need for researchers to understand the full impact these devices have on the driving task. Many studies have been conducted to quantify the impact cell phones have on drivers’ crash risks. However, little or no research has been performed that examines the physical placement of the cell phone while the driver is engaged in the primary task of operating the vehicle. For determination of the most frequent cell phone resting locations and containers, a visual evaluation was performed during this study of the video collected during the 100-Car Naturalistic Driving Study between January 2003 and July 2004. In the video, drivers used a cell phone at any point during a trip. Knowing the placement of cell phones can aid researchers in designing experiments focused on in-vehicle cell phone use and can improve the design and technology of handheld devices and vehicles that attempt to integrate cell phone use safely.

Investigating Improper Lane Changes: Driver Performance Contributing to Lane Change Near-Crashes

We investigated the contributing factors that led to the lane change near-crashes recorded in the 100-Car Naturalistic Driving Study using a case-crossover experimental design. Drivers’ visual behavior and vehicle control were compared across a sample of lane change near-crashes and matched baselines. Baseline lane changes were sampled if they occurred prior to the near-crash, had a similar maneuver as the near-crash (including direction and speed), occurred within ± 2 hours from the time of day, occurred in similar light conditions, occurred on a similar day of the week (weekday vs. weekend), occurred on a road that had a similar number of lanes, had a similar placement of surrounding vehicles, and were made by the same driver. A total of 18 lane change near-crashes and 33 baseline lane changes were identified. Left lane change near-crashes appear to have resulted in part because drivers tended to slow down at the start of the maneuver and were less likely to use their rearview mirror. Right lane change near-crashes appeared to have occurred because of more aggressive maneuvering, infrequent turn signal use, and because drivers were less likely to look over their shoulder. Deficiencies in judging the distance and approach rate to adjacent vehicles, as well as circumstances in the environment, may also have played a contributing role.

Individual driver risk assessment using naturalistic driving data

Driving risk varies substantially among drivers. Identifying and predicting high-risk drivers will greatly benefit the development of proactive driver education programs and safety countermeasures. The objective of this study is twofold: (1) to identify factors associated with individual driver risk and (2) predict high-risk drivers using demographic, personality, and driving characteristic data. The 100-Car Naturalistic Driving Study was used for methodology development and application. A negative binomial regression model was adopted to identify significant risk factors. The results indicated that the driver's age, personality, and critical incident rate had significant impacts on crash and near-crash risk. For the second objective, drivers were classified into three risk groups based on crash and near-crash rate using a K-mean cluster method. The cluster analysis identified approximately 6% of drivers as high-risk drivers, with average crash and near-crash (CNC) rate of 3.95 per 1000miles traveled, 12% of drivers as moderate-risk drivers (average CNC rate=1.75), and 84% of drivers as low-risk drivers (average CNC rate=0.39). Two logistic models were developed to predict the high- and moderate-risk drivers. Both models showed high predictive powers with area under the curve values of 0.938 and 0.930 for the receiver operating characteristic curves. This study concluded that crash and near-crash risk for individual drivers is associated with critical incident rate, demographic, and personality characteristics. Furthermore, the critical incident rate is an effective predictor for high-risk drivers.

How Dangerous Is Looking Away From the Road? Algorithms Predict Crash Risk From Glance Patterns in Naturalistic Driving

In this study, the authors used algorithms to estimate driver distraction and predict crash and near-crash risk on the basis of driver glance behavior using the data set of the 100-Car Naturalistic Driving Study. Driver distraction has been a leading cause of motor vehicle crashes, but the relationship between distractions and crash risk lacks detailed quantification. The authors compared 24 algorithms that varied according to how they incorporated three potential contributors to distraction--glance duration, glance history, and glance location--on how well the algorithms predicted crash risk. Distraction estimated from driver eye-glance patterns was positively associated with crash risk. The algorithms incorporating ongoing off-road glance duration predicted crash risk better than did the algorithms incorporating glance history.Augmenting glance duration with other elements of glance behavior--1.5th power of duration and duration weighted by glance location--produced similar prediction performance as glance duration alone. The distraction level estimated by the algorithms that include current glance duration provides the most sensitive indicator of crash risk. The results inform the design of algorithms to monitor driver state that support real-time distraction mitigation systems.

Analysis of Naturalistic Driving Event Data

Naturalistic driving studies have been conducted over the past 5 years or more and have commonly reviewed video and kinematic data to identify and analyze crash, near-crash, and critical-incident events. But statistical methods that are applicable to these event data are needed. This paper addresses two issues in model development for naturalistic driving event data: the test for omitted-variable bias and the exploration of the advantages of hierarchical model structures in data analysis. With roadway departure event data from the 100-Car Naturalistic Driving Study conducted at Virginia Tech Transportation Institute, Blacksburg, Virginia, logit models were used to estimate the probability that a crash or a near crash would occur, rather than a critical incident. The models indicated a substantial omitted-variable bias for estimation of the effect of context variables but little difference for driver variables. These tests indicated that modeling of naturalistic event data should have included variables that described the attributes of the event, the driver, and the context to reduce the likelihood of bias. Hierarchical model structures offer the advantage of driver-level predictors to parameterize the effects of event attributes and contexts. The models thus reflect how driver decisions are executed: drivers with particular characteristics (one level) find themselves in contexts in which they execute specific driving maneuvers (second level), which lead to certain outcomes. Suggestions for further research include testing with additional data sets and potential applications to analysis of crash surrogates.

Near Crashes as Crash Surrogate for Naturalistic Driving Studies

Naturalistic driving is an innovative method for investigating driver behavior and traffic safety. However, as the number of crashes observed in naturalistic driving studies is typically small, crash surrogates are needed. A study evaluated the use of near crashes as a surrogate measure for assessment of the safety impact of driver behaviors and other risk factors. Two metrics, the precision and bias of risk estimation, were used to assess whether near crashes could be combined with crashes. The principles and exact conditions for improved precision and unbiased estimation were proposed and applied to data from the 100-Car Naturalistic Driving Study. The analyses indicated that a positive relationship exists between the frequencies of contributing factors for crashes and for near crashes. The study also indicated that analyses based on combined crash and near-crash data consistently underestimate the risk of contributing factors compared to use of crash data alone. At the same time, the precision of the estimation will increase. This consistent pattern allows investigators to identify true high-risk behaviors while qualitatively assessing potential bias. In summary, the study concluded that the use of near crashes as a crash surrogate provides definite benefit when naturalistic studies are not large enough to generate sufficient numbers of crashes for statistical analysis.

Urban Freeway Crash Analysis

Associations between the number of crashes and the severity of crashes with a long list of potentially contributing factors, such as geometric and operational characteristics and weather conditions, were investigated. Both aggregate and disaggregate databases with crash statistics were used. Severity was estimated with constants for the different impact (cost) of fatalities, injuries, and material damages. Noteworthy associations included the negative correlation with the speed limit (spots with lower speed limits have more, or more severe, crashes), existence of merge–diverge pavement markings (suggesting the possibility of an illegal driving maneuver), downhill grade sections, and curves. Several temporal variables such as months, days, or time of day with a higher frequency of crashes worked well in models, but those were findings of local interest. An important outcome of this analysis is the absence of any major correlation, which implies that crashes on the modern and mildly congested Attica Tollway in Greece are events largely attributable to driver behavior and other circumstances that cannot be accounted for by geometric, operational, or other temporal variables. Findings of this research suggest that fixed and temporal roadway characteristics and the presence of rain or wet pavements may explain (and likely contribute to) about 5% to 10% of the crashes and severity of crashes observed in 2004 and 2005 on the Attica Tollway. The 2006 report from the 100-Car Naturalistic Driving Study concluded that nearly 80% of all crashes involved driver inattention, lending support to the finding that roadway and environmental factors are relatively minor causes of crashes.

Driver Inattention: A Contributing Factor to Crashes and Near-Crashes

Driver distraction, or inattention, has been receiving wide media attention recently as many state legislatures are considering various levels of restricting cell phone use. Research has been conducted using a variety of experimental methods to determine the level of risk associated with driving inattention. While most of this research suggests that inattention impairs driving, there have been no studies to directly link driving inattention to crashes. Data from the 100-Car Naturalistic Driving Study, an instrumented vehicle study for which data was collected on 100 drivers in the Washington, DC metropolitan area for 12 months, were used in the following analyses. Crashes and near-crashes were identified in the data using post-hoc triggers based upon driving performance metrics, (i.e. hard braking). Results suggest that inattention contributed to 78% of all crashes collected over the 12 month data collection period.

Naturalistic Data Collection of Driver Performance in Familiar and Unfamiliar Vehicles

The 100-Car Naturalistic Driving Study was the first large-scale instrumented vehicle study with no special driver instructions, unobtrusive data collection instrumentation, and no in-vehicle experimenter. The final data set includes approximately 2,000,000 vehicle miles, almost 43,000 hours of data, 241 primary and secondary drivers, 12 to 13 months of data collection for each vehicle, and data from a highly capable instrumentation system. In addition, 78 of 102 vehicles were privately owned and 22 were leased. After 12 months, leased vehicles were provided to 22 private vehicle drivers who then drove the leased vehicles for an additional four weeks. Driving performance for the same drivers in familiar and unfamiliar instrumented vehicles was then compared. Results provided evidence of increased relative risk for the same driver for weeks 1 through 4 of driving an unfamiliar leased vehicle as compared to the same period of driving their privately owned vehicle.