Feasibility of Using In-Vehicle Video Data to Explore How to Modify Driver Behavior That Causes Nonrecurring Congestion (2011)

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.

C H A P T E R 2 Existing Studies Using In-Vehicle Video CamerasDigital video cameras have rapidly evolved since the 1990s. Tapeless video cameras make it possible to use relatively small hardware equipment to record and save large-sized data. The quality of images has greatly improved and the editing process is simplified, allowing nonlinear editing systems to be widely deployed on desktop computers. Even though image sizes are small, it is easy to recognize the movements of targets and the ongoing background environment and acquire the information needed. Because of their advanced capability, digital cameras have recently been used in multiple transportation safety research projects to capture drivers’ behaviors in a naturalistic driving environment. Table 2.1 lists the studies discussed in this report. Because of difficulties associated with older data collection methods (e.g., videocassette tapes), emphasis was placed on research conducted during recent years. A short description of each project follows Table 2.1.Project 1: Sleeper Berth Conducted by the Virginia Tech Transportation Institute (VTTI), this study examined drivers’ sleeping habits and drowsiness with respect to crash risk. Two tractor trailers were instrumented and loaned to trucking companies for their extended runs. Data from 41 drivers were used. In total, 47 males and 9 females were involved in the study. The average age was 43, with ages ranging from 28 to 63. On average, the drivers had 13 years of driving experience. Data collection runs lasted up to 240 h (6 to 10 days). Continuous video data were recorded on four channels: driver’s face, forward roadway, left rear, and right rear. Data were saved only for predefined critical incidents. If a specified kinematic trigger was activated, the data acquisition system (DAS) saved video and parametric data for 90 s before and 30 s after the event. Predefined events included: • Steering wheel moved faster than 3.64 rad/s; • Lateral acceleration was greater than 0.3 g;8• Longitudinal acceleration was greater than 0.25 g; • Critical incident button was pressed; • Vehicle crossed solid lane border; • Time to collision (TTC) of 4 s or less; • PERCLOS (percent eyelid closure) of 8% for 1 min; • Driver subjectively assessed drowsiness as “extremely fatigued or difficult to stay awake” or did not respond; • Lane departure followed by a steering event (disabled if turn signal was on); and • A baseline data file, collected every 45 to 75 min. In addition to the video-recorded data, the data collected included vehicle network information (speed, accelerator, brake pedal, and steering wheel); environmental monitoring (temperature, illumination, vibration, and noise in decibels); X, Y, and Z acceleration; and lane orientation using a SafeTRAC lane tracking system, as well as some data generated after the data reduction, such as eyeglance behavior and road type and geometry (1–3). Project 2: Automotive Collision Avoidance System Field Operational Test The Automotive Collision Avoidance System (ACAS) FOT was led by General Motors (GM) under a cooperative agreement with the U.S. Department of Transportation (DOT). The FOT involved exposing a fleet of 11 ACAS-equipped Buick LeSabre passenger cars to 12 months of naturalistic driving by drivers from southeastern Michigan. The ACAS included a forward crash warning (FCW) and an adaptive cruise control (ACC) system. The FOT’s goal was to determine the feasibility of the ACAS for widespread deployment from the perspective of driving safety and driver acceptance. Ninety-six drivers partic- ipated, resulting in 137,000 mi of driving. Results indicated that the ACC was widely accepted by drivers, but the accept- ance of the FCW was mixed (due to false alarms) and not found to be significantly related to the FCW alert rate.

9Time Frame (Year) Study (Institute That Conducted the Research) 97 98 99 00 01 02 03 04 05 06 07 08 09 10 1. Sleeper Berth (VTTIa) 2. Automotive Collision Avoidance System Field Operational Test (UMTRIb) 3. Quality of Behavioral and Environmental Indicators Used to Infer the Intention to Change Lanes (Chemnitz University of Technology and INRETS)c 4. Lane Change Field Operational Test (VTTI) 5. Road Departure Crash Warning System Field Operational Test (UMTRI) 6. The 100-Car Study (VTTI) 7. Drowsy Driver Warning System Field Operational Test (VTTI) 8. Naturalistic Truck Driving Study (VTTI) 9. Naturalistic Driving Performance During Secondary Tasks (UMTRI) 10. Effect of In-Vehicle Video and Performance Feedback on Teen Driving Behavior (Iowa) 11. Naturalistic Teen Driving Study (VTTI) 12. Cooperative Intersection Collision Avoidance System for Violations (CICAS-V)—Infrastructure (VTTI) 13. Pilot Study to Test Multiple Medication Usage and Driving Functioning (NHTSAd) 14. Older Driver Field Operational Test (ongoing study) (VTTI) 15. Cooperative Intersection Collision Avoidance System for Violations (CICAS-V)—Pilot Field Operational Test (VTTI) 16. Volvo Driving Behavior Field Operational Test (ongoing study) (Volvo and SAFERe) a Virginia Tech Transportation Institute b University of Michigan Transportation Research Institute c French National Institute for Transport and Safety Research (Institut National de Recherche sur les Transports et leur Sécurité) d National Highway Traffic Safety Administration e SAFER Vehicle and Traffic Safety Centre, Chalmers University of Technology, Sweden Table 2.1. List of Studies Using In-Vehicle CamerasDriver image data were collected for 8 s (4 s before and 4 s after the event) when the ACAS was activated. Alerts consisted of ACC engagements, collision warnings, hard braking, and hard steering. In addition to the video data, 250 separate data signals were collected, including data for trip time history, trip transition, trip summary (dura- tion and counts), trigger summary, buffered Controller Area Network (CAN) packets, video time and image, and audio data. The data set also has a valid time stamp and global posi- tioning system (GPS) data that can be used to link to weather and other environmental data, such as traffic count and work zones (4). Project 3: Quality of Behavioral and Environmental Indicators Used to Infer the Intention to Change Lanes The study was conducted by the Chemnitz University of Tech- nology (Chemnitz, Germany) and the Institut National de Recherche sur les Transports et leur Sécurité (INRETS) inBron, France. It focused on the examination and comparison of selected behavioral and environmental indicators that pre- dict the intention to change lanes. Data were collected on a multilane motorway between Bron and the Lyon Interna- tional Airport (in both directions), with a length of 31 mi in an area of central France around the city of Lyon. The study included 22 participants aged 24 to 58; among them, 10 were female and 12 were male. Their driving experience ranged from 2 to 39 years and annual driving distance ranged between 1,243 and 31,075 mi. Participants were required to drive a Renault Scenic equipped to record and synchronize sensor data and videos on this route. Video signals were recorded from five sources: (1) stereo- vision camera with radar for distance estimation to obstacles; (2) front view; (3) rear view; (4) view from the left outside mir- ror down to the road surface; and (5) view of the participant’s head with indications of the eye tracker. Data other than video data included speed, acceleration, deceleration, yaw rate, driver’s eye movement, steering wheel position, pedal use, turn signal, and inclination. Environmen- tal data that included the distance to the car ahead and GPS data were also gathered (5).

10Project 4: Lane Change Field Operational Study The main purpose of this VTTI study was to examine lane change behavior. The study monitored the commutes of 16 participants for approximately 1 month. Drivers commuted on Route 460 through the New River Valley or on Interstates 81 and 581 in the Roanoke Valley. Commutes were 25 or more miles in each direction to and from work. Data would begin recording when a vehicle reached 35 mph and stopped recording when the vehicle slowed to 25 mph. In all, 24,000 vehicle miles of travel (VMT) data were collected. More than 8,000 lane changes were identified in the full data set and then classified by urgency and severity. Approximately 500 of the more urgent lane change events were analyzed in depth. Video data were recorded on five channels: driver’s face, forward roadway, rear view, and two side views. Data were saved using 8-mm videotapes. Besides the video data, the vehi- cle network information collected speed, accelerator, brake pedal, steering wheel, and turn signal data, as well as lateral acceleration, radar-collected information (one front and two rear sensors), and reduced data, such as eyeglance behavior and road type and geometry (6). Project 5: Road Departure Crash Warning System Field Operational Test The project was conducted under a cooperative agreement between U.S. DOT and the University of Michigan Trans- portation Research Institute (UMTRI) and its partners: Vis- teon Corporation and AssistWare Technologies. This FOT was designed to assess a Road Departure Crash Warning System (RDCWS). Two areas were assessed: safety-related changes in driver performance that may have been attributed to the sys- tem and levels of driver acceptance in key areas. Testing involved 11 passenger sedans equipped with the RDCWS and a DAS that collected performance, video, and audio data. Seventy-eight drivers participated for 4 weeks each, and the resulting data set captured 83,000 mi of driving. Analysis showed that drivers improved lane-keeping and reduced lane excursions while the RDCWS was active. Driver acceptance of the system was relatively positive, especially for the lateral drift component of the system. Two video cameras were mounted on the vehicle’s A-pillar: one forward-looking and one aimed at the driver. The inside camera also had a set of infrared light-emitting diodes (LEDs) that provided nighttime illumination of the driver’s face. The images of the driver’s face camera were captured in three modes: at 0.5 Hz when the data system was on, during an RDCWS alert captured for 8 s (4 s before and 4 s after theevent), and a batch of 50 driver images spaced every 0.2 s for 5 min at the beginning of the trip and every 5 min thereafter. In addition to the video data, roughly 400 separate data sig- nals were collected, including data for vehicle and driver infor- mation, vehicle position, heading and motion, driver control inputs, RDCWS driver displays, RDCWS intermediate values, roadway environment, RDCWS and subsystem diagnostic information, RDCWS radar data, and audio data. The data set has a valid time stamp that can be used to link to weather data and to link valid GPS data to other environ- mental data, such as traffic count and work zones (7). Project 6: The 100-Car Study The 100-Car Study was undertaken by VTTI, which collected large-scale naturalistic driving data from 100 vehicles in north- ern Virginia for approximately 18 months (12 to 13 months per vehicle). Drivers were given no special instructions, and the majority (78 out of 100) drove their own vehicles. The result- ing data set has 6.4 terabytes (TB) of approximately 2 million VMT from 241 primary and secondary driver participants with a 12- to 13-month data collection period for each vehicle. The 8,295 incidents recorded included 15 police-reported crashes, 67 other crashes, and 761 near crashes. A variety of crash risk factors were analyzed. Continuous video was collected on four channels at 30 Hz: driver’s face, instrument panel (over driver’s right shoulder), forward roadway, and rear roadway. The final data set contains 43,000 h of video. Vehicle network information (speed, brake pedal, throttle, and turn signal); GPS (latitude, longitude, and heading); and X, Y, and Z acceleration were also collected. For- ward radar and rear radar were used to collect surrounding information. Data reduction generated other data, such as driver status, traffic flow, vehicle status, seat belt usage, and road type and geometry. Because of a malfunction in the GPS subsystem, the time data are unreliable. Consequently, it is not possible to link some environmental data from external databases, such as work zone data and traffic condition data. The weather variable that has been coded in the reduced data set is available (8–9). Project 7: Drowsy Driver Warning System Field Operational Test The Drowsy Driver Warning System (DDWS) study was con- ducted by VTTI. The main purpose was to examine the effec- tiveness of a mechanism that alerted drivers that they were about to fall asleep (monitored using a PERCLOS meter). VTTI instrumented 34 trucks with an experimental warning system, video cameras, and a DAS. The final data set included 2.3 million VMT, 12 TB of data, and 46,000 h of driving.

11Continuous video collected data on four channels at 30 Hz: driver’s face, forward roadway, left rear, and right rear. Addi- tional data included vehicle network information (speed, brake pedal, throttle, and turn signal), GPS (latitude, longitude, and heading), lateral and longitudinal acceleration, forward radar- collected data, and sleep quantity (measured by an activity wristwatch). After the data reduction, 16 crashes and 136 near crashes were identified. Data were reduced to identify events based on such information as: • Lateral acceleration; • Longitudinal acceleration; • Lane deviation; • Normalized lane position; and • Forward TTC. The following events were identified: baseline driving epoch, crash, crash: tire strike (defined as any physical contact of tires with other objects on the road), near crash (evasive maneuver), crash-relevant conflict (evasive maneuver), crash-relevant conflict (no evasive maneuver), or nonconflict. Other vari- ables, such as seat belt usage, date, time, light, weather, work zone, roadway condition, and traffic density, were also coded. The status of the vehicle and driver before events was coded as well (10). Project 8: Naturalistic Truck Driving Study Conducted by VTTI, the Naturalistic Truck Driving Study (NTDS) attempted to examine the crash risk factors of commercial truck drivers. VTTI instrumented eight tractor trailers to monitor truck-driving behavior. The data set includes 735,000 VMT data, which amounts to 6.2 TB and 14,500 h of driving. Almost 3,000 critical events, such as crashes, illegal maneuvers, and unintentional lane deviations, were analyzed. Continuous video data were collected on four channels: driver’s face, forward roadway, left rear, and right rear. Addi- tionally, the final data set has vehicle network information (speed, brake pedal, throttle, and turn signal), GPS (latitude, longitude, and heading), lateral and longitudinal accelera- tion, forward and rear radar-collected data, and sleep quan- tity (measured by an activity wristwatch). Data were reduced based on the following triggers: • Longitudinal acceleration (LA); • Swerve; • TTC; • Lane deviation;• Critical incident button; and • Analyst identified. Events identified from the data reduction included crash, crash: tire strike, near crash, crash-relevant conflict, uninten- tional lane deviation, and illegal maneuver. After the data reduction, five crashes, 61 near crashes, 1,586 crash-relevant conflicts, 1,215 unintentional lane deviations, and 5,069 base- lines were identified (11). Project 9: Naturalistic Driving Performance During Secondary Tasks The purpose of the study, which was conducted by UMTRI, was to determine the frequency and conditions under which driv- ers engage in secondary behaviors and to explore the relation- ship these behaviors might have to driving performance. Data from 36 drivers involved in a naturalistic driving study were divided into three age-groups and analyzed to determine the frequency and conditions under which drivers engage in sec- ondary behaviors, such as eating, drinking, and using a cellular phone. Mean ages for drivers in this study were 25.1, 45.6, and 64.2 for the younger, middle, and older age-groups, respec- tively. The data collected were also analyzed to explore the rela- tionship these behaviors might have to driving performance. A video camera was mounted to the inside of the vehicle’s A-pillar and captured 5-s images of the driver’s face at 10 frames/s at 5-min intervals. Researchers examined a repre- sentative sample of 18,281 video clips from the FOT. The sample was not associated with any RDCWS alerts, repre- sented driving at least 25 mph, and included drivers with at least 10 qualifying video clips. Researchers coded 1,440 5-s video clips of the drivers’ faces for the occurrence of specific secondary behaviors and the duration of glances away from the forward scene. Other performance data from instrumented vehicles were used to calculate the variability of the steering angle, the mean and the variability of lane position, the mean and the variabil- ity of throttle position, and the variability of speed (12). Project 10: Effect of In-Vehicle Video and Performance Feedback on Teen Driving Behavior The study was conducted with 26 participants from a high school in rural Iowa. Study periods consisted of a 9-week base- line period followed by 40 weeks of video collection and feed- back and 9 weeks of video collection without immediate feedback. The study found that teen drivers showed a statisti- cally significant decrease in triggering behaviors between the

12feedback and nonfeedback conditions, possibly indicating that drivers became aware of their unsafe driving behaviors and learned to improve their driving. The study used a DriveCam camera mounted to the wind- shield underneath the rearview mirror. The DriveCam is a palm-sized device that integrates two cameras (in-cab and for- ward view) and a wireless transmitter. Video data are continu- ously buffered 24 h per day but only write to memory when a threshold in latitudinal or longitudinal force is exceeded. Twenty seconds of data (10 before and 10 after each “event trigger”) were recorded. Event triggers included any event that exceeded g-forces of .055 for lateral movement or 0.50 for lon- gitudinal movement. If an event occurred, the drivers were given immediate feedback. In this data set, weather was coded as clear or cloudy; fog; rain; mist; snow, sleet, or hail; or smoke or dust. Because no GPS data were collected, location-related environmental data cannot be linked to the data. However, the data were reduced such that extensive environmental data (e.g., traffic condition, work zones, and driver behavior data) are coded in the reduced database by reductionists (13–14). Project 11: Naturalistic Teen Driving Study The Naturalistic Teen Driving Study (NTNDS) was conducted by VTTI. The primary purpose of the study was to evaluate and quantify crash risk among teen drivers. VTTI installed DASs in 42 cars primarily driven by newly licensed teenage drivers in the New River Valley area of Virginia. Naturalistic driving data of the teens and a parent of each teen were collected during the course of 18 months. The resulting data set has 500,000 VMT, amounting to 5.1 TB of data. Continuous video was collected on four channels: driver’s face, instrument panel (over driver’s right shoulder), forward roadway, and rear roadway. Two additional cameras would periodically activate for a few seconds at a time. These cameras provided views of the vehicle’s entire cabin (blurred to protect passenger identities) and the lap area of the back seat. Other data, such as GPS (latitude, longitude, and heading); X, Y, and Z acceleration; forward radar-collected data; and video-based lane tracking data, as well as reduced data (e.g., eyeglance behavior, time-of-day and ambient lighting, road type and geometry, and traffic density), were available in the resulting database (15). Project 12: Cooperative Intersection Collision Avoidance System for Violations Infrastructure During the Cooperative Intersection Collision Avoidance Sys- tem for Violations (CICAS-V), a VTTI and Collision Avoid-ance Metrics Partnership (CAMP) collaborative project, the first study was an infrastructure-based effort monitoring sig- nalized and stop-controlled intersections. The study was undertaken to model stopping behavior and the kinematic factors that could lead to intersection violations. Continuous video was collected. Stop-controlled inter- sections generally had one camera focused on one particular approach. Signalized intersections had four channels of video, one for each approach. In total, 1.5 TB of video and radar data were collected. Other data collected included lateral speed, lateral and longitudinal acceleration, lateral and longitudinal range, and lane tracking for approaching objects (16–17). Project 13: Pilot Study to Test Multiple Medication Usage and Driving Functioning The study was performed by TransAnalytics for the National Highway Traffic Safety Administration (NHTSA). Its purpose was to explore the relationship between polypharmacy and driving functioning through separate but related research activ- ities. Driver performance evaluations, brake response time, and functioning screening measures were conducted for the study sample; drug profiles were documented through a “brown bag” review by a licensed pharmacist. Field evaluation occurred on routes in residential communities in the Hockessin, Delaware, and Parkville, Maryland, vicinities. Two miniature video cameras were used: one for the driver’s face view and one for the forward road view. Cameras were used in the field study of driver performance of 44 older adults. Additionally, cameras were used in private cars of a subsample of five individuals. The video data included the Advanced System Format (ASF) with 704 × 496 resolutions and a 12-Hz frame rate. Each trip was recorded in 10- to 100-s snippets (depending on the amount of motion in the video), which were later combined and rendered in postprocessing to produce single clips for subsequent video coding analysis. Recorders were set to start recording automatically when powered on and to stop recording when no motion was detected in the driver face view camera for at least 30 s. Other data, such as driving speed, brake response time, GPS, onboard diagnostics (including vehicle speed, throttle position, and engine speed), and date and time, were also recorded. The lane-changing behavior of the drivers was manually recorded by researchers in this study (18). Project 14: Older Driver Field Operational Test The purpose of the FOT, which is being conducted by VTTI, is to study older drivers’ driving behavior. The data collection process is still ongoing. The estimated resulting data set should have 4,867 h of video data and 2.5 TB of

13video and binary data collected from 131,400 vehicle miles of travel. Four cameras are used to collect driving data: forward, rear, driver’s face, and instrument panel over the driver’s shoulder. Other data collected include latitude and longitude acceleration, forward radar-collected data, lanetracker data that tracks lane changing, GPS location, and acceleration. Project 15: Cooperative Intersection Collision Avoidance System for Violations Pilot Field Operational Test The Cooperative Intersection Collision Avoidance System for Violations (CICAS-V) study was conducted by VTTI. This was the second study performed during the CICAS-V project. It was a pseudonaturalistic field test of a collision warning system for both effectiveness and driver acceptance. Continuous video was collected on four channels: driver’s face, instrument panel (over the driver’s right shoulder), for- ward roadway, and rear roadway. The study collected 214 giga- bytes (GB) of data, which amounted to 194 h of data. Other information, including vehicle network information (speed, brake pedal, throttle, and turn signal); GPS (latitude, longi- tude, and heading); X, Y, and Z acceleration; forward and rear radar-collected data; and reduced data, such as eyeglance behavior and map-matching variables, were also available. For applicable intersections, only the distance to stop bar, time to intersection crossing, lane number, and signal phase were also gathered (19). Project 16: Volvo Driving Behavior Field Operational Test The Swedish manufacturer Volvo is conducting an ongoing study to compile a variety of data about driving behavior. The research project is part of the European Union (EU) project called EuroFOT, in which Volvo Cars and the SAFER Vehicle and Traffic Safety Centre at Chalmers University of Technol- ogy are engaged. The overall goal is to develop a safer, cleaner, and more efficient road transportation system in Europe. The study started in May 2008 and is expected to last 3 years. Approximately 100 Volvo V70 and XC70 cars will be involved in the data collection. Cameras to record the driver’s head and eye movements, as well as the view of the road and behind the car are installed in the car to collect video data. A data logger will also be used to record the information from the safety features in the car. Sys- tems to be tested include Collision Warning with Auto Brake (CWAB), ACC, Lane Departure Warning System (LDWS), Driver Alert Control (DAC), and Blind Spot Information Sys- tem (BLIS).Concluding Remarks As demonstrated in this chapter, there have been significant efforts to gather naturalistic driver behavior using video and other sensor systems. These data sources will be analyzed in more detail in Chapter 3. References 1. Dingus, T., V. Neale, S. Garness, R. Hanowski, A. Keisler, S. Lee, M. Perez, G. Robinson, S. Belz, J. Casali, E. Pace Schott, R. Stickgold, and J. A. Hobson. Impact of Sleeper Berth Usage on Driver Fatigue. Report RT-02-050. FMCSA, 2001. 2. Impact of Sleeper Berth Usage on Driver Fatigue: Final Report. Report FMCSA-MCRT-02-070. FMCSA, 2002. 3. Neale, V. L., G. S. Robinson, S. M. Belz, E. V. Christian, J. G. Casali, and T. A. Dingus. Impact of Sleeper Berth Usage on Driver Fatigue, Task 1: Analysis of Trucker Sleep Quality. Report DOT MC 00 204. Office of Motor Carrier Safety, FHWA, 1998. 4. University of Michigan Transportation Research Institute. Auto- motive Collision Avoidance System Field Operational Test Report: Methodology and Results. Report DOT HS 809 900. NHTSA, 2005. 5. Henning, M., O. Georgeon, and J. Krems. The Quality of Behavioral and Environmental Indicators Used to Infer the Intention to Change Lanes. Proc., 4th International Driving Symposium on Human Factors in Driver Assessment, Training, and Vehicle Design, Stevenson, Wash., 2007, pp. 231–237. 6. Lee, S. E., E. C. B. Olsen, and W. W. Wierwille. A Comprehensive Examination of Naturalistic Lane Changes. Report DOT HS 809 702. NHTSA, 2004. 7. University of Michigan Transportation Research Institute. Road Departure Crash Warning System Field Operational Test: Methodology and Results. NHTSA, 2006. 8. Dingus, T. A., S. G. Klauer, V. L. Neale, A. Petersen, S. E. Lee, J. Sud- weeks, M. A. Perez, J. Hankey, D. Ramsey, S. Gupta, C. Bucher, Z. R. Doerzaph, J. Jermeland, and R. R. Knipling. The 100-Car Naturalistic Driving Study, Phase II: Results of the 100-Car Field Experiment. Report DOT HS 810 593. NHTSA, 2006. 9. Neale, V. L., S. G. Klauer, R. R. Knipling, T. A. Dingus, G. T. Holbrook, and A. Petersen. The 100-Car Naturalistic Driving Study, Phase 1: Experimental Design. Report DOT HS 809 536. NHTSA, 2002. 10. Hanowski, R. J., M. Blanco, A. Nakata, J. S. Hickman, W. A. Schaudt, M. C. Fumero, R. L. Olson, J. Jermeland, M. Greening, G. T. Hol- brook, R. R. Knipling, and P. Madison. The Drowsy Driver Warning System Field Operational Test, Data Collection: Final Report. Report DOT HS 811 035. NHTSA and Virginia Tech Transportation Institute, Blacksburg, Va., 2008. 11. Blanco, M., J. S. Hickman, R. L. Olson, J. L. Bocanegra, R. J. Hanowski, A. Nakata, M. Greening, P. Madison, G. T. Holbrook, and D. Bowman. Investigating Critical Incidents, Driver Restart Period, Sleep Quantity, and Crash Countermeasures in Commercial Vehicle Operations Using Naturalistic Data Collection. FMCSA, 2008. 12. Sayer, J., J. Devonshire, and C. Flanagan. Naturalistic Driving Perfor- mance During Secondary Tasks. Presented at 4th International Dri- ving Symposium on Human Factors in Driver Assessment, Training, and Vehicle Design, Stevenson, Wash., 2007. 13. McGehee, D., M. Raby, C. Carney, J. D. Lee, and M. L. Reyes. Extending Parental Mentoring Using an Event-Triggered Video Intervention in Rural Teen Drivers. Journal of Safety Research, Vol. 38, 2007, pp. 215–222.

1414. McGehee, D. V., C. Carney, M. Raby, J. D. Lee, and M. L. Reyes. The Impact of an Event-Triggered Video Intervention on Rural Teenage Driving. Presented at 4th International Driving Symposium on Human Factors in Driver Assessment, Training, and Vehicle Design, Stevenson, Wash., 2007. 15. Lerner, N., J. Jenness, J. Singer, S. G. Klauer, S. Lee, M. Donath, M. Manser, and M. Ward. An Exploration of Vehicle-Based Monitoring of Novice Teen Drivers: Draft Report. Virginia Tech Transportation Insti- tute, Blacksburg, Va., 2008. 16. Doerzaph, Z. R., V. L. Neale, J. R. Bowman, and K. I. Wiegand. Live Stop-Controlled Intersection Data Collection. Virginia Transportation Research Council, Charlottesville, Va., 2007.17. Doerzaph, Z. R., V. L. Neale, J. R. Bowman, D. C. Viita, and M. A. Maile. Cooperative Intersection Collision Avoidance System Limited to Stop Sign and Traffic Signal Violations: Subtask 3.2 Interim Report. NHTSA, 2008. 18. Staplin, L., K. H. Lococo, K. W. Gish, and C. Martell. A Pilot Study to Test Multiple Medication Usage and Driving Functioning. Report DOT HS 810 980, NHTSA, 2008. 19. Neale, V. L., Z. R. Doerzaph, D. C. Viita, J. R. Bowman, T. Terry, R. Bhagavathula, and M. A. Maile. Cooperative Intersection Collision Avoidance System Limited to Stop Sign and Traffic Signal Violations: Subtask 3.4 Interim Report. NHTSA, 2008.