Nearly 80 percent of car crashes and 65 percent of near crashes involve some form of driver inattention within three seconds before the event.
The ?100-Car Naturalistic Driving Study,? conducted by the National Highway Traffic Safety Administration (NHTSA) and Virginia Tech Transportation Institute (VTTI), tracked drivers in the metropolitan Washington, D.C., area over an 18-month period.
Previous studies have pointed to the high incidence of crashes and near crashes as a result of driver inattention. These studies, however, rely on data collected from incident reports and simulated driving environments. In the 100-car study, unobtrusive devices were used to observe drivers in a natural setting over an extended period.
To determine what behaviors contributed to driver inattention, VTTI installed a system of video cameras and sensor devices throughout 100 vehicles, which included five model sedans and one SUV.
The data collection system consisted of a number of sensor networks, a series of cameras, a data acquisition system (DAS), and real-time incident tracking devices. The DAS unit was mounted under the package shelf for the sedans and behind the rear seat in the SUVs.
Doppler radar antennas were mounted behind special plastic license plates on the front and rear of the vehicle (Fig. 1) .
The sensor network consisted of dozens of sensors positioned in various locations throughout the car. A vehicle network box interacted with the vehicle network. An accelerometer box obtained longitudinal and lateral kinematic information. A headway detection system provided information on leading or following vehicles. A side obstacle system detected lateral conflicts. An incident box let drivers flag incidents for the research team. And, a video-based lane-tracking system measured lane-keeping behavior. Each of the sensing subsystems in the car was independent, so any failures that occurred were constrained to a single sensor type.
Digital video devices provided a visual window into driver behavior. Five cameras monitored the driver's face and driver side of the vehicle, the forward view, the rear view, the passenger side of the vehicle, and the area over the driver's shoulder. Video compression software provided the research team with synchronization, simultaneous display, and archiving of footage.
Real-time incident tracking devices completed the portrait of the participants' driving behavior. Automatic collision notification informed the research team of the possibility of a collision. The research team also used cellular communications to communicate with vehicles on the road to determine system status and position. System initialization equipment automatically controlled system status. And, a Global Positioning System (GPS) subsystem collected information on vehicle position.
An incident pushbutton also was included to provide the driver with the means to record unusual events. The pushbutton device consisted of a camera that captured a view of the driver's face and an audio channel that recorded the driver's voice explaining the nature of the incident (Fig. 2) .
Despite knowledge of these devices, the 241 drivers of the vehicles were involved in 82 crashes, 761 near crashes, and 8295 critical incidents. The most common distraction for drivers was the use of cell phones.