In 2005, Congress authorized and funded the second Strategic Highway Research Program (SHRP2) Naturalistic Driving Study (NDS) with the goals of improving safety and reliability for motorists and workers, enabling transportation agencies to improve their infrastructure more quickly, targeting and efficiently allocating resources, and enhancing existing processes. The SHRP2 NDS is the largest study of its kind with over 3,100 primary drivers and 3,000 vehicles across 6 sites within the United States. The Virginia Tech Transportation Institute (VTTI) developed the data acquisition systems installed in participating vehicles and continues to house all data acquired throughout the NDS. Data included but were not limited to Global Positioning System coordinates, speed, brake and acceleration behavior, driver demographics, detailed event descriptions, and video feeds to the front and rear of the vehicle and on the driver’s face and hands. The accompanying Roadway Information Database (RID) includes detailed roadway data collected on more than 12,000 centerline mi of highways in and around the site, including but not limited to crash histories, traffic and weather conditions, road type, and present signage. Used together, data collected during the NDS and for the RID can be leveraged for a new perspective on driving behaviors. Intersections provide the setting for a large number of traffic incidents. Overall, 40 percent of crashes in the United States occur at intersections; a total of 57 percent of fatalities from 1997— 2004 occurred at stop-controlled intersections, of which 61 percent occurred in rural areas. Factors believed to contribute to these incidents include inadequate surveillance, failure to obey/yield, driver inattention, and speed. In 2000, researchers in Kansas hypothesized that the majority of such collisions occur because drivers “did not see oncoming vehicles or failed to accurately estimate the speeds of oncoming vehicles on the major roadway.” A naturalistic environment provides an opportunity for new insight into stopping and scanning behaviors at intersections. The main objective of this research was to produce actionable insight into transportation safety by leveraging the SHRP2 databases. Specifically, this research aimed to explore and quantify the stopping and scanning behaviors of drivers as they approached and entered rural high-speed intersections. The use of NDS data represents an important addition to the body of knowledge concerning driver behavior at intersections by leveraging the higher degree of ecological validity relative to driving simulators. Secondary objectives include assessing SHRP2’s ability to address further questions of safety and increasing awareness and understanding of relevant analysis techniques and methods. Overall, 40 percent of crashes in the United States occur at intersections; a total of 57 percent of fatalities from 1997—2004 occurred at stop-controlled intersections, of which 61 percent occurred in rural areas. Factors believed to contribute to these incidents include inadequate surveillance, failure to obey/yield, driver inattention, and speed. This study used naturalistic driving data collected under the second Strategic Highway Research Program to explore drivers’ brake and glance patterns on approach to rural high-speed, stop-controlled intersections from the minor route. Brake distance was found to be sufficiently predicted by brake speed (the speed at which the driver was moving upon initial brake activation). At an average brake speed of 61.7 mi/h, participants first applied the brakes at an average distance of 328.7 ft from the intersection. Older participants (ages 45 to 84) applied the brakes farther upstream, especially at higher speeds, than their younger counterparts (ages 18 to 44). The probability of making a complete stop was found to vary significantly with average annual mileage (AAM) and expressed risk associated with performing rolling stops. Participants with higher AAM were found more likely to make complete stops. Intersection approaches were divided into five 98.4-ft segments, and total glance duration to eight regions of interest (ROIs) within each segment were analyzed. Drivers spent nearly the entire approach glancing to the forward ROI until they were 98.4 ft from the intersection. Between 0 and 98.4 ft, drivers spent an average of 5.1 s scanning the intersecting roadway; a total of 86.5 percent of all intersection scanning occurred in the last 98.4 ft of the approach. A novel difference was found among intersection crossings according to the type of stop performed. Drivers who came to a complete (0 mi/h) stop spent just 39.2 percent of their prestop time scanning the intersection, while rolling stoppers spent 74.5 percent. This suggests that complete stoppers focus on getting to the intersection and then stop, scan, and proceed, whereas rolling stoppers scan the intersection prior to arrival so that they can proceed at higher speeds while maintaining a perception of safety. (Author/publisher)
Samenvatting