Development and analysis of vehicle trajectories and speed profiles along horizontal curves
Vehicle crashes at horizontal curves account for 42 percent of lane departure crashes on roadways. The risks of crashes on horizontal curves have been reported to be about 1.4 to 5 times greater than that for tangent sections. Twenty-five percent of the total number of fatal crashes in the United States occurred on horizontal curves. Driver error due to situational complexity has been cited as major factors run-off-road or cross-centerline crashes at horizontal curves. Major causes of crashes on horizontal curves include abrupt changes in vehicle operating speed and failure to maintain proper lane position within the horizontal curve, as well as other factors. The literature documents many investigations of these factors. This dissertation presents a methodology to develop and analyze vehicle trajectories and speed profiles at horizontal curves. Key research objectives included evaluating appropriate methods and equipment to gather and analyze data for individual vehicles, developing statistical models to predict vehicle speed and lateral position, and comparing speeds and lateral positions of different types of vehicles as they traverse a horizontal curve.
To compare the accuracy and efficiency of commonly used traffic data collection methods, pneumatic road tubes and video data recording systems were used to gather data related to vehicle speed and lateral position. After initial tests on a closed parking lot, a more extensive experiment was conducted on a horizontal curve that was closed to traffic. The setup for this had seven stations with each station consisting of a set of pneumatic tubes placed in a z-configuration, and a trailer with the video equipment placed on the roadway shoulder. The video equipment consisted of 4 cameras: 2 cameras on a mast at about 35 feet above the roadway and 2 cameras placed on the trailer at a height of about 7 feet above the roadway surface. Observations were made during day time and night time based on capabilities afforded by the equipment. Potential impacts of the data collection equipment on driver behavior were also examined. A repeated measures analysis of variance was used to compare vehicle trajectories and speed profiles to evaluate the accuracy of the measurements. The results from this study showed that pneumatic road tubes could be used to obtain vehicle speeds, lateral positions and trajectories along horizontal curves at a high degree of accuracy in a cost effective and time efficient manner. The presence of the trailer with the video recording equipment seemed to affect driver behavior. Also, it was felt that 7 stations (with z-configurations) may affect driver behavior. So, fewer stations would minimize such impacts on driver behavior.
In the second study, vehicle, roadway, and environmental factors at two horizontal field locations are analyzed (one urban and one rural). An OLS model is developed to predict vehicle speed and lateral position along a curve based on a vehicle's curve entry speed and lateral position. Variables such as direction of travel, time of day, type of vehicle, pavement condition (wet/dry), and opposing vehicles are found to significantly influence vehicle trajectories and speeds.
The third study of this dissertation presents an exploratory analysis of vehicle trajectories and speed profiles. Trajectory data from both curve sites primarily reveal cutting or normal pathways. The analyses show that deviations vehicle speeds and lateral position varied as drivers traverse the curve, and that these changes were statistically significant at the 95 percent confidence level. Further, the means of the maximum speed and lateral deviation of passenger cars were greater than those for heavy vehicles along the horizontal curves.