Lateral and Tangential Accelerations of Left Turning Vehicles from Naturalistic Observations 2019-01-0421
When reconstructing collisions involving left turning vehicles at intersections, accident reconstructionists are often required to determine the relative timing and spacing between two vehicles involved in such a collision. This time-space analysis frequently involves determining or prescribing a path and acceleration profile for the left turning vehicle. Although numerous studies have examined the straight-line acceleration of vehicles, only two studies have presented the tangential and lateral acceleration of left turning vehicles. This paper expands on the results of those limited studies and presents a methodology to automatically detect and track vehicles in a video file. The authors made observations of left turning vehicles at three intersections. Each intersection incorporated permissive green turn phases for left turning vehicles. The authors recorded video of left turning vehicles at each intersection from a small unmanned aerial system (sUAS), and that video was analyzed with a convolutional neural network designed to detect vehicles. The detected vehicles were then tracked over time and the results were analyzed. A total of 86 left turning vehicles were analyzed. In 23 of the observed turns, an oncoming vehicle was also visible in the video. The spatial relationship between the oncoming vehicles and the left turning vehicles was analyzed and the relationship between gap acceptance and acceleration is presented. Accident reconstructionists and traffic engineers can use this data to prescribe realistic values or ranges to accelerations of left-turning vehicles.
Citation: Carter, N., Beier, S., and Cordero, R., "Lateral and Tangential Accelerations of Left Turning Vehicles from Naturalistic Observations," SAE Technical Paper 2019-01-0421, 2019, https://doi.org/10.4271/2019-01-0421. Download Citation
Author(s):
Neal Carter, Steven Beier, Rheana Cordero
Affiliated:
Kineticorp LLC
Pages: 17
Event:
WCX SAE World Congress Experience
ISSN:
0148-7191
e-ISSN:
2688-3627
Related Topics:
Neural networks
Vehicle acceleration
Accident reconstruction
Vehicle dynamics /flight dynamics
SAE MOBILUS
Subscribers can view annotate, and download all of SAE's content.
Learn More »