Identifying hazardous obstructions within an intersection using unmanned aerial data analysis

Intersections are places where two or more roadways intersect with one another. Federal Highway Administration (FHWA) quoted that more than fifty percent of crashes occur at or near intersections. An at-grade intersection is more accident prone compared to grade separated intersections. Intersection-related crashes are almost 335-times more likely to occur as compared to non-intersection-related crashes caused by vehicles turned with obstructed views. Approach and departure sight triangles are two types of sight triangles that are considered to provide clear visibility for drivers entering and exiting an intersection. Presently, there is no common practice followed by transportation or public work agencies other than considering accident history and public complaints as indicators for the need to improve intersection conditions. This approach warrants a safe and rapid methodology that can effectively identify obstructions within the intersection sight triangles. Unmanned aerial vehicles coupled with close range photogrammetry (UAV-CRP) technology offers a safe, quick, and efficient alternative for obstruction identification compared to traditional methods. In this study, a UAV based methodology is developed to identify obstructions in intersections from the 3-dimensional models developed using the imagery collected from unmanned aerial surveys. Two case studies including a T-intersection and a railroad crossing were considered to demonstrate the developed methodology. This research successfully validated the developed UAV methodology by analyzing departure sight triangles at T-intersection and approach sight triangles at railroad crossing. It is expected to be applied in many other civil engineering applications that are deemed critical due to the presence of obstructions and can potentially save human lives.

Language

  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 01746518
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jun 9 2020 3:04PM