Measuring the Impact of Roadside Features on Road-Departure Crashes and Prioritizing Safety Improvement Projects
-
2024-09-01
-
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final Sept 2023 to Nov 2024
-
Corporate Publisher:
-
Abstract:This study investigated the impact of roadside features on road departure (RD) crashes and provided recommendations for safety improvements on Utah’s highways. The research utilized a computer vision model originally developed in the UDOT project, Automated Safety Assessment of Rural Roadways Using Computer Vision (Mashhadi et al., 2023), which was initially trained on images from Mandli. The model was retrained on Pathway images to extract roadside features from across seven Utah roads, including non-interstate routes (US-6, SR-10, SR-12, US-40, SR-150) and interstate highways (I-15 and I-80). The retrained model achieved high classification accuracies, enabling the extraction of key roadside features such as clear zones, rigid obstacles, side slopes, and safety barriers using advanced computer vision techniques. Safety ratings were assigned based on the algorithm developed in accordance with FHWA roadside rating guidelines. The roadside feature dataset was merged with the crash database obtained from UDOT using the milepoints information to create a comprehensive view of crash occurrences and severities in relation to roadside features. Statistical analysis, including Spearman’s Rank Correlation Coefficient, was conducted to identify the correlation between these features and RD crash rates. The study identified hotspots and provided targeted recommendations for safety improvements, such as installing safety barriers, improving clear zones, implementing high-friction surface treatments, and considering additional safety measures to enhance overall roadway safety.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type: