Curve Safety Improvements Using Mobile Devices and Automatic Curve Sign Detection – Phase I
-
2019-11-01
-
Details:
-
Creators:
-
Corporate Creators:
-
Corporate Contributors:
-
Subject/TRT Terms:
-
Publication/ Report Number:
-
Resource Type:
-
Geographical Coverage:
-
Edition:Final; October 2018 – November 2019
-
Corporate Publisher:
-
Abstract:A disproportionally high number of serious vehicle crashes (25% of fatal crashes) occur on horizontal curves (FHWA, 2019), even though curves represent only a fraction of the roadway network (5% of highway miles) (FHWA, 2016). The MUTCD (Manual on the Uniform Traffic Control Devices) (FHWA, 2012) requires various horizontal alignment warning signs (curve signs) to ensure curved roadway safety. However, current transportation agencies’ practices for inventorying the locations and types of existing curve signs are largely a manual procedure, which is costly, labor-intensive and time-consuming. This report presents a cost-effective, live curve sign inventory system for meeting MUTCD requirements using intra-agency, low-cost mobile devices (e.g. smartphones), existing vehicles, and deep learning and crowdsourcing technologies with a special focus on 1) critically validating an automatic curve sign detection and classification method and 2) assessing and delivering a smartphone-based data collection module. The outcomes will strongly complement current transportation agencies’ curve sign placement operations for meeting MUTCD requirements. A case study, using 13 centerline miles of roadway on State Route 2 and consisting of 471 curve signs in Georgia, was conducted to critically validate the accuracy of detecting and classifying curve signs using the proposed method. Results show 471 curve signs were correctly detected and it has demonstrated that the developed automatic curve sign detection and classification method using deep learning is very promising for implementation. Finally, conclusions and recommendation for future research along with a roadmap for validating, refining, and implementing a cost-effective, live and sustainable curve sign system, are presented.
-
Format:
-
Funding:
-
Collection(s):
-
Main Document Checksum:
-
Download URL:
-
File Type: