| Title | Automatic intersection map generation task 10 report. |
|---|---|
| Record ID | 61914 |
| Personal Name Creator |
Zheng, Jianfeng; Liu, Henry X.; Parikh, Jay |
| Corporate Creator | Crash Avoidance Metrics Partners |
| Corporate Contributor |
United States. Federal Highway Administration; United States. Joint Program Office for Intelligent Transportation Systems |
| Publisher | United States. Federal Highway Administration; United States. Joint Program Office for Intelligent Transportation Systems |
| Publication Date | 20160229 |
| Language | English |
| Abstract | This report describes the work conducted in Task 10 of the V2I Safety Applications Development Project. The work was performed by the University of Michigan Transportation Research Institute (UMTRI) under contract to the Crash Avoidance Metrics Partners LLC (CAMP) Vehicle to Infrastructure (V2I) Consortium. Participating companies in the V2I Consortium were FCA US LLC, Ford, General Motors, Hyundai-Kia, Honda, Mazda, Nissan, Subaru, Volvo Truck, and VW/Audi. This project investigated the feasibility of automatically generating intersection maps in SAE J2735 MAP format using Basic Safety Messages (BSMs) received by Roadside Units (RSUs). A procedure for classifying vehicle trips through the intersection and subsequently estimating lane centerlines was developed. In addition, a method for associating vehicle movements to the green signal phase was also developed using Signal Phase and Timing (SPaT) messages. BSM data from five intersections used in the Safety Pilot Model Deployment Project were analyzed. The estimated maps were compared to reference maps produced from LIDAR surveys of the intersections. With 48 traffic lanes in total, 43 lanes and associated vehicle movements were correctly identified using the estimation approach. Five lanes were not successfully identified due mainly to a lack of BSM data. For the identified lanes, two measurements of accuracy were calculated: the mean distance of the estimated geometry node points to closest points in the surveyed lane centerline geometry and the maximum distance of estimated points and surveyed lane geometry. On average, the mean distance was found to be 0.5 meter and maximum distance was found to be 1.2 meters between the estimated maps and reference maps. Overall, this project demonstrated the significant potential of using BSM data for estimation of an intersection map as well as association of the vehicle lanes (vehicle movements) to the traffic signal phases. |
| Rosap ID | dot:3601 |
| Rosap URL | https://rosap.ntl.bts.gov/view/dot/3601 |
| TRT Terms | Accuracy; Automatic vehicle detection and identification systems; Center lines; Data collection; Feasibility analysis; Intersections; Laser radar; Mapping; Maps; Traffic lanes; Traffic signal phases |
| General Subjects | Connected vehicle; intersection map; map generation; V2I system; roadside equipment; basic safety message; vehicle to infrastructure communications |
| Geographical Coverage |
United States |
| TRIS Online Accession No |
1640098 |
| Contract Number | DTFH6114H00002 |
| Report Number | FHWA-JPO-16-416 |
| Availability | Intelligent Transportation Systems Joint Program Office |
| Resource type | Tech Report |
| URL | https://ntlrepository.blob.core.windows.net/lib/61000/61900/61914/FHWA-JPO-16-480_20180418.pdf |
| Format | |
| Database | NTL Digital Repository |