Title Application Scenarios and Pilot Run of Vehicle and Pedestrians Traffic Flow Information Using UAV Aerial Videography (1/3) – Establishing a Test Plan
Dept Transportation Safety Division
Year 2022
Month 5
Price 500
Summary        This research is the first year of a 3-year (2021-2023) project. The purpose of this project is to continue the results of the previous research and development of traffic conflict analysis tools. In addition to conducting a yearly pilot test program to verify the maturity and limitations of the technology, the project also aims to cooperate with local governments to promote the application of the tool.
In order to analyze the traffic conflicts at intersections and the behavior characteristics of drivers, the preliminary research from 2018-2020 combined new technologies such as drone aerial videography and artificial intelligence image analysis. It used traffic conflict risk analysis technology to establish a traffic conflict analysis tool. In order to verify the maturity and limitations of relevant technologies, this research took an inventory of six common traffic conflicts at intersections through expert and scholar symposiums and surveys by local authorities, including: opportunities for left turn, intersection crossing conflicts, right-turn conflicts, non-signal intersections, opportunities to turn left (advanced analysis), conflicts between people and vehicles, etc. Two pilot testing plans will be conducted every year from 2021-2023.
       This year, it cooperated with 4 municipalities directly under the Central Government to handle two pilot testing plans of opportunity left turn and intersection crossing conflict, a total of 14 intersections, and the analysis of 4 high-risk intersections.
The pilot testing plan proposes three categories of driving behaviors: risky driving behaviors, illegal behaviors, and traffic conflicts, and develops relevant trajectory judgment logic to complete the detection of risky driving behaviors or illegal behaviors, procedures to provide a reference for subsequent application of analytical tools.
       This research also continued to optimize related analysis technologies, including the development of Oriented YOLOv4 to simplify the image analysis process, shorten the processing time, and improve the accuracy of vehicle identification. The traffic conflict analysis software added a vehicle speed heat map, as well as the scale and number of seconds difference in traffic signs.
In addition, in conjunction with the two pilot testing plans of opportunity left turn and intersection crossing conflict, relevant analysis modules and charts have been added to further analyze the possible causes of traffic conflicts to assist road management agencies with early intersection traffic safety analysis and to improve this issue.
Post date 2022/06/14
View count: 113