Institute of Transportation, MOTC

       Unsignalized intersections play a significant role in the domestic transportation network. Due to the absence of traffic signal control and the complexity of traffic flow, these intersections encounter issues related to safety and efficiency. According to statistics from the Ministry of Transportation, in the year 2022, nearly 376,000 traffic accidents occurred in Taiwan, with approximately 213,000 of these incidents transpiring at intersections. Within intersection-related accidents, 47% took place at unsignalized intersections, underscoring the noteworthy safety concerns associated with these intersections. Nonetheless, there has been limited research on unsignalized intersections, potentially stemming from the complexity of traffic flow and the challenges in collecting traffic characteristics data, making quantitative analysis difficult for subsequent studies.

       In recent years, with the rapid advancement of unmanned aerial vehicles (UAVs) and the improvement of image recognition and tracking technologies, an increasing number of studies have utilized UAV-captured traffic trajectory data for analyzing traffic characteristics and behaviors. In comparison to traditional methods of collecting traffic data, trajectory data not only offer temporal insights into traffic patterns but also provide spatial information, allowing for a more comprehensive exploration of variations in traffic characteristics at different locations on the road. Therefore, in order to quantitatively assess the traffic characteristics of unsignalized intersections, this research uses traffic trajectory data collected by UAVs and thoroughly investigate the resultant traffic patterns and behaviors. The trajectory data, with its high fidelity, allows for the generation of various traffic characteristics for each time frame, including position, speed, acceleration, and vehicle trajectory paths. Additionally, risk-related and unsafe driving behaviors, such as speeding, crossing double yellow lines, abrupt deceleration, and two-vehicle conflicts, can be computed based on set criteria.