| 英文摘要 |
In freeway weaving areas, the four different types of traffic streams from and to the mainline and ramps interact within the same segment, generating turbulence that reduces operational capacity. Because the traffic flow characteristics of weaving areas are highly complex and empirical data are difficult to collect, previous studies on weaving-section traffic remain limited. Benefiting from recent advances in survey technology, this study employs an unmanned aerial vehicle (UAV)–based aerial photography approach combined with computer-vision image recognition to analyze traffic flow characteristics and collects data from three weaving sections of different configurations. The results of the study revealed the following: (1) Drivers tend to increase lane-changing maneuvers after passing the gore nose of the channelization island, resulting in weaving behavior concentrated in the upstream portion of the section; (2) The average speed of the four traffic flows within the weaving section differs, and past methods that combined the four flows into non-weaving flows (freeway-to-freeway, ramp-to-ramp) and weaving flows (freeway-to-ramp, ramp-to-freeway) may oversimplify the situation; (3) Weaving sections may experience traffic congestions for all lanes or only some of the lanes, and this distinction should be considered when evaluating their level of service. |