Merging Area Design of Dedicated Lanes for Autonomous Vehicles Based on the Human-Like Lane-Changing Characteristics

With the development of intelligent transportation, setting up dedicated lanes (DL) for autonomous vehicles (AVs) has received extensive attention to ensure the operational efficiency of AVs. This necessitates the establishment of dedicated merging areas. In this paper, based on the Ubiquitous Traffic Eyes trajectory dataset, this study first uses a combination of the K-Medoids clustering algorithm and Dynamic Time Warping (DTW) algorithm to classify the cooperation level of the following vehicle on the target lane during the lane change. Next, this study analyzes the lane-changing behavior characteristics under different levels of cooperation. Finally, taking the cooperative lane change behavior as the characteristic of human-like lane change behavior for AVs, a method for calculating the length of merging zones for DL for AVs is proposed. An example validation is conducted, and comparative work on the length of merging zones for DL for AVs is carried out by utilizing similar scenarios in highway merge areas. The example validation demonstrates that the proposed calculation method accurately calculates the length of merging zones for dedicated lanes for AVs.