Observed lane-choice and lane-changing behaviors on an urban street network using video-capture-based approach and suggested structures of their models.

摘要

Multiple vehicle trajectories provide a sound base for analyzing vehicle-based travel behaviors and thus are used for microscopic modeling of traffic on an urban street network. In the past, however, the difficulty of collecting vehicle trajectory data has been an obstacle to deriving verifiable models from empirical data. The research presented in the dissertation proposes the video-capture-based methodology for extracting multiple vehicle trajectories with a computer software package, Vehicle Video-Capture Data Collector (VEVID), developed by the author. With the use of a video-capture board and VEVID, it is feasible to effectively, easily, and inexpensively collect multiple vehicle trajectory data simultaneously under multiple-lane conditions. Secondly, the author conducted videotaped field observations, and classified characteristics of lane-choice and lane-changing behaviors on an urban street network based on key findings from observations. Then rules and heuristic structures of a lane-choice model, a lane-changing model, and a lane-assignment model were presented. Meanwhile, values corresponding to parameters used in these models were recommended. Thirdly, the author explored the integration of a lane-assignment model into a prospective ATIS (Advance Travel Information System)-oriented simulation system on a theoretical basis. Finally, based on analysis of observation-based inherent characteristics of route-choice and route-switch phenomena that occur in an urban street network, some thoughts and related model structures are proposed. This information will help simulation system developers to clearly specify the relationship and structure of vehicle-based travel behaviors to the system being simulated so that a developed simulation system may represent travel behaviors and dynamic traffic assignment at the level of lanes.