Video -Based Vehicle Trajectory Data Collection

Vehicle trajectory data provide detailed information on microscopic phenomena that can be used for behavioral modeling of car-following, lane-changing, gap acceptance, cooperative driving, etc. Vehicle trajectories also provide detailed data needed for safety research. Usually, the vehicle trajectory data are extracted from high-resolution images - either from video or camera based data collection to develop a holistic dataset of vehicle positions. In recent years, video data collection and subsequent transcription of video data to vehicle trajectories are being attempted by many transportation researchers. While there have been multiple video data collection efforts conducted across the world, there has been little guidance and recommendations on the data platform, data formats and data elements for vehicle trajectory data collection. As part of the Next Generation SIMulation (NGSIM) program, the Federal Highway Administration’s conducted an initial prototype, and three vehicle trajectory data collection efforts. This paper documents the guidelines developed for the NGSIM data collection efforts for obtaining efficiently transcribed vehicle trajectories.