Complete Vehicle Coverage Trajectory data is of fundamental importance to traffic signal control strategy and Intelligent Transport System (ITS). There are mainly two categories in traffic data collection: stationary and mobile sensing. However, neither can offer near complete spatial and temporal coverage, especially at traffic signals. This study first proposes a full-scale Automated and Connected UAV (AC-UAV) system, which consists of transformable UAVs with automated landing and take-off capabilities, cooperative charging piles, and a fleet management center. Secondly, based on the developed AC-UAV system, introduce a 3-steps methodological framework: Multiple Vehicle Detection algorithm (MVD) based on deep learning, Multiple Vehicle Tracking (MVT) algorithm based on data feature association and trajectories reconstruction in the form of polylines varying in time and direction, as real-time input for traffic flow analysis and service patrol incident detection. The field experiments with an AC-UAV system and real-time surveillance video analysis were conducted in Xi’an, China for complete vehicle coverage trajectory reconstruction at traffic signals. The results show that the proposed framework based on AC-UAV system is capable of conducting mobile complete traffic data analysis tasks and also feasible for large-scale automated city applications.

CICTP