The development of transportation system theory is always constrained and driven by the ability to acquire traffic measurement data, which is determined by the experimental nature of traffic science. In the 1940s, traffic data collection was mainly done manually through counting. With the development of perception and the emergence of new advanced equipment, traffic data collection has gradually changed. Nowadays, information technology development represented by mobile internet, ubiquitous sensing and computing, and artificial intelligence is rapidly emerging, providing new opportunities for the intergenerational transformation of traffic science theories. It has become a common consensus in the field of traffic engineering in both academia and industry. The core of the transformation revolution is that people's measurement capabilities for traffic systems have evolved from sampling and aggregate-based measurement to full-scale, real-time, and individual-label-based measurement. In this context, new traffic science datasets are the key to the development of new traffic theories.
To promote research on new traffic science theories, our team proposed the concept of holographic traffic flow data, which describes the behavior of individual movements (moving-stationary) through complete observation and obtains individual continuous motion trajectories through vehicle travel path reconstruction. Finally, a city-level holographic traffic flow dataset was formed through virtual measurement, covering complete and compliant travel behavior information and complete spatial information of individuals in Xuancheng, China for a continuous month.
Experiments have shown that this data reveals that the distribution of flow on the main roads during the morning rush hour in the city is radial, and through correlation calculation, it is found that this data has high consistency with other data sources in terms of cross-section flow and road travel time. In addition, the holographic traffic flow also reflects the day-night characteristics of traffic flow operation, such as lower vehicle speeds at night, which is consistent with prior knowledge. From the perspective of travel, by statistically analyzing individual travel times, it is possible to clearly distinguish different populations with different travel patterns (Figure 1). Through specific statistical analysis of vehicle travel, it was found that 20% of vehicles in the city accounted for 80% of travel, which was verified by the Lorenz law.
The holographic traffic flow data has been made public through the data open platform figshare (https://doi.org/10.6084/m9.figshare.c.5796776.v1). At the same time, to meet researchers' needs for customized data acquisition, our team has developed and opened the Traffic Virtual Measurement Platform (http://vsensor.openits.cn/) (Figure 2), which supports traffic flow detection simulation from both Eulerian and Lagrangian perspectives (such as detecting traffic flow and vehicle speed through virtual loop detectors, and simulating real-time position floating vehicle vb detection). Our team's publicly available city-scale holographic traffic flow dataset can to some extent meet the high-quality data needs of cutting-edge transportation research, promote the development of the transportation field, such as traffic control, path planning, and decision-making processes research.