In a groundbreaking study poised to reshape urban traffic management, researchers at The Hong Kong Polytechnic University, in collaboration with New York University, have developed a novel framework to estimate citywide traffic flows using Global Open Multi-Source (GOMS) data. This research, published in Communications in Transportation Research on November 18, 2025, promises to enhance the accuracy and applicability of traffic flow predictions across diverse urban landscapes.
The study addresses a critical challenge in smart city development: the high cost and limited coverage of traditional traffic sensors. Existing methods often rely on proprietary datasets from individual cities, leading to a trade-off between accuracy and generality. The new framework, however, leverages publicly available global data to overcome these limitations.
Revolutionizing Traffic Flow Estimation with GOMS Data
Associate Professor Wei Ma’s team, including Dr. Zhenjie Zheng and Dr. Zijian Hu, has identified a strong correlation between road traffic flow and various urban characteristics. These include urban building structures, human activity patterns, infrastructure connectivity, and dynamic traffic conditions. By integrating these elements, the researchers have developed an attention-based graph neural network capable of fusing and extracting geographical and demographic information from GOMS maps.
The model’s effectiveness was validated using data from 15 heterogeneous cities across Europe and North America, demonstrating stable and satisfactory estimation accuracy. The results show that the average, minimum, and maximum mean absolute percentage errors among all cities are 23%, 17%, and 27%, respectively.
The Power of GOMS Map Images
Unlike traditional tabular data, GOMS map images provide a more comprehensive representation of traffic flow dynamics. These images naturally include spatial relationships, such as road topology and bottleneck locations, which are crucial for extracting rich spatial information. Moreover, GOMS maps contain extensive contextual information, offering insights into geographic features and demographic patterns that are difficult to capture in tabular form.
This innovative approach allows for efficient handling of large datasets, enhancing the scalability of traffic flow estimation across extensive networks. The use of deep learning methods trained on map images also ensures robust generalization across various urban studies, even under unseen network conditions.
Implications for Smart Cities and Beyond
The introduction of GOMS data into traffic flow estimation marks a significant advancement in urban planning and smart city development. By breaking the trade-off between accuracy and generality, this framework offers a scalable solution that can be applied globally, regardless of local data constraints.
Preliminary research utilizing GOMS data has also shown promise in estimating land-use patterns and air pollution, indicating its potential for broader applications in urban studies. The integration of such data can lead to more informed decision-making and improved urban infrastructure planning.
Expert Opinions and Future Directions
Experts in the field have praised the study for its innovative use of open-source data and its potential to transform urban traffic management. Professor Monica Menendez from New York University, a collaborator on the project, emphasized the importance of such interdisciplinary efforts in addressing complex urban challenges.
“The use of GOMS data in traffic flow estimation is a game-changer for smart cities. It allows us to harness global data resources to create more accurate and adaptable models, ultimately leading to more efficient and sustainable urban environments,” said Professor Menendez.
Looking ahead, the research team aims to refine the model further and explore its application in other areas of urban planning. The continued development of GOMS-based frameworks could pave the way for more comprehensive and integrated smart city solutions.
About the Publishing Journal
Communications in Transportation Research, the journal that published this study, is co-published by Tsinghua University Press and Elsevier. It is recognized for its high-quality, peer-reviewed research that contributes significantly to emerging transport systems. The journal has been indexed in several prestigious databases and is known for its commitment to open access, with Tsinghua University Press covering publication fees for 2025.
Launched in 2021, the journal aims to foster academic exchange and development between China and the global community, showcasing innovative achievements in transportation and related fields. It has consistently ranked highly in international journal rankings, reflecting its impact and relevance in the field.
As urban environments continue to evolve, the integration of global data into traffic management systems will be crucial for building smarter, more efficient cities. This study represents a significant step forward in that direction, offering a scalable and adaptable solution to one of the most pressing challenges in urban planning today.
