by Paolo Nesi, Pierfrancesco Bellini and Marco Fanfani (UNIFI, DINFO, DISIT lab, Snap4City)
Snap4City is a digital twin platform for decision makers to operate and plan city activities, using contextual and historical data to support planning and generate problem-solving suggestions. The platform integrates ML/AI analytics, providing seamless suggestions and XAI annotations for decision makers. Recently, interactive 3D digital twin models have been provided for real-time simulation of changes and their effects on KPIs like decongestion and emissions.
Snap4City [L1] is a digital twin platform [L2] designed for decision makers to perform activities of operation and planning in the cities and control rooms. Operation means at least to monitor, control, predict and react in real time to current operational conditions and related events. For this purpose, data are collected and processed and early warning is computed by using a range of data analytics and AI processes. Planning means to assess the city’s condition in terms of contextual and historical data, the desiderata/objectives and goals / key performance indicators to provide support in planning, generating suggestions, solutions to the problems, etc. The operation is always to be performed in real time or quasi real time, while the plan may take time. Recently, what-if analysis and optimisation tools, typically applied in plan are also used in operation for fast reaction to critical conditions, and the time to plan has been considerably shortened. Snap4City provides an integrated solution for data gathering, indexing, computing, and information distribution, thus realising a continuously updated digital twin of the urban environment at global and local scales for monitoring operation and planning. It addresses 3D building models, road networks, Internet of Things entities, points of interest, paths, as well as results from analytical processes for traffic density reconstruction, pollutant dispersion, predictions, and what-if analysis for assessing impact of changes, all integrated into a freely accessible interactive 3D web interface, enabling stakeholder and citizen participation in city decision processes.
The architecture of Snap4City is shown in Figure 1. The multidomain platform allows collecting and automatically processing and representing complex data of several different kinds. For example, traffic flows which are graphs changing over time, origin destination matrices which are maps/graphs changing over time, time series, heatmaps, trajectories and routings, scenarios to be assessed, events and intervention of restoring in the city, etc. The more relevant domains that are presently strongly interconnected are: mobility and transport, environmental, and energy.
Figure 1: Snap4City digital twin conceptual architecture.
Snap4City is an open-source IoT platform coordinated by the DISIT Lab of the University of Florence [L4]. Snap4City is an official FIWARE (an open-source initiative that provides a universal set of APIs that allow the development of smart applications in multiple sectors) and EOSC (European Open Science Cloud) platform and it includes a set of Node-RED libraries. It is at present in operational use on several federated installations. The Snap4City platform is able to manage multiple tenants and billions of data per day with the key focus on interoperability, decision making, in operation and plan. The Snap4City framework is implemented in several smart cities and areas in Italy (Merano, Cuneo, Florence, Lonato del Garda, etc.) and across Europe (Malta, Rhodes, Varna, Limassol, Valencia, Pont Du Gard, Dubrovnik, Mostar, and West Greece, etc.), as well as on almost all continents. The largest installation of the platform manages multi-tenant, advanced smart city IoT/IoE applications and involves 20 organisations, 40 cities and thousands of operators and developers.
The main drivers for sustainability are decongestion of traffic and services, decarbonisation, improvement of services’ accessibility and of security/safety, and possibly the improvement of quality of life. For example, (i) traffic decongestion can be obtained by reducing the number of cars in stop-and-go conditions – this would also imply reducing emissions, depending on the kind of vehicles that are typically in the city, thus stimulating the decarbonisation with e-vehicle diffusion (the actual solutions could be several, from changing the transportation infrastructure, optimising the semaphore cycles, etc.); (ii) making mobility services more accessible would imply having more rides or different paths of public transportations and in some cases increasing emissions; (iii) reducing the critical conditions could improve safety, maybe at the expense of monitoring more carefully all the city areas and thus spending more energy and computational power. A list of artificial intelligence Snap4City tools (AI, explainable AI, XAI tools) for decision makers in improving sustainability of their city is available at [L3].
The first step of those activities is the data ingestion and modelling, field interoperability and data processing area capabilities for collecting data from any source and exchanging data in push toward any brokers, gateways, and services. This area is interoperable with a large number of protocols and formats and enables federation of smart cities. Real-time data, as well as event-driven streams, are ingested using IoT brokers and indexed and shadow stored into a graph database (based on Km4City ontology, and data model for digital twins) and time series storage cluster, thus making them accessible to other consumer processes and brokers. Internal brokers are based on Orion Broker NGSI (also compatible with smart data models, and data spaces). Data can be retrieved in pull or in an event-driven/push way. Node-RED flows are used to enable data interoperability with third party services such as: GIS (Geographic Information Systems), ITS (Intelligence Transport Systems), TV cam services, CKAN open data networks, BIM servers, social media, data gateways, etc. Specific libraries of microservices for data transformation has been developed and made freely available to develop custom-made processing logic.
The ML/AI analytics are in most cases seamless for the decision makers, who are receiving suggestions from AI and also via some XAI annotations. Recently, the provision of interactive 3D digital twin models is actually requested by decision makers for its immediate representation to perceive an awareness of the situations, and to perform real-time simulation of changes, estimating their effects in terms of KPIs on decongestion, travel time, emissions, etc. The specific action is founded as CAI4DSA cascade project of FAIR PE on Future Artificial Intelligence, PNRR.
Links:
[L1] https://www.snap4city.org
[L2] https://digitaltwin.snap4city.org
[L3] https://www.snap4city.org/997
[L4] https://www.disit.org
References:
[1] L. Adreani et al., “Smart city digital twin framework for real-time multi-data integration and wide public distribution,” IEEE Access, IEEE, 2024. pp: 1-27, ISSN: 2169-3536, ISSN: 2169-3536, DOI: 10.1109/ACCESS.2024.3406795
[2] P. Bellini et al., “Snap4City Dashboard Manager: a tool for creating and distributing complex and interactive dashboards with no or low coding,” SoftwareX, Elsevier, 2024. https://www.sciencedirect.com/science/article/pii/S2352711024001006. https://doi.org/10.1016/j.softx.2024.101729
[3] L. Adreani et al., “Smart city scenario editor for general what-if analysis,” Sensors, MDPI, 2024. https://www.mdpi.com/1424-8220/24/7/2225/pdf
Please contact:
Paolo Nesi, University of Florence, DINFO, DISIT, Snap4City, Italy
