Swiss Mobility System - Digital Twin

Project idea

To support research and policy making, we are creating a field-​specific Digital Twin of Switzerland and its interactions with neighbouring countries. The Digital Twin of the Swiss mobility system will make use of the data sets and models available on Switzerland. In the long run, it will not only make the unique role and contribution of the center to society visible, but it will also allow different groups at ETH and beyond to build on a high-​quality open-source base for their individual and joint research. We are currently developing an Open Digital Twin Platform for this purpose that enhances and supports Open Research Data (ORD) practices in the wild. 

Project background

The concept of Digital Twins was coined in 2003 to conceptualise the matching of the physical world with a digital representation through some kind of data collection process in an industrial context.¹ The idea stuck around and slowly emancipated itself from its industrial roots expanding to various other fields and disciplinary contexts where the simple description makes sense to abstract processes like in smart cities.² While Digital Twins have found wide applications, they have remained underspecified and carry a very different meaning in different communities. For example, in aviation, Digital Twins stand for similitude of the digital representation to the highest possible physical degree.³ Whereas for Smart Cities, Digital Twins only require eventual consistency,² i.e., representing some past abstract state of the city correctly such as traffic load or the population in contrast to properly simulating the physical movement of cars and people as similitude would require.

Are these diverse understandings epistemologically incompatible or can they be studied in a common framework that defines Digital Twins as an open standard? Previous work on Digital Twins is limited to prototypes that demonstrate the idea and variations thereof.^3,4,5,6,7,8 At the same time, industry is creating standards for Digital Twins that are proprietary and could become the norm by the virtue of simply being there like early web browsers and operating systems. We identified the unique opportunity to create an open standard for Digital Twins right now because we are at the crossroads where the required technologies for Digital Twins are maturing, and we have no default yet.

The ongoing debates have identified the Semantic Web⁹ as a core technology to make information exchangeable across systems through knowledge graphs upon which Digital Twins could be based.^4,6 Whereas an ontology is a crucial component to interact with unknown Digital Twins based on a generic format¹⁰ - a Digital Twin is more than just a knowledge exchange platform.¹¹ The systematic nature of Digital Twins makes them a natural candidate to implement FAIR (findable, accessible, interoperable, and reusable) data standards that guide ORD practices and enable reproducible research.¹² In the best case, Digital Twins as an open standard can be a driver of development like the Internet. In the worst case, they could be bogged down in a battle of standards like HD-DVD vs BluRay that slow down adaptation because of uncertainty about which format will eventually become the default. Our project has the opportunity to shift the development towards the best case by developing a consistent open standard that is applicable across domains in cooperation with many stakeholders of a diverse background at the Center for Sustainable Future Mobility (CSFM) and other partners in academia, industry and government.

Industrial players from large multinational companies to consultants are keen to offer their Digital Twin solutions in the hope to fence off significant market shares from the imagined application fields of Digital Twins. The reasons why there is no open standard yet are multitude but include these wild-west assumptions of a large untapped opportunity that can be conquered and fenced off.¹³ Another issue is that the underlying technologies are only now maturing and are often still under development themselves. This makes early Digital Twins fickle and prone to breaking down. Most current actors hope to be a defining force for Digital Twins to come and therefore try to create facts on the ground. In this fast-paced environment, only few truly open initiatives have been formed. Mostly, they revolve around specific EU projects such as DUET⁷ or industry platforms such as Bosch’s open IoT ecosystem DITTO. These approaches are often hardware-specific, task-specific, and platform-specific, and ultimately fail to provide a common standard that can be generalised across all Digital Twin applications from mobility over construction to industrial production.

Digital Twins are attractive for many communities because they inherently couple data collection, data analysis and data presentation and provide users ranging from policy makers and researchers to citizens with an integrated view on information. As a point of departure, we will focus on the center’s main area of expertise: mobility. This includes the generation of synthetic populations (SynPops) which are a key component for mobility modelling based on the “New Census” strategy of Switzerland.¹⁴ A large number of synthetic population models is available, but their comparison is currently not possible due to the differences in assumptions. Embedding these models in a Digital Twin will allow us to make them comparable and advance the discourse on “SynPops” through validation as well as giving mobility modelling in Switzerland a stronger and open foundation. However, at the same time, we engage other research fields such as energy and land use to come up with a holistic standard that may be deployable across disciplinary boundaries.

On the one hand, the Digital Twin-based ORD practices that were hoped for so far are still lacking.¹³ On the other hand, we are close to building Open Digital Twins as the constituent components of Digital Twins (see following figure for a detailed explanation of our model) can be implemented with existing ORD solutions.¹¹ The physical and data environment can be provided by projects that implement the Federal Open Data Strategy such as NADIM¹⁵ (Nationale Datenvernetzungsinfrastruktur Mobilität), opendata.swiss¹⁶ and municipalities and cantons who openly integrate and aggregate mobility knowledge across Switzerland from all available open sources and some access-limited sources as well. The analytical environment of a Digital Twin could be provided by an open-source knowledge infrastructure for collaborative and reproducible data science such as Renku¹⁷ implementing, for example synpop models. The virtual environment to access, interact with, and visualise the data can be based on classical dashboards in an Open Street Map¹⁸ or SwissTopo¹⁹ environment. Eventually, it might be possible to implement them with modern visualisation approaches based on extended reality (XR) applications.¹¹ At the core of this project will be to provide ORD solutions for the connection environment such as developed in the Experiments as Code paradigm that allow us to interlink these ongoing efforts.²⁰ Thereby, we will enable a simple interaction with mobility data and enable reproducible workflows that push ORD practices in mobility research to the next level.

Implementation

The CSFM has joined together with the external pageSwiss Data Science Center (SDSC) ORD teamcall_made to develop the first version of the Digital Twin in 2023. We have received funding support from Swissuniversities in the form of the external pageSwiss Open Research Data Grantscall_made which will start in Spring 2023 pending some conditional clarifications. We aim to bring together state of the art ORD practices across domains to ensure an Open Digital Twin Platform that works beyond mobility with an accompanying Open Digital Twin Standard.