Interpretable and Robust Machine Learning for Mobility Analysis
Artificial intelligence (AI) is revolutionizing many areas of our lives, leading a new era of technological advancement. Particularly, the transportation sector would benefit from the progress in AI and advance the development of intelligent transportation systems. Building intelligent transportation systems requires an intricate combination of artificial intelligence and mobility analysis. The past few years have seen rapid development in transportation applications using advanced deep neural networks. However, such deep neural networks are often difficult to interpret and lack robustness, which slows the deployment of these AI-powered algorithms in practice. To improve their usability in deployment, an increasing research effort has been devoted to developing interpretable and robust machine learning methods, among which the causal inference approach recently gained traction as it can provide interpretable and actionable information. However, most methods are developed for image or sequential data which cannot satisfy the specific requirements of mobility data analysis. These unique requirements have been intensively studied in the Geographic Information Science (GIScience) field but have not yet been well utilized in developing machine learning models. Through the collaboration with the Swiss Data Science Center, we aim to bring together the knowledge of GIScience and Machine Learning, advancing our understanding of how interpretable and robust machine learning methods can be tailored to mobility analysis with the support of causal inference. The outcome of this research will deepen our understanding of how to integrate AI technologies and GIScience for mobility analysis, making AI in the transportation sector more interpretable and reliable. Ultimately, we aim to facilitate the deployment of AI in intelligent transportation systems and build a safer, more efficient, and more sustainable transportation system in the future.
Internal Researchers: Prof. Dr. Martin Raubal, Dr. Yanan Xin, Ye Hong
Project Start: 01.11.2021
Funding: The Hasler Foundation, Hasler Responsible AI Research Program
E-bike city
The E-Bike City project is an interdisciplinary research project of the Department of Civil, Environmental and Geomatic Engineering (D-BAUG) at ETH Zürich. A group of 26 researchers is collaborating over three years to investigate the effects of redesigning our urban streets to give absolute priority to people traveling by bicycle, e-bike, micro-mobility and public transport. The research will inform a new city design which will improve urban living and increase quality of life in a way that has never been done in Switzerland.
The MIE lab’s works on a core part of the project: the bike network design. What is a good bike network, and how can we optimise the bike network while minimising the impact on other modes such as cars? We develop optimization methods and an evaluation framework for bike network planning, with a focus on finding the best trade-off between bike and car travel times.
Internal Researchers: Prof. Dr. Martin Raubal, Ayda Grisiute, Nina Wiedemann
Project Start: 01.11.2022
Funding: Departement of Civil, Environmental and Geomatic Engineering (D-BAUG), The Swiss Federal Office of Energy (SFOE),
Finished Projects
V2G4CarSharing – Mobility-Aware V2G Optimization for Car–Sharing
Car-sharing and Vehicle-to-Grid (V2G) are promising innovations to decarbonize the transport sector. Integrating car-sharing and V2G offers new opportunities to accelerate the market penetration of individual innovation due to their complementary roles in increasing the asset use of electric vehicles (EVs). However, there are still many unanswered questions in integrating car-sharing and V2G regarding its feasibility and potential benefits. In this research project, we aim to develop and evaluate optimal strategies to integrate car-sharing and V2G through close collaboration with our research partners Hive Power and Mobility. This project will enrich our understanding of the feasibility and benefits of combining car-sharing and V2G, and provide guidance to support industrial implementations.
Internal Researchers: Prof. Dr. Martin Raubal, Dr. Yanan Xin, Nina Wiedemann
Project Start: 01.10.2021
Funding: The Swiss Federal Office of Energy (SFOE), Mobility Research Programme
EIM – Empirical use and Impact analysis of MaaS
Individual private car use is inherently unsustainable when compared to shared modes. Yet, it is one of the main travel modes around the world, often due to a lack of alternatives. The introduction of novel shared travel modes and their integration with public transport (Mobility as a Service, MaaS), promises to enable seamless intermodal travel, thus facilitating a behavioral change from private car use to more sustainable, shared modes.
Core of the EIM project is YUMUV , a real world MaaS case-study conducted by the Swiss Federal Railways (SBB) and the public transport providers in the cities of Basel, Bern and Zurich. The Mobility Information Engineering (MIE) Lab at the Chair of Geoinformation Engineering and the Institute for Transport Planning and Systems (IVT) provide academic support for the design and the analysis of this study.
We use the data generated in this study to advance our so-far limited scientific understanding of how MaaS changes travel behavior (e.g., mode choice, car ownership) or how more sustainable travel behavior can be encouraged (e.g., through optimized service bundling or real-time mobility prediction).
We plan to develop an integrated representation of relevant mobility and context data based on person-specific graphs. This allows the application of both state-of-the-art research methods in transport behavior (hybrid choice and multivariate probit models) and geographic information science (spectral clustering, graph neural networks) on such heterogeneous data.
Internal Researchers: Prof. Dr. Martin Raubal, Henry Martin
Project Start: 01.08.2019
Funding: The EIM project is funded by the ETH Mobility Initiative
COMMIT – Context-Aware Mobility Mining Tools
Location-acquisition technologies provide ample opportunities regarding the monitoring, management, and regulation of human mobility behavior, thereby avoiding some potential negative effects (e.g. traffic jams, fuel consumption). COMMIT project aims to develop serials of context-aware computational methods for evaluating, predicting, and analyzing human mobility behavior based on massive trajectory data and related contextual data. It can be expected to have high societal significance and relevance for a broad range of scientific disciplines and related industrial sectors.
Internal Researchers: Prof. Dr. Martin Raubal, Dominik Bucher, Henry Martin, Ye Hong
Project Start: 01.10.2018
Funding: SDSC (Swiss Data Science Center)
SCCER Mobility
The Swiss Competence Center for Energy Research – Efficient Technologies and Systems for Mobility (SCCER Mobility) aims at developing the knowledge and technologies essential for the transition of the current fossil fuel based transportation system to a sustainable one, featuring minimal CO2 output and Primary Energy Demand as well as virtually zero-pollutant emissions. Read more!
Internal Researchers: Prof. Dr. Martin Raubal, Dominik Bucher
Project Start: 01.08.2014
Funding: CTI (Commission for Technology and Innovation)
Decision Support System for Personalized Ride-Sharing Services
Various studies have demonstrated that ridesharing and carpooling are beneficial to alleviate traffic congestions and to reduce CO2 emissions, fuel/energy consumption as well as travel costs. This project aims to offer personalized, short-term, and automated rideshare planning based on passively tracked movement data, considering travel time uncertainty and users’ travel preferences and schedules. This project will provide tangible support for the fulfillment of the “Energy Strategy 2050” objectives via an information service for sustainable mobility choices from the perspective of sharing mobility.
Internal Researchers: Prof. Dr. Martin Raubal, Dominik Bucher
Project Start: 01.05.2019
Funding: Innosuisse – Swiss Innovation Agency
Optimizing the potential impact of personal and autonomous electric mobility on grid stability
The high power demand of electric vehicles (EVs) during charging forces us to study Swiss mobility and energy systems as a whole and makes the consideration of the grid impact a requirement. Vehicle-to-grid (V2G) opens an avenue for the successful integration of EVs in the energy system. This project will offer support for the fulfillment of the Swiss Energy Strategy 2050 from the perspective of energy infrastructure and smart charging. The outcome of this project will provide optimal charging schedules for electric vehicles by investigating the interaction between the grid and electric vehicles.
Internal Researchers: Prof. Dr. Martin Raubal, Jannik Hamper, Henry Martin, Dr. Yanan Xin
Project Start: 01.05.2019
Funding: Innosuisse – Swiss Innovation Agency
SBB Green Class
In the context of integrated mobility offers currently gaining in importance, SBB is collaborating with BMW Switzerland to develop a combined mobility solution, and test it with 140 test customers in a one-year pilot project. We provide the lead scientific support for this new research project SBB Green Class, and will use surveys, interviews, and spatio-temporal movement data analyses to examine how such a comprehensive multi-modal mobility offer influences people’s mobility behavior. Read more!
Internal Researchers: Prof. Dr. Martin Raubal, Henry Martin, Dominik Bucher
Project Start: 27.10.2016
Funding: Schweizerische Bundesbahnen (SBB)
Mobility Profiles and Logistics Concept for Small-scale Food Producers
Especially in rural areas of Switzerland, transport costs are a critical factor for small-scale food producers, who frequently transport relatively small amounts of goods over large distances. In collaboration with alpinavera, we will conduct a GNSS-assisted travel survey to assess the current stage of transport behaviour and needs, and build on the resulting insights to develop and test a range of alternative scenarios for their potential economic and ecological benefits.
Internal Researchers: Prof. Dr. Martin Raubal, Dominik Bucher
Project Start: 01.02.2017
Funding: alpinavera
OMLETH I
In the OMLETH project (“ortsbezogenes mobiles Lernen an der ETH Zürich”), a generic platform for location-based mobile learning has been developed. The prototype platform was evaluated together with students and lecturers of a selected course provided by the Chair for the History of Urban Design, belonging to the department of architecture D-ARCH over two semesters. The platform enables lecturers to easily create location-based learning modules with a web-based interface. The developed learning modules are made available to students via a mobile browser “App” which supports them during individual learning sessions and excursions. Read more!
Internal Researchers: Prof. Dr. Martin Raubal, Christian Sailer, Dr. Peter Kiefer
Project Duration: 01.08.2014–31.12.2015
Funding: Innovedum ETH Zürich
GoEco!
Current mobility patterns are dominated by car use, even though a number of alternative and energy-efficient mobility options are already available. The GoEco! app can help to make this transformation happen. We conduct a study with 200 active users from Canton Ticino and the City of Zurich. The GoEco! smartphone app helps the test persons to make their mobility lifestyle more sustainable. It tracks the trips they perform and uses game elements to challenge them to modify their mobility behavior: progressing towards their goal and competing with their friends, they learn how to become more sustainable. Read more!
Internal Researchers: Prof. Dr. Martin Raubal, Dominik Bucher, Dr. David Jonietz
Project Start: 01.01.2015
Funding: SNF – Abteilung Nationale Forschungsprogramme
Project Results: The results of the study can be accessed via the research database P3 of the Swiss National Science Foundation: http://p3.snf.ch/project-153743.
OMLETH II
Building on the findings of OMLETH I the platform will be technically adapted and augmented with spatio-temporal visualization functionality to enable the analysis of learners’ movement trajectories. The resulting system will be evaluated in the course of numerous field tests, which will involve large groups of ETH students as well as several secondary level school classes. The gained insights will be incorporated into an audio-visual knowledge platform for location-based mobile learning which is continuously updated. New applications of the system in courses hosted by D-USYS and the Material Archive of Baubibliothek will provide further insights. Potential participants (ETH interns) are kindly invited to sign up. Read more!
Internal Researchers: Prof. Dr. Martin Raubal, Christian Sailer, Dr. Peter Kiefer
Project Start: 01.01.2016
Funding: Innovedum ETH Zürich