New CEUS paper online: “Graph based mobility profiling”

Our new paper on “Graph based mobility profiling” was accepted at Computers, Environment and Urban Systems (CEUS) and is now available online. We propose a graph based workflow to identify groups of persons with similar mobility behavior based on person specific graphs that describe the mobility behavior. Our approach is privacy friendly, does not depend on a specific clustering algorithm, is robust against the choice of hyperparameters, does not require specific labels in the dataset, and is not limited to specific types of tracking data. We show in the paper how this can be used to evaluate the impact of new mobility offers.

The paper is open access and the source code of the project is available on our Github.

New TGIS Paper is Online – “Street-level Traffic Flow and Context Sensing Analysis”

Our new paper entitled “Street-level traffic flow and context sensing analysis through semantic integration of multisource geospatial data” by Yatao Zhang and Martin Raubal is now online at TGIS.

Sensing urban spaces from multisource geospatial data is vital to understanding the transportation system in the urban context. However, the complexity of urban context and its indirect interaction with traffic flow deepen the difficulty of exploring their relationship. This study proposes a geo-semantic framework first to generate semantic representations of multi-hierarchical urban context and street-level traffic flow, and then investigate their mutual correlation and predictability using a novel semantic matching method. The results demonstrate that each street is associated with its multi-hierarchical spatial signatures of urban context and street-level temporal signatures of traffic flow. The correlation between urban context and traffic flow displays higher values after semantic matching than those in multi-hierarchies. Moreover, we found that utilizing traffic flow to predict urban context results in better accuracy than the reversed prediction. The results of signature analysis and relationship exploration can contribute to a deeper understanding of context-aware transportation research.

Kick-off of the e-bike city lighthouse project

What if 50% of the existing urban road space was allocated to e-bikes and the slow modes? Seven chairs at the department for civil engineering at ETH join forces to analyze the opportunities and effects of an urban future giving priority to cycling, micromobility and public transport. Our group will contribute with the developement of spatial optimization methods for redistributing the street lanes to design the new bike lane network, considering constraints such as (multi-modal) accessibility and exposure levels of the lanes.

New SIGSPATIAL Paper Is Online – “Improving next location prediction”

We are delighted to announce that our paper “How do you go where? Improving next location prediction by learning travel mode information using transformers” by MIE Lab members Ye Hong, Henry Martin, and Martin Raubal is now online at arXiv and will be presented at ACM SIGSPATIAL, November 1–4, 2022, Seattle, WA, USA conference.

In this work, we propose a transformer decoder-based neural network to predict the next location an individual will visit based on her historical locations, time, and travel modes. In particular, we design an auxiliary task to jointly predict the next travel mode, with the aim of guiding the learning process of the network. We conduct extensive experiments on two real-world GPS tracking datasets and conclude that considering additional aspects of travel behaviour significantly increases the performance of next location prediction. The overall architecture of the proposed model is shown below.

Best paper award at AGILE conference

Our paper “Unlocking Social Network analysis methods for studying human mobility” received the Best Paper Award at the 25th AGILE Conference on Geographic Information Science! Nina Wiedemann presented the paper in Vilnius, Lithuania, and it was elected by the audience out of three papers pre-​selected by the conference committee. In the paper we demonstrate how network modelling methods can help to gain insights into the topology and spatial layout of mobility graphs. For more details, see the publication in the conference proceedings.

New article in Renewable & Sustainable Energy Reviews published

How well can you charge your battery electric vehicle with solar power generated from your own roof?

Our latest journal article “Using rooftop photovoltaic generation to cover individual electric vehicle demand—A detailed case study” shows that this works surprisingly well even without a dedicated battery home storage.
The article was also featured in the ETH News and other media outlets. Check out the press release or the journal article published in Renewable & Sustainable Energy Reviews (open access).

3rd place in the Traffic4cast 2021 extended challenge

Nina Wiedemann of MIE Lab achieved 3rd place in the NeurIPS 2021 Traffic4cast competition!
In this year’s extended challenge, the task was to predict traffic volume and speed in unknown cities, based on movie-format spatio-temporal data. Our team could improve the generalisation ability of the convolutional neural network with a patch-based approach. For more details on the method, check out our preprint on arXiv.

New Survey Paper Is Published – “Applications of Deep Learning in Congestion Detection, Prediction and Alleviation: A Survey”

How can deep learning be used to make road travel safer and faster at the same time? Can we achieve these two objectives while ensuring fairness across individuals or user groups? How important is the percentage of data available for better model performance? Are there potential conflicts of interest if a small number of private players cater to the route suggestion requirements for the majority of the population? Check out the latest publication from MIE-lab- “Applications of deep learning in congestion detection, prediction and alleviation: A survey”, authored by Nishant Kumar and Prof. Dr. Martin Raubal. The paper attempts to answer these questions based on the current state of research. The paper is available open access at TR:C.

As the title suggests, the paper covers three specific tasks. In the first part (congestion detection), we summarize how the deep learning models that were initially developed for computer vision tasks are being applied in detecting traffic congestion. A clear distinction is seen between the models and data sources used in different geographical locations. In the second part (congestion prediction), we summarize why congestion prediction is a more difficult task than traffic prediction. We discuss specific examples of how researchers have leveraged the domain knowledge from transportation (such as the heterogeneity of road networks) to improve the overall prediction accuracy of deep learning models. In the third part (congestion alleviation), we summarize how deep learning is being used for alleviating congestion. We review the challenges in implementing demand-side solutions argue in favour of system-level optimizations with the policymaker in the loop.

Finally, we draw attention to the bigger picture and the potential clash of interests between our efforts to alleviate recurrent and non-recurrent congestion. Presented below is a hypothetical U-curve showing the current understanding of how traffic congestion and road safety are related.

A Clustering-Based Framework for Individual Travel Behaviour Change Detection

In our recently published 2021 GIScience paper “A Clustering-Based Framework for Individual Travel Behaviour Change Detection“, we propose a clustering-based pipeline to delineate travel behaviours and detect possible change periods/points from raw GPS recordings. In particular, considering trip mode, trip distance, and trip duration as travel behaviour dimensions, we measure the similarities of trips and group them into clusters using hierarchical clustering. Two different methods are then proposed to detect changes in an individual’s observed usage proportion of trip clusters. Through testing the framework on a large-scale longitudinal GPS tracking dataset, we demonstrate its effectiveness in detecting change periods/points by jointly considering multiple travel behaviour dimensions.

The code is openly available on GitHub, with the possibility to reproduce the framework on the Geolife dataset. The study is to be presented at GIScience 2021 and the pdf version is available here.