Physics and Collective Dynamics of Future Mobility

The densely populated State of North Rhine-Westphalia in Germany has a large correlated motorway network. We began big data analyses of the measured data on flows and velocities by employing and transferring modern methods for correlated systems which we partly developed in the context of finance. Some of our first results are: We show how antipersistence of traffic flow time-series impacts the duration of traffic congestion on a wide range of time scales. We find a large number of short lasting traffic jams, which implies a large risk for rear-end collisions. To capture non-stationarities, we identify quasi-stationary states in temporal correlations, revealing structures in time, due to the rich non-Markovian features of traffic. We use a machine learning approach and apply k-means clustering to temporal correlation matrices. We relate this to free or congested traffic phases (in Kerner's theory) in space and time. Focusing on the spatial correlations, we present a method to identify and quantify collective behavior, i.e. coherent motion in the whole network or in large parts of it. We show that collectivity throughout the network cannot directly be related to the traffic states (free, synchronous and congested) in Kerner's theory. Hence, the degree of collectivity provides a new, complementary observable to characterize the motorway network. This is joint work of the groups Guhr and Schreckenberg at Duisburg-Essen.

Individual transport in cities is mostly enabled by private cars, a status quo that is both environmentally and socially unsustainable. Especially on inner-city routes, cycling is a widely accessible and more sustainable alternative. Promoting cycling critically relies on a sufficiently developed bicycle infrastructure. Here, we aim at identifying bike path networks that enable fast and safe cycling in cities by explicitly considering the demand distribution and route choice of cyclists based on safety preferences. We reverse the process of network formation by starting from a complete network of bike paths, successively removing the less important bike paths and continuously updating cyclists' route choices. Even a few bike paths can lead to a highly cycle-friendly network. Our framework may thus enable the demand-driven design of efficient infrastructures for more sustainable private transport.

Electric vehicles may dominate motorized transport in the next decade, yet the impact of the collective dynamics of electric mobility on long-range traffic flow is still largely unknown. We demonstrate a type of congestion that arises if charging infrastructure is limited or electric vehicle density is high. This congestion emerges solely through indirect interactions at charging infrastructure by queue-avoidance behavior that— counterintuitively—induces clustering of occupied charging stations and phase separation of the flow into free and congested stations. The resulting congestion waves always propagate forward in the direction of travel, in contrast to typically backward-propagating congestion waves known from traditional traffic jams. These results may guide the planning and design of charging infrastructure and decision support applications in the near future.

Ride sharing -- the bundling of simultaneous trips of several people in one vehicle -- may help to reduce the carbon footprint of human mobility. However, the complex collective dynamics pose a challenge when predicting the efficiency and sustainability of ride-sharing systems. Standard door-to-door ride sharing services trade reduced route length for increased user travel times and come with the burden of many stops and detours to pick up individual users. Requiring some users to walk to nearby shared stops reduces detours, but could become inefficient if spatio-temporal demand patterns do not well fit the stop locations. Here, we present a simple model of dynamic stop pooling with flexible stop positions. We analyze the performance of ride sharing services with and without stop pooling by numerically and analytically evaluating the steady state dynamics of the vehicles and requests of the ride sharing service. Dynamic stop pooling does a-priori not save route length, but occupancy. Intriguingly, it also reduces the travel time, although users walk parts of their trip. Together, these insights explain how dynamic stop pooling may break the trade-off between route lengths and travel time in door-to-door ride sharing, thus enabling higher sustainability and service quality.

Understanding the patterns underlying human mobility in urban areas is essential to better plan cities, engineering traffic, and mitigating diseases. However, existing studies have characterized only some spatial features of mobility — such as travel distance — overlooking an important temporal feature: how frequently do we visit a particular place? And, how is this visitation frequency related to the distance we traveled? Modeling the interplay between spatial and temporal features of mobility is critical; for instance, it can provide urban planners with the information to best place a shopping mall to attract customers. The analysis of over 8 billion human mobility traces collected over four continents reveal that humans have a natural tendency of trading off travel distance with frequency: the further we travel, the less frequently we do it, according to a “visitation law” that can be described through a simple mathematical law. We will then show how this striking discovery can be used to better locate businesses and facilities in urban spaces, and find more effective containment strategies for disease spreading.

There is a contradiction at the heart of our current understanding of mobility patterns. On one hand, a highly influential stream of literature driven by analyses of massive empirical datasets finds that human movements show no evidence of characteristic spatial scales. There, human mobility is described as scale-free. On the other hand, in geography, the concept of scale, referring to meaningful levels of description from individual buildings through neighborhoods, cities, regions, and countries, is central. Here, we resolve this apparent paradox by showing that human mobility does indeed contain meaningful scales, corresponding to spatial containers restricting mobility behavior. The scale-free results arise from aggregating displacements across containers. We present a simple model, which given a person’s trajectory, infers their neighborhoods, cities and so on. We find that the containers characterizing the trajectories of more than 700,000 individuals worldwide do indeed have typical sizes. We show that our description improves on the state-of-the-art in modeling, and allows us to better understand effects due to socio-demographic differences and the built environment

Urban transport systems are gaining in importance, as an increasing share of the global population lives in cities and mobility-based carbon emissions must be reduced to mitigate climate change. Furthermore, pollution and congestion caused by car traffic generate severe adverse health effects. Thus, the share of public transport systems in urban traffic must be increased. However, building and maintaining public transport systems is expensive and congestion effects lower their quality, raising the question of how to optimise them to cope with these challenges. We analyse the optimal shape of urban transport networks under economical constraints with the objective to minimise the average travel time for commuters in the city. We propose a versatile numerical approach to find the optimal network geometry for different spacial city structures and congestion models.

We introduce a generic meta-population framework by which the spreading dynamic in both levels of inter-population and intra-populations (mobility network) evolve. Then we show how to estimate A) where, B) when the dynamic had started and C) calculate the cut-off errors in this frame. Finally we redefine the effective distance concept and show that the effective distance vs overtaking time, when the inter-population dynamic overtakes the intra-population dynamic, presents a universal geometry, namely a line, for those neighbours of the origin whose reachability via intermediate nodes are negligible in comparison to the direct effective paths. This line has a universal slope for any disease or network of mobility or origin node. The empirical data of COVID-19 epidemics in Iran and the United States of America as well as the H1N1 pandemic in the world confirms our theory.

Demand-responsive ride-pooling (DRRP), if operated efficiently, can be much more sustainable than individual traffic by private car. However, since customers are only willing to incur limited pooling delays, the degree of poolability of ride requests via unimodal DRRP (single vehicle per trip) is limited. Fortunately, if we allow for transfers between different vehicle types, the pooling potential increases. In this project, we develop a model of bimodal on-demand transportation, where a DRRP system is combined with a fixed schedule line service. We identify the critical control parameters that determine system performance, which is measured via emissions and quality. We solve this multi-objective optimization problem by computing Pareto fronts, thereby characterizing the possible tradeoffs between quality and emissions under varying demands. Our model quantifies under what circumstances bimodal public transportation is feasible, both in terms of convenience and ecological footprint. Moreover, the model yields estimates for how to set up and operate a bimodal transportation system optimally, and as such, may guide urban planners and public transportation services when considering the transition towards sustainable mobility.

Urban traffic congestion is a worldwide problem affecting the livability of cities and causing environmental and health problems. The relation between the flow of cars and their spatial density informs the expected travel time in a single street. However, the macroscopic characterization of traffic dynamics at urban scale remains an elusive task. Here we unravel this complex phenomena under various conditions of demand and translate it to the travel time of the individual drivers. First, we start with the current conditions, showing that there is a characteristic time τ that takes a representative group of commuters to arrive to their destinations from the maximum density of cars at the traffic peak hour. While this time differs from city to city, it can be explained by a quantity Γ, defined as the ratio of the total vehicle demand to their available street capacity. Modifying Γ can improve τ and directly inform planning and infrastructure interventions. Moreover, we systematically characterize the dynamic of the system by increasing volume of cars in the network, keeping the road capacity and the empirical spatial dynamics from origins to destinations unchanged. We identify three states of urban traffic, separated by two distinctive transitions. The first describing the appearance of the first bottle necks, and the second the transition to a complete collapse of the system. The transition to the second state measures the resilience of the various cities and is characterized by a non-equilibrium phase transition.

Ride-sharing — the combination of multiple trips into one — may substantially contribute towards sustainable urban mobility. It is most efficient at high demand locations with many similar trip requests. However, here we reveal that people’s willingness to share rides does not follow this trend. Modeling the fundamental incentives underlying individual ride-sharing decisions, we find two opposing adoption regimes, one with constant and another one with decreasing adoption as demand increases. In the high demand limit, the transition between these regimes becomes discontinuous, switching abruptly from low to high ride-sharing adoption. Analyzing over 360 million ride requests in New York City and Chicago illustrates that both regimes coexist across the cities, consistent with our model predictions. These results suggest that even a moderate increase in the financial incentives may have a disproportionately large effect on the ride-sharing adoption of individual user groups.

Ride pooling (RP) is an emerging transport mode that uses on-demand shuttle buses to combine the trips of multiple users into one vehicle. It has the potential to drastically reduce carbon emissions, traffic and the need for owning a private car. The extent to which this potential is exploited is quantified by the system's efficiency, which is best understood as a tradeoff between increased vehicle occupancy and thus entailed detours. Here, we introduce a modelling framework that predicts RP efficiency without relying on simulation results, and demonstrate how the minimum fleet problem can be solved with it. Furthermore, we verify its predicted scaling behavior via extensive simulations using a variety of street networks and dispatch algorithms. It turns out that under a broad class of the latter two, the relationship between fleet size $N$ and the request frequency $f$ it can serve follows a power law. The exponent of this power law is approximately 1.15 ($f\propto N^{1.15}$) which confirms the intuition that RP outperforms private car use and non-shared taxis ($f\propto N^{1.0}$) as user demand increases.

Many of our routines and activities are linked to our ability to move; be it commuting to work, shopping for groceries, or meeting friends. Yet, factors that limit the individuals' ability to fully realise their mobility needs will ultimately affect the opportunities they can have access to (e.g. cultural activities, professional interactions). One important aspect frequently overlooked in human mobility studies is how gender-centred issues can amplify other sources of mobility disadvantages (e.g. socioeconomic inequalities), unevenly affecting the pool of opportunities men and women have access to. In this work, we leverage on a combination of computational, statistical, and information-theoretical approaches to investigate the existence of systematic discrepancies in the mobility diversity (i.e. the diversity of travel destinations) of (1) men and women from different socioeconomic backgrounds, and (2) work and non-work travels. Our analysis is based on datasets containing multiple instances of large-scale, official, travel surveys carried out in three major metropolitan areas in South America: Medellín and Bogotá in Colombia, and São Paulo in Brazil. Our results indicate the presence of general discrepancies in the urban mobility diversities related to the gender and socioeconomic characteristics of the individuals. Lastly, this paper sheds new light on the possible origins of gender-level human mobility inequalities, contributing to the general understanding of disaggregated patterns in human mobility.

Reducing the reliance on personal cars is a key requirement for reducing emissions, congestions and making cities more livable. On-demand ridepooling, little more than a concept 8 years ago, is now a usable mode of transport in various locations all over the world. By bundling spatio-temporally correlated requests together in a single vehicle, such services can offer a high level of passenger satisfaction in an efficient and sustainable manner. In this talk, we will explore using theory and simulations how the efficiency of on-demand ridepooling systems are affected by factors such as trip density, fleet size and network topology. Based on extensive user surveys done on the users of MOIA, the biggest commercially running ridepooling service in Europe, we will present which factors encourage people to choose ridepooling as a mode of transport.

We are spending much time in traffic jams. Would it help to build additional roads? As D. Braess has shown in 1968 this is not necessarily true! In a road network of selfish users additional roads can lead to higher travel times for all users! We have extended the scenario studied by Braess on a macroscopic level to the microscopic level by considering the dynamics of vehicles explicitly. This more realistic traffic dynamics allows to include stochasticity both in the vehicle motion and the route choices of the drivers. Different examples are investigated to determine whether Braess paradox still occurs in such more realistic settings. Furthermore we consider the effect of traffic information (e.g. by navigation apps) which allows for more intelligent route choices to determine if these a able to resolve Braess paradox.

Sustainable mobility is among the primary challenges for human society and ensuring it is indispensable for a viable future. Compared to individual mobility, public transport offers an alternate transport mode with substantially lower emissions and many other economic and social benefits. However, despite these benefits, public transport is not widely adopted, yet achieving the same is critical for a sustainable future. I qualitatively demonstrate how interactions between the users' mode choice behavior and the service provider may lead the system to be stuck in a sub-optimal state, causing higher average travel times. A simplified game-theoretic model reveals that this hysteresis effect is prevalent in public transport systems, generically emerging as cities expand. I observe the same hysteresis effect in public transport usage data from major urban areas in the UK, potentially contributing to the declining public transport ridership and explaining opposite trends in London. These findings indicate that public transport systems may self-organize to sub-optimal conditions as cities expand, limiting public transport ridership and impeding the shift to sustainable urban mobility.

The description of complex human mobility patterns is at the core of many important applications ranging from urbanism and transportation to epidemics containment. Data about collective human movements were once scarce, limited to surveys describing commuters' flows, but with the advent of the ICT a deluge of passively collected information has become available thanks to new sources such as Phone CDR, GPS devices, or Smartphone apps. However, most studies in mobility and epidemics often rely on a dataset coming from a single source and assume that their conclusions are universal. Here, we test such an overarching assumption on an unprecedented scale by comparing human mobility datasets obtained from 7 different data-sources, tracing over 500 millions individuals in 145 countries. We report wide quantifiable differences in the resulting mobility networks and, in particular, for the displacement distribution previously thought to be universal. These variations also impact processes taking place on these networks: we consider epidemic spread and show that, depending on the dataset used, the spreading dynamics can lead to completely different outcomes and even exhibit a variable number of incidence peaks. Our results point to the crucial need for disclosing the data processing and, overall, to cross-check empirical results for assessing the robustness of simulation studies or the presence of universal behavior.

Latent information about an individual's mobility can be present in the patterns of their social contacts as well as those that they co-locate with. We develop a "colocation" network to distinguish the mobility patterns of an ego's social ties from colocators, individuals not socially connected to the ego but who nevertheless arrive at a location at the same time as the ego. We apply entropic measures to analyse and bound the predictive information of an individual's mobility pattern and the flow of that information from their social ties and colocators. While social ties generically provide more information than colocators, we find that significant information is present in their aggregation: 3-7 colocators can provide as much predictive information as the top social tie, and can replace up to 85% of the predictive information about an ego, compared with social ties that can replace up to 94% of the ego's predictability. The presence of such information transfer raises privacy concerns: given the increasing availability of real-time mobility traces from smartphones, individuals sharing data may be providing actionable information not just about their own movements but the movements of others whose data are absent.

The talk will reflect on the experience of the impact of the pandemic on the transport system in Switzerland. Emphasis will be on speed changes observed in the network during the phases of the pandemic. The results will be contrasted with the analysis of 100+ MFD (macroscopic fundamental diagrams) estimated from 43 cities world wide. The analysis os focussed on the link between network topology and network capacity. The talk will close with a discussion of the current dilemma of transport planning between induced demand, accessibility and global warming.

Making our cities better and sustainable is key to solving the climate and urban transport crises. To this end, urban data science offers new tools to quantify societal problems in cities and to propose human-centric solutions to policy makers. In this talk I outline our recent and ongoing efforts towards that end, focusing on urban transport and vulnerable road users. I discuss inequalities between transport modes through mobility space and collision threat distributions, and automated methods based on network science to generate and merge bicycle networks and to identify their gaps, for different urban development stages. All research comes with concrete policy recommendations that significantly increase urban livability, public health, and decarbonization.

On-demand mobility services commonly employ dynamic pricing schemes. Like Uber's surge pricing, these pricing schemes are meant to balance demand and supply by incentivizing drivers to offer their service during times and in areas of high demand. However, surge pricing may instead cause temporary supply shortages as drivers try to trigger artificial price surges to increase their profit. We capture the underlying incentives in a minimal game-theoretic model of the supply dynamics, illustrating that they are a generic consequence of dynamic pricing schemes. Furthermore, we identify the emergence of such self-organized price surges by analyzing price estimate time series for Uber trips from 137 locations in 59 urban areas across six continents and identifying locations with strong, repeated price surges.

Traffic congestion increases CO2 emissions and environmental pollution, is detrimental to health, and causes substantial economic losses. Digital route choice support is supposed to help customers to navigate in complex street networks and avoid congestion. However, generic feedback delays are unavoidable as time passes between collecting information about current traffic flow and the moment customers enter the recommended street segments. Here we illustrate how routing based on delayed information may cause oscillations and destabilize traffic flow. Above a critical delay, recommendations provided to avoid congestion may even induce congestion. We propose two ways of counteracting this collective phenomenon: (i) reducing the impact of temporally localized traffic fluctuations by averaging traffic data over sufficiently long time intervals and (ii) reducing the size of the responding population, i.e. informing fewer customers about the traffic state. Intriguingly, less or less exact information may thus help digital route choice support to actually achieve the desired effect - less congestion.

The progress in automation of driving abilities of cars is enormous. But we will have a mixture of human drivers and automated vehicles for a longer time. The problems arising out of this mixture are not know in full extend. Therefore simulation based on realistic models are needed in order to mimic future traffic dynamics on the roads. In addition communication between actors gives rise to further impact on traffic patterns. The talk tries to give some insights into this viral question.