Expert Talk: Standardization in Mobility Assessment – there’s more than the last mile missing.

November 16th, 2022 | by GEONATIVES

(8 min read)

Mobility assessment is a fascinating topic. And so is standardization. But where would you find the experts and audience to combine both subjects? Correct. Here at GEONATIVES!

In our experts’ voices series, we had the pleasure to talk to Dr. Steffen Axer of Volkswagen Group whose main job is around impact assessment of future mobility systems. We talked to him as traffic engineer and transport modeling expert on a private basis. His academic background reaches from economics to traffic management, and he holds a master’s degree in traffic engineering. His PhD efforts focused on ITS applications and, among other, the analysis of traffic infrastructure data from recorded trajectories.

One of the key questions Steffen is trying to address is: what data need to be collected to assess the quality of mobility and the impact of new mobility services objectively? What will be the future scope of mobility e.g., in a city or metropolitan area?

Currently a commonly agreed basis for mobility assessment is missing. In many of today’s assessments, bias is part of the result. Often, information is missing about what exactly has been measured and how this was done.

According to Steffen, the resource dimensions time, money, energy, and space usage should be part of every assessed mobility system. Unfortunately, there is no reference implementation (and standardization) available to compare various investigations. This means that studies on mobility tend to pick only on a small subset of indicators that could have been modeled or observed by ex-post data analytics. But mobility is a fairly complex system.

Dimensions of mobility: people, planet, profit. Or are other dimensions more important?

Take one indicator, for example: the TomTom traffic index. It focuses on motorized individual road traffic and how much longer your average car journey takes compared to “free flow”. Quite often the quality of urban mobility gets expressed by this indicator. This indicator is for sure reasonable, but it describes the quality of transportation only for one particular mode of transport. Whose interests are represented by this kind of index – or, in other words: cui bono? Who would be affected by any measures taken to improve the traffic index?

Mobility simulations are a powerful method to model current traffic systems in their complexity and to predict potential impacts of many changes to the status quo. They allow thus, to overcome the limitations of historical mobility data analysis and enable us to look into future mobility system constellations. Commercial and open-source software packages are available for modeling on nanoscopic, microscopic and macroscopic level. The big question, though, is what is behind these models. Can or are they being assessed independently so that whoever extracts indicators are not taking advantage of a certain modeling approach that is favorable to their own purpose and extracts only the indicators serving their own purpose. One large step into a more comparable world of mobility simulations would be an assessment technique that does not prefer any kind of simulation technology, Steffen says. Such an assessment enforces data interfaces that allows at least to evaluate the same mobility indicators. Moreover, such an approach would also require reference implementations of data generation for different simulation techniques to ensure data result similarity.

Within our talk with Steffen, we clearly identified a major gap in today’s mobility simulation landscape: there’s no such thing like an independent entity verifying large scale mobility simulations and derived results. According to Steffen, transparent and open documentation are necessary for several stages during the mobility modeling process. That would cover aspects of data sourcing, decision and service modeling etc. – for sure, a very challenging task, but necessary, if we really want to reach comparable results of different software and modeling approaches.

If we reached such a level of model comparability, one might ask whether the resulting transparency would be everybody’s darling? For the very first time, mobility plans and current mobility system performance of various cities could be somehow ranked and benchmarked. That makes clear, that even public authorities are “particular” stakeholders: They, for sure, may not want to see an official label stating that their mobility concept is worse than others. For Steffen as a traffic engineer, reaching such a state is in the end the desired objective. At this level, society and authorities starts to optimize their systems due to a certain level of competition which is currently not there.

Software is good and open source is better if your goal is to have transparency about models and data. One example providing transparency and a consistent modeling from microscopic to macroscopic levels is MATSim by TU-Berlin and ETH Zürich. It also facilitates the exchange of individual models (transport layer, motion/dynamics) and can simulate huge regions with millions of individual agents. Companies like SBB and MOIA are contributing code and methods in order to set standards that are able to be challenged and thus allows other experts to understand granularity and level of details, how mobility challenges could be handled. MATSim delivers especially transparency when it comes transport, service dynamics and choice modeling. However, when it comes to the task of demand synthesis, MATSim required exogenous inputs. The main reason for this is the very heterogeneous and manifold data situation that is available around the globe. Anyway, even for this kind of a problem several activities raised up in the last years, which might be somehow merged in the feature, Steffen says.

Coming from models to the second and no less important aspect: data. Who owns the data, who collects them, and how can they be exchanged?

In terms of ownership, the situation seems to be clear within the European Union and the US for data that have been collected by spending public funds: they belong to the public. But as we have seen in one of our previous talks, this does not really seem attractive for commercial providers. Even some kind of publicly founded data is even not available for commercial companies, Steffen says. For instance, anonymized trip diary surveys (MID 2017) are typically geocoded that allows mobility modelers, even in companies, to understand the geographical context of a trip and potential mode alternatives. Unfortunately, geocoded information of trip diary surveys is not available for commercial activities, Steffen says.

Regarding data we are not only talking about road and rail networks, timetables and distribution of population but especially about anonymous trip data, which describe when and where someone is traveling and originates mainly from smartphone apps or cellphone networks.

Some data are collected only by commercial entities, but they are provided for free by the individuals creating them. Yes, these commercial entities (read: tech and mapping companies) also provide free services, but the data could also be used for many other purposes if only they were available to a broader public. Just take the example mentioned above about the assessment of mobility simulation: more real-world data might indeed help getting a better picture of a tool’s performance.

Data you collect is just a chunk of bits and bites until you can do two things: first, put them into a format that can be processed by a significant number of stakeholders, and second, assess their fitness for a given purpose.

In terms of data formats that are widely used, the issue seems almost solved. Even though some formats may have emerged from commercial applications, they are available publicly and a significant amount of relevant data is being stored and exchanged. Two examples mentioned in our expert talk with Steffen are RailML and the General Transit Feed Specification (GTFS). GTFS is used by public transport entities to store, among other, schedules and make them available in real time so that you can perform your typical Google search and find the best connection from point A to point B. RailML offers possibilities to model schedules as well but most users such as the large European railway companies are mainly using the topology layer for managing their rail networks.

Nevertheless, companies still using their own, proprietary formats need to see incentives to migrate to one of the open standards. The usual arguments for migrating to open standards and open source would be that overall maintenance efforts for software are expected to be lower than for a fully vertically integrated proprietary solution. Often, a strong stakeholder is needed to get a standard established (see GTFS). An inherent participation in technological advancement is yet another point. But the key argument within the context of this expert talk (see our headline) is another one: trust.

Each party in the mobility sector has its own interests. Unless it is fully clear what these interests are and how these interests affect data and models used and/or provided by a party, the level of trust will be limited. How do you create trust? First, you lay open what you have (see the prior discussion about open-source software and data). Second, you involve trusted independent parties in the dialog between the special interest groups.

No party is fully independent, though, but publicly funded universities usually provide a good basis to perform the relevant research and investigation that might go into reference implementations for inter-modal mobility models and the extraction of the respective indicators. A trustee would be particularly helpful, if data is collected by various stakeholders, to get a “normalized” result. This, and in addition the provision of publicly accepted and used standards for data formats may form an excellent basis for an un-biased assessment of mobility solutions.

At the end of our talk, our discussion tried to address some non-technical aspects of mobility, which are harder to put into exact numbers. A big question is how and why individuals decide for a specific transport means if they are given multiple options. Socio-economic aspects may influence this decision and these aspects will vary among regions (e.g., a preference for bicycles in Dutch cities vs. the famous metro in Paris – just to name two prejudices we Germans usually have). For these aspects, a standardized method for collecting and assessing data is required, too.

Standardized questionnaires about the travel behavior would be helpful to compile standardized travel diaries, which are objective and transparent. Regularly updated travel diaries are thus a fundamental source for every well-designed mobility simulation as it lays the ground for mode choice modeling. This travel diary data with a limited geocoded resolution should be made available to everybody, even commercial companies. Because such data is the catalyst that allows the evaluate and development of new mobility services and business concepts that are necessary to reach the so called “Verkehrswende” (turnaround in transport policy). At the end of the day companies must provide services that fits to the requirement of the population. Restricting such central part of mobility modeling data simply cause a loss of innovation. No data, no “Verkehrswende”, Steffen says.

Mobility components are similar across the entire world; there are trains, light rail, cars, (mini) busses, bicycles, scooters almost everywhere. Sometimes you may find auto rickshaws instead of taxis. The mix and its optimization, though, is different in each region. Therefore, assessing and comparing mobility solutions across regions is yet another challenge. But if you think out of the box, you may find inspiration: The Economist has come up with the so-called “Big Mac index” to compare currencies’ valuations with their official exchange rates (the idea is that a product that is globally available in identical form can serve as a reference against which to measure other indicators).

Now, what would be the “Big Mac” of mobility against which to measure individual solutions and by which to see the excellence of one region over another? The accessibility of any given point from any other given point might be a good candidate. But, frankly speaking, we don’t know the answer yet and current research often only works on only one part of the problem in a too conventional way. Steffen sees it that ways too, the complexity of mobility and its intercedences is hard to expressed by a single number.

Thanks again to Steffen for this great talk and for spending his time with us. Special thanks to him for reviewing our draft of this post and adding significant details.