MovingLab – Mobility Data Acquisition with Smart Devices
René Kelpin, DLR Institute of Transport Research
MovingLab combines generic mobility survey methods with main technical functionalities, namely a sensor data based trip and transport mode detection. The paper describes the overall approach, first data collecting field tests and respective findings.
As state of the art, transport, household and mobility surveys apply well-established survey meth-ods: Questionnaires are printed on paper to be individually filled and sent to data acquisition insti-tutes; digitalised questionnaires are made available via the Internet for online surveys; interviewees use digitalised questionnaires embedded into enhanced survey systems for guided telephone interviews. This is also true for large European public survey campaigns, such as survey Mobility in Germany 2016, where mobility aspects of around 130,000 households are being investigated. Weaknesses of methods mentioned, such as missed short trips or underestimated distances and durations, are intensively discussed within the research community, as it was at ETC2016 and successors. At the same time public authorities, e.g. the German Ministry of Transport, request new and innovative data collection methods minimising or even excluding known weaknesses. For a few years, alternative data gathering methods are being discussed; first piloting projects have already been applied on smaller scales – with clear conclusions and recommendations: With smartphone and GPS based methods innovative ICT services can be combined perfectly with enhanced survey designs. Relevant questions can be asked at the best possible moment, e.g. when a trip is just finished and a point of individual interest is reached by survey participants. Trip or activity based knowledge is best available and can be gathered “at first hand”. However, most large European mobility surveys still apply “old-fashioned” methods.
With the DLR facility “MovingLab” such a survey data gathering tool is currently being set up. Hereby, the main development will be the MovingLab App. Two main functionalities will be integrated into the App: First, based on smart devices sensor data (GPS, acceleration, gyroscope, etc.) an enhanced mode and trip detection algorithm constantly detects applied transport modes and performed trips. This self-learning algorithm jointly takes into account mobile devices’ sensor data as well as offline infrastructure data, such as dedicated transport modes’ networks (railway systems) and public transport stations, for most precise transport mode recognition and differentiation. And second, a generic survey tool allows implementing and context based triggering of project specific research questions (by means of questionnaires). In consequence, this App is to be considered a tool – a survey skeleton attached to a main GPS based technical functionality. For any given project purpose or approach, for any given project research question, for any given project survey design, this skeleton can be “bodied” and trained to specific projects’ needs.
MovingLab is intended to be used as an open facility for mobility survey projects. Therefore, in a pilot phase, lasting from November 2016 to December 2017, the overall design and usability is tested by an initial group of 1,000 users. In the main phase by end of 2018, the recruitment of a German representative group of 10,000 users is included. In consequence, the tool MovingLab enables large scale mobility surveys in Germany while granting access to a representative group of test users. However, the MovingLab system can easily be adjusted to and applied by other European survey projects - also facilitating existing user panels.
An initial test setup, a first stable App version and a well-trained mode detecting algorithm, is ex-pected to be on place in summer 2016; a first field test focussing on combined system usability and survey design applicability is planned for a three months’ period from July 2016.
Based on data collected in this field test, the ETC paper will present the general MovingLab approach, the overall facility status and first data analyses available in October 2017. A special focus of the paper will be laid on the mode and trip detection algorithm and its degree of reliability. The paper and its presentation at the ETC2017 also intend to add new aspects to research community’s discussions on alternative mobility data acquisition and to create viable co-operation with European research institutes which are already active on the field of mobility and transport surveys.
Association for European Transport