In the project CoMobility (Co-designing Inclusive Mobility), researchers from the Warsaw University of Technology work on the methods of machine learning and Big Data analysis. These modern tools are used for data on Warsaw and its agglomeration (such as public transport routes, survey results and data from the transport model). They cooperate, inter alia, with specialists on social research and transport modelling (from Warsaw University and SGH Warsaw School of Economics), with a team dealing with air pollution modelling (from the Norwegian Institute for Air Research) and with experts on joint development of local community initiatives (Foundation Na Miejscu).

First results

– At present, among other things, we are identifying the scale of the impact of various factors on transport choices made by citizens – explains Professor Maciej Grzenda. – We have developed a system for collecting data on inter alia timetables and GPS traces of of public transport vehicles, using  Apache NiFi, Apache Flink and Apache Hadoop platforms and other software modules, including novel models developed by our team. The system architecture also includes proposed new modules for modelling transport choices with machine learning methods.

The developed solutions are already operational and provide useful information.

– They allow to determine characteristics such as frequency of connections of different types for various starting points and destinations, or occurrence of possible deviations from planned transport times, which may be caused by, for example, road incidents – lists Professor Grzenda.

Combining the data with survey results has allowed to identify the relationships between the characteristics of, for example, public transport connections, distances and time of travel, reason for travelling and the chosen means of transport. The results allow to answer the question in what situations the citizens choose to travel in an environmentally-friendly way: on foot, by bike or public transport. Identifying these dependencies is an important objective of the CoMobility project.

How to use the knowledge?

Our researchers have presented the results of their work at prestigious events for machine learning specialists: Marcin Luckner, PhD, at the ICCS conference (International Conference on Computational Science) in London in June, and Professor Maciej Grzenda and Przemysław Wrona, MSc, in September at the ECML-PKDD conference (European Conference on Machine Learning and Principles and Practice of Knowledge Discovery in Databases) in Grenoble in France. Grzegorz Bagrowski, Łukasz Brzozowski and Adrian Trzeciak are also part of the research team.

– The work done so far has been connected with, among other things, the use of drift detectors to analyse the data stream of vehicle locations in order to distinguish temporary fluctuations in vehicle punctuality on specific routes from significant changes – explains Professor Grzenda. – We are planning, inter alia, to use and develop methods dedicated for concept drift, which in this case means, e.g., evolution of transport choices in time, resulting from, for instance, increasing awareness of the impact of the choice of transport methods on air pollution. We are also planning to develop transfer learning methods to analyse transport issues, i.e., methods which enable the use of knowledge collected in, e.g., one agglomeration to develop more accurate models for another agglomeration.

The CoMobility (Co-designing Inclusive Mobility) project  is conducted by an interdisciplinary team of research units from Norway and Poland, which comprises, apart from the Warsaw University of Technology, Warsaw University (consortium leader; Principal Investigator: Anna Nicińska, PhD), Norwegian Institute for Air Research, SGH Warsaw School of Economics, Fridtjof Nansen Foundation at Polhøgda, Foundation Na Miejscu and the City of Lublin.

The CoMobility benefits from a 2.05 million€ grant from Iceland, Liechtenstein and Norway through the EEA Grants. The aim of the project is to provide a package of tools and methods for the co-creation of sustainable mobility in urban spaces.