Research and development of models and algorithms based on machine learning processes, aiming to detect traffic jams at stops or segments, predict the time of arrival of a tram and provide users with other relevant information to reduce traffic jams and improve user experience with additional content for passengers and other entities in the urban mobility network.
A platform for urban mobility was developed as a result of research and development conducted within the EU project “Istraživanje beacona u svrhu izgradnje mreže kretanja – razvoj platforme za urbanu mobilnost” (Beacon research aiming to build a mobility network – urban mobility platform development). The project was implemented by the company Trilix d.o.o. from Zagreb and the Faculty of Electrical Engineering, Computer Science and Information Technology Osijek (FERIT). The total value of this RDI project was HRK 3,526,917.02, and it was co-financed by the European Union’s Regional Development Fund in the amount of HRK 2,241,146.27. Trilix d.o.o. from Zagreb and the Faculty of Electrical Engineering, Computing and Information Technologies Osijek contributed with HRK 1,285,770.75. The project started on 1 June, 2018 and ended on 2 September, 2020.
For testing purposes, the company Gradski prijevoz putnika d.o.o. Osijek (GPP d.o.o. Osijek) provided the project beneficiaries with an access to a part of its transport infrastructure. The platform for urban mobility is based on low-energy devices known as beacons, which enable Bluetooth communication and precise determination of a relative distance between the users. The beacons monitor vehicle and passenger movement, as well as other information. Approximately fifty beacons were installed along the Zeleno polje – Višnjevac route and around ten beacons were installed in trams travelling that route. Mobile application, developed as an integral part of the urban mobility platform, allows the users to enter information on traffic conditions, such as the number of passengers at stops or the time elapsed travelling between specific stops. At the same time, the app automatically collects data from surrounding beacons, and all the data serves for the development of models and machine learning processes, which should be able to predict the future traffic conditions as accurately as possible.
To collect a sufficient amount of data, the students of FERIT tested the urban mobility platform by entering data on traffic conditions in March, June and July. The collected data were analysed by researchers of FERIT Osijek, who used the results of the analysis to develop models and algorithms based on machine learning processes. Those models and algorithms aim to detect jams at tram stops or segments, predict the time of tram arrival, and provide the user with other information that could help reduce traffic jams and enhance the user experience, with additional content for passengers and other urban mobility participants.