In cooperation with the Nara Institute of Science (Japan) and the Istanbul IT and Smart City Technologies Inc. (Turkey), BME’s MEDIANETS research laboratory will develop solutions that will be able to predict the intensity of future traffic and forecast potential congestions using artificial intelligence. These predictions will be used to intervene in the control of traffic lights in real time in response to the traffic conditions identified. This will help reduce travel time and harmful emissions.

This solution is significant and unique as no similar solution has been tested in real life situations in a large urban environment in Europe or in Japan before. A key objective of the project is to have the partners test this solution in real life traffic conditions too, in one of the world’s largest cities, Istanbul. The reference solution created in Istanbul in the course of the three-year project is also expected to strongly support the solution’s future marketing as a large number of cities are interested in a similar solution.

Duration: 2024-2027
Source of funding: European Interest Group CONCERT-Japan programme.

https://concert-japan.eu/research/multi-input-deep-learning-for-congestion-prediction-and-traffic-light-control/

The Miracle ITS R&D project aims to address traffic congestion while promoting sustainability through the implementation of dynamic toll pricing. The proposed solution calculates toll rates based on variables such as distance traveled and vehicle occupancy, encouraging carpooling as part of a broader strategy for congestion reduction and sustainable development. High-occupancy vehicles will benefit from lower toll rates, incentivizing more efficient use of road space. Our team is leveraging AI technologies to automate the passenger counting process, ensuring accurate verification of vehicle occupancy and preventing fraudulent passenger registrations. Moreover, the intended solution is smartphone-based, eliminating the challenges and high costs of the traditional road toll collection infrastructure.

Duration: 2024-
Source of funding:

The project investigates the use of machine learning in vehicles on the road to facilitate decision-making in vehicle maneuvering and traffic flow control. The goal is to enable human intervention-free levels (SAE Level 4 and Level 5) of autonomous driving vehicles. Besides developing novel vehicle control models using multi-agent reinforcement learning algorithms for traffic light-free autonomous intersections and dynamic lane allocation on road sections, security aspects in the context of both the RL algorithms and communications are also in the focus of the project.

Duration: 2023-2025
Source of funding: CELSA Research Fund (Central Europe Leuven Strategic Alliance)

The aim of the research is for the CrySyS and MediaNets laboratories todevelop, in collaboration with the Special Service for National Securityunder the umbrella of Artificial Intelligence National Laboratory Program,the competence to improve the security of connected vehicles. As part ofthis, the implementation of different network communications and thesecurity vulnerabilities of the internal vehicle controllers (ECUs) areinvestigated. Particular emphasis is put on analyzing the implementationof Vehicle-to-Everything (V2X) communication, as it is an essential basisfor safe transportation in the future. Advanced algorithms using featureslike sensor data sharing of collective perception services are designed todiscover, filter, and remove misbehaving entities from the Public KeyInfrastructure-based V2X trust domain. Detailed simulation modelscomprising realistic perception pipelines and precise 3D vehicular sensorsuites are developed to evaluate the proposed schemes and techniques.

Duration: 2021, 36 months
Funding: Artificial Intelligence National Laboratory Program

The project developed a federated learning-driven traffic flow categorization approach, aiming to shift from data to model/algorithm sharing. The developed framework provides a common platform for computer network traffic flow measurement, feature computation, and federated model learning, sharing, and deployment. This innovative approach aims to fuel future improvements in operations and management, enhancing the quality and reliability of computer networks and services.

Duration: 2022, 6 months
Source of funding: GÉANT Innovation Programme

The goal of this R&D project is to develop an intelligent transportation system, a new solution for the needs of urban mobility based on artificial intelligence, which will enable substantial progress in real-life control of the city traffic. Our team is focusing on the stream processing of big data, machine vision and prediction of future traffic conditions, based on historical data, using machine learning and classification methods. We are also involved in the intervention strategies (V2X communication and adaptive traffic light control), this way the system will enable city authorities to intervene in a timely and adequate manner and prevent traffic problems before they occur in the city. Our pilot implementation is currently under testing in the city of Pécs in Hungary.

Duration: 2019-2023

Source of funding: National Research, Development and Innovation Office, Hungary

The V2X Communication Research Group of our Laboratory is contracted by Magyar Közút as a partner for the third year now involved in Cooperative Intelligent Transport Systems (C-ITS) harmonization, evaluation and deployment within the framework the European C-ROADS Platform. Besides that our research group is also a member of the two consortia led by Utiber and Roden to design the M9 and M76 smart motorways by the order of the National Infrastructure Developing Ltd. Within the framework of the above collaborations, our colleagues support high-level decision-making, handle ITS-G5 network design tasks, analyze the development of standards, specify test cases, and implement various C-ITS/V2X simulations for comprehensive technology evaluation.

Duration: 2018-

Sources of funding: Magyar Közút, National Infrastructure Developing Ltd.

The AWARD platform increases efficiency of the logistic process from warehouse to last-mile delivery.The easy to implement solution uses intelligent planning algorithms, machine learning and smart robotics (i.e. drones and AGVs). Logistic processes can be enhanced through automation. The AWARD platform does this by improving both the processes in a warehouse and delivery to the end customer. It provides state-of-the-art planning algorithms and a hybrid infrastructure of automated delivery by robots. By implementing AWARD, logistics service providers can streamline their process through increased predictability and reliability, thereby saving on labor costs. For instance, buffers that are used in a warehouse for timely delivery can be reduced. All the while, customers are treated with a modern-day customer experience.

Duration: 2018-2019

Source of funding: EIT Digital