Hansestadt Rostock - Evaluation and further development of the classification of traffic situations with machine learning algorithms

Project manager: Dr.-Ing. Birgit Jaekel

For any inquiries please contact:

Project duration: 07/2019 - 06/2021

External oroject partners: Amt für Verkehrsanlagen der Hanse- und Universitätsstadt Rostock, Straßenbauamt Stralsund

Brief description: The main road network of the Hanseatic and university city of Rostock is often loaded beyond its capacity limits on weekdays and during the peak travel season in summer. Criticism of the resulting traffic jams is often directed at overly rigid traffic control in the city's metropolitan area.

In fact, there is currently no situation-dependent or traffic volume-dependent control, and signal programs are only switched according to predefined schedules. While peak traffic loads based on planning are still handled in the best possible way during rush hour, traffic peaks in other situations usually lead to a collapse of traffic.

In contrast to the other federal states, the traffic volume in Mecklenburg-Vorpommern is extremely influenced by tourism and behaves in the opposite direction to the general trend during the summer months. While traffic figures drop in most metropolitan areas during the vacation season, they reach peak levels in the Rostock metropolitan area. In addition, unpredictable weather-dependent traffic peaks and local disruptions due to events and accidents occur.

The Hanseatic City of Rostock intends to control traffic signals and coordination routes on the main traffic axes according to demand. To this end, time-dependent control is to be replaced by a dynamic system that switches the existing signal programs according to their performance. Control decisions must be made fully automatically on the basis of traffic situations classified by neural networks. These are based on online data such as travel time and traffic density, which are collected by Bluetooth sensors and level-counting stations. The goal is to increase the performance of the overall network with the existing infrastructure.

Assigned research projects:

• Neural Networks
• Artificial Intelligence
• Travel time measurement
• Traffic Simulation
• Traffic control