Despite increasing digitalisation and networking, current traffic systems are largely characterised by incomplete information of all road users. Due to this lack of information, especially about non-motorised road users such as pedestrians and cyclists, they often cannot be sufficiently taken into account in traffic control and road safety measures. Although the quality of individual information for motorised road participants is increasing due to the continuous improvement of sensors and situation analysis capabilities of modern driver assistance systems, the availability of this data for central traffic control systems is usually not given. On the public side, this reduces the possibilities for global optimisation and increased efficiency for all road users and thus leads to unnecessary waiting times due to imperfect data availability and information. The present project proposal addresses this issue and attempts to achieve increased efficiency in the areas of traffic control and road safety by raising this previously mostly inaccessible data across all road users. Through data fusion and the use of AI processes, it is being examined how traffic light systems can be optimised through more comprehensive, real-time information about all road users and how road safety can be increased, especially at intersections as accident hotspots.

In the context of the growing digitalisation and networking of transport systems, the potentials of AI for multimodal traffic control in urban transport are shown and demonstrated in this project. To this end, new traffic control systems will be developed using AI methods, applied and tested in the existing traffic system of the city Ingolstadt under real-life network-wide and local conditions. The control systems should react strategically to the current traffic situation by using a wide variety of data sources, some of which are new, such as vehicle fleets, public transport vehicles, cyclists and locally recorded sensor data from pedestrians and other road users in situ and with minimal latency or also network-related. This is intended to increase traffic safety on the one hand and optimise the traffic flow and the performance of the infrastructure on the other hand through a two-way interaction between vehicles and traffic signals. This contributes to the reduction of traffic-related emissions.

The project is thus divided into a global consideration of the traffic flow in the network and a local consideration of a selected and limited High Definition Test Field (HDT), which is additionally equipped with stationary sensors in order to record all road users with high precision at multi-modal intersections. Depending on the use case, the AI methods are implemented online (for ongoing analysis and optimisation) or offline (to analyse recurring situations in the data collected over a longer period of time or to evaluate the effectiveness of changes in traffic control).