Smart Traffic Signal Systems

  Logos of project participants
 

Contact

Photo of Dr.-Ing. Adrian Fazekas © Copyright: ISAC

Name

Adrian Fazekas

Chief Engineer

Phone

work
+49 241 80 25227

Email

E-Mail
  Chart explaining the project Copyright: © ISAC

Problem Definition

Pedestrians differ elementarily in the respective walking speeds at which they cross a street: Certain groups of people need a longer time window than others, for example elderly, blind and handicapped people. This repeatedly leads to stressful and dangerous situations for the affected groups of people. A significant part of today's traffic signal control systems does not take this into account to the required extent, both with regard to the release time start or release duration and in connection with conditionally compatible traffic flows. The latter in particular leads to a significant proportion of accidents with personal injury occurring at intersections with traffic signal systems (LSA).

Project Objective

The aim of the project is to increase the safety and comfort of pedestrians and thus to predictively prevent accidents. This is to be realized by a newly developed traffic signal control system based on intelligent sensor technology. This will take into account differences in pedestrian walking speeds, the traffic situation and composition, as well as patterns in the occurrence of dangerous situations.

Implementation

The pedestrian crossing area is detected with camera sensors. This requires a higher resolution to detect relevant pedestrian features. Intersections and their arms are captured using thermal cameras to obtain only the data needed to extract trajectories of road users. The protection of personal data is thus ensured. Based on the sensor data, a newly developed data processing takes place.

Firstly, AI algorithms are to be used to derive a crossing request at an early stage based on the distance traveled or recognized movement patterns of the pedestrians and, for the first time, also to predict the time expected for the crossing. Secondly, image-processing algorithms will be used to detect and classify road users based on the thermal images and to record their trajectories. On the one hand, this makes it possible to provide data such as the motor vehicle traffic volume, the number and composition of waiting road users or similar, to the traffic light control system without additional detection equipment, e.g. induction loops and the civil engineering required for this. On the other hand, a real-time online safety analysis can be performed.

Based on this data base, an LSA control system with a significantly higher flexibility than before is to be developed. Such a traffic light control system should make it possible to flexibly start or end the green times to be switched for pedestrians, as well as to dynamically shorten or lengthen them, depending on the walking speeds, the crossing requests and the traffic situation, thus increasing the comfort and safety of pedestrians. In addition, the LSA control allows for the first time appropriate and demand-driven interventions based on real-time safety analysis.