SMaRt
Supporting the Man-in-the-loop in Roadtunnel
Problem Definition
In recent years, the workload for personnel in tunnel control centers has grown steadily. This can result in overloads that potentially affect traffic safety. Currently, only two-stage systems for monitoring traffic flow in road tunnels are in use: Firstly, automated incident detection is carried out by means of sensors, which is forwarded to the monitoring personnel for assessment. The final detection of the incident or the cause of the incident is therefore carried out by the employees in the monitoring centers. Depending on the reliability of the automated incident detection systems used, this means more or less considerable work for the so-called "man-in-the-loop" employees.
Modern detection systems are usually extremely reliable and helpful. As an example, we would like to refer to the detection system of aveVerkehrs- und Informationstechnik GmbH, which is also used in this project. In line with the latest research, it is based on the sensor intelligent induction loop. Nevertheless, the aim is to achieve a significant innovative leap by means of modern, AI-based algorithms as well as data fusion with a video camera system.
Project Goal
The integration of adaptive algorithms, both in the automated incident detection and in the combination of the robust alarms generated in the process - in short, the integration of AI in the data fusion - is intended to significantly simplify the working conditions of the monitoring personnel. In this way, a significant contribution will also be made to the safety of traffic flow in the tunnel.
The selected consortium consists of an SME (ave GmbH), a research institution (ISAC/RWTH), and a Autobahn GmbH, a tunnel and infrastructure operator (Northern Bavaria branch). This cooperation covers all essential levels and thus forms an optimal prerequisite for successful achievement of the objectives.
Implementation
In this project, adaptive algorithms (AI) are developed, enhanced and applied. This is being done with the aim of making both intelligent induction loop detection and post-designed video detection more robust, efficient and reliable. The individually improved systems will subsequently be combined in a common detection system in order to achieve further optimization in this way. The core work of the research project is, among other things, the transfer of knowledge between the ISAC of RWTH Aachen University and ave GmbH in the areas of AI and video detection.