Stefan Feit

Adaptive traffic light prediction via Kalman filtering

Current fields of research in the automotive sector are dealing with the development of new driving-assistance-functions that aim to improve security, efficiency and comfort of vehicles. A significant field of study represents the prediction of traffic signals ahead that enable innovative functionalities such as Green Light Optimal Speed Advisory (GLOSA) or efficient start-stop control. This paper deals with the challenges of predicting future signals of traffic-adaptive traffic lights. First of all, we extract important characteristics of adaptive traffic lights and the underlying traffic situation at crossings relying on historical data of several Munich traffic lights. Based on these insights, we present and evaluate a generic model to predict future traffic-adaptive traffic signals at crossings. We show that with the proposed model, 95% of future signals can be predicted with an accuracy of 95% at best. On average, 71% of future signals can be predicted with an accuracy of 95% for the considered traffic lights.

Learning Traffic Light Parameters with Floating Car Data

The knowledge of traffic light parameters, such as cycle plan or future signal phase and timing information (SPaT) of traffic lights is the base for a vast number of use scenarios. A few examples are traffic signal adaptive routing, green light optimal speed control, red light duration advisory or efficient start-stop control. The basis for all these functionalities is the knowledge on the correct traffic light cycle time, i.e. the periodicity of the traffic lights signaling sequence. With a correct cycle time given, green start and end times can be derived from periodically reoccurring movement patterns. In this paper, we propose a method to reconstruct a traffic lights cycle plan through the interpretation of the recorded information on a vehicles movement pattern (trajectory) in the intersection area. The recorded trajectories are temporarily sparse and and the cycle plan changes frequently. Therefore, we propose a model that focuses on the performance on very limited available trajectory data and yet is robust with regard to estimation errors. We show that our approach is able to detect the correct cycle time with already 30 trajectories at an accuracy of 99%.

The Research Initiative UR:BAN

The automobile and road traffic have made in the course of their joint development great progress. Still, to be ready for the future, the entire transport system has continuously to meet new and higher requirements. Each trip shall be traveled safely, efficiently and comfortably. The progressive urbanization leads to more and more people living in urban areas. The additional large number of commuters and supply transport should not be underestimated. This moves mobility spaces increasingly into the urban space and individual traffic using cars represents the majority.

Stop Line Detection Using Satellite-Image Segmentation

Current fields of research in the automotive sector are dealing with the development of new driving-assistance-functions that aim to improve safety, efficiency and comfort of vehicles. All these assisting functionalities necessitate a considerable amount of information from the vehicles environment. Autonomous cruise control (ACC) or efficient start-stop control for example require knowledge on the position of stop lines of crossings. In this paper we focus on the detection of those stop lines using satellite image segmentation. We thereby merge topological information from multiple sources (navigable maps, satellite images) for a region-of-interest-determination. Besides, we exploit knowledge of certain characteristics of stop lines, such as color, form, minimal estimated distance and relative position to the center of a crossing. We develop an automated image processing system that detects stop lines and provides the according geo-coordinates.

Traffic Light Assistance – Ein innovativer Mobilitätsdienst im Fahrzeug

Die immer weiter zunehmende Verkehrs- und CO2-Belastung in urbanen Raumen einerseits, die Verfugbarkeit von immer mehr Verkehrs- und Infrastrukturinformationen andererseits begunstigen die Entwicklung neuer Assistenzfunktionen wie den Traffic Light Assistant im Automobil. Derartige Funktionen benotigen vielerlei unterschiedliche Verkehrsinformationen, die derzeit noch nicht auf einer Marktplattform angeboten werden. Mit dem Traffic Light Assistant und dem Mobilitatsdatenmarktplatz MDM als Plattform fur den Austausch von Verkehrsdaten wird ein wichtiger Schritt in diese Richtung unternommen. Wir stellen den Traffic Light Assistant im Rahmen unseres Forschungsprojektes Alpha Munchen vor und zeigen die damit einhergehenden Chancen fur neue Geschaftsfelder und Businessmodelle im Rahmen des MDM auf.