Investigating the opportunities and challenges of increasing vehicle automation for monitoring components in traffic management applications
Accurate traffic state estimation and prediction is important for effective traffic management measures. At the moment, Eulerian sensors, in many cases loop-detectors, are the primary sources of information for traffic state estimation and prediction. Due to the development in and increasing availability of technology, new types of sensors are being used in traffic. For instance, information from smartphones and navigation systems is already widely used in commercial navigation services. Within vehicles an increasingly number of interesting sensors is installed. These sensors can be used by the vehicle itself to perform automated vehicle tasks. An example of an automated function is the Adaptive Cruise Control (ACC), which requires information on the vehicle speed, distance to and relative speed of the downstream vehicle. The traffic information gathered by sensors may be valuable for estimation/prediction within a traffic management application. Therefore, I want to investigate which opportunities there are for using sensor information which can potentially be used due to increasing vehicle automation for traffic state estimation/prediction. However, increasing vehicle automation also poses challenges. For estimation/prediction traffic flow models are used which describe the behaviour of the traffic flow. Due to vehicle automation this behaviour can change, which can lead to more inaccurate models and consequently traffic state estimation/prediction. Therefore, this challenge will also be addressed.
Fusion of heterogeneous data-types, describing traffic behaviour in (new) traffic flow models.
Governments invest large amount of money to improve the mobility in their countries. A part of these investment goes directly to installing and maintaining road-side sensors used for traffic state estimation. Traffic state estimation is essential for traffic management application which contribute to a better mobility. In this research, I opt to develop estimation methodologies which reduce the reliance on these expensive sensors by using existing vehicle sensors. Furthermore, I will consider changes in driver behaviour due to increasing vehicle automation and opt to make our traffic state estimation methodologies ready for these developments.
Paul B.C. van ErpStart/end date: 01/09/2015 – 01/09/2019
Daily Supervisor(s): Victor Knoop
Promotor: Serge Hoogendoorn, Marieke Martens (Universiteit Twente)