Uncertainty in travel time is considered a larger discomfort by travelers than the travel time itself. An understanding of the mechanisms that cause traffic congestion is the key to alleviate the impact of congestion. This is the reason Daphne van Leeuwen of CWI’s Stochastics group focused on the following research question during her PhD: How to design effective control mechanisms to reduce or prevent congestion road networks in the omnipresence of uncertainty?
Current technological developments, such as advances in automated driving and developments in sensor technology, allow for better regulation of road traffic in transportation systems. Accurate monitoring of the network state combined with the coordination and cooperation of individuals, paves the way for more reliable and efficient usage of infrastructure.
In order to get a complete view of this complex issue and the impact of uncertainty, Van Leeuwen took into account the variability of traffic demand and road capacity in the deployment of roadside systems, as well as the behaviour and interaction of travelers. She also investigated how to partition the network to determine the optimal control points to manage traffic.
During her PhD Van Leeuwen combined models from queueing theory to the application of road traffic. She extended classical models from transportation theory with techniques known from queueing theory. The results give structural insights into the impact of variability under varying conditions, such as large or small capacity roads, begin and end of the peak hour. Moreover, a peak spreading algorithm has been developed, to study the impact of departure time shifts by a central coordinator on congestion levels. The system collects all the departure time information and schedules of participating travelers according to their given departure time preferences. This approach shows that a large reduction in average queue length can be achieved when only a small percentage of travellers participate and that the participating travelers benefit the most from this approach.
The developed models lead to new types of modelling approaches that can more accurately capture the traffic state. Approximation techniques are developed that increase the computational speed, making it possible to apply for road traffic applications.