Decentralized Approaches to Adaptive Traffic Control

Traffic systems are highly complex multi-component systems suffering from instabilities and non-linear dynamics, including chaos. This is caused by the non-linearity of interactions, delays, and fluctuations, which can trigger phenomena such as stop-and-go waves, noise-induced breakdowns, or slower-is-faster effects. The recently upcoming information and communication technologies (ICT), including cheap optical, radar, video, or infrared sensors and mobile communication technologies promise new solutions leading from the classical, centralized control to decentralized approaches in the sense of collective (swarm) intelligence and ad hoc networks. Such concepts reduce the problem of data flooding by restricting to the locally relevant information only and reach more adaptiveness, flexibility, resilience and robustness with respect to local requirements and temporary failures.

One focus of my talk will be adaptive cruise control (ACC) systems, which do not only increase the comfort and safety of car passengers, but also enhance the stability of traffic flows and the capacity of the road. We call this ``traffic assistance''. Specifically, I will present an automated driving strategy that adapts the operation mode of an ACC system to the autonomously detected, local traffic situation.  The impact on the traffic dynamics is investigated by means of a multi-lane microscopic traffic simulation.

In the future, vehicles will become automatic traffic state detection, data management, and communication centers when forming ad hoc networks

through inter-vehicle communication (IVC). The applicability of short-range inter-vehicle communication for the detection of dynamic congestion fronts on freeways will be shortly discussed. Adaptive, self-organized traffic control in urban road networks is another interesting application field for decentralized traffic optimization strategies. Therefore, this talk will also discuss control principles that allow one to reach a self-organized synchronization of traffic lights. It turns out that a local optimization fails to reach high performance, and that a stabilizing strategy is not successful as well. A suitable combination of the two, however, can perform superior to centralized, cycle-based approaches.

The focus of this talk will be adjusted to the interest of the audience.

Talk