Methods for Improving the Flow of Traffic

Abstract

We describe a distributed and self-regulated approach for theself-organisation of a large system of many self-driven, mobile objects,i.e., cars in traffic. Based on methods for mobile ad-hoc networks usingshort-distance communication between vehicles, and ideas fromdistributed algorithms, we consider reactions to specific trafficstructures (e.g., traffic jams.) Building on current models from trafficphysics, we are able to develop strategies that significantly improvethe flow of congested traffic. Results include fuel savings up to 40%for cars in stop-and-go traffic; we present a number of simulationresults illustrating the underlying mechanisms. In addition, we discussthe organic structure of urban traffic, and hint at how self-healingmethods can lead to improvements in rush-hour traffic.