When vehicular traffic is dense but flowing, phantom traffic jams are likely to arise: waves of brake lights travel up the highway without apparent reason. Mathematical models reveal how traffic is similar to water waves, dunes, and cosmic clouds; and how just a few self-driving cars can change traffic flow completely.

Traffic flow is a fantastic example of a system that exhibits complex collective dynamics, while being composed of many agents (vehicles with drivers) that are governed by local vehicle-vehicle interactions (safety-driven) as well as large-scale decisions (route choice, average speed choice, etc.). Key research fields concerned with vehicular traffic flow are: (i) traffic assignment, i.e., how many drivers take which trips at what time, and which routes to take; (ii) traffic estimation and control, i.e., how can sensors yield a real-time picture of the traffic state on the road network and how can ramp metering, variable speed limit signs, and traffic lights be controlled to optimize throughput; and (iii) traffic flow theory, i.e., how do the emergent flow structures on the road evolve and depend on the local vehicle interactions. This last aspect is the focus of this video.

A great reading introduction to traffic flow theory is given by various wikipedia articles, including those on traffic flow https://en.wikipedia.org/wiki/Traffic_flow, traffic jams and the fundamental diagram of traffic flow. Multiple video lectures on traffic flow theory are provided by the video tutorials on traffic flow by the Institute for Pure and Applied Mathematics and UCLA.

Specifically related to the video, one can find a layman’s description on phantom traffic jams and traffic waves, with more graphics and videos, on the traffic modeling website and the website. Moreover, one can find a video that explains how “Just a few autonomous cars can improve traffic for everyone”. The experiment shown in the video was the first experimental demonstration that with fewer than 5% vehicles with automated velocity control, traffic waves can be dampened and prevented from happening, resulting in the overall fuel consumption of traffic reduced by 40%, and emissions of nitrogen oxides (pollutants) reduced by 70%. The detailed results can be read in the original research article and its preprint.

Very cool simulation tools that enable anyone to experiment with traffic flow dynamics (and experience its complexity and its beauty) are the simple graphical traffic simulators, e.g., by M. Treiber and D. Helbing or by A. Volkhin and well as the complex simulation software (free to use) SUMO which is powerful, yet simple enough to be usable with only a small amount of reading/training.

Next Section »