Demand-Responsive Decentralized Urban Traffic Control. Part I: Single-Intersection Policies

摘要

Common logic as well as analytical studies indicate that demand-responsive traffic control can very significantly improve urban street network performance. Yet effective traffic-responsive control has eluded transportation system researchers for many years--the main stumbling blocks being the inadequacy of predicted data from past traffic behavior and the unmet requirements for real-time optimal control techniques. This study develops innovative approaches to the urban traffic control problem using on-line traffic models which relax the reliance on predicted data, and decentralized optimization strategies which are amenable to real-time execution. The new methodologies are particularly suited for implementation via existing microprocessor technologies. This report presents the studies undertaken and results achieved in Phase I of the research. It describes in detail the development and testing of models and strategies for real-time computation of optimal control policies for single intersections. These strategies lay a foundation for the establishment of demand-responsive decentralized urban traffic control on a systemwide basis.