Adaptive signal control is the subject of an increasing amount of research, as well as development and implementation. Most existing adaptive control systems achieve coordination by applying system control as a constraining layer on top of local control. Some researchers have suggested that, with the right local-control logic, coordination might be achieved as an dynamically emergent phenomenon without the need for a management layer. This paper explores the potential of a self-organizing signal control algorithm using a variety of performance measures. First, the initially reported algorithm performance is reproduced in an idealized environment; next, the algorithm is applied in a realistic road network to compare its performance against actuated-coordinated control, with and without pedestrian phases. Comparisons are made under (1) the same base volumes used to design the actuated-coordinated timing plan; and (2) a variant volume. Self-organizing control is found to be more flexible than coordinated control, and induces a tradeoff in performance among different movement types. Delay reductions of 38–56% are seen in an environment without pedestrian phases. However, with pedestrian phases in recall, self-organizing control performs worse (39% increase in delay) under base volumes, and achieves a weak benefit (6% reduction in delay) under the variant volume. Because of the large total delay reductions in some scenarios, the results show promise for future development.
Traffic signals, adaptive control, self-organization, performance measures
Date of this Version