Vehicle Detection Design (Coverage and Accuracy) and the Performance of Congested Traffic Networks

This paper explores the effects of detection (coverage and accuracy) on the performance of an oversaturated network with traffic signals controlled by a next-generation multi-agent-based system that reacts to demands in real time. Several scenarios on a realistic simulated network are used to illustrate changes in performance when detector inputs are inaccurate and generate discrepancies in the estimation of link demands. Then, these results are compared to the network with accurate detection but a reduced number of sensors. Equivalent system performance from comprehensive sensor coverage but with inaccuracies and reduced number of sensors with increased accuracy can be found. This approach is proposed to be used for detection design, where tradeoffs between performance and coverage are taken into account. Results from different patterns of sensor inaccuracies are also discussed, providing insights on patterns that may have less impact than others. Future work on the effects of demand levels and turning movements is also introduced, as well as a counteracting feedback solution for monitoring detection inputs.