This paper reports on the effects of pragmatic design decisions concerning early on-line motorway traffic monitoring equipment on its ability to support accurate travel time estimation during congestion, by comparing on-line travel time estimates on a number of routes with observed travel times from a licence-plate survey. It turns out that the combined effects of design decisions aimed at producing a reliable no-frills monitoring system, and the technology used, have curtailed the flexibility of such traffic monitoring systems and now limit their suitability for new tasks such as traffic state estimation and travel time estimation. One of the test sites used in DACCORD is described, and results obtained with the DACCORD algorithms here are reported. Next the effects of instrumentation peculiarities on the quality of travel time estimates, compared to the true travel times obtained from a license plate survey, are discussed. This paper shows that the effects from such peculiarities can be remedied by estimating traffic speed during heavy congestion from a speed-flow diagram. (A*)
Abstract