A data fusion framework for travel time estimation in urban traffic networks
At Mathematics in Transport Conference IV.
01 Sep 2005.
Full text not available from this repository.
Underlying all attempts to manage urban traffic congestion is the need for a comprehensive knowledge and understanding of the state of all parts of the network at all times. This need has given rise to a diverse range of real time traffic detection methodologies and while much research on network state estimation has been carried out based on these data sources, the different characteristics of the detection datasets often potentially produce differing (if not conflicting) estimates of either the absolute value or the short term trend in urban travel times.
This paper presents a theoretical data fusion framework which enables two of the commonest forms of real time traffic data to be combined to create a single best estimate of travel time along an urban road segment. It is proposed that rather than the unidirectional evolutionary approach of many existing travel time estimation systems, improvements in the absolute accuracy of the travel time estimates and reductions in time lag effects can be achieved by combining the currency (but limited spatial relevance) of inductive loop data with the accuracy (but post-event nature) of number plate recognition and matching data.
This is achieved by applying a principle of memory to the estimated travel time series where, rather than being discarded as the estimated series evolves, previous estimates are reassessed when their accuracy is revealed at a future moment in time, with the revealed error being translated back to the current time point to improve the accuracy of future estimates. This paper proposes a generic data fusion framework based on this principle, building on both existing and emerging real time traffic detection systems and existing research into urban travel time estimation.
Conference or Workshop Item
|Venue - Dates:
||Mathematics in Transport Conference IV, 2005-09-01 - 2005-09-01
||23 Jul 2008
||16 Apr 2017 17:52
|Further Information:||Google Scholar|
Actions (login required)