On Sensors in Transportation Asset Management
: Applications of Data Science and Mathematical Modelling

  • Tom J Strain

Student thesis: Doctoral ThesisDoctor of Philosophy (PhD)

Abstract

Fixed, remote survey, and ad-hoc sensor deployments are investigated to examine how intelligent transportation technologies can enrich the transportation asset management process. Initially, we consider whether data captured from a structural health monitoring system installed on a
bridge might also provide an estimate of the number of vehicles that travel across it. A case study of the Clifton Suspension Bridge in Bristol, UK is developed. Overall, on a withheld test set, our spectrogram thresholding-based method achieves a 1.36 vehicle accuracy, outperforming an existing state-of-the-art system ROCKET.
Second, a computer vision-based decision support system, designed to automatically verify an inventory of roadside assets, is developed and tested with assets along an eight-kilometre section of the A27 highway in England, UK. To develop the system, remote survey data from a vehicle equipped with a forward-facing camera and a GPS-enabled inertial measurement unit (IMU),
aerial highway imagery, and the asset inventory, are fused. Overall, 91% of assets in a withheld test set are automatically verified, thus greatly reducing an analyst’s manual workload. We also explore how visual simultaneous localisation and mapping (vSLAM) systems might make our method more robust, by refining or replacing missing IMU data in GPS-denied environments.
Finally, we consider a case study of how the Highways England traffic officer (TO) fleet might capture asset data across the strategic road network in England, UK, alongside their primary function of incident response. The TOs patrol under one of two distinct regimes: one that aims to minimise the fleet’s incident response time, and one that aims to maximise the fleet’s coverage (for
asset data capture). Comparison of the network coverage and incident response times achieved under each regime shows that fleets (of various sizes) can successfully be deployed for asset data capture while only increasing incident response times by a few minutes.
Date of Award21 Jun 2022
Original languageEnglish
Awarding Institution
  • The University of Bristol
SupervisorAndrew Calway (Supervisor) & R E Wilson (Supervisor)

Cite this

'