This paper presents a data flow approach to predict transit vehicle arrival/departure. The prescriptive approach breaks the prediction tasks into several component parts. Three components are identified (a tracker, a filter and a predictor), that are necessary to use automatic vehicle location (AVL) data to position a vehicle in space and time and then predict the arrival/departure at a selected location. Data, starting as an AVL stream, flows through the three components, each component transforms the data, and the end result is a prediction of arrival/departure. This prescription provides a framework that can be used to describe the steps in any prediction scheme. The authors describe a Kalman filter for the filter component, and present two examples of algorithms that are implemented in the predictor component. AVL data from Seattle, Washington, and Portland, Oregon, are used to create examples of transit vehicle prediction systems. The examples demonstrate that there is a significant gain in information for passengers using the AVL data with an implementation of this prescription. Results also show that the overall prescription can be useful in a transit setting that has a fleet of transit vehicles, each equipped with a transmitter, that travels along prescribed routes, and that has a transit database that defines the schedule times and the geographical layout of every route.


  • English

Media Info

Subject/Index Terms

Filing Info

  • Accession Number: 00961893
  • Record Type: Publication
  • Files: TRIS, ATRI
  • Created Date: Sep 1 2003 12:00AM