Intelligent Automatic Vehicle Location (AVL) Techniques for Fleet Management

Automatic Vehicle Location (AVL) systems have become increasingly prevalent in a variety of fleet monitoring/management applications, particularly for transit (i.e., tracking buses) and for probe vehicle operation. To date, the majority of these AVL systems report position (and sometimes other vehicle parameters such as velocity) on a periodic basis to a fleet management center. Because data communication costs can be high, sending data on a periodic basis is inherently inefficient. Instead, it is possible to use intelligent filtering techniques that send trajectory data on a non-periodic basis using information on the spatial aspects of the roads and velocity of the vehicle. Thus for the same or less communication cost, greater information can be received and the route representation and/or time-of-arrival predictions can be improved. In this paper, we describe real-time filtering algorithms that have been designed, developed, and implemented in an AVL system. A vehicle was instrumented with the AVL system and subsequently operated on a variety of different roadways, including freeways, arterials, and residential streets. Based on this mixture of driving conditions, it was found that positional trajectories can be effectively recreated (with low error) when up to 50% of the data is removed through the intelligent data filtering method. For velocity data, the effect of filtering was less, where approximately 35% of the data could be removed. Naturally this effectiveness depends on the types of roadways, their curvature, and traffic conditions. However these preliminary experiments show that significant savings (both in communications and data storage) can occur.

Language

  • English

Media Info

  • Media Type: Print
  • Features: CD-ROM; Figures; References;
  • Pagination: 12p
  • Monograph Title: Proceedings of the 12th World Congress on Intelligent Transport Systems

Subject/Index Terms

Filing Info

  • Accession Number: 01015772
  • Record Type: Publication
  • Files: TRIS
  • Created Date: Jan 31 2006 9:56AM