ISPRS, Voi.34, Part 2W2, “Dynamic and Multi-Dimensional GIS", Bangkok, May 23-25, 2001 
A STUDY ON VEHICLE POINT CORRECTING ALGORITHM IN GPS/AVL SYSTEMS 
HongShan NIU 1 Jie XU 2 Hong LI 3 
1,3:College of Electronic Science and Technology, 
2:lnstitute of Information Engineering, 
National University of Defense Technology, Changsha, 
Hunan, China, 410073 
E-mail: 1 hsnie@371.net 2 xjxj@371.net 
KEY WORDS: AVL System, GPS Data, GIS, Maplnfo Map, and Correcting Algorithm 
ABSTRACT 
AVL system is a successful application of the integration of GIS and GPS. When GPS data is displayed on a GIS map, logicallythe 
vehicle point should be displayed on the roads. But because of the effect of errors, there are always warps between the true vehicle 
point and the displayed point on the map. It is necessary to correct these warps. In this paper, we put forward two algorithms: Linear 
Abstraction Algorithm (LAA) and its improved one--Buffer Linear Abstraction Algorithm (BLAA). Experimental results show that both 
algorithms have perfect precision. And the shortcomings of the algorithms are also pointed out. 
1. INTRODUCTION 
Nowadays, the application of 3S (GIS, GPS and RS) is 
developing towards the direction of integration. In this kind of 
integration, GPS is used to provide spatial position information of 
objects real-timely and quickly; RS is used to provide semantic 
or non-semantic information of objects and their environment 
real-timely or quasi- real-timely, while GIS is used to do some 
integrating disposal > integrating management and dynamic 
access to multi-source temporal and spatial data. GPS/AVL 
system is a successful application of the integration of GIS and 
GPS. 
In a GPS/AVL system, when a GPS mobile object (a vehicle, for 
example) is displayed on a map, logically the vehicle point 
should be displayed on the roads. But because the effect of 
low-precision of map, errors of map digitizer, errors of GPS 
system, errors of GPS receiver, there are always warps between 
the true vehicle point and the displayed point on the map. But the 
precise position of a mobile object is a key to the system, so it is 
badly necessary to correct these warps. The authors put forward 
a correcting algorithm- Linear Abstract Algorithm (LAA) and its 
improved one—Buffer Linear Abstract Algorithm (BLAA). 
2. A STUDY ON THE ALGORITHM 
In brief, LAA and BLAA are such algorithms that we should first 
make linear abstract of the road layer (Put these line segment 
objects into a single Maplnfo hidden layer) of a map and then 
make correction based on the hidden layer. Before correction, 
we process the map according to the request of algorithm design, 
which includes creating a linear layer to simulate the road layer 
and creating a layer which contains objects representing 
crossroads and so on. 
2.1 LINEAR ABSTRACTION OF ROADS 
Linear abstraction of roads is the foundation of LAA and BLAA. 
In a Maplnfo map, roads are represented as region objects, so 
when we create the linear road layer (We call it linear layer), it’s 
naturally for us to apply central lines to simulate road objects. 
Thus we get the linear layer. It is not necessary to be very 
precise when you draw the lines, but they must be able to 
represent the roads. If a road is not very straight, we use several 
line segments to simulate it. The lines in the linear layer are used 
to calculate the position of corrected vehicle point. 
2.2 GET THE CROSSROAD LAYER 
Crossroads are special points for vehicles: vehicles may turn left 
or turn right, vehicles may stop before a red light and the names 
of roads may change and so on. So we must deal with these 
conditions specially. Just like the linear abstract of road layer, we 
should also create a crossroad layer, that is, create a layer that 
contains objects situating at the positions of the crossroads on 
the road layer. This layer will be used to judge the spatial 
relationship between vehicle points and crossroads. Fig. 2 shows 
the crossroad layer.