x
erre ä
données
nées de
iler dans
5 oblige
données
gnements
ré. Une
aide de
rmation.
ettre au
érer les
| la. U.S,
1 Change
osant de
emes de
1 Change
inateur,
auses et
comprend
ctés et
e Earth
X. Par
nal sont
n Earth
créé le
lito aux
données
ensuite
réseau,
lans des
ent des
es GCDIS
INTEGRATING DIFFERENTIAL GPS WITH AN INERTIAL NAVIGATION SYSTEM (INS)
AND CCD CAMERAS FOR A MOBILE GIS DATA COLLECTION SYSTEM
El-Sheimy N. and K. P. Schwarz
Department of Geomatics Engineering, The University of Calgary
2500 University Drive, N. W. , Calgary, Alberta, Canada T2N 1N4
ISPRS Commission II, WG II/1
KEY WORDS: Highway survey system, Sensor integration, Geographical information system (GIS) , CCD cameras, Inertial
navigation system , Differential Positioning with GPS.
ABSTRACT
The creation of a Geographic Information System (GIS) for road networks requires large amounts of data which currently are
obtained by manually digitizing existing maps with some update from conventional photogrammetric and surveying
techniques. In order to collect digital data faster, the use and integration of digital sensors is required which work in kinematic
mode, i.e. from a moving vehicle. The University of Calgary has developed a precise mobile survey system for road inventory
and general GIS applications. The system integrates a cluster of CCD cameras, a GPS receiver, and an Inertial Navigation
System (INS), to automatically collect data in a road corridor at velocities of 50 - 60 km/h and to store this data in a GIS
system format. The updated GPS/INS information is used to geometrically correct the images collected by the CCD cameras
which record all details along the highway within a corridor of about 50 m. The shutters of the cameras and the output of the
INS system are synchronized by the clock of the GPS receiver.
In this paper an overview of the sensor integration and the mathematical transformations required to convert the image
coordinates into 3-D coordinates will be given. Also, some data storage, merging and manipulation problems will be
addressed. Special emphasis is given to the contribution of each subsystem to the overall error budget of the derived 3-D
coordinates which have been assessed using test results recently obtained.
1. INTRODUCTION
During the last decade, the demand for GIS in management
and design applications has greatly increased. The
realization of a GIS in many application areas still suffers
from data acquisition problems. To be of value to the user, a
GIS must be updated regularly, so that the information in the
data base correctly represents the real world. However, the
acquisition of up-to-date GIS data by conventional survey
techniques is prohibitive in cost and has therefore limited
the applicability and usefulness of GIS to potential users.
Described in this paper is an attempt to overcome this
problem.
A data acquisition system has been designed and
implemented at The University of Calgary (U of C) that can
be used to selectively update GIS data bases very quickly and
inexpensively. The system integrates a cluster of Charged-
Coupled-Devices (CCD) cameras, an Inertial Navigation
System (INS), and satellite receivers of the Global
Positioning System (GPS). Figure 1 shows the system in
schematic form. The overall objective was the development
of a precise mobile highway survey system that could be
operated at a speed of 60 km per hour and achieve an accuracy
of 0.3 m (RMS) with respect to the given control and a
relative accuracy of 0.1 m (RMS) for points within a 50 m
radius. This accuracy is required in all environments
including inner cities, where a stand-alone GPS is not
reliable. Sensor integration has been optimized to reach the
requirements of the survey market. The data flow has been
streamlined to facilitate the subsequent feature extraction
process and transfer into a GIS system. The new system is
named VISAT and derives its name from the fact that it
utilizes Video images, an INS system, and the GPS Satellite
system. For further details on the design of the VISAT
system, see Schwarz et al (1993a) and El-Sheimy and
Schwarz (1993); for the workstation design, see Li et al
(1994).
This paper gives an overview of the most important
components of the VISAT system, discusses the underlying
principle of extracting three dimensional (3-D) coordinates
from the video images through the use of INS/GPS data,
highlights some of the integration problems, and analyses
results recently obtained in field and laboratory tests.
2. SYSTEM CONCEPT AND FUNCTIONALITY
The hardware component of the VISAT system consists of
three sensor systems, two ASTECH PXII GPS receivers, a
LTN 90-100 strapdown INS, and two COHU 4912 CCD
cameras.
The GPS is capable of providing very accurate position and
velocity under ideal conditions. However, such conditions
do not often exist. Independent GPS navigation requires at
least four satellites. To reach the accuracy required for the
VISAT system, it has to be used in differential mode, e.g.
using double differencing techniques. The major drawback of
241
