-30- 
4 EXPERIMENTS 
We now describe the organization of the survey and the data 
collected. 
4.1 Organization of the survey 
The first part of the survey lasted for 2 days, for a total of about 
16 hours of work. We chose to take the scans in a real situation 
of territorial scale surveying. Thus, we conducted the 
experiments in an urban area, dealing with problems like traffic, 
poor visibility of the GPS satellites and the planning of 
numerous surveying stations. 
In accordance with the above-mentioned premises, the 
surveying session was organized as follows: 
identification of an appropriate station point for the master 
GPS antenna. The characteristics of this point must ensure a 
good distance from obstacles, like buildings and vegetation, 
and from reflecting surfaces (which could produce 
multipath errors); moreover, it should be in the centre of the 
area to be surveyed so as to have short baselines and, if one 
wishes to use the survey for cartographic and cadastral 
purposes, it should be a vertex of known WGS84 
coordinates (if a vertex of known coordinates is not 
available, one can always carry out a subsequent survey to 
frame the master station in a pre-existing network or one 
can use the international IGS network of permanent GPS 
stations). 
positioning of the laser scanner station on the preset point; 
analysis of the scanner’s field of view and decision about 
the points on which to position the three adapters with the 
rover GPS antennas. The laser scanner is equipped with a 
digital camera that acquires the image of the area to be 
surveyed; with this image, it is possible to decide on the 
areas in which to position the adapters; 
positioning of the adapters and acquisition of the GPS data. 
For each antenna, we set a sample rate of 3 seconds; with a 
session duration of at least 15 minutes, this provides a sufficient 
number of epochs for determination (with centimetric precision) 
of the GPS point in rapid-static mode. To exploit the potential 
range of the instrument (up to 250 m), we tried to choose station 
points of the rover GPS antennas that were well distributed (also 
in depth) within the field of view of the single scan; in this way, 
we could reduce the hinge effects that can result if they are too 
close together (even a slight rotation of the model in the 
junction zone generates a high linear error if multiplied by an 
arm of 250 m). 
- after the adapters were positioned and their effective 
presence in the scanner’s field of view was checked (e.g. 
with a rapid wide-grid scan), we began the actual scanning. 
Assuming that the aim of territorial scale surveying is large- 
scale reproduction (1:500; 1:1000), it is necessary to have a 
precision of the measured detail points of 10 cm. 
As a first approximation, the precision of surveying performed 
with the laser scanner can be estimated as the combination of 
the intrinsic precision of restitution of the single collimated 
point (s ±6 mm) and the spacing of the projective grid. In fact, 
due to the nature of the laser scan, the object is defined by a 
huge mass of indiscriminate points; recognition of the 
characteristics of the object (e.g. restitution of the edges of a 
building) is performed by interpreting the point cloud and 
exploiting the level of detail acquired. We chose to survey all 
the buildings in the field of view with a grid of at least 5 cm. 
This was achieved by setting the projective grid spacing at 5 cm 
at the distance of the farthest building; 
execution of automatic recognition of the targets when 
possible; in fact, for distances less than 80 m, the Cyrax™ 
2500 laser scanner can recognize the shape and reflectance 
of the dedicated flat targets and carry out fine recognition 
with a millimetric point density; 
manual recognition of the most distant targets; this 
important innovation of the scanning procedures of the 2500 
model (with respect to the previous model) was 
fundamental for our experiments. In fact, it was possible to 
manually collimate the target on the point cloud (or the zone 
in which it was situated) and then perform a fine scan of it 
with millimetric point density, thus measuring the same 
level of detail as with the automatic procedure. The same 
procedure would have been very difficult with the previous 
instrumentation because collimation of the scanning zone 
was based only on the low-resolution digital image acquired 
by the scanner and recognition of a small object (like the 
target) at great distances was practically impossible. 
Figure 8. Fine scan on the adapter and the antenna 
Figure 7. The three antennas in the scanner’s field of view