36 
Figure 9 Observed linear trend in the difference of mile 
chaînage values 
Figure 10 Comparison of the mile chainage measured by the 
odometer in moving and Stop&Go mode. 
4.3 Comparison between GPS and odometer 
From the GPS co-ordinates a mile chainage was computed in 
correspondence of every reference point. Different procedures 
for extracting the mile chainage from the 3D co-ordinates were 
carefully evaluated. The results achievable with different 
approaches strongly depend on GPS data acquisition interval, 
the speed of the vehicle, the regularity of the trajectory and of 
course, the accuracy of the GPS co-ordinates. In particular, the 
procedure for computing the mile chainage by accumulating the 
length of the differential vectors between two subsequent 
instantaneous positions of the vehicle was firstly applied. The 
results show that only with accurate data (dual frequency Post- 
Processed) the mile chainage estimates can be considered 
reliable for positioning also along very irregular paths. In 
general, when dealing with meter accuracy data the cumulative 
procedure results in the accumulation of the errors, too. A 
simple approach to reduce the accumulated error is to use the 
minimum number of GPS positions necessary to describe the 
variation of trajectory from a straight line. To satisfy more 
stringent requirements accurate and dense data can be adopted 
in a smoothing procedure with the aim of limiting the errors in 
the mile chainage. 
In this work in consideration of the regularity of the speed of 
the vehicle and the highway path, a simple approach, based on 
selecting no more than ten GPS positions every kilometre was 
adopted for mile chainage extracting. 
In the following the mile chainage measured by the odometer 
working in Stop&Go mode are taken as reference values. 
Figure 11 shows the results obtained from the comparison of 
the Stop&Go GPS mile chainages and the reference values. 
Figure 11 Comparison between the mile chainage measured in 
Stop&Go mode by GPS receivers and the reference values. 
The systematic effect of about 3 meters can be explained by an 
overestimation of the GPS mile chainage due to residual 
cumulative errors. In order to obtain better results from the dual 
frequency GPS data, a dedicated fitting procedure has to be 
implemented, eventually including also the integration of data 
from other sensors. 
In order to analyse the quality of positioning in a normal 
operational environment the results from data collected in 
dynamic mode are compared. The comparison between the mile 
chainage measured in RTCM and Post-Processed mode by Pro 
XR/XRS GPS receiver and the odometric mile chainage (Figure 
12) shows the same situation discussed before. 
Figure 12 Comparison between the mile chainage measured in 
Stop&Go mode by GPS Pro XR/XRS receiver and the 
odometric mile chainage. 
4.5 A configuration of a GPS based Positioning system for 
road survey 
The experience gained during the field tests has been utilised 
for designing an optimal hardware and software configuration 
for a GPS-based positioning system, shown in Figure 13. The 
proposed configuration refers to the receivers and softwares 
adopted in the experiments (Trimble) and includes both Real- 
Time and Post-Processing positional capabilities.