' tunnels, 360? 
60? profiles is 
aser beam over 
» fulfill the de- 
ination with a 
> distance bet- 
m) the rotation 
ystem parame- 
anner are com- 
ion. The final 
he tunnel tube 
nspection, sur- 
ofile 
  
its designed to 
inder different 
ted theoretical 
rence between 
he variation of 
ed. 
est scenes by 
reflectances R 
aboratory. The 
through win- 
amps mounted 
; spotlights be- 
al. 
urements is to 
and true range. 
For a target point, let d be the range measurement reported by 
the scanner, and let r be the true range from the geometric ori- 
gin of the scanner, measured with a high precision reference 
system. Under ideal conditions, we expect to observe that d = 
r, but nonlinear effects with range measurements, offset errors 
in the analog unit, surface material of the target, and changing 
lighting conditions cause errors with range measurement. 
5 T 7 
  
range error (true range - measured range) [mm] 
  
  
  
0 i 1 1 
2000 4000 6000 8000 10000 
true range [mm] 
Fig. 4: Sensor accuracy for different materials 
To determine range accuracy we acquired range measurements 
of targets with various surface materials and reflectance within 
the ambiguity interval of the scanner. The target is positioned 
at known distances with an angle of incidence a. of about 90°, 
i.e. normal to the target surface. For each target position (0 ... 
15 m; 0.5 m steps) with all standardized reflectances R, a sam- 
ple of j = 1000 consecutive range measurements is acquired. 
Experimental results for linearity, and worst-case deviations e 
= max (ej e, = d. - r) between measured range and reference 
range r are illustrated in Fig. 4. After calibrating the look-up- 
tables diojgo 7 f(d, 0/80) of both measurement channels, high 
accuracy with linearity (max. error 0.1 % of max. range) and 
resulting range errors of less than +10 mm for worst-case re- 
flectance of R = 2 % are achieved. 
Range drift 
To measure range drift, a single target is placed (d = 6m; o = 
90°; R = 50%) in front of the laser scanner (Ÿ = 18?C) and 
range data is acquired over several hours. Due to thermal hea- 
ting effects in the measuring system (laser head and high fre- 
quency unit) a drift of 4 cm in range measurement occured in 
the first 10 min. After this "heating time", no significant range 
drift (Ad € e) can be detected for hours. Compared [20] to 
other laser range sensors, the laser range measuring system 
shows good long-time stability and high accuracy with range 
measurements. 
4.2 Precision 
For characterizing the precision of the range measurements the 
variation of repeated range measurements is identified. Am- 
bient lighting conditions, surface material of the target, distan- 
ce from the scanner to the target, beam incidence angle at the 
target, and ambient temperature are varied in this paper. Under 
each of these conditions 1000 range measurements are sam- 
pled at each target and each position. Mean and variance of the 
depth measurements are computed, and final precision is 
quantified as the standard deviation of the distribution of the 
measurements. 
Effect of ambient light 
To study the effect of ambient light on sensed range, a black 
target (R = 5%) is placed at a fixed distance, samples are 
taken, and mean and variance are computed. This procedure is 
repeated for target positions between 2 m and 10 m (2 m inter- 
val) in front of the sensor under different indoor lighting con- 
ditions: during night, during a sunny day, and during a cloudy 
day, all with and without spotlights. 
Evaluation of experiments shows that precision is nearly inde- 
pendent from intensity of ambient illumination. Only when 
sunlight or a spotlight directly radiates into the scanner, noise 
with range measurement increases, therefore decreasing pre- 
cision with measurements. 
Effect of surface material 
Surface material and therefore the reflectance of a target sur- 
face has a definite effect on the range measurements. To in- 
vestigate the effect of object surface material, targets are po- 
sitioned at fixed distances and reflectance is varied. Fig. 5 
shows the mean and standard deviation of range measurements 
for all standardized reflectances R within the range from 2 to 
10 m (0.5 m interval). 
  
    
  
   
10000 : 
9000 |--- | | | 
8000t-+----- eL ud -; a 
7000 CU bur usu in 2 
6000 +--------- 
5000 D rum ete ur mpeg 
4000 ----------—— ee 
3000 
mean of range measurements [mm] 
  
  
  
1000 i i 
2000 4000 6000 8000 10000 
distance [mm] 
5a) Mean of range measurements 
eo 
O1 
  
co 
e 
T 
=a N on 
N 
O1 
N 
[e] 
ol 
ce 
al 
  
  
  
standard deviation 3c of range measurements [mm] 
0 1 i | 
2000 4000 6000 8000 10000 
distance [mm] 
5b) Standard deviation 36 of range 
475