Page 93 - Autonomous Mobile Robots

76 Autonomous Mobile Robots

(b)

Probability –30 30
1
0.5
0
20
–20
10
–10
Distance (m)
0 0
10 –10
Distance (m)
20 –20
30 –30

FIGURE 2.18 Continued.

the noise power by considering more than one range bin (Equation [2.16]). The
target presence probability-based feature extraction, unlike the CFAR detector,
is not a binary detection process as is shown in Figure 2.17c. This method
of feature detection is useful in data fusion as the feature representation is
probabilistic.

2.7 MULTIPLE LINE-OF-SIGHT TARGETS — RADAR PENETRATION
At 77 GHz, millimeter waves can penetrate certain nonmetallic objects, which
9
sometimes explains the multiple line-of-sight objects within a range bin. This
limited penetration property can be exploited in mobile robot navigation in
outdoor unstructured environments, and is explored further here.
For validating the target penetration capability of the RADAR, tests were
carried out with two different objects. In the section of the RADAR scan,
2
shown in Figure 2.24a, a RADAR reflector of RCS 177 m and a sheet of
9 Although it should be noted that these can be the results of specular and multiple path
reflections also.

© 2006 by Taylor & Francis Group, LLC

FRANKL: “dk6033_c002” — 2006/3/31 — 17:29 — page 76 — #36