Page 61 - Autonomous Mobile Robots

44 Autonomous Mobile Robots

features are added together with the RADAR losses. Finally, Section 2.9 shows
full predicted range spectra and the results are compared with the measured
range bins in the initial stages of a simple SLAM formulation.

2.2 RELATED WORK

In recent years RADAR, for automotive purposes, has attracted a great deal
of interest in shorter range (<200 m) applications. Most of the work in
short-range RADAR has focused on millimeter waves as this allows narrow
beam shaping, which is necessary for higher angular resolution [5]. Some
of the work to date in autonomous navigation using MMW RADAR is
summarized here.
Boehmke et al. [8] succeeded in producing three-dimensional (3D) terrain
◦
maps using a pulsed RADAR with a narrow beam of 1 and high sampling rate.
◦
The 1 RADAR beam width has a large antenna sweep volume and its physical
size is large for robotic applications. The efforts by Boehmke et al. show the
compromise between a narrow beam and antenna size, where a narrow beam
provides better angular resolution.
Steve Clark [9] presented a method for fusing RADAR readings from
different vehicle locations into a two-dimensional (2D) representation. The
method selects one range point per RADAR observation at a particular bearing
angle based on a certain received signal power threshold level. This method
takes only one range reading per bin which is the nearest power return to
exceed that threshold to the RADAR, discarding all others. Clark [10] shows
a MMW-RADAR-based navigation system which utilizes artificial beacons
for localization and an extended Kalman filter for fusing multiple observa-
tions. The fixed threshold can be used when the environment is known with no
1
clutter. However, in a realistic environment (containing features having various
RCS) fixed thresholding on raw data will cause an exorbitant number of false
alarms if the threshold is low or missed detections if the threshold is too high.
Manual assistance is required in adjusting the threshold as the returned signal
power depends on various objects’ RCS. This method of feature detection is
environment-dependent.
Foessel [11] shows the usefulness of evidence grids for integrating uncer-
tain and noisy sensor information. Foessel et al. [12] show the development of a
RADAR sensor model for certainty grids and also demonstrate the integration
of RADAR observations for building 3D outdoor maps. Certainty grids divide
the area of interest into cells, where each cell stores a probabilistic estimate
of its state [13,14]. The proposed 3D model by Foessel et al. has shortcom-
ings such as the necessity of rigorous probabilistic formulation and difficulties

1
Clutter in this research is assumed to be the backscatter from land and is difficult to model. Land
clutter is dependent on the type of terrain, its roughness, and dielectric properties.

© 2006 by Taylor & Francis Group, LLC

FRANKL: “dk6033_c002” — 2006/3/31 — 17:29 — page 44 — #4