IEEE TRANSACTIONS ON CIRCUITS AND SYSTEMS—I: REGULAR PAPERS, VOL. 53, NO. 4, APRIL 2006 769
Single Ping—Multiple Measurements:
Sonar Bearing Angle Estimation Using
Spatiotemporal Frequency Filters
Matthew A. Clapp, Member, IEEE, and Ralph Etienne-Cummings
Abstract—Presented is a mixed-signal full-custom VLSI chip
designed to receive sonar return signals from an ultrasonic mi-
crophone array, and extract input bearing angles of the incoming
signals. Processing utilizes simple low-power analog spatiotem-
poral bandpass filters to extract wavefront velocity across the
array, which translates to input bearing angle. Processing uses
phase information of array signals, not onset or offset of ultrasonic
burst. With such synchronous processing, multiple angle readings
from different returns of the same ultrasonic transmit burst are
possible. Compatible microphone arrays are compact in size—test
array has a total baseline of 26.5 mm. In a test with ultrasonic
beacon 65 cm from a microphone array, angular precision of 1
was demonstrated in most instances in the range 60 to 60 .
Applications include sonar localization of remote objects, sonar
imaging, and improved interference rejection between objects
within the field of view of the sensor microphones. The chip was
fabricated on a standard 3M2P CMOS process with a 0.5- m
feature size.
Index Terms—Acoustic arrays, acoustic beam steering, acoustic
imaging, acoustic signal analysis, analog processing circuits, array
signal processing, bandpass filters, continuous-time systems,
frequency-domain analysis, intelligent sensors, linear arrays,
mobile robot motion-planning, object detection, phased arrays,
robot sensing systems, sensor processing, signal processing, sonar
arrays, sonar imaging, sonar signal processing, spatial filters,
spatiotemporal filters, synchronous detection, time-frequency
analysis.
I. INTRODUCTION
FOR A VARIETY of applications, sonar is a popularchoice for remote object sensing and remote imaging.
Underwater sensing of ships at sea is possibly the oldest and
most well-known large-scale deployment of sonar [1]. In the
past few decades, sonar technology has increasingly come to
be used on smaller scales. Anyone who has seen a sonogram
of a developing fetus is familiar with sonar’s widespread use
in medical imaging [2]. Diagnostic ultrasonic waves are med-
ically benign to the patient, involving no radiation exposure
Manuscript received May 2004; revised March 2, 2005. This work was sup-
ported by the National Science Foundation under ERC cooperative agreement
EEC9731478, and by the DARPA under Award N00014-00-C-0315. This paper
was recommended by Associate Editor T. S. Lande.
M. A. Clapp was with the Department of Electrical and Computer Engi-
neering, The Johns Hopkins University, Baltimore, MD 21218 USA. He is now
with Agere Systems, San Jose, CA 95134 USA (e-mail: mclapp@gmail.com).
R. Etienne-Cummings is with the Department of Electrical and Computer En-
gineering, The Johns Hopkins University, Baltimore, MD 21218 USA (e-mail:
retienne@jhu.edu).
Digital Object Identifier 10.1109/TCSI.2005.859613
as X-rays do. In addition, ultrasonic imaging is both vastly
more portable and lower-cost than other medical imaging
technologies such as magnetic resonance imaging (MRI). Such
advantages are also of prime importance to designers of mobile
robots. For these applications, the low cost and low processing
requirements of a sonar ranging system are clear advantages
over other popular methods used to sense the environment.
One alternative, laser range-finding, is usually seen as more
accurate—but its common drawbacks include higher cost,
higher power consumption, and faster processing requirements.
A mobile robot built to navigate its environment with sonar
sensors is usually faced with a few well-known challenges.
The first is the attenuation of ultrasonic sound frequencies in
air, which is considerably more than the attenuation in water.
Greater attenuation causes the ultrasonic signal to fall off more
quickly, and reduces the strength of received reflections from
distant objects. The second issue facing designers of ultrasonic
systems is the wide angle of sensitivity of most ultrasonic mi-
crophones. In an ideal sense, it is convenient to think of sound
waves emanating out in a straight line and reflecting back on
the same straight line. In reality, however, most real-world
ultrasonic sources have beams which encompass a relatively
wide area. Reflections occur from almost every object and
reflect specularly in multiple directions, because all objects
appear as smooth surfaces to long-wavelength (up to 8.5 mm
in air) near-audible sound waves. Ultrasonic microphones
receptive over a large angle receive all of these reflections. A
naive processor thinking that a sonar return can only be from an
object directly in front of the sonar sensor will give erroneous,
puzzling results in this all-too-common scenario [3]. In the
mobile robot community, much effort has been devoted to
solving this problem. More and more complicated systems in-
volving multiple large sonar transducers, complicated heuristic
or statistical processing, and large processing units have been
described in the literature [4]–[7]. The trend in these systems
is to increase the amount of information available to the pro-
cessing units by using multiple sensors instead of individual
ones. The logical conclusion to such work is array processing,
a mature, decades-old field which has already been effectively
used to enhance the reception of both electromagnetic and
sound waves [8].
Medical and ship-based sonar systems have been using clas-
sical sonar phased arrays for some time now. Phased array sys-
tems use multiple receivers separated in space to receive a signal
from a single source. The angle of the incoming signal relative
to the array determines the phase of the signal at each point in
1057-7122/$20.00 © 2006 IEEE