IEICE TRANS. FUNDAMENTALS, VOL.E91A, NO.

4 APRIL 2008
1031
PAPER
Special Section on Selected Papers from the 20th Workshop on Circuits and Systems in Karuizawa
Cross-Correlation by Single-bit Signal Processing for Ultrasonic
Distance Measurement
Shinnosuke HIRATA
a)
, Student Member, Minoru Kuribayashi KUROSAWA

, Member,
and Takashi KATAGIRI

, Nonmember
SUMMARY Ultrasonic distance measurement using the pulse-echo
method is based on the determination of the time of ight of ultrasonic
waves. The pulse-compression technique, in which the cross-correlation
function of a detected ultrasonic wave and a transmitted ultrasonic wave is
obtained, is the conventional method used for improving the resolution of
distance measurement. However, the calculation of a cross-correlation op-
eration requires high-cost digital signal processing. This paper presents a
new method of sensor signal processing within the pulse-compression tech-
nique using a delta-sigma modulated single-bit digital signal. The proposed
sensor signal processing method consists of a cross-correlation operation
employing single-bit signal processing and a smoothing operation involv-
ing a moving average lter. The proposed method reduces the calculation
cost of the digital signal processing of the pulse-compression technique.
key words: ultrasonic distance measurement, pulse-compression tech-
nique, delta-sigma modulation, cross-correlation, moving average lter
1. Introduction
Ultrasonic sensors are widely used in industrial applications
to measure the distance of objects in air. The advantages
of ultrasonic sensors include their low cost, small size, and
simple hardware. Autonomous mobile robots require many
kinds and relatively large numbers of sensors to measure
distances to obstacles and to recognize their own positions,
and the computational ability for each sensor is limited.
Therefore, ultrasonic sensors have been used for environ-
ment recognition in autonomous mobile robots [1][3].
The measurement of object distance using ultrasonic
sensors is based on the pulse-echo method which determines
the time of ight (TOF) of an echo. The distance from the
transmitter and the receiver to the object is calculated from
the TOF and the acoustic velocity. An ultrasonic wave is
generated, transmitted by the transmitter to the object, and
an echo from the object is then detected by the receiver. Fre-
quently, the TOF is computed using the threshold method.
In the threshold method, the echo is evaluated against a
given amplitude level [1]. This method allows for a low-
cost system, but suers from poor resolution of the distance
measurement in a noisy environment.
For eective improvement of the resolution of dis-
Manuscript received July 2, 2007.
Manuscript revised October 12, 2007.

The authors are with the Department of Information Pro-

cessing, Tokyo Institute of Technology, Yokohama-shi, 226-8502
Japan.

The author is with the Sutekina Inc., Komagane-shi, 399-4117

Japan.
a) E-mail: hirata.s.ab@m.titech.ac.jp
DOI: 10.1093/ietfec/e91a.4.1031
tance measurement in such an environment, the pulse-
compression technique has been employed. Pulse compres-
sion is used in the elds of radar systems, communication
engineering, and so on. With this technique, the received
signal is correlated with a reference signal, which is the
transmitted signal. The resolution of the TOF computed by
the cross-correlation function is improved because the de-
tection of the echo doesnt rely on the signal magnitude but
on its shape [4]. The transmitted signal must be a frequency-
modulated signal or a signal modulated by pseudo-random
sequences that have high autocorrelation [2], [3].
The typical signal processing method for the pulse-
compression technique is a cross correlation of the received
signal and the reference signal. Cross-correlation opera-
tions, consisting of huge iterations of multiplications and
accumulations, require high-cost digital signal processing.
Therefore, real-time distance measurement by the pulse-
compression technique is dicult given the limited compu-
tational ability of autonomous mobile robots.
In this paper, a sensor signal processing method using
a delta-sigma modulated single-bit digital signal is proposed
[5], [6]. The proposed cross-correlation operation by single-
bit signal processing can reduce the calculation cost of dig-
ital signal processing. The cross-correlation function ob-
tained by the proposed method was examined by computer
simulation.
2. Measurement System
A system for ultrasonic distance measurement using the
pulse-echo method is illustrated in Fig. 1. In the measure-
ment system, a transmitted signal is generated. This signal
Fig. 1 Design of a measurement system for ultrasonic distance
measurement using the pulse-echo method.
Copyright c 2008 The Institute of Electronics, Information and Communication Engineers
1032
IEICE TRANS. FUNDAMENTALS, VOL.E91A, NO.4 APRIL 2008
is transmitted as an ultrasonic wave by transmitter T to an
object, and the ultrasonic wave that includes an echo from
the object is then detected by receiver R. The transmitted
signal and the detected signal of receiver R are digitized by
A/D converters as a reference signal and a received signal,
respectively, and the TOF of the echo is computed by sen-
sor signal processing. Distance d from transmitter T and
receiver R to the object is then calculated based on the TOF
and the acoustic velocity in air.
3. Cross-Correlation Operation
In systems for ultrasonic distance measurement using the
pulse-compression technique, a chirp signal is generated as
the transmitted signal which is transmitted in the form of
an ultrasonic wave by transmitter T to an object, while the
ultrasonic wave that includes an echo from the object is de-
tected by receiver R. The transmitted signal and detected
signal are digitized by A/D converters as a reference sig-
nal and a received signal. The cross-correlation function of
the reference signal and the received signal is computed by
sensor signal processing. The time at the peak of the cross-
correlation function is the TOF of the echo.
3.1 Sensor Signal Processing
The design of a cross-correlation-operation systemusing the
conventional sensor signal processing method is illustrated
in Fig. 2. In the conventional method, the transmitted signal
is converted by the A/D converter into a multi-bit signal h (i)
as the reference signal, and the detected signal that includes
the echo is converted into a multi-bit signal x (t) by the A/D
converter. The received signal x (t i) is correlated with
the reference signal h (N i). The cross-correlation func-
tion C (t) of the reference signal and the received signal is
expressed as
C (t) =
1
N
N1

i=0
h (N i) x (t i) . (1)
The calculation of the cross-correlation operation of Eq. (1)
Fig. 2 Design of a cross-correlation operation using the conventional
sensor signal processing method.
requires huge numbers N of multiplications and accumu-
lations, and is repeated with every new sample. N is the
number of samples in the duration of the transmitted signal,
approximately several tens of thousands at a several MHz
sampling rate. Therefore, the conventional signal processing
method for cross-correlation operation entails a high calcu-
lation cost.
The design of a cross-correlation-operation system us-
ing the proposed sensor signal processing method employ-
ing single-bit signal processing is illustrated in Fig. 3. In
the proposed method, the transmitted signal and the de-
tected signal are digitized not by A/D converters but by a
delta-sigma modulator and a comparator. Here, the trans-
mitted signal is converted into a single-bit signal h
1
(i) by
the comparator as the reference signal, and the received sig-
nal is converted into a single-bit signal x
1
(t) by the delta-
sigma modulator. The received signal x
1
(t i) is corre-
lated with the reference signal h
1
(N i). Single-bit sig-
nals are discrete signals which have only two states, 1 and
1; hence, multiplications of single-bit signals are replaced
with logical-operations EXOR, NOT. Therefore, the cross-
correlation function C
1
(t) is expressed as
C
1
(t) =

4
1
N
N1

i=0
h
1
(N i) x
1
(t i) . (2)
Multiplications and accumulations by single-bit signal pro-
cessing are replaced with logical-operations EXOR, NOT,
and their summation. The calculation of the cross-
correlation operation of Eq. (2) entails huge numbers N of
the above operations and their summation, and is repeated
with every new sample. Because the logical operations
of single-bit signals are far simpler than the calculation of
multi-bit signals, the proposed method employing single-
bit signal processing reduces the calculation cost of cross-
correlation compared with the conventional method.
3.2 Simulation Result
The cross-correlation function C
1
(t) was evaluated by a
computer simulation using MATLAB/Simulink. In the sim-
ulation, the transmitted signal was a linear chirp signal from
Fig. 3 Design of cross-correlation operation using the proposed sensor
signal processing method of single-bit processing.
HIRATA et al.: CROSS-CORRELATION BY SINGLE-BIT SIGNAL PROCESSING FOR ULTRASONIC DISTANCE MEASUREMENT
1033
Fig. 4 The cross-correlation function C
1
(t) obtained by logical-
operations EXOR, NOT and their summation, as compared to the cross-
correlation function obtained by multiplications and accumulations using
real numbers.
55 kHz down to 25 kHz, and the duration of the signal was
5 ms. The sampling frequency of the delta-sigma modulator
was 2 MHz; hence, the number of samples N in the duration
of the transmitted signal was 10000. The distance d from
the transmitter and the receiver to the object was 2 m, and
the acoustic velocity was 343.6 m/s in air at 20 degrees C;
therefore, the TOF was 5.82 ms. The attenuation of the ul-
trasonic wave in air is 2.11 dB/m [7]. The delta-sigma mod-
ulator was a 2nd-order delta-sigma modulator.
Figure 4 shows the cross-correlation function C
1
(t) ob-
tained by logical-operations EXOR, NOT and their summa-
tion, and a close-up of the region near the peak of the cross-
correlation function C
1
(t). The abscissa of Fig. 4 is the time,
and the ordinate of Fig. 4 is a cross-correlation function
C
1
(t) normalized by an autocorrelation function. The time
at the peak of the cross-correlation function C
1
(t) shows ap-
proximately 5.8 ms, but the cross-correlation function C
1
(t)
contains high-frequency noise as compared with a cross-
correlation function obtained by multiplications and accu-
mulations by real numbers. The high-frequency noise de-
creases the accuracy of the cross-correlation function C
1
(t).
4. Smoothing Operation
To improve the accuracy of the cross-correlation function
C
1
(t), a smoothing operation is required to cancel the high-
frequency noise. The high-frequency noise of the cross-
correlation function C
1
(t) is the quantized error of the
received signal x
1
(t), because the delta-sigma modulator
shifts the quantized error into a high-frequency band. The
quantized error of the delta-sigma modulated single-bit sig-
nals could be reduced by a low-pass lter. The methods
of multiplication to cancel the quantized error by a moving
average lter as the low-pass lter have been proposed in
Refs. [8][10].
4.1 Sensor Signal Processing
A low-pass lter was used to reduce the high-frequency
noise and improve the accuracy of the cross-correlation
function. The cross-correlation function C
S
(t) of the ref-
erence signal and the received signal smoothed by a moving
average lter as the low-pass lter is expressed as
C
S
(t) =

4
1
N
N1

i=1
h
1
(N i)
_

_
1
M
M1

k=0
x
1
(t i k)
_

_
=

4
1
N
N1

i=0
h
1
(N i) x
M
(t i) (3)
M is the length of the moving average lter. The received
signal x
1
(t), which is a single-bit signal, is converted into
a multi-bit signal, x
M
(t), by the moving average lter. The
cross-correlation operation of Eq. (3) consists of multipli-
cations and accumulations of the multi-bit signal x
M
(t i)
and the single-bit signal h
1
(N i). Smoothing the received
signal x
1
(t) by the moving average lter increases the calcu-
lation cost of the combined cross-correlation and smoothing
operations.
The cross-correlation function C
S
(t) is also expressed
as
C
S
(t) =

4
1
N
N1

i=0
h
1
(N i)
_

_
1
M
M1

k=0
x
1
(t i k)
_

_
=
1
M
M1

k=0
C
1
(t k) . (4)
Equation (4) is a transformation of Eq. (3). The calculation
of Eq. (4) achieves the smoothing of the cross-correlation
function C
1
(t) by the moving average lter. Therefore, the
cross-correlation operation of Eq. (4) consists of the logical-
operations EXOR, NOT, and their summation of single-bit
signals, while the smoothing operation of Eq. (4) consists of
the moving average of the cross-correlation functions. The
smoothing method of Eq. (4) reduces the calculation cost of
the combined cross-correlation and smoothing operations.
The design of the sensor signal processing method for the
1034
IEICE TRANS. FUNDAMENTALS, VOL.E91A, NO.4 APRIL 2008
Fig. 5 Design of a sensor signal processing method involving cross-correlation operation using
single-bit signal processing and a moving-average-lter smoothing operation.
Fig. 6 The smoothed cross-correlation function C
S
(t) as compared to
the cross-correlation function obtained by multiplications and accumula-
tions using real numbers.
cross-correlation operation and the smoothing operation is
illustrated in Fig. 5.
4.2 Simulation Result
The cross-correlation function C
S
(t) was evaluated by a
computer simulation using MATLAB/Simulink. In the sim-
ulation, the transmitted signal was a linear chirp signal from
55 kHz down to 25 kHz, and the duration of the signal was
5 ms. The sampling frequency of the delta-sigma modulator
was 2 MHz; hence, the number of samples N in the duration
of the transmitted signal was 10000. The distance d from
the transmitter and the receiver to the object was 2 m, and the
acoustic velocity was 343.6 m/s in air at 20 degrees C; there-
fore, the TOF was 5.82 ms. The attenuation of the ultrasonic
wave in air was 2.11 dB/m [7]. The delta-sigma modulator
was a 2nd-order delta-sigma modulator. The length of the
moving average lter M for the smoothing operation was 25
taps.
Figure 6 shows the smoothed cross-correlation func-
tion C
S
(t) obtained by smoothing the cross-correlation
function C
1
(t) by the moving average lter and a close-up
of the region near the peak of the cross-correlation function
C
S
(t). The abscissa of Fig. 6 is the time, and the ordinate
of Fig. 6 is the cross-correlation function C
S
(t) normalized
by the autocorrelation function. The time delay of the mov-
ing average lter is calibrated by subtracting the constant
time delay of the lter. The time at the peak of the cross-
correlation function C
S
(t) shows approximately 5.82 ms,
and the high-frequency noise of the cross-correlation func-
tion C
S
(t) is reduced as compared with the cross-correlation
function obtained by multiplications and accumulations us-
ing real numbers. The smoothing operation improved the
accuracy of the cross-correlation function C
S
(t).
5. SNR of a Cross-Correlation Function
The accuracy of the cross-correlation function C
S
(t) is im-
HIRATA et al.: CROSS-CORRELATION BY SINGLE-BIT SIGNAL PROCESSING FOR ULTRASONIC DISTANCE MEASUREMENT
1035
Fig. 7 The SNR of smoothed cross-correlation function C
S
(t) as com-
pared to the cross-correlation function obtained by multiplications and ac-
cumulations using multi-bit digital signal.
proved by the moving average lter. The accuracy of this
function is expressed as the signal-to-noise ratio (SNR) of
the cross-correlation function C
S
(t). The SNR of the cross-
correlation function C
S
(t) was changed by the length of the
moving average lter M.
The SNR of the cross-correlation function C
S
(t)
was evaluated by a computer simulation using MATLAB/
Simulink. In the simulation, the transmitted signal was a
linear chirp signal from 55 kHz down to 25 kHz, and the
duration of the signal was 5 ms. The distance d from the
transmitter and the receiver to the object was 2 m, and the
acoustic velocity was 343.6 m/s in air at 20 degrees C; there-
fore, the TOF was 5.82 ms. Attenuation of the ultrasonic
wave in air was 2.11 dB/m [7]. The delta-sigma modulator
was a 2nd-order delta-sigma modulator. The sampling fre-
quency of the delta-sigma modulator ranged from 1 MHz to
16 MHz.
Figure 7 shows the SNR of the cross-correlation func-
tion C
S
(t) as compared to the cross-correlation function ob-
tained by multiplications and accumulations using 8 bit dig-
ital signals and 16 bit digital signals at a 200 kHz sampling
rate. The abscissa of Fig. 7 is the length of the moving
average lter M and the ordinate is the SNR of the cross-
correlation function C
S
(t). The signal amplitude of the
SNR is the root-mean-square of a cross-correlation func-
tion obtained by multiplications and accumulations by real
numbers and smoothed by the moving average lter. The
noise amplitude of the SNR is the root-mean-square error
between the cross-correlation function C
S
(t) and the cross-
correlation function obtained by multiplications and accu-
mulations by real numbers and smoothed by the moving av-
erage lter.
The zero frequency f
zero
of the moving average lter
was determined by the sampling frequency f
S
of the delta-
sigma modulator and the length of the moving average lter
M. The zero frequency f
zero
of the moving average lter is
expressed as
f
zero
=
f
S
M
{1 + 3 + 5 + } . (5)
The minimum zero frequency f
zero
of the moving average
lter must be higher than the central frequency, 40 kHz, of
the chirp signal, otherwise the moving average lter elim-
inates both the high-frequency noise and the peak of the
cross-correlation function C
S
(t).
The SNR of the cross-correlation function C
S
(t) at ev-
ery sampling frequency contains each peak and dip. The
length of the moving average lter M
p
at the peak of the
SNR of the cross-correlation function C
S
(t) at 2 MHz sam-
pling is 25 taps, and the minimum zero frequency is approx-
imately 80 kHz, while the length of the moving average l-
ter M
d
at the dip of the SNR of the cross-correlation func-
tion C
S
(t) at 2 MHz sampling is 51 taps, the minimum zero
frequency is approximately 40 kHz. Therefore, the optimal
length of the moving average lter M was determined by the
sampling frequency, 116 MHz, of the delta-sigma modu-
lator and twice the central frequency, 80 kHz, of the chirp
signal.
6. Reduction of Calculation Cost
In the simulation, the transmitted signal was a linear chirp
signal from 55 kHz down to 25 kHz, and the duration of the
signal was 5 ms. The transmitted signal was converted into
a single-bit signal h
1
(i) by the comparator as the reference
signal. The reference signal has 399 zero-cross points, and
the same values, 1 or 1, in several tens of samples, S
Z
samples, exist between two zero-cross points. The dier-
ence of C
1
(t) and C
1
(t 1) consists of subtraction and ad-
dition of single-bit samples. The subtracted samples are a
new sample x
1
(t), the oldest sample x
1
(t N), and double
of x
1
_
t
_
Z1
i=0
S
399i
_
, when Z is even. The added samples
are double of x
1
_
t
_
Z1
i=0
S
399i
_
, when Z is odd. Therefore,
the calculation, logical-operations EXOR, NOT, and their
summation, is replaced with the summation of the recursive
dierence of the cross-correlation functions, as illustrated in
Fig. 8.
The cross-correlation function C
1
(t) of the summation
of the recursive dierence of the cross-correlation functions
is expressed as
C
1
(t) =

4
1
N
N1

i=0
h
1
(N i) x
1
(t i)
= C
1
(t 1) +

4
1
N
[x
1
(t)
+2 {x
1
(t S
399
) x
1
(t S
399
S
398
) +
+ x
1
_

_
t
398

i=0
S
399i
_

_
_

_
x
1
(t N)
_

_
. (6)
Equation (6) is a transformation of Eq. (2). Therefore, the
cross-correlation function C
1
(t) obtained by Eq. (6) and the
cross-correlation function C
1
(t) obtained by Eq. (2) are nu-
merically equivalent.
1036
IEICE TRANS. FUNDAMENTALS, VOL.E91A, NO.4 APRIL 2008
Fig. 8 Design of cross-correlation operation by the summation of recursive dierence of the
cross-correlation function C
1
(t).
The calculation of the cross-correlation operation of
Eq. (6) requires 401 summations of single-bit samples. The
number of zero-cross points of the linear chirp signal, 399,
is constant and independent of the sampling frequency of
the delta-sigma modulator. Therefore, the cross-correlation
operation of Eq. (6) reduces the calculation cost of digital
signal processing compared with the cross-correlation oper-
ation of Eq. (2).
7. Conclusion
Sensor signal processing using a delta-sigma modulated
single-bit digital signal is proposed for real-time ultrasonic
distance measurement. The proposed sensor signal process-
ing method is composed of a cross-correlation operation by
single-bit signal processing and a smoothing operation per-
formed by a moving average lter. The calculation cost
of the cross-correlation operation was reduced by single-
bit signal processing, and the SNR of the cross-correlation
function was improved by the moving average lter.
In the simulation, the eects of reverberations in the
environment or echoes from other objects arent considered.
Therefore, experimental determination of the eect for sake
of measurement accuracy is necessary. Furthermore, in the
proposed method, the distance d from the moved transmitter
T to the moved receiver R couldnt be measured, because
the frequency of the linear chirp signal is doppler-shifted
when the transmitter T or the receiver R moves. Analyzing
doppler-shift frequency is necessary to measure the distance
d from the moved transmitter T and the moved receiver R to
the object.
References
[1] Y. Nagashima and S. Yuta, Ultrasonic sensing for a mobile robot
to recognize an environment Measuring the normal direction
of walls, Proc. IEEE/RSJ International Conference on Intelligent
Robots and Systems, vol.2, pp.805812, Raleigh, NC, U.S.A, July
1992.
[2] K.W. Jorg and M. Berg, Sophisticated mobile robot sonar sensing
with pseudo-random codes, Robotics and Autonomous Systems,
vol.25, no.3, pp.241251, Nov. 1998.
[3] J. Klahold, J. Rautenberg, and U. Ruckert, Continuous sonar sens-
ing for mobile mini-robots, Proc. 2002 IEEE International Confer-
ence on Robotics and Automation, vol.1, pp.323328, Washington,
DC, U.S.A., May 2002.
[4] D. Marioli, C. Narduzzi, C. Oelli, D. Petri, E. Sardini, and A.
Taroni, Digital time-of-ight measurement for ultrasonic sensors,
IEEE Trans. Instrum. Meas., vol.41, no.1, pp.9397, Feb. 1992.
[5] T. Fukui, M. Segawa, M. Kurosawa, K. Oka, and T. Higuchi, A
control system with single-bit digital signal processing, Proc. 4th
Int. Conf. on Control, Automation Robotics and Vision, pp.1992
1996, Singapore, Dec. 1996.
[6] S.R. Norsworthy, R. Schreier, and G.C. Temes, Delta-Sigma Data
Converters Theory, Design, and Simulation, IEEE Press, Piscat-
away, 1997.
[7] E.J. Evans and E.N. Bazley, The absorption of sound in air at audio
frequencies, Acoustica, vol.6, pp.238245, 1956.
[8] F. Maloberti, Non conventional signal processing by the use of
sigma delta technique: A tutorial introduction, Proc. IEEE Inter-
national Symposium on Circuits and Systems, vol.6, pp.26452648,
San Diego, CA, U.S.A., May 1992.
[9] H. Fujisaka, M. Sakamoto, and Morisue, Bit-stream signal pro-
cessing circuits and their application, IEICE Trans. Fundamentals,
vol.E85-A, no.4, pp.853860, April 2002.
[10] Y. Orino, M.K. Kurosawa, and T. Katagiri, Single-bit digital multi-
plier based on delta-sigma modulation, 16th Workshop on Circuits
and Systems in Karuizawa, pp.117122, Karuizawa, Japan, April
2003.
HIRATA et al.: CROSS-CORRELATION BY SINGLE-BIT SIGNAL PROCESSING FOR ULTRASONIC DISTANCE MEASUREMENT
1037
Shinnosuke Hirata was born in Aomori,
Japan, on October 14, 1982. He received the
B.Eng. degree in electrical and electronic engi-
neering, and the M.Eng. degree in information
processing from Tokyo Institute of Technology,
Tokyo, in 2005 and 2007, respectively. He is
a doctoral student in the Interdisciplinary grad-
uate school of science and engineering, Tokyo
Institute of Technology. His current research in-
terests include ultrasonic sonar sensing, sensor
signal processing, and single-bit digital signal
processing and its application to control systems. He is a member of the
Acoustical Society of Japan, and IEEE.
Minoru Kuribayashi Kurosawa (for-
merly Kuribayashi) (M95) was born in Nagano,
Japan, on April 24, 1959. He received the
B.Eng. degree in electrical and electronic en-
gineering, and the M.Eng. and D.Eng. degrees
from Tokyo Institute of Technology, Tokyo, in
1982, 1984, and 1990, respectively. Beginning
in 1984, he was a research associate of the Pre-
cision and Intelligence Laboratory, Tokyo Insti-
tute of Technology, Yokohama, Japan. From
1992, he was an associate professor at the De-
partment of Precision Machinery Engineering, Graduate School of Engi-
neering, The University of Tokyo, Tokyo, Japan. From 1999, he has been
an associate professor at Department of Advanced Applied Electronics, In-
terdisciplinary Graduate School of Engineering, Tokyo Institute of Tech-
nology, Yokohama, Japan. Since 2003, he belongs to Department of Infor-
mation and Processing in the same Institute. His current research interests
include ultrasonic motor, micro actuator, PZT thin lm, SAW actuator, and
single bit digital signal processing and its application to control systems.
Dr. Kurosawa is a member of the Acoustical Society of Japan, IEEE, the
Institute of Electrical Engineers of Japan, and the Japan Society for Preci-
sion Engineering.
Takashi Katagiri was born in Nagano,
Japan, on March 21, 1960. He received the
B.Eng. degree in electronics engineering from
Nagoya Institute of Technology, Aichi, in 1982,
and the M.Eng. Degrees in Physical electron-
ics engineering from Tokyo Institute of Tech-
nology, Tokyo, in 1984, respectively. Begin-
ning in 1984, he had worked for Sankyo Seiki
Mfg.Co., Ltd. (Now NIDEC Sankyo Corp.) and
TAMAGAWA SEIKI Co., Ltd. In 1999, he
founded his own company, Prime Motion Inc.
His current research interests include motion controller and motion sensor.