Home

Search

Collections

Journals

About

Contact us

My IOPscience

A traveling wave piezoelectric beam robot

This content has been downloaded from IOPscience. Please scroll down to see the full text.
2014 Smart Mater. Struct. 23 025013
(http://iopscience.iop.org/0964-1726/23/2/025013)
View the table of contents for this issue, or go to the journal homepage for more

Download details:
IP Address: 132.210.90.71
This content was downloaded on 08/01/2014 at 14:36

Please note that terms and conditions apply.

Smart Materials and Structures

Smart Mater. Struct. 23 (2014) 025013 (8pp)

doi:10.1088/0964-1726/23/2/025013

A traveling wave piezoelectric beam robot

H Hariri, Y Bernard and A Razek
Laboratoire de Genie Electrique de Paris (LGEP)/SPEE-Labs, CNRS UMR 8507, SUPELEC, Universite
Pierre et Marie Curie P6, Universite Paris-Sud 11, 11 rue Joliot Curie, Plateau de Moulon F91192 Gif sur
Yvette CEDEX, France
E-mail: haririhas@gmail.com
Received 3 September 2013, revised 10 November 2013
Accepted for publication 28 November 2013
Published 23 December 2013
Abstract

In this paper, the operation principles of a traveling wave piezoelectric beam robot are
presented. A prototype consisting of an aluminum beam structure, with two non-collocated
piezoelectric patches bonded on its surface, was fabricated and tested to demonstrate the
generation of a traveling wave on the beam based on the one mode excitation and the two
mode excitation operation principles for propulsion. A numerical model was developed and
used to study and optimize the generated motion of the piezoelectric beam robot.
Experimental characterization of the robot for the two types of operation has been carried out,
a comparison between them is made and results are given in this paper.
Keywords: piezoelectric beam robot, one mode excitation, two mode excitation, modeling,
experimental characterization
(Some figures may appear in colour only in the online journal)

1. Introduction

Hernandez [6] used two Langevin piezoelectric transducers to create a traveling wave on a finite beam; the
traveling wave is created by actuating the two piezoelectric
transducers (vibratorvibrator: two mode excitation) and also
by actuating one transducer while the other is used as an
absorber (vibratorabsorber: one mode excitation). He used
the traveling wave to realize a linear pump system. Kim
and Park [7] used two piezoelectric Langevin transducers as
vibrators to create a traveling wave on a finite beam. They
studied theoretically the change in traveling wave direction
according to frequency and verified it experimentally. Loh and
Ro [4] demonstrated experimentally the possibility to generate
a traveling wave on a finite length using two piezoelectric
Langevin transducers as vibrators (two mode excitation).
Some experimental tests are performed in their works to
characterize this linear traveling wave ultrasonic motor.
Another type of traveling wave linear ultrasonic motors
using piezoelectric patches bonded on an elastic structure as
actuators instead of Langevin actuators is also presented in
the literature. These types of motors use many piezoelectric
patches bonded on one or both sides of the elastic structure
and also teeth on the structure to generate the traveling wave.
As examples of this type of motors we cite [8, 9].
Dual piezoelectric actuators for the traveling wave
ultrasonic motor are presented in [10]; this motor uses dual

Our idea is to generate a traveling wave in a beam structure

to move it on a solid substrate using piezoelectric patches
bonded on the beam surface. This idea is inspired by the
linear traveling wave ultrasonic motor [1, 2] and it is applied
to robotic systems to move all of the system instead of
moving the slider as in the case of linear traveling wave
ultrasonic motors. Several configurations were reported in
the literature to excite traveling waves in finite structures.
Among them we can cite the one mode excitation presented
first by Kuribayashi et al [3] and the two mode excitation
presented first by Loh et al [4]. Both methods are presented
for ultrasonic linear motors. Other methods used to generate
traveling waves in finite beam structures, like the feedback
control method, active control method and adaptive control
method are also present in the literature but they were not in
our field of interest in this paper. Interested readers can refer
to [5], which presents an important review of the excitation of
traveling waves in finite beam structures.
The following are some key examples of traveling waves
generated on finite beam structures using two Langevin
piezoelectric transducers; the principle of these following
works is inspired by linear traveling wave ultrasonic motors
theory.
0964-1726/14/025013+08$33.00

c 2014 IOP Publishing Ltd Printed in the UK

Smart Mater. Struct. 23 (2014) 025013

H Hariri et al

Figure 2. Schematic figure of the one mode excitation (1ME).

when it hits the boundaries, which creates a mixture of

standing and traveling waves; as a standing wave is generated
by superposition of two traveling waves with the same
amplitude and frequency but moving in opposite directions.
An additional piezoelectric patch can be used to avoid wave
reflection. Our system consists of a beam structure, with two
non-collocated piezoelectric patches attached to its surface.
One patch produces the mechanical displacement of the beam
by applying an electrical voltage, while the other converts this
mechanical displacement into electrical energy which is then
dissipated through a passive RL electrical network to avoid
wave reflection. The device studied is presented in figure 2.
It can be noted that, by switching the role of both
piezoelectric patches, the motion direction is reversed.
Series RL connection was chosen in this work as shown
in figure 2, but this is not the only way to absorb waves.
It was chosen due to the simplicity of its modeling and
implementation. In order to obtain a great traveling wave rate,
i.e. to avoid the maximum possible wave reflection, R and
L must be calculated to obtain the maximum ratio of power
dissipated over the power provided. The power dissipated
is the power consumed by the electrical shunt circuit and
the power provided is the power consumed by the patch
actuator. The ratio of the power dissipated over the power
provided must be calculated because not only the power
dissipated depends on the charge (RL values) but also, the
power provided depends on the charge, the applied frequency
and the piezoelectric patch positions.
As stated above, many other passive methods exist in
the literature. For example, Hariri et al present in their
paper [13] a comparison between open circuit, R shunt
circuit, RL series connection and RL parallel connection for
damping vibrations of the same asymmetric system (two noncollocated piezoelectric patches bonded on a thin beam). Also
other passive techniques like SSDS (synchronized switch
damping on a short circuit) and SSDI (synchronized switch
damping on an inductor) are compared in [14]. Semi-passive
techniques are given in [15] for damping applications. The
advantage of the RL series or parallel technique compared to
other passive or semi-passive techniques is the simplicity of its
modeling and implementation but its major drawback is that
it is not very effective at low frequencies. In our case we are
not interested in working at low frequencies; in contrast, we
need a high frequency in order to get enough contact points
between the robot and the ground to move the robot. Also we
take into account that the transverse displacement of the beam
decreases with frequency.
Others types like the feedback control method, active
control method and adaptive control method are reviewed

Figure 1. Schematic figure of the traveling wave piezoelectric beam

robot. 3D view of the top and side view of the bottom.

piezoelectric patches bonded on the beam structure but it uses

teeth also to generate this traveling wave using the two mode
excitation method.
For robotic applications, Hariri et al [11] presented
an overview of the locomotion principles for piezoelectric
miniature robots. In this overview, it is noted that there is
no robot based on the generation of a traveling wave on the
beam to move all of the system. The only robot cited in this
work uses a standing wave with legs to generate motion [12].
This robot has the same structure as our robot. It consists
of a piezoelectric layer bonded on a metal layer. Our robot
consists of only two piezoelectric patches bonded on a beam
layer. The originality of our system compared to this robot
and to all linear ultrasonic motors presented is that we use
two piezoelectric patches without using teeth or legsin
robotic notationto generate the traveling wave as presented
in figure 1. What we mean by robotic application is that we are
interested in moving the entire system and not a slider (rotor)
on the elastic beam (stator). The motion is generated using
one mode excitation (1ME) or two mode excitation (2ME)
operation principles.
After introducing the operation principles (1ME and
2ME) of the traveling wave piezoelectric beam robot,
modeling of this robot will be presented in both cases (1ME
and 2ME). Then a preliminary design will be studied in order
to obtain a better traveling wave performance of the robot. A
prototype will be fabricated to verify experimentally the robot
motion. At the end, robot speed versus applied voltage and
versus embedded masses will be measured on a smooth glass
flat surface; robot speed for different mechanical loads will be
measured and results are given in this paper.
2. Operation principle case for one mode excitation
(1ME)

We repeat that our idea is to generate this traveling wave

in a beam structure to move it on a solid substrate. The
traveling wave is generated without using legs, with two
non-collocated piezoelectric patches bonded on the beam
surface. As we know, pure linear traveling waves are usually
observed on long structures and rarely on normal finite
structures. In finite structures like beams, the excited vibration
wave produced by one piezoelectric patch is partially reflected
2

Smart Mater. Struct. 23 (2014) 025013

H Hariri et al

beam using the one mode excitation (1ME) and the two
mode excitation (2ME) methods. Based on EulerBernoulli
assumptions for a beam structure, the linear constitutive
relations and by applying Hamilton principles, we obtain
the variational equation governing the mechanical and
piezoelectric part of the system. Then Hermite elements are
used to obtain the variational equation in matrix form, taking
into account the damping behavior of the real system.
Take first the case of one mode excitation. In this case one
patch (p2) is acting as an actuator where a sinusoidal voltage
is applied (V sin(wt)) while the other (p1) is connected to a
series RL connection. The equation is given as follows:

mm 0 0
U i
Mmm 0 0
U i
ze

0 2 0

0 0 0 tp Q
p1 +
tp tp Qp1
p2
p2
tp Q
0 00
tp Q
0
0 0

1
1
Keq
Kmvp1 Kvvp1
Kmvp2 Kvvp2
Ui

K 1 K
Kvvp1
0
+
vvp1 vmp1
tp Qp1
1
1
tp Qp2
Kvvp2
K
0
Kvvp2
vmp2

V
E3p2 (t) = sin(wt)
tp

Figure 3. Schematic figure of the two mode excitation (2ME).

in [5] for damping application also. It should be noted that

the active control method is somehow the two mode excitation
(2ME) method; it will be discussed in section 3.
3. Operation principle case for two mode excitation
(2ME)

In one mode excitation, one patch is used as an actuator to

produce vibration on the beam while the other is used as a
sensor to convert this mechanical vibration into heating in the
goal to produce a traveling wave on the beam. In two mode
excitation, the two patches are used as actuators to produce
the mechanical displacement of the beam in order to create
the traveling wave. For this aim, the two piezoelectric patches
produce the mechanical displacement of the beam by applying
simultaneously two neighboring natural mode shapes of the
beam at the same frequency but with a phase difference of
90 . The vibration can be approximated as the superposition
of these two modes; this is called a two mode excitation.
It becomes clear now that, the principle of two mode
excitation is based on the excitation of the two patches,
at a frequency between two resonance frequencies. This
principle is necessary to generate the traveling wave on the
beam. At the resonance frequency, two progressive waves
with the same amplitude propagating in opposite directions
cancel each other, resulting in a standing wave on the
beam, so the beam robot will stop moving. Below or above
the resonance frequency, one progressive wave is excited
more than the other. The resulting waves propagate in the
same direction as the waves with the greater amplitude
propagate [4]. This result was demonstrated theoretically
and verified also experimentally in [7] for linear traveling
wave ultrasonic motors. They demonstrated that the generated
traveling wave changes direction according to the excitation
frequency. Taking now the case where the two patches are
actuated by V sin(wt) and V sin(wt + ) respectively, when
is equal to 90 , the traveling wave reaches its maximal speed.
The motion direction can be reversed by changing the phase
difference from 90 to 90 without causing the motion speed
to decrease. This result is demonstrated experimentally in [4].
A schematic figure for the two mode excitation (2ME) at
equal to 90 is represented in figure 3.

1
1
Keq = Kmm Kmvp1 Kvvp1
Kvmp1 Kmvp2 Kvvp2
Kvmp2

(1)

p1
E3p1 tp = ze Q

(2)

ze = jwL + R.

Take now the case of two mode excitation. In this case the two
patches are acting as actuators where two sinusoidal voltages
are applied with the same amplitude and a phase difference
of 90 (V sin(wt) and V cos(wt)).The equation is given as
follows:

Mmm 0 0
U i
Cmm 0 0
U i

0 0 0 tp Q
p1 + 0 0 0 tp Q
p1 +

0 00
tp Qp2
0 00
tp Qp2

1
1
Kmm Kmvp1 Kvvp1 Kvmp1 Kmvp2 Kvvp2 Kvmp2

1
Kvmp1
Kvvp1

1
Kvvp2 K
vmp2

1
1
Kmvp1 Kvvp1
Kmvp2 Kvvp2

1
Kvvp1
0

1
0
Kvvp2

Ui

V
E3p1 (t) = cos(wt)

(3)
tp Qp1 =
tp

V
tp Qp2
E3p2 (t) = sin(wt)
tp
where Mmm , Cmm and Kmm are the structural mass matrix, the
structural damping matrix and the structural stiffness matrix
respectively, Kmvp1 and Kmvp2 are the piezoelectric stiffness
matrices for piezoelectric patches 1 and 2, Kvmp1 and Kvmp2
are the transpose of Kmvp1 and Kmvp2 , and Kvvp1 and Kvvp2

4. Modeling, design and fabrication of the

piezoelectric beam robot

A finite element model is developed in order to be used

to demonstrate the generation of a traveling wave on the
3

Smart Mater. Struct. 23 (2014) 025013

H Hariri et al

Figure 4. Transverse displacement versus piezo-thickness for

different beam materials.

Figure 5. First resonant frequency versus piezo-thickness for

are the dielectric stiffness for piezoelectric patches 1 and 2.

The mass and stiffness matrices included all the geometric
parameters and physical properties of the system. Ui is the
vector with nodal structural displacements, Qp1 and Qp2 are
the electrical charge for piezoelectric patches 1 and 2, E3p1
and E3p2 are the electric field for piezoelectric patches 1 and 2,
V is the applied voltage, tp is the thickness of the piezoelectric
patches, ze is the electrical impedance of the series RL circuit
and w is the applied frequency. Readers can refer to [13] for
more details.
Our goal is to demonstrate the generation of a traveling
wave on the beam; a prototype has been manufactured due to
this aim. The dimensions of the beam (180 mm 17 mm
0.5 mm) and the positions of the piezoelectric patches (xp1 =
24 mm and xp2 = 124 mm) were fixed; we still have to
know the dimensions of the PZT piezoelectric patches, and
the type of elastic material used for the beam. The width
of the PZT patches was chosen to be the same as the beam
width. We are concerned with the case of a thin beam where
only the x-direction is considered for modeling; therefore,
it is good to have the length of the patch more than its
width; practically, two times its width. Consequently, the
length of the PZT patches was selected to be between 30 and
35 mm. The thickness of the PZT patches and the type of
material used for the beam were chosen using the developed
model to obtain maximum transverse displacement (maximal
electro-mechanical conversion) and maximum frequency to
be nearest the ultrasonic range. The optimum thickness for
the two patches is determined for a constant electric field
applied to a patch, while the other is kept in open circuit.
This particular case does not lose the generality of the study
because the optimum thickness does not change if the voltage
is applied to the first patch, the second or both patches at
the same time. By varying the thickness of the patches and
calculating the transverse displacement at a given point for
different rigidities of the beam, we obtain the curves of
figure 4. The decrease in each curve of figure 4 shows that
when the thickness of the piezoelectric patches becomes large,

the bending stiffness of the system becomes more important

than the bending moment generated by the PZT patches [16].
Figure 5 represents the first resonant frequency depending on the thickness of the PZT patches for different rigidities
of the beam case of freefree boundary conditions (B.C).
As we are interested in the material that gives the best
compromise between maximum transverse displacement and
maximum frequency, aluminum was chosen at its optimal
thickness to be the better elastic material in our study. For
this material and based on figure 4, the optimal PZT patches
thickness is equal to 0.27 mm compared to 0.5 mm thickness
of the beam.
We are looking for the frequency that gives the better
traveling wave performance on the beam structure (better
waveform, higher transverse displacement) in both modes of
operation. Using the model of the piezoelectric beam robot
given in the equation (1) case of 1ME operation principle,
R and L optimal values were determined at each resonance
frequency. Simulations show that the better traveling wave
performance is given at the 17th resonance frequency (f17 =
11.8 kHz). At this frequency the inductance L was the
inductance tuning the shunt effect to the resonance frequency
and the resistance R was simulated at this frequency to get
the maximum ratio of the power dissipated over the power
provided after the inductor L is tuned as shown in figure 6.
Using the model given in the equation (3) case of
2ME operation principle, iterative simulations are performed
at each frequency between two successive resonance
frequencies. This frequency is taken in simulation as
the middle of the two successive resonance frequencies.
Simulations show that the middle of the 16th and 17th
resonance frequencies (f1617 = 11.1 kHz) is the optimal
operating frequency.
Results showing the traveling wave performance in the
1ME and the 2ME operation principles are given in table 1.
By comparing the figures in table 1 to the pure traveling
wave and pure standing wave, we get the traveling wave
performance (better traveling waveform, higher transverse

different beam materials.

Smart Mater. Struct. 23 (2014) 025013

H Hariri et al

Table 1. Traveling wave performance for both modes of operation.

Figure 7. Robot structure for the 1ME operating principle (1: power

amplifier, 2: robot body, 3: series RL synthetic inductor).

length (x) in mm and time (t) in ms of a flexural beam at

its transverse displacement. The figures on the right show
the same wave but in another representation; it represents
transverse displacement in m through the length of the beam
in mm at each instant. As we can see from the figures, at
the location of the contact surface between the beam and the
PZT patches, the vibration is less progressive than the place

Figure 6. Optimal R values at the 17th resonance frequency.

displacement) at each mode of operation. The figures on

the left represent a mixture wave between a traveling wave
and a standing wave and it is projected on a plane of
5

Smart Mater. Struct. 23 (2014) 025013

H Hariri et al

Table 2. 1ME transverse displacement (m) through the length of the beam (mm) at each instant at the 6th, 12th, 17th and 19th resonant
frequencies.

table 1 show a higher traveling wave ratio in the case of

1ME. The optimal operating frequency is obtained using
iterative simulations at each resonance frequency in the case
of 1ME. Some frequencies are chosen in table 2 to justify the
choice of the optimal operating frequency in this case. It is
clear from the table that the 17th resonant frequency has the
highest transverse displacement corresponding to the better
waveform. The same methodology was carried out in the case
of 2ME using iterative simulations at each frequency between
two successive resonance frequencies.

Figure 8. Robot structure for the 2ME operating principle (1: power

Experimental implementation of the one mode and two

mode excitation is shown in figures 7 and 8 respectively.
In figure 7, a power amplifier is used to amplify the signal
supplied by the signal generator to the first PZT patch, and
then a series RL circuit with a synthetic inductor [17] is
connected to the second PZT patch. In figure 8, two power
amplifiers with a phase shifter are used to amplify the signal
provided by the signal generator to the PZT patches.

amplifiers, 2: robot body).

near the middle of the beam. In 2ME, we excite the two

patches at a frequency between two resonances mode while in
1ME, one patch is excited at the resonance frequency and the
other is used to absorb displacement in order to avoid wave
reflection. That explains why the transverse displacements
in 2ME are more important than in the case of 1ME even
if in 1ME we excite at the resonance frequency as shown
in table 1. Wave propagation through the beam in 2ME is
less progressive than in the case of 1ME. The figures in

Table 3 shows a comparison between the simulation and

experiment results. The robot has the highest speed at these
values.
6

Smart Mater. Struct. 23 (2014) 025013

H Hariri et al

Table 3. Simulation and experimental comparison.

Piezoelectric beam robot

1ME
fopt (kHz)
Lopt (mH)
Ropt ()

2ME

By simulation

Experimentally By simulation

Experimentally

11.8
9
9.75

11.6
8.2
Btw 5 and 15

11.3

11.1

Figure 9. Robot speed versus applied voltage on a smooth glass flat

Figure 11. Robot speed versus embedded mass on a smooth glass

surface for the one mode and two mode excitation.

flat surface for the one mode and two mode excitation.

Figure 10. Embedded mass on the robot body.

5. Robot characterization
Figure 12. Robot speed measured on an inclined plane.

Maximum applied voltage was chosen equal to 30 V

according to the maximum peak to peak electric field beyond
which the piezoelectric material used is depolarized. All
measurements here are taken for the one mode and two mode
excitation at its optimal values obtained experimentally. Speed
as a function of voltage, speed as a function of embedded
masses and speed as a function of dragged load are important
in mobile robotics applications. We measured first the robot
speed on a smooth glass flat surface for different applied
voltages. Figure 9 shows that robot speed varies linearly
with the applied voltage. The robot speed increases with
voltage, which is logical because with increasing voltage, the
transverse displacement increases and this increase is reflected
by an increase of the speed of the robot.
It also shows that the speed of the robot in the case of
2ME is higher than the 1ME. This result is in agreement with

the simulation. At 30 V amplitude robot 1 reaches 80 mm s1

in the 1ME and 131 mm s1 in the 2ME according to figure 9.
Having shown the influence of the applied voltage on the
velocity of the robot, we will see the influence of embedded
masses on the velocity as shown in figure 10 at 30 V amplitude
for the two mode operating principle. Figure 11 shows that
the variation is linear between embedded masses and robot
speed. It also shows that the slope is greater in the case of
2ME. This means that when we increase the mass, the robot
speed decreases much more in the case of 2ME than in the
case of 1ME. This is because the traveling wave ratio in the
case of 1ME is higher than in the case of 2ME. To determine
the nominal operating point of the robot, we measured robot
7

Smart Mater. Struct. 23 (2014) 025013

H Hariri et al

traveling wave performance is studied in this paper at given

piezoelectric patch positions. It is necessary in future work
to study the influence of piezoelectric patch positions on the
optimal operating frequency of the robot i.e. on the transverse
displacement and traveling wave performance.
References
[1] Sashida T and Kenjo T 1993 An Introduction to Ultrasonic
Motors (Oxford: Clarendon)
[2] Ueha S and Tomikawa Y 1993 Ultrasonic Motors, Theory and
Applications (Oxford: Oxford Science)
[3] Kuribayashi M, Ueha S and Mori E 1985 Excitaion conditions
of a flexural traveling waves for a reversible ultrasonic
linear motor J. Acoust. Soc. Am. 77 14315
[4] Loh B G and Ro P I 2000 An object transport system using
flexural ultrasonic progressive waves generated by two
modes excitation IEEE Trans. Ultrason. Ferroelectr. Freq.
Control 47 9949
[5] Gabai R and Bucher I 2009 Excitation and sensing of multiple
vibrating traveling waves in one-dimensional structures
J. Sound Vib. 319 40625
[6] Hernandez C 2010 Realization of piezoelectric micropumps
Thesis LGEP, University Paris-sud, Paris, France
[7] Kim G H, Park J W and Jeong S H 2009 Analysis of dynamic
characteristics for vibration of flexural beam in ultrasonic
transport system J. Mech. Sci. Technol. 23 142834
[8] Bein T, Breitbach E J and Uchino K 1997 A linear ultrasonic
motor using the first longitudinal and the fourth bending
mode Smart Mater. Struct. 6 61927
[9] Roh Y, Lee S and Han W 2001 Design and fabrication of a
new traveling wave-type ultrasonic linear motor Sensors
Actuators A 94 20510
[10] Suybangdum P, Smithmaitrie P and Laoratanakul P 2009 Dual
piezoelectric actuators for the traveling wave ultrasonic
linear motor 4th Int. Conf. on Experimental Mechanics
(Nov.)
[11] Hariri H, Bernard Y and Razek A 2010 Locomotion principles
for piezoelectric miniature robots ACTUATOR 10: Proc.
ACTUATOR 10 (June) ed D E Bremen, pp 101520
[12] Son K J, Kartik V, Wickert J A and Sitti M 2006 An ultrasonic
standing-wave-actuated nano-positioning walking robot:
piezoelectricmetal composite beam modeling J. Vib.
Control 12 1293309
[13] Hariri H, Bernard Y and Razek A 2011 Finite element model
of a beam structure with piezoelectric, patches using RL
shunt circuits AC2011, 14th Int. Conf. on Active Systems for
Dynamics Markets pp 12431
[14] Corcolle R et al 2008 Modeling of a beam structure with
piezoelectric materials: introduction to SSD techniques
COMPEL-Int. J. Comput. Math. Electr. Electron. Eng.
29 20514
[15] Badel A 2006 Recuperation denergie et controle vibratoire
par e lements piezoelectriques suivant une approche non
lineaire PhD Thesis Universite de Savoie, France
[16] Wang Q M and Cross L E 1998 Performance analysis of
piezoelectric cantilever bending actuators Ferroelectrics
215 187213
[17] Moheimani S O R and Andrew F J 2006 Piezoelectric
Transducers for Vibration Control and Damping (Berlin:
Springer)

Figure 13. Speed versus dragged load for different voltages for the

one mode and two mode excitation.

speed versus a mechanical load at a given applied voltage. For

this aim, this mechanical load is considered to be the weight
of the robot on an inclined plane as shown figure 12.
We have measured the robot speed for different positions
of the inclined plane for different mechanical loads. Figure
13 shows robot speed versus mechanical load for different
applied voltages, where it becomes easy to determine the
nominal operating point of the robot. At the same voltage, the
maximum mass pulled is higher in the case of 1ME than in the
case of 2ME. In case of 2ME, this robot can provide 432 W
at its nominal operating point (7.2 mN, 60 mm s1 ) and it
can provide 360 W at its nominal operating point (9 mN,
40 mm s1 ) in the case of 1ME. Figure 13 shows that speed
decreases with the mechanical load since the increase of the
mechanical load means an increase in the inclination angle,
therefore increasing the force opposite to the motion direction.
6. Conclusion

We presented in this paper a one mode excitation (1ME)

and two mode excitation (2ME) traveling wave piezoelectric
beam robot. Modeling and experimental studies have been
conducted. In the case of the 1ME operation principle, this
robot has an optimal operating frequency equal to 11.6 kHz,
and needs a passive RL circuit of 10  and 8.2 mH. In the
case of the 2ME operation principle, this robot has an optimal
operating frequency equal to 11.3 kHz. At 30 V amplitude and
without embedded mass, this robot travels at 81 mm s1 in the
case of 1ME and at 131.5 mm s1 in the case of 2ME. At the
same voltage, this robot can provide 360 W at its nominal
operating point (9 mN, 40 mm s1 ) in the case of 1ME and
432 W (7.2 mN, 60 mm s1 ) in the case of 2ME. The better