Anda di halaman 1dari 6

2008 IEEE/RSJ International Conference on Intelligent Robots and Systems

Acropolis Convention Center


Nice, France, Sept, 22-26, 2008

Gait Transition by Tuning Muscle Tones using Pneumatic Actuators in


Quadruped Locomotion
Katsuyoshi Tsujita, Toshiya Kobayashi, Takashi Inoura and Tatsuya Masuda
Dept. of Electrical and Electronic Systems Engineering, Osaka Institute of Technology,
Ohmiya 5-16-1, Asahi-ku, Osaka 535-8585, Japan
E-mail : tsujita@ee.oit.ac.jp

Abstract The development of an oscillator controller for a


quadruped robot with antagonistic pairs of pneumatic actuators
is reported. Periodic motions of the legs switch between the
swinging and supporting stages based on the phase of the
oscillators. The oscillators receive touch sensor signals at the
end of the legs as feedback when the leg touches the ground
and compose a steady limit cycle of the total periodic dynamics
of quadruped locomotion. And also muscle tone is adaptively
controlled according to the dynamic state of the main body.
This system can generate gait transition from one to another by
changing locomotion speed and muscle tone. The effectiveness
and performance of the proposed controller were evaluated with
numerical simulations and experiments with the hardware.

I. INTRODUCTION
Locomotion is one of the basic functions of a mobile
robot and one of the strategies it uses to accomplish this
is using its legs, which allows it to move over rough terrain.
Therefore, a considerable amount of research has been done
on controlling the motion of legged locomotion robots. This
article discusses the control of motion in quadruped robots,
with an emphasis on dynamic gait transition control.
In the future, walking robots will be required that can
carry out various tasks over unstructured terrain. Walking
robots will be required to achieve real-time adaptability
to changing environments and maneuverability to generate
voluntary motion according to contexts within the changing
environments.
Biological research over the last few years has made
great contributions to overcoming such difficulties. During
spontaneous motion such as walking straight, many joints
and muscles are organized into a collective unit that is
controlled as though it had fewer degrees of freedom (DOFs)
but that still needs to retain the necessary flexibility for
changing environments[1]. During voluntary motion, on the
other hand, such as in turning walks, another organized collective unit emerges selectively according to the commanded
signal.
This knowledge has inspired robotics researchers and a
considerable amount of research has been done[5]-[12] on
biologically inspired control systems for walking robots that
enable them to adapt to variations in the environment based
on the Central Pattern Generator (CPG) principle[2]-[3].
However, not that much research has been done on voluntary and autonomous gait transition in quadruped robots.

978-1-4244-2058-2/08/$25.00 2008 IEEE.

To make voluntary motion such as tones control to change


motions, it is very important for robots to voluntary motions.
Leg motions in locomotion have two essential stages. One
is the swinging stage and the other is the supporting stage. In
the swinging stage, the actuator forces are relaxed; the joints
become less stiff and more passive. In the supporting stage,
stiffness of the joints increases due to forces generated by
the antagonistic pair of actuators. By controlling and tuning
the stiffness of the joints through the balanced adjustment
of the generated force of such a pair of actuators, the robot
is expected to become more adaptive to variations in the
environment and in the surface of the ground. The idea
of the architecture of tones control of the actuators are
inspired by the biological studies. According to them, the
tones are actively and adaptively controlled by the neuronal
system in the basal ganglia [4]. And this neuronal system
stabilizes the posture, and obtains the stable locomotion by
also controlling the oscillation of central pattern generator in
the spinal cord.
In this article, a new system is proposed for controlling
the gait transition of a quadruped robot with nonlinear oscillators. Development of an oscillator controller for quadruped
robots with antagonistic pairs of pneumatic actuators is
shown in this paper. It consists of a spontaneous locomotion
controller and a voluntary tones controller. The spontaneous
locomotion controller is designed as an oscillator network
that has mutual interactions among oscillators with feedback
signals from external sensors. Periodic motions of the legs
are switched between the swinging and supporting stages
based on the phase of the oscillators. Oscillators contain
network architecture, interact mutually with each other, and
receive touch sensor signals as feedback signals at the end
of the legs when the leg touches the ground. At the moment
the leg makes contact, the oscillator phase is reset, and the
swinging stage is forced to change to the supporting stage.
These dynamic interactions make possible mutual entrainments between oscillators and create a steady limit cycle of
the total periodic dynamics of quadruped locomotion.
The voluntary tones controller, on the other hand, controls
the posture of the main body according to locomotion speed
and body state. Monitoring the locomotion speed and the
amplitude of the rolling motion and pitching motion of the
main body, the stiffness of the trunk, especially the joint
stiffness of the spine are controlled to stabilize and reduce the

2453

Authorized licensed use limited to: Khajeh Nasir Toosi University of Technology. Downloaded on December 21, 2009 at 05:23 from IEEE Xplore. Restrictions apply.

Leg 3

Rear
body

Front
body

r0
Leg 2

Leg 4
Inertial space
Leg 1

Fig. 1.

Schematic model of a quadruped robot

divergence of the amplitude of rolling and pitching motion


of the main body to keep the posture. To the results, the
dynamic interactions of the periodic motions of the legs
through the motion of the main body changes according to
the dynamic state of the system, and appropriate gait pattern
emerges according to the locomotion speed and body motion.
The performance of the proposed control system was verified to be good through numerical simulations and hardware
experiments. The results of this study shows the effectiveness
of the proposed system.

Fig. 2.

Assignment of actuators
Locomotion command

LF

RF
Rhythm generator

Tones adjuster
LR

RR

II. MODEL
Consider the quadruped robot shown in Fig. 1, which has
four legs and a main body. Each leg consists of three links
that are connected to each other through a one degree of
freedom (DOF) rotational joint. The main body is composed
of two parts, front body and rear body. The front body and the
rear body are connected through a rotational joint. Each leg
is connected to the main body through a one DOF rotational
joint. Legs are enumerated from leg 1 to 4, as shown in Fig.
1. The joint of the main body at waist is numbered as joint
0, and the joints of each leg are numbered as joint 1, 2, and
3 from the main body toward the end of the leg. We define
(0)
(0)
ri and i (i = 1, 2, 3) as the components of position
vector and Euler angle from inertial space to the coordinate
system which is fixed on the main body, respectively. (B)
is defined as the joint angle of the rear body to front body
(i)
in yaw axis. We also define j as the joint angle of link j
of leg i.
The state variable is defined as follows;


(1)
q T = rk(0) k(0) (B) j(i)
(i = 1, , 4, j = 1, 2, 3, k = 1, 2, 3)
Equations of motion for state variable q are derived using
Lagrangian formulation as follows;
M q + H(q, q)
= G+T +

Robot body
Sensory feedback

Fig. 3.

Sensory feedback

Architecture of the control system

gravity term. T is the input torque at the actuator. is the


reaction force from the ground at the point where the tip
of the leg makes contact. We assume that there is no slip
between the ends of the legs and the ground.
III. CONTROL SYSTEM
Figure 3 shows the control scheme of the proposed system.
The controller has a nonlinear oscillator network with
individual oscillators assigned to joints with the oscillator
phase defined as (k) (k = 1, , 4).
z (k) = exp(j(k))

(2)

where M is the generalized mass matrix and the term M q


expresses the inertia. H(q, q)
is the nonlinear term which
includes Coriolis forces and centrifugal forces. G is the

Leg motion
generator

Tones controller

j : Imaginary unit

(3)

where z (k) is the state of oscillator of leg k. is a


function of duty ratio which determines the ratio of stance
phase during the total time period for walking cycle, i.e. this

2454

Authorized licensed use limited to: Khajeh Nasir Toosi University of Technology. Downloaded on December 21, 2009 at 05:23 from IEEE Xplore. Restrictions apply.

Oscillator phase
0

phi_A

IV. NUMERICAL SIMULATIONS


A. Simulation conditions

2*pi

Numerical simulations are implemented to verify the effectiveness of the proposed control system and to investigate
the relationship between the locomotion speed and stable gait
pattern. Table 1 shows the physical parameters of the robot.
They are corresponding to the hardware model.
Liftoff

Swing
Fig. 4.

Touchdown

Stance

Table 1. Physical parameters of robot


Length of front body
Width of front body
Length of rear body
Width of rear body
Length of thigh (between joints)
Length of shank (between joint and toe)
Total height (at nominal stance)
Mass of front body
Mass of rear body
Mass of thigh
Mass of shank
Total mass

Actuation sequence

parameter controls locomotion speed.


=

2(1 )
Tf

(4)

where, Tf is nominal time period for swinging stage, and


it is constant.
The antagonistic pairs of pneumatic actuators are driven
by timing signals as functions of the oscillator phases. The
contact sensor signals are feedback for the oscillator network. These dynamic interactions cause the entrainment and
generate a stable limit cycle for locomotion. The oscillator
network can be expressed in the following equations;
(k)

+ K((l) (k) lk ) + (Ak (k))(5)

(k)
Tmn

F ((k))

spine
Tmn
= Q(, ave|(0) |)

In the simulations, the control conditions are given as


follows;
(i) Fixed nominal locomotion speed
In this case, the nominal gait patterns and nominal duty
ratios and spine stiffness Ksp are fixed as follows;
Transverse walk
Trot

(6)

where, K is constant number and Ak is the nominal value


of oscillator ks phase at the moment of leg ks contact
with the ground(i.e. touch down moment in fig. 4). is
the impulsive function. lk is the nominal phase difference
(k)
between oscillators l and k. Tmn is the On/Off timing of the
air valve of actuator n (n =extensor or flexor) of joint m
of leg k and is given as a function of the oscillator phase.
In concrete the actuators are driven for the motion sequence
in the locomotion, as shown in figure 4. In the figure 4,
illustrated actuators are actively shrinking according to the
supplied air pressure which are controlled in eq. (6). The
motion primitives in figure 4 is following the idea of [14].
On the other hand, the muscle tone at the spine is voluntarily controlled in voluntary controller. When the robot has to
change its locomotion speed according to some situations
surrounding the robot, the muscle tone of the postural
actuators are actively controlled. In this study, the stiffness of
the trunk is controlled according to the locomotion speed by
controlling the antagonistic pairs of actuators at t he spine in
the roll and the pitch axes are actively controlled , as follows;
(7)

spine
, (m = roll, pitch, n = flexor, extensor)
where, Tmn
is the stiffness of the actuator. This tones controller changes
the stiffness of the joints and that at the spine according
to the dynamic state of the main body and the locomotion
speed.

0.174 [m]
0.120 [m]
0.168 [m]
0.120 [m]
0.150 [m]
0.185 [m]
0.320 [m]
2.30 [kg]
2.40 [kg]
0.351 [kg]
0.218 [kg]
6.98 [kg]

= 0.75
= 0.50

Ksp = 10 [Nm/rad]
Ksp = 40 [Nm/rad]

In each gait pattern, nominal phase difference in eq. (5)


is given as;
Transverse walk

13 = 31 =
(8)
2

24 = 42 =
(9)
2
Trot
13
24

= 31 = 0
= 42 = 0

(10)
(11)

(ii) Variable nominal locomotion speed


In this study, the nominal duty ratio is the control parameter that changes as a function of time t as follows;
(t)

(t)

(t)

W ALK = 0.75 0 t < TS


0.25(t TS )
TS t < TE
0.75
TE TS
T ROT = 0.50 TE t

(12)
(13)
(14)

Spine stiffness is also controlled according to eq. (7). In


this study, to simplify the problem, the spine stiffness is
controlled as a function of only duty ratio as follows;
Ksp = 40 30

T ROT
W ALK T ROT

The nominal gait patterns are not specified.

2455

Authorized licensed use limited to: Khajeh Nasir Toosi University of Technology. Downloaded on December 21, 2009 at 05:23 from IEEE Xplore. Restrictions apply.

(15)

0.03

0.02

Pitch angle [rad]

Time period [sec]

0.9

0.8

0.7

0.01

-0.01

-0.02

0.6

-0.03
-0.06

0.5
0

10

20

30
40
Step number

50

60

-0.04

-0.02

0.02

0.04

0.06

0.02

0.04

0.06

Roll angle [rad]

70

(a) Stable walk

(a) Stable walk


0.03

1
0.02

Pitch angle [rad]

Time period [sec]

0.9

0.8

0.01

-0.01

0.7
-0.02

0.6
-0.03
-0.06

-0.04

-0.02

0
Roll angle [rad]

0.5
0

10

20

30
Step number

40

50

(b) Stable trot

60

Fig. 6.

(b) Unstable walk


Fig. 5.

Time history of time period for walking cycle

Roll and Pitch angles

RF

LR

B. Simulation results

RR

Figure 5 shows time history of time period for walking


cycle in fixed transverse walk pattern (case (i)). If the
stiffness is too much high, the pattern and locomotion itself
becomes unstable (figure (b)). However, if we choose the
appropriate stiffness at the spine, the robot can continue the
stable walk (figure (a)). Figure 6 shows the limit cycle in roll
and pitch motion (case (i)). We can find that the proposed
system can generate stable locomotion both in walk and
trot patterns. And the amplitudes of the rolling and pitching
motion in walk pattern are larger than in trot pattern. So
this amplitude of the robot can detect the state of the robot
and can be utilized to control the stiffness of the spine to
stabilize the locomotion. Figure 7 shows gait transition from
the transverse walk pattern to trot pattern by changing its
locomotion speed and appropriately controlled stiffness of
the spine according to the locomotion speed and amplitude
of the roll and pitch motion (case (ii)). From these results,
the proposed system is effective and can stably generate its
gait transition from one to another by controlling its speed
and stiffness of the spine.

LF
10

12

14

16

18

20

22

24

Time [sec]

Fig. 7.

Gait tansition (foot print)

C. Stability
The steady locomotion of the quadruped robot is periodic
and is characterized by a limit cycle in the state space.
The stability of the limit cycle is examined in the following
way: First, 8 variables are selected as state variables.
X R8

(16)

T

(0) (0) (B) (B) (0) (0) (B) (B)
(17)
X = 1 2 2 3 1 2 2 3
(0)

(0)

The variables 1 and 2 are roll and pitch angles of the


(B)
(B)
main body. And 2 and 3 are pitch and yaw angles of
the trunk.
Stability of the robots locomotion is examined by checking the eigen values k (k = 1, , 8) of Poincare map in

2456

Authorized licensed use limited to: Khajeh Nasir Toosi University of Technology. Downloaded on December 21, 2009 at 05:23 from IEEE Xplore. Restrictions apply.

B=0.50
B=0.55
B=0.60
B=0.65
B=0.70
B=0.75

Max. Norm of eigen value

1.5

0.5

0
10

15

20

25

30

35

40

Stiffness of the trunk [Nm/rad]

Fig. 8.

Fig. 10.

Maximum norm of Poincare maps eigen value

terms of variable X. If the maximum norm of eigenvalue


of the steady state is less than one, motion of the system is
considered stable. Since the system involves state transition
in this study, the stability of the system is checked in terms
of the state before and after the transition, not during the gait
transition.
Figure 8 shows maximum norm of Poincare maps eigen
value at various duty ratio and trunks joint stiffness Ksp .
We can note that when duty ratio is around 0.75 (typical
duty ratio for walk pattern), the locomotion is stable at low
joint stiffness of the trunk. To the contrary, when duty ratio
is around 0.50 (typical duty ratio for trot pattern), the
locomotion is stable at high joint stiffness of the trunk. The
midrange duty ratio and the joint stiffness are not suitable for
stable locomotion. This facts shows the gait transition occurs
from a stable attractor to another one by passing through the
unstable parameter region.
V. HARDWARE EXPERIMENTS
Figure 9 shows the quadruped robot with pneumatic actuators to drive the musculo-skeletal system. In the figure, we
can see the antagonistic pairs of actuators, especially, in the
trunk.

Front view

Side view
Fig. 9.

Robot hardware

Hardware system

host computer on real-time using proposed control method,


and sent to actuator drivers (iXs research corp., iMCs03)
via USB. The electric valves (SMC corp., SYJ3340-6MM5) controls air supply from air compressor (Jun-Air International, 3-4 Minor) to the actuators. The pressured air
is supplied at 0.6[MPa] through 8[mm] tube from the
compressor to multi-port valves, and is distributed through
4 [mm] tube from the valves to the actuators. Each actuator
(Hitachi Medico Co., Air muscle, 20[mm], L =150 [mm])
generates driving torque and stiffness at the joint.
Figure 11 are the results of hardware experiment. The
figures show the system can continue stable locomotion both
in walk and trot pattern. And they are in limit cycles.
Figure 12 shows the gait transition from transverse walk to
trot pattern by changing locomotion speed and the stiffness
appropriately according to the speed. We can find that the
gait pattern smoothly changed from walk to trot and they are
very stable patterns. From these results, the effectiveness of
the proposed system is verified.
VI. CONCLISIONS
We developed a quadruped robot with antagonistic pairs
of pneumatic actuators controlled by a nonlinear oscillator
network and muscle tone controller. Periodic motions of
the legs are switched between the swinging and supporting
stages based on the phase of the oscillators. The oscillators
receive touch sensor signals at the end of the legs as feedback
when the leg touches the ground and compose a steady
limit cycle of the total periodic dynamics of quadruped
locomotion. And tones control for spine actuators according
to the locomotion speed or body state makes autonomous
gait transition and they are smooth and stable. We can note
that the voluntary gait transition according to the locomotion
condition requires appropriate tones control to keep the
posture. The effectiveness and performance of the proposed
controller for the quadruped robot were evaluated with
numerical simulations and experiments with the hardware.
VII. ACKNOWLEDGMENTS

Figure 10 shows architecture of the hardware system.


CPU on the host computer Intel Xeon 3.0[GHz]. Sampling
frequency is 1.0 [kHz]. Signals from contact sensors are
input to 12bit A/D converters and are sent to the host
computer. Control commands for actuators are calculated on

This work has been partially supported by a Grant-inAid for Scientific Research on Priority Areas Emergence
of Adaptive Motor Function through Interaction between
Body, Brain and Environment from the Japanese Ministry
of Education, Culture, Sports and Technology.

2457

Authorized licensed use limited to: Khajeh Nasir Toosi University of Technology. Downloaded on December 21, 2009 at 05:23 from IEEE Xplore. Restrictions apply.

0.03

Pitch angle [rad]

0.02

0.01

-0.01

-0.02

-0.03
-0.06

-0.04

-0.02

0.02

0.04

0.06

0.02

0.04

0.06

Roll angle [rad]

(a) Stable walk


0.03

Pitch angle [rad]

0.02

0.01

-0.01

-0.02

-0.03
-0.06

-0.04

-0.02

0
Roll angle [rad]

(b) Stable trot


Fig. 11.

Roll and Pitch angles

RF

LR

RR

LF
10

12

14

16

18

20

22

24

Time [sec]

Fig. 12.

Gait tansition (foot print)

R EFERENCES
[1] N.A.Berstein, Co-ordination and regulation of movements, Oxford,
Pergamon press, New York, 1967
[2] S.Grillner, Locomotion in vertebrates:Central mechanisms and reflex
interaction, Physiological Review, No.55, pp.367-371, 1977
[3] S.Grillner, Neurobiological bases of rhythmic motor acts in vertebrates, Science, Vol.228, pp.143-149, 1985
[4] K.Takakusaki, T.Habaguchi, J.Ohtinata-Sugimoto, K.Saitoh and
T.Sakamoto, Basal ganglia efferents to the brainstem centers controlling postural muscle tone and locomotion: A new concept for understanding moto disorders in basal ganglia dysfunction, Neuroscience,
119, pp.293-308, 2003
[5] G.Taga, A model of the neuro-musculo-skeletal system for human
locomotion: I. Emergence of basic gait, Biological Cybernetics,
No.73, pp.97-111, 1995
[6] Y.Fukuoka, H.Kimura and A.Cohen, Adaptive Dynamic Walking of a
Quadruped Robot on Irregular Terrain Based on Biological Concepts,
The International Journal of Robotics Research 22, No. 3, pp.187-202,
2003

[7] A.J.Ijspeert, A connectionist central pattern generator for the atuatic


and terrestrial gaits of a simulated salamander, Biological Cybernetics, Vol.84, pp.331-348, 2001
[8] K. Akimoto, S. Watanabe and M. Yano, A insect robot controlled
by emergence of gait patterns, Proc. of International Symposium on
Artificial Life and Robotics, Vol. 3, No. 2, pp. 102-105, 1999
[9] R.J.Full, Biological inspiration: Lessons from many-legged locomotors, Robotics Research 9, J.M.Hollerbach and D.E.Koditscheck,
Eds.Springer London, pp.337-341, 2000
[10] S.Aoi and K.Tsuchiya, Stability Analysis of a Simple Walking Model
Driven by an Oscillator With a Phase Reset Using Sensory Feedback,
IEEE Trans. Robotics, Vol.22, No.2, pp.391-397, 2006
[11] K. Tsujita, K. Tsuchiya and A. Onat, Adaptive Gait Pattern Control
of a Quadruped Locomotion Robot, Proc. IEEE/RSJ IROS 2001,
pp.23182325, 2001
[12] K.Tsuchiya, S.Aoi and K. Tsujita, Locomotion Control of a Biped
Locomotion Robot using Nonlinear Oscillators, Proc. IEEE/RSJ IROS
2003, pp.17451750, 2003
Ekeberg,A combined neuronal and mechanical model of fish
[13] O.
swimming, Biol. Cybern. 69, pp.363-374, 1993
Ekeberg and K. Pearson, Computer Simulation of Stepping in the
[14] O.
Hind Legs of the Cat: An Examination of Mechanisms Regulating
the Stance-to-Swing Transition, J. Neurophysiol., 94, pp.4256-4268,
2005
[15] T. McGeer, Passive dynamic walking, International Journal of
Robotics Research, Vol.9, No.2, pp.62-82, 1990.
[16] M. Wisse and J.Frankenhuyzen, Design and Construction of MIKE;
a 2-D Autonomous Biped Based on Passive Dynamic Walking, Proc.
of Adaptive Motion of Animals and Machines 2005, pp.143-154, 2005
[17] K.Ono and X.Yao, Simulation Study of Self-Excited Walking of a
Biped Mechanism with Bent Knee, Adaptive Motion of Animals and
Machines, pp.131-142, 2005
[18] Y.Sugimoto and K.Osuka, Walking control of quasi-passive-dynamicwalking robot Quartet III based on delayed feedback control, Proc.
of 5th CLAWAR, pp.123-130, 2002
[19] S.Collins, A.Ruina, R.Tedrake and M.Wisse, Efficient bipedal robots
based on passive-dynamic walkers, SCIENCE, Vol.307, pp.10821085, 2005
[20] B.Verrelst et al., The pneumatic biped lucy actuated with pleated
pneumatic artificial muscles, Autonomous Robots, Vol.18, No.2,
pp.201-213, 2005
[21] R.Q.Linde, Design, analysis and control of a low power joint for
walking robots, by phasic activation of mckibben muscles, IEEE
Trans. Robotics and Automation, Vol.15, No.4, pp.599-604, 1999
[22] H. F. Schulte, The characteristics of the McKibben Artificial Muscle,
The Application of External Power in Prosthetics and Orthotics,
Publication 874, National Academy of Sciences - National Research
Council, Washington DC, Appendix H, pp. 94-115, 1961
[23] K. Inoue, Rubbertuators and applications for robotics, Proc. of the
4th International Symposium on Robotics Research, pp. 57-63, 1987
[24] M. Ozkan, K. Inoue, K. Negishi, T. Yamanaka, Defining a neural network controller structure for a rubbertuator robot, Neural Networks,
13(4-5), pp.533-544, 2000
[25] B.Tondu, V.Boitier and P.Lopez, Naturally compliant robot-arms
actuated by McKibben artificial muscles, Proc. of 1994 IEEE International Conference on Systems, Man, and Cybernetics, Vol. 3, pp.
2635-2640, 1994
[26] D.G.Caldwell, N.Tsagarakis, D.Badihi and G.A.Medrano-Cerda,
Pneumatic muscle actuator technology: a lightweight power system
for a humanoid robot, IEEE ICRA 1998, pp. 3035-3058, 1998
[27] D.G.Caldwell, G.A.Medrano-Cerda, and C.J.Bowler, Investigation of
Bipedal Robot Locomotion using Pneumatic Muscle Actuators, Proc.
of the IEEE ICRA 1997, 1997
[28] D.G.Caldwell, G.A.Medrano-Cerda and M.J.Goodwin, Control of
pneumatic muscle actuators, IEEE Control Systems Magazine, Vol.
15(1), pp. 40-48, 1995
[29] S.T. Davis and D.G.Caldwell, The Bio-mimetic Design of a Robot
Primate using pneumatic Muscle Actuators, CLAWAR 2001, pp.197204, 2001
[30] K.Tsujita, T.Inoura, A.Morioka, K.Nakatani, K.Suzuki and T.Masuda,
Oscillator-controlled Bipedal Walk with Pneumatic Actuators,
Proc.of MOVIC 2006, pp.670-675, 2006

2458

Authorized licensed use limited to: Khajeh Nasir Toosi University of Technology. Downloaded on December 21, 2009 at 05:23 from IEEE Xplore. Restrictions apply.

Anda mungkin juga menyukai