Mechatronics 25 (2015) 37–43

Contents lists available at ScienceDirect

Mechatronics
journal homepage: www.elsevier.com/locate/mechatronics

Development of a monolithic compliant SPCA-driven micro-gripper

Dapeng Zhang ⇑, Zhengtao Zhang, Qun Gao, De Xu, Song Liu
Research Center of Precision Sensing and Control, Institute of Automation, Chinese Academy of Sciences, Beijing, China

a r t i c l e i n f o a b s t r a c t

Article history: This paper develops a monolithic compliant SPCA-driven micro-gripper for micro-assembly and describes
Received 27 January 2014 the mechanism design, kinematic model, static model, control strategy and experimental verification of
Accepted 23 November 2014 micro-gripper. The paper includes the following points: (1) the MCM integrates the lever mechanism and
Available online 31 December 2014
parallelogram mechanism so as to magnify the SPCA input and transform the linear input of SPCA into the
clamping movement of gripping jaw. In addition, the kinematic model of the MCM is established to
Keywords: describe the relationship between the SPCA input and the output displacement of gripping jaw. (2)
Micro-gripper
The static model of movement jaw is established to analyze the inherent relationship between the
Compliant mechanism
Kinematic model
gripping force and the strain angle of flexible hinge. (3) The control strategy based on the incremental
Static model PID algorithm is applied to control the gripping manipulation of micro-gripper. The control strategy
selects the visual or force feedback signal according to the contacting condition between the gripping
jaw and clamped micro-object. Finally, a series of experiments were performed to validate the kinematic
model, static model and control strategy of micro-gripper. In addition, a glassy micro-tube with 150 lm
diameter was clamped non-destructively by the developed micro-gripper.
Ó 2014 Elsevier Ltd. All rights reserved.

1. Introduction resolution [12–14]. Therefore, many researchers developed the

SPCA-driven micro-grippers to clamp the micro-object [1–5,15–
It is widely recognized that high precision position systems are 20]. In the above-mentioned micro-grippers, the monolithic
required for many applications from micro-assembly to medical compliant mechanisms (MCM) based on the flexible hinges were
instruments, and hence have received increasing attention by generally utilized to amplify the SPCA input and transform the lin-
many researchers. One of particular interests in this area is the ear input of SPCA into the clamping movement of gripping jaw. The
development of robust micro-grippers that can perform many researchers generally design the MCM by FEM, which can simulate
micro-scaled tasks ranging from micro-assembly processes to min- the output motion of the MCM under the SPCA pushing. Then
imally invasive surgery [1–3]. Micro-gripper is the terminal manip- according to the simulation results, the designers spent plenty of
ulator contacting the micro-object directly, which can decide the time in adjusting the geometrical parameters of the MCM repeat-
micro-assembly task successful or not. The primary requirement edly so as to achieve the expected clamping effect [18,19]. More-
of the micro-gripper in micro-assembly system is to be able to over, the MCM generally adopts the symmetrical structure to
grasp micro-object of different shapes and textures with high accu- clamp the micro-object [5,18–20]. Namely, when SPCA drives the
racy and controlled grasping force [4–6]. This is to ensure that the MCM, two gripping jaws symmetrically clamp the micro-object.
micro-gripper is able to satisfactorily achieve the goal of grasping However, it is difficult that the output displacements of gripping
without damaging the micro-scaled object which generally would jaws are totally equal due to the fabrication and assembly error
be sensitive to external perturbation and gripping force. of micro-gripper.
Various types of micro-grippers have been proposed in the The clamped micro-object is generally delicate, so the gripping
literature, such as piezoelectric, electrostatic, magnetic, electro- force between the gripping jaw and micro-object should be mea-
thermal and shape memory micro-grippers [7–11]. As a typical sured so as to prevent the micro-object from damaging during
smart actuator, the stack piezoelectric ceramic actuator (SPCA) the clamping process. At present, there are two methods to detect
has drawn great attention due to several intrinsic advantages the gripping force: self-sensing method and affixing strain gauges.
such as high force to weight ratio, fast response and high motion Timothy and Yuichi developed the piezoelectric tweezer-type
effector with self-sensing capability so that the piezoelectric
⇑ Corresponding author. micro-grippers could be used as a sensor and an actuator simulta-
E-mail address: [email protected] (D. Zhang).
neously [21–26]. The piezo-electric self-sensing bridge circuit has

http://dx.doi.org/10.1016/j.mechatronics.2014.11.006
0957-4158/Ó 2014 Elsevier Ltd. All rights reserved.
38 D. Zhang et al. / Mechatronics 25 (2015) 37–43

been applied to measure the displacement of gripping jaw and

gripping force. However, the circuit requires exact capacitive
matching of piezoelectric actuator, which has been proven difficult
in practice due to the capacitance variation [21,22]. In addition, the
self-sensing method was not appropriate for the micro-assembly
due to the charge error accumulation during the integration of
the current signals. Consequently, very few works refer to the
use of self-sensing method for the control of micro-assembly
[27]. In addition, some researchers affixed the strain gauges on
the flexible hinges of the MCM so that the gripping force was mea-
sured by detecting the strain angle of flexible hinge [5,28–30].
However, the inherent relationship between the gripping force
and strain angle of flexible hinge has not been discussed.
In the aforementioned micro-grippers, the micro-grippers were
controlled by the single feedback signal such as the input force of Fig. 1. 3D model of the SPCA-driven micro-gripper.
SPCA, the input displacement of SPCA or the gripping force
between the micro-object and gripping jaw [5,31–33]. However,
in order to clamp the micro-object rapidly and safely, the control other units; III. SPCA: provides the MCM with input displacement;
strategy of micro-gripper should select the visual or force feedback IV. SPCA cover: connects the SPCA with platform base; V. MCM:
signal according to the contacting condition between the gripping transforms the linear input of SPCA into the clamping motion; VI.
jaw and clamped micro-object so that the micro-gripper can Strain sensor: measures the strain angle of flexible hinge.
provide the micro-object with the desired gripping motion and In general, the MCM is divided into two units: lever mechanism
gripping force. (ABC) and parallelogram mechanism (EFGH), as shown in Fig. 2.
This paper describes the mechanism design, kinematic model, The lever mechanism (ABC) is composed of the corner-filleted flex-
static model, control strategy and experimental validation of a ible hinge B, the rigid link AB and BC. When the voltage is applied
SPCA-driven micro-gripper. The paper includes the following to the SPCA controller, SPCA drives the MCM at the point A. Then
points: (1) the MCM integrates the lever mechanism and parallel- the lever mechanism rotates around the axial center of flexible
ogram mechanism in order to amplify the SPCA input and trans- hinge B, which can amplify the linear input of the SPCA in accor-
form the linear input of SPCA into the clamping movement of dance with the lever principle. The output movement of lever
gripping jaw. The MCM adopts asymmetric structure to clamp mechanism (ABC) is transmitted to parallelogram mechanism
the micro-object so as to improve the gripping accuracy. Moreover, (EFGH) by the rigid link CD. The parallelogram mechanism includes
the kinematic model of the MCM is established to describe the four rigid links: EF, FG, GH and HE. The connecting nodes of adja-
relationship between the SPCA input and the output displacement cent links E, F, G and H adopt the structure of single-notch flexible
of gripping jaw. According to the kinematic model, the designer hinges, as shown in Fig. 2. When the rigid link CD drives the paral-
can calculate the magnification ratio by the geometrical parame- lelogram mechanism (EFGH), the movement jaw moves toward the
ters of MCM so as to avoid modifying the parameters repeatedly. stationary jaw due to the rotation of the flexible hinge E, F, G and H
(2) The static model of movement jaw is established to analyze around their central axes [21–22]. The parallelogram mechanism is
the inherent relationship between the gripping force and strain able to further magnify the input motion of rigid link CD by
angle of flexible hinge. According to the static model, the output modifying the length of rigid link DE and DF.
of gripping force can be adjusted by modifying the geometrical
parameters of movement jaw. (3) The control strategy based on 3. Modeling and analysis of the micro-gripper
the incremental PID is applied to control the clamping manipula-
tion of micro-gripper. The control strategy selects the visual or 3.1. Kinematic model of the MCM
force feedback signal according to the contacting condition
between the gripping jaw and clamped micro-object. Finally, a ser- When establishing the kinematic model of the MCM, we
ies of experiments were performed to validate the kinematic assume that: (1) the flexible hinge is considered as two rigid links
model, static model and control strategy. A glassy micro-tube with and movable hinge with torsion spring; (2) the deformation on the
150 lm diameter was clamped non-destructively by the developed flexure hinge is assumed to be pure rotation without expansion
micro-gripper. and contraction and the rotational angle is less than 10[5]. There-
The paper is organized as follows: Section 2 describes the fore, the kinematic model of MCM can be described by several rigid
structure and principle of SPCA-driven micro-gripper. Section 3
discusses the kinematic model of the MCM, the static model of
movement jaw and control strategy of clamping manipulation.
Section 4 focuses on the experiments to validate the established Parallelogram Stationary jaw
kinematic model, static model and control strategy. Section 5 mechanism
summarizes the research and suggests the possible directions for
the future research.
H G

A E D F
2. Structure and principle of the SPCA-driven micro-gripper
B
C
Single notch Movement jaw
The 3D model of developed micro-gripper consists of pre-
flexible hinge
adjustment screw, platform base, SPCA, SPCA cover, MCM and Corner-filleted Lever mechanism
strain sensor, as shown in Fig. 1. The functional description of var- flexible hinge
ious components is as follows. I. Pre-adjustment screw: provides
the SPCA with preload; II. Platform base: installs the MCM and Fig. 2. The structure of developed MCM.
 D. Zhang et al. / Mechatronics 25 (2015) 37–43 39

links and flexure hinges H1  H5, as shown in Fig. 3. Where din dP dG

denotes the SPCA input, dP and dG indicate the corresponding out-
put of parallelogram mechanism and movement jaw respectively. H G
x
When the SPCA pushes the MCM, the flexible hinges H1  H5 l5
o '
rotate around their central axes so as to achieve the clamping y DD
d in l4 E
manipulation of the micro-gripper. The solid lines indicate the ini- A B C' F
tial position of the MCM, the dashed lines denote new equilibrium A 'Δθ B H1 θ o C Δθ o
position accordingly, as shown in Fig. 4. The parallelogram mecha-
l1 l2 l3
nism only exhibits one stable equilibrium position due to the small
rotational angle of flexible hinges [34,35]. As shown in Fig. 4, the Fig. 4. The kinematic model of the MCM.
point A denotes the contacting point between the SPCA and
MCM. When the SPCA provides the MCM with input displacement
0
din, the point A moves to A due to the rotation of flexible hinge H1
γ
around the point B (the center of flexible hinge H1). DhB indicates
0
the rotational angle between the vector AB and A B. Correspond- dG dG dG fG
0
ingly, the endpoint of rigid link CD varies from the point C to C ,
the motion distance Ds (point C) can be calculated by the Eq. (1). micro-object

ðl2 þ l3 Þdin L
Ds ¼ ð1Þ H3
l1
When the parallelogram mechanism rotates Dh0 around the
point O (the midpoint of rigid link EH), the corresponding displace- dP dP dP δ
ment of point C along x and y axis can be calculated by the Eq. (2).
 (a) (b) (c)
DxC ¼ x0C  xC ¼ jOCj½sinðh0 þ Dh0 Þ  sin h0  ¼ l4 Dh0
ð2Þ Fig. 5. The grasping process of micro-gripper.
DyC ¼ y0C  yC ¼ jOCj½cosðh0 þ Dh0 Þ  cos h0  ¼ l3 Dh0
where xc and yc denote the initial coordinates of point C, x0 c and y0 c
indicate the new equilibrium coordinates. l1  l5 denote the size of
several rigid links. h0 indicates the original angle between the vector the output of gripping force can be calculated by measuring
OC and link HE. Dh0 denotes the rotational angle due to the position the strain angle of flexible hinge. Namely, the gripping force
variation of point A, which is generally far less than 10. So it can be of micro-gripper can be regulated by adjusting the geometrical
assumed that: cos(Dh0)  1 and sin(Dh0)  0. parameters of the movement jaw. The grasping process of
According to the Eqs. (1) and (2), the SPCA input din can be indi- micro-gripper can be divided into three stages, as shown in
cated by and l1  l5 and Dh0, as shown in the Eq. (3). Additionally, Fig. 5. The movement jaw can be equivalent to cantilever beam,
according to the geometrical structure of MCM, the output the clamped micro-object is assumed to be a cylinder. First, the
displacement of parallelogram mechanism dP can be represented movement jaw does not contact the gripped micro-object, nei-
by the Eq. (4). ther the gripping force nor the strain angle of flexible hinge
qffiffiffiffiffiffiffiffiffiffiffiffiffi H3 occurs. At this time, the displacement direction of dG and
2 2
l1 l3 þ l4 dP is parallel, c denotes the original distance between the move-
din ¼ Dh0 ð3Þ ment jaw and the micro-object, as shown in Fig. 5(a). Then
l2 þ l3
when the movement jaw initially contacts the micro-object,
dP ¼ l5 Dh0 ð4Þ the gripping force fG and the strain angle of flexible hinge H3
are still zero, the output displacement of movement jaw
Therefore, the theoretical magnification ratio k of MCM can be dG = c, as shown in Fig. 5(b). Finally, the parallelogram mecha-
defined by the Eq. (5). nism continues going forward with the increase of SPCA input,
however, the movement jaw is prevented by the clamped
dP l5 ðl2 þ l3 Þ
k¼ ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffi ð5Þ micro-object. Therefore, dP is gradually greater than dG (d = dP
din 2 2
l1 l3 þ l4  dG) and the movement jaw rotates around the central axis
of flexible hinge H3, as shown in Fig. 5(c). In this case, the
clamping force fG and the strain angle of flexible hinge H3
3.2. Static model of the movement jaw appear. The micro-object is clamped tightly by the movement
jaw and stationary jaw.
The static model of the movement jaw is established to dis- Based on the cantilever theory, the static model of the move-
cuss the inherent relation between the gripping force and the ment jaw describing the inherent relationship between the grip-
strain angle of flexible hinge H3. Based on the static model, ping force and strain angle of flexible hinge H3 is established in
the Eq. (6).

dP H4 dG
9pr 1=2 L
H5 u ¼ pffiffiffi 5=2 f G ð6Þ
2 2Ebt
H G
H2 H3
d in A E D F where u indicates the strain angle of flexible hinge H3 around the
H1 central axis. r and t represent the radius and thickness of flexible
B
C hinge H3 respectively. b indicates the width of flexible hinge H3. L
denotes the length of cantilever beam. E is the elasticity modulus
Fig. 3. The kinematic scheme of the MCM. of aluminum alloy.
40 D. Zhang et al. / Mechatronics 25 (2015) 37–43

3.3. Control strategy of the clamping manipulation 4. Experimental results and discussions

In order to clamp the micro-object firmly and safely, the 4.1. Experimental setup
gripping force and the output displacement of movement jaw
should be controlled accurately. The increment PID algorithm can The developed micro-gripper includes SPCA, MCM, strain sen-
avoid cumulative error and response rapidly according to the feed- sor, SPCA cover and platform base, as shown in Fig. 7. The basic
back signals. Therefore, the increment PID algorithm is adopted to operation of micro-gripper commences by applying the desired
control the clamping manipulation of micro-gripper. The control voltage from the SPCA controller to the preloaded SPCA to activate
strategy of micro-gripper selects the visual or force feedback signal the output motion of movement jaw. The maximum travel and
according to the contacting condition between the gripping jaw motion resolution of SPCA (type: PI-P841.40) are 60 lm and
and clamped micro-object, as shown in Fig. 6. Where cini indicates 0.6 lm respectively. The MCM was made of aluminum alloy (type:
the original distance between the micro-object and movement jaw. 7075T6) which was very suitable for compliant mechanism due to
c(k) denotes the real-time distance between the micro-object and high flexibility, and the MCM was fabricated by wire electro-
movement jaw, which is acquired by extracting the feature of discharge machining (EDM) technology. Strain sensor (type:
movement jaw from captured microscopic-vision images. e(k) is SAK120-1-C11-P003M-V2M, Gage Factor: 2.13 ± 1%, Gage length:
the difference between the cini and c(k), k is sampling number. 1 mm, Gage Resolution: 120 ± 0.5 OHM) was sticked on the central
KP, KI and KD represent the proportional, integral and derivative axis of flexure hinge H3 to measure the strain angle during the
coefficient respectively. T0 denotes the sampling period, Ti and Td clamping process.
indicate the integral and derivative constant respectively.
" #
1X k
eðkÞ  eðk  1Þ 4.2. Verification of the kinematic model of the MCM
uðkÞ ¼ K P eðkÞ þ T 0 eðk  iÞ þ T d ð7Þ
T i i¼0 T0
The experiment was performed to verify the kinematic model of
First, when the movement jaw does not contact the micro- the MCM established in the Eq. (5). The input displacement of the
object, the SPCA input is controlled according to the visual feed- MCM was controlled by SPCA, and the corresponding output dis-
back. In this case, the clamping force f(k) = 0, the control variable placement of MCM could be calculated by extracting the edge of
of SPCA controller u(k) is represented by the Eq. (7) according to movement jaw from microscopic-vision images captured by
the literature [36,37]. Then when the movement jaw contacts the microscopic lens (type: Navitar Zoom 6000) and CCD camera (type:
micro-object, the clamping force f(k) is not equal to zero. In this AVT GC2450). Fig. 8(a) demonstrated the original position of the
case, the control variable u(k) is calculated according to the clamp- movement and stationary jaw. With the increase of SPCA input,
ing force f(k) which is measured by strain sensor affixed on the the movement jaw gradually approached the stationary jaw, as
central axis of flexible hinge H3. Finally, when the difference shown in Fig. 8(b).
between the maximum allowable gripping force fmax (Which is According to the geometrical parameters of the MCM listed in
determined by the material properties of clamped micro-object.) Table 1, the theoretical magnification ratio was calculated by the
and f(k) is less than the setting threshold, the ‘‘Switch’’ of clamping Eq. (5). In the clamping experiment, the SPCA input varied from
manipulation is turned off so as to protect the clamped micro- 0 to 30 lm, the corresponding output displacement of MCM was
object. calculated by extracting the edge of movement jaw from the
In order to avoid the cumulative error of controlling variable, microscopic-vision images. As shown in Fig. 9, the experimental
the incremental input Du(k) is defined to control the SPCA input, and theoretical results were indicated by the red and blue lines
as shown in the Eq. (8). Where integral coefficient KI = T0/Ti, deriv- respectively, the slope of lines denotes the magnification ratio. As
ative constant KD = Td/T0. From the Eq. (8), we can see clearly that shown in Fig. 9, the experimental result of amplification ratio
the incremental input Du(k) only includes three proximate sam- was approximately 6.0, the theoretical result calculated by the
pling value, which would effectively avoid the accumulation error. Eq. (5) was 7.8. The theoretical magnification ratio was greater
Moreover, the clamping manipulation of micro-gripper is generally than the experimental value. The reason arising from error
slow so as to protect the clamped micro-object. Therefore, the between the theoretical and experimental results: (1) when
derivative adjustment can be ignored in the control algorithm of
gripping manipulation. So the increment input Du(k) of SPCA
controller can be represented in the Eq. (9). SPCA cover Platform base

DuðkÞ ¼ uðkÞ  uðk  1Þ

¼ K P feðkÞ  eðk  1Þ þ K I eðkÞ þ K D ½eðkÞ  2eðk  1Þ þ eðk  2Þg
ð8Þ

DuðkÞ ¼ K P ½eðkÞ  eðk  1Þ þ K I eðkÞ ð9Þ

Feature extration Image capture

(k)
_ KP
γini + e(k) u(k) SPCA
Switch KI Micro-gripper
controller
KD
_
fmax + f(k) Filtering and Strain signal
amplification acquisition SPCA MCM Strain sensor

Fig. 6. Control strategy of gripping manipulation. Fig. 7. The developed micro-gripper.

 D. Zhang et al. / Mechatronics 25 (2015) 37–43 41

Vertical
cylinder

Stationary
stationary movement stationary movement jaw
jaw jaw jaw jaw
Movement
jaw

Strain
sensor

Force
sensor
(a) (b)
Fig. 8. The microscopic images of grasping process.
Fig. 10. The experimental setup to verify the static model.

Table 1
force would be transmitted to the force sensor by vertical cylinder.
Geometrical parameters of the MCM.
Therefore, the gripping force between the micro-tube and move-
Parameters Value (mm) Parameters Value (mm) ment jaw could be measured accurately. In addition, the strain
l1 11.75 l4 5 angle under the gripping force was detected by the strain sensor
l2 14.75 l5 30 affixed along the central axis of flexible hinge H3. The experimental
l3 2.75
result describing the relationship between the gripping force and
strain angle of flexible hinge H3 was described in blue lines, as
shown in Fig. 11. Moreover, according to the geometrical parame-
250 ters of movement jaw listed in Table 2, the theoretical relationship
between the gripping force and strain angle of flexible hinge H3
output displacement of movement jaw (μm)

Experimental
Theretical was illustrated in red lines, as shown in Fig. 11.
200 When the strain angle of flexible hinge was less than 2  103
rad, the theoretical value of gripping force calculated from static
model of the movement jaw generally tracked the experimental
150 result, as shown in Fig. 11. When the strain angle of flexible hinge
was larger than 2  103 rad, the slopes of two curves were
approximately equal. Moreover, with the increase of strain angle,
100
the variation of experimental and theoretical values is identical.
Therefore, it can make a conclusion that the established static
model of the movement jaw can describe the inherent relationship
between the gripping force and strain angle of flexible hinge H3.
50
The several factors arising from the errors between the theoretical
and experimental result: (1) the static model assumes the move-
ment jaw to be a cantilever beam. When the bending angle of
0
0 5 10 15 20 25 30 movement jaw under the gripping force is tiny, the assumption
may be reasonable. However, with the increase of bending angle,
input displacement of SPCA (μm)

Fig. 9. Comparison of magnification ratio between experimental and theoretical

results. 3
Theoretical
Experimental
2.5
establishing kinematic model of the MCM, the flexible hinges are
assumed to be pure bend without the expansion and contraction.
However, the flexure hinges would expand and contract along their 2
axial direction and the rotation centres of flexure hinges may devi-
gripping force (N)

ate from the ideal position. (2) The SPCA does not lie in the central 1.5
line of the MCM, which can also lead to the error between the
experimental and theoretical magnification ratio.
1

4.3. Validation of static model of the movement jaw 0.5

The experiment to clamp the diameter of 150 lm was

0
performed to validate the static model of the movement jaw
established in the Eq. (6). The experimental setup mainly consists
of force sensor, vertical cylinder, strain sensor and micro-gripper, -0.5
0 0.5 1 1.5 2 2.5 3 3.5 4 4.5 5
as shown in Fig. 10. The vertical cylinder connected with the force -3
strain angle of flexible hinge (rad) x 10
sensor (type: ATI Nano17, sensing range: Fx = Fy = 12 N, Fz = 17 N,
resolution: 3.9 mN) was in touch with the stationary jaw. When Fig. 11. Theoretical and experimental relationship between the gripping force and
the micro-tube was clamped by the micro-gripper, the gripping strain angle.
42 D. Zhang et al. / Mechatronics 25 (2015) 37–43

Table 2 350
Geometrical parameters of movement jaw.
300

Gripping Force of Micro-gripper (mN)

Parameters Value (mm) Parameters Value (mm)
L 7 r 1
t 0.5 b 3 250

200

150
120
Output Displacement of Micro-gripper (μm)

100
100
50
80
0
60
-50
0 20 40 60 80 100 120 140
40 Time (ms)

Fig. 13. The variation curve of gripping force.

20

-20
Micro-tube
0 10 20 30 40 50 60 70 80
Time (ms)

Fig. 12. The variation curve of output displacement.

the cantilever beam assumption is not suitable for the static model
of movement jaw. (2) The installation of strain sensor may cause
error. Ideally, the central axis of strain sensor should coincide with
the axis of flexible hinge H3 so that the strain sensor can detect the
strain angle accurately. However, when the strain sensor is affixed
to the flexible hinge, an angle between two axes exists inevitably. Fig. 14. The clamping manipulation of micro-tube.
(3) The signal processing may result in the error. The signals of
strain angle need to be filtered and amplified so as to perform Another experiment to clamp a glassy micro-tube with a diam-
the subsequent processing, which can cause the deviation between eter of 150 lm was performed to validate the stability of gripping
the theoretical and experimental values. force. In this case, the controlling goal of gripping force was
300 mN, the proportional coefficient KP = 0.05, internal constant
4.4. Characteristics of the control strategy KI = 0.8. The experimental result was shown in Fig. 13, the variation
curve of gripping force was approximately divided into three
The control strategy of micro-gripper was verified by testing phases. First, when the movement jaw did not contact the
motion precision of movement jaw and stability of gripping force. micro-tube, the SPCA input was controlled according to the visual
First, an experiment to control the movement jaw to motion feedback and the gripping force was zero. Then when the move-
100 lm was proposed to validate the motion accuracy of move- ment jaw touched the micro-tube, the SPCA input was controlled
ment jaw. The input displacement of micro-gripper was controlled according to the feedback signals of gripping force. At this time,
by SPCA, the corresponding output of movement jaw was calcu- the gripping force reached 300 mN within 3 ms. Finally, when
lated by extracting the edge of movement jaw from microscopic- the gripping force reached 300 mN (the maximum allowable grip-
vision images. According to the visual feedback acquired from ping force), the ‘‘Switch’’ of clamping manipulation was turned off
microscopic-vision images, the proportional constant KP and to protect the clamped micro-tube. At this time, the SPCA stopped
integral coefficient KI were adjusted repeatedly so that the output driving the MCM further, and the clamping force was stabilized at
displacement of movement jaw could be controlled accurately. 300 mN, as shown in Fig. 13. Therefore, based on the proposed con-
Ultimately the proportional constant KP = 0.1, internal coefficient trol strategy of micro-gripper, the gripping force between the
KI = 0.8. The experimental result was shown in Fig. 12, when the movement jaw and micro-tube could be controlled precisely and
SPCA did not drive the micro-gripper, the output displacement of stabilized continuously, which could avoid damaging the clamped
movement jaw initially was zero. With the increase of SPCA input, micro-object due to the overshoot of gripping force. The glassy
the output displacement of movement jaw reached 100 lm within micro-tube with a diameter of 150 lm was clamped non-destruc-
5 ms. Then the SPCA did not drive the micro-gripper further tively by the developed micro-gripper, as shown in Fig. 14.
according to the control strategy. In this case, the output displace-
ment of movement jaw remained approximately constant, the 5. Conclusion and future work
overshoot of output displacement was less than 5 lm. Therefore,
based on the proposed control strategy, the motion of movement This paper develops a monolithic compliant SPCA-driven micro-
jaw could be controlled precisely and stabilized continuously, gripper for micro-assembly and describes the kinematic modeling,
which would prevent the clamped micro-object from damaging mechanism design, static model, control strategy and experimental
on account of the output overshoot of movement jaw. verification of micro-gripper. The paper includes the following
 D. Zhang et al. / Mechatronics 25 (2015) 37–43 43

points: (1) the MCM integrates the lever mechanism and parallel- [9] Giouroudi I, Hotzendorfer H, Andrijasevic D, Ferros M, Brenner W. Design of a
microgripping system with visual and force feedback for MEMS application.
ogram mechanism in order to amplify the SPCA input and trans-
The Institution of Engineering and Technology Seminar on MEMS Sensors and
form the linear input of SPCA into the clamping movement of Actuators; 2006. p. 243–50.
gripping jaw. The MCM adopts asymmetric structure to clamp [10] Lopez-Wall B, Gauthier M, Chaillet N. Principle of a submerged freeze gripper
the micro-object to improve the gripping accuracy. Moreover, the for microassembly. IEEE Trans Robotics 2008;24(4):897–902.
[11] Lin Che-Min, Fan Chen-Hsien, Lan Chao-Chieh. A shape memory alloy actuated
kinematic model of MCM is established to describe the relationship microgripper with wide handing ranges. In: IEEE/ASME international
between the SPCA input and output displacement of gripping jaw. conference on advanced intelligent mechatronics; 2009. p. 12–17.
(2) The static model of movement jaw is established to analyze the [12] Edamana B, Oldham KR. Optimal low-power piezoelectric actuator control
with charge recovery for a microrobotic leg. IEEE Trans Mechatronics 2013;
inherent relationship between the gripping force and strain angle 18(1):251–62.
of flexible hinge. According to the static model, the output of [13] Zhang Yanliang, Han Mingli, Shee Cheng Yap. Calibration of piezoelectric
gripping force can be adjusted by modifying the geometrical actuator-based vision guided cell microinjection system. In: IEEE/ASME
international conference on advanced intelligent mechatronics; 2008. p.
parameters of movement jaw. (3) The control strategy based on 808–12.
the incremental PID is applied to control the clamping manipula- [14] Zhou Yong, Zhou Beiyue, Li Shi. Driving performance analysis of a novel
tion of micro-gripper. The control strategy selects the visual or piezoelectric actuator. In: International conference on measuring technology
and mechatronics automation; 2010. p. 34–7.
force feedback signals according to the contacting condition [15] Grossard M, Boukallel M, Chaillet N, Rotinat-Libersa C. Modeling and robust
between the gripping jaw and clamped micro-object. The ‘‘Switch’’ control strategy for a control optimized piezoelectric microgripper. IEEE Trans
of gripping manipulation is utilized to protect the clamped micro- Mechatronics 2011;16(4):674–83.
[16] Grosssard M, Rotinat-Libersa C, Chaillet N, Boukallel M. Mechanical and
object. Finally, a series of experiments were performed to validate
control-oridented design of a monolithic piezoelectric microgripper using a
the kinematic model, static model and control strategy. A glassy new topological optimization method. IEEE Trans Mechatronics 2009;
micro-tube with 150 lm diameter was clamped non-destructively 14(1):32–45.
by the developed micro-gripper. In addition, the reasons arising [17] Nashrul Mohd, Zubir Mohd, Shirinzadeh Bijan. Development of a high
precision flexure-based micro-gripper. Precision Eng 2009;33:362–70.
from error between the theoretical and experimental results have [18] Zubir MNM, Shirinzadeh B. Development of a high precision flexure based
been discussed. microgripper. Precision Eng 2009;33(4):362–70.
The experimental results have showed that: (1) the integration [19] Zubir MNM, Shirinzadeh B, Tian YL. Development of a novel flexure-based
microgripper for high precision micro-object manipulation. Sensors and
of lever mechanism and parallelogram mechanism can amplify the Actuators 2009;150(2):257–66.
SPCA input and achieve the gripping motion. The magnification [20] Nah SK, Zhong ZW. A microgripper using piezoelectric actuation for micro-
ratio of MCM calculated by kinematic model generally matches object manipulation. Sensors and Actuators 2007;133:218–24.
[21] Mcpherson Timothy, Ueda Jun. Piezoelectric self-sensing technique for
the experimental result. (2) The static model of movement jaw tweezer style end-effector. In: IEEE international conference on intelligent
could reflect the relation between the gripping force and strain robots and systems; 2011. p. 1940–5.
angle of flexible hinge H3. (3) The control strategy of micro-gripper [22] Kurita Yuichi, Sugihara Fuyuki, Ueda Jun, Ogasawara Tsukasa. Piezoelectric
Tweezer-type end effector with force and displacement sensing capability.
based on the incremental PID can satisfy with the demanding of IEEE/ASME Trans Mechatron 2012;17(6):1039–48.
the gripping manipulation. The motion precision of movement [23] Hideyuki Ikeda, Takeshi Morita. High-precision position using a self-sensing
jaw and stable gripping force are achieved accurately. piezoelectric actuator control with a differential detection method. Sensors
and Actuators 2011;A 170:147–55.
In the future, the micro-gripper would be further explored for
[24] Kawamata A, Kadota Y, Hosaka H, Morita T. Self-sensing piezoelectric actuator
downsizing, and hysteresis is also considered in control strategy using permittivity detection. J Ferroelectrics 2008;368:194–201.
so as to improve the gripping manipulation. In addition, the devel- [25] Ivan IA, Rakotondrabe M, Lutz P, Chaillet N. Current integration force and
oped micro-gripper serves as a foundation to research the dynamic displacement self-sensing method for cantilevered piezoelectric actuators. Rev
Sci Instrum 2009;80(12):126103.
response so as to obtain more robust and industrially-oriented [26] Ivan IA, Rakotondrabe M, Lutz P, Chaillet N. Quasistatic displacement self-
design. sensing method for cantilevered piezoelectric actuators. Rev Sci Instrum
2009;80(6):065102.
[27] Rakotondrabe Michy, Ivan Ioan Alexandru, Khadraoui Sofiane, Lutz Philippe,
Chaillet Nicolas. Simultaneous displacement/force self-sensing in piezoelectric
Acknowledgements
actuators and applications to robust control. IEEE/ASME Trans Mechatron
2012;17(6):1039–48.
The author would like to thank the National Natural Science [28] Xu Qingsong. A new compliant microgripper with integrated position and
force sensing. In: IEEE/ASME International Conference on Advanced Intelligent
Foundation of China (Grant Nos. 61303177 and 61105036) which
Mechatronics; 2013: p. 591–6.
supported the development of micro-gripper. [29] Carrozza MC, Eisinberg A, Menciassi A, Campolo D. Towards a force-controlled
microgripper for assembling biomedical microdevices. J Micromech Microeng
2000;10:271–6.
References [30] Kemper M. Development of a tactile low-cost microgripper with integrated
force sensor. In: International Conference on Control Application; 2004. p.
1461–6.
[1] Bolopion A, Regnier S. A review of haptic feedback teleoperation systems for
[31] Chang Ren Jung, Cheng Chih Yi. Vision-based compliant-joint polymer force
micromanipulation and microassembly. IEEE Trans Automat Sci Eng
sensor integrated with microgripper for measuring gripping force. In:
2013;10(3):496–502.
International conference on advance intelligent mechatronics; 2009. p. 18–23.
[2] Qingsong Xu. Precision position/force interaction control of a piezoelectric
[32] Beyeler F, Neild A, Oberti S. Monolithically fabricated microgripper with
multimorph microgripper for microassembly. IEEE Trans Automat Sci Eng
integrated force sensor for manipulating microobjects and biological cells
2013;10(3):503–14.
aligned in an ultrasonic field. IEEE/ASME J Microelectromech Syst 2007;
[3] Sharma HD, Bhardwaj S, Singh RK, Naruka GS, Tholia A. Automated alignment,
16(1):7–15.
focusing and control of nanomanipulators in microassembly workcell. In:
[33] Jayaram K, Joshi SS. Development of a flexible-based force sensing
International conference on mechatronics and automation; 2011. p. 177–82.
microgripper for micro-object manipulation. J Microelectromech Syst 2010;
[4] Rakotondrabe M, Ivan IA. Development and force/position control of a new
20(1):15–25.
hybrid thermo-pieoelectric microgripper dedicated to micro-manipulation
[34] Zubir Mohd, Nashrul Mohd, Shirinzadeh B. Development of a compliant-based
tasks. IEEE Trans Automat Sci Eng 2011;8(4):824–34.
microgripper for microassembly. In: IEEE/ASME international conference on
[5] Wang DH, Yang Q, Dong HM. A monolithic compliant piezoelectric-driven
mechatronic and embedded systems and applications; 2008. p. 522–7.
microgripper: design, modeling and testing. IEEE Trans Mechatronics
[35] Xuefeng Yang, Wei Li, Yuqiao Wang, Gao Ye. Output displacement analysis for
2013;18(1):138–47.
compliant single parallel four-bar mechanism. In: International conference on
[6] Komati B, Rabenorosoa K, Clevy C, Lutz P. Automated guiding task of a flexible
mechatronics and automation; 2010. p. 1354–7.
micropart using a two-sensing-finger microgripper. IEEE Trans Automat Sci
[36] Widyotriatmo A, Rauzanfiqr S K, Suprijanto S. A modified PID algorithm for
Eng 2013;10(3):515–24.
dynamic control of an automatic wheelchair. In: IEEE conference on control
[7] Qingsong Xu. Adaptive discrete-time sliding mode impedance control of a
system and industrial information; 2012. p. 64–8.
piezoelectric microgripper. IEEE Trans Robotics 2013;29(3):663–73.
[37] Tzafestas S, Papanikolopoulos NP. Incremental fuzzy expert PID control. IEEE
[8] Boudaoud M, Haddab Y. Modeling and optimal force control of a nonlinear
Trans Ind Electron 1990;37(5):365–71.
electrostatic microgripper. IEEE Trans Mechatronics 2013;18(3):1130–9.