design and development of a monolithic gripper with flexure...
TRANSCRIPT
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 109
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
Abstract— This paper presents a robotic gripper design
which relies on flexural joints rather than prismatic and
revolute joints to ensure gripping action. The designed
flexure-based gripper exhibits large displacement that fall
within the millimetre scale range. With large displacements,
the gripper is used to handle kiwifruit and at the same time
determine its softness for sorting purposes. Flexural joint
grippers with kiwifruit-size range of displacement are
currently non-existent in the industry. Flexural joint
mechanism is based on the intrinsic compliance of a
material to achieve structural elastic deformation to achieve
displacement. The gripper is monolithic in nature to
properly accommodate flexure joints and also any further
works in microelectronics process integration. The
monolithic model exhibits more lumped compliancy
displacement than distributed compliancy. The feasibility
of the design is examined through a series of numerical
analysis and simulations on COMSOL platform. Crucial
design standards factored in are low fabrication and
material costs, parallel gripping motion of fingers, large
finger displacement, low joint stress concentrations and low
output forces. A prototype is realised through subtractive
fabrication of the prototype. Through a series of
experiments the maximum range of motion for the gripper
is found to be 20mm, half of which is contributed by each
finger. Another set of experiments were done to investigate
whether the gripper can accurately affirm the softness of
the manipulated kiwifruits. The flexural gripper was able to
successfully grasp and manipulate kiwifruit, at the same
time determining fruit softness.
Index Term— flexure joints, large displacement, softness
detection, monolithic gripper.
I. INTRODUCTION
The implementation of flexural joints in conventional robotic
gripping has not been conceived because of the small
displacements associated with flexure joints [1]–[3].
Notwithstanding, the development of a monolithic structure
becomes possible with the use of flexural joint mechanism. The
most common type of grippers are mechanical grippers,
vacuum grippers, pneumatic grippers, hydraulic grippers and
magnetic grippers [4][5]. Newer technologies involve the use
of soft grippers from highly compliant materials such as
silicone, rubber and other elastomeric polymers [6]. The use of
flexural joints mechanism is highly prevalent in
Micro Electromechanical Systems (MEMS) technology [7].
The major reason for the developing the gripper as a monolithic
structure is because it accommodates any possible future
integration with microelectronics. In addition, a monolithic
structure eradicates the need for assembly to construct the
gripper and consequently relative reduction in fabrication costs
[8]. Furthermore, any friction related ramifications like
backlash, wear and loss of accuracy are not a factor.
With flexure joints producing either lumped or distributed
compliancy [9], various lumped flexure joints have been
developed, and a study to compare the performances of one
against another has been conducted [8], [10], [11]. As presented
in [8], flexure hinges of rectangular, circular and parabolic form
were evaluated against each other to determine which one has
the edge above the others in terms of maximizing deformation
from applied forces, stiffness profile along the joints and the
stress analysis. Other authors have transformed already existing
prismatic and revolute joints into their flexure joints
counterparts [1] using the rigid-compliant mechanism
transformation process. The pseudo-rigid-body model is
employed as a license to apply similar kinematic analysis
procedures as the ones used for traditional rigid joints.
Many designs have been introduced and customized to suit
the very purposes for which they were developed [7], [10], [12],
[13], with [7] producing an atlas of most flexure-based grippers
already designed. The closest to achieving macroscale gripping
with flexure joints was [14], obtaining a maximum of 0.14mm
from circular flexure joints. Most flexure based grippers are
normally open grippers. Therefore, their fingers are designed
with a gap that is slightly larger than the object to be
manipulated [2]. The extra space is within the gripping stroke
of both fingers. Other types of grippers like the one presented
in [15] has articulated joints and variable stiffness fingers based
on structure controlled principle. The gripper possesses the
capacity to change the stiffness of the fingers based on the
mechanical object changes.
Softness detection has been implemented on fruits before and
other objects like meat and human touch. Sensory technology
for the detection of human touch and hardness was developed
in [16]. Fruit softness detection has been done using various
methods like Laser Doppler Vibrometer technology [17] and by
use of different types of penetrometers [18],[19], which is the
prevalent way of obtaining the softness of fruits. Various fruits
have been tested and a review has been done on non-destructive
fruit firmness sensors [20] in which kiwifruit was among the
Design and Development of a Monolithic
Gripper with Flexure Joints for Handling
Kiwifruit
Ngonidzaishe N. Mrewa, Ahmed M. R. Fath El-Bab and George N. Nyakoe
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 110
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
fruits tested. Tactile sensing technology has been explored for
determining the softness of soft tissues. The tactile sensor has
been tested on specimen chips, silicone rubber and meat parts
like chicken legs and the heart [21]–[23]. The technology was
further advanced after modifications of the sensor where used
for testing the ripeness of mango fruits [24].
In this paper, the development of the flexure-based gripper
had to undergo materials analysis, structural optimization and
boundary and loading conditions consideration. Plexiglas
material was used, and the gripper has 4-bar linkages with
circular flexure joints and carries out friction type gripping.
With large displacements, the gripper can manipulate kiwifruit.
Following 3D CAD design, analytical and numerical simulation
(using ANSYS and COMSOL), a prototype was fabricated
using laser machining. Hence, this paper discusses the design
criteria implemented in order to realize the gripper model, the
executed FEA computation and the fabrication and
performance of the gripper prototype.
II. GRIPPER DESIGN
In this section, the design criteria is introduced.
A. Design and Material Selection Criteria
1) The gripper will be a monolithic structure with flexural
joints to avoid friction, backlash and wear which are
prevalent in revolute and prismatic joints. The monolithic
structure further accommodates any possible future
integration with microelectronics.
2) The gripper material should be available and be of low
cost to allow for disposable replacement of the gripper if
needed for contamination prevention.
3) The dimensions of the gripper customized based on
kiwifruit dimensions.
4) The gripper should exhibit low element stresses, small
driving forces and high mechanical advantage
(displacement amplification factor).
5) The gripper jaws displacement range will be maximized
to cover the large range of the kiwi size.
6) The gripper structure should ensure that the gripper jaws
are parallel during grasping and releasing the fruit to ensure
a more secure, centralized grasp and also avoid fruit damage
7) The gripper will have a force sensor and a set
displacement determination criteria for the purpose of
detecting kiwi softness during grasping. This is for real-time
sorting of kiwi.
B. Gripper Materials Selection
This selection was done based on criteria 1 and 2.
A material to be used for flexural joint implementation should
give a high acceptance to any deflection related motion and high
resistance to failure, fracture or loss of elasticity. To achieve this, a highly flexible and strong material is required. The
flexibility of any material is largely dependent on its Young’s
modulus [8]. Therefore, the material strength-to-Young’s
modulus ratio, termed the modulus of resilience, and the
material weight are the material properties considered.
Plexiglas has a resilience which is higher than most engineering
materials given in Table 1. It has the lowest density value
amongst the materials. Hence, Plexiglas is used for the gripper material [25].
Table I
Physical properties of common engineering materials
Material Strength
(MPa)
Young’s
Modulus
(MPa)
Resilience Material
density
(kg/m3)
Plexiglas 69 2.8 0.025 1119
Steel 1641 207 0.0079 8050
Aluminium 503 71.7 0.007 2700
E-glass 1640 56 0.029 1900
Plexiglas is a low cost material (The cost of a single 2X1.35m
sheet of 3mm thickness is about $60.00). Besides this, it possesses good machinability [26], [27] by laser cutting hence
allowing low cost fabrication process. Finally, it is readily
available in the market thereby making it easy for the industry
to reproduce the gripper. Plexiglas is a common material
available in the local, global and online market. It comes in
different standard thicknesses of 1, 3, 6 and 10mm, making the
thickness choice process easy.
C. Geometrical Analysis
In this section, every aspect of the gripper is specially designed
in order to achieve the design criteria stipulated from 3 to 7.
Nonetheless, trade-offs were made for other criterion so as to
maintain the strength of other implemented measures, as will be
discussed in the upcoming subsections.
1) Fingers/tips
Based on criteria 3, the dimensions of the fingers are selected,
as shown in Fig 1. Kiwifruit has a varying range of radial
diameter of 30mm – 70mm [28]. A short length of 40mm was given to the fingers to avert high finger deformation leading to
corresponding reduction in grasp firmness [29].
The gripper is a normally open gripper. The distance separating
the two fingers depends upon the object of manipulation and the
gripping stroke of each finger. A distance of 52mm exists
between the two fingers when they are in repose, which is
roughly the average diameter of kiwifruit. With the addition of
the circular force sensor provision, the distance between the
fingers becomes 46mm.
The parallel orientation of the gripper portrays the translational
movement expected of the gripper. The fingers are rectangular
shaped, with the same thickness as that of the whole gripper. Therefore, the grasped object is solely clenched by friction
retention during manipulation [30].
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 111
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
Fig. 1. Modelled gripper with flexure joints for macroscale grasping
2) Gripper active elements
The angle of inclination which the active elements make with
the transverse axis of the gripper is essential to increase the displacement amplification factor (criteria 4). When the angle
of inclination from the transverse axis is small, more planar
motion of the fingers towards each other is achieved. This is
because the x-component of the applied force, Fcosθ, is at an
elevated margin when angle θ is small. In this case, the elements
are inclined at a 45° angle.
3) Flexure joints
Flexure joints either produce lumped or distributed compliancy.
This can be depicted in Fig 2. The monolithic gripper exploits
on lumped compliancy more than distributed compliancy in the
ploy to achieve maximum deformation in the material structure.
The choice of notch type selected was based on the attainable motion range and associated stress concentration, the
compactness of the joint, robustness and the off-axis stiffness
[31]. Circular notch type hinges were adopted as its
characteristics were most suitable in this situation.
The stiffness equation of a joint was adopted from [32]. The
variables for the dimensions can be determined from Fig 3.
𝐾𝑏 =2𝐸𝑏𝑡2.5
9𝜋𝑟0.5 (1)
Hence, each joint has a stiffness of 6.3 N/m. Criteria 4 and 5
were regarded for determining the optimum joint thickness
befitting this design.
Fig 2: (a) lumped compliance (b) distributed compliance [9]
Fig. 3. Variable assignment of joint dimensions [32]
Subjecting only one joint to stresses required to attain the
desired displacement would be irrational. Hence, generating
multiple active flexure joints along the gripper links results in
stress and deformation distribution amongst the joints. The
distribution profile, however, is somehow unevenly arrayed and
it depends on the number of joints used, separation distance,
and the type of joint. This design employed 8 joints for the inner
links and 9 for the outer links.
4) Gripper links
The length, angle and nature of the links are fundamental
features to promote huge deformations. The links are populated with multiple flexure joints to further maximize displacement
(criteria 5). As a way to ensure the parallel translational motion
of the gripper fingers during gripping and releasing the kiwi and
hence strategic dexterity of grasping action, a parallelogram
mechanism using a 4 bar linkage was used (criteria 6). The
mechanism enhances planar motion in one direction, thereby
inhibiting any imminent toggle movement.
The stiffness of the structure (K) and the moment (M)
encountered is directly associated with the length (L) of the
member. This is based on (2) and (3).
𝐾 =𝐸𝐴
𝐿 (2)
𝑀 = 𝐹𝐿 (3) The longer the link, the lesser the stiffness translating to more
deformation. Hence from (2), increasing L makes K small. E
and A is material’s Young’s Modulus and cross sectional area
respectively. From (3), increasing L makes M bigger, with force
F being a constant. The moment is directly proportional to
length, which translates to more deformation as well. Links of
112mm vertical dimensions are used. Each set of links forming
a parallelogram was brought closer together to increase both
stiffness and deflection. Because the links are so close together,
they operate almost as a single thick layer hence making it
stiffer. The links are separated by a distance of 5mm.
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 112
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
The links incline at an angle of 65° to the transverse axis and
69° inner angle to the active elements. Maximizing the
capitalization of these angles was impeded by the set distance
between the fingers and the size and shape optimization of the
overall gripper.
5) Actuator
The actuator used was chosen based on the level of control to be implemented, the cost, and the anticipated input and output
displacement. With the need for precision gripping and force
control in order to ensure meticulous fruit gripping without
affecting the integrity of the fruit by damaging it, as well as
softness determination of the fruit, an electric linear actuator is
used [33]. A miniature linear motion actuator from Firgelli
Company is used, having a 50mm stroke, a step size of 0.1mm
and a gear ratio of 256:1. The mode of operation is either
through a computer or through pulse width module. All these
modes require a power source of 3.3V for the operation of the
actuator. The actuator bracket is developed for input force transfer to the
gripper. It is an interface between the actuator and the gripper
itself. Its connections does not constrain any movement of the
gripper’s active elements. It is fashioned to match the type of
actuator used.
6) Force sensing and control
Based on criteria 7, a force sensor is used to make the gripper
smart by becoming capable of determining the force being
applied and consequently the softness of the fruit. This
simplifies the sorting process. The force sensor is aligned along
the profile of the fingers, being placed on an 18mm cylindrical
disc as shown in Fig 1. Force-sensitive resistors (FSR) will be used. Their resistance changes in proportion to the amount of
applied pressure. It is of low cost and measure forces ranging
from 0-100N, which is within the operating range of the fruit
gripping.
A combination of the linear actuator (LAC) control board and
Arduino microcontroller are used for force and positioning
control. From the applied force between the fingers and the
fruit, the resistance of the FSR decreases. The resistance is
changed into voltage for the purpose of producing an analogue
input to the controller. This feedback signal determines whether
the actuator should stop, actuate more or retract.
Fig. 4. LAC control board and Arduino Uno microcontroller combination
forming the gripper control board.
D. Boundary and Loading Conditions
Rotational motion has been inhibited. The fingers can only
operate in one plane. The orientation of approach of the gripper
to the object of manipulation is threefold:
Vertical Gripper - Horizontal Fruit
Vertical Gripper - Vertical Fruit
Horizontal Gripper - Vertical Fruit
These orientations are as shown in Fig 5 for both flat and v-
shaped fingers.
Fig. 5. Various object – gripper orientation during grasping[34]
The base is stationary and the links and the active elements are
dynamic. The actuating force is exerted on the active elements
through the actuating point, which then transmit the force to the
links and eventually the fingers. The retraction motion of the
actuator causes a closure of the gripper fingers, thereby
gripping the fruit. Actuation causes the gripper to relax its grasp
on the fruit and hence dropping the fruit after manipulation. The
force on the fruit is enough to prevent slippage and fruit damage
simultaneously. A slight load due to the weight of the fruit is experienced inside the inner profile of the fingers, and is
accounted for during simulations.
E. Kinematic Analysis
To analyse the kinematics of the gripper, the Pseudo-Rigid-
Body Model (PRBM) is implemented[8]. The dynamic and kinematic analysis for force determination were adopted from
[34]. The analysis involving displacement considerations and
stress evaluation were adopted from [14], [35]. The model
assists in determining the grasping force necessary to hold the
object (fruits) in place during manipulation. The gripper
mechanism for the modelled gripper in Fig 1 is simplified in Fig
6. A classical approach of static force analysis is adopted for the
purpose of determining the relationship between the actuation
force and the output gripping force. The same approach is used
to determine the reaction forces at the 4 major joints of the 4
bar linkage. Only the 4 major corner joints reaction forces are
determined because they are the ones facing the most forces and stresses. Hence, the links are considered primarily as producing
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 113
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
distributed compliance. The gripping mechanism in Fig 6 is set
to determine the input-output/P-F force relationship.
Fig. 6. Gripping mechanism of the modelled flexural gripper
Considering vector force equilibrium at point c;
�̅�𝐿 + �̅�𝐿 + �̅� = 0 (4)
𝐹𝐿 =𝑃
2 sin𝛾 (5)
FL is the resultant force from the actuator input force P acting
on either link. Hence, considering one side of the gripping
mechanism due to symmetry principle, and applying the
moment equilibrium about point a;
𝐹
𝐹𝐿=
𝑋𝑖
𝑋𝑜 (6)
Letting the pivot distance ratio Xi/Xo be X, F-P ratio becomes
X/2sin γ.
𝐹
𝑃=
𝑋
2 sin𝛾 (7)
𝐹 = 𝑅1𝑅2𝑅3𝑚𝑔 (8)
Where F is the gripping force
R1 is factor of safety (ranges from 1.2 - 2.0)
R2 is 1+ A/g (A is object acceleration during manipulation) R3 is transfer coefficient of gripper
Therefore, for this gripper, the tensile strength is 69MPa, and
the allowable applied stress shall be 46MPa, hence giving a
factor of safety of 1.5. The least possible factor of safety has
been chosen because Plexiglas is not a highly compliant
material. So allowing for the highest possible flexibility will
ensure more displacement.
An arbitrary maximum object acceleration of 0.04m/s2 is used.
This is despite the fact that the fruits being grasped by the
designed gripper shall be stationary.
Hence, R2 =1.0041
The transfer coefficient for the gripper orientation of;
1. Horizontal gripper, vertical object
2. Vertical gripper, vertical object
These orientations demand a transfer coefficient of 1
2𝜇 [34],
where μ is the friction coefficient. Making use of the Amonton-Coulomb friction model;
R3 = 1.25
Therefore, with a maximum weight of kiwifruit of 0.12kg [36],
the necessary grasping force during manipulation is 2.2N.
All the links for the 4-bar mechanism of the gripper on each
side reacts to the force exerted FL. The outer links also
experience a torque as a result of the force input.
Fig. 7. Detailed free body diagram of the RHS 4-bar linkage
Force FL = 3.4707N, XT = 112mm, which is the distance from
the base to the upper links, and Xi = 96.08mm, which is the
distance from the base to the point of action of the force. The
4-bar mechanism on each side of the gripper is depicted in Fig 7. Links 1 and 3 are parallel, and so are links 2 and 4. The
linkage is exploded to reveal the forces on all links as shown in
Fig 8.
Forces acting on link 4; FL, F34 and F14.
Forces acting on link 3; F23 and F43.
Forces acting on link 2; F32 and F12, with torque T.
Forces acting on link 1; F21 and F41.
By the graphical method of virtual work application to static
force analysis, and also observing equilibrium and
superposition principles;
F43 = F32 = -F34 = -F23 = -F12 = -2.457 N
F14 = 3.032 N Torque T = 275.184 Nmm (clockwise)
The linear contribution of each joint to the displacement of the
fingers is derived from the angular displacement of that
particular joint, a sequel to the force applied to the member or
the moment it generates. The total linear distance travelled
either is calculated from the total angular distance or plotted in
CAD design and measured. Due to the length and complexity
of these equations which are given in [14], most of the
calculations where done using MATLAB software.
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 114
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
Fig. 8. Exploded view of the 4-bar linkage showing all reaction forces
The active elements are inclined at γ = 134° to the transverse
axis.
Since FL = 3.4707 N. the two symmetric elements experience
the same force. From [37], analysis of flexure joints is done to
determine the joints’ capacity and precision of displacement.
The analysis for circular joints is done as well. The compliance
of the joints are a major contributor to the displacement
achieved overall. The equations governing the in-plane
compliances are given in [37]. The static displacement
equations for flexure mechanisms are provided. In the case of
the designed flexure jointed gripper in Fig 1, whilst considering
one symmetrical half, each of the flexure jointed links is a planar serial compliant mechanism, the two of them are a
parallel compliant mechanism, and considering the whole 4-bar
mechanism is therefore a planar hybrid compliant mechanism.
The output displacement is given by;
𝑈𝑥 = −𝐹 ∑ 𝐶𝑥2𝑖𝑛2𝑖 − [∑ 𝑙𝑖
𝑛2𝑖 (𝐹𝑇𝑙𝑖)(𝐶𝜃2𝑖)] (9)
Where i is from 0 to n, and n is the number of flexure joints
available. F is the force component acting on the inner link in
the x direction = 2.411 N. T is the torque acting on the outer
link. Cθ is the in-plane compliance for rotational movement, and
Cx is the in-plane compliance for horizontal translational
movement. The distance from which the displacement is being
taken from and the flexure joint contributing to the displacement is designated li. The compliance of all the joints
is the same since the gripper is a monolithic structure fabricated
from a common material in addition to them having the same
following properties: symmetrical joints, same radius of
construction, same thickness, same width, and same angle of
inclination. Joint displacement summary is given in Table 2.
As expected, the lowest flexure joints at the base area
experience more displacement than the ones above them.
Furthermore, the displacement, though unequally shared, is
distributed across all the joints, which satisfies the design
requirements. The inner links of the parallelogram setup evidently experience a slightly higher deformation than the
outer links because they are directly linked to the applied force.
The total displacement per finger amounts to 10.76mm.
Table II
Translational displacement and joint location contribution for each joint
Joint
Number
Outer
Distance
from F (mm)
Inner
Distance
from F (mm)
Outer
Displ
(mm)
Inner
Displ
(mm)
1 18.66 18.66 0.1872 0.1872
2 1.11 14.31 0.0097 0.1388
3 13.04 27.9 0.1253 0.2993
4 26.63 41.5 0.2832 0.4879
5 40.23 55.09 0.4691 0.7042
6 53.82 68.69 0.6828 0.9486
7 67.42 82.28 0.9246 1.2208
8 81.01 92.02 1.1942 1.4330
9 93.42 - 1.4647
F. Gripper Prototype and Development
1) Model
After the implementation of the design criteria, dimensions
and various orientations were realized. A 3D model was
designed, as shown in Fig 1.
2) Fabrication
Subtractive manufacturing using laser machining was
employed. 3mm thickness Plexiglas was used. The fabricated
prototype is shown if Fig 9.
The actuator and the actuator bracket have been installed and
fastened successfully.
Fig. 9. Fabricated gripper prototype with auxiliary components
III. SIMULATION AND EXPERIMENTS
A. Simulation
The allowable stress for the gripper elements is 46MPa.
However, the stress values experienced in the gripper are kept
further away from this nominal value so as to increase the
number of cycles the gripper can make in its lifetime.
COMSOL platform was used to conduct simulations. The
maximum input force was used to initiate the simulation, and
all boundary conditions observed. The simulation was done for
both actuation and retraction motions in order to grasp the full
extension of the fingers. Plexiglas was chosen during material
selection. The solid mechanics physics was used and a
stationary study conducted. The results of the study reflect how
each part of the gripper reacts to the applied force. This reaction
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 115
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
is shown in terms of the displacement achieved, stresses
involved and strain. All these parameters are shown against
time.
Subsequently, a stress-based fatigue analysis is conducted on
both ANSYS and COMSOL, in order to get the number of life
cycles and usage factor of the gripper. These are conducted with
the assumption that the gripper is to be operated at high
intensity levels of repetition and finger extension. The Wohler
S-N curve revealing the alternating stress amplitude versus the
number of cycles to failure for Plexiglas was adopted from
[38]. A mechanism with 1 - 10 000 cycles before failure is
considered to have low cycle fatigue. 10 000 - 100 000 cycles
is considered to be a high cycle fatigue mechanism. The usage
factor indicates whether the stress limit has been exceeded and
failure is expected or if the component will hold for the
expected fatigue life.
B. Experiments
After fabrication, the first stage of experiment was to determine
the maximum range of displacement the fingers are able to
attain during actuation and retraction motions. With a known
initial finger separation distance of 49mm, the separation
distance is measured by a ruler after actuation and retraction.
The second part of experiments involves using the gripper to
manipulate kiwifruits from one point to the other. The forces
involved during this process are monitored to see if they
correspond to the expected values based on the fruit’s size and
softness. The gripper is used to move kiwifruits between 2 points half a metre apart. Three different kiwifruits were tested
during this experiment, and the process was repeated on each
fruit 3 times. The tested fruits varied in size and softness in
order to check whether these parameters affect the applied force
during manipulation, and to what degree. The experiment
comprised of the following activities;
i. Gripper approaches the fruit
ii. Gripper positioned with fruit in between the fingers
iii. Gripper grasps the fruit
iv. Gripper moves the fruit from Point A to Point B
v. Gripper is well positioned at point of release
vi. Gripper releases fruit
When the gripper is perfectly positioned, having the kiwifruit stationed between the gripper fingers, the retraction motion of
the actuator forces the fingers to approach the fruit. After
contact has been established, the fingers apply force to the fruit
as the fingers continue closing up. The reaction force is
measured by the FSR sensor, processed through Arduino Uno
and displayed on a computer. The fruits are radially grasped for
uniformity purposes. Step size input is given to the actuator
until the fruits are firmly grasped.
The respective properties of the fruits used for experimentation
are given in Table 3. The dimensions in bold are the dimensions
of the longer radial axis which was considered for all the experimental grasping operations.
Table III
Mechanical properties of kiwifruit used for experimentation
Fruit Dimensions
(mm)
Weight
(g)
Nominal Force
(N)
1 53 x 44 x 55 110.2 2.30
2 51 x 46 x 54 105.4 1.90
3 53 x 46 x 55 112.0 2.35
The third phase of experiments involves fruit softness detection. The determination of the softness of the kiwifruit is done by
finding the stiffness of the fruit during grasping and
manipulation. During the grasping of kiwifruit, the gripper
applies a force on the kiwi surface in an attempt to grasp the
kiwi. The kiwi deforms due to this load. The level of
deformation is based on the following parameters;
i. Kiwi softness
ii. Magnitude of the force
iii. Gripper stiffness
iv. Diameter of the contact surface
Gripper fingers also deform and based on the same principles.
Only one half of the gripper is considered due to symmetry. The
free body representation of one half of the gripper shown in Fig
7 is given in Fig 10, with K1 and K2 being stiffness variables for
the kiwifruit and the gripper respectively. This representation is
due to the arrangement shown in Fig 11. Since one half of the
gripper is being considered, the LHS fixed end is due to the fact that no movement is possible for the kiwifruit to the left due to
the equal and opposite force exerted by the other gripper finger.
Force F is between the kiwifruit and the gripper finger and its
mechanism. The RHS fixed end is due to the base attachment
of the links.
Fig. 10. Free body diagram of one half of the gripper during grasping action
𝐹 = 𝐾𝑒𝑞𝑥 (10)
𝐾𝑒𝑞 = 𝐾1 +𝐾2 (11)
K2 is obtained from analytical and simulation results
Analytical: 1.235649 N/mm
Simulation: 1.217778 N/mm
Therefore, the stiffness of the kiwifruit is given by;
𝐾1 =𝐹
𝑥− 𝐾2 (12)
The force (F) is gotten from the FSR sensor reading.
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 116
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
Fig. 11. Flexural gripper arrangement with kiwifruit showing supports and
loading conditions
The displacement (x) is retrieved from the relationship which
exists between the control percentages of the actuator
configuration utility software, the actuator displacement, and
the output displacement at the fingers. The range of
displacement values for which the gripper operates within is
tabulated and shown in Table 4. The displacement used is half
the computed distance due to one-side symmetry considerations in effect.
Table IV
Input-output displacement relationship comparison against control percentage
input
control
percentage
(%)
actuator
stroke (mm)
Finger
separation
distance
(mm)
proportional
finger
separation
distance (mm)
25.0000 23.000 42.000 42.000
27.1875 24.000 43.422 43.250
29.3750 25.000 44.844 44.500
31.5625 26.000 46.266 45.750
33.7500 27.000 47.688 47.000
35.9375 28.000 49.110 48.250
38.1250 29.000 50.532 49.500
40.3125 30.000 51.954 50.750
42.5000 31.000 53.376 52.000
44.6875 32.000 54.798 53.250
45.0000 32.143 55.000 53.429
46.8750 33.000 55.819 54.500
49.0625 34.000 56.840 55.750
51.2500 35.000 57.861 57.000
53.4375 36.000 58.882 58.250
55.6250 37.000 59.903 59.500
57.8125 38.000 60.924 60.750
60.0000 39.000 62.000 62.000
In order to have a glimpse on the true meaning of the
determined stiffness value for the kiwifruit, the value has to be
compared against already known values which translates to the
softness of the fruit. Hence, a non-destructive radial
compression test was conducted using a Universal Testing
Machine (UTM) on kiwifruits of varying sizes, weight and
softness. Fig 12 shows a compression test being conducted on
one of the kiwifruits. The experimental readings are obtained
after the probe contacts the fruit’s surface. The experimental
readings are time, displacement and force applied. This information was used to determine the stress, strain and
eventually the modulus of elasticity (E) of the various
categories of softness. An extract of the relevant Young’s
modulus is shown in Fig 13. The Young’s modulus is then used
to compute the stiffness of the fruits, which is the set
comparable platform for softness detection.
Fig. 12. Compression test on kiwifruit using a UTM
Prior to the actual experiments, preliminary procedures involve
knowing the contact area of the probe, the radial diameter and
weight of the fruits to be subjected to experiments. The
computed parameters which are given by the UTM are the
applied force and the stroke of the probe, both of which are obtained against time. This in turn leads to a series of calculated
parameters of stress, strain and consequently the Young’s
modulus. With knowledge of the Young’s modulus, the
stiffness K1* is realised using (13) [39].
𝐾1∗ =
2𝑎𝐸𝐶𝑘
1−𝑣2 (13)
Where;
a is the radius of the contact probe, and is made to be 5mm for
K1* consistency.
ν is the poisson’s ratio of kiwifruit (0.33).
Ck is a factor of scaling that is dependent on the radial length of
the kiwi, achieved deformation, and the poisson’s ratio of kiwi.
For this analysis, the value of Ck is taken to be 1.
Because the maximum output force of the gripper is very small (2.2N), the considered computed values have a 4N set
benchmark as the upper limit.
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 117
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
Fig. 13. Stress-strain graph for hard, soft and partially soft kiwi over a 4N
span.
From Fig 13, all the curves have not yet attained the degree of
linearity expected from the elastic range of any stress-strain
graph, which it eventually achieves at slightly higher force
values. This is attributed to lack of full contact at those forces,
probe stability and fruit surface conditions. The curves also
show how the Young’s modulus differs with varying softness
levels. However, because the Young’s modulus is a factor of stress,
which is an attribute of force, fruits having the same softness
level might have different elasticities recorded as a result of the
force variation applied to each. This is due to the fact that
smaller fruits require less force application during
manipulation. This can be seen in Fig 14, where a single
modulus can overlap into 2 different fruit size categories.
Therefore, the elasticity is realised according to sizes.
The fruit sizes are on the basis of their weights.
Small 75g and less
Medium-size 75g – 100g Large 100g and above
Fig. 14. Summary of Young’s Modulus of kiwifruits of varying sizes and
softness
The softness index is placed into 3 categories; hard, soft and
partially soft. 9 different kiwifruits were tested, 3 per each
category in order to cater for all sizes. The obtained results are
summarized in Fig 14.
IV. RESULTS AND DISCUSSION
Simulation results for stress and displacement are shown in Fig
15 and Fig 16. The 3D safe stress profile is given in Fig 15. The
heat map beside the structure indicates the extremities of stress
concentration, with the maximum stress regions being indicated
by red colour, and the least stressed regions are in blue colour.
The maximum stress regions lie at the top flexure joints. This is
because in the parallelogram setup, those are the joints
responsible for maintaining the translational motion of the
gripper throughout the deformation process. After grasping, the
uppermost joints will experience reaction forces as well. Hence,
in the event of failure, the gripper is likely to fracture in those
areas. The lower joints also experience greater stress, and the
stress deteriorates as one ascends up the links until it reaches its
lowest joint stress at the flexure joint closest to the active
elements. The maximum stress is 23.7MPa, which is slightly
higher than half the allowable stress. Hence, the gripper
operates in a safe region.
The 3D safe displacement profile is given in Fig 16. The colour
bar beside the gripper is a deformation indicator, with the
maximum deformation being indicated by red colour, and the
least deformation being shown blue colour. As desired, the
maximum deformation is experienced by the gripper fingers.
Fig. 15. Simulation results for stress analysis
Fig. 16. Simulation results showing displacement distribution within the
gripper
The fatigue analysis in Fig 17 reveals the number of cycles the
0
0.002
0.004
0.006
0.008
0.01
0.012
0.014
0.016
0.018
0 0.005 0.01 0.015
Stre
ss (
MP
a)
Strain
hard partially soft soft
0.5
1
1.5
2
2.5
3
3.5
small medium largeYou
ng'
s M
od
ulu
s, E
(M
Pa)
Kiwifruit Size
soft kiwi P.soft kiwi hard kiwi
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 118
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
gripper has to undergo before failure can be expected. The
fatigue analysis done revealed the gripper mechanism can go
to cycles of 104 magnitude if it is to be operated at the
maximum force of 5N throughout its lifetime. Therefore, the
gripper is a high fatigue mechanism as long as it is operated
within the confines of the stipulated stresses. Fig 18 shows the
fatigue usage factor of the gripper. The gripper shows a usage
factor of 0.6, which is way below 1. Therefore, the stresses in
the gripper during its operations are far from the stress limit
the mechanism can take.
Fig. 17. Stress-based fatigue analysis of the modelled gripper showing number
of cycles to failure
Fig. 18. Stress-based fatigue analysis of the modelled gripper showing the
fatigue usage factor
The first experiment shows that the gripper displacement range
is 20mm. This displacement is enough to grasp kiwifruit. A
summary of displacement results is given in Fig 19 realised
through various methods.
Fig. 19. Graph of comparison between experimental, CAD, FEA and
analytical maximum output displacement.
Results of the second phase of experiment shows how the force
obtained during the manipulation process varies with time. This
is plotted in Fig 20, Fig 21 and Fig 22 for fruit 1, 2 and 3
respectively.
The expected curves reveal how the grasping process would be
like if the required force is arrived at instantaneously, the manipulation is conducted without any fluctuations, and the
fruit dropped instantly. The maximum force for the expected
curves show the maximum force required to translate a solid
object of that particular weight without causing damage to the
fruit. It can be seen from Fig 20 that the curve for fruit
manipulation is slightly lower than the ideal case. This reveals
that the force needed to successfully complete the manipulation
process is lower, signifying that the fruit was slightly softer and
not solid hard. Fig 21 shows a curve which is basically in the
same range of maximum force as the expected case. This
indicates that the fruit was translated without any damaging
effects on the fruit. The graphical results shown in Fig 22 are for an experiment conducted on a very soft kiwifruit. The peak
force is largely lower than the expected case.
Fig. 20. Manipulation process of fruit 1 between 2 points half a metre apart
Fig. 21. Manipulation process of fruit 2 between 2 points half a metre apart
109.26
11.88 10.76
0
2
4
6
8
10
12
experimental simulation CAD analytical
0
0.5
1
1.5
2
2.5
3
0 20 40
Forc
e (N
)
Time (S)
experiment
expected
0
0.5
1
1.5
2
2.5
0 20 40
Forc
e (N
)
Time (S)
experiment
expected
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 119
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
Fig. 22. Manipulation process of fruit 3 between 2 points half a metre apart
From Fig 23, 24 and 25, experimental results for softness
detection are shown. The stacked shaded areas show the region
which the Young’s modulus of a particular kiwifruit size fall,
which is merely a disintegration of columns shown in Fig 14.
E1, E2 and E3 represent the 1st, 2nd and 3rd experimental
repetition for any given fruit being experimented on. E ideal
shows the Young’s modulus of the particular fruit in question if
the actual force which should be used to manipulate the fruit is
achieved. However, E ideal doesn’t normally reflect the desired
elasticity because the force used does not take in consideration
the softness of the fruit. As anticipated, the each fruit size falls within the region of its softness, and all subsequent experiments
(2nd and 3rd) likewise. E ideal for a small partially soft kiwi in
Fig 24, however, does not fall in its region. It falls in the region
of a medium sized kiwi rather than in the small kiwi region.
This goes to prove that knowledge of the amount of force
applied alone is insufficient to conclude the level of softness of
fruits, which is the reason why softness detection was initiated.
Fig. 23. Young’s Modulus experimental determination of soft kiwifruits
Fig. 24. Young’s Modulus experimental determination of partially soft
kiwifruits
Fig. 25. Young’s Modulus experimental determination of hard kiwifruits
V. CONCLUSION
The analytical design of the gripper reveal how applicable the
model can be at macroscale. The total displacement achievable
in each finger is approximately 10mm when considering both
extension and retraction actuator motions, which is sufficient to
conduct successful conventional grasping operations. With
such a huge displacement for flexure joint type grippers, middle
sized fruits can be grasped and manipulated successfully
without any friction or backlash occurring within the gripper
structure. The achieved parallel motion of the fingers by
implementation of the 4-bar linkage makes the grip on the fruits
more secure, thereby preserving the quality of the fruits during
the manipulation process. The stress factor of the gripper is kept
below half the allowable stress of the gripper structure material.
The simulation results were closely approaching the analytical
results. It can be seen, however, in the simulation results that a
slight deflection in the y-direction occurs. The magnitude of
this displacement at the finger area is less than 1 mm (the
maximum is 0.9mm). Therefore at macroscale, this
displacement is small enough to cause any significant rotational
movement during grasping hence, negligible. Softness
0
0.5
1
1.5
2
2.5
3
0 10 20 30 40
Forc
e (N
)
Time (S)
experiment
expected
0.6
0.8
1
1.2
1.4
1.6
1.8
1 2 3
You
ng'
s M
od
ulu
s, E
(M
Pa)
Fruit Number
small size medium size large size E ideal E1 E2 E3
1.1
1.3
1.5
1.7
1.9
2.1
1 2 3
You
ng'
s M
od
ulu
s, E
(M
Pa)
Fruit Number
small size medium size large size E ideal E1 E2 E3
1.11.31.51.71.92.12.32.52.72.93.1
1 2 3
You
ng'
s M
od
ulu
s, E
(M
Pa)
Fruit Number
small size medium size large size E ideal E1 E2 E3
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 120
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
detection has been incorporated as part of the gripper’s
functionality as a preservative quality measure of the kiwifruits
during manipulations. After successive experiments, the
gripper has shown tremendous accuracy in determining the
softness of grasped fruits and by making use of the softness
index. The softness index is based on the Young’s modulus of
the fruits, which is compared against a developed scale from
compression test data. Future work can involve the use of a
displacement sensor in conjunction with the FSR sensor to
further enhance the precision of softness detection. The gripper
is open to more optimization techniques and improvements.
ACKNOWLEDGEMENT
The first author would like to thank Pan African University
for the scholarship of studying at the institution and provide the
platform for higher education. The same gratitude should
extend to Egypt Japan University of Science and Technology
(E-JUST), MEMS Lab, for allowing the first author to conduct
experiments and other research procedures at their institution.
They opened up their staff and the resources for the purposes of
this research.
REFERENCES [1] Y. Tsai, S. H. Lei, and H. Sudin, “Design and analysis of planar
compliant microgripper based on kinematic approach,” J.
Micromechanics Microengineering, pp. 143–156, 2005.
[2] A. Nikoobin and H. M. Niaki, “Deriving and analyzing the effective
parameters in microgrippers performance,” Sci. Iran. B, vol. 19, no.
6, pp. 1554–1563, 2012.
[3] S. K. Nah and Z. W. Zhong, “A microgripper using piezoelectric
actuation for micro-object manipulation,” Sensors Actuators, A
Phys., vol. 133, no. 1, pp. 218–224, 2007.
[4] A. A. Escriva, “Design of a Smart Gripper for Industrial
Applications,” Tampere University of technology, 2016.
[5] T. Bock, J. Herbst, C. Balaguer, and M. Abderrahim, “Design of a
Gripping System for the Automated Assembly of Large Building
Modules,” in Proceedings of the 17th IAARC/CIB/IEEE/IFAC/IFR
International Symposium on Automation and Robotics in
Construction, 2017, pp. 1–8.
[6] C. I. Basson, G. Bright, and A. . Walker, “Testing Flexible Grippers
for Geometric and Surface Grasping Conformity in Reconfigurable
Assembly Systems,” South African J. Ind. Eng., vol. 29, no. May,
pp. 128–142, 2018.
[7] M. Verotti, A. Dochshanov, and N. P. Belfiore, “A Comprehensive
Survey on Microgrippers Design: Mechanical Structure,” J. Mech.
Des., vol. 139, no. 6, pp. 1–46, 2017.
[8] L. L. Howell, Compliant mechanisms. Wiley, 2001.
[9] J. A. Gallego and J. Herder, “Synthesis Methods in Compliant
Mechanisms: An Overview,” in Proceedings of IDETC/CIE ASME
2009 International Design Engineering Technical Conferences and
Computers and Information in Engineering Conference, 2010, pp.
193–214.
[10] S. Krishnan and L. Saggere, “Design and development of a novel
micro-clasp gripper for micromanipulation of complex-shaped
objects,” Sensors Actuators, A Phys., vol. 176, pp. 110–123, 2012.
[11] I. Q. Meng, “A Design Method for Flexure-Based Compliant
Mechanisms on the Basis of Stiffness and Stress Characteristics,”
2012.
[12] Y. S. Oh, H. E. Stephanou, and G. D. Skidmore, “Design,
Optimization and Experiments of Compliant Microgripper,” in
Proceedings of IMECE’03 2003 ASME International Mechanical
Engineering Congress, 2016, pp. 1–6.
[13] O. F. Linkage-type, “An Atlas of Linkage-Type Robotic Grippers,”
Mech. Mach. Theory, vol. 32, no. 7, pp. 811–833, 1997.
[14] F. Dirksen, “Non-intuative Design of Compliant Mechanisms
Possessing Optimized Flexure Hinges,” 2013.
[15] X. Li, W. Chen, W. Lin, and K. H. Low, “A Variable Stiffness
Robotic Gripper Based on Structure-Controlled Principle,” IEEE
Trans. Autom. Sci. Eng., vol. 15, no. 3, pp. 1104–1113, 2018.
[16] T. Shimizu, M. Shikida, K. Sato, K. Itoigawa, and Y. Hasegawa,
“Micromachined active tactile sensor for detecting contact force and
hardness of an object,” Sensors Actuators A, vol. 103, pp. 213–218,
2003.
[17] S. Terasaki, N. Wada, N. Sakurai, N. Muramatsu, R. Yamamoto,
and D. J. Nevins, “Nondestructive Measurement of Kiwifruit
Ripeness Using a Laser Doppler Vibrometer,” Food Process Eng.
Inst. ASAE, vol. 44, no. 1, pp. 81–87, 2001.
[18] I. J. Davie, N. H. Banks, P. B. Jeffery, C. J. Studman, and P. Kay,
“Non-destructive measurement of kiwifruit firmness,” New Zeal. J.
Crop Hortic. Sci., vol. 24, no. 2, pp. 151–157, 1996.
[19] F. R. Harker, J. H. Maindonald, and P. J. Jackson, “Penetrometer
Measurement of Apple and Kiwifruit Firmness: Operator and
Instrument Differences,” J. Am. Soc. Hortic. Sci., vol. 121, no. 5,
pp. 927–936, 1996.
[20] F. J. García-Ramos, C. Valero, I. Homer, J. Ortiz-Cañavate, and M.
Ruiz-Altisent, “Non-destructive fruit firmness sensors: a review,”
Spanish J. Agric. Res., vol. 3, no. 1, pp. 61–73, 2005.
[21] A. Fouly, M. N. A. Nasr, A. M. R. Fath El Bab, and A. A.
Abouelsoud, “Design and modeling of micro tactile sensor with
three contact tips for self-compensation of contact error in soft
tissue elasticity measurement,” IEEJ Trans. Electr. Electron. Eng.,
vol. 10, pp. 144–150, 2015.
[22] A. M. R. F. El Bab et al., “Design and Simulation of a Tactile
Sensor for Soft-Tissue Compliance Detection,” IEEJ Trans. Sensors
Micromachines, vol. 128, no. 5, pp. 186–192, 2008.
[23] A. M. R. F. El Bab, K. Sugano, T. Tsuchiya, O. Tabata, M. E. H.
Eltaib, and M. M. Sallam, “Micromachined tactile sensor for soft-
tissue compliance detection,” J. Microelectromechanical Syst., vol.
21, no. 3, pp. 635–645, 2012.
[24] C. T. Nnodim, A. M. R. FathElbab, B. W. Ikua, and D. N. Sila,
“Estimation of the Modulus of Elasticity of Mango for Fruit
Sorting,” Int. J. Mech. Mechatronics Eng., vol. 19, no. 2, pp. 1–10,
2019.
[25] “Physical Properties of Acrylic Sheets.”
[26] M. O. Helmy, A. M. Fath El-Bab, and H. El-Hofy, “Elimination of
Clogging in PMMA Microchannels Using Water Assisted CO2
Laser Micromachining,” Appl. Mech. Mater., vol. 799–800, pp.
407–412, 2015.
[27] M. O. Helmy, A. R. F. El-Bab, and H. A. El-Hofy, “Fabrication and
characterization of polymethyl methacrylate microchannel using dry
and underwater CO 2 laser,” Proc. Inst. Mech. Eng. Part N J.
Nanomater. Nanoeng. Nanosyst., vol. 232, no. 1, pp. 23–30, 2018.
[28] Kiwifruit. OECD, 2008.
[29] N. N. Mrewa, A. M. R. F. Elbab, and G. N. Nyakoe, “Design and
Development of a Monolithic Gripper with Flexural Joints for
Macroscale Manipulation,” in 2019 International Sustainable
Research and Innovation (SRI) Conference, 2019, pp. 1–7.
[30] G. Causey, “Guidelines for the design of robotic gripping systems,”
Assem. Autom., vol. 23, no. 1, pp. 18–28, 2005.
[31] E. Boudreault and C. Gosselin, “Design of Sub-centimetre
Underactuated Compliant Grippers,” in Proceedings of IDETC/CIE
ASME 2006 International Design Engineering Technical
Conferences and Computers and Information in Engineering
Conference, 2017, pp. 1–9.
[32] M. Liu, X. Zhang, and S. Fatikow, “Design of flexure hinges based
on stress-constrained topology optimization,” J. Mech. Eng. Sci.,
vol. 0, no. 0, pp. 1–11, 2016.
[33] “Miniature Linear Motion Series Datasheet,” 2016.
[34] F. Y. Chen, “Force Analysis and Design Considerations of
Grippers.,” Ind. Rob., vol. 9, no. 4, pp. 243–249, 1982.
[35] D. Kern, M. Rösner, E. Bauma, W. Seemann, R. Lammering, and T.
Schuster, “Key features of flexure hinges used as rotational joints:
Transition from concentrated to distributed compliances,” Forsch.
im Ingenieurwesen/Engineering Res., vol. 77, no. 3–4, pp. 117–125,
2013.
[36] “Standards for seeds, tractors, forest, fruit and vegetables - OECD.”
[Online]. Available: http://www.oecd.org/tad/code/. [Accessed: 16-
Feb-2019].
[37] N. Lobontiu, Compliant Mechanisms_ Design of Flexure Hinges.
CRC Press, 2002.
[38] S. Issa Salih, “Investigation of Fatigue and Compression Strength
for the PMMA Reinforced by Different System for Denture
Applications,” Int. J. Biomed. Mater. Res., vol. 3, no. 1, p. 5, 2015.
[39] A. Fouly, A. M. R. FathEl-Bab, M. N. A. Nasr, and A. A.
International Journal of Mechanical & Mechatronics Engineering IJMME-IJENS Vol:19 No:04 121
193304-7575-IJMME-IJENS © August 2019 IJENS I J E N S
Abouelsoud, “Modeling and experimental testing of three-tip
configuration tactile sensor for compensating the error due to soft
tissue surface irregularities during stiffness detection,” Meas. J. Int.
Meas. Confed., vol. 98, pp. 112–122, 2017.