A Mechanical Device for Harvesting Crocus Sativus (Saffron)


14 Νοε 2013 (πριν από 3 χρόνια και 6 μήνες)

73 εμφανίσεις

A Mechanical Device for Harvesting
(Saffron) Flowers

M. Ruggiu, A. Manuello Bertetto

Published in Applied Engineering in Agriculture Vol. 22(4): 491
498 ( Copyright

2006 American Society of Agricultural and Biological Engineers ).

Submitted for review in January 2005 as manuscript number PM 5736; approved for
publication by the Power & Machinery Division of ASABE in March 2006.

The authors are
Maurizio Ruggiu,
tant Professor in Applied Mechanics, and
Manuello Bertetto,
Full Professor in Applied Mechanics, Department of Mechanical
Engineering, University of Cagliari, Cagliari, Italy.
Corresponding author:
Ruggiu, Piazza d


09123, Cagliari, I
taly; phone:++390706755716;
fax:++390706755717; e
mail: ruggiu@dimeca.unica.it.

A prototype for harvesting Crocus Sativus (saffron) flowers was developed. The
device has a simple design and is capable of performing the cutting procedure with j
ust one
actuated degree of freedom. A cam driven by a revolute motor drags the flower towards the
cutting area where the flower is severed. The operating principle and some theoretical
analyses of the device are presented along with details of the mechanic
al design and the
results of laboratory tests. These are instrumental for developing the final design, its
geometry, kinematics and kinetics. Finally, field tests are described.

Mechanical harvesting, Planar kinematics, Floral crops.

A number o
f devices are currently being used as an aid to, or even a replacement of, human
workers for harvesting crops (Petrucci et al., 1983; Manfredi, 1990; Giametta, 1992; Kondo
and Monta, 1999a; Pilarski et al., 2002). Flower crops are almost always harvested m
regardless of whether they are grown in a greenhouse or in a field. Still, some innovative
prototypes have been proposed for picking different types of flowers (Williames, 1986;
Savoia, 1985; Melidis and Vatterott, 1986; Kondo and Monta, 1999b; Va
lk and Vos Marinus,
2003). This work is generally laborious because flowers are structurally delicate and small in

For the case of
Crocus Sativus
(saffron) flowers, value of the powder extracted from its
stamens makes development of mechanical devic
es worthwhile. In industrialized countries,
processed powder costs as much as

10,000/kg, owing to the fact that approximately
150,000 flowers are needed to produce 1 kg of the spice.

Crocus Sativus
(saffron) flowers must be harvested in the early hours
of the day and then
only during a 15

to 20
day period in November. Thus, the harvest requires a great number of
workers for a short period of time. The work is tiring and repetitive. The flowers have a short
stem, requiring pickers to bend over repeatedly
. To pick the flowers, workers have to locate
the flower to be picked, separate it from its leaves, then tear the flower by applying a slight
force between with their thumb and index fingers.

This article describes a mechanical device designed to harvest
Crocus Sativus
flowers. To facilitate mechanization, the operating kinematics of the device is different than
the kinematics of manual harvesting.

Operating Principle

Since mimicking human picking would have required a complex robot capable of locating the
flower, separating it from the leaves, and cutting the stem, an innovative and very different
picking process which is easily mechanized was proposed and is described

First, two components are drawn together, thereby seizing both the flower and its leaves
between them. Then a linear oscillation takes place causing the flower to be cut by a series of
torsional loads applied to the stem. During the process the leaves a
id the cutting owing to
their rough surfaces. The cutting process thus involves two steps: approach and oscillation
(fig. 1).

Figure 1. The kinematics of the cutting process: (a) approach step, (b) oscillation step.

Although the cutting procedure consi
sts of two distinct steps, only one actuator, and hence
only one degree of freedom is required making the device a very simple mechanism. The two
cutting steps were obtained because of the geometry chosen for the cam which, while
rotating, drags the flower

and its leaves towards a passive striker. The mechanism is shown in
figure 2.

Figure 2. Cutting phases using the proposed cam
striker mechanism.

The cam, actuated by a revolute motor, moves closer to the flower (fig. 2a).

As the spacing
decreases, the cam first touches and then drags the flower toward the torsion zone (figs. 2b
and c). Here, an oscillating motion cuts the stem by torsion (fig. 2d). Then, the flowers
blooms are harvested by intake. For this reason, an inlet

on the striker, connected to a small
exhauster, was designed and tested as described in a subsequent section.

The cam is formed by two curves, joined by a line. Its geometry can mathematically be
expressed as:




are the coordin
ates of a polar reference system located at the cam axis and
the spiral constant.

Equation 1a represents an Archimedean spiral curve which causes the flower to be moved to
the space between the rotating cam and the curved striker by reducing the spac
e between the
components. The arc shaped profile of equation 1b is the path by which the flower is moved
during the cutting process.

Kinematics and Kinetics

The ability of the Archimedean spiral curve in reducing the space left between the cam and
the st
riker while performing the approach step can be described geometrically as shown in
figure 3.

Figure 3. Details of the approach step: (a) initial, (b) actual.

For an arbitrary position ? of the cam, the space left becomes:


The distance between t
he cam and striker is:


The rate of distance change
can simply be computed by differentiating equation 2 with
respect to time:


Equation 4 shows that
is proportional to the rotational velocity ? of

the cam. Moreover,
it is to be noted that the cam is accelerating when moving from the resting state throughout
the approach step, thus reducing

Once the flower stem is held between the cam and the striker, the flower meets the arc
shaped prof
ile of the cam. In order to correctly locate the flower, throughout the oscillating
step a relative motion between the flower stem and the cam has to occur. Thus, the flower can
remain fixed in the position reached at the end of the approach step. In figur
e 4, P
the position that the flower should have reached if it were moving together with the cam that
is at the end of the spiral, whereas P is the position the flower is required to be in at right at
the beginning of the oscillating step.

igure 4. Oscillating step: polar coordinate reference system for the flower stem kinematics.

Assuming a moving frame of reference [ ? ,

] rotating with the cam, the velocity of the
flower can be expressed as:


where v
is the velocity of the flower relative to the [ ? ,

] frame of reference. Equation 5
may be satisfied by an opportune friction condition between the flower stem and the contact
materials of either of the cam or the striker.

For the purpose of selecting
an appropriate driving motor for the cam, a kinetics analysis was

Figure 5. Cam and striker free
body diagrams.

With reference to figure 5, where the mechanism was disassembled in order to indicate the
forces on each component, the equilibri
um with respect to each pivot is:


is the moment exerted by a preload spring,
are the components of the
force that cam and striker exert on each other,
I o
is the moment of inertia with respect to the
pivot axis,
is the motor to
rque. Sign of the moment of the tangential force T depends on
the direction in which the cam rotates. The sign of the inertial term,
, depends on the
sign of
. The relationship between T and N depends on whether friction condition


f s
represent the dynamic and static friction coefficients, respectively. Equation 7a
is valid when friction exists between the flower stem and the components surfaces, equation
7b may be used when the stem rolls on the surfaces with no slipping.

e the oscillating step is being performed, the cam motion may be modeled as a harmonic
function of time, with ? as the amplitude as follows:


Finally, by combining equations 6, 7, and 8, the driving torque
may be computed:


Equations 9a and

b contain four solutions depending on combination of signs. Figure 7 shows
the maximum values of C
, C
when realistic design parameters, listed in table 1, were
used in the calculation.

Table 1. Design parameters used into the C
, C


? (rad)

k (N/m)



L (m)



? (rad/s)


p /2








Figure 6. Motor torque vs. cam rotation.

Prototype and Design Details

Main Components

A cam
striker mechanism was made of 15
mm thick aluminum. The main parts of the device
are shown in figure 7.

Figure 7. Main components of the cam
striker mechanism: (a) cam, (b) striker, and (c) frame.

The frame (fig. 7c) was constructed with a rectan
gular opening to allow for the mounting of
the motor. Figure 8 shows the assembled cam
striker mechanism.

Figure 8. Cam
striker mechanism assembly.

Loading Bending
Torsion Spring

Since a contact force is required for cutting, a pre
loaded bending
torsion spring was
mounted on the mechanism. Figure 9 shows the mechanism with the spring.

Figure 9. Cam
striker mechanism with the bending
torsion spring.

The bending
torsion spring c
onsists of a 3
mm diameter L
shaped harmonic steel wire. The
axis of the spring is attached to the frame and is first parallel to the axis of the motor and then
bends through 90

, thus extending along the plane of the striker. The pre
loading torque is
rovided by rotating the whole spring with respect to the axis of attachment. Additionally, a
loading bending moment is accomplished as well, by moving a slider along the bent
portion of the axis of the spring.

Materials for the Friction Surfaces


was taken to choose the materials used on the cam and striker friction surfaces. The
main requirement for these surfaces was to allow the flower to move according to the
predicted kinematics. That is, the flower stem has to roll over a very small part of
Archimedian spiral and then slip on the arc
shaped part of the cam. Additionally, this
property must be insensitive to changing environmental conditions. An iterative trial
error procedure was used for selecting both the appropriate surface types,
done with different
values of torque derived from the bending
torsion pre
loaded spring. The surfaces finally
selected are shown in figure 10. They are a vinyl no
slip tape for the cam and neoprene
rubber tape for the striker. The static friction coefficie
nts between the surfaces were
approximately 1.2 or 1.0 depending on whether they were dry or wet. As shown in figure
10b, the surface on the striker has several slits to allow the water to quickly drain away.

Figure 10. Friction surfaces of the cam
ker mechanism: (a) cam, (b) striker.

Secondary Components

In order for the device to work properly, some secondary components were required as well.

First, the cam was equipped with a highly deformable rod, referred to as the finger, which
aids in the movement of the flower toward the torsion area. Second, a stop on the frame
prevents the finger from moving further once it has moved the flower into pos
ition. Figure 11
shows the prototype with the flexible rod, and a schematic of its operation.

Figure 11. Secondary components: (a) overview, (b) operating principle of the high
deformable rod.

Third, a stop was mounted on the striker in order to fix th
e striker in its rest position so the
cam and striker were in contact with each other only after the cam had been sufficiently

Laboratory Tests

Cutting Process Measurements

Tests were carried out to establish the types of friction surfaces, the

load and stiffness of
the bending
torsion spring, and other operating parameters. Depending on the season, saffron
flowers were not available during the laboratory tests, thus another flower type, having
mechanical characteristics similar to saffron,
concerning the cutting process, was selected.
The flower chosen was the viola x wittrockiana. When still young, these flowers seem to
have toughness and strength of the stem comparable to that of the saffron. Figure 12 shows a
photo of the experimental set

Figure 12. Experimental setup: (a) load cell, (b) rubber band, and (c) potentiometer.

Figure 12 shows a load
cell (a) which measured the force applied by the rubber band (b)
simulating the bending
torsion spring behavior. The rotary potentiometer (
c) measured the
angular position of the cam.

Results of the experimental tests are presented in figure 13.

Figure 13. Results of the laboratory cutting test.

Figure 13 shows the presence of a pre
load of the rubber band at the beginning of cutting.
en the rubber band tension follows an oscillating pattern towards the end. During the
oscillating step, the contact force
between cam and striker can be simply calculated by
recalling the equilibrium of the striker with respect to its pivot (eq. 6) and t
he relationship of
the contact and tangential forces (eq. 7a and b).


are the rubber band force and its moment arm respectively. A mean value of
about 10 N in case a or about 5 N in case b ( ? = 0.5) was obtained.

Flower Inlet Apparatus

Numerous tests were carried out to design and test the flower inlet apparatus. This consists of
an inlet placed on the striker connected by a tube to an exhauster. The tests were conducted
on a flower model made of paper, similar to

saffron flower both in terms of shape, i.e. petal
size, and of weight, as shown in figure 14a. The main value of mass was 0.3 grams for the
model and 0.5 grams for the real flower. The main value of the model surface area was 24 cm
whereas it was 22 cm
for saffron. Figure 14b shows the test rig for measuring the force
required to intake the flower.

Figure 14. Flower intake test: (a) flower simulator, (b) test rig.

As seen in figure 14b, the test apparatus consisted of the inlet (1) on the striker,
the flower
simulator (2) at one end of a wire connected to a load cell (3) which measured the force
needed to draw the flower. The load cell was connected to a spherical joint (4) in order to let
the flower move freely when drawn by the intake air. The tes
ts were carried out placing the
paper flower anywhere between cam and striker, then increasing the motor power by a linear
potentiometer and meanwhile collecting the data from the load cell. Figure 15 shows the
force as a function of the power consumption
of the exhauster motor.

Figure 15. Intake force vs. motor power.

From data collected it can be observed that the flower is sucked by the exhauster with 300 W
of the motor power. The corresponding force may be estimated similar to the flower weight
from the flower equilibrium. However, according to a visual observation o
f the flower
behavior when subjected to the aspiration force field a mean value of 40.7 mN with a
standard deviation of 4.4 mN was estimated to be certainly able to take the flower in.

Tests in a Field of Saffron Flowers

Following laboratory testing, the

device was tested in the field. This was done to test the cam
striker mechanism performance while cutting under actual weather conditions. Most of the
flowers were bloomed since few days (10 days) with a height of approximately 15 cm, few
flowers were onl
y partially opened or even not opened yet and their height was about 5 cm.

For testing, the mechanism was mounted on the end of a hollow tube with the electronic
circuitry for motor control and the power battery pack were both placed inside a box fixed to

the tube. A knob on the box connected to a potentiometer, allowed adjustment position of the
cam throughout the cutting process. Figure 16 shows the equipment used for field testing, and
figure 17 shows the cutting steps during a test. The cutting procedu
re consists of carrying the
device under the flower in order to locate the stem anywhere in the space between cam and
striker. Once positioned, the cam was operated forcing the flower to execute the kinematics
required. During the cutting the device was he
ld stationary. The flowers harvested had a mass
variation of about 10% with an average value of 0.5 grams. The size variation was higher,
approximately 20% to 25%, depending on the growing phase of the flower. The mean value
was found to be 22 cm

gure 16.
Crocus Sativus
(saffron) flower cutting equipment: (a) overview, (b) cam

Figure 17. Steps of the cutting process in the tests on the saffron flower field: (a) flower
location, (b) approach, (c) oscillating and cutting.


2 summarizes the results of the in field tests.

Table 2. Results of the in
field tests.

Flowers detached


Flowers picked with success


Flowers broken by the mechanism


Flowers unpicked


The three flowers that were not picked occurred, at the outset of the tests when the pre
loading bending
torsion spring had not been tuned. The spring torque was increased of about
a half of its initial pre
loading moment. Time needed for each cut was less

than 10 s. During
the cutting tests, adjustment of the spring pre
loading was somewhat time


The device described throughout the article is based on a

operating principle strongly
different than the method used for the manual harvesting of the
Crocus Sativus
flowers. The
principle proposed has been easily mechanized into a cam
striker mechanism. The device
was first tested in the laboratory to establish

its operating parameters and then in the field for
to observe the cutting performance under actual environment conditions. As a result of these
tests, the device and the equipment appear be suitable for mechanical harvesting of the
Crocus Sativus
since almost 87% of the flowers were successfully picked.

There is scope for future work focused on testing the inlet arrangement in a saffron field and
on complementing the device with a motor for the intake and a bag for receiving the cut


This research was funded by Italian Ministry of Research (MIUR). The authors would like to
thank Dr. Bruno Todde for his contribution in running the experimental tests.


Giametta, G. 1992. Mechanical harvesting of olives: Present s
ituation and prospects.
321: 510

Kondo, N., and M. Monta. 1999a. Fruit harvesting robotics.
J. of Robotics and Mechatronics
11(4): 321

Kondo, N., and M. Monta. 1999b. Chrysanthemum cutting sticking robot system.
J. of
s and Mechatronics
11(3): 220

Manfredi, E. 1990. Mechanical harvesting tomatoes for processing.
Acta Horticulturae
277 :

Melidis, P., and K. H. Vatterott. 1986. Method and apparatus for harvesting bloom parts of
crocus flowers. D.E. Patent No
. 3407517.

Petrucci, V. E., C. D. Clary, and M. O'Brien. 1983. Grape harvesting systems. In
and Practices for Harvesting and Handling Fruits and Nuts
, 525
574. Westport, Conn.: AVI

Pilarski, T., M. Happold, H. Pangels, M. Ollis, K
. Fitzpatrick, and A. Stentz. 2002. The
demeter system for automated harvesting.
Autonomous Robots
13(1): 9

Williames, G. A. 1986. Flower harvesters. U.S. Patent No. 4761942.

Savoia, G. 1985. Camomile flower picking or harvesting machine. G.B. Patent

No. 2153197.

Valk, J., and D. Vos Marinus. 2003. Flower harvesting, transport and singling method, by
transporting flowers in rake shaped holders and transferring them to fork shaped holders in
singling device. N.L. Patent No. 1020432C.

ASABE Site Map



Meetings & Events


Career Resources


Contact Us

About Us

News & Public Affairs



Awards & Landmarks

Technical Library

American Society of Agricultural and Biological Engineers

2950 Niles Road, St. Joseph, MI 49085

Phone: (269) 429

Fax: (269) 429


Copyright 2011 American Society of Agricultural and Biological Engineers