You are on page 1of 13

Computers and Electronics in Agriculture

32 (2001) 167– 179 www.elsevier.com/locate/compag

Fruit contact pressure distributions —


equipment
Bernd Herold a,*, Martin Geyer b, Clifford J. Studman c
a
Institut für Agrartechnik Bornim e.V., Horticultural Engineering Di6ision, Max-Eyth Allee 100,
D-14469, Potsdam-Bornim, Germany
b
Institute of Agricultural Engineering Bornim, Potsdam, Germany
c
Institute of Technology and Engineering, Massey Uni6ersity, Palmerston North, New Zealand

Received 1 October 2000; received in revised form 9 March 2001; accepted 20 March 2001

Abstract

A commercially available tactile sensing system allows the pressure distribution between
contacting surfaces to be determined in real time. This sensor has been used to determine the
pressure distribution between fruits, mainly apples, in contact with flat and curved surfaces.
The sensor consists of a thin flexible plastic film containing a grid of sensitive material that
responds according to the pressure at each point on the mesh. Details of the sensor and
examples of experimental results obtained with the system are described. The device was
found to have a non-linear output, and the data needed to be calibrated to improve the
accuracy of the results. Examples of the surface pressure distribution generated in the contact
area as apple fruit were loaded to failure, and the effects of commercial packaging materials
on pressure distribution are presented. © 2001 Elsevier Science B.V. All rights reserved.

Keywords: Apple fruit; Contact area; Matrix sensor; Packaging material; Pressure distribution

1. Introduction

After harvest, biological products such as fruit and vegetables are subjected to a
wide range of dynamic and static loads as they are transported, graded and
packaged. Handling damage caused by impact or quasi-static compression is a
significant problem for many perishable crops, and a large number of empirical

* Corresponding author.
E-mail address: bherold@atb-potsdam.de (B. Herold).

0168-1699/01/$ - see front matter © 2001 Elsevier Science B.V. All rights reserved.
PII: S 0 1 6 8 - 1 6 9 9 ( 0 1 ) 0 0 1 6 0 - 0
168 B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179

experiments have been conducted to establish the extent of damage which can result
during commercial operations (e.g. Klein, 1987; Schulte et al., 1990, 1991; Hyde,
1997; Thompson, 1996; Studman, 1999). In order to explain the results, several
simple models to describe the fruit loading process, and the resulting volume of
damage, have been developed (e.g. Holt and Schoorl, 1977; Chen and Sun, 1981;
Jones et al., 1991; Garcia et al., 1995; Roudot et al., 1991). While damage volume
estimates based on energy considerations can be useful, in practice, the critical issue
is not so much the extent to which the product is loaded or even damaged by the
loading cycle, but rather the consequences of that damage on subsequent storage
life and acceptability of the product to the consumer (Hung, 1993; Shewfelt and
Prussia, 1995; Studman, 1997). In this respect, visible damage and surface area of
damage may be more important than the total volume of damage (Pang et al., 1992;
Studman, 1996, 1999). For a clearer understanding, a model of the mechanical
behaviour of the fruit itself is needed. Damage is essentially related to the stresses
developed within the product during loading. These are related to the contact
pressures, rather than the total applied force (Schoorl and Holt, 1980; Johnson,
1985). Most theoretical studies relating to contact area and pressure estimation
have been based on elastic contact theory (Hertz, 1882; Timoshenko and Goodier,
1970), despite clear limitations of this theory in describing materials as complex as
fruit and vegetables, which exhibit both anisotropic properties (Khan and Vincent,
1990, 1993; Studman, 1997) and visco-elastic behaviour (Mohsenin, 1970). Studies
have shown that elasticity theory is a reasonably reliable indicator of the interac-
tions between key variables such as force and time of contact, although errors in
estimates of contact area can be around 20% (Siyami et al., 1988). Experimentally,
contact pressures are difficult to measure. However, a commercial tactile sensor is
now available, which allows the pressure distribution between contacting surfaces
to be determined in real-time (Tekscan, 2000)1. It was originally developed to help
dentists determine how well a patient’s teeth were contacting. It consists of a thin
flexible plastic film containing a grid of sensitive material that responds according
to the pressure at each of the points on the mesh.
This paper presents details of an experimental system used and its performance.
A simplified description has been presented elsewhere (Geyer et al., 1998). Exam-
ples of the surface pressure distribution generated in the contact area as apple fruit
were loaded to failure are also given. A quantitative analysis of the contact area will
be presented elsewhere (Studman et al., in preparation).

2. Equipment

2.1. Tactile sensor and measuring system

Pressure distribution data were determined using a 0.1 mm thick matrix-based


tactile sensor Type Tekscan No. 5051 with a saturation pressure of 2.4 MPa

1
Tekscan, Inc., 307 West First Street, Boston, MA 02127-1342, USA.
B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179 169

(Tekscan, 2000). This sensor consisted of an assembly of two thin, flexible


polyester sheets that had patterns of electrically conductive electrodes. The inside
surface of one sheet formed a 44 row pattern, while the inner surface of the
other employed a 44 column pattern. The spacing between the rows and columns
was 1.3 mm. Thin semi-conductive coating (ink) formed an intermediate layer
between the electrical contacts (rows and columns). This ink provided the electri-
cal resistance change at each of the intersecting points, which created the sensing
locations (Fig. 1). The sensor was connected to a scanning handle that scanned
across the sensor matrix to measure the resistance at each individual sensing
location and digitised the measured data using an 8 bit analog-to-digital con-
verter (Fig. 2). The digitised information of each sensing point was transmitted
from the scanning handle to a Tekscan data acquisition card in a PC (ISA slot)
via a serial communication protocol. The data acquisition card acted as a dy-
namic memory management device and handled all communication between the
PC bus and the scanning handle.
The Tekscan software enabled pressure information to be displayed in real
time on a PC screen, at a maximum sampling rate of 100 frames per second,
and enabled the magnitude, temporal characteristics and location of forces on its
surface to be determined.
The software also enabled calculation of mean pressures over a selected sec-
tion of the contact area to be made, and also calculated the total force on the
section.

2.2. Test configuration

Compression tests were conducted on selected apple cultivars. Each sample


was embedded in moist sand, with the stem– calyx axis horizontal, and com-
pressed from above, with the sensor placed between the compression surface and
the fruit (Fig. 3).

Fig. 1. Matrix based tactile sensor. (a) Schematic view of extracted structure parts. (a) Real view of
sensor with square sensing area (side length 56 mm, matrix of 44 rows and 44 columns) connected to the
scanning handle.
170 B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179

Fig. 2. Simplified sensor and scanning handle electronics schematic (Tekscan 2000).

The sensor system scanned the sensing array and recorded the pressure distribu-
tion throughout the loading sequence. In most of the experiments, the sampling rate
was set at 10 frames per second. Measurements were saved for subsequent analysis.
The samples were loaded with a universal testing machine Type Zwicki 1120 with
a 500 N load cell. The fruits were compressed at a constant crosshead displacement
of 30 mm per minute to a maximum force of 250 N, and then unloaded. The Zwicki
load was recorded onto a second computer as a function of time at the same
sampling rate as the Tekscan, for comparison with the calculated force recorded by
the tactile sensor.

3. Performance of the tactile sensor

The sensor was checked for consistency of output by applying a uniform pressure
to a selected section of the sensor when it was placed between two flat parallel
surfaces. This was achieved by applying a load through a flat ended cylindrical
plunger 20 mm in diameter, which was cushioned with a foam rubber plate 3 mm
thick and 18 mm in diameter. The load was applied at several locations across the
film (at the corners and at the centre) and the resulting output compared with the
B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179 171

Fig. 3. Arrangement for compression test on apple fruit.

load measured by the universal testing machine (Fig. 4). These tests showed that the
relationship between the tactile sensor output and the total applied load was
non-linear, and a suitable calibration was required to correct for this. Measure-

Fig. 4. Comparison between tactile and force sensor data from calibration tests with a cylindrical
plunger.
172 B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179

ments at different locations also indicated that the sensitivity over the whole area of
the sensor was not uniform, but varied slightly across the film. One side of the film
(near locations 3 and 4), at the right edge of the sensor in direction from the
measurement head had a higher response compared to the left edge. However, there
was reasonable uniformity along the film. The output varied with position over a
range of 9 8.3% of the force recorded at the central part of the sensor (location 5).
Hence, in order to minimise variability, experiments were conducted at the same
point on the sensor each time, or a correction was necessary.
If the raw data from the Tekscan were used to calculate the total force taken
from tests conducted around the centre of the sensing area, then there was a
reasonably linear relationship between these two data sets (R 2 = 0.92). For this
comparison, data points were selected at identifiable events such as Bioyield or peak
load recorded by the Universal testing machine and the Tekscan. Calibration to
remove these non-linearities is discussed below.

4. Typical uncalibrated results

The results shown below are typical results taken from a series of experiments
conducted on apples grown in the Potsdam region of Germany. The examples
shown all involved Jonica apples harvested during September and October, and
stored in a commercial cool store for up to 4 weeks. Samples were then selected at
random from the bins and transported to the laboratory, where they were either
placed in a cool store at 4°C, or placed for 24 h in a room at 229 1°C and 50%
relative humidity, until tested. Tests were conducted at room temperature, or on
fruit taken straight from the cool store. The typical fruit weight was about 180 g,
and the average Penetrometer firmness was 5094 N.
Fig. 5 shows a typical example of force versus time during compression of an
apple fruit by a flat plate, including two characteristic failure events. Some points
on this curve are marked by numbers for comparison with measurements of
pressure distribution (Fig. 6).
The general pressure pattern for apples pressed against a flat steel surface was
similar for all varieties. Initially, the pressure increased steadily in the contact area
as the applied load increased (Fig. 6). The pressure was greatest in the centre of the
contact area, falling away towards the edge in a roughly parabolic shape, as
predicted by elastic contact theories. However, when the load in the centre exceeded
the maximum sustainable, there was a sudden drop in pressure at the centre,
corresponding to the Bioyield point (Mohsenin 1970). This was accompanied by an
increase in contact area. After this failure, the pressure distribution changed so that
the greatest pressure was at the edge of the contact circle. The average central
pressure fell to about half the level just prior to failure, while the edge pressure
could still reach the peak pressure before failure. Subsequently, as the load
continued to increase, there were further increases in contact area. Pressures near
the edge of the circle remained the highest for the rest of the loading period. While
the pressure near the centre varied, it was generally more evenly distributed
B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179 173

Fig. 5. Typical universal testing machine recording of force versus time for compression of a Jonica fruit
by flat plate. The pressure distribution at each marked point is indicated in Fig. 6. Points 1 and 2 were
before failure, points 3 and 4 just after, and points 5 and 6 were before and after a significant failure.
Fruit stored in cool store, then 24 h at 22 9 1°C, and tested at 22 91°C.

throughout the central region of the contact area and did not regain the values
reached just before the initial failure. Major failure events were common at all
stages during loading. The greatest pressures (usually at points on the edge of the
contact circle) occurred before failure, but were usually equal to, or lower than, the
pressure at the first failure. After a large failure event (for example, when the
applied force fell noticeably), the pressure distribution in the central region ap-
peared to be more uniform than if a relatively small failure event occurred. Failures
were not necessarily spread over the entire contact area, but could involve the
collapse of only a part of the rim. Frequently partial failure could be observed from
the sensor output, even though there was no evidence of failure from the force
output from the testing machine load cell. However, the pressure distribution over
the entire contact area would usually be affected by such collapses. On unloading,
the pressure became more uniform over the entire contact area, but the contact area
was much larger at a given load than during loading, as would be expected when
plastic deformation had occurred.

Fig. 6. Tekscan records of pressure distribution on a Jonica fruit surface under increasing load by flat
plate (same run as Fig. 5). The horizontal plane agrees with the sensing area, while the vertical axis
shows the raw pressure data. The pressure distributions correspond to the marked points in Fig. 5.
Points 1 and 2 were before failure, points 3 and 4 just after, and points 5 and 6 before and after a
significant failure. Fruit were stored in cool store, then for 24 h at 22 91°C and tested at 22 9 1°C.
174 B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179

Fig. 6. (Continued)
B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179 175

For fruit removed from cool storage (4°C) and tested immediately, rupture events
were generally more distinctly marked. The maximum pressures were higher overall,
both before and after failure events, when compared with fruit stored at room
temperature for 5 days. Other apple varieties behaved in a similar fashion, although
failure pressures varied (Geyer et al., in preparation).

5. Discussion

These results confirm Mohsenin’s view that the Bioyield maximum pressure is an
important feature of the failure behaviour of fruit cell tissue. However, the value
cannot be determined easily by conventional sensing techniques, unless an assump-
tion is made about the pressure distribution.
Although the pressure distribution changed dramatically, the differences between
total force (calculated by the Tekscan software by simple summation of the
pressures) before and after rupture were surprisingly small. In many cases, raw
tabulated data of force produced by the system before and after a failure event
recorded from the sensor output showed a small increase (0– 10%). This result was
clearly erroneous and was almost certainly a consequence of the non-linearity of the
Tekscan output, as indicated in Fig. 4. In the software supplied with the system, the
Tekscan estimate is obtained by summing all readings at all the contact points,
while the shape of the calibration curve demonstrated that the Tekscan sensitivity
decreased with increasing pressure. Hence, a decrease in pressure coupled with a
balancing increase in area (and hence the number of low force contact points
contributing to the force sum) would result in an overestimate of the summed force.
This was confirmed by making an appropriate calibration adjustment for the
output and applying it to selected data points. If average values are taken from the
data used in the calibration to obtain Fig. 4, it can be shown that the curve can be
fitted by the relationship:

F = %all pixels(0.000105Q 2 +0.008965Q),

where F is the total force, and Q is the pixel numerical value (0–255).
The pressure on a single pixel can be obtained directly from this relationship
since each pixel occupies an area of 1.69 mm2.
Thus, the pressure in kPa is given by

P =0.06202Q 2 +5.304Q.

This correction was applied to the data to correct for the non-linearity in the
output, using the results from Fig. 4 to correct the individual readings. This resulted
in a dramatic fall in the force calculated by the Tekscan, as would be expected. For
example, a failure event during an experimental run on a pear, which reduced the
original output by only 8.7%, resulted in a fall in the force of 17.1% after the
correction was applied.
176 B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179

Fig. 7. Comparison of commercial cushioning materials 40 mm diameter cylinder pressed against flat
plate, with 6.6 mm cushioning layer, load 450 N. Firm cushioning material, Cellular Urethane, Poron
4701-12-20250-1604. Medium cushioning material, Cellular Urethane, Poron 4701-01-15250-1604
(Rogers Corp., Rogers/Conn.).

It is also possible that there was some hysteresis in the Tekscan output, which
would have had a further effect. However, as Fig. 4 shows, this correction would be
relatively minor in comparison.
In practice, the Tekscan system offers some very good opportunities for a better
understanding of contact pressure distributions. For comparative purposes, the raw
output is adequate. However, for precise measurements and pressure estimates, the
non-linearity of the sensor is a serious problem, and a correction is required. The
calibration given above is based on a simple averaging approach, and a more
accurate procedure is preferable. Truppel et al. (in preparation) have written a
program to correct the output, and further work is in progress to develop an
on-line linearising function to adjust the Tekscan output on a point-by-point basis,
to correct for these variations. However, these programs are not yet available
commercially, and for the present studies, the force measured by the Zwicki
Universal Tester was used for analysis.
The sensor has also been used to measure the pressure distribution between fruit
and other curved surfaces, including fruit to fruit contact (Geyer et al., in
preparation, Pang et al., 1992). These situations are very common during harvest-
ing, handling and storage operations. Fruit-to-fruit contacts are complicated by the
fact that one or both fruit can be damaged during contact, and while the total
volume of bruising may be determined by the severity of loading, the damage is
distributed unevenly between the two fruit. In some cases, where severe damage
occurs, one fruit may still be undamaged (Pang et al., 1992; Studman et al., 1997).
Similar results were obtained to the results for single apple loading against a flat
plate, although the rim effect was sometimes reduced. Some care must be taken to
avoid damage to the sensor in this type of experiment.
B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179 177

The sensor can also be used for the assessment of cushioning materials used in
packaging. Fig. 7 shows the pressure distribution when a cylinder was pressed
against a flat plate, with layers of commercial cushioning interposed. The figure
demonstrates the situation at an advanced state of loading, when the applied
pressure has exceeded the cushioning capabilities of the material. There is a large
increase in pressure in the centre of the contact area, even though the pressure is
distributed over a similar or slightly larger area. The other material has dis-
tributed the pressure more effectively over the contact area, and so the maxi-
mum pressure in the centre is lower when compared to the other material. This
work has been described elsewhere (Geyer et al., in preparation, Studman et al.,
in preparation).
Fig. 8 shows the pressure distribution produced between an apple and a flat
surface, when a layer of commercial foam netting was placed between them. The
pressure was completely taken by the netting and hence by the region of the
fruit directly beneath the netting. Shortly afterwards, the apple tissue failed. This
was evident from the resulting fall in pressure and an increase in the contact
area.

Fig. 8. Effect of foam mesh (3.0 mm diameter foam in 12 mm grid) layer between the apple and a flat
surface. Pressure distribution just before tissue failure (averaged data). Jonagold at room temperature.
Applied force 60 N; contact area 4.1 cm2.
178 B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179

6. Conclusions

The Tekscan sensor enabled pressure distribution data to be obtained with a


spatial resolution of about 1.3 mm. It was capable of a maximum sampling
frequency of 100 frames per second. This sensing technique has been used to
determine the real-time development of pressure over the contact area between
apples and and other surfaces. The advantage of this sensing technique is the ability
to measure the contact pressure distribution between the surfaces. The sensor
enabled the simultaneous acquisition of data on contact area, pressure, and total
force throughout the loading cycle.
In comparison to a calibrated force sensor, the accuracy of the total force
measurement varied by up to 10%. However, this error could be reduced by
allowing for the non-linearity of the output. For reproducible data, a suitable
calibration has yet to be developed.
The studies showed that the mechanical resistance and failure behaviour of apple
fruit can be studied in more detail with this technique than by conventional means.
It was possible to detect partial failure behaviour within the contact area. The
grid-based pressure sensor has provided experimental data to compare with existing
models of fruit cell tissue failure. It has been used successfully in a number of
different loading configurations. In practical applications, the tactile sensor can be
used to provide information on the risk of damage during handling and packaging
operations.

Acknowledgements

We would like to thank the German Federal Ministry for Food, Agriculture and
Forestry, and the New Zealand Ministry of Research, Science and Technology for
support for this project under the International Sciences and Technology Linkages
scheme.

References

Chen, P., Sun, Z., 1981. Impact Parameters Related to Bruise Injury in Apples. Paper No. 81-3041.
American Society of Agricultural Engineers, St. Joseph, MI.
Garcia, J.L., Ruiz-Altisent, M., Barriero, P., 1995. Factors influencing mechanical properties and bruise
susceptibility of apples and pears. J. Agric. Eng. Res. 61 (1), 11 – 18.
Geyer, M., Herold, B., Studman, C.J., 1998. Fruit Contact Pressure Distributions. Conference Proceed-
ings Ag Eng 98: Oslo, Paper 98-F-001.
Geyer, M., Studman, C.J., Herold, B., in preparation. Contact pressure distributions during slow
compression loading of fruit and vegetables.
Hertz, H., 1882. Uber die Beruhrung fester elastischer Korper und uber die Harte (On the contact of
elastic solids and on hardness). J. Reine und angewandte Mathematik 92, 156 – 171 English
translation in Jones and Scholl (Eds.), 1896. Miscellaneous Papers by H. Hertz, vol. 5. Macmillan,
London, pp. 90 –156.
Holt, J.E., Schoorl, D., 1977. Bruising and energy dissipation in apples. J. Texture Stud. 7, 421 – 432.
B. Herold et al. / Computers and Electronics in Agriculture 32 (2001) 167–179 179

Hung, Y.C., 1993. Latent damage-a systems perspective. In: Shewfelt, R.L., Prussia, S.E. (Eds.),
Postharvest Handling — a Systems Approach. Academic Press, San Diego, CA Chapter 10.
Hyde, G.M., 1997. Bruising-impacts, why apples bruise, and what you can do to minimize bruising.
Treefruit Postharvest J., Wash. State Univ. 8 (4), 9 – 12.
Johnson, K.L., 1985. Contact Mechanics. Cambridge University Press, New York.
Jones, C.S., Holt, J.E., Schoorl, D., 1991. A model to predict damage to horticultural produce during
transport. J. Agric. Eng. Res. 50 (4), 259 – 272.
Khan, A.A., Vincent, J.F., 1990. Anisotrophy of apple parenchyma. J. Sci. Agric. 52, 455 – 466.
Khan, A.A., Vincent, J.F., 1993. Anisotrophy in the fracture properties of apple flesh as investigated by
crack opening tests. J. Mater. Sci. 28, 45 – 51.
Klein, J.D., 1987. Relationship of harvest date, storage conditions, and fruit characteristics to bruise
susceptibility of apples. J. Am. Soc. Hortic. Sci. 112 (1), 113 – 118.
Mohsen in, N.N., 1970. Physical Properties of Plant and Animal Materials, vol. 1. Gordon and Breach,
New York, pp. 88 –100, 341 – 365, 401 – 430.
Pang, D.W., Studman, C.J., Ward, G.T., 1992. Bruising damage in apple-to-apple impact. J. Agric. Eng.
Res. 52 (4), 229 –240.
Roudot, A.C., Duprat, F., Wenian, C., 1991. Modelling the response of apples to loads. J. Agric. Eng.
Res. 48 (4), 249 –259.
Schoorl, D., Holt, J.E., 1980. Bruise resistance measurements in apples. J. Texture Stud. 11, 389 – 394.
Schulte, N.L., Timm, E.J., Brown, G.K., Marshall, D.E., Burton, C.L., 1990. Apple damage assessment
during intrastate transportation. Appl. Eng. Agric. 6 (6), 753 –758.
Schulte, N.L., Timm, E.J., Armstrong, P.A., Brown, G.K., 1991. Apple Bruising — a Problem During
Hand Harvesting, Paper No. 91-1021. ASAE, St. Joseph, MI.
Shewfelt, R.L., Prussia, S.E., 1995. Postharvest Handling — a Systems Approach. Academic Press, San
Diego, CA Chapter 2.
Siyami, S., Brown, G.K., Burgess, G.J., Gerrish, J.B., Tennes, B.R., Burton, C.L., Zapp, R.H., 1988.
Apple impact bruise prediction models. Trans. Am. Soc. Agric. Eng. 31 (41), 1038 – 1046.
Studman, C.J., 1996. Model of fruit bruising. In: Proceedings of the 2nd Australasian Postharvest
Conference, Science and Technology for the Fresh Food Revolution, Monash University, Mel-
bourne. Institute for Horticultural Development, Department of Natural Resources and the Environ-
ment, pp. 241 –246.
Studman, C.J., 1997. Factors affecting the bruise susceptibility of fruit. In: Jeronimidis, G., Vincent,
J.F.V. (Eds.), Proceedings of Conference on Plant Biomechanics. University of Reading, Reading,
UK, pp. 273–281.
Studman, C.J., Brown, G.K., Timm, E.J., Schulte, N.L., Vriegg, M., 1997. Bruising on blush and
non-blush sides in apple to apple impacts. Trans. Am. Soc. Agric. Eng. 40 (6), 1655 – 1663.
Studman, C.J., 1999. Handling Systems and Packaging. In: Bakker-Arkema, F.W. (Ed.), CIGR
Agricultural Engineering Handbook, vol. IV.3. American Society of Agricultural Engineers, St.
Joseph, MI, pp. 291 –340 Chapter 3.
Studman, C.J., Geyer, M., Herold, B., in preparation. Prediction of Contact Area and Pressure
Distribution for Compression Loading of Fruit and Vegetables. In Preparation. Tekscan 1993.
Technical Information on Tactile Thin Film Sensor System I-Scan. Tekscan, South Boston, MA.
Tekscan, 2000. Tekscan technology. http://www.tekscan.com/technology.html
Thompson, A.K., 1996. Postharvest Technology of Fruit and Vegetables. Blackwell, Oxford, pp. 1 – 25,
42 –44.
Timoshenko, S.P., Goodier, J.N., 1970. Theory of Elasticity, third ed. McGraw-Hill, New York.
Truppel, I., Herold, B., Geyer, M., in preparation. An automatic calibration algorithm for Tekscan
pressure sensor.

You might also like