Embroidered Resistive Pressure Sensors: A Novel Approach for Textile Interfaces

Full text

Embroidered Resistive Pressure Sensors: a Novel Approach for
Textile Interfaces
Roland Aigner∗, Andreas Pointner∗, Thomas Preindl∗, Patrick Parzer, Michael Haller
mi-lab@fh-hagenberg.at
Media Interaction Lab, University of Applied Sciences Upper Austria
a b c d
Figure 1: The workow of implementing an embroidered pressure sensor: stitching an electrode to a base material using
a conductive yarn (a), adding a resistive fabric (b), and stitching another electrode with conductive yarn (c). The resulting
sensor’s resistance readings may be used for arbitrary controls (d).
ABSTRACT
We present a novel method for augmenting arbitrary fabrics with
textile-based pressure sensors using an o-the-shelf embroidery
machine. We apply resistive textiles and conductive yarns on top
of a base fabric, to yield a exible and versatile continuous sensing
device, which is based on the widespread principle of force sensitive
resistors. The patches can easily be attached to measurement and/or
computing devices, e.g. for controlling accessories. In this paper, we
investigate the impacts of related design and fabrication parameters,
introduce ve dierent pattern designs, and discuss their pros and
cons. We present crucial insights and recommendations for design
and manufacturing of embroidered pressure sensors. Our sensors
show a very low activation threshold, as well as good dynamic
range, signal-to-noise ratio, and part-to-part repeatability.
CCS CONCEPTS
•Human-centered computing →Interaction devices
;
Haptic
devices.
KEYWORDS
Embroidered Force Sensitive Resistance; Embroidery; Space-Filling
Patterns; Textile Sensor; Smart Textiles
Permission to make digital or hard copies of all or part of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for prot or commercial advantage and that copies bear this notice and the full citation
on the rst page. Copyrights for components of this work owned by others than ACM
must be honored. Abstracting with credit is permitted. To copy otherwise, or republish,
to post on servers or to redistribute to lists, requires prior specic permission and/or a
fee. Request permissions from permissions@acm.org.
CHI ’20, April25–30, 2020, Honolulu, HI, USA
©2020 Association for Computing Machinery.
ACM ISBN 978-1-4503-6708-0/20/04. . . $15.00
https://doi.org/10.1145/XXXXXXX.XXXXXXX
ACM Reference Format:
Roland Aigner
∗
, Andreas Pointner
∗
, Thomas Preindl
∗
, Patrick Parzer, Michael
Haller. 2020. Embroidered Resistive Pressure Sensors: a Novel Approach
for Textile Interfaces. In CHI Conference on Human Factors in Computing
Systems (CHI ’20), April 25–30,2020, Honolulu, HI, USA. ACM, New York, NY,
USA, 11 pages. https://doi.org/10.1145/XXXXXXX.XXXXXXX
1 INTRODUCTION
The handcraft of textile production is sometimes attributed as one
of the rst technologies of human history. Consequently, textiles
are omnipresent in humans’ environment for millennia, through-
out all parts and cultures of the world, most visibly in the form of
clothing. By now, we are literally surrounded by textiles, since they
are also found on furniture, walls, and oors, in vehicles, but also
in very much concealed forms, such as ber-reinforced compos-
ites. This quality of ubiquity and versatility poses an ideal vantage
point for pushing Mark Weiser’s frequently quoted vision of Ubiq-
uitous Computing [
34
]. In this spirit, professionals and hobbyists
have recently moved on from e-textiles (incorporating elements
for heating, lighting, etc.), and started incorporating interactive
functionalities into textiles by weaving, knitting, and embroidering.
Apart from availability, numerous aspects of textiles pose great
advantages for function and comfort, such as exibility, elasticity,
breathability, and pleasant haptic qualities.
In this paper, we focus on the process of embroidery, which is
the craft of applying yarn to a base fabric, originally used for aes-
thetic purposes. The development of machine-embroidery made it
a fundamental tool of modern textile nishing for mass production,
∗Authors contributed equally to this research.

CHI ’20, April25–30, 2020, Honolulu, HI, USA Aigner, Pointner, Preindl, et al.
still providing advantages like customization and repeatability. Ar-
guably, this caused its spread to technical applications in the elds
of engineering [
31
], medicine [
2
], and smart textiles [
19
]. Beyond
advantages in industrial production, we deem the technique ideal
for rapid prototyping of textile sensors and circuitry.
We present a method for implementing pressure sensor patches
on top of existing weaves or knits. In contrast to solutions with
discrete pressure sensitive points [
22
], we aim for sensor areas, en-
abling to create arbitrary shapes and sizes. We utilize a commercial
embroidery machine (cf. Figure 1), which allows to specify a variety
of parameters to control the behavior of the resulting sensor. We
therefore examine these parameters with respect to their impact
on the sensor’s responsiveness and dynamic range. Moreover, we
investigate a number of patterns, which are highly scalable in size,
and report on their advantages and disadvantages. We highlight
best practises to embroider resistive textile sensors and complete
with a description of two applications demonstrating potential use
cases.
In summary, the main contributions of this paper are:
•
A description of a method for rapid prototyping of pressure
sensors via a commercially available embroidery machine.
•
Findings of an experiment, investigating design and manu-
facturing parameters, which are aecting the sensors’ per-
formance.
•
Study and analysis of ve space-lling patterns for generat-
ing scalable pressure sensor patches.
•
Implications and best practices for design and manufactur-
ing, in particular regarding handling of conductive yarn in
context of embroidery.
•
Example applications demonstrating the embroidered pres-
sure sensors and the feasibility of our approach.
2 RELATED WORK
Textile sensing is an extensive area of research addressing the ac-
quisition of analog or digital information from textiles by means of
electronic measurement. A subdomain is dealing with the detection
of changes in pressure, predominately using capacitive and resistive
approaches. Both sensing methods are viable for pressure and touch
sensing with their respective benets and drawbacks originating
from their fundamentally dierent working principles. The suitable
sensing method therefore often depends on the actual use cases.
In our work, we focus on resistive pressure sensors utilizing the
advantages that come with that approach.
2.1 Textile Pressure Sensors
Resistive textile sensors are usually composed of two conductive
electrodes separated by a resistive sheet [
27
,
32
,
36
]. Often, they aug-
ment everyday objects with sensing capabilities. Rofouei et al. [
27
]
constructed an array of pressure sensors using a resistive eece,
placed between two conductive layers. The sensor grid could infer
shape, position and weight of an object placed on it. Using the same
resistive textile eCushion [
36
] built a smart cushion for sitting pos-
ture analysis. Similarly, Shu et al. [
16
] implemented a resistive tex-
tile pressure sensor array for in-shoe sensing, electrically connected
by conductive yarn, which is adhered to the readout electronics.
Researchers have developed a multitude of resistive textile sensor
arrays based on a multi-layer approach [
6
,
7
,
23
,
29
,
32
,
38
]. Smart-
Mat [
32
] is integrating textile pressure sensors into exercise mats
to recognize and count several types of exercises. GestureSleeve [
29
]
presents a touch-enabled textile placed on the forearm, enabling
interaction via gestures, such as swipes or taps. FlexTiles [
23
] uses
a stretchable three-layer sensor, to detect pressure on prosthetic
limbs [
15
], or detecting a wide range of gestures on textiles [
24
].
There are notable exceptions to these multi-textile-layer approaches.
RESi [
22
] combines conductive and resistive properties on a yarn
level. Saenz-Cogollo et al. [
28
] built a resistive sensor mat by embroi-
dering conductive yarn onto a non-conductive textile and stamping
a conductive polymer onto the yarn junctions to enable pressure
sensing.
Our sensor design builds upon mentioned research and oers
comprehensive design details for applications where multi-scale
sensors and a comprehensible manufacturing process from fabric
to function are required.
2.2 Machine Embroidery
Embroidery has been one of the rst fabrication techniques used
in the eld of smart textiles. In their pioneering work, Post et
al. [
26
] used it to create conductive electrodes for capacitive sens-
ing, inspiring other works and thereby casting the seed for the
prospective eld of smart textiles. Embroidery was utilized for
generating multi-layered capacitive sensing structures [
20
], con-
nections between textile wiring and rigid electronic substrates [
17
],
integrating electromyography electrodes into clothes [
30
], creating
antennas [
1
,
5
,
13
,
33
] and fabricating textile coils for magnetic res-
onance sensors [
21
]. Also interaction-centered applications, such
as pinch gesture recognition on textiles [
11
,
14
], have been in-
vestigated. Researchers also combined aesthetics with function in
embroidered pattern resistors [
10
] and realized imaginary com-
puting technologies by embroidering logic gates with ornamental
patterns [25].
Addressing the wide range of application specic papers, we
seek to oer a fabrication centered approach, to facilitate the design
and manufacturing process of textile pressure sensors for future
researchers, practitioners, and makers.
2.3 Customized fabrication
From early on, the eld of smart textiles showed a strong over-
lap of academic research, art practices, education, and the DIY
community. The extensive collection of projects from Satomi and
Perner-Wilson, combining textile crafts with electronics on their
platform Kobakant
1
, demonstrated the virtue of hands-on fabri-
cation in the domain of personal fabrication. Acknowledging the
demand for personal fabrication methods in the eld of smart tex-
tiles, a growing group of researchers addresses this topic in the
context of weaving [9] and embroidery [12].
Aligning with their goals, we try to narrow the gap between
textile fabrication and electronic engineering to foster collaboration
and enhance mutual understanding in both elds.

Embroidered Resistive Pressure Sensors: a Novel Approach for Textile Interfaces CHI ’20, April25–30, 2020, Honolulu, HI, USA
non-cond. thread
conductive thread
resistive sheet
base material
bobbin thread
Figure 2: The basic components of our embroidered pressure sensor (left), showing bobbin thread holding the conductive yarn.
The degree of contact and compression of yarn and resistive material is the most contributing factor to the change of resistance
(right): when pressure is applied, the contact area between yarn and resistive material rises (right).
RV
RS1 RS2
conductive thread
resistive sheet
Ω
Figure 3: The overall resistance is the sum of volume re-
sistance and two surface resistances, resulting in Rtot a l =
RV+RS1+RS2.
3 FORCE SENSITIVE RESISTANCE
Our textile sensor is based on the principle of Force Sensitive Resistors
(FSR), which are used in dierent applications for several decades
[
8
]. An FSR represents a continuous electrical controller, whose
electric resistance decreases gradually as pressure is applied. In our
implementation, two electrodes are put in contact with a resistive
material (cf. Figure 3).
From an electrical point of view, the sensor’s overall resistance
is the sum of the material’s volume resistance
RV
and the surface
resistances
RS1
and
RS2
resulting from the contact areas of each
of the electrodes. While volume resistance is mostly constant for
a given sensor, the surface resistance is varying when pressure is
applied, as a result of the interface eect [35].
3.1 Embroidered Force Sensitive Resistor
The most common stitch in machine embroidery is the lockstitch,
which interlocks an upper and a lower (bobbin) thread using a
rotary hook.
For the embroidered sensors, we augment a base fabric substrate
with pressure sensing capability, employing a conductive yarn as
upper and a non-conductive yarn as bobbin thread (cf. Figure 2).
In one of three possible approaches, the base fabric is covered by
the resistive material, which is then xated along the edges using a
regular, non-conductive yarn, in order to prevent it from shifting out
of place. Subsequently, the electrodes are stitched on top. Contrary,
1www.kobakant.at/DIY/
electrodes can also be stitched below the resistive material or in
alternating order, as we will demonstrate later.
3.1.1 Base fabric. A crucial property for manufacturing is the
thread count of the base fabric, as it aects the precision of the
embroidered pattern: denser materials enable for more accurate
positioning of stitches. While not aecting the sensor’s electrical
properties, thread count aects the minimum electrode distance
that can be achieved. Moreover, a so-called backing or stabilizer
(usually a non-woven fabric) is commonly applied to the backside
of the base material, to prevent warping caused by the pull of the
embroidery yarns, which in our case would result in stressing the
sensor, introducing noise indistinguishable from actual applied
pressure. Also, stretching the base fabric can cause the electrodes to
tear, rendering the sensor permanently defective. For our sensors,
we decided for the BadgeTex 2900 twill with 330 g/m
2
, as it allows
for sub-millimeter precision in positioning stitches and shows little
tendency to frail when trimmed. It comes already with the top cloth
pressed onto a polyester eece stabilizer, to prevent warping.
3.1.2 Resistive fabric. Regarding the resistive fabric, we investi-
gated several materials suitable for embroidery. Options include
knitted and woven fabrics, eeces, and foams. We closely inspected
the Eeonyx EeonTex LTT-SLPA 20 k knit and the SEFAR CAR-
BOTEX 03-82 CF weave (cf. supplementary material). During our
tests, we noticed that the elastic Eeonyx shows anisotropic perfor-
mance, due to the inherent surface structure of a knit. Additionally,
the behavior largely varies, depending on which side is facing up.
Therefore, we decided to use the CARBOTEX, since it did not show
direction dependent behavior. Note that there are also weave vari-
ants not fully made of carbonized threads, which potentially renders
them direction dependent and inhomogeneous.
3.1.3 Stitching of electrodes. We informally evaluated 39 commer-
cially available conductive yarns, inspecting electrical resistance,
mechanical abrasion, and suitability for machine embroidery. De-
tailed information and results of this evaluation can be found in the
supplementary material. We found the Madeira HC40 as the ideal
yarn for our requirements, since it is specially designed for embroi-
dery machines and proved durable during our tests. Moreover, it
showed relatively low electrical resistance of 127
Ω
/m. As a bobbin
thread, we used an Amann ISABOB 100% Polyester pre-wound
bobbin thread.
As the resistive fabric is of uniform thickness, and assuming
parallel electrodes of equal length, the resulting resistor can be

CHI ’20, April25–30, 2020, Honolulu, HI, USA Aigner, Pointner, Preindl, et al.
modeled as a cuboid with specic electrical resistance
ρ
. In this
case, the longitudinal resistance between the two electrodes is
R=ρd
tl ,(1)
where
t
is the resistive material’s thickness,
l
is the length of the
respective electrode’s trace length on-top of the resistive material,
and dis the distance between the two electrodes (cf. Figure 4).
4 SENSOR EVALUATION
In the following, we describe an evaluation of our pressure sensor
design with respect to design and fabrication parameters. In partic-
ular, we were interested in how these parameters aect resistance
once we apply pressure (cf. Figure 4a). Therefore, we tested vari-
ations of a primitive sensor by placing weights on top. For these
patches, we modied several parameters we considered relevant
for their performance: due to the general denition of electrical
resistance (cf. Equation 1) and since material thickness
t
is con-
stant, the most relevant properties in our scenario are electrode
distance
d
(1 mm, 2 mm, 4 mm, . . . , 32 mm) and electrode length
l
(10 mm, . . . , 60 mm) (cf. Figure 4a,b). Furthermore, we investigate
stitch length
s
(1 mm, 2 mm, 5 mm), which is the distance between
two stitches, mostly aecting the compression of yarn and fabric
at rest
2
(cf. Figure 4c). We also tested dierent modes of electrode
layering, meaning if both electrodes are stitched (i) on top, or (ii) be-
low the CARBOTEX sheet, or (iii) in mixed, i.e. alternating order (cf.
Figure 4d). As discussed later, some of our pattern layouts feature
electrode intersections and double stitches, therefore we also tested
those: intersections (none vs. one) require mixed electrode layering
and are established by guiding the upper electrode across the lower
one (cf. Figure 4e). Double stitches (single vs. double) refer to parts
of the electrode trace that are stitched twice, e.g. by performing a
U-turn and following the exact same trace backwards (cf. Figure
4f).
4.1 Apparatus
For our experiments, we manufactured patches consisting of two
parallel straight electrodes of equal length and xed the CARBOTEX
resistive fabric with a non-functional thread (65
×
65 mm
2
). We
fabricated 6 samples for each variation to cancel out fabrication
imprecisions in our data, resulting in an overall of 108 patches. Note
that for testing electrode lengths, we varied the size of the resistive
sheet instead of the electrode and xed it only at the outer sides,
left and right (cf. Figure 4b). We did this to precisely control the
eective trace length, i.e. the extent of the stitch intersecting with
the CARBOTEX, so imprecisely trimmed seam would not aect the
results.
For evaluation, we mounted each of the samples at on a rubber
sheet for good support, and placed a PMMA-plate (60
×
60 mm
2
, so it
would not rest on the xation stitch) on top, for equally distributing
pressure all over the sensor area (cf. Figure 4a). All sample patches
were generated with a Tajima SAI
3
, in combination with the Creator
software from Pulse Microsystems Ltd.
2
Machine controlled thread tension is also contributing to compression, however it
has to be properly balanced to yield a sound stitch and therefore allows little scope for
variation.
3MDP-S0801C
4.2 Procedure
For measuring resistance values, we used a Tektronix Keithley
2401 SourceMeter Instrument, attached with crocodile clips for
reliable and consistent connection. We rst measured each sample
without pressure applied, then we put on the PMMA-plate (12 g)
and successively added weights of 100 g, 200 g, 500 g, 1000 g, and
2000 g on top of the plate. In order to compensate for drift, we
waited for 5 s before removing and placing the next weight. For
data analysis, we chose a measurement value after 3 s of inaction,
so the resistance value was reasonably settled with RSD
4
< 0.5 % in
±0.5 s.
4.3 Results
In particular, we investigated the resistance dierence between
zero-load and maximum weight, henceforth used as a measure for
dynamic range. We judged high ranges as superior since our in-
tention was to implement a textile sensor similar to widespread
FSRs (e.g. FSR 402), which exhibit similar characteristics, and are
therefore easily applied in combination with straightforward mea-
surement electronics (e.g. via voltage dividers). Another aspect we
consider is the resistance when the sensor is not stressed, which
we call the resting resistance. While not necessarily a measure of
sensor performance, it may aect the choice of readout electron-
ics. Beyond absolute values, the relative dynamic range provides
information about the sensitivity at low and high pressure levels.
We calculate relative values by normalizing measurements by the
respective resting resistance.
Our results are presented in Figure 5. Overall, highest resistance
values were observed for patches with electrode layering methods
mixed and below, with a high RSD of up to 69 % at rest. Otherwise,
RSD within patch conguration was mostly below 20 %, with an
exception of patches for testing electrode length, which showed
noticeably high RSD of 18 % to 31 %. Samples with
d
= 1 mm were
also highly erratic (RSD = 15 %), which we speculate is a result of
their tendency to short, especially when stressed, as their electrodes
are too close. Due to those inconsistencies, we decided to remove
outliers for further data analysis, so we deleted two measurements
per conguration at maximum, which resulted in 12 % of our total
patches.
4.3.1 Electrode Distance. We observed a strikingly linear relation
between resistance and electrode distance, conrming factor
d
in
Equation 1. Dynamic range increased slightly with higher electrode
distances, providing benecial conditions for sensing.
Based on the presented results, we recommend to use this pa-
rameter to control the resting resistance. Furthermore, we infer
that high electrode distances are preferable, as they provide higher
dynamic range. The maximum distance is however depending on
the use case, since it should not exceed the minimum actuator size.
4.3.2 Electrode length. The data conrms the inverse relation of
resistance and electrode length
l
in Equation 1. Our measurements
show inconsistencies though, which we attribute to manufacturing
imprecisions.
4
We use the Relative Standard Deviation for our evaluations, as it is more aligned with
the relevant notion of signal-to-noise ratio.

Embroidered Resistive Pressure Sensors: a Novel Approach for Textile Interfaces CHI ’20, April25–30, 2020, Honolulu, HI, USA
(f)
(e)
(d)
s
(c)
(b)
l
(a)
d
F
actuator
l
Figure 4: Variables of the sensor evaluation apparatus: electrode distance (a), electrode length (b), stitch length (c), electrode
layering (d), electrode intersections (e), and double stitches (f).
We derive that electrode length can be used to counterbalance
electrode distance, when designing a sensor patch. E.g. when higher
electrode distance is required to ll larger sensor areas, increas-
ing the electrode length is a means of cancelling out the rise of
resistance, provided the nature of the pattern allows to do so.
4.3.3 Stitch length. Results of the stitch length experiment showed
that patches with longer jumps showed higher resistance values
at rest, which dropped signicantly faster with pressure, when
compared to shorter stitch lengths.
We argue that larger stitch lengths are preferable, as smaller
jumps cause strong contact of yarn and resistive sheet already in
resting state, which results in reduced dynamic range. However,
choosing the stitch lengths too large may result in slack yarn, de-
pending on the material, so it must be chosen with care. In contrast
to electrode distance and length experiments, we observed rela-
tively minor changes in the sensor’s resting resistance when stitch
length is modied (+8 % for 2 mm and +26 % for 5 mm, when
compared against 1 mm patches).
4.3.4 Electrode Layering. We found that electrode layering sub-
stantially inuences the sensing performance.
Patches with both electrodes on top showed the least dynamic
range, since they were already in good contact with the resistive
material at resting state. Samples with mixed order exhibited sig-
nicantly higher drops in resistance as well as higher resting resis-
tances. This trend extended to samples with both electrodes below,
where both dynamic range and resting resistance were highest.
Furthermore, we observed a remarkably high sensitivity, with rel-
ative resistance drops of 61 % (below) and 47 % (mixed), already
when applying the actuator plate (12 g), against only 0.45 % for the
reference patches (on top).
Although the below approach shows highest dynamic range and
lowest sensitivity, the patch is of rather slack quality, as it tends
to wrinkle and lose contact easily, which reects in highly incon-
sistent readings for resting resistance (RSD = 26 %). Consequently,
we consider the mixed approach to be a superior compromise in
practice, as it also shows excellent performance and does not suer
from the issue of wrinkling, since the resistive material is xed
already by stitching an electrode on top. Additionally, it enables
sensor patterns featuring intersecting electrode traces.
4.3.5 Intersections. Samples with a single intersection in general
showed lower resistance when compared to ones with no intersec-
tions, with resistance drops of 18 % (resting), up to 58 % (2000 g).
This result was expected due to the electrodes’ low distance at the
intersection point. Also, the dynamic range is aected, rising by
15 % in our tests, when comparing absolute dierences between 0 g
and 2000 g.
Summarizing, we found that intersections have to be handled
with care, as they pose signicant inuence on both resting resis-
tance and dynamic range. We therefore recommend keeping the
intersection count as low as possible.
4.3.6 Double Stitches. Double stitched electrodes also signicantly
aect both dynamic range and resting resistance. Resistance values
drop by 28 % (2000 g), up to 32 % (resting). Dynamic range (again
comparing absolute dierences between 0 g and 2000 g) drops by
60 %, when contrasted to the single stitched patches. We hypothe-
size this reduction in resting resistance and dynamic range results
from more compression of yarn and resistive sheet, which coincides
with similar observations for a small stitch length, as elaborated
above.
In light of these results, we recommend avoiding trace sections
requiring double stitches as much as possible.

CHI ’20, April25–30, 2020, Honolulu, HI, USA Aigner, Pointner, Preindl, et al.
0 100 200 500 1000 2000
weight [g]
0.5
1
1.5
2
2.5
3
3.5
4
on top mixed below
R [kΩ]
Electrode Layering
0.60
0.65
0.70
0.75
0.80
0.85
0.90
0.95
125
R [kΩ]
Stitch Length [mm]
0
0.5
1
1.5
2
2.5
3
3.5
4
4.5
5
0 10 20 30
R [kΩ]
d [mm]
Electrode Distance
0
0.5
1
1.5
2
2.5
3
3.5
4
5 25 45 65
R [kΩ]
l [mm]
Electrode Length
0.20
0.40
0.60
0.80
1.00
1.20
1.40
none one
R [kΩ]
Intersections
0.45
0.5
0.55
0.6
0.65
0.7
0.75
0.8
single double
R [kΩ]
Double Stitches
F
s
F
Figure 5: The results of our experiment with respect to the
investigated parameters. Note that weight values on x-axes
do not include the 12 g of the actuator plate.
5 SPACE-FILLING PATTERNS
As discussed earlier, we aim for nding sensor layouts, in order
to cover multi-scale areas. Henceforth, we will call these layouts
patterns, which are composed of two electrode traces (cf. Figure 6a).
Depending on the geometric features, the resulting patterns may be
used either as single sensor patches or as multiple linked (i.e. array
or matrix) sensors (cf. Figure 6b). From related work, we identied
the following, most essential requirements for our sensor layout:
Layout of alternating traces:
To avoid areas with less sensi-
tivity, the pattern should maximize the intervals where one
electrode’s trace is directly adjacent to the other electrode’s
trace, in contrast to running in parallel to itself.
Minimize segments requiring double stitches:
Sections of
double stitched traces (cf. Figure 6c) need to be minimized, as
they not only require additional yarn but may also weaken
the structural integrity of the resistive material, e.g. by per-
forating the sheet when stitch length is chosen too low. Also,
we discovered an overall resistance drop, as can be seen
in the previous chapter, eectively reducing the sensor’s
dynamic range.
Suitability for linkage of multiple sensors:
Start and end
of electrode traces are preferably located at the area’s perime-
ter, so wiring of separate elements (e.g. electronics or adja-
cent patches) can be performed easily.
Uniform responsiveness:
Ideally, the pattern provides a uni-
form, location-independent pressure response across the
whole sensor area.
Feasibility for arbitrary size:
Patterns allowing for arbitrary
dimension are preferred, so patches can be easily adapted to
t in spatially constraining scenarios.
Minimum length of tracks:
The lengths of electrode traces
should be minimized, as additionally yarn increases costs
and manufacturing time.
Support of arbitrary outlines and topologies:
For maximum
exibility, the pattern should be adjustable for any outline
shape, ideally even for concave shapes. In a best-case sce-
nario, the pattern also supports arbitrary topologies, thus
supporting shapes with holes and even nested sensitive
areas.
5.1 Candidate Patterns
To infer an optimal layout, we investigated arrangements used
in prior work [
3
,
37
] in terms of the presented requirements. We
closely examined several patterns (cf. Figure 7), including (i) In-
terdigitated Electrodes, (ii) Boustrophedon Path, (iii) Meander, (iv)
Fermat Spiral, and (v) patterns derived from fractal space-lling
curves, in particular from Hilbert Curve.
In order to generalize these patterns, we introduce the parameter
trace distance
e
(cf. Figure 7), which is the minimum orthogonal
distance between electrode traces. Note that due to the rather com-
plex electrical implications of the patterns’ geometries, both trace
lengths and trace distance do not directly translate to the previously
used parameters electrode length
l
and distance
d
from Equation 1.
5.1.1 Interdigitated Electrodes. We included the Interdigitated Elec-
trodes (IDE), as it is widely used in commercial (printed) FSR sensors
(cf. Figure 7a). It shows uniform responsiveness all over the area and
can easily be adapted for many outline shapes. While optimal for
printed sensors, in the context of embroidery there is the downside
of large sections requiring double stitching, i.e. along the "prongs".
5.1.2 Boustrophedon Path. The Boustrophedon pattern features
numerous electrode intersections, uniformly distributed across the
sensor area, but does not require double stitches (cf. Figure 7b). Due
to its traces running vertically and horizontally, a matrix layout
can easily be implemented.
5.1.3 Meander. Inspired by the Greek fret design, the Meander
pattern is widely used in capacitive sensors [
18
]. The pattern is
comparatively yarn saving as it does not require double stitches
whatsoever (cf. Figure 7c). A critical disadvantage is its highly non-
uniform responsiveness due to the electrode intersection in the
center, which prevents inferring objective pressure data. Like the
Boustrophedon however, it can be easily used to create a matrix
layout.

Embroidered Resistive Pressure Sensors: a Novel Approach for Textile Interfaces CHI ’20, April25–30, 2020, Honolulu, HI, USA
a b c
Figure 6: Top and bottom sides of the Hilbert Curve pattern (a). Depending on the pattern design, multiple tiles can be linked
to arrays or matrices (b). Also, dierent scales can be combined, to gain areas with higher resolution. Single stitched electrode
traces should be preferred over double stitched ones (c).
e
upper electrode
lower electrode
e
eeee
ee
e
e
Interdigitated Electrodes Boustrophedon Meander Fermat Spiral Hilbert Curve
Figure 7: We investigated ve dierent patterns, using 3 dierent sensor area sizes, while keeping trace distance econstant.
5.1.4 Fermat Spiral. A straightforward adaption of the Meander
pattern, avoiding the unfavorable intersection, is the Fermat Spiral,
with a division in the center to yield two separate electrode traces
(cf. Figure 7d). In case of linking multiple patches, the trace ends
should be located at the perimeter; therefore, the electrode must be
stitched in both ways, thus requiring double stitching of the entire
traces.
5.1.5 Fractal Space-Filling Curves. In geometry, numerous can-
didate space-lling curves exist, such as Peano, Hilbert, Moore,
Sierpiński, Morton etc. Arguably, most of these patterns are very
much interchangeable in our use case, since the respective elec-
trodes’ layout on the sensor area may be of little relevance, as long
as trace lengths and trace distance are similar. Consequently, we
selected based on a manufacturing point of view, e.g. Gosper Curve
features a non-quadratic footprint and Morton Curve is hard to
complement with a second electrode. We further evaluated both
Peano and Hilbert curves and ultimately decided to choose the latter
as its Euclidean length grows with curve order with 2
n
, in contrast
to the Peano’s, which grows with 3
n
. Therefore, the Hilbert curve
is superior to the Peano curve in terms of scaling behavior, while
similar in stitching trace and yarn usage.
6 PATTERN EVALUATION
We evaluated the ve chosen patterns in a formal study. In total,
we implemented 15 dierent samples (5 patterns
×
3 sizes) and fab-
ricated 3 samples for each to cancel out fabrication imprecisions in
our data, resulting in an overall of 45 patches. We chose mixed elec-
trode layering for all the pattern types, due to its good performance
in the sensor evaluation, but also to be consistent with Meander
and Boustrophedon, which inherently require this design, given
the intersections. For all of the pattern types, we manufactured the
most basic variants, thus disregarding additional double stitches
of Fermat Spirals and Hilbert Curves, which would be required for
linking multiple tiles.
Similar to the sensor evaluation, we xed the CARBOTEX resis-
tive fabric with a square of non-conductive yarn (42
×
42 mm
2
). For
evaluation, we mounted each of the samples at on rubber sheet,
and placed a PMMA-plate (5 g, 40
×
40 mm
2
, so it would not rest
on the xation stitch) on top, for equally distributing pressure all

CHI ’20, April25–30, 2020, Honolulu, HI, USA Aigner, Pointner, Preindl, et al.
0
200
400
600
800
1000
1200
0 500 1000 1500 2000
R [Ω]
weight [g]
IDE
0
20
40
60
80
100
120
140
160
0 500 1000 1500 2000
R [Ω]
weight [g]
Boustrophedon
0
100
200
300
400
500
600
700
800
0 500 1000 1500 2000
R [Ω]
weight [g]
Meander
0
200
400
600
800
1000
1200
1400
1600
0 500 1000 1500 2000
R [Ω]
weight [g]
Fermat Spiral
0
200
400
600
800
1000
1200
0 500 1000 1500 2000
R [Ω]
weight [g]
Hilbert
8
16
32
width [mm]
Figure 8: The results of our patches experiments show that Boustrophedon has particularly low resistance values. Especially
trends for IDE, Fermat Spiral, and Hilbert Curve atten out beyond 1000 g, making it particularly dicult to distinguish
between high pressure values. Note that weight values on x-axes do not include the 5 g of the actuator plate.
over the sensor area. Again we placed weights of 100 g, 200 g, 500 g,
1000 g, and 2000 g.
Building on our earlier ndings, we chose trace distance
5e
= 2 mm
and stitch length
6s
= 4 mm, and varied sizes with
w
= 8 mm, 16 mm,
and 32 mm, yielding three dierent curve orders (cf. Figure 7).
6.1 Results and Discussion
Overall, we did not remove outliers, as RSD was < 0.2% for all
measurements. Figure 8 provides an overview of the results of the
mean resistance values using dierent pressure levels. We observed
that Fermat Spiral showed the highest dynamic range; the resistance
dropped by 853
Ω
, 177
Ω
, and 47
Ω
for the sizes 8 mm, 16 mm,
and 32 mm respectively. Contrarily, relative dynamic ranges of all
patterns are in similar ranges. As expected, Boustrophedon showed
a very low resting resistance of 150
Ω
, 76
Ω
, and 77
Ω
for pattern
sizes of 8 mm, 16 mm, and 32 mm, due to numerous electrode
intersections. This results in a particularly low signal-to-noise ratio
and potentially challenges the measurement electronics. Similarly,
the 32 mm variant of the IDE patch showed very low resistance
values, starting at 81 Ωwhen no load applied.
We also noticed that the smallest samples (8 mm) showed high-
est deviation. We speculate that this is a result of manufacturing
imprecisions, causing higher contributions to errors due to shorter
traces, while similar aws are canceled out in larger samples.
As expected, overall resistance dropped quickly with patch width,
due to the quadratic growth of area and therefore trace lengths. In
contrast to our previous experiment, we were not able to observe a
proportional relation between resistance and electrode length. This
seems reasonable given the high complexity of the current ow,
resulting from intertwined electrode traces, high numbers of trace
corners etc. We plan to further evaluate this in future work.
7 DESIGN RECOMMENDATIONS
Based on our two experiments that conrmed existing knowledge
and brought new insights, we present the following design im-
plications, for designers to consider when implementing pressure
sensors using an embroidery machine.
5
Note that with the exception of Boustrophedon,
e
is similar to electrode distance
d
as used in the sensor evaluation, however it does not directly translate.
6
We state maximum stitch length, since not all stitches can have the same length;
pattern layouts may occasionally dictate shorter stitches due to trace corners, dead
ends, and intersections.
1 mm 1 mm1 mm
a b c
Figure 9: A too low stitch length may perforate the resistive
sheet, seriously aecting its durability (a). When electrode
yarn is trimmed, hardly visible fringes may cause shorts, so
they must be carefully removed (b). Stitches must meet a
reasonable distance and bobbin thread tension must be cho-
sen accordingly, so the electrodes’ loops at the patches’ back-
sides do not short (c).
7.1 D1: Careful embroidery process
In general, embroidery is a very stressful process for both conduc-
tive and bobbin threads. Consequently, it causes yarn breakages,
resulting in segmented circuit traces [
12
,
26
]. Therefore, it is im-
portant to nd a conductive thread which is explicitly designed
to be used with an embroidery machine. As mentioned before, we
found the Madeira HC40 as the most promising conductive yarn
and during the implementation of all prototypes and samples, it
has never broken. Additionally, we reduced the embroidery speed
from a maximum of 800 stitches/min to 300 stitches/min, to avoid
generated heat and friction. On the other hand, we noticed too low
speeds causing the embroidery frame to oscillate, impairing the
precision of stitches.
7.2 D2: Use a stabilizer
Stabilizer fabrics are used to enhance production quality and stabil-
ity by reducing shimmy eects. Additionally, they provide adequate
rigidity, which improves the sensor measurement quality. During
our experiments, we tested numerous stabilizer fabrics that were
placed at the bottom of the fabric. Even though it can be removed
after the embroidery process, we keep it attached, since it prevents
from warping the sensor, which would introduce noise in the resis-
tance readings.

Embroidered Resistive Pressure Sensors: a Novel Approach for Textile Interfaces CHI ’20, April25–30, 2020, Honolulu, HI, USA
Figure 10: The Meander pattern can be used to build sen-
sor matrices, as can any pattern with electrode intersections.
Stitching a second, slightly oset backup trace for each elec-
trode can improve durability.
7.3 D3: Optimize stitching settings
Depending on base and resistive materials, short stitches may be
problematic as they would harm (e.g. perforate) the fabric (cf. Fig-
ure 9a). We established a borderline stitch length of 1 mm for the
stitch we used in our test. Larger stitch lengths of 4 mm proved to
be optimal for our sensor patches, as they resulted in sensors with
reasonable dynamic range.
Thread tension balance should be considered to achieve a clean
stitch backside. This is opposed to conventional embroidery for
aesthetic means, where the backside is purely functional. Usually,
the bobbin thread is supposed to hold the top yarn in place, e.g. by
providing inwards tension in corners. In our case however, a clear
trace is also required on the backside, as electrode yarn pulled out
of place by the bobbin may cause shorts (cf. Figure 9c). Obviously,
this issue is most prominent when using small electrode distances.
7.4 D4: Manufacture and nish carefully to
avoid aws
Trimmed yarn must be thoroughly veried and nished, as resid-
ual yarn is causing shorts. Some machines provide related system
settings, however residue was still a few centimeters in length in
our case. Even when we cut manually, hardly visible fringes easily
cause shorts (cf. Figure 9b). This manual eort may be avoided by
adapting the design to trim at an adequate distance or at safe areas,
if possible.
For improved reliability, electrode traces may be stitched with
two traces, running next to each other (cf. Figure 10). This way, a
backup trace is added, so the sensor remains functional in case the
electrode thread breaks. However, implications of this action must
be considered, as it eectively reduces the trace distance and cuts
the electrode resistances in half. Note that in contrast to double
stitches, the two traces here run slightly oset, minimizing the
damage inicted on the resistive material.
7.5 D5: Choose an appropriate pattern
In our experiments, we showed that the sensing behavior of the
patch can largely be controlled by choosing the pattern design,
i.e. without adjusting components’ materials or manufacturing
procedures whatsoever. This gives some freedom of choice, since
parameters can be adjusted so the resulting resistance range ts
the requirements of the readout electronics. However, as we at-
tribute the majority of pressure sensitivity to the interface eect,
the electrode distance must be smaller than the minimum actuator
size, since pressing between two traces will not reect in measured
resistance, and therefore dead spots will arise. The minimum trace
distance is however dictated by the base and resistive materials’
structure, e.g. yarn count—for our materials, we identied a 2 mm
limit, unless mixed electrode layering is used, where even smaller
distances may be feasible.
When trying new pattern designs, we recommend doing so with
our earlier formulated layout requirements in mind.
7.6 D6: Linking of sensor patterns
The patterns presented in this paper pose dierent tness for link-
ing, which is a result of their electrode traces. Meander and Bous-
trophedon are suitable for building sensor matrices, as their traces
run in orthogonal manner, thus vertically vs. horizontally, which
is due to their intersection(s) (cf. Figure 10). Traces of IDE and
Fermat Spiral on the other hand run in parallel, which is optimal
for chaining, e.g. to build arrays. While fractal space-lling curves,
such as Hilbert curves, are also chainable, they arguably are more
appropriate for stand-alone patches.
7.7 D7: Fixing the hardware on the textile
To avoid fragile connectors, we recommend mounting the hardware
to the fabric by embroidering (cf. Figure 11). The most critical
part when xing the hardware on the textile with the embroidery
machine is the proper alignment and xation of the board, so the
needle does not punch beyond the hole, causing damage.
8 APPLICATIONS
In this section we show three dierent example applications enabled
by embroidered pressure sensors. To demonstrate the feasibility
and scalability of this approach, we challenged two developers and
one designer to implement three prototypes within one day each.
Beforehand, we developed a custom PCB with connector holes of
3 mm diameter, with the purpose of directly attaching it to the
textile, similar to the LilyPad Arduino mainboard (cf. Figure 11a),
to avoid fragile connections between electrodes and hardware. The
PCB features 16 transmitter and 16 receiver connectors, power and
ground, as well as an I
2
C interface. The transmitter and receiver
ports enable to scan a sensor matrix with a maximum of 16 columns
and 16 rows via a shift register (74HC595D) and a multiplexer
switch (ADG1438BRUZ), resulting in 256 individual sensors. The
board is designed for usage in combination with an ESP32 MCU,
which provides a readout rate of 100 Hz. Measurements are taken
by the MCU’s ADC via a voltage divider, calibrated with a digital
potentiometer (MCPT41050), acting as a pull-down resistor.
The group received the evaluation results of this paper as a
foundation. All three demonstrators were fabricated using tools
commonly available in a fablab.
In a rst application, the group implemented a slider interface,
consisting of eight sequentially chained Fermat Spirals (cf. Figure
11b). Accordingly, the textile did not behave as a smooth slider, but

CHI ’20, April25–30, 2020, Honolulu, HI, USA Aigner, Pointner, Preindl, et al.
a b
c d
Figure 11: Our custom PCB was embroidered directly onto
the fabric (a), to avoid fragile electrode connectors. Based on
our ndings, a simple slider (b), a motorcycle glove (c), and a
seat cover (d) were created for demonstrating combinations
of our sensors at dierent scales.
more like a set of individual buttons, which could be controlled
individually; alternatively, a number of densely placed patterns
(e.g. narrow IDE or Boustrophedon) could yield a higher spatial
resolution.
In a second application, the same group designed and imple-
mented a pressure-sensitive motorcycle glove (cf. Figure 11c). The
team placed a 40
×
40 mm
2
2
nd
order Hilbert pattern on the palm
and 16
×
16 mm
2
1
st
order Hilbert patterns on each of the glove’s
nger tips, as the positions of the pattern’s electrode inlets are ideal
for the wiring conditions at the ngertips. Also, the Hilbert pattern
setup kept the amount of required conductive yarn low. All the
sensor patterns were individually wired to the PCB, as the spatial
separation does not require patch linking.
Finally, to demonstrate a large-scale setup, the group imple-
mented a motorcycle seat cover (cf. Figure 11d), with the overall
goal to analyze the biker’s posture while riding. To provide an
adequate spatial resolution, they linked sensors in a matrix lay-
out for the seating area as well as for left and right thigh. While
the reason for choosing mixed electrode layering in the rst two
was due to the higher resistivity, in this case it was inherently re-
quired, as electrode intersections are necessary to enable the matrix
arrangements.
9 DISCUSSIONS AND LIMITATIONS
We discussed our sensor layout, design recommendations, and po-
tential application scenarios. Expectedly, an embroidered pressure
sensor does show limitations. Most of all, careful manufacturing
is required. We also noticed hysteresis and drift; however, those
eects were negligible, when compared to resistance readings. E.g.
after a 10 s settling time, we observed a drift of just 0.6 % over
a period of 1 h. More details can be found in the supplementary
material.
The main limitation we identied is a rather inconsistent and
non-deterministic performance, when compared to printed sensors.
Reasons for this are found in irregularities of the manufacturing
process, as well as material imperfections on a microscopic level,
but also in physical composition innate to textile materials. E.g.
after crimping or shearing, the sensor will hardly return to its exact
previous state; hence slight shifts in characteristics are expected.
Yet, the patches using space-lling patterns showed low standard
deviation and therefore good part-to-part repeatability.
During our tests, we placed our patches on a rigid substrate, as
otherwise applied force would be dodged, e.g. when applied on a
pillow. Arguably, one of the key benets of textiles – exibility – is
therefore limited, however numerous benecial qualities remain,
e.g. tactile quality. The resulting design possibilities are vast and
beyond the scope of this paper.
Stretchability, which is a quality intrinsic to knits, is limited for
our sensors due to the stabilizer required for the base material.
Stretching would reect in sensor readings indistinguishable from
actual pressure. Consequently, the value of a stretchable pressure
sensor using this approach is questionable.
We mentioned linking of multiple patterns to arrays or matrices,
however this topic raises several challenges that are to be addressed
in future work. Figure 6b shows an example of dierent scaled
Meander patterns, linked on a single patch. Note that merging
multiple smaller tiles into a big one does reduce yarn consumption
but does not reduce readout complexity.
10 CONCLUSION & FUTURE WORK
We demonstrated the process of embroidering pressure sensitive
sensors onto stabilized fabric, harnessing inherent advantages that
derive from textiles, providing omnipresent, unobtrusive, breath-
able, exible, and comfortable surfaces. We demonstrated that the
practice of embroidering sensors provides huge benets for rapid
prototyping and bears great potential for customization.
For future work, we plan to investigate combinations of embroi-
dered sensors with non-functional materials, not only for protec-
tive purposes, but to add valuable tactile qualities, e.g. by using
deformable and elastic foams (e.g. ESD foam) as resistive mate-
rial. Moreover, we plan to improve durability and possibilities for
linking patches and connecting peripheral electronics. Finally, we
strongly believe that adequate authoring tools are very essential
for designers to come up with exciting ideas [
4
]. Therefore, we plan
to implement a software solution for assisting in the processes of
design, manufacturing, calibration, and operation.
We hope that our work, including complementary insights, will
advance, encourage, and inspire the development of this excit-
ing eld, and that it will enable design and research communi-
ties to eectively implement textile pressure sensors by means of
embroidery.
11 ACKNOWLEDGEMENTS
This research is part of the COMET project TextileUX (No. 865791,
which is funded within the framework of COMET – Competence
Centers for Excellent Technologies by BMVIT, BMDW, and the
State of Upper Austria. The COMET program is handled by the
FFG.

Embroidered Resistive Pressure Sensors: a Novel Approach for Textile Interfaces CHI ’20, April25–30, 2020, Honolulu, HI, USA
REFERENCES
[1]
Bahaa Abbas, Salam K. Khamas, Alyani Ismail, and Aduwati Sali. 2019. Full Em-
broidery Designed Electro-Textile Wearable Tag Antenna for WBAN Application.
Sensors 19, 11 (May 2019), 2470. https://doi.org/10.3390/s19112470
[2]
Mohsen Akbari, Ali Tamayol, Veronique Laforte, Nasim Annabi, Alireza Hassani
Najafabadi, Ali Khademhosseini, and David Juncker. 2014. Composite Living
Fibers for Creating Tissue Constructs Using Textile Techniques. Advanced Func-
tional Materials 24, 26 (July 2014), 4060–4067. https://doi.org/10.1002/adfm.
201303655
[3]
Humza Akhtar and Ramakrishna Kakarala. 2014. A comparative analysis of
capacitive touch panel grid designs and interpolation methods. In 2014 IEEE
International Conference on Image Processing (ICIP ’14). IEEE, 5796–5800. https:
//doi.org/10.1109/ICIP.2014.7026172
[4]
Lea Albaugh, Scott Hudson, and Lining Yao. 2019. Digital Fabrication of Soft
Actuated Objects by Machine Knitting. In Proceedings of the 2019 CHI Conference
on Human Factors in Computing Systems (CHI ’19). ACM Press, New York, NY,
USA, 1–13. https://doi.org/10.1145/3290605.3300414
[5]
Nicolas Brechet, Galatee Ginestet, Jeremie Torres, Elham Moradi, Leena Ukkonen,
Toni Bjorninen, and Johanna Virkki. 2017. Cost- and time-eective sewing
patterns for embroidered passive UHF RFID tags. In 2017 International Workshop
on Antenna Technology: Small Antennas, Innovative Structures, and Applications
(iWAT ’17). IEEE, 30–33. https://doi.org/10.1109/IWAT.2017.7915289
[6]
Jingyuan Cheng, Mathias Sundholm, Bo Zhou, Marco Hirsch, and Paul Lukowicz.
2016. Smart-surface: Large scale textile pressure sensors arrays for activity
recognition. Pervasive and Mobile Computing 30 (Aug. 2016), 97–112. https:
//doi.org/10.1016/j.pmcj.2016.01.007
[7]
Maurin Donneaud, Cedric Honnet, and Paul Strohmeier. 2017. Designing a
Multi-Touch eTextile for Music Performances. NIME 2017 Proceedings of the
International Conference on New Interfaces for Musical Expression (2017), 7–12.
https://www.nime.org/proceedings/2017/nime2017_paper0002.pdf
[8]
Franklin N. Evento. 1983. Electronic pressure sensitive force transducer. Patent
No. US4489302A, Filed Sept. 24th., 1979, Issued March 15th., 1983.
[9]
Mikhaila Friske, Shanel Wu, and Laura Devendorf. 2019. AdaCAD: Crafting
Software For Smart Textiles Design. In Proceedings of the 2019 CHI Conference
on Human Factors in Computing Systems (CHI ’19). ACM, New York, NY, USA,
Article 345, 13 pages. https://doi.org/10.1145/3290605.3300575
[10]
Ramyah Gowrishankar and Jussi Mikkonen. 2013. Pattern resistors. In Pro-
ceedings of the 17th annual international symposium on International sympo-
sium on wearable computers (ISWC ’13). ACM Press, New York, NY, USA, 137.
https://doi.org/10.1145/2493988.2494341
[11]
Nur Al-huda Hamdan, Jerey R Blum, Florian Heller, Ravi Kanth Kosuru, and
Jan Borchers. 2016. Grabbing at an angle: menu selection for fabric interfaces. In
ISWC. ACM Press, 1–7. https://doi.org/10.1145/2971763.2971786
[12]
Nur Al-huda Hamdan, Simon Voelker, and Jan Borchers. 2018. Sketch&Stitch. In
Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems
(CHI ’18). ACM Press, New York, NY, USA, 1–13. https://doi.org/10.1145/3173574.
3173656
[13]
Branimir Ivsic, Davor Bonefacic, and Juraj Bartolic. 2013. Considerations on
embroidered textile antennas for wearable applications. IEEE Antennas and
Wireless Propagation Letters 12, January (2013), 1708–1711. https://doi.org/10.
1109/LAWP.2013.2297698
[14]
Thorsten Karrer, Moritz Wittenhagen, Leonhard Lichtschlag, Florian Heller, and
Jan Borchers. 2011. Pinstripe. In Proceedings of the 2011 annual conference on
Human factors in computing systems (CHI ’11). ACM Press, New York, NY, USA,
1313. https://doi.org/10.1145/1978942.1979137
[15]
Joanne Leong, Patrick Parzer, Florian Perteneder, Teo Babic, Christian Rendl,
Anita Vogl, Hubert Egger, Alex Olwal, and Michael Haller. 2016. proCover:
Sensory Augmentation of Prosthetic Limbs Using Smart Textile Covers. In Pro-
ceedings of the 29th Annual Symposium on User Interface Software and Technology
(UIST ’16). 335–346. https://doi.org/10.1145/2984511.2984572
[16]
Lin Shu, Tao Hua, Yangyong Wang, Qiao Li, David Dagan Feng, and Xiaoming
Tao. 2010. In-Shoe Plantar Pressure Measurement and Analysis System Based
on Fabric Pressure Sensing Array. IEEE Transactions on Information Technology
in Biomedicine 14, 3 (May 2010), 767–775. https://doi.org/10.1109/TITB.2009.
2038904
[17]
Torsten Linz, René Vieroth, Christian Dils, Mathias Koch, Tanja Braun,
Karl Friedrich Becker, Christine Kallmayer, and Soon Min Hong. 2008. Em-
broidered Interconnections and Encapsulation for Electronics in Textiles for
Wearable Electronics Applications. Advances in Science and Technology 60 (Sept.
2008), 85–94. https://doi.org/10.4028/www.scientic.net/AST.60.85
[18]
Bob Lee Mackey. 2006. Sensor patterns for a capacitive sensing apparatus. Patent
No. US7129935B2, Filed June 2nd., 2003, Issued Oct. 31st., 2006.
[19]
Viktorija Mecnika, Melanie Hoerr, Ivars Krievins, Stefan Jockenhoevel, and
Thomas Gries. 2015. Technical Embroidery for Smart Textiles: Review. Ma-
terials Science. Textile and Clothing Technology 9 (Nov. 2015), 56. https://doi.org/
10.7250/mstct.2014.009
[20]
Jan Meyer, Bert Arnrich, Johannes Schumm, and Gerhard Troster. 2010. Design
and Modeling of a Textile Pressure Sensor for Sitting Posture Classication. IEEE
Sensors Journal 10, 8 (Aug. 2010), 1391–1398. https://doi.org/10.1109/JSEN.2009.
2037330
[21]
Robert H. Morris, Glen McHale, Tilak Dias, and Michael I. Newton. 2013. Em-
broidered coils for magnetic resonance sensors. Electronics 2, 2 (2013), 168–177.
https://doi.org/10.3390/electronics2020168
[22]
Patrick Parzer, Florian Perteneder, Kathrin Probst, Christian Rendl, Joanne
Leong, Sarah Schuetz, Anita Vogl, Reinhard Schwoediauer, Martin Kaltenbrunner,
Siegfried Bauer, and Michael Haller. 2018. RESi: A Highly Flexible, Pressure-
Sensitive, Imperceptible Textile Interface Based on Resistive Yarns. In Proceedings
of the 31st Annual ACM Symposium on User Interface Software and Technology
(UIST ’18). ACM, New York, NY, USA, 745–756. https://doi.org/10.1145/3242587.
3242664
[23]
Patrick Parzer, Kathrin Probst, Teo Babic, Christian Rendl, Anita Vogl, Alex
Olwal, and Michael Haller. 2016. FlexTiles: A Flexible, Stretchable, Formable,
Pressure-Sensitive, Tactile Input Sensor. In Proceedings of the 2016 CHI Conference
Extended Abstracts on Human Factors in Computing Systems (CHI ’16). 3754–3757.
https://doi.org/10.1145/2851581.2890253
[24]
Patrick Parzer, Adwait Sharma, Anita Vogl, Alex Olwal, and Michael Haller. 2017.
SmartSleeve: Real-Time Sensing of Surface and Deformation Gestures on Flexible,
Interactive Textiles, Using a Hybrid Gesture Detection Pipeline. In Proceedings
of the 30th Annual ACM Symposium on User Interface Software and Technology
(UIST ’17). https://doi.org/10.1145/3126594.3126652
[25] Irene Posch and Ebru Kurbak. 2016. CRAFTED LOGIC Towards Hand-Crafting
a Computer. In Proceedings of the 2016 CHI Conference Extended Abstracts on
Human Factors in Computing Systems (CHI EA ’16). ACM, New York, NY, USA,
3881–3884. https://doi.org/10.1145/2851581.2891101
[26]
Ernest Rehmi Post, Margaret Orth, Peter Russo, and Neil Gershenfeld. 2000. E-
broidery: Design and fabrication of textile-based computing. IBM Systems Journal
39, 3.4 (2000), 840–860. https://doi.org/10.1147/sj.393.0840
[27]
Mahsan Rofouei, Wenyao Xu, and Majid Sarrafzadeh. 2010. Computing with
Uncertainty in a Smart Textile Surface for Object Recognition. In Proceedings
of the IEEE International Conference on Multisensor Fusion and Integration for
Intelligent Systems. 174–179. https://doi.org/10.1109/MFI.2010.5604473
[28]
Jose Saenz-Cogollo, Massimiliano Pau, Beatrice Fraboni, and Annalisa Bonglio.
2016. Pressure Mapping Mat for Tele-Home Care Applications. Sensors 16, 3
(March 2016), 365. https://doi.org/10.3390/s16030365
[29]
Stefan Schneegass and Alexandra Voit. 2016. GestureSleeve: using touch sensitive
fabrics for gestural input on the forearm for controlling smartwatches. Proceedings
of the 2016 ACM International Symposium on Wearable Computers (ISWC ’16),
108–115. https://doi.org/10.1145/2971763.2971797
[30]
Ali Shafti, Roger B. Ribas Manero, Amanda M. Borg, Kaspar Althoefer, and
Matthew J. Howard. 2017. Embroidered Electromyography: A Systematic Design
Guide. IEEE Transactions on Neural Systems and Rehabilitation Engineering 25, 9
(2017), 1472–1480. https://doi.org/10.1109/TNSRE.2016.2633506
[31]
Axel Spickenheuer, Martin Schulz, Konrad Gliesche, and Gert Heinrich. 2008.
Using tailored bre placement technology for stress adapted design of composite
structures. Plastics, Rubber and Composites 37, 5 (June 2008), 227–232. https:
//doi.org/10.1179/174328908X309448
[32]
Mathias Sundholm, Jingyuan Cheng, Bo Zhou, Akash Sethi, and Paul Lukow-
icz. 2014. Smart-mat: Recognizing and counting gym exercises with low-cost
resistive pressure sensing matrix. Proceedings of the 2014 ACM International Joint
Conference on Pervasive and Ubiquitous Computing (UbiComp ’14) (2014), 373–382.
https://doi.org/10.1145/2632048.2636088
[33]
Aris Tsolis, William Whittow, Antonis Alexandridis, and J. Vardaxoglou. 2014.
Embroidery and Related Manufacturing Techniques for Wearable Antennas:
Challenges and Opportunities. Electronics 3, 2 (May 2014), 314–338. https:
//doi.org/10.3390/electronics3020314
[34]
Mark Weiser. 1999. The Computer for the 21st Century. SIGMOBILE Mob. Comput.
Commun. Rev. 3, 3 (July 1999), 3–11. https://doi.org/10.1145/329124.329126
[35]
Karsten Weiss and Heinz Wörn. 2005. The working principle of resistive tactile
sensor cells. In IEEE International Conference Mechatronics and Automation. IEEE,
471–476. https://doi.org/10.1109/ICMA.2005.1626593
[36]
Wenyao Xu, Ming Chun Huang, Navid Amini, Lei He, and Majid Sarrafzadeh.
2013. ECushion: A textile pressure sensor array design and calibration for
sitting posture analysis. IEEE Sensors Journal 13, 10 (2013), 3926–3934. https:
//doi.org/10.1109/JSEN.2013.2259589
[37]
Haisen Zhao, Fanglin Gu, Qi-Xing Huang, Jorge Garcia, Yong Chen, Changhe
Tu, Bedrich Benes, Hao Zhang, Daniel Cohen-Or, and Baoquan Chen. 2016.
Connected Fermat Spirals for Layered Fabrication. ACM Trans. Graph. 35, 4,
Article 100 (July 2016), 10 pages. https://doi.org/10.1145/2897824.2925958
[38]
Bo Zhou, Mathias Sundholm, Jingyuan Cheng, Heber Cruz, and Paul Lukowicz.
2016. Never skip leg day: A novel wearable approach to monitoring gym leg
exercises. In 2016 IEEE International Conference on Pervasive Computing and
Communications. IEEE, 1–9. https://doi.org/10.1109/PERCOM.2016.7456520