Content uploaded by Florian Heller
Author content
All content in this area was uploaded by Florian Heller on Feb 19, 2018
Content may be subject to copyright.
HapticToolkit - Easily Integrate and
Control Vibration Motor Arrays for
Wearables
Jan Thar
RWTH Aachen University
thar@cs.rwth-aachen.de
Florian Heller
Hasselt University - tUL - imec
florian.heller@uhasselt.be
Sophy Stoenner
RWTH Aachen University
stoenner@cs.rwth-aachen.de
Jan Borchers
RWTH Aachen University
borchers@cs.rwth-aachen.de
Permission to make digital or hard copies of part or all of this work for personal or
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full citation
on the first page. Copyrights for third-party components of this work must be honored.
For all other uses, contact the Owner/Author.
Copyright is held by the owner/author(s).
ISWC’17, September 11–15, 2017, Maui, HI, USA
ACM ISBN 978-1-4503-5188-1/17/09...$15.00
https://doi.org/10.1145/3123021.3123066
Abstract
Haptic input is a common input method for navigation aids
for visual impaired people, leveraging an otherwise unused
sensory channel depending on the body region, but building
systems with large numbers of vibration motors is rather
complex.
For this purpose we developed a system to easily and
quickly build systems with largen numbers of vibration mo-
tors. With only low requirements on manual skills as well
as tools, such wearables with huge numbers of vibration
motors can be reproduced, e.g., at a local Fab Lab or Mak-
erspace.
Author Keywords
e-textiles; wearable computing; construction kit;
ACM Classification Keywords
B.4.m [Input/Output and data communications (e.g., HCI)]:
Toolkit
Introduction
Several approaches using arrays of vibration motors to pro-
duce a tactile image on the human body have been pre-
sented before. From sparse ones (4 ×4 motors) as in [2],
more broader ones like the one for gaming [1] with 4 ×12
motors, to a matrix of 16 ×8 motors as in [3] to directly
249
ISWC '17, SEPTEMBER 11–15, 2017, MAUI, HAWAII, USA
feel a depth image on the abdomen. While all these sys-
tems are reproducible, the amount of work to control large
amounts of vibration motors and implement them into a
wearable form should be reduced. Furthermore weight,
power consumption and noise has to be reduced to be able
to wear such a system.
Figure 1: A belt with 16*8 vibration
motors and a depth camera on the
front side
Figure 2: Detail: Wiring,
3D-printed clips, and
PWM-I/O-expander boards
Design considerations
The toolkit has to simplify the design for wearable haptic
feedback systems, it has therefore to be scalable to differ-
ent layouts, easy to reproduce without special tools, and
furthermore be cost sensitive and working reliably. We
started testing several approaches for driving the vibra-
tion motors. The designs both for electronics as well as
3D-design-files can be found at 1.
Serial connection - WS2811
The WS2811 chip which is used to drive the popular LED
stripes can be used in combination with an additional driv-
ing circuit (transistor and free wheeling diode) to drive a
vibration motor. Either each color is replaced by a motor
or a combination of motor and signal LEDs can be used.
While driving the motors would be rather easy, e.g., using
the existing libraries for WS2812, the electronic consists
of multiple components. This increases costs and size, but
the main disadvantage is the reliability of the system: If one
board fails, the remaining boards that are logicaly behind it
will also not work anymore.
Parallel connection
From a reliability perspective, having all motors with their
logic circuits on a parallel bus system would be the most
promising option, as long as the bus system itself it not
damaged or short-circuited. Having a dedicated micro-
controller for each motor results, again, in rather big and
1https://github.com/JanThar/HapticHardware
expensive circuitry per motor. This can be reduced by driv-
ing multiple vibration motors with one board, partly negating
the parallel set up (resulting in a combination of parallel bus
and star topology). One minor advantage is also the pos-
sibility to drive the motors in more complex patterns due to
the dedicated micro-controller, as well as having the pos-
sibility to use, e.g., linear resonant actuators instead of DC
vibration motors. Still, the complexity to built the system
from the electronic hardware side in large quantities, as well
as the need to program each controller makes such a sys-
tem only feasible for real mass production, not for self made
singular wearables.
I2C connection
We therefore choose the PCA9685 as I2C-I/O-expander.
It is used in the popular Adafruit servo driver board, there-
fore a large number of software examples exist, lessening
the burden for programming. On the PCB we add driver
circuits (ULN2803) for the vibration motors as well as an
adjustable voltage converter. The address of each board
is selected on the backside with solder jumper. Pads on
the side are used to either directly solder wires or solder a
pin header and then use wires with crimped connectors for
easier adaption of the setup if required. Small indentations
on the outer sides are used to fix a 3D-printed clip, which
is used to attach the PCB to the fabric. Alternatively, holes
can be used to sew the board directly on it. The resulting
boards can be seen in Figure 2 on the right side, the PCB
layout is shown in Figure 4. This design was the easiest to
built with a low number of components, and it is reliable and
scalable (from 16 motors with one board up to 992 with 62
boards on one I2C bus).
Motors
Instead of using standard DC vibration motors with external
excenters we wanted to use one with an integrated excen-
250
ISWC '17, SEPTEMBER 11–15, 2017, MAUI, HAWAII, USA
ter to reduce the size of an external housing. The common
pancake vibration motors where not sufficiently reliable
while performing a long test run of 48 hours, therefore we
switched to less common encapsulated vibration motors,
which are still easily and cheaply obtainable. These motors
come with a standard ribbon cable as connection to the I2C
boards.
Figure 3: Backside: Battery pack
and processing unit
Figure 4: PCB-Layout
Assembly
For an easy assembly of the belt, a rectangular hole is
laser cut for each motor. A small 3D-printed housing (in two
halves, see Figure 5) is then used to clip the motor into the
hole. Similar clips are designed to hold the wires in place,
as seen in Figure 2, for a clean design. The use of the clip-
ping system reduces the need of sewing, and allows an
easy change of components if necessary. All 3D-designs
are made for 3D-prints without support material, allowing
the use of basic 3D-printers without the need to manually
remove a support structure afterwards.
Camera, Processing and Battery
For testing the system, we used a Intel RealSense camera
together with an Up board as processing unit, which then
controls the vibration motors over the I2C boards. Since
Up board, camera, and I2C boards with integrated voltage
regulators run on 5V, a common power supply can be used,
again reducing complexity and access for everyday makers,
who should be able to rebuild the system. Both Up-Board
and Battery pack are mounted within 3d-printed housings at
the backside, sewn onto the textile directly above the Velcro
connector, as shown in Figure 3.
Building and evaluation
Instead of using a vest we choose to build a belt with 16
×8 motors as in [3]. The number of motors results in a 4
cm spacing between these, being close to the distinguish-
able resolution of the human abdomen. On the other hand,
the use of 16×PWM-I/O expanders to drive them allows a
simplified setup with this number of motors. The belt form
with all components included allows an easier testing setup
with multiple people in short time, while it is also possible to
wear the belt below clothing with only the small camera out-
side the clothing, rendering the whole system invisible. For
the belt itself, we choose a more visible concept to highlight
the technique from a design perspective, the whole system
can be seen in Figure 7. Since nearly all components are
clipped into (for convenience) laser-cut holes in the belt, a
fast and precise assembly is possible and no special tools
besides a 3D printer are required. The most complicated
step is the assembly of the SMD-components on the driver
boards. While it is possible to manufacture both the boards
and populate them by hand, an automatic assembly or ob-
taining the board might be the easier way.
For wearability at the backside, only flat sides of 3D-printed
parts are in contact with the human body (see Figure 6).
The used 3D-net mesh textile which is highly breathable
and stretchable also improves wearability.
Furthermore, the belt system allows to test the system with
camera and vibration motors on the human back, to enable
a person to feel what happens behind herself without look-
ing back, e.g., as safety feature for industry 4.0.
The belt was now in use on several fairs, proving the sys-
tem to be simple to use (people walk around with closed
eyes after few minutes testing), and both power consump-
tion (roughly 6h with one battery pack, depending on the
average distance to objects over time), and noise are within
acceptable borders. Furthermore, the system runs reliably
for more than 80h on faires alone, not considering testing
phases before and in between without any fixes (besides
software upgrades).
251
DESIGN EXHIBITION
Building other wearables with haptic feedback
Figure 5: 3D-design motor clip
Figure 6: Complete system -
Body-side
Figure 7: Complete system
Designing another wearable with multiple vibration motors
will still begin with choosing the right vibration motor. De-
pending on vibration duration and intensity, even pancake
motors are possible, otherwise choose for the easiest setup
with encapsulated motors. Optimally these do not come
with wires but short springs as contacts for SMD mount-
ing — these were the easiest to solder and most reliable
in our experience. Next, adapt the 3D-printed motor cas-
ing if necessary. We will provide a customizable version for
different motor sizes in a next step. Cutouts for the motors
directly depend of the size of the housing, therefore they
have to be adapted as well if needed. The vector graphic
template provided shows the setup for our vest - for other
wearables a basic outline from a picture or design file can
be used. The cutouts for the motors are then placed at
the desired places, then the cutouts for the PCBs on the
sides are placed (one for 16 vibration motors). Finally, the
wiring can be hold in place with additional cutouts for the
cable clips. Thereafter, the cutouts are cut with a laser or by
hand, the clips and housings are 3D-printed, and a number
of PCB driver board assembled - each with a unique I2C
address configured with a solder jumper on the backside.
Each board connects to 16 vibration motors with wires,
while 4 additional pins are used for I2C data connection
with a central unit, and power line. All PCBs can be con-
nected in parallel. The internal voltage regulator is set to
the normal configuration for 3.3V vibration motors, other
values up to 5V are also possible. This setup allows, as
mentioned, the use of a normal 5V power bank, and only
minor wiring with soldering wires at the vibration motors
and crimped connectors (or soldering) at the other end is
necessary. The PCB is designed for automatic assembly.
Both 3D-printed clips and laser cut holes furthermore dras-
tically reduce assembly time of the whole wearable and also
allows easy maintenance.
Since the driver PCBs use the same I2C I/O expander chip
as the Adafruit Servo motor boards, programming can be
then done, e.g., with Adafruit’s library and examples for the
Arduino, which allows an easy access for programming the
setup for non-experts.
Design and Optic
Overall the design is purely functional — eventually, the
system should be covered with another layer of textile and
become invisible. Therefore, all components are designed
to be as easy as possible to produce with tools one can find
in a local makerspace, if not at home. On the other hand,
cable routing and motor and electronic mounting with 3d-
printed clips allows a clean set up. Highlighting the clips
with orange plastic produces a clearly visible structure,
making the technical parts more visible for the audience,
while both gray cables and textile stays in the background,
which allows a good visibility for exhibitions, where a cov-
ered version would be almost invisible and not easy to un-
derstand for the user. Even with the visible setup, explana-
tions are necessary, and for presentations an even more
visible setup might be interesting, but out of scope of the
toolkit.
Conclusion and Future Work
The proposed setup allows building and controlling of wear-
able systems with large amounts of vibration motors. While
the system is already proven to work stable, it has to be
easier to adapt to different sizes/kinds of vibration motors
to not be constrained to a single supplier. Main obstacle
for everyday-users to built such a system might be solder-
ing the PCBs by hand, which can be easily manufactured
with a pick-and-place machine. All housings and clips are
3D-printable without support material, and laser cutting the
textile (although not required) should also be possible with
access to a Makerspace. Therefore, the resulting system
252
ISWC '17, SEPTEMBER 11–15, 2017, MAUI, HAWAII, USA
can be used as easy prototyping base for wearables with
haptic feedback and large numbers of vibration motors.
For presentation on fairs we will add another layer with
LEDs above the belt to visualize the vibration image for by-
standers, with the reduced noise generation it is already
hard to catch the functionality of the system without wearing
it.
Acknowledgements
This toolkit was funded by the German Federal Ministry
of Education and Research as part of the Photonics Re-
search Germany (13N14065). Special Thanks to David An-
ton Sanchez, who developed the first version of the vest.
REFERENCES
1. Sean Benson. 2015. 3D Haptic Vest for Visually
Impaired and Gamers. (August 2015).
https://hackaday.io/seanbenson
2. Dimitrios Dakopoulos, Sanjay K. Boddhu, and Nikolaos
Bourbakis. 2007. A 2D vibration array as an assistive
device for visually impaired. In Proc. of BIBE ’07. IEEE,
930–937.
3. Philipp Wacker, Chat Wacharamanotham, Daniel
Spelmezan, Jan Thar, David A. Sánchez, René Bohne,
and Jan Borchers. 2016. VibroVision: An On-Body
Tactile Image Guide for the Blind. In Proceedings of the
2016 CHI Conference Extended Abstracts on Human
Factors in Computing Systems (CHI EA ’16). ACM,
New York, NY, USA, 3788–3791. DOI:
http://dx.doi.org/10.1145/2851581.2890254
253
DESIGN EXHIBITION
