nt face=arial size=-1 color=black>

STReSS: A Practical Tactile Display System with One Millimeter Spatial Resolution and 700 Hz Refresh Rate
STReSS: A Practical Tactile Display System
with One Millimeter Spatial Resolution
and 700 Hz Refresh Rate
J´
er
ome Pasquero and Vincent Hayward
Centre for Intelligent Machines & Dept. of Electrical and Computer Eng.
McGill University, Montr´
eal, Qc, H3A 2A7, Canada
{jay,hayward}@cim.mcgill.ca
Abstract. A tactile display system is described which can produce tac-
tile movies, that is, rapid sequences of tactile images refreshed at a rate
of 700 Hz. The display uses an array of one hundred laterally moving
skin contactors designed to create a time-varying programmable strain
eld at the skin surface. The density of the array is of one contactor per
millimeter square, resulting in a device with high spatial and temporal
resolution. The paper describes the construction method and the drive
electronics. It also reports informally on initial test patterns and on the
resulting tactile sensations.
1
Introduction
Tactile displays are devices meant to articially create sensations that resemble,
for example, those arising from sliding a ngertip on a textured surface, or from
brushing over Braille characters. For a variety of reasons, such devices are not
yet available. Nevertheless, the demand for tactile displays capable of creating
a reasonable subset of the gamut of all possible tactile sensations is signicant.
Potential applications include virtual training for surgeons, remotely touching
materials via the internet, sensory substitution devices among many others.
While most tactile displays are designed to stimulate the skin of the ngertip,
there exists vibrotactile devices that cause tactile sensations at various places of
the body including the back, the arm, the phalanges or the feet [12, 17, 5, 24].
Other displays often employ miniature transducers arrays to cause tactile sen-
sation via skin indentation and other methods. Typical stimulation mechanisms
involve arrays of moveable pins or inatable miniature bladders to either indent
the skin or vibrate it locally [16, 6, 27]. Actuation techniques include electrome-
chanical actuators, piezoceramics, servomotors, shape memory alloys, uids, and
others [11, 29, 18]. Other systems, generate friction electrostatically when a user
slides a nger over the display [20]. Some devices avoid direct solid contact with
the skin by using air jets [1]. Many devices are or were based on electrostimula-
tion, e.g. [2]. Devices that use electrogel bristles brushing against the skin have
been introduced [13]. Yet some others require miniature magnets to be glued to
Proc. of Eurohaptics 2003. Dublin, Ireland, July 2003. pp. pp. 94-110 the skin for electromagnetic activation [26]. For a more complete survey, please
refer to [23].
In [8], it was suggested that normal skin indentation and normal vibrations
are not essential to many tactile sensations. Lateral skin strain induced by an
array of discrete contactors moving laterally can give rise to a variety of sensa-
tions, including those given by small-scale geometrical features moving against
the skin. This observation motivated the development of a tactile display ar-
ray which relied on the deformation of a membrane to determine the swinging
movements of a dense array of pins [8]. In the intervening period, it was found
that inherent structural and manufacturing limitations made it dicult to use
this technique to achieve sucient strain levels at the skin surface [19]. This and
other factors prompted the development of a new generation of surface strain
devices. This paper introduces the Stimulator for Tactile Receptors by Skin
Stretch or STReSS, an ecient and practical tactile device which achieves high
spatial and temporal resolution.
2
Requirements for Practical Tactile Displays
The slow development of tactile displays can be attributed to poor usability.
Most existing systems are cumbersome, bulky, expensive, and often focused on
the optimization of one single feature. In order to be successful, devices should
conform to a large set of requirements and be capable of causing a variety of sen-
sations. Moreover, they should stimulate a reasonable skin area, say of ngertip
size, and be capable of refreshing it at a high rate to match the skins sensitive
bandwidth (about one kHz).
They must be safe. They must be compact so as to be embedded in other
structures such as computer mice or steering wheels. Most crucially, they should
be able to resist prolonged exposure to skin abrasion and be impervious to
various pollutants including dirt and skin secretions.
It is generally accepted that in order to give meaningful tactile information,
displays should conform to a spatial resolution of one transducer per mm
2
. The
central problem is then to pack of the order of 100 robust transducers in about
one cm
3
, each capable of substantial movement at about one kHz.
The skin receptors adapt rapidly as can be readily observed when attending
to the sensations coming from a stationary nger pressing on fabric or on paper.
Therefore, a tactile display which by principle remains stationary with respect
to the skin it contacts, should be capable of replicating the interaction of the
skin sliding on a surface. More generally, it is clear that active touch which
involves voluntary movement provides more information than passive touch
does, although in some cases both result in the same tactile sensation for the
same surface [14]. So ideally, the display should allow the freedom of active
exploration.
The present prototypes were designed empirically because a preferred design
depends on many unknown factors which fall under the following headings:
Proc. of Eurohaptics 2003. Dublin, Ireland, July 2003. pp. pp. 94-110 Biomechanical factors. Fingers are mechanically highly variable. Their prop-
erties vary according to the owners age group, occupation, habits, gender,
and most evidently, plain individual dierences. The skin is both tough and
deformable and has peculiar biomechanical properties [28]. These properties
do not provide much useful design information at this stage of the develop-
ment. Presently, we are interested in large strains (of the order of 10%) for
small 1 mm
2
patches.
Neuroanatomical factors. The nature, distribution, and specicity of various
touch receptors is still an open issue, see for example [22, 21]. Therefore, this
gives little information as to a preferred design.
Psychological and behavioral factors. The respective roles of biomechan-
ics and neuroanatomy to conscious touch perception is also open to debate,
see for example [4]. Previously published studies of the behavioral response
to skin stretch were obtained with dierent stimuli (e.g. only one contac-
tor [3]). The stimuli that we are creating result from an ensemble of orga-
nized time-varying patterns which are presumably subject to further process-
ing centrally or at the periphery (e.g. [7]). This also, provides little design
guidelines.
Cognitive factors. It is highly probable that learning, training, and more gen-
erally, prior knowledge of the sensations given by touch and other modalities
play a role in touch perception.
Factors related to the application. Even if all of the above issues where
settled, designs would depend on the application. For example, a display
designed for representing ne textures would obey dierent rules than one
designed for reading Braille, since an actuator design relies, among others,
on a displacement-bandwidth tradeo.
3
Actuator Array Structure
Piezoceramic actuators constitute a practical choice to build an actuator array.
They can be operated over a large bandwidth and are relatively easy to form
in a desired miniature structure. Moreover, they are widely available from a
variety of sources for a reasonable price. Unfortunately, piezoceramic actuators
still require high operating voltages. However, the technology is mature and
piezoceramic properties continue to improve steadily year after year.
The objective is to create a deformable structure capable of causing pro-
grammable strain elds in a patch of skin in contact with it. The exposed side
of this structure should thus be made of an array of contactors each moveable
tangentially. Among the dierent piezoelectric elements available, bimorphs can
achieve substantial displacement via bending of a cantilever. This bending mo-
tion can directly be used to stretch the skin without the need f