HAL Id: hal-00844954
https://hal.archives-ouvertes.fr/hal-00844954
Submitted on 2 Oct 2019
HAL is a multi-disciplinary open access
archive for the deposit and dissemination of sci-
entic research documents, whether they are pub-
lished or not. The documents may come from
teaching and research institutions in France or
abroad, or from public or private research centers.
L’archive ouverte pluridisciplinaire HAL, est
destinée au dépôt et à la diusion de documents
scientiques de niveau recherche, publiés ou non,
émanant des établissements d’enseignement et de
recherche français ou étrangers, des laboratoires
publics ou privés.
Pseudo-haptic feedback: can isometric input devices
simulate force feedback?
Anatole Lécuyer, Sabine Coquillart, Abderrahmane Kheddar, Paul Richard,
Philippe Coiet
To cite this version:
Anatole Lécuyer, Sabine Coquillart, Abderrahmane Kheddar, Paul Richard, Philippe Coiet. Pseudo-
haptic feedback: can isometric input devices simulate force feedback?. VR: Virtual Reality, Mar 2000,
New Brunswick, NJ, United States. pp.83-90, �10.1109/VR.2000.840369�. �hal-00844954�
Pseudo-Haptic Feedback :
Can Isometric Input Devices Simulate Force Feedback?
Anatole L´ecuyer
Aerospatiale Matra CCR
Sabine Coquillart
INRIA Rocquencourt
Abderrahmane Kheddar
CEMIF-SC, UEVE
Paul Richard
LRP, CRIIF
Philippe Coiffet
CNRS, LRP
Abstract
This paper considers whether a passive isometric input
device, such as a
TM
, used together with visual
feedback, could provide the operator with a pseudo-haptic
feedback.
For this aim, two psychophysical experiments have been
conducted. The first experiment consisted of a compliance
discrimination, between two virtual springs hand-operated
by means of the
TM
. In this experiment, the stiff-
ness (or compliance) JND turned out to be 6%. The second
experiment assessed stiffness discrimination between a vir-
tual spring and the equivalent spring in reality. In this case,
the stiffness (or compliance) JND was found to be 13.4%.
These results are consistent with previous outcomes on
manual discrimination of compliance. Consequently, this
consistency reveals that the passive apparatus that was used
can, to some extent, simulate haptic information.
In addition, a final test indicated that the proprioceptive
sense of the subjects was blurred by visual feedback. This
gave them the illusion of using a non isometric device.
1. Introduction
Isotonic or isometric 3D input devices
1
are clever de-
vices for the purpose of 3D interactions and 3D manipula-
tions of objects. They are compatible with almost all ma-
Corresponding author: Anatole L´ecuyer. Aerospatiale Matra,
Centre Commun de Recherche Louis Bl´eriot, 12 rue Pasteur, 92152
Suresnes, France. Email: anatole.lecuyer@aeromatra.com or ana-
tole.lecuyer@inria.fr
Corresponding author: Sabine Coquillart. INRIA, Domaine de
Voluceau, Rocquencourt - BP 105, 78153 Le Chesnay Cedex, France.
Email: sabine.coquillart@inria.fr
1
Zhai classified the input devices into two categories : ISOMETRIC
devices (they offer resistance and stay put while you exert force on them)
and ISOTONIC devices (they offer no significant resistance and are used
to track users as they move around the virtual world) [21].
jor CAD softwares available on the market. Their user-
friendliness has shown their potential usability as a tool
for off-line robot programming and teleoperation, or vir-
tual prototyping. For instance, to teleprogram the Mars
Pathfinder Sojourner robot, the operator uses a
TM
as an input tool along with a virtual reality interface [1].
The
!
TM
is an isometric input device with six de-
grees of freedom (dof) which is now commercialized by the
Spacetec company [3]. The Magellan
!""#%$'&()
TM
[2]
- another 6dof isometric device - was successfully used by
the DLR - the German Space Agency - operators and astro-
nauts to teleoperate a space robot within the context of the
well-known ROTEX experiment [11].
Since isometric or isotonic input interfaces are com-
pletely passive, they have never been regarded as being able
to return forces. How the use of the properties of an isomet-
ric input device, the Spacetec
!"*!+
TM
2003C model,
together with visual feedback to provide force information
to the operator is the subject of the following paragraph.
To begin with, there is to take advantage of the mechani-
cal characteristics of the isometric device : its internal stiff-
ness and its thrust. Those characteristics are combined with
visual feedback to provide a kind of pseudo-force feedback.
For example, let us assume that one manipulates a virtual
cube in a 3D virtual environment (VE). The cube must be
inserted inside a narrow duct. As the cube penetrates the
duct, its speed is reduced. In other words, the
TM
output resolution, which controls the cube motion, is de-
creased. Consequently, the user will instinctively increase
its pressure on the ball which results in the feeding back of
an increasing reaction force by the static device. The cou-
pling between the slowing down of the object on the screen
and the increasing reaction force coming from the device
gives the user the illusion of a force feedback as if a friction
force between the cube and the duct was directly applied to
him.
This “illusion” of force feedback was first qualitatively
estimated with a group of 18 people during an experiment
called the swamp. The subjects were told to manipulate a
virtual cube, displayed on the horizontal plane, and to cross
square areas (see Figure 1). When over these areas, the
speed of the cube was either accelerated or slowed down.
The subjects were told to describe their sensations when the
cube was crossing the areas, and to compare the sensations
they felt when using either the
!
TM
or a classical
2D mouse. While using the
!"*!+
TM
, their accounts re-
Figure 1. Swamp experiment: cube crossing
a slowing down area.
vealed that the subjects felt “something” as the cube crossed
these areas. Most subjects experienced a sense of friction,
gravity or viscosity when the cube’s movement was slowed
down. They found that forces were much more perceptible
with the
!
TM
than with the 2D mouse. The per-
ception of forces was a bit less sharp when the cube was
accelerated. This is probably due to the fact that the reac-
tive force from the
TM
is more perceptible during
compression phases.
Those qualitative indications revealed the potentialities
of this concept, but they did not measure or identify the
characteristics of such an illusion. It was necessary to eval-
uate the feedback more quantitatively. A compliance dis-
crimination task between a real and a virtual spring was
chosen as a simple evaluation task. The real spring was
tested first in the real environment (RE), then the virtual
spring was tested in the VE. The virtual spring was graph-
ically displayed on a computer screen and was dynami-
cally animated when pushing the
TM
ball. If the
!"*!+
TM
, used together with visual feedback, allowsone
to discriminate a virtual spring stiffness from a real one,
then the whole system may thus be fit for feeding haptic in-
formation which was supposedly difficult to provide with-
out a force feedback interface.
First, an overview of previous work in the field of 3D in-
put interfaces evaluation and sensory illusions will be pre-
sented. It will be followed by a description of the exper-
imental system which was set up for the evaluation of the
stiffness of the real and virtual springs. Previous works
concerning compliance or stiffness discrimination and JND
(Just Noticeable Difference) will also be mentioned. The
two following sections describe two psychophysical exper-
iments: the compliance discrimination between two virtual
springs, and the compliance discrimination between a vir-
tual spring and a real one. The first experiment is a VE
evaluation; while the second experiment is the main subject
of the study. The paper ends with a general conclusion and
a reference to further work.
2 Previous work
Force/tactile interfaces have been developed in recent
years [6] in order to provide force/touch feedback to users.
They receive motor actions from the user and send haptic
images to him. These interfaces are used to simulate a wide
range of object dynamics such as hardness and elasticity.
Yet, today, they are still expensive and complex.
The perception of real or virtual environments is not re-
stricted to the intra-sensory interpretation cues. Cues sent
by different senses are somehow interpreted together. For
instance, manipulating objects combines tactile, kinesthetic
sensations and often vision [7]. Given these complexities,
it would seem more appropriate to investigate the “plural-
istic” nature of sensory perception, rather than one isolated
sense. Aldridge [4] observed that the visual representation
of a virtual object has some effect on the integration of the
touch feedback. He stated that further experiments needed
to be carried out in order to explore the extent of such “vi-
sual dominance”.
Previous work on visual dominance showed that multi-
ple cues offer a high level of redundancy and can improve
signal-to-noise ratios. For instance, it has been shown that
lip-reading modifies the auditory cortex, and enhances audi-
tory perception [14]. One interesting issue is how these dif-
ferent sources of information are all combined to form what
might be called holistic perceptions. A famous example of
visual dominance is the Ventriloquist’s effect [19]. Diderot
(1749), offered early support for the existence of sensory
dominance [13]. Several researchers have demonstrated a
dominance of vision over taction [7]. Lee and Lishman pro-
vided evidence that vision plays an integral role in human
stance control (balance). This “visual proprioceptive con-
trol” is shown to dominate over non-visual information. Lee
and Lishman described also the tuning role that visual pro-
prioception plays in learning a new stance (i.e. ankle-foot
proprioception). This suggests that vision plays a major role
in making things feel the way they do.
But vision may sometimes make things feel different
than they are. Katz [7] observed that different materials
(paper, rubber, leather, etc..) can easily be interpreted dif-
ferently by blindfolded subjects. Srinivasan [16] found that
vision could also mislead someone during a compliance dis-
crimination task between two springs . The displacement
of the springs was visually observed on a computer screen,
while springs were pressed manually by means of a me-
chanical apparatus. Srinivasan observed that an inapropri-
ate vision feedback can totally invert the stiffness percep-
tion and the result of the discrimination, which ushers in
the illusion concept.
Illusion plays a central role in a VE perception. Illusion
is a non veridical perception, a mistake made by the brain
and not by our senses. Well-known optical illusions such as
the M¨uller-Lyer illusion are extensively described in scien-
tific works [10]. Some haptic illusions may also be revealed
by simple experiments. For example, Weber first observed
that the temperature of an object influences the haptic per-
ception of its weight: a cold coin seems heavier than the
same coin when warmer [15]. Another haptic illusion is the
size-weight illusion: a large radius ball seems heavier than
a ball of the same weight, but with a smaller radius. Re-
cently, Ellis and Lederman [8] established the size-weight
illusion as a primarily haptic phenomenon, despite its hav-
ing been more traditionally considered an example of vision
influencing haptic processing. The resort to intra and inter-
sensory illusions and dominance can be relevant when used
in VR applications. The following paper concentrates on
the use of these potentialities.
3. Experimental set-up
As already stated, the aim of the experimental system
described is to measure the capacity to feed back haptic in-
formation by means of a passive isometric input device and
vision feedback, which is the key issue of the scheme. It
is coupled with the force applied on the
!"*!+
TM
, and
any change in the visual feedback generates a difference
in force perception. The perception of the stiffness of the
spring involves a multi-modal combination of force and dis-
placement. Thus, it seemed to be an appropriate model to
demonstrate the general concept.
An experiment of compliance discrimination of springs
was then chosen because of the simplicity of the model of
stiffness, relevant previous works on manual discrimina-
tion of compliance, and the fact that springs are a classical
and fundamental element in computer graphics and com-
puter haptics modeling. The stiffness discrimination lies
between a virtual spring and a real one. In the VE case,
the spring displacement is visually displayed on a computer
screen. The force information of the spring stiffness is in-
herent to the reactive force from one’s interaction with the
!"*!+
TM
. Those two independent sensory cues (virtual
displacement motion and reactive force of the
!
TM
)
should allow the user to discriminate the stiffness of a vir-
tual spring from that of a real one. If this works out, the con-
cept is apt to provide haptic information which was a priori
not to be simulated without an actual force feedback device.
Real springs are tested in RE conditions. Each spring looks
like a trumpet piston (see Figure 2). Three real springs were
Figure 2. Real spring embedded in a piston
used with different degrees of stiffness: 249, 363 and 544
N/m. Their stiffness was empirically derived by measur-
ing spring displacement when fixed weights were applied
on each of them. Friction effects inside the real piston were
nearly canceled by directly applying lubricant on the spring
and inside the static part of the piston.
The visual display of the virtual spring is fundamental:
spatial reference must be the same one when comparing
the virtual spring motions and the real ones. The virtual
spring is thus visually displayed on a monoscopic worksta-
tion screen, as similar as possible to the real one (see Figure
3), and of the same size as the real spring. Special attention
was given to many graphical features (color, texturing, etc.)
in order to recreate the virtual piston with the highest pos-
sible realism. A
TM
ball was also rendered on the
left side of the screen to facilitate the comprehension of the
scale factor between VE and RE.
The displacement of the virtual spring
,
virtual
is deduced
from the force applied by the user
-
user
using the well-
known equation 1, in which
.
virtual
is the virtual spring
stiffness. The force applied by the user on the ball is mon-
itored by internal
*
TM
sensors. The
TM
(force applied by user )/(sensors output) profile was man-
ually identified with a dynamometer. A maximum 10%
uncertainty in the output data was observed. A maximum
pushing limit is indicated on the virtual display by a red
mark on the moving part of the piston. It corresponds to the
sensing limit of the
!"*!+
TM
’s force sensors in the case
of the stiffest virtual spring. The red mark is also printed on
the real spring to keep the same visual aspect.
,
virtual
/
-
user
0
.
virtual
(1)
In order to obtain similar tactile and grasping sensations in
the real and virtual cases, the same moving part of the piston
was fixed on the
TM
, by means of two plastic links
(see Figure 4). The grasping of the virtual spring is thus
similar to the real one, thanks to the plastic upper link on
the
TM
on which the subject can put his forefinger
and his middle finger (see Figure 4).
Figure 3. Visual display of a virtual spring
Finally, for the testing of the real spring, subjects grasp
the real piston as shown in Figure 2, and push the moving
part of the piston with an active motion of the thumb. For
the testing of the virtual spring, the subject applies a force
on the
!"*!+
TM
by pushing the moving part of the piston
fixed on the
!"*!+
TM
’s base with the thumb (see Figure
4). He/she looks at the screen so as to see the displacement
of the virtual spring resulting from his/her actions.
Figure 4. “Modified” isometric device
4. Previous work on compliance discrimination
In the RE case, the compliance discrimination has been
widely studied. The Just Noticeable Difference (JND) is
the just detectable increment (or decrement) of intensity for
a specific stimulus. Jones and Hunter [12] found that the
compliance discrimination JND in forearms was 23% of the
intensity of the stimulus. Tan studied the manual discrim-
ination of the compliance with active motion of the finger
by using an electromechanical apparatus called the ‘linear
grasper’. She first found a JND of 8% in the case of a fixed
squeezing distance [18]. But after reducing terminal force
cues by using a roving squeezing distance, the JND reached
up to 22% [18]. When the mechanical work cues were also
eliminated through an equal-work-force-displacement pro-
file, the JND was found to vary between 15% and 99% [17].
In the VE case, Tzafestas found a JND of 44% with the
help of the LRP dextrous hand master which is an exoskele-
ton glove with 14 active dof [20]. The virtual discrimination
was made between two virtual balls displayed on a com-
puter screen. The balls were pressed alternatively with the
thumb and the forefinger of the master glove.
As far as the authors of this paper know, no studies on
compliance discrimination have ever been carried out using
simultaneously a real spring and a virtual one.
5. Compliance discrimination between two vir-
tual springs
It seems at first that the experimental set-up presents
many uncertainties linked to the identification of the
*
TM
’s force/output profile, graphical approxima-
tions, manual evaluation in RE of the stiffness of the
springs, small differences in graspings and frictions be-
tween RE and VE. Therefore the reliability of the VE has to
be taken into account first. In order to evaluate the virtual
model of the spring, a compliance discrimination experi-
ment between two virtual springs is first carried out.
5.1. Experimental procedure
4 people, from the age of 21 to 38, took part in this exper-
iment. There were 3 men and 1 woman with no known per-
ception disorders. All the subjects were right-handed and
used their dominant hand to perform the grasping task.
The psychophysic method used was a constant stimuli
method with a forced choice and (+,-) paradigm (see [9] for
a description of the method). During each trial the subject
had to choose between two virtual springs displayed on the
same computer screen and to say which one of the two was
the stiffer.
Three values of virtual reference stiffness were used:
249, 363 and 544 N/m. Each spring was compared with five
possible stiffer springs whose stiffness varied from the ref-
erence stiffness by +0, +5, +10, +15 and +20 percent. Each
subject tested all the possible pairs. For each subject each
pair appeared 25 times in random order. The total amount of
trials was then 100 a pair and the total amount of trials was
1500. For each trial the reference stiffness was randomly
