A New Control Strategy for the Improvement of Contact Rendering
with Encounter-type Haptic Displays
Oscar De La Cruz Fierro
1,2,3,4
, Wael Bachta
2,3,4
, Florian Gosselin
1
and Guillaume Morel
2,3,4
1
Interactive Robotics Laboratory, CEA, LIST, F-91190, Gif-sur-Yvette, France
2
Institut des Systèmes Intelligents et de Robotique, Sorbonne Universités, UPMC Univ. Paris 06, F-75005, Paris, France
3
Institut des Systèmes Intelligents et de Robotique, CNRS, UMR 7222, F-75005, Paris, France
4
Institut des Systèmes Intelligents et de Robotique, Equipe Agathe, INSERM, ERL U1150, F-75005, Paris, France
Keywords: Robot Control, Haptic Interface, Encounter-type Haptic Display (ETHD).
Abstract: Encounter-type haptic interfaces are used to interact physically with virtual environments. They allow
controlling the position of an avatar in the simulation while perceiving the forces applied on it when it interacts
with the surrounding objects. Contrary to usual force feedback devices, the interface tracks the real user’s
finger without touching it when the user’s finger avatar moves in free space. Only when a contact occurs in
the virtual environment, the interface comes in contact with the user to display the mechanical properties of
the encountered objects. This way, the device’s behaviour is more natural as simulated contacts really occur
in the real world. Existing control laws for such devices exhibit however limitations, especially when contacts
occur at high speed. In such cases, the device tends to bounce against the user’s finger, which decreases the
realism of the interaction. In this paper, we propose a new control strategy where the interface is first stabilized
against the obstacles before the user touches its end-effector. This way, contacts appear more natural, even at
high speeds, as confirmed by preliminary user-tests made with an existing 2 DoF encounter type haptic
interface at different speeds with the state of the art control law and the novel approach we propose here.
1 INTRODUCTION
Haptic interfaces allow motion interactions with
virtual or remote environments with a reproduction of
the sense of touch, using kinesthetic (force/position)
and cutaneous (tactile) receptors (Hannaford and
Okamura, 2008). We can distinguish four methods for
creating haptic sensations artificially: vibrotactile
devices, force-feedback systems (discussed in this
paper), surface displays and distributed tactile
displays (Hayward and Maclean, 2007).
Force-feedback systems are robotic mechanisms
capable to measure the user’s movements and deliver
a force signal to the operator’s hand, usually through
a pen-like interface, a knob or a thimble (Campion,
2011). A non-exhaustive list of application cases are
computer-aided design (Nahvi et al., 1998),
maintenance and assembly tasks (McNeely et al.,
1999), games (Martin and Hillier, 2009) and virtual
reality task simulations (Sagardia et al., 2015) as well
as teleoperation (Gosselin et al., 2005).
In an ideal force-feedback system, the user should
be able to move in free space without feeling any
force and the device should prevent him/her to move
in the constraint direction if a stiff object is being
touched. In the mentioned application contexts
however, force-feedback interfaces usually require
the user to be mechanically linked to them. This link
has a non-negligible influence: the user experiences
the friction, inertia and vibrations of the mechanical
structure even when moving in free space, which
reduces the realism of the interaction. In addition, the
difference between free space and contact is less
distinctively felt than in real world.
Encounter-Type Haptic Displays (ETHDs)
propose, as a solution to this problem, to remove the
mechanical link between the interface and the
operator (McNeely, 1993), (Tachi et al., 1994). This
principle allows a perfect transparency in free space
motion as the user touches the haptic device, usually
with the fingertip, only when a contact occurs in the
virtual/remote environment (see Figure 1).
(Yoshikawa and Nagura, 1997) and (Yoshikawa and
Nagura, 1999) use for example a set of optical glass-
fiber on-off sensors for measuring the position of the
operator’s finger, respectively in 2D space with a
Fierro, O., Bachta, W., Gosselin, F. and Morel, G.
A New Control Strategy for the Improvement of Contact Rendering with Encounter-type Haptic Displays.
DOI: 10.5220/0006474704710480
In Proceedings of the 14th International Conference on Informatics in Control, Automation and Robotics (ICINCO 2017) - Volume 2, pages 471-480
ISBN: Not Available
Copyright © 2017 by SCITEPRESS – Science and Technology Publications, Lda. All rights reserved
471
ring-like end-effector and in 3D space with a cap-like
end-effector. The position of the finger is however
only roughly estimated. In (Gonzalez et al., 2015a) a
ring-like end-effector instrumented with infrared
proximity sensors, mounted on a 2 Degrees of
Freedom (DoF) interface, is used to reconstruct the
shape of the finger and precisely estimate its position
using distance measurements (Gonzalez et al.,
2015b). We find as well hand exoskeletons
(Nakagawara et al., 2005), (Fang et al., 2009) where
the position of a thin reflecting plate, pushed by the
nail, is recorded thanks to an optical sensor. Force felt
between the plate and the finger is negligible.
Figure 1: Encounter-type haptic display principle.
Special attention should be given to the control law
that governs ETHDs, particularly to the transitions
between free space and contact modes. As shown in
(Gonzalez et al., 2015a), control strategies usually
implemented on ETHDs rely on an abrupt transition
between these two modes, potentially generating
vibrations and non-realistic impact forces at that
moment. To cope with this issue, (Gonzalez et al.,
2015a) proposes a smooth transition-based control.
This solution was implemented on a 2DoF ETHD. It
proves stable and more realistic, especially when
finger interactions occur at low speeds (≈0.2 m/s).
However at higher speeds the problem is not
completely tackled and the sensation felt may be non-
realistic at the moment of contact.
In this paper, we propose a new control strategy
aimed at allowing natural transitions between free
space and contact modes, even at higher speeds (>0.2
m/s). It includes a bilateral damping allowing the
stabilization of the robot’s end-effector before
application of force feedback. This comes at the price
of a slight shift of the virtual wall, which remains
however imperceptible for most users as proved by
the results of our evaluations. The state of the art
control law implemented in (Gonzalez et al., 2015a)
is first briefly presented in section 2. The proposed
upgraded control strategy is then explained in section
3. Section 4 presents the results of the experiments
performed to validate the potential of our approach.
Finally conclusions are given in section 5.
2 SMOOTH TRANSITION-BASED
CONTROL
In free space, the ETHD should closely track the
finger’s position without touching it. When the user’s
avatar enters in contact with a virtual wall, the
resulting contact force should be displayed to the
user. A control law, which to our knowledge answers
the most closely the aforementioned requirements,
was proposed in (Gonzalez et al., 2015a) and
implemented on a 2DoF ETHD. It will be briefly
described in the following lines.
2.1 Control Algorithm for Finger
Tracking in Free Space
We note here
the position
error between the ring center and the finger center
(see Figure 2). Being small as close tracking of the
finger is desired, it can be expressed in joint space:
(1)
where
are the joints positions and
the robot’s jacobian matrix expressed in its
global reference frame
.
Figure 2: 2DoF ETHD with ring center
and finger
center
(adapted from (Gonzalez, 2015)).
ICINCO 2017 - 14th International Conference on Informatics in Control, Automation and Robotics
472
Error minimization is achieved with a
Proportional Derivative controller, which provides
the robot with a reference torque:
(2)
where
is the equivalent impedance,
and
the proportional and derivative gains respectively.
With this controller, a link equivalent to a spring and
damper system is created between the centers of the
user’s finger and of the ring (see Figure 3).
Figure 3: Spring damper coupling between the center of the
finger
and the center of the ring
in free space.
2.2 Control Algorithm for Force
Rendering at Contact
When a virtual object is encountered, the interface
must render the corresponding interaction force
,
which is defined as an unilateral constraint. A
viscoelastic compliant virtual environment without
tangential friction is assumed and a modified Kelvin-
Voigt model (Achhammer et al., 2010) is used to
calculate the resulting interaction forces (see Eq. 3).
Figure 4: At contact the influence of the tracking force
has
been diminished by a factor and a spring and damper
coupling (
) is created between the ring and
the wall to render the interaction force
.
Let
be the distance between the ring’s avatar
inner circumference and the closest point of a virtual
object,
the position of a vertical wall (see Figure 4),
a unitary vector normal to the surface of contact,
the speed of the ring along . The environment
interaction force
can be expressed as follows:
(3)
The transition between free space and contact is
achieved by reducing the influence of the tracking
force
by a factor nearby the obstacles, i.e. at a
distance
from the virtual object (VO)
placed at position
. Equation 4 shows how varies
in function of
, the distance between the finger’s
avatar and the VO (see Figure 3). Factor cannot be
totally cancelled at the proximity of the wall as in this
case the ring would not follow the finger when it
moves away from it.
(4)
Here
,with
chosen so
that
,
. This way by the moment the
user encounters the ring, the tracking effect is almost
canceled and
. The updated Eq. (2) can now
be expressed as:
(5)
While allowing a smooth transition between free
space and contact, this algorithm presents in practice
an undesired behaviour when impacting virtual
objects at high speeds (> m/s): oscillations appear
when the ring encounters a virtual object and
therefore an unnatural contact is perceived by the user
when his/her finger encounters the ring. It may give
the impression of touching a moving object instead of
a static one as in the real life.
3 OFFSET TRANSITION-BASED
CONTROL
When the user’s finger encounters the ring, he/she
should feel as touching a static object. The offset
transition-based control introduced in this paper
proposes therefore to first apply a dissipating force
when the ring’s avatar penetrates into the VO in order
to stop it before displaying the VO properties
.
Further details will be given below.
A New Control Strategy for the Improvement of Contact Rendering with Encounter-type Haptic Displays
473
3.1 Virtual Environment Force Estimation
and Rendering
In free space, the user can move the interface freely,
i.e. no interaction force exists. When the ring’s inner
periphery penetrates in a virtual object, we propose to
completely stop it before displaying the VO
properties. Therefore a dissipating force is applied on
the ring until the interface is static (in practice until
, with
an experimentally tuned
threshold introduced to cope with the speed signal’s
noise). When the mentioned condition is true, the VO
properties (
and
) are rendered to the user. The
new virtual wall position
is defined as the
coordinate of the distal point on the inner periphery
of the ring once it is static.
This algorithm was implemented using a Finite
State Machine (FSM) as shown in Figure 5. The initial
state is called transparent: in this mode only the
tracking force acts on the ring and
. As soon as
the interface approaches the VO and the inner
periphery of the ring penetrates into it (i.e.
),
the braking state becomes active. The applied
bilateral force
exerted on the ring is shown in
equation 6 with
the dissipative gain.
n
(6)
When
the VO state becomes active.
The speed threshold is fixed to
,
which in practice corresponds to a static interface
with a finger inside the ring. In this state the VO
properties are the same as implemented in the original
control law.
Figure 5: Finite State Machine governing the proposed
control law.
Equation 7 defines
in reference to the new wall at
position
, where
is the distance between the
inner ring periphery and it.
(7)
3.2 Tracking Force
As previously explained, the influence of the tracking
force must diminish when the ring touches the wall,
however it must be strong enough for the interface to
follow his/her finger when moving away from the
VO. Also, it is important to have a continuous
tracking force to ensure that the interface will behave
correctly during transitions between free and contact
modes. In the original control law
decreases by a
factor , which is function of
, to ensure its
continuity (see section 2.2).
The control algorithm presented in section 3.1
requires a different strategy to make sure that the
tracking force remains continue. still varies in
function of
as in (4) but we make here
so that the minimum value of
when the ring’s
avatar penetrates in the reference wall is ensured.
Indeed, because of the tracking error, the finger’s
center position is always in advance to that of the ring
when it is in movement. This means that if we
consider a finger avatar the size of the ring, it will
always penetrate first in the reference (
) or offset
(
) VO position. When the VO takes its new value
at
, the augmented finger’s avatar is already
penetrating into it. At this moment varies in
function of
to ensure the continuity of and
therefore that of
. When moving away from the VO,
the augmented finger avatar can be far enough from
it in order too fully reactivate . When the ring
comes back in transparent state, this is when
, we do to vary in function of
again. This
way the continuity of the tracking force is ensured
during all state transitions (see Figure 6).
Figure 6: Variation of for a typical encounter and related
active state at each stage. represents the wall offset.
ICINCO 2017 - 14th International Conference on Informatics in Control, Automation and Robotics
474
Figure 7: Offset transition-based control. interaction force. : estimation of the robot’s end-effector speed
with
negative sign and of the distance
.
As we can observe in equation 8, the distance
is
also valid when the VO state is active. Here
means that will be function either of
or
,
according to the active state (see Figure 6). The control
explained in sections 3.1 and 3.2 is resumed in the
control block diagram from Figure 7.
(8)
4 EXPERIMENTS AND RESULTS
To validate the benefits of the new approach, called
VO-B (Virtual Object B) in the following, we have
compared its performances with that of the original
control law, called VO-A (Virtual Object A).
4.1 Experimental Setup
The robot used for these experiments is an optimized
version of a 2D substructure of a parallel 6DoF haptic
interface developed at CEA, LIST for tele-surgery
(Gosselin et al., 2005). This 2 DoF robot is composed
of two links m long each (see Figure 8). Its
workspace lies in a vertical plane. Actuation is
provided by two Maxon RE-35 DC motors and cable
capstan reducers, allowing a particularly transparent
behaviour. 1000ppt encoders are used for position
sensing and counterweights mounted on each axis
allow gravity compensation. The 2D ring-shaped
encounter-type end effector has an inner diameter of
24 mm, sufficient to track a finger at medium speeds.
Sixteen Vishay VCNL4000 infrared proximity
sensors distributed over the inner side of the ring
make possible the estimation of a finger’s center
position at a rate of 300 Hz with a mm precision.
Figure 8: 2DoF ETHD.
ATMega328P microcontrollers retrieve and send
proximity sensor measurements to the haptic
interface controller trough a fast serial bus at a rate of
400 kbps. Estimation of the finger’s location is
computed as the center of the polygon obtained from
the measurement. The controller is composed of a
PC104 computer running Xenomai realtime
operating system and a servo-drive controlling both
motors. A telerobotics library (Gicquel et al., 2001)
acquires the state of the robot, computes the finger’s
position and sends the reference torques to the servo-
drive at a rate of 1 KHz. Rate mismatch between the
control loop (1 KHz) and the estimation of the
finger’s center (300 Hz) is handled by a Kalman filter
committed to extrapolating the finger’s position.
A New Control Strategy for the Improvement of Contact Rendering with Encounter-type Haptic Displays
475
4.2 Practical Comparison between
Smooth Transition-based and
Offset Transition-based Control
In order to compare algorithms VO-A and VO-B, we
made a typical encounter with a vertical virtual wall.
We asked therefore a participant to move the ETHD
horizontally from the right to the left with his index
finger until he taps on a vertical wall located a few
centimetres on the left of the initial position. In order
to ensure an as natural as possible contact, we asked
the user to keep his index finger straight and
perpendicular to the working plane of the robot, with
the pulp oriented to the left. These experiments were
made both at low speed, where the existing control
law is assumed to work properly, and at higher
speeds, for which the device’s behaviour becomes
unnatural. The participant had a sufficient time to
understand these instructions and familiarize with the
interface. We checked both visually during the
experiments and during data post-processing that the
gesture was performed properly. It is worth noting
that in practice, due to the limited dynamics of the
robot, the user enters in contact with the ring before
he encounters the wall for speeds higher than m/s.
We chose therefore m/s as high speed. A value of
m/s was chosen for the low speed so as to remain
significantly lower than the high speed.
The robot’s gains were defined experimentally, so
as to be the highest possible while remaining stable.
Their values in free space and during contacts are
given in equations 9 and 10 respectively. We use the
same gains
,
and
,
in both conditions VO-
A and VO-B. The dissipative gain
is defined in
equation 11. A high value is chosen in order to stop
the interface as fast as possible (in practice in less than
20ms, see Table 1). With these values, no instabilities
were observed in practice.
(9)
(10)
(11)
In both cases, we denote
the instant of time
at which the ring’s inner periphery encounters the
object’s reference constraint,
the instant of time
when the user’s finger contacts the ring,
the
amplitude of the very first rebound of the end-effector
at contact and
as the reference constraint. In
condition VO-B,
represent the modified constraint
which is actually
plus the offset . Regarding
speeds, we denote
the speed of the ring’s
center when it encounters the reference wall
,
its speed when it encounters the user’s
finger and
its minimal speed just after
the finger encountered the ring.
4.2.1 Low Speed Case
The results obtained at low speed are given in Figures
9 and 10. No significant discrepancy can be observed
between VO-A and VO-B. Both
and
look similar. Only the speed
, i.e. the
rebound speed, is slightly smaller in condition VO-B.
As a whole, the behaviour of both control laws is very
similar. This is not a surprise as VO-A has already
good performances at low speeds.
Figure 9: Typical encounter with a vertical virtual wall at a
low speed in condition VO-A.
Figure 10: Typical encounter with a vertical virtual wall at
a low speed in condition VO-B.
4.2.2 High Speed Case
A clear difference between VO-A and VO-B can be
observed in this case (see Figure 11 and Figure 12),
considering both
, which is smaller for VO-B
(observable oscillations appear in VO-A) and
, which is closer to zero in VO-B, indicating
that in this case the ring is quasi-static, as expected
ICINCO 2017 - 14th International Conference on Informatics in Control, Automation and Robotics
476
(the observed oscillations prove on the contrary that
this is not the case in condition VO-A).
Rebound speed
is also higher in
VO-A than in VO-B. These observations lead us to
make the hypothesis that with the new algorithm
proposed in this article the sensation felt by the user
will be more realistic since
and
tend to be smaller at high speeds than in
condition VO-A.
Figure 11: Typical encounter with a vertical virtual wall at
a high speed in condition VO-A.
Figure 12: Typical encounter with a vertical virtual wall at
a high speed in condition VO-B.
4.3 User Tests
According to (Samur, 2012), many factor studies can
be used to assess the benefits of haptic feedback on
sensory-motor tasks: peg-in-hole, tapping, targeting,
etc. In the present work, we are more particularly
interested in comparing how much the interface
stabilizes when contacting a virtual object and how
natural the contact is perceived by the user. As a
consequence, we chose a tapping test to qualify the
behaviour of the interface at both low and high speeds
using VO-A and VO-B. The metrics introduced in
section 4.2 will guide our analysis. The perception of
the interaction was also evaluated through a survey.
4.3.1 Methodology
Four right-handed volunteers (3 men, 1 woman), aged
23-31, were invited to perform the tapping test. A
printed document describing the experiment was
given to each participant and he/she was asked to sign
a letter of consent. The experiment was performed in
an isolated room. The participant was standing, facing
a screen and wearing an anti-noise helmet. The haptic
interface was placed on a table so that his/her right
index finger can be placed comfortably inside the end
effector and so that a horizontal movement to the left
can be performed easily.
A devoted graphical user interface (GUI) was
developed for this experiment. It displays a 2D virtual
environment in which the free space appears as a
black vertical rectangle, surrounded by a thick green
contour representing four virtual walls. The user is
asked to tap on the left wall at low and high speeds,
as previously defined in section 4.2, in conditions
VO-A and VO-B. A white circle represents his/her
finger and a vertical line indicates where to start
before each tap (see Figure 13). In both conditions the
finger’s avatar is stopped against the wall, even if the
robot, hidden by a vertical barrier, goes further, in
order to avoid the influence of visual cues.
Figure 13: Setup of the experiment.
Low speed tests were always performed first, the
order of presentation of each case being alternative,
i.e. either training with VO-A low and then perform
tests in condition VO-B low then VO-A low or
training with VO-B low and then perform cases VO-
A low then VO-B low. The same principle was used
in high speed conditions.
It is worth noting that it is of crucial importance
that the user keeps the index finger straight and
perpendicular to the working plane of the robot
throughout the experiments, with the pulp oriented to
the left, so that it can make a full and proper contact
with the wall. The non-respect of this gesture may
A New Control Strategy for the Improvement of Contact Rendering with Encounter-type Haptic Displays
477
impact the user’s perception and the quality of the
recorded data. To avoid this, the participants had a
sufficient time to familiarize with the interface and
practice taping at low and high speeds. Also, data was
recorded for each single tap and its exploitability was
verified in situ (as for the results given in section 4.2,
we checked both visually during the experiments and
during data post-processing performed just after each
tap that the gesture was performed properly). Each
participant was asked to perform taps until we get at
least three valid taps (correct speed and absence of
contact with the ring before the obstacle).
4.3.2 Results
Between 10 and 15 taps were necessary in each case
to obtain three valid taps (one of the subject being
unable to perform taps at high speeds, his data were
discarded for the analysis).
Figure 14 illustrates the contact speed in each case
(the median value appears in red, the box represents
the first and third quartiles and the lower and higher
bars the extremal values over the 9 trials, i.e. 3 users
with 3 taps each). We can observe from these results
that our data set is close to the low and high speeds
defined in section 4.2 (i.e. and m/s).
Figure 14: Speed of the ring center at time
.
Figures 15, 16 and 17 illustrate respectively the speed
of the ring center when the finger touches it (it should
be as low as possible to realistically simulate the wall,
which is fixed), the amplitude of the first rebound of
the ring against the wall and the highest value of the
speed of the ring after contact with the virtual wall
(both should be as small as possible).
These results confirm that the behaviour of the
interface in conditions VO-A and VO-B is very
similar at low speeds. It differs only for high speeds.
In this case, the speed of the ring when it encounters
the user’s finger tends to be closer to zero in condition
VO-B (see Figure 15). At least 25% of the samples
show positive speed, which means that at contact the
ring and the finger were moving in the same direction.
This is preferable than having a ring moving against
the finger at contact. A clear difference can also be
observed if we take into account the rebound
and the speed
. Figure 16 shows that
conditions VO-A produces a very important rebound
compared to VO-B. The same tendency is observed
for speeds in Figure 17, the absolute value of the
speed in VO-A being much higher than in the VO-B
case. The rebound amplitude and speed in VO-B
shows a considerable reduction, confirming that
conditions VO-B allows a more realistic simulation
of a fixed wall, even at high speeds.
Figure 15: Speed of the ring center at time
.
Figure 16: Rebound amplitude
.
Figure 17: Speed of the ring center at a time
.
Further details on the behaviour of the interface in
condition VO-B are given in Table 1. Results show
that in average the ring stabilizes in about ms at low
speed and ms at high speed, the constraint offset
remaining below 3 mm. We can expect that a human
operator wouldn’t realize these differences when
performing a tap (according to (Knorlein et al., 2009)
ICINCO 2017 - 14th International Conference on Informatics in Control, Automation and Robotics
478
and (Vogels, 2004), a visuo-haptic delay is
imperceptible if it is lower than 45 ms).
Table 1 : Mean and standard deviation for and .
mean(std)
(ms)
(mm)
Low speed
3.111(2.804)
0.388(0.251)
High speed
14.556(1.424)
2.434(0.357)
The participants were asked to answer three questions
after the completion of the tests in each case. Q1 asks
if the user perceived the contact before (score 1 or 2),
just when (3) or after (4 or 5) the finger’s avatar
touched the virtual wall. It provides information on
the perception of the visuo-haptic delay (ideal result
is 3). Q2 asks if at contact the touched wall was
perceived as moving to the left (score 1 or 2), being
static (3) or moving to the right (4 or 5). It tell us if
the user was perceiving the rebound (ideal result is
also 3). Finally, Q3 asks if the sensation at contact
was felt very natural (1), natural (2), neutral (3),
unnatural (4) or very unnatural (5). Results are given
in Figure 18 (mean scores for the three participants).
They show almost no difference at low speed, as
expected. At high speed however, conditions VO-B
gives better results. The ring appears static while in
VO-A it appears slightly moving. Also, the contact is
perceived as being more natural in condition VO-B.
Figure 18: Survey scores.
5 CONCLUSIONS
In this paper, we introduced a new control law
intended to improved contact rendering with
encounter type haptic displays. The results of our
experiments show that this offset transition-based
control allows to reduce the speed of the end effector
before the user’s finger encounters it, as well as its
rebound amplitude against the obstacles. As a
consequence, the contact is perceived as more natural.
REFERENCES
Achhammer, A., Weber, C., Peer, A., Buss, M. (2010).
Improvement of model-mediated teleoperation using a
new hybrid environment estimation technique, in: Proc.
IEEE Int. Conf. on Robotics and Automation,
Anchorage, U.S.A., pp. 5358–5363.
Campion, G. (2011) Literature Review. In: Campion, G.
(ed) The Synthesis of Three Dimensional Haptic
Textures: Geometry, Control, and Psychophysics,
Springer Series on Touch and Haptic Systems, London,
England., Springer-Verlag, pp. 7–44.
Fang, H., Xie, Z., Liu, H. (2009). An exoskeleton master
hand for controlling DLR/HIT hand, in: Proc.
IEEE/RSJ Int. Conf. on Intelligent Robots and Systems,
St. Louis, U.S.A., pp. 3703–3708.
Gicquel, P., Bidard, C., Coulon-Lauture, F., Measson, Y.,
Desbats, P. (2001). Tao: 2000: A generic control
architecture for advanced computer aided teleoperation
systems, in: Proc. ANS Int. Topical meeting on Robotics
and remote systems, Seattle, U.S.A.
Gonzalez, F. (2015). Contributions au développement
d’une interface haptique à contacts intermittents (phd
thesis). Université Pierre et Marie Curie - Paris VI.
Gonzalez, F., Bachta, W., Gosselin, F. (2015a). Smooth
transition-based control of encounter-type haptic
devices, in: Proc. IEEE Int. Conf. on Robotics and
Automation, Seattle, U.S.A., pp. 291–297.
Gonzalez, F., Gosselin, F., Bachta, W. (2015b). A 2-D
Infrared Instrumentation for Close-Range Finger
Position Sensing, IEEE Trans. on Instrumentation and
Measurement. 64(10), 2708–2719.
Gosselin, F., Bidard, C., Brisset, J. (2005). Design of a High
Fidelity Haptic Device for Telesurgery, in: Proc. IEEE
Int. Conf. on Robotics and Automation, Barcelona,
Spain, pp. 205–210.
Hannaford, B. & Okamura, A.M. (2008). Haptics, in:
Siciliano, B. & Khatib O. (eds.) Handbook of Robotics,
Berlin, Germany, Springer, pp. 719–739.
Hayward, V. & Maclean, K.E. (2007) Do it yourself
haptics: part I. IEEE Robotics and Automation
Magazine. 14(4), 88–104.
Knorlein, B., Di Luca, M., Harders, M. (2009). Influence of
Visual and Haptic Delays on Stiffness Perception in
Augmented Reality, in: Proc. IEEE Int. Symp. on Mixed
and Augmented Reality, Washington, USA, pp. 49–52.
Martin, S. & Hillier, N. (2009) Characterisation of the
Novint Falcon Haptic Device for Application as a
Robot Manipulator. In: Proc. Australasian Conf. on
Robotics and Automation, Sydney, Australia.
McNeely, W.A. (1993). Robotic graphics: a new approach
to force feedback for virtual reality, in: IEEE Virtual
Reality Annual Int. Symp., Lafayette, Louisiane, pp.
336–341.
McNeely, W.A., Puterbaugh, K.D. & Troy, J.J. (1999). Six
Degree-of-freedom Haptic Rendering Using Voxel
Sampling, in: Proc. Annual Conf. on Computer
Graphics and Interactive Techniques, Los Angeles,
U.S.A., pp. 401–408.
A New Control Strategy for the Improvement of Contact Rendering with Encounter-type Haptic Displays
479
Nahvi, A., Nelson, D.D., Hollerbach, J.M. & Johnson, D.E.
(1998) Haptic manipulation of virtual mechanisms
from mechanical CAD designs, in: Proc. IEEE Int.
Conf. on Robotics and Automation, Leuven, Belgium,
pp. 375–380.
Nakagawara, S., Kajimoto, H., Kawakami, N., Tachi, S.,
Kawabuchi, I., (2005). An Encounter-Type Multi-
Fingered Master Hand Using Circuitous Joints, in:
Proc. IEEE Int. Conf. on Robotics and Automation,
Barcelona, Spain, pp. 2667–2672.
Sagardia, M., Hertkorn, K., Hulin, T., Schätzle, S., Wolff,
R., Hummel, J., Dodiya, J., Gerndt, A. (2015). VR-
OOS: The DLR’s virtual reality simulator for
telerobotic on-orbit servicing with haptic feedback, in:
Proc. IEEE Aerospace Conf., Montana, USA, pp. 1–17.
Samur, E. (2012) Performance Metrics for Haptic
Interfaces. London, Springer-Verlag.
Tachi, S., Maeda, T., Hirata, R., Hoshino, H. (1994). A
construction method of virtual haptic space, in: Proc.
Int. Conf. on Artifitial Reality and Tele-Existence,
Tokyo, Japan.
Vogels, I.M.L.C. (2004). Detection of temporal delays in
visual-haptic interfaces. Human Factors. (46)1, 118–
134.
Yoshikawa, T., Nagura, A. (1997). A touch and force
display system for haptic interface, in: Proc. Int. Conf.
on Robotics and Automation, New Mexico USA, pp.
3018–3024.
Yoshikawa, T., Nagura, A. (1999). A three-dimensional
touch/force display system for haptic interface, in:
Proc. IEEE Int. Conf. on Robotics and Automation,
Detroit, USA, pp. 2943–2951.
ICINCO 2017 - 14th International Conference on Informatics in Control, Automation and Robotics
480
