J. M. Azor
ın, J. M. Sabater, N. Garc
ıa, F. J. Mart
ınez, L. Navarro
Dpto. de Ingenier
ıa de Sistemas Industriales, Universidad Miguel Hern
Avenida de la Universidad s/n, Elche (Alicante), 03202 Spain
R. J. Saltar
DISAM, ETSII, Universidad Polit
ecnica de Madrid
e Guti
errez Abascal 2, Madrid, 28006 Spain
Haptic interfaces, robot-assisted surgery, telesurgery, surgical simulation.
This paper presents a comparative of different non-specific haptic interfaces that could be used for robot-
assisted surgery. The purpose of this analysis is to determine which master interface has the best performance
for a specific task in which the master-slave scale factor is less than one. Three haptic interfaces have been
considered: two commercial masters, one with serial configuration, the PHANToM 1.5 prototype master, and
one with spherical setup, the Microsoft Force Feedback 2 Sidewinder; and other one non commercial with a
parallel architecture designed in our laboratory, the Magister-P. Two experiments performed to measure the
fidelity of the haptic interfaces have been described and the results obtained have been discussed on this paper.
Robot-assisted surgery has became a new research fo-
cus of the robotics community during the last years.
The followed objective is to develop “a partnership
between man (the surgeon) and machine (the robot)
that seeks to exploit the capabilities of both to do a
task better than either can do alone” (Taylor et al.,
1991). Telesurgery and surgical simulators belong to
this new robotics area. Telesurgery allows surgeons to
perform remote surgical operations using robots and
haptic interfaces. On the other hand, surgical sim-
ulators are based in virtual environments where the
surgeon uses a haptic interface to control the med-
ical robot. These virtual environments incorporate
accurate and reliable mathematical models of the hu-
man body part that is going to be operated and of the
rigid bodies of the medical apparatus involved. Sur-
gical simulators can be applied to surgical training
uhnapfel et al., 2000), avoiding the use of other ex-
pensive learning techniques as experiments with ca-
davers, and to biomedical research. In addition, these
surgical simulators become an indispensable tool in
telesurgery, since any action performed by the sur-
geon over the patient can be verified in the simulator
before it would be executed by the remote robot.
Nowadays, a great research effort is being made
on the robot assisted surgery, and several universi-
ties and enterprisers have presented their develop-
ments with applications on orthopedic, microsurgery,
laparoscopy,...(Cleary and Nguyen, 2001). Neverthe-
less, as it is suggested on (Taylor and Stoianovici,
2003), it is necessary to made significant advances on
several aspects, like the interface technology, where
better surgeon-machine interfaces are needed. The
work presented here is focused on this aspect, and it
presents some discussion of a comparative analysis of
different general purpose haptic interfaces that would
be able to be used as robot-assisted surgery input de-
vices. Although near almost the actual developments
use particular devices designed for an specific task, as
for example the Laparoscopic Engine of Immersion
for endoscopic surgical systems, our interest is made
such tasks with general purpose haptic interfaces, so
we can take benefit of the different kinematic struc-
tures of master and slave to improve the ergonomics
and/or the manipulability of the surgical tools. Two of
the used interfaces are commercially available, while
the third one has been designed in our laboratory.
To our knowledge, there are not many research
works in the literature that analyzes which haptic in-
terface achieves a better behavior in determined surgi-
cal interventions. Some work has been done trying to
obtain general indexes that help us to classify the per-
formance of haptic devices (Moreyra and Hannaford,
1998),(Hayward and Astley, 1996). The purpose of
the current analysis is to discuss the desirable char-
acteristics of a haptic interface to be used for robot-
M. Azorín J., M. Sabater J., García N., J. Martínez F., Navarro L. and J. Saltarén R. (2005).
In Proceedings of the Second International Conference on Informatics in Control, Automation and Robotics - Robotics and Automation, pages 375-378
DOI: 10.5220/0001159103750378
Figure 1: Masters considered in the analysis: Microsoft
Force Feedback (right), Phantom (center), and Magister-P
assisted surgery.
The paper is organized as follows. Section 2 de-
scribes the three haptic interfaces that have been used
in the comparative analysis, explaining its character-
istics. The experiments performed and the results ob-
tained are showed in section 3. Finally the main con-
clusions of the paper are summarized in section 4.
In this section, the three haptic interfaces used on this
work are presented. All the experiments are limited
to a 1 DOF of the interfaces, trying to mimic the force
sensed on the movement along the axis of a typical
endoscopic device. We assume that a minimally in-
vasive surgery (MIS) is simulated. In MIS the op-
eration is performed with instruments and viewing
equipment inserted in the human body through small
incisions (typically less than 10mm in diameter). The
main advantage of this technique is the limited trauma
to healthy tissue, reducing the post-operative hospital
stay of the patient (Ortmaier, 2002). However this
technique has a difficult learning process since the
surgeon must learn to navigate using visual informa-
tion provided by cameras with non orthogonal optical
to the scene observed.
The three master used are the low-price spherical
Microsoft Force Feedback (MMF- fig. 1 - a)), the well
known 6 d.o.f. serial arm Phantom master 1.5 Proto-
type (fig. 1 - b)), and a parallel configuration device,
the Magister-P (fig. 1 - c)). The spherical and the
serial master are commercially available, while the
parallel master has been designed in our laboratory
(Sabater et al., 2004). All devices incorporate force
feedback. Force feedback improves the task perfor-
mance in spite of the necessity of having a measure of
the force/torque sensor in the robot that executes the
operation. Without force feedback, the forces that the
surgeon feels only depends on the master character-
istics, independently of the remote environment situ-
ation. This way, e.g., the surgeon can not determine
the quantity of forces to exert in a suture operation if
only exists visual feedback.
On many of the recent developed systems, the force
reflection is neglected. This could be accepted by the
moment that, in a real classic MIS technique, the real
forces sensed by the surgeon are minimal. Neverthe-
less, if our future goal is not to copy the classical MIS
techniques, but to give the surgeon new tools that al-
low him to increment the useful information that way
the surgical procedures can be improved, it is neces-
sary to have an accurate force reflection system.
About the number of DOF, the Phantom and
Magister-P masters have 6 DOF, while the Microsoft
Force Feedback master has only 3 DOF (spherical
motion with 2DOF force feedback). Therefore any
device have the same number of DOF that the laparo-
scopic instrument (4 DOF). Next, how these masters
can be used in an endoscopic operation is explained:
Microsoft Force Feedback Master. This device has
only 3 DOF to control the 4 DOF of the laparo-
scopic slave, so the use of additional master but-
tons is necessary. The ascend and descend motion
of the tool can be achieved with the motion of the
joystick Y axis (fig. 1-a)). In this way, the force
feedback is possible when the tool interacts with
the deformable object. The Z-axis rotation move-
ment can be achieved with the Z-axis joystick rota-
tion movement, and the XY rotation movement of
the tool around the fulcrum point can be achieved
with the movement XY of the joystick at same time
a button is pressed.
Phantom Master. At this device it is necessary to
constraint (or neglect) 2 DOF in order to control
the 4 DOF of the surgical tool. The ascend and de-
scend motion of the tool can be linked with the Z
movement of the Phantom final effector. The Z-
axis movement can be achieved with the Z-rotation
of the Phantom, and the fulcrum XY rotation can
be represented by the XY translation movement of
the device. That way 2 rotation DOF of the effector
of the Phantom are neglected.
Magister-P Master. The ascend and descend motion
of the tool corresponds to the Z movement of the
lower platform (fig. 1-c)). The 3 orientation DOFs
of the surgical tool are given by the orientation of
the lower platform. The X and Y position of the
final effector are neglected, and are only used to
increase the ergonomic feeling of the surgeon.
The most important features of the devices that are
related with the task they are going to be used are
summarized on table 1.
Table 1: Main features of used devices
MFF Phantom Magister
DOF 3 6 6
Configuration Spherical Serial Parallel
Prog. Interface DirectX Ghost C-interface
Range Forces 0-4 N 0-8.5 N 0.3-90 N
Bandwidth low 1000Hz 320 Hz
Linearity almost
at low range no linear linear no linear
The three master devices have enough bandwidth
for the designed experiments. Nevertheless, for a high
fidelity haptic rendering, devices must have at least
300 Hz or higher, so the MFF is not an appropriate
device. Besides, the DirectX library does not permit
to modify the frequency of the haptic loop. All the
same, the designed experiments try to measure the
performance of the kinesthetic rendering (at frequen-
cies lower than 0.5 Hz) in a low velocity task, as a
surgical operation, so the three devices are valid ones
from the point of view of bandwidth. Mechanically
speaking, the Magister-P has some fabrication errors
that make that we need to overcome a friction thresh-
old to render forces.
At this section the setup of the comparative experi-
ments is explained. The results are plotted and some
comments for their use as master devices in an endo-
scopic robotic surgery are made.
Two psychophysical kind of experiments have been
performed over a group of users with previous ex-
perience with the three master devices. Users have
made each experiment twice, with at least one day
between them. A total of 10 users (6 male and 4 fe-
male) have done a number of 40 experiments. The
goal of the first group of experiments is to get data
of the force/torque interaction of the surgeon with the
master device, measuring the minimum force sensed,
what is related with the capability of detecting little
deformations on a deformable tissue. Similarly, the
second group of experiments tries to determine the
minimum difference on the increment of force/torque
that can be sensed, what is related with the ability of
distinguish between similar tissues.
Forces and torques are needed to be introduced to
the devices to make the experiments. In the Phan-
tom case, the Ghost library allows directly to program
them in cartesian space. In the Magister-P case, our
own library takes care of the calibration test to allow a
Figure 2: Experiment 1 results
direct input of forces/torques in cartesian space. Nev-
ertheless, in the MFF case, the DirectX library does
not allow to introduce force/torque values on carte-
sian or joint spaces. Instead of that, the own library
units are used. That way, a previous experiment of
dynamometry-calibration to find the relation of these
units with the force values must to be done. Next,
each one of the experiments is detailed and the results
are plotted.
3.1 Minimum sensed force value
On this experiment, starting with a null value of the
displayed force/torque, the value is being increased
each 2 seconds a determined value F
until the
users is being able to sense the force, recording the
value of this force. The process has been made with
two different increment values for each master. Figure
2 shows the results of the minimum force experiment.
Some remarkable aspects are:
The first time the experiment is done (upper part of
figure 2), the operators use to detect the force ear-
lier when the increment is the higher one. Besides,
there are significant differences when the increment
is one or another.
The second time the experiment is done (down part
of figure 2), the difference between the two incre-
ments is reduced, and the torque is detected earlier
in general.
In the Magister-P experiments, the minimum value
is above 0.4 N, due to the friction threshold that this
device has.
3.2 Minimum increment of force
perceived by the user
The goal of this experiment is to get the minimum
difference between consecutive forces that a user can
feel. This experiment tries to evaluate the ability of
the devices to display the changes on the stiffness of
the tissues.
To make this experiment, an initial force/torque
nominal value (F
) has been chosen for each
Figure 3: Experiment 2 results
master. This value is kept constant during a period of
time (at least one second) and next the force/torque
increment (F
) is added to this value and
rendered. If the operator feels the increment, the
value of the increment is reduced a certain quantity
), and the experiment in repeated with the
new values. This process is repeated while the opera-
tor feels the increment. The last value of increment is
noted as a result. Figure 3 plots the results obtained
on the two experiments made with the different mas-
ters. The average data are:
In general, users detect a lower torque increment
) when the nominal initial value is the lowest.
On the second experiment, the values are lower
than in the first one.
3.3 Comparative aspects
Some conclusions can be obtained from the previous
In the fist experiment, in all the devices, when the
increment of forces is small enough, and working
in low frequencies, some users get used to this in-
crement, and they do not detect the force until its
value is significantly higher than when the incre-
ment is given in bigger steps. This is because the
human sensors that work at low frequencies are
fast adaption sensors, and they get used to the new
value of the force.
The mechanical configuration of the devices plays
an important role in the ability of the users to feel
forces. The workspace is also a key parameter.
The ability of translation in the Phantom and in
the Magister-P provokes that the user can move his
hand, hiding some forces, and so that force values
are smaller on the spherical configuration.
In all the cases, on the second turn of the exper-
iments, values are smaller, due to the experience
acquired by users on the first turn.
A comparative analysis of general purpose haptic de-
vices in a robotic assisted surgery environment has
been made. The studied devices are not task-specific
devices, and they can be used on other teleoperation
All the master devices have the capacity of ren-
dering forces to the operator. Using this feature, the
comparative is focused on this rendering ability, and it
tries to determine the minimum change in the stiffness
of a tissue that could be perceived by the operator-
surgeon. For that, the experiments have been limited
to the input of a determined force/torque on an axis
for each device.
Future work must include the rendering of forces
on arbitrary directions, considering the isotropy of
each device. Besides, the workspace and the er-
gonomics of each device are also parameters that must
be included on a deeper study.
Cleary, K. and Nguyen, C. (2001). State of the art in sur-
gical robotics: clinical applications and technology
challenges. Comput Aided Surg., 6 (6):312–328.
Hayward, V. and Astley, O. R. (1996). Performance mea-
sures for haptic display. In G.Giralt, G. H. S.-V., ed-
itor, Robotic Research: The 7th International Sympo-
sium., pages 195–207.
uhnapfel, U., Camak, H., and Maab, H. (2000). Endo-
scopic surgery training using virtual reality and de-
formable tissue simulation. Computer and Graphics,
Moreyra, M. and Hannaford, B. (1998). A practical mea-
sure of dynamic response of haptic devices. In Proc.
o the IEEE International Conference on Robotics and
Automation, IEEE, pages 369–374.
Ortmaier, T. (2002). Motion compensation in minimally
invasive robotic surgery. PhD thesis, Lehrstuhl
fur Realzeit-Computersysteme Technische Universitat
Sabater, J. M., Saltar
en, R., and Aracil, R. (2004). Design,
modelling and implementation of a 6 urs parallel hap-
tic device. Robotics and Autonomous System, 47(1):1–
Taylor, R., Paul, H., Kazanzides, P., Mittelstadt, B., Hanson,
W., Zuhars, J., Williamson, B., B.Musits, Glassman,
E., and Bargar, W. (1991). Taming the bull: Safety
in a precise surgical robot. In Proc. IEEE Int. Conf.
on Advanced Robotics (ICAR), number 1, pages 865–
Taylor, R. H. and Stoianovici, D. (2003). Medical robotics
in computer-integrated surgery. IEEE Transactions on
Robotics and Automatioin, 19 (5):765–781.