• Article •
Multi-dimensional force sensor for haptic interaction: a
review
Aiguo Song1*, Liyue Fu1
1. School of Instrument Science and Engineering, Southeast University, Nanjing, 210018, China
* Corresponding author, [email protected]
Received: 24 February 2019 Accepted:
Supported by Natural Science Foundation of China under Grant Number U1713210.
Abstract Haptic interaction plays an important role in the virtual reality technology, which let a person
not only view the 3D virtual environment but also realistically touch the virtual environment. As a key part
of haptic interaction, force feedback has become an essential function for the haptic interaction. Therefore,
multi-dimensional force sensors are widely used in the fields of virtual reality and augmented reality. In
this paper, some conventional multi-dimensional force sensors based on different measurement principles,
such as resistive, capacitive, piezoelectric, are briefly introduced. Then the mechanical structures of the
elastic body of multi-dimensional force sensors are reviewed. It is obvious that the performance of the
multi-dimensional force sensor is mainly dependent upon the mechanical structure of elastic body.
Furthermore, the calibration process of the force sensor is analyzed, and problems in calibration are
discussed. Interdimensional coupling error is one of the main factors affecting the measurement precision
of the multi-dimensional force sensors. Therefore, reducing or even eliminating dimensional coupling error
becomes a fundamental requirement in the design of multi-dimensional force sensors, and the decoupling
state-of-art of the multi-dimensional force sensors are introduced in this paper. At last, the trends and
current challenges of multi-dimensional force sensing technology are proposed.
Keywords Haptic interaction; Virtual reality; Multi-dimensional force sensor; Elastic body; Decoupling;
Force sensor calibration
1 Introduction
At present, virtual reality technology has been widely used in various fields, such as industrial assembly,
telerobot, service, medical training, rehabilitation, education, entertainment, and so on. As one of the key
techniques, haptic interaction has been paid increasingly attentions by researchers[1-3]
. Multi-dimensional
force sensor is a foundational element for haptic interaction systems[4-5]
. For a typical haptic interaction
system, it consists of haptic display device, audio-video devices, three-dimensional geometric model,
physical-based haptic rendering, and force sensor, as shown in Figure 1.
Human
operator
Audio-video devices
Multi-
dimensional
force sensor
Haptic
display
device
3D geometric model
Collision detection
Physical-based haptic
rendering
Virtual
environment
Figure 1 A typical haptic interaction system.
According to the measurement principle, the multi-dimensional force sensors can be classified into the
following types: resistance strain type[6]
, piezoelectric type[7]
, capacitive type[8]
, optical fiber type[9]
, etc.
Most of the multi-dimensional force sensors are based on strain gauges.
Metal-foil strain gauge is the sensing element of resistance strain type multi-dimensional force sensor
that detects the force signal based on strain-resistance effect. Shape change of wire gauze subjected to
forces or stresses results in its resistance variation, and accordingly transformation to electric signal output.
Among the multi-dimensional force sensors based on many measured principles, resistance strain type
force sensor is the most widely used one at present. It has many advantages, such as high sensitivity, wide
measurement range, high reliability, proven technique, etc. However, it has some disadvantages including
poor dynamic response, great influence of external factors on Wheatstone bridge. The key issue of the
design of this kind of sensor is how to design a reasonable and optimum mechanical structure of the
elastomer[10-11]
.
Piezoelectric multi-dimensional force sensor is a kind of potentiometric sensor which realizes the
measurement of force/torque (F/T) based on piezoelectric effect[12]
. The piezoelectric effect is usually also
used for tactile sensor fabrication[13-14]
. At present many researchers have studied on the piezoelectric
component as a sensing element to directly measure the multidimensional forces/torques[15]
. Liu et al.
proposed a piezoelectric six-dimensional heavy F/T sensor based on Stewart structure[16]
. The sensor
achieves the effect of enhancing the heavy load sharing ability of piezoelectric wafer while the radius of the
manipulator is increasing. Li et al. studied large-range six-dimensional F/T sensor, but no universal
analytical mathematical model was obtained, and it is difficult to realize the miniaturization of the force
sensor[17]
. The main characteristic of piezoelectric material is that it has high inherent frequency, which is
especially suitable for dynamic measurement. However the main disadvantages of piezoelectric material
are that it is insensitive to the static force and it is easy subjected to the external electromagnetic noise.
Capacitive multi-dimensional force sensors can measure the multi-dimensional force/torque by setting a
couple of capacitors and changing the relative gap between electrodes. Because of high sensitivity and
flexibility, many researchers have conducted research on the capacitive force sensor. Kim et al. presented a
capacitive type six-axis F/T sensor with parallel plate capacitors embedded in a flexure structure[18]
. Lee et
al. developed a capacitive type six-axis F/T sensor in which the sensing elements are aligned in plane, and
can be fabricated on a single printed circuit board[19]
. Accordingly, the sensor’s structure is simple and
processing is easy. Kim et al. presented a novel capacitive six-axis F/T sensor with six capacitive sensor
cells for robot applications, and proposed a sensor design with parallel and orthogonal arrangements of
sensing[20]
. Over the past decade, micro-electromechanical systems (MEMS) capacitive F/T sensors have
been used in specific domains, such as high temperature and magnetic fields. This is attributed to the
attractive features of MEMS capacitive sensors, such as high sensitivity, low noise, low power
consumption, and insensitivity to temperature, and so on[21]
. In order to resolve the problem of high cost of
current commercial F/T sensors, Jacob et al. presented a new design for an inexpensive and robust F/T
sensor using MEMS barometers[22]
. This sensor can be assembled in two days for less than 20 USD, but its
accuracy is much lower than that of commercial sensors. Brookhuis et al. designed a silicon capacitive
three-axis F/T sensor that can be used to measure the interaction forces between a human finger and the
environment, but the resolution of the sensor need to be improved[23]
. In spite of their advantages, the
application of the capacitive multi-dimensional F/T sensors is restrained because of their large parasitic
capacitance interference.
The objective of this review paper is to give an overview about the resistance strain type multi-
dimensional force sensor which is the most widely used for haptic interaction at present. In section 2, the
development of mechanical structures of the multi-dimensional force sensors are introduced. Then
calibration method and problems in calibration are described in section 3. This is followed by presentations
of coupling error and static decoupling methods in section 4. At last, we indicate the trends and challenges
of the multi-dimensional force sensors.
2 Mechanical Structures of Multi-dimensional Force Sensor
The key mechanical component of all kinds of multi-dimensional force sensors is the elastic body usually
called elastic beam, which responds to the external acting F/T by deformation. Depending on the
mechanical structure of the elastic beam, the mechanical structures of multi-dimensional force sensor are
usually classified into two types: integrated structure and Stewart parallel structure. Multi-dimensional
force sensor with integrated structure includes cylindrical elastic beam structure, cross elastic beam
structure, asymmetric radial elastic beam structure, and composite elastic beam structure. The comparisons
of main features of six-dimensional F/T sensors with various mechanical structures are given in Table 1.
For six-dimensional F/T sensors, the main principle is to test three directional forces and three dimensional
moments by measuring the related strains of the elastic beam by strain gauges.
A typical cylindrical elastic beam structure of 6 degree of freedom (DOF) F/T sensors are illustrated in
Figure 2, which is called Watson force sensor developed by DRAPER Laboratory in the United States[24]
. It
consists of three vertical thin beams that are milled from a cylinder and distributed in the circumferential
direction of 120 degrees. The measuring circuit of the sensor is placed in the hollow cylinder. The F/T
sensor has simple structure and high sensitivity but large interdimensional coupling errors, thus it needs
decoupling calculation to obtain precise six-dimensional forces and torques. And the anti-overload ability
of the sensor is very low, which means it is vulnerable to damage.
Table 1 Comparisons of main features of F/T sensors with various mechanical structures
Mechanical Pros Cons Typical
structure design
Cylindrical beam Large load capacity, good shock
resistant ability
Low sensitivity in vertical
direction, serious
interdimensional coupling
[25][26]
Cross beam
high symmetry, compact structure,
large rigidity, easy to machine,
simplified mechanical model due to
a flexible link
Existence of interdimensional
coupling and radial effect [27][28]
Asymmetric
radial beam
Large rigidity, easy to set the
coordinate system
Difficult to machine, and
existence of nonlinear
interdimensional coupling
[29]
Composite beam No-coupling measurement can be
realized.
Complicated structures with
combination of elements; Greatly
affected by machining and
assembly precision.
[30]
Stewart platform Large rigidity, super stability, large
load capacity
It is difficult to achieve the forces
and moments isotropy. [31]
A six-dimensional F/T sensor early developed by Stanford Research Institute (SRI) is depicted in Figure
3, which is also a vertical elastic beam based cylinder structure[25]
. It is made of an aluminum cylinder with
75 mm diameter and has eight narrow and long elastic beams. Each beam has a small groove in the neck to
make the neck only transmit forces, and make the torque effects be very small. Strain gauges are attached
on both sides of the other end of each beam.
Figure 2 Draper Watson’s 6 DOF F/T sensor[24]. Figure 3 SRI 6 DOF F/T sensor[25].
Figure 4 and 5 depicted typical crossbeam structure and non-radial three-beam structure of 6 DOF F/T
sensors, respectively. The crossbeam structure shown in Figure 4 was brought forward by Japan Daiwa
Balance Co., Ltd. based on the wrist force sensor developed in JPL Laboratory[26]
. It is an integral spoke
Figure 4 Crossbeam structure based 6 DOF F/T
sensor[26]. Figure 5 Non-radial three-beam based 6 DOF F/T
sensor[28].
x
y z
structure with a flexible link between the crossbeam and flange, thus simplified mechanical model of the
elastomer as a cantilever beam. Strain gauges attached on the four main beams compose 6 Wheatstone
bridges for six-dimensional force/torque measurement. The SAFMS 6 DOF F/T sensors, developed in 1987
by Hefei institute of intelligent machinery, in collaboration with Southeast University, adopted this
mechanical structure[27]
, seen in Figure 6.
A sensor with a non-diametric three-beam centrosymmetric structure is shown in Figure 5. The inner
edge and outer edge of the sensor are fixed on the arm and claw of a robot respectively, and the force
exerted to the sensor is transmitted by three beams tangle to the inner edge. Two pair of strain gauges are
attached to top, bottom, left, and right sides of each beam, so that six groups of half-bridges can be formed.
Exact solution of the six-dimensional forces/torques can be obtained by decoupling the six groups of
bridges. This kind of force sensor has high rigidity. It was first proposed by Carnegie Melon University,
and also studied by Huazhong University of Science and Technology, that is HUST-FS6 6 DOF F/T
sensor[28]
. The aforementioned structures are cross-beam or vertical-beam structures with simple monolithic
elastic elements, whereas interdimensional coupling is serious. In 1982, Schott proposed a six-dimensional
F/T sensor with double-ring composite elastic beam structure[28]
, as depicted in Figure 7. This structure
with combination of elements can reduce or eliminate the interdimensional coupling.
Figure 8 shows a classic structure of another commonly used 6 DOF force sensor based on Stewart
platform[30]
, a 6 DOF parallel mechanism proposed by Stewart in 1965. The six spherical hinges of the
upper or lower platform are distributed around the same circle. They are semi-symmetric, i.e. spherical
hinges 1, 3, 5 and 2, 4, 6 distributed at intervals of 120°, respectively. Six elastic beams between the upper
and the lower platform produce strains when six components of force apply to the platform. Therefore the
6 DOF forces/torques can be obtained from the output of the Wheatstone bridge comprised of strain gauges
attached to the ring of elastic beams. The difference from the aforesaid two kinds of vertical beam structure
is non-integral structure that the upper and lower of the platform are connected with the six elastic beams
by spherical hinges. The spherical linkage means the low inherent frequency, so that the Stewart parallel
structure based multidimensional force sensor is only suitable for the static force/torque measurement.
The above mentioned mechanical structures are classical structures of elastic beams. In recent years,
many researchers presented some novel mechanical structures of elastic beam on the basis of the classical
Figure 6 Crossbeam structure based SAFMS 6
DOF F/T sensor. Figure 7 Schematic view of a six-dimensional F/T
sensor with composite elastic beam[29].
structures to improve the performance of multi-dimensional force sensor. A large-range six-axis force
sensor made by machining dumbbell grooves in cross beams is developed by Changchun Institute of Optics,
which has high sensitivity while ensuring a large measurement range[31]
. Schematic diagram of the elastic
beams of the sensor is depicted in Figure 9. Mastinu et al. designed a six-axis F/T sensor with a novel
mechanical structure, seen in Figure 10a. The sensing element of this sensor is a quasi-statically determined
three spoke structure constrained by virtue of elastic sliding spherical joints, seen in Figure 10b, which is
designed to avoid friction[32]
. Experimental results demonstrated that the sensor have a good performance in
linearity, crosstalk and dynamic behavior. Liang et al. developed a novel miniature four-dimensional force
sensor with elastic elements consisted of circular diaphragm and cantilever beam, seen in Figure 11[33]
. The
sensor comprises of a sensor tip, an upper cover, a base frame and elastic elements. Hu et al. presented a
novel elastic element design method of a six-dimensional F/T sensor with a floating beam that changing the
floating beam to H-beam to increase the stiffness of the sensor and improve the dynamic performance[34]
.
Song et al. proposed an improved mechanical structure of elastomer design for three-dimensional force
sensor miniaturization, which combines the cross beam and central straight beam to achieve the sensation
of 3 dimensional forces, shown in Figure 12. It is suitable for haptic interaction owing to its small size with
28mm diameter[35]
.
u1
u6u5
u2u3
u4
U1
U2U3
U4
U5 U6
x
y
z
o
Figure 8 Schematic diagram of six-axis F/T sensor
based on Stewart platform[30].
Figure 9 A large-range six-axis F/T sensor with
dumbbell grooves[31].
Figure 10 (a) Six-dimensional F/T sensor with a novel structure; (b) Circular spoked structure as an elastic spherical
sliding[32].
a b
Dumbbell groove Floating beam
Strain beam
Figure 11 (a) Modular structure of a four-dimensional force sensor; (b) The elastic element[33].
Figure 12 (a) Mechanical structure of a three-dimensional force sensor; (b) The prototype of three-dimensional
force sensor.
3 Calibration of Multi-dimensional Force Sensor
The static calibration of sensors plays an important role in testing, and it is a necessary step to ensure the
high precision of sensors. For a multi-dimensional F/T sensor, static calibration is to exert standard forces
of different magnitudes and directions on the sensor to obtain the corresponding output, and to evaluate and
correct its performance by establishing the relationship between them based on curve fitting.
3.1 Calibration Method
At present, the static calibration methods of multi-dimensional force sensors mainly include tension-
compression dynamometer method, standard weight method and other standard force generator consisting
of force generating equipment and high precision single-axis sensor. Due to the diversity of sensor
structures, there is no universal calibration device. Standard weight method is mainly introduced in this
section. The calibration test bench was developed by Robot Sensor and Control Lab of Southeast
University. Figure 13 shows the force/torque loading test bench.
a b
X
Y
Z
O
1
2
3
4
5 6
7
8
9
1-calibration shaft, 2-sensor, 3-rotatable indexing plate, 4,5-sliding
rods, 6-pulley, 7-base, 8-additional pulley rod, 9-weights, 10-steel wire
10
Figure 13 The force/torque loading test bench.
The six-dimensional F/T sensor to be calibrated is fixed on the scalable and rotatable indexing plate,
which is fixed on the base of the test bench, and the calibration shaft passes through the central axis of the
sensor. The sliding rods on both sides of the test bench can be adjusted and fixed to a certain height, so that
the steel wire suspending the standard weights presents a horizontal straight line between the calibration
shaft and the pulley, that is, it is parallel to the Y axis in Figure 13.
On the test bench, six linear independent standard force/torque components are applied to the six-
dimensional F/T sensor, respectively[36]
, and the output voltage values of the corresponding channels are
recorded. Ideally, in the static calibration experiment, each force/torque component can be loaded only
once under the condition that the applied standard force/torque is unbiased and the sensor is pure linear.
However, due to the deviation of loading force/torque and the non-linearity of six-dimensional F/T sensor
in the actual calibration process, the error of data obtained by loading only once is large. In practice, the
method of average value obtained by loading many times is often adopted.
For micro F/T sensors, such as submillimeter multi-dimensional F/T sensor developed using silicon
microfabrication technology, it is usually calibrated by applying gravitationally generated force onto its
structure. This method, however, is difficult to implement. Electromagnetic force was used to calibrate the
F/T sensor in reference [37]. An aluminum cross-beam is suspended over, and placed at the center of the
sensor. During calibration, the sensor is placed in a magnetic field and current is driven through the
aluminum beams to produce force electromagnetically. Schematic view of the sensor design and calibration
method is illustrated in Figure 14. Precise control of the magnitude of the electromagnetic force can be
achieved owing to its relationship to current and magnetic field strength.
Figure 14 Schematic view of calibration method of the submillimeter three-dimensional F/T sensor.
3.2 Problems in Calibration
For a multidimensional F/T sensor, due to the resistant-strain effect based measurement principle and
manufacturing errors, there is a mutual coupling among the output channels of the sensors. The inter-
dimensional couplings are complex and difficult to describe accurately in theory. The experimental method
is usually used to calibrate the relationship. However, due to the limitation of the experimental equipment
and method, the force/torque cannot be directly applied to the center point of the force sensor, or the force
direction is deviated a bit from the theoretical direction when the sensor is calibrated. This will result in
interference among the six dimensional force components and limit the calibration precision of the sensor.
4 Coupling errors and Decoupling Methods
For an ideal six-axis F/T sensor, the voltage value of each output channel only depends on the applied
force/torque on this channel, and is completely insensitive to the force/torques on the other five channels.
However, due to the structural design of the sensor, the accuracy of mechanical processing, bonding
technique of strain gauges, transverse effect of strain gauge and detection method[37]
, almost every
component of force/moment acting on the sensor will affect all the output signals of the force sensor, which
is called inter-dimensional coupling. The measurement error due to the coupling between dimensions is
called inter-dimensional coupling error. The inter-dimensional coupling problem is main factor limiting
measuring accuracy of multi-dimensional F/T sensors, so we must decouple the output signals to reduce or
eliminate coupling errors.
Generally, there are two decoupling methods for multi-dimensional F/T sensors. One is hardware
decoupling method, which starts with structural design and mechanical manufacturing process to eliminate
the coupling errors. But this method is difficult to implement and greatly increases the cost of sensor
manufacturing. Song et al. developed a novel four-dimensional F/T sensor with a novel self-decoupled
mechanical structure for human–computer interaction [39]
, seen in Figure15. The experimental results
demonstrated that the four-axis F/T sensor with self-decoupled structure have the maximum measurement
error of 1.5% F.S. Zhao et al. proposed a mechanical decoupling method for parallel three-dimensional
force sensors, which uses rolling friction instead of sliding friction to reduce coupling[40]
. Wu Bo et al.
Polysilicon cross-beam
Aluminum cross-beam B
B: magnetic field
q: distributed electromagnetic
designed a six-dimensional force sensor with self-decoupling structure[41]
, but precise slip clearance and
groove are needed, and the contact force between elastomer and groove side wall will produce additional
coupling. The other method is software decoupling. The traditional static decoupling method, which is most
widely used at present, is to solve decoupling matrix based on the least square method. This method is
simple in principle and can realize online decoupling. Commercial sensors such as six-dimensional F/T
sensors produced by ATI and AMTI Companies adopt this method to decouple. However, because of the
matrix inversion in this method, it is easy to appear ill-conditioned matrix, so that a small perturbation of
experimental data may make the decoupling accuracy change greatly. With the wide application of neural
networks in many fields in recent years, various improved neural networks have been used in the static
decoupling research of multi-dimensional force sensors[42-44]
. Song et al. proposed some static decoupling
methods based on coupling error modeling[45]
, combination of SVM (Support Vector Machine) and
coupling error modeling[46]
. The inter-dimensional coupling error can be reduced to less than 1.5% F.S.
Figure 15 (a) self-decoupled mechanical structure of the 4 DOF F/T sensor; (b) prototype of the 4 DOF F/T sensor.
5 Trends and Challenges
Multi-dimensional F/T sensors technology has been studied for more than 40 years. With the rapid progress
of the robot, haptics, human-computer interaction and virtual reality techniques, the force sensor as one of
the important sources of the interactional information has been continuously improved. The development
trends and challenges of the force sensor research are discussed as follows.
5.1 Trends
With the emergence of various new materials, as well as the development of mechanical manufacturing
technology, MEMS and 3D printing technologies, the precision and reliability of multi-dimensional force
sensor will be greatly improved.
Design of new mechanical structures of elastic beam for the purpose of higher sensitive and lower
interdimensional coupling is always the direction of multi-dimensional force sensor research. However, as
the progress of semiconductor technology and MEMS technology, multi-dimensional force sensors develop
towards miniaturization[47]
, which leads to the force sensor will be more widely used in biomedicine and
virtual reality fields such as minimally invasive surgery, diagnostics tools, remote surgery, rehabilitation,
virtual training, haptic system, and so on[48]
.
On the one hand, many tasks like the force feedback of the maglev control system[49]
and the haptic
interaction of the minimally invasive robotic surgery[50]
, require the real-time measurement of multi-
dimension force signal, which makes the dynamic performance of multi-dimensional force sensor be paid
more attentions than before. On the other hand, as the wearable haptic devices become the pursued
objective of haptic interaction design, the multi-dimensional F/T sensor has been called for miniaturization,
softness and wearability. Furthermore, the 3D printing technology starts applying to fabrication of multi-
dimensional force sensors. Figure 16 shows a 3D-printed cross elastic beam of 6 DOF F/T sensor by using
PEEK material manufactured by Robot Sensor and Control Laboratory at Southeast University, which
demonstrated to be more sensitive to the acting force/torque.
Figure 16 Physical picture of a 3D-printed cross elastic beam of 6 DOF F/T sensor by using PEEK material.
5.2 Challenges
As the development of virtual reality and augment reality, the devices supporting the multi-perception and
interaction become more and more important, which requires the force sensors have higher precision, better
dynamic performance and lower cost. Therefore, the multi-dimensional force sensors for haptic interaction
faces some challenges.
1. Dynamic Calibration and Compensation
For haptic interaction or robot grasp task, the efficiency of online force sensing and feedback is highly
dependent on the performance of multi-dimensional force sensor. The real-time haptic interaction or robot
grasp task requires good dynamic performance of the force sensor. However, due to the low stiffness of
elastic element and the adhesive strain gauges, the inherent frequency of resistive strain gauge based multi-
dimensional force sensor is usually no more than 20KHz. Although it completely meet the dynamic
requirement of haptic interaction or robot grasp tasks, but the inherent frequency will be descended as the
load increases owing to the equivalent mass increasing. That means once the force sensor is installed on
force feedback device or manipulator, its working bandwidth will be reduced a lot. Therefore, the dynamic
performance of multi-dimensional force sensor needs to be compensated to improve its dynamic
performance.
Currently, the research on multi-dimensional force sensor mainly concentrates on the improvement of
mechanical structures of elastic element, static calibration, and decoupling methods, whereas dynamic
calibration and performance analysis are little. And there is no dynamic calibration guideline for multi-
dimensional force sensors. PTB in Germany has done a lot of research on dynamic calibration of uniaxial
force sensor in recent years[51-54]
, and published guidelines for dynamic calibration of uniaxial force
measuring devices in 2017[55]
. It has important reference value for dynamic calibration of multi-
dimensional force sensor.
2. Dynamic Coupling and Decoupling
The interdimensional coupling interference between the output signals of multi-dimensional force sensor
exists not only in the static measurement but also in the dynamic measurement. At present, most of the
dynamic decoupling methods are based on the transfer function analysis, in which the decoupling network
is simple and low precision, but the dynamic accuracy requirement for the force sensor is also urgently
needed. Because of the error of measurement, interference and limitation of identification method, the
dynamic coupling model is often inaccurate. In 2001, Song et al. proposed a dynamic decoupling method
based on diagonal predominance compensation, which can be approximately decoupled by designing a
steady compensator, but it doesn’t reveal the dynamic decoupling relationship of multi-dimensional force
sensor in theory[56]
. Combined with identification and niche genetic algorithm, Ding et al. established and
optimized the dynamic decoupling network for calibration data. The decoupling network has certain
robustness[57]
. In recent years, Xu et al. has studied the Blind Source Separation (BSS) method and applied
it to the dynamic decoupling of multi-dimensional force sensors[58]
. In general, there are few researches on
dynamic decoupling of multi-dimensional force sensor. However, with the wide application of multi-
dimensional force sensors in the field of dynamic measurement such as haptic interaction, there is a
growing need for dynamic decoupling of multi-dimensional force sensor.
6 Applications
6.1 Robotic Applications
With the development of robotic technology, the application fields of robot have been greatly expanded,
and its functions been significantly improved. Intelligentization has become the trend of robotics, and
sensor technology is foundation of intelligent robot. As one of the most basic sensors of robot, multi-
dimensional F/T sensor is mainly applied in the wrist and finger. Force and tactile measurement and
feedback are indispensable when performing reliable and non-destructive manipulation of parts[58]
.
Consequently, multi-dimensional F/T sensors for intelligent robot have proliferated over the past few
decades.
In 1992, a small cylindrical F/T sensor in robot manufacturing was studied by Little[60]
. Park et al.
developed a six-dimensional F/T sensor for an intelligent robotic gripper[61]
. Luo et al. proposed a Stewart
platform 3-dimensional force sensor with distinctive structure of ball joints for robotic fingers[62]
. Wood et
al. designed and fabricated a two-dimensional force sensor with parallel dual cantilever modules
configuration to characterize the lift/drag force from a robotic insect[63]
.
6.2 Biomedical and Biomechanics Applications
Accurate F/T measurements are required in many biomedical applications, such as MIS (minimally
invasive surgery)[64-65]
, diagnostics tools[66-67]
, prosthetics and rehabilitation medicine. The surgeon exploits
the finger palpation capabilities to characterize tissue hardness and to measure pulsating vessels[65]
. During
MIS, the applied force on tissue estimated by its deformation through visual feedback will lead to a serious
limitation for the effectiveness of the operations. Consequently instruments capable of measuring and
delivering F/T to the surgeon are required. Seibold et al. developed a Stewart platform based six-
dimensional F/T sensor for MIS[68]
, which is shown in Figure 17. The F/T sensor can measure manipulation
forces at the instrument’s tip. Dalvand et al. proposed an actuated modular force feedback-enabled
laparoscopic instrument for robotic-assisted surgery[69]
. The instrument is able to measure tip-tissue lateral
interaction forces as well as normal grasping forces.
Figure 17 Instrument for minimally invasive surgery[68]
Biomechanics is a discipline that studies relationship between forces and motion, physiology, pathology
of organisms. As a force measuring element, F/T sensor plays an important role in biomechanics.
Krouglicof et al. developed 6 DOF force sensor for biomechanics and sports medicine based on Stewart
platform[70]
. Song et al. presented a novel three-axis force sensor for measuring the throwing forces of shot-
put athletes[71]
. Hybrid Assistive Limb was developed to assist the disabled in walking by Tsukuba
University[72]
. There are multi-dimensional F/T sensors in aforementioned mechanics instruments to
achieve force measurement and force feedback.
6.3 Haptic Interaction
Haptic interfaces enable person-machine communication through touch, and most commonly, in response
to user movements. Haptics refers to the capability to sense a mechanical environment through touch.
Haptics also consists of kinesthesia, as well as the ability to perceive one’s body position, movement and
weight[73]
. As illustrated in Figure 1, multi-dimensional force sensor is a foundational element for haptic
interaction systems. Haptics applications such as computer interaction, teleoperation, teaching and training,
games, and multimedia require a particular haptic device structure for various purposes[4]
. Prattichizzo et al.
Gripper
F/T sensor
designed a 3-DOF wearable device for cutaneous force feedback[74]
. It consists of a static platform which
placed on the back of the finger and a mobile platform which is responsible for applying forces at the finger
pad. Force sensors on the mobile platform measure the normal component of the force applied to the
fingertip. Hamid et al. developed a flexible endoscopic sensing module for force haptic feedback
integration to measure the directional force[75]
.
7 Conclusions
Multi-dimensional force sensors have been widely applied in the haptic interaction areas, which support the
virtual reality and augment reality technology. Because the performance of multi-dimensional force sensor is
largely dependent upon the mechanical structures of elastic beams, the design of elastic beam with new
mechanical structure becomes main content of multi-dimensional force sensor research. Some typical
mechanical structures of elastic beams of multidimensional force sensor are introduced, and the inter-
dimensional coupling error is analyzed briefly. The static calibration as well as dynamic calibration of the force
sensor are discussed in this paper. At last, we point out the developing trends and challenges of the dimensional
F/T sensor research in future.
References
1 Seth A, Vance J M, Oliver J H. Virtual reality for assembly methods prototyping: a review. Virtual reality, 2011, 15(1):
5-20
DOI: 10.1007/s10055-009-0153-y
2 Ebrahimi E, Babu S V, Pagano C C, et al. An empirical evaluation of visual-haptic feedback on physical reaching
behaviors during 3D interaction in real and immersive virtual environments. ACM Transactions on Applied Perception
(TAP), 2016, 13(4): 19
DOI: 10.1145/2947617
3 Song A, Wu C, Ni D, et al. One-therapist to three-patient telerehabilitation robot system for the upper limb after stroke.
International Journal of Social Robotics, 2016, 8(2): 319-329
DOI :10.1007/s12369-016-0343-1
4 Vu M H, Na U J. A new 6-DOF haptic device for teleoperation of 6-DOF serial robots. IEEE Transactions on
Instrumentation and Measurement, 2011, 60(11): 3510-3523
DOI: 10.1109/tim.2011.2164285
5 Hayward V, Astley O R, Cruz-Hernandez M, et al. Haptic interfaces and devices. Sensor Review, 2006, 24(1): 16-29
DOI: 10.1108/02602280410515770
6 Ma J, Song A. Fast estimation of strains for cross-beams six-axis force/torque sensors by mechanical modeling. Sensors,
2013, 13(5): 6669-6686
DOI: 10.3390/s130506669
7 Qin L, Jiang C, Liu J, et al. Design and calibration of a novel piezoelectric six-axis force/torque sensor. Seventh
International Symposium on Precision Engineering Measurements and Instrumentation. International Society for Optics
and Photonics, 2011, 8321(4):15
DOI: 10.1117/12.903717
8 Dobrzynska J A, Gijs M A M. Polymer-based flexible capacitive sensor for three-axial force measurements. Journal of
Micromechanics and Microengineering, 2012, 23(1): 15009-15019
DOI: 10.1088/0960-1317/23/1/015009
9 Puangmali P, Liu H, Seneviratne L D, et al. Miniature 3-axis distal force sensor for minimally invasive surgical
palpation. IEEE/ASME Transactions On Mechatronics, 2012, 17(4): 646-656
DOI: 10.1109/TMECH.2011.2116033
10 Fu L, Song A. Dynamic characteristics analysis of the six-axis force/torque sensor. Journal of Sensors, 2018, 2018
DOI:10.1155/2018/6216979
11 Kang M K, Lee S, Kim J H. Shape optimization of a mechanically decoupled six-axis force/torque sensor. Sensors and
Actuators A: Physical, 2014, 209: 41-51
DOI:10.1016/j.sna.2014.01.001
12 Zhong X, Zhang X. Review of multi-dimensional force/torque sensor for robots. Transducer and Microsystem
Technologies, 2015, 34(5): 1-4
DOI: 10.13873/J.1000-9787(2015)05-0001-04
13 Song A, Han Y, Hu H, et al. A novel texture sensor for fabric texture measurement and classification. IEEE Transactions
on Instrumentation and Measurement, 2014, 63(7): 1739-1747
DOI: 10.1109/TIM.2013.2293812
14 Song A, Han Y, Hu H, et al. Active perception-based haptic texture sensor. Sensors and Materials, 2013, 25(1): 1-15
15 Li M, Liu J, Qin L, et al. Research of parallel piezoelectric six-axis force/torque sensor’s static characteristics. Journal of
mechanical engineering, 2014, 50(6): 1-7
DOI: 10.3901/JME.2014.06.001
16 Liu W, Li Y J, Jia Z Y, et al. Research on parallel load sharing principle of piezoelectric six-dimensional heavy
force/torque sensor. Mechanical Systems & Signal Processing, 2011, 25(1): 331-343
DOI: 10.1016/j.ymssp.2010.09.008
17 Li Y J, Sun B Y, Zhang J, et al. A novel parallel piezoelectric six-axis heavy force/torque sensor. Measurement,
2009,42(5): 730-736.
DOI: 10.1016/j.measurement.2008.12.005
18 Kim D, Lee C H, Kim B C, et al. Six-axis capacitive force/torque sensor based on dielectric elastomer. Electroactive
Polymer Actuators and Devices (EAPAD), 2013, 8687
DOI: 10.1117/12.2009970
19 Lee D H, Kim U, Jung H, et al. A Capacitive-Type Novel Six-Axis Force/Torque Sensor for Robotic Applications. IEEE
sensors Journal, 2016, 16(8): 2290-229
DOI: 10.1109/JSEN.2015.2504267
20 Kim U, Lee D H, Kim Y B, et al. A Novel Six-Axis Force/Torque Sensor for Robotic Applications. IEEE/ASME
Transactions on Mechatronics, 2017, 22(3): 1
DOI: 10.1109/TMECH.2016.2640194
21 Liang Q, Zhang D, Coppola G, et al. Multi-Dimensional MEMS/Micro Sensor for Force and Moment Sensing: A
Review. IEEE sensors Journal, 2014,14(8):1
DOI: 10.1109/JSEN.2014.2313860
22 Guggenheim J W, Jentoft L P, Tenzer Y, et al. Robust and Inexpensive Six-Axis Force–Torque Sensors Using MEMS
Barometers. IEEE/ASME Transactions on Mechatronics, 2017, 22(2): 838-844
DOI: 10.1109/TMECH.2017.2654446
23 Brookhuis R A, Wiegerink R J, Lammerink T S J, et al. Three-axial force sensor with capacitive read-out using a
differential relaxation oscillator. In: IEEE Conference on Sensors, Baltimore, MD, 2013: 1-4
DOI: 10.1109/ICSENS.2013.6688398
24 Watson P C, Drake S H. Method and apparatus for six degree of freedom force sensing. U. S. Patent 4 094 192, June 13,
1978
25 Coiffet P, Chirouze M. An introduction to robot technology. New York: McGraw-Hill, 1983
26 Yoshikawa T, Miyazaki T. A six-axis force sensor with three-dimensional cross-shape structure. In: IEEE International
Conference on Robotics & Automation, IEEE, 1989
DOI: 10.1109/ROBOT.1989.99997
27 Yu G, Wu Z, Ge Y. State of arts and development trends toward application-oriented force/torque sensors. Robot, 2003,
25(2): 188-192
DOI: 10.1023/A:1021351800842.
28 Huang X. Hu J, Wang J. A non-diametric three-beam structure and its optimum design for six-axis wrist force sensor.
Robot, 1992, 14(5):1-7
29 Schott J. Tactile sensor with decentralised signal conditioning. The 9th IMEKO world Congress, Berlin, 1982.
30 Dwarakanath T A, Dasgupta B, Mruthyunjaya T S. Design and development of a Stewart platform based force–torque
sensor. Mechatronics, 2001, 11(7):793-809
DOI: 10.1016/s0957-4158(00)00048-9
31 Han K, Wang Z, et al. Design of big-scale six-axis force sensor and study on calibration test. Transducer and
Microsystem Technologies. 2016, 35 (5): 87-90
DOI: 10.13873/J.1000-9787(2016)05-0087-04
32 Mastinu G, Gobbi M, Previati G. A new six-axis load cell. Part I: Design. Experimental Mechanics, 2011,51(3):.373-388
DOI: 10.1007/s11340-010-9355-1
33 Liang Q, Zhang D, et al. A novel miniature four-dimensional force/torque sensor with overload protection mechanism.
IEEE Sensors Journal. 2009,9(12): 1741-1747
DOI: 10.1109/jsen.2009.2030975
34 Hu S, Wang H, Wang Y, et al. Design of a novel six-axis wrist force sensor. Sensors, 2018, 18, 3120
DOI:10.3390/s18093120
35 Zhang Q, Song A, et al. Design of a 3 dimensional force sensor. Acta Metrology Sinica, 2017, 39(1): 52-55
DOI: 10.3969/j.issn.10001158.2018.01.12
36 Li H, Duan Z, Zhou J. Approach to static calibration based on coordination transformation for six-dimension force
sensor. Transducer and Microsystem Technologies. 2006, 25(3): 74-76, 80
DOI: 10.1016/S1005-8885(07)60041-7
37 Jin W L, and Mote C D. Development and calibration of a submillimeter three-component force sensor. Sensors and
Actuators A (Physical), 1998, 65(1):89–94.
DOI: 10.1016/s0924-4247(97)01594-x
38 Chen X, Yao Y, Yuan Z. The interference and calibration method of 6-axis force/moment sensors. Journal of transducer
technology, 1995, 2, 37-40
DOI: 10.13873/j.1000-97871995.02.010
39 Song A, Wu J, Qin G, et al. A novel self-decoupled four degree-of-freedom wrist force/torque sensor. Measurement,
2007, 40 (9): 883-891
DOI: 10.1016/j.measurement.2006.11.018
40 Zhao Y, Jiao L, Weng D, et al. Decoupling Principle Analysis and Development of a Parallel Three-Dimensional Force
Sensor. Sensors, 2016, 16(9): 1506
DOI: 10.3390/s16091506
41 Wu B, Cai P, Decoupling analysis of a sliding structure six-axis force/torque sensor, Measurement Science Review,
2013, 13 (4): 187–193
DOI: 10.2478/msr-2013-0028
42 Lei J, Qiu L, Liu M, et al. Application of neural network to nonlinear static de- coupling of robot wrist force sensor, in:
World Congress on Intelligent Control & Automation, IEEE, 2006, 5282–5285
DOI: 10.1109/WCICA.2006.1714077
43 Fu L, Song A. An optimized BP neural network based on genetic algorithm for static decoupling of a six-axis
force/torque sensor. IOP Conference Series: Materials Science and Engineering, 2018, 311:1-9.
DOI: 10.1088/1757-899X/311/1/012002
44 Li Y, Wang G, Yang X, et al. Research on static decoupling algorithm for piezoelectric six axis force/torque sensor
based on LSSVR fusion algorithm. Mechanical Systems and Signal Processing, 2018, 110:509-520
DOI: 10.1016/j.ymssp.2018.03.015
45 Ma J, Song A, Wu J. Research and application of static decoupling for 3-axis wrist force sensor. Acta Metrologica Sinica,
2011, 32(6): 517-521
DOI: 10.3969/j.issn.1000-1158.2011.06.08
46 Ma J, Song A, Xiao J, A robust static decoupling algorithm for 3-axis force sensors based on coupling error model and
𝜀-SVR. Sensors, 2012, 12(12): 14537–14555
DOI: 10.3390/s121114537
47 Zhang W, Lua K B, Truong V T, et al. Zhou G Y. Design and characterization of a novel T-shaped multi-axis
piezoresistive force/moment sensor. IEEE Sensors Journal, 2016, 16(11): 4198-4210
DOI: 10.1109/JSEN.2016.2538642
48 Liang Q K, Zhang D, Coppola G, Wang Y N, Wei S, Ge Y J. Multi-dimensional MEMS/Micro sensor for force and
moment sensing: a review. IEEE Sensors Journal, 2014, 14(8): 2643-2657
DOI: 10.1109/JSEN.2014.2313860
49 Feng H M, Song F Z, Song B. Dynamic analysis of coupling system of multi-axis force sensor structure and its
measuring circuits. In: International Conference on Transportation, IEEE, 2012, 899-902
DOI: 10.1109/TMEE.2011.6199347
50 Kim C, Lee C H. Development of a 6-DoF FBG force–moment sensor for a haptic interface with minimally invasive
robotic surgery. Journal of Mechanical Science and Technology, 2016, 30(8): 3705-3712
DOI: 10.1007/s12206-016-0732-2
51 Ch Schlegel, G Kieckenap, et al. Traceable periodic force calibration. Metrologia, 49 (2012) 224-235
DOI:10.1088/0026-1394/49/3/224
52 Kobusch M. Characterization of force transducers for dynamic measurements. PTB-Mitteilungen, 2015, 125(2): 43-51
53 Vlajic N, Chijioke A. Traceable dynamic calibration of force transducers by primary means. Metrologia, 2016, 53(4):
S136-S148
DOI: 10.1088/0026-1394/53/4/S136
54 Kobusch M, Sascha Eichstädt. A case study in model-based dynamic calibration of small strain gauge force transducers.
ACTA IMEKO, 2017, 6 (1): 3-12
DOI: 10.21014/acta_imeko.v6i1.433
55 DKD-R 3-10: Dynamic calibration of uniaxial force measuring devices and testing machines (basic principles), PTB.
https://doi.org/10.7795/550.20171212A
56 Song G, Zhang W, Zhai Y. Study of dynamic decoupling of sensor based on diagonal predominance compensation.
Chinese Journal of Scientific Instrument, 2001, 22(z4): 12-14
57 Ding M, Liang H, Wang Q, Wang C. Dynamic decoupling method for multi-axis sensor based on niche genetic
algorithm. Chinese Journal of Sensors and Actuators, 2006, 19(3): 667-671.
DOI: 10.3969/j.issn.1004-1699.2006.03.032
58 Wang Y, Xu D, Zhang J, et al. Research on dynamic decoupling of six-axis force sensor base on the procedure neural
networks. Journal of Chongqing University of Arts and Sciences, 2016, 35(5): 34-40
DOI: 10.3969/j.issn.1673-8004.2016.05.009
59 Fahlbusch S and Fatikow S. Micro-force sensing in a micro-robotic system, in Proc. IEEE Int. Conf. Robot. Autom.,
Seoul, Korea, 2001, pp. 3435-3440.
DOI: 10.1109/ROBOT.2001.933149
60 Little R. Force/Torque sensing in robotic manufacturing. Sensors, 1992,9(11):346-348.
61 Park J. Development of the 6-axis force/moment sensor for an intelligent robot’s gripper. Sensors and Actuators A:
Physical, 2005,118(1):127-134.
DOI: 10.1016/j.sna.2004.07.013
62 Luo M, Luo X, and Pan C. A Stewart platform-based 3-axis force sensor for robot fingers, Second International
Conference on Mechanic Automation & Control Engineering. IEEE, 2011.
DOI: 10.1109/MACE.2011.5986851
63 Wood R J, Cho K J, and Hoffman K. A novel multi-axis force sensor for microrobotics applications, Smart Mater.
Struct., 2009, 18(12): 125002.
DOI: 10.1088/0964-1726/18/12/125002
64 Puangmali P, Althoefer K, Seneviratne L D, et al. State-of-the-art in force and tactile sensing for minimally invasive
surgery. IEEE Sensors Journal, 2008, 8(4): 371–381.
DOI: 10.1109/JSEN.2008.917481
65 Puangmali P, Liu H, Seneviratne L D, et al. Miniature 3-axis distal force sensor for minimally invasive surgical
palpation. IEEE/ASME Transactions on Mechatronics, 2012, 17(4): 646–656.
DOI: 10.1109/TMECH.2011.2116033
66 Menciassi A, Eisinberg A, Scalari G, et al. Force feedback-based microinstrument for measuring tissue properties and
pulse in microsurgery. In: Proceedings 2001 ICRA, IEEE International Conference on IEEE, 2001, 626–631.
DOI: 10.1109/ROBOT.2001.932620
67 Papakostas T V, Lima J, and Lowe M. A large area force sensor for smart skin applications. In: Proceedings IEEE
Sensors, Orlando, FL, USA, 2002: 973–978.
DOI: 10.1109/ICSENS.2002.1037366
68 Seibold U, Kubler B, Hirzinger G. Prototype of Instrument for Minimally Invasive Surgery with 6-Axis Force Sensing
Capability. In: Proceedings of the 2005 IEEE, ICRA 2005, Barcelona, Spain, 2005:496-501.
DOI: 10.1109/ROBOT.2005.1570167
69 Dalvand M M, Shirinzadeh B, et al. An actuated force feedback-enabled laparoscopic instrument for robotic-assisted
surgery. The International Journal of Medical Robotics and Computer Assisted Surgery, 2014, 10(1):11-21.
DOI: 10.1002/rcs.1503
70 Krouglicof N, Alonso L M, Keat W D. Development of a mechanically coupled, six degree-of-freedom load platform for
biomechanics and sports medicine. Synthese, 2004, 5: 4426-4431.
DOI:10.1109/ICSMC.2004.1401228
71 Song G, Yuan H, Tang Y, et al. A novel three-axis force sensor for advanced training of shot-put athletes. Sensors and
Actuators A: Physical, 2006, 128(1):60-65.
DOI:10.1016/j.sna.2006.01.016
72 Kazerooni H. Exoskeletons for Human Power Augmentation. IEEE/RSJ International Conference on Intelligent Robots
& Systems, IEEE, 2005.
DOI:10.1109/IROS.2005.1545451
73 Hayward V, Astley O R, Cruz-Hernandez M, et al. Haptic interfaces and devices. Sensor Review, 2006, 24(1):16-29.
DOI: 10.1108/02602280410515770
74 Prattichizzo D, Chinello F, Pacchierotti C, et al. Towards Wearability in Fingertip Haptics: A 3-DoF Wearable Device
for Cutaneous Force Feedback. IEEE Transactions on Haptics, 2013, 6(4):506-516.
DOI: 10.1109/TOH.2013.53
75 Hamid N, Willem B H, Stefano S, et al. A flexible endoscopic sensing module for force haptic feedback integration.
2018 9th Cairo International Biomedical Engineering Conference (CIBEC), 158-161.
DOI: 10.1109/CIBEC.2018.8641817