CHAPTER 19 Haptic Interfaces Using Electrorheological Fluidsservices.eng.uts.edu.au/cempe/subjects_JGZ/ems/ems... · CHAPTER 19 Haptic Interfaces Using Electrorheological Fluids

1

CHAPTER 19Haptic Interfaces Using

Electrorheological Fluids

Constantinos MavroidisRutgers University, The State University of New Jersey

Yoseph Bar-Cohen Jet Propulsion Laboratory/California Institute of Technology

Mourad BouzitRutgers University, The State University of New Jersey

19.1 Introduction / 119.2 Electrorheological Fluids / 319.3 Haptic Interfaces and Electrorheological Fluids / 819.4 MEMICA Haptic Glove / 10

19.4.1 Electrically Controlled Stiffness Elements / 1019.4.2 Electrically Controlled Force and Stiffness (ECFS)

Actuator / 1119.4.3 MEMICA Haptic Glove and System / 13

19.5 ECS Element Model Derivation / 1519.5.1 Static Case / 1619.5.2 Dynamic Case / 17

19.6 Parametric Analysis of the Design of ECS Elements / 1919.7 Experimental ECS System and Results / 2119.8 Conclusions / 2319.9 Acknowledgments / 2419.10 References / 24

19.1 Introduction

Over the past 50 years, it has been known that there are liquids that respondmechanically to electrical stimulation. These liquids, which have attracted a greatdeal of interest from engineers and scientists, change their viscosityelectroactively. These electrorheological fluids (ERF) exhibit a rapid, reversible,and tunable transition from a fluid state to a solid-like state upon the application

2 Chapter 19

of an external electric field [Phule and Ginder, 1998]. Some of the advantages ofERFs are their high-yield stress, low-current density, and fast response (less than1 ms). ERFs can apply very high electrically controlled resistive forces whiletheir size (weight and geometric parameters) can be very small. Their long lifeand ability to function in a wide temperature range (as much as �40°C to+200°C) allows for the possibility of their use in distant and extremeenvironments. ERFs are also nonabrasive, nontoxic, and nonpolluting, meetinghealth and safety regulations. ERFs can be combined with other actuator typessuch as electromagnetic, pneumatic, or electrochemical actuators so that novel,hybrid actuators are produced with high-power density and low-energyrequirements. The electrically controlled rheological properties of ERFs can bebeneficial to a wide range of technologies requiring damping or resistive forcegeneration. Examples of such applications are active vibration suppression andmotion control. Several commercial applications have been explored, mostly inthe automotive industry, such as ERF-based engine mounts, shock absorbers,clutches and seat dampers. Other applications include variable-resistanceexercise equipment, earthquake-resistant tall structures, and positioning devices[Phule and Ginder, 1998].

While ERFs have fascinated scientists, engineers, and inventors for nearly 50years, and have given inspiration for developing ingenious machines andmechanisms, their applications in real-life problems and the commercialization ofERF-based devices has been very limited. There are several reasons for this. Dueto the complexity and nonlinearities of their behavior, their closed-loop control isa difficult problem to solve. In addition, the need for high voltage to controlERF-based devices creates safety concerns for human operators, especially whenERFs are used in devices that will be in contact with humans. Their relativelyhigh cost and the lack of a large variety of commercially available ERFs withdifferent properties to satisfy various design specifications have made thecommercialization of ERF-based devices unprofitable. However, research onERFs continues intensively and new ERF-based devices are being proposed[Tao, 1999]. This gives rise to new technologies that can benefit from ERFs. Onesuch new technological area which will be described in detail here is virtualreality and telepresence, enhanced with haptic (i.e. tactile and force) feedbacksystems, and for use in medical applications, for example.

In this chapter, we first present a review of ERF fundamentals. Then, wediscuss the engineering applications of ERFs and, more specifically, theirapplication in haptics. We describe in detail a novel ERF-based haptic systemcalled MEMICA (remote mechanical mirroring using controlled stiffness andactuators) that was recently conceived by researchers at Rutgers University andthe Jet Propulsion Laboratory [Bar-Cohen et al., 2000a]. MEMICA is intended toprovide human operators an intuitive and interactive feeling of the stiffness andforces in remote or virtual sites in support of space, medical, underwater, virtualreality, military, and field robots performing dexterous manipulation operations.MEMICA is currently being used in a system to perform virtual telesurgeries asshown in Fig. 1 [Bar-Cohen et al., 2000b]. The key aspects of the MEMICA

Haptic Interfaces Using Electrorheological Fluids 3

system are miniature electrically controlled stiffness (ECS) elements andelectrically controlled force and stiffness (ECFS) actuators that mirror thestiffness and forces at remote/virtual sites. The ECS elements and ECFSactuators, which make use of ERFs, are integrated on an instrumented glove.Forces applied at the robot end-effector will be reflected to the user using thisERF device where a change in the system viscosity will occur proportionally tothe force to be transmitted. This chapter, as a case study on the design of ERF-based devices, describes the design, analytical modeling, and experiments thatare currently underway to develop such ERF-based force feedback elements andactuators.

Figure 1: Performing virtual reality medical tasks via the electrorheologcal fluidbased MEMICA haptic interface.

19.2 Electrorheological Fluids

Electrorheological fluids are fluids that experience dramatic changes inrheological properties, such as viscosity, in the presence of an electric field.Winslow first explained the effect in the 1940s using oil dispersions of finepowders [1949]. The fluids are made from suspensions of an insulating base fluidand particles on the order of 0.10 to 100 µm in size. The volume fraction of theparticles is between 20% and 60%. The electrorheological effect, sometimescalled the Winslow effect, is thought to arise from the difference in the dielectricconstants of the fluid and particles. In the presence of an electric field, theparticles, due to an induced dipole moment, will form chains along the field lines,as shown in Fig. 2. The structure induces changes to the ERF�s viscosity, yieldstress, and other properties, allowing the ERF to change consistency from that ofa liquid to something that is viscoelastic, such as a gel, with response times tochanges in electric fields on the order of milliseconds. Figure 3 shows the fluid

4 Chapter 19

state of an ERF without an applied electric field and the solid-like state (i.e. whenan electric field is applied). Good reviews of the ERF phenomenon and thetheoretical basis for ERF behavior can be found in Block and Kelly [1988], Gastand Zukoski [1989], Weiss et al. [1994], and Conrad [1998].

Figure 2: Particle suspension forms chains when an electric field is applied.

Figure 3: Electrorheological fluid at reference (left) and activated states (right)[Courtesy of ER Fluid Developments Ltd, UK].

Under the influence of an electric field, the ERF alters its state from aNewtonian oil to a non-Newtonian Bingham plastic. As a Bingham plastic, theERF exhibits a linear relationship between stress and strain rate like a Newtonianfluid, only after a minimum required yield stress is exceeded. Before that point, itbehaves as a solid. At stresses higher than this minimum yield stress, the fluidwill flow, and the shear stress will continue to increase proportionally with theshear strain rate (see Fig. 4), so that:

yτ = τ + µγ� , (1)


where τ is the shear stress, τy is the yield stress, µ is the dynamic viscosity, andγ is the shear strain. The dot over the shear strain indicates its time derivative,the shear rate. An interesting phenomenon that has been observed by manyresearchers is that the dynamic viscosity becomes negative at high fields (seeFig. 4). This phenomenon can be explained by assuming that fewer, or weaker,bonds are formed at higher shear rates, thus giving a smaller total yield stress andthe effect of negative dynamic viscosity [ER Fluids Developments Ltd., 1998].

Both the yield stress τy and the dynamic viscosity µ are two of the mostimportant parameters that affect the design of ERF-based devices. The dynamicviscosity µ is mostly determined by the base fluid and the electric field. The fieldinduced yield stress τy depends on the electric field strength. For this dependence,some theoretical models have been derived but neither one is yet able to reflectthese relations properly. As a rule of thumb, one can assume that the yield stressincreases quadratically with the electric field strength [Lampe, 1997].

There are two important values for the yield stress: the static yield stress τy,s

and the dynamic yield stress τy,d. The static yield stress is defined as the value ofstress needed to initiate flow, i.e., the stress needed to change from solid toliquid. The dynamic yield stress is the value of stress needed in zero-strain rateconditions to go from a liquid to solid. Which one is larger is different from fluidto fluid. Most of the time, the static yield stress is higher than the dynamic yieldstress. This phenomenon, called �stiction,� is highly dependent on the particlesize and shape [Weiss et al., 1994].

Shear Stress vs Shear Rate

0

0.5

1

1.5

2

2.5

3

3.5

0 1000 2000 3000 4000 5000 6000 7000

Shear rate, 1/sec

Shea

r Str

ess,

kPa

00.81.62.43.24

Figure 4: Relationship between shear stress and shear strain rate in an ERF[Reprinted with permission of ER Fluids Developments Ltd.].

6 Chapter 19

Another important parameter that needs to be known for ERFs is the currentdensity J, defined as the current per unit electrode area. This parameter is neededto estimate the power consumption of ERF-based devices. Measurement ofelectric current through ERF materials is believed to be the result of chargeleakage between particles [Weiss et al., 1994].

ERF properties change with temperature and this can have a tremendouseffect on the performance of ERF-based devices. �Good� ERFs should showconstant properties over a large range of temperatures. There is no unified modeldescribing the temperature dependence of the parameters of ERFs. Thistemperature dependence changes from fluid to fluid. The biggest temperatureproblem for ERFs results from the huge increase of current density withincreasing temperatures. This increases power consumption but also increasessafety concerns for human operators of ERF devices.

A good database of commercially available ERFs, including propertycomparison tables, can be found in Lampe [1997]. A review of the materialcompositions for ERF patents can be found in Weiss et al. [1994]. As an exampleof ERFs we present here the electrorheological fluid LID 3354, manufactured byER Fluids Developments Ltd. [ER Fluids Developments Ltd, 1998] which isused in the work presented in the next sections.

LID 3354 is an electrorheological fluid made of a 35% volume of polymerparticles in fluorosilicone base oil. It is designed for use as a general-purpose ERfluid with an optimal balance of critical properties. Its physical properties aredensity: 1.46 × 103 kg/m3; viscosity: 125 mPa sec at 30°C; boiling point: >200°C; flash point: >150°C; insoluble in water; freezing point: < �20°C. Thefield dependencies for this particular ERF are

( ), ,y s s refC E Eτ = − 2, ,y d dC Eτ = 2 ,o vC Eµ = µ − (2)

where µo is the zero field viscosity; Cs, Cd, Cv, and Eref are constants supplied bythe manufacturer. The subscripts s and d correspond to the static and dynamicyield stresses. The formula for static yield stress is only valid for fields greaterthan Eref. Figures 5(a), (b), and (c) are a graphical representation of Eq. (2) for theERF LID 3354. Figure 5(d) shows the dependence of the current density at 30°Cas a function of the field. Figure 5(e) shows the coefficient Cd of Eq. (2) as afunction of temperature.

Control over a fluid�s rheological properties offers the promise of manypossibilities in engineering for actuation and control of mechanical motion.Devices that rely on hydraulics can benefit from ERF�s quick response times andreduction in device complexity. Their solid-like properties in the presence of afield can be used to transmit forces over a large range, and they have a largenumber of other applications. A good description of the engineering applicationsof ERFs can be found in Duclos et al. [1992] and Coulter et al. [1994]. Devicesdesigned to utilize ERFs include engine mounts [Sproston et al., 1994], active


0

1

2

3

4

5

6

7

8

9

0 1 2 3 4

Field, kV/mm

Yiel

d st

ress

, kPa

0

0.5

1

1.5

2

2.5

3

3.5

4

4.5

0 1 2 3 4

Field, kV/mm

Yiel

d st

ress

, kPa

(a) Static Yield Stress at 30°C (b) Dynamic Yield Stress at 30°C

-200

-150

-100

-50

0

50

100

0 1 2 3 4

Field, kV/mm

Plas

tic v

isco

sity

, mPa

.sec

-2

0

2

4

6

8

10

12

0 1 2 3 4

Field, kV/mm

Cur

rent

Den

sity

, m

icro

A/s

q.cm

(c) Dynamic Viscosity at 30°C (d) Current Density at 30°C

0

0.05

0.1

0.15

0.2

0.25

0.3

0.35

0.4

-20 0 20 40 60 80

Temperature, °C

A,

kPa.

mm

2/kV

2

(e) Temperature Dependence of Cd

Figure 5: Technical information diagrams for the ER Fluid LID 3354 [ER FluidsDevelopments Ltd., 1998]. [Reprinted by permission of ER Fluids DevelopmentsLtd.]

8 Chapter 19

dampers and vibration suppression [Choi, 1999], clutches [Bullough et al., 1993],brakes [Seed et al., 1986], and valves [Coulter et al., 1994]. An importantengineering application of ERFs is vibration control; a good review of the subjectcan be found in Stanway et al. [1996]. The application of ERFs in robotics hasbeen very limited. They have been used mainly as active dampers for vibrationsuppression [Ghandi et al., 1987; Furusho et al., 1997]. Recently, an ERF-basedsafety-oriented mechanism has been proposed for human-robot cooperativesystems [Arai et al., 1998].

19.3 Haptic Interfaces and Electrorheological Fluids

To address the need for interacting with remote and virtual worlds, theengineering community has started developing haptic (tactile and force) feedbacksystems [Burdea, 1996]. At the present time, haptic feedback is a less-developedmodality for interacting with remote and virtual worlds compared with visual andauditory feedback. Thus, realism especially suffers when remote and virtual tasksinvolve dexterous manipulation or interaction in visually occluded scenes. A verygood description of the current state of the art in haptic and force feedbacksystems can be found in Burdea [1996] and Bar-Cohen et al. [2000c].

Tactile sensing is created by skin excitation that is usually produced bydevices known as �tactile displays.� These skin excitations generate the sensationof contact. Force-sensitive resistors, miniature pressure transducers, ultrasonicforce sensors, piezolectric sensors, vibrotactile arrays, thermal displays, andelectrorheological devices are some of the innovative technologies that have beenused to generate the sensation of touch. While tactile feedback can be conveyedby the mechanical smoothness and slippage of a remote object, it cannot producerigidity of motion. Thus, tactile feedback alone cannot convey the mechanicalcompliance, weight, or inertia of the virtual object being manipulated [Burdea,1996].

Force-feedback devices are designed to apply forces or moments at specificpoints on the body of a human operator. The applied force or moment is equal orproportional to a force or moment generated in a remote or virtual environment.Thus, the human operator physically interacts with a computer system thatemulates a virtual or remote environment. Force-feedback devices includeportable and nonportable interfaces. Force-feedback joysticks, mice [ImmersionCorp., 1999; Haptic Technologies, 1999], and small robotic arms such as thePhantom [Sensable Technologies, 1999] are nonportable devices that allow usersto feel the geometry, hardness, and/or weight of virtual objects.

Portable systems are force-feedback devices that are grounded to the humanbody. They are distinguished as arm-exoskeletons if they apply forces at thehuman arm and as hand masters if they apply forces at a human's wrist and/orpalm. Portable hand masters are haptic interfaces that apply forces to the humanhand while they are attached at the human operator forearm. In most cases, thesesystems look like gloves where the actuators are placed at the human forearm,and forces are transmitted to the fingers using cables, tendons and pulleys. The


CyberGrasp is an example of such a system. It is a lightweight, force-reflectingexoskeleton glove that fits over a CyberGlove and adds resistive force feedbackto each finger via a network of tendons routed around an exoskeleton [VirtualTechnologies, 1999]. The actuators are high-quality dc motors located in a smallenclosure on the desktop. The remote reaction forces can be emulated very well;however, it is difficult to reproduce the feeling of �remote stiffness.�

To date, there are no effective commercial unencumbering haptic feedbackdevices for the human hand. Current hand master haptic systems, while they areable to reproduce the feeling of rigid objects, present great difficulties inemulating the feeling of remote/virtual stiffness. In addition, they tend to beheavy and cumbersome with low bandwidth, and they usually allow limitedoperator workspace.

During the last 10 years, some researchers have proposed the use of ERFs inan effort to improve the performance of haptic interfaces. There are manyproperties of ERFs that can greatly improve the design of haptic devices. Theirhigh-yield stress, combined with their small sizes, can result in miniature hapticdevices that can fit easily inside the human palm without creating anyobstructions to human motion. ERFs do not require any transmission elements toproduce high forces, so direct-drive systems can be produced with less weightand inertia. The possibility of controlling the fluids' rheological properties givesdesigners of ERF-based haptic system the possibility of controlling the systemcompliance, and hence, mirror accurately remote or virtual compliance. Finally,ERFs respond almost instantly�in milliseconds�which can permit very highbandwidth control important for mirroring fast motions. The only concern that adesigner of ERF-based haptic interfaces may have is the need for high voltages todevelop the forces and compliance required. This has two consequences: (a) itincreases the complexity of the electronic system needed to develop the highvoltage and (b) it raises safety concerns for the human operator. Both issues canbe solved easily with modern electronic circuit design techniques. Nowadays,low-power, small-size circuits can be used to generate the required high voltageusing a very low current on the order of micro-amps. Consequently, the requiredpower becomes extremely low�in the order of milliwatts�posing no hazard forhuman operators.

Kenaley and Cutkosky were the first to propose the use of ERFs for tactilesensing in robotic fingers [1989]. Based on that work, several workers proposedthe use of ERFs in tactile arrays to interact with virtual environments [Wood,1998] and also as assistive devices for the blind to read the Braille system. Thefirst to propose this application of ERFs was Monkman [1992]. Continuing thiswork, Professor and his group at the University of Hull, UK, developed andtested experimentally a 5 × 5 ERF tactile array [Taylor et al., 1996]. Furusho andhis group at Osaka University in Japan developed an ERF-based planar force-feedback manipulator system that interacts with a virtual environment[Sakaguchi and Furusho, 1998]. This system is actuated by low-inertia motorsequipped with an ER clutch. An ERF-based force-feedback joystick has beendeveloped at the Fraunhofer-Institute in Germany. The joystick consists of a ball

10 Chapter 19

and socket joint where ERF has been placed in the space between the ball and thesocket. The operator feels a resistive force to his/her motion resulting from thecontrolled viscosity of the ERF [Böse et al., 2000].

Finally, researchers at Rutgers University and JPL are developing an ERF-based force-feedback glove called MEMICA [Bar-Cohen et al., 2000; Mavroidiset al., 2000; Pfeiffer et al., 1999]. This system is described in detail in the nextsection of this chapter.

19.4 Memica Haptic Glove

This section presents the development of a haptic interfacing mechanism that willenable a remote operator to �feel� the stiffness and forces at remote or virtualsites. These interfaces are based on novel mechanisms that were conceived byJPL and Rutgers University investigators, in a system called MEMICA [Bar-Cohen et al., 2000; Mavroidis et al., 2000; Pfeiffer et al., 1999]. The key aspectsof the MEMICA system are miniature ECS elements and ECFS actuators thatmirror the forces and stiffness at remote/virtual sites. The ECS elements andECFS actuators that make use of ERFs to achieve this feeling of remote/virtualforces are placed at selected locations on an instrumented glove to mirror theforces of resistance at the corresponding locations in the robot hand. In thissection, the mechanical design of the ECS elements and of the ECFS actuatorsand their integration on the MEMICA glove are presented.

19.4.1 Electrically Controlled Stiffness Elements

The ECS element stiffness is modified electrically by controlling the flow of anERF through slots on the side of a piston (Fig. 6). The ECS element consists of apiston that is designed to move inside a sealed cylinder filled with ERF. The rateof flow is controlled electrically by electrodes facing the flowing ERF whileinside the channel.

Figure 6: ECS element and its piston.


To control the �stiffness� of the ECS element, a voltage is applied betweenelectrodes facing the slot, affecting the ability of the liquid to flow. Thus, the slotserves as a liquid valve since the increased viscosity decreases the flow rate ofthe ERF and varies the stiffness that is felt. To increase the stiffness bandwidthfrom free flow to maximum viscosity, multiple slots are made along the pistonsurface. To wire such a piston to a power source, the piston and its shaft are madehollow and electric wires are connected to electrode plates mounted on the sideof the slots. The inside surface of the ECS cylinder surrounding the piston ismade of a metallic surface and serves as the ground and opposite polarity. Asleeve covers the piston shaft to protect it from dust, jamming or obstruction.When a voltage is applied, potential is developed through the ERF along thepiston channels, altering its viscosity. As a result of the increase in the ERFviscosity, the flow is slowed significantly and resistance to external axial forcesincreases. Section 19.5 presents the dynamic modeling equations for such anECS element. Section 19.6 presents a parametric study of the design of the ECSelement while Sec. 19.7 presents experimental results obtained from a large-scaleprototype of an ECS element.

19.4.2 Electrically Controlled Force and Stiffness (ECFS)Actuator

To produce complete emulation of a mechanical �tele-feeling� system, it isessential to use actuators in addition to the ECS elements in order to simulateremote reaction forces. Such a haptic mechanism needs to provide both activeand resistive actuation. The active actuator can mirror the forces at thevirtual/remote site by pulling the finger or other limbs backward. The mainobjective of this section is to present a small linear actuator that utilizes the ECSconcept developed in Sec. 19.4.1. This actuator operates as an inchworm motor(as shown in Fig. 7) and consists of active and passive elements, i.e., two brakesand an expander, respectively. One brake locks the motor position onto a shaftand the expander advances (stretches) the motor forward. While the motor isstretched forward, the other brake clamps down on the shaft and the first brake isreleased. The process is repeated as necessary, inching forward (or backward) asan inchworm does in nature.

Using the controllability of the resistive aspect of the ERF, a brake can beformed to support the proposed inchworm. A schematic description of the ECFSactuator is shown in Fig. 8. The actuator consists of two pistons (brake elements)and two electromagnetic cylinders (pusher elements). Similar to ECS, each pistonhas several small channels with a fixed electrode plate. When an electric field isinduced between the piston anode and cylinder cathode, the viscosity of the ERFincreases and the flow rate of the fluid though the piston channel decreases,securing the piston to the cylinder wall.

12 Chapter 19

Figure 7: Concept of the inchworm motor.

Figure 8: ECFS actuator configuration.

Each of the electromagnetic cylinders consists of a coil and a ferromagneticcore integrated within the piston. When a current impulse is passed through thewinding, an electromagnetic field is induced and, depending on the currentdirection, the cylinder moves forward or backward. This actuation principle isshown as a set of sequence diagrams in Fig. 9. In the first step, piston P1 is fixedrelative to the cylinder by activating its electrode; then, triggering theelectromagnetic cylinder moves piston P2 forward. The ERF located between the

ER Fluid

Ferromagneticcylinder

EC1 winding

EC2 winding

Piston 1

Anode plate

Piston 2

Cylinder (Cathode)

Brake MoverBrake

Shaft


two pistons is then displaced backward through the channels of piston P2. Ahorizontal channel is added at the surface of a ferromagnetic cylinder to increasethe flow rate of the fluid. In the second step, the ERF in the channels of piston P2is activated and P2 becomes fixed to the cylinder while P1 is disconnected fromthe cylinder. The current in the first winding is then reversed, changing thepolarization of the magnetic cylinder, pushing P1 forward relative to P2.

Figure 9: Sequence diagram of ECFS actuator operation.

At each cycle, the pistons move forward or backward with very smalldisplacement (<1.5 mm). The duration of each cycle is close to a millisecond,corresponding to the response time of the ERF. The ECFS actuator can thenreach a speed higher than 15 cm/s with a piston displacement equal to 0.5 mm at3-ms cycle duration. The electromagnetic cylinder is designed to produce thesame force as the resistive force of the piston inside the ERF, which isabout 15 N.

19.4.3 MEMICA Haptic Glove and System

A haptic exoskeleton integrates the ECS elements and ECFS actuators at variousjoints. As shown in Fig. 10, the actuators are placed on the back of the fingers,out of the way of grasping motions. The natural motion of the hand is thenunrestricted. Also, this configuration is capable of applying an independent force(uncoupled) on each phalange to maximize the level of stiffness/force feedbackthat is �felt� by the operator.

Different mounting mechanisms are evaluated and some of them areschematically represented in Fig. 11. Of these solutions, the most ergonomicseems to be the arched actuator [Fig. 11(a)], which has better fitting with thefinger motion and geometry. Since ERF viscosity is higher than air, there is noneed for tight tolerance for the ECFS piston and its cylinder. The secondproposed solution uses a curved sliding rail, which is also suitable for a fingermotion. The third solution uses a flexible tendon connected directly to the piston

ERFluid

ERFluid

ER Fluid ER Fluid

V

t

t

t

t

P2

EC2I

EC1I

VP1

1 2 1 2 1 21 2

Displacement

21

P1

P2

EC1

EC2

14 Chapter 19

inside the cylinder where the tendon length can be adjustable to the user phalangelength.Once the appropriate exoskeleton mechanism is chosen, we will prototype acomplete haptic glove with 16 actuators: 3 actuators for the thumb (2 for theflexion motion and 1 for the abduction/adduction motion), 4 actuators for theindex (3 for the flexion motion and 1 for the abduction/adduction motion) and 3actuators for the three other fingers (flexion only). Figure 12 contains 3Ddrawings of the exoskeleton glove showing the positioning of all actuators on thefingers. For precision grasping with the thumb and the index finger whereabduction/adduction motion is involved, it is necessary to integrate actuators thatwould resist this motion.

Figure 10: Mounting of an ECFS actuator on the finger phalange.

Figure 11: Different exoskeleton mechanisms.

ECFS Actuator

Joint

Curved sliding railArched ECFS Acutor

Joint

Sensor ECFS ActuatorAdjustable tendon

Load cell

Sequeezingvelcro

Flexiblesmetal strip

Finger phalange

Foam

ECFS Actuator

JointHall sensor

Adjustable ring

Load cellandHall sensor Load cell

ECFS Actuator

θθθθ

F

Joint


Figure 12: 3D View from the MEMICA system and close-up view of the glove.

19.5 ECS Element Model Derivation

In this section we present the dynamic modeling equations of an ECS element.These equations calculate the force that a human operator will feel when he/sheis haptically interacting with an ECS element. The calculated force is thefunction of the applied voltage to the ERF, the geometry of the ECS element, andthe characteristics of the ERF.

The field E is provided by an applied voltage V. Considering a Gaussiansurface A of radius r and length l between the inner ri, and outer ro radii (see Fig.9), Gauss�s law is used to find the electric field. A charge of q is assigned to thecharged core and the field is known to be in the radial direction, so that the dotproduct becomes the scalar product, and

o

qE dA⋅ =ε�

��

� ⇔ ( )2o

qEA E rl= π =ε

⇔ �( )2 o

qE r rrl

=πε

�

, (3)

where oε is the electrical permittivity of free space.This expression for the electric field can now be used in the definition of a

difference in potential, computing the difference between the inner and outerwalls, V:

Close-up views ofthe glove

MEMICA system overall view

16 Chapter 19

� ln2 2

i

o

ro

o o ir

rq qV r drrl l r

� � � � � �= − ⋅ =� � � � � �πε πε� � � � � ��

� . (4)

Equations (3) and (4) are combined to relate the electric field directly to theapplied voltage and geometry:

1 1 1ln2

ln ln

o

o i o o

i i

rq VEl r r rr r

r r

� ��

� �� = =� �� πε � � � ��

. (5)

The force Fapp applied by the operator is equal to the reaction force FR he orshe will feel. This reaction force is the sum of three forces: a shear force Fτ, apressure force Fp, and a friction force Ff. Assuming that the interface between thepiston and cylinder is frictionless, the total reaction force can be computed as thesum of the remaining components.

R app p f pF F F F F F Fτ τ= = + + ≈ + . (6)

19.5.1 Static Case

Since the pressure force is a result of the flow of the ERF through the channels,there will be no pressure force term for the static force.

The shear force term is calculated by considering the entire surface area incontact with ERF, which is the surface area of all of the channels. Since shearstress for a given voltage is a function of radius, the shearing force acting on eachchannel will be the sum of a term calculated at the outer radius, a term calculatedat the inner radius, and twice a term integrated with respect to the area of the sidewalls of the channel. The area of an inner or outer channel wall, i.e., Arw(ri) orArw(ro) respectively, is the product of the channel length L and the arc subtendedby the radius through the angular width of the channel θ: The area Asw of eachone of the side walls is equal to the product of the channel length L and thechannel width ∆r (i.e., difference in radii):

( ) ( )rwA r r L= θ swA L r= ∆ swdA Ldr= . (7)

Since the contribution to the total reaction force from each channel is thesame, the total reaction force�the product of shear stress and area�can befound by multiplying the contribution from one channel by the number ofchannels N:


( ) ( ) ( ) ( ) ( ), 2o

i

r

R s s o rw o s i rw i s swr

F N r A r r A r r dA� �

= τ + τ + τ� ��

� . (8)

The expressions of the shear stresses are calculated from Eqs. (1) and (2).Since the static case is being considered, the shear rate is zero in Eq. (1).Combining the information from Eqs. (1), (2), (8) and after mathematicalprocessing, the following expression is obtained for the reactions force in staticmode:

( )( ),22 2

lnR s s o i ref

o

i

F NC L V r r r Err

� ��

θ� �� = + − ∆ + θ +� ��

. (9)

19.5.2 Dynamic Case

In order to calculate the reaction force felt by the operator when the piston ismoving, the dynamic shear stress and the pressure force must be considered.First, the shear force Fτ,d is calculated as in Eq. (8), replacing the subscript s withd:

, ( ) ( ) ( ) ( ) 2 ( )o

i

r

d d o rw o d i rw i d swr

F N r A r r A r r dAτ

� �= τ + + τ� �

� �� τ . (10)

Referring back to Eq. (1), we note that the shear rate is equal to the velocitygradient given by ν/∆r, with the velocity of the ERF equal in magnitude to thevelocity v of the piston. Now using the expressions for dynamic yield stress andplastic viscosity from Eq. (2) in Eq. (1) and the expression for the electric fieldfrom Eq. (5). and subsequently substituting in Eq. (10), the following equation isobtained:

2

, 21 1 1 12

ln

2 .

d d vo i i o o

i

o io

v VF NL C Cr r r r r r

r

r rNL vr

τ� �� = − θ + + −� �� ∆� � � � � � � ��

� ��

� �+� �+ µ + θ� �� ∆� ��

(11)

The next step in calculating the reaction force for the dynamic case iscalculating the pressure force Fp,d. The pressure force can be determined by

18 Chapter 19

finding the pressure gradient in the channels, which is found through a forcebalance of a differential fluid element with some acceleration, a, equal inmagnitude to the acceleration of the piston The differential fluid element isconsidered to have a differential mass dm, a length dx and an area Af :

,( ) df

F dpdm a dx A p p dxNL dx

τ � �� = − − +� ��

. (12)

The differential mass element dm can be written as a function of the densityρ:

fdm A dx= ρ . (13)

The area of the fluid is written as

( ) ( )2 2 2 2

2 2f o i o iA r r r rθ θ= π − π = −π

. (14)

Simplifying Eq. (12) and solving for the pressure gradient,

,d

f

Fdp adx NLA

τ� �− = − ρ� ��

. (15)

The pressure force is found by multiplying the pressure drop by the pistonarea Ap:

,p d p p pdp dpF pA p p L A LAdx dx

� �� = ∆ = − + = −� ��

, (16)

Where the area of the piston is given by:

( ) ( )2 2 2 2 2 2

2 2p o o i o o iN NA r r r r r rθ θ= π − π − π = π − −

π(17)

So,

( )( )

22 2 2

, ,2 2

12

2

op d d o o i

o i

r NF F L r r r aN r rτ

� �� π θ� �= − − ρ π − −� � � �θ � �� −� ��

. (18)


The total reaction force for the dynamic case will be the sum of these twoforces. After mathematical manipulation,

( )

( )

2, 2 2

2

2

2 2 2

2

2 2 2

ln

2

oR d

o i

o id v o

o i i o oi

o o i

rF NLN r r

r rv VC C vr r r r r rr

r

NL r r r a

� �� π

= � �θ� �−� ��

� �� +θ θ � �� − + + − + µ + θ� �� ∆ ∆� � � ��

θ� �−ρ π − −� ��

(19)

19.6 Parametric Analysis of the Design of ECS Elements

The analytical Eqs. (9) and (19) will be used to evaluate how various geometricand input factors can influence the reaction forces felt by the user. The resultsfrom this study are very important for the design of the ECS elements.

Human studies have shown that the controllable maximum force that ahuman finger can exert is between 40 and 50 N [Burdea, 1996]. However,maximum exertion forces create discomfort and fatigue to the human operator.Comfortable values of exertion forces are between 15% and 25% of thecontrollable maximum force exerted by a human finger. Hence, the designobjective is to develop an ECS element that will be able to apply a maximumforce of 15 N to the operator. We are primarily interested in the dependence ofthe reaction forces from the ECS when the following parameters are changing:voltage applied V, motion characteristics imposed by the user such as the velocityv, acceleration a, geometric characteristics of the piston such as geometry of thechannel defined by the inner and outer diameters ri and ro, and the angle of thechannel θ. Therefore, in our study it is desired to find the ranges for theseparameters that will result in the desired maximum force output of 15 N.

The parameters related to the fluid ERF LID 3354, shown in Table 1, weredetermined from the manufacturer's specifications [ER Fluids DevelopmentsLtd., 1998]. The default geometric parameters of the ECS element shown inTable 2 have been determined from the dimensions of commercially availablesensors and electronic equipment that will be used for measuring and actuatingthe device, as well as any manufacturing and machinability constraints. In thefirst prototype that is presented in this work (see Sec. 19.7), no effort forminiaturization was made since the goal was to prove the concept that ERFs canbe used to create haptic feedback. The default values for motion characteristics

20 Chapter 19

were selected based on representative values of the maximum velocities andaccelerations that a human finger can develop (see Table 3).

Table 1: ERF LID 3354 parameters.(see Sec. 19.2 for parameter definition)

Cd 0.00026Cv 0.198 E-7µ0 0.125ρ 1460 kg/m3

Table 2: Values for the geometric parameters.(see Section 19.5 for parameter definition)

L 0.0254 mri 0.011316 mro 0.012065 m∆r 0.000749 mN 12θ 0.47 rad (27°)

Table 3: Values for the motion characteristics.(see Sec. 19.5 for parameter definition)

a 0.01 m/s2

v 0.1 m/s

Voltage is the principal parameter of interest in this study since it will beused for controlling the compliance of the ERF. We calculate the maximumvoltage needed for achieving a reaction force of 15 N. Setting the default valuesin Eqs. (9) and (19) and changing the voltage the force has been calculated and isshown in Fig. 13. As expected, the relationship of force to voltage is linear in thestatic case and parabolic in the dynamic case. In the static case, a voltage ofapproximately 2 kV is needed to achieve the desired force of 15 N. In thedynamic case, the desired force output of 15 N is reached using 1 kV.

A similar parametric study revealed that the reaction force is almostindependent of the velocity and acceleration imposed by the user. This is due tothe fact that the velocity and acceleration contributions in the reaction force aremuch smaller than the voltage-related term.

The piston geometry is another important factor that affects the reaction forcefelt by the user. In the results presented in this section, the outer diameter of thepiston changes from 0.012 m to 0.014 m while the voltage is equal to 1 kV. Allother parameters take the default values shown in Tables 1, 2, and 3. Thecalculated reaction forces are shown in Fig. 14. It is clearly seen that, as the outerdiameter increases, the reaction force decreases dramatically. On the other hand,as the outer diameter approaches the value for the channel inner diameter, the


reaction forces become infinite. The thinner the piston channels are, the larger thereaction force is, hence the required voltage can be reduced.

In a similar way, the channel angle changed from 0 to 0.5 rad and the forcewas calculated. Under a certain minimum value of θ, the reaction force dropsdramatically. Also there is a maximum limit for θ after which the reaction forceis constant. Therefore, optimal values for θ are around 0.4 rad (i.e. 30 deg).

The parameters N and L affect the reaction force in a linear way. Increasingthese parameters increases the force for a given voltage. However, thedimensions of the piston limit L and the values of θ limit the number of channelsN.

Figure 13: Force as a function of voltage.

Figure 14: Force (N) vs. outer diameter ro .

19.7 Experimental ECS System and Results

To test the concept of controlling stiffness with a miniature ECS element, alarger-scale test bed has been built at the Rutgers Robotics and MechatronicsLaboratory. This test bed, shown in Figs. 15 and 16, is equipped withtemperature, pressure, force, and displacement sensors to monitor the ERF's state.The cylinder is mounted on a fixed stainless steel plate to maintain rigidityduring normal force loading. The top plate is also stainless steel and serves as thebase for the weight platform. Beneath the platform, around the stainless steelshaft, is a quick release collar that allows the force to be released by the operator.

22 Chapter 19

Figure 15: Experimental test bed.

Figure 16: Actual prototype system.

The shaft, which transmits the force down into the cylinder, is restrained to onlyone-dimensional motion through a linear bearing mounted to the top plate. Thehalf-inch solid shaft is reduced by an adapter to a quarter-inch aircraft steelhollow shaft. At this junction there is a load cell and flange bracket mounted forthe wiper shaft of the displacement sensor. The quarter-inch shaft inserts throughthe ERF chamber�s top plate and a small bundt cup that is needed to minimizeleaking from the chamber during operation. Within the chamber, theexperimental piston is attached to the shaft with e-clips secured at the top andbottom of the piston. The chamber itself is a 1-in. internal diameter beaded Pyrexpiping sleeve, 6 in. in length. Pyrex allows visual observation of the ERF duringactuation. In order to apply voltage to the fluid, supply wires are run down


through the hollow shaft and into the piston, where the electrical connections aremade to the channel plates. Threaded into the bottom plate of the chamber is thedual pressure and temperature sensor. The final sensor is mounted alongside thechamber and affixed with a flanged bracket to the chamber.

Six system parameters are measured during experimentation: voltage,current, force, displacement, pressure and temperature. All sensor signals areinterfaced directly to analog-to-digital boards located in a Pentium II PC and areprocessed using the Rutgers WinRec v.1 real-time control and data acquisitionWindows NT-based software. In addition, all sensors are connected to digitalmeters located inside the interface and control box. Sensor excitation voltages aresupplied by 5 V from the PC or by the meter provided with the sensor itself.

Extensive experimental tests are currently underway to determine therelationship of the reaction force to the applied voltage, human motion,temperature, and pressure changes and to verify Eqs. (9) and (19). Representativeresults from these tests are shown in Figs. 17(a) and (b). In Fig. 17(a), no voltageis applied to the device. Four different weights equal to 2.75 lb, 5.50 lb, 8.25 lb,and 11 lb are placed individually on the weight platform. Each time the quickrelease collar is released, the piston displacement induced by the weight isrecorded. A very fast descent of the piston is observed for all the weights. In Fig.17(b), the same procedure is followed but this time a voltage of 2 kV is appliedon the ERF. It can clearly be seen that the piston is showing a very slow descentand for the lightest weight (i.e., the 2.5 lb) no motion is observed. Thisexperiment shows that when the electrical field is enabled, the viscosity of theERF is such that the ECS element can resist the gravity forces from the weights.

Figure 17: Piston displacement. (a) ER fluid with no field; (b) ER fluid with fieldenabled (2 k V dc).

19.8 Conclusions

Using EAPs as smart materials can enable the development of many interestingdevices and methodologies. In this chapter, the authors presented the conceptualapplication of electrorheological fluids (ERFs) to address the need for hapticinterfaces in such areas as automation, robotics, medicine, games, or sports.Using such EAP fluids, one may be able to construct a system that allows one to�feel� the environment compliance and reaction forces at remote or virtualrobotic manipulators. The ability to have a human operator control a remote

24 Chapter 19

robot in the sense of telepresence addresses the realization that there are sometasks that can be best performed by a human but may be too hazardous forphysical presence. Using a haptic interface as described in this chapter allowshuman operators to perform the tasks without the associated risks.

A new mechanism was described that can allow operators to sense theinteraction of stiffness forces exerted on a robotic manipulator. An analyticalmodel was reviewed and experimental data were described. A key to the newhaptic interface is the so-called electrically controlled stiffness (ECS) elementwhich has been demonstrated in a scaled-size experimental unit that proves thefeasibility of the mechanism. A conceptually novel ERF-based haptic systemcalled MEMICA that is based on such ECS elements was described in thischapter. MEMICA is intended for operations that support space, medical,underwater, virtual reality, military, and field robots performing dexterousmanipulations. For medical applications, virtual procedures can be developed assimulators for training doctors, an exoskeleton system can be developed toaugment the mobility of handicapped or ill persons, and remote surgery can beenabled.

19.9 Acknowledgments

Work at Rutgers University was supported by NASA's Jet Propulsion Laboratory(JPL) and CAIP�Center for Advanced Information Processing. Research atJPL/Caltech was carried out under a contract with the National AeronauticsSpace Agency. The authors would like to thank Mr. Charles Pfeiffer, Mr. AlexPaljic, Ms. Jamie Lennon, Ms. Sandy Larios, and Ms. Sarah Young, RutgersUniversity, for providing assistance during the development of the MEMICAconcept. Also, the authors would like to thank Dr. Benjamin Dolgin, Ph.D., JPL,Pasadena, CA, Dr. Deborah L. Harm, Ph.D., NASA JSC, Houston, TX, Mr.George E. Kopchok, Harbor-UCLA Medical Center, Los Angeles, CA andRodney White, M.D., Harbor-UCLA Medical Center, Los Angeles, CA, for theirhelpful comments and suggestions.

19.10 References

Bar-Cohen, Y., Pfeiffer C., Mavroidis C., and Dolgin B., �MEMICA: a conceptfor reflecting remote-manipulator forces,� NASA Tech Briefs, Vol. 24, No. 2,pp. 7a-7b, 2000a.

Bar-Cohen, Y., Mavroidis C., Bouzit M., Dolgin B., Harm D., Kopchok G., andWhite R., �Virtual reality robotic operation simulations using MEMICAhaptic system,� SmartSystems 2000: The International Conference for SmartSystems and Robotics for Medicine and Space Applications, Houston, Texas,September 6 to 8, 2000b.


Bar-Cohen, Y., Mavroidis, C., Pfeiffer C., Culbert C. and Magruder D., �Hapticinterfaces,� Chapter in Automation, Miniature Robotics and Sensors for Non-Destructive Testing and Evaluation, Y. Bar-Cohen, ed., The AmericanSociety for Nondestructive Testing, Inc. (ASNT), pp. 461-468, 2000c.

Bar-Cohen, Y., Mavroidis C., Bouzit M., Pfeiffer C. and Dolgin B., �Remotemechanical mirroring using controlled stiffness and actuators (MEMICA),�Rutgers Docket Number 99-0056 A US and International PCT patentapplication was filed by Rutgers University in September 2000d.

Block, H. and Kelly, J. P., �Electro-Rheology,� Journal of Physics, D: AppliedPhysics, Vol. 21, 1988, pp. 1661.

Böse, H., Berkemeier J. and Trendler A., �Haptic system based onelectrorheological fluid,� Proceedings of the ACTUATOR 2000 Conference,19-21 June 2000, Bremen GERMANY.

Bullough, W. A., Johnson A. R., Hosseini-Sianaki A., Makin J., and FiroozianR., �Electro-rheological clutch: design, performance characterisitics andoperation,� Proceedings of the Institution of Mechanical Engineers. Part I,Journal of Systems and Control Engineering, Vol. 207, No. 2, 1993, pp. 87-95.

Burdea, G., Force and Touch Feedback for Virtual Reality, New York: JohnWiley and Sons, 1996.

Choi, S. B., �Vibration control of a flexible structure using ER dampers,�Transactions of the ASME, Journal of Dynamic Systems, Measurement andControl, Vol. 121, 1999, pp. 134-138.

Conrad, H., �Properties and design of electrorheological suspensions,� MRSBulletin, Vol. 23, No. 8, August 1998, pp. 35-42.

Coulter, J. P., Weiss K. D., and Carlson D. J., �Engineering applications ofelectrorheological materials,� in Advances in Intelligent Material Systemsand Structures-Volume 2: Advances in Electrorheological Fluids, KohudicM. A., ed., Technomic Publishing Company, Inc., Lancaster, PA, 1994, pp.64-75.

Duclos, T., Carlson, J., Chrzan, M. and Coulter, J. P., �Electrorheological fluids �materials and applications,� in Intelligent Structural Systems, Tzou andAnderson, eds., Kluwer Academic Publishers, Netherlands, 1992, pp. 213-241.

ER Fluids Developments Ltd., �Electro-rheological fluid LID 3354,� TechnicalInformation Sheet, United Kingdom, 1998.

Furusho, J., Zhang G. and Sakaguchi M., �Vibration suppression control of robotarms using a homogeneous-type electrorheological fluid,� Proceedings of the1997 IEEE International Conference on Robotics and Automation,Albuquerque, NM, 1997, pp. 3441-3448.

26 Chapter 19

Gast, A. P., and Zukoski, C. F., �Electrorheological suspensions as colloidalsuspensions,� Advances in Colloid and Interface Science, Vol. 30, 1989, pp.153.

Ghandhi, M. V., Thompson B. S., and Shakir S., �Electro-rheological fluid basedarticulating robotic systems,� in Advances in Design Automation: Vol. 2;Robotics, Mechanisms and Machine Systems, ASME DE Vol. 10-2, 1987,pp. 1-10.

Haptic Technologies, http://www.haptech.com/prod/index.htm, 1999.Immersion Corp., http://www.force-feedback.com/research/research.html, 1999.Kenaley, G. L. and Cutkosky M. R., �Electrorheological fluid-based robotic

fingers with tactile sensing,� Proceedings of the 1989 IEEE InternationalConference on Robotics and Automation, Scottsdale, AZ, pp. 132-136.

Lampe, D., Materials Database on Commercially Available Electro- andMagnetorheological Fluids (ERF and MRF), http://www.tu-dresden.de/mwilr/lampe/HAUENG.HTM, updated on 01/30/1997.

Mavroidis, C., Pfeiffer C. and Bar-Cohen Y., �Controlled compliance hapticinterface using electro-rheological fluids,� Proceedings of the 2000 SPIEConference on Electro-Active Polymer Actuators and Devices (EAPAD 2000,SPIE Proc. Vol. 3987, 2000a, pp. 300-310.

Mavroidis, C., Pfeiffer C., Celestino J. and Bar-Cohen Y., �Design and modelingof an electro-rheological fluid based haptic interface,� Proceedings of the2000 ASME Mechanisms and Robotics Conference, Baltimore MD,September 10-13, 2000b, Paper DETC2000/MECH-14121.

Mavroidis, C., Pfeiffer C., Lennon J., Paljic A., Celestino J., and Bar-Cohen Y.,�Modeling and design of an electro-rheological fluid based haptic system fortele-operation of space robots,� Proceedings of the ROBOTICS 2000Conference: The 4th International Conference and Exposition/Demonstrationon Robotics for Challenging Situations and Environments, February 27-March 2, 2000c, Albuquerque, NM, pp. 174-180.

Monkman, G. J., �Electrorheological tactile display,� Presence, MIT Press, Vol.1, No. 2, 1992.

Pfeiffer, C., Mavroidis C., Bar-Cohen Y., and Dolgin B., �Electrorheologicalfluid based force feedback device,� Proceedings of the 1999 SPIETelemanipulator and Telepresence Technologies VI Conference, SPIE Proc.Vol. 3840, 1999, pp. 88-99.

Phule, P. and Ginder J., �The materials science of field-responsive fluids,� MRSBulletin, August 1998, pp. 19-21.

Sakaguchi, M. and Furusho J., �Force display system using particle-typeelectrorheological fluids,� Proceedings of the 1998 IEEE InternationalConference on Robotics and Automation, Leuven, Belgium, May 1998a, pp.2586-2590.


Sakaguchi, M. and Furusho J., �Development of ER actuators and theirapplications to force display systems,� Proceedings of the 1998 IEEE VirtualReality Annual International Symposium (VRAIS), Atlanta, GA, 1998b, pp.66-70.

Seed, M., Hobson G. S., Tozer R. C., and Simmonds A. J., �Voltage-controlledelectrorheological brake,� Proceedings of the IASTED InternationalSymposium on Measurements, Processes and Controls, Sicily, September1986, pp. 280-284.

Sensable Technologies, http://www.sensable.com/products/desktop.htm, 1999.Sproston, J. L., Stanway R., Williams E. W. and Rigby S., �The

electrorheological automotive engine mount,� Journal of Electrostatics, Vol.32, 1994, pp. 253-259.

Stanway, R., Sproston, J. L., and El-Wahed, A. K., �Applications of electro-rheological fluids in vibration control: a survey,� Smart Materials andStructures, Vol. 5, No. 4, 1996, pp. 464-482.

Tao, R., ed., Proceedings of the Seventh International Conference on ER Fluidsand MR Suspensions, Honolulu, Hawaii, July 19-23, 1999, World ScientificPublishing Company.

Taylor, P. M., Hosseini-Sianaki A. and Varley C. J., �An electrorheologicalfluid-based tactile array for virtual environments,� Proceedings of the 1996IEEE International Conference on Robotics and Automation, Minneapolis,MN, April 1996a, pp. 18-23.

Taylor, P. M., Hosseini-Sianaki A. and Varley C. J., �Surface feedback forvirtual environment systems using electrorheological fluids,� InternationalJournal of Modern Physics B, Vol. 10, No. 23 & 24, 1996b, pp. 3011-3018.

Virtual Technologies, http://www.virtex.com/ 1999.Weiss, K. D., Carlson D. J., and Coulter J. P., �Material aspects of

electrorheological systems,� in Advances in Intelligent Material Systems andStructures-Volume 2: Advances in Electrorheological Fluids, Kohudic M.A., ed., Technomic Publishing Company, Inc., Lancaster, PA, 1994, pp. 30-52.

Winslow, W. M., �Induced fibrillation of suspensions,� Journal of AppliedPhysics, Vol. 20, 1949, pp. 1137.

Wood, D., �Editorial: tactile displays: present and future,� Displays-Technologyand Applications, Vol. 18, No. 3, 1998, pp. 125-128.

Documents

CHAPTER 19 Haptic Interfaces Using Electrorheological Fluidsservices.eng.uts.edu.au/cempe/subjects_JGZ/ems/ems... · CHAPTER 19 Haptic Interfaces Using Electrorheological Fluids