ROBOTIC ARM

Chapter 27

ROBOTIC ARM Gabor Kovacs, Patrick Lochmatter, Michael Wissler, Claudio Iseli and Lukas Kessler EMPA, Swiss Federal Laboratories for Material Testing and Research, Dübendorf, Switzerland

Abstract An arm wrestling robot was developed at the Swiss Federal Laboratories for Materials Testing and Research (EMPA) in order to demonstrate the potential of the dielectric elastomer (DE) actuator technology in a humanoid robotic system, where the unique properties of this material system come to a significant design parameter. The EMPA was one of the three participating organizations in the first arm wrestling match of electroactive polymer (EAP) robotic arm against human, which was held during the EAP-in-action session of the EAPAD conference in San Diego, 7 March 2005. Acrylic based elastomer was used as the dielectric layer in the DE actuators, due to its excellent mechanical and electrical properties. More than 250 rolled actuators were arranged in two groups according to the human protagonist– antagonist operating principle in order to achieve an arm-like rotation movement in both directions: The active movement from the horizontal arm position to the upright starting position and the reverse ‘arm wrestling motion’ exhibiting maximal force. The robot was powered by a computer-controlled high voltage device. The rotary motion of the arm was executed by electrically activating respectively deactivating the corresponding actuator group. The focus of this chapter is on the main development and manufacturing technique of the rolled DE actuators and on the engineering procedure for the robot design. The design of the DE actuators to produce a work against an external force and to display a working loop by an adequate arrangement of a pre-stressed actuator system is of primary interest. Keywords: Artificial muscles, dielectric elastomer, electroactive polymers, robotics.

27.1

INTRODUCTION

In the field of robotics most actuators are supposed to perform mechanical work against external loads for displacement of elements and in some cases, for structural shape-changing. The principle of operation of conventional actuators such as electric or piezoelectric motors, hydraulic or pneumatic cylinders is based on the active displacement of two stiff bodies in opposite directions (e.g. stator and rotor, cylinder and piston, etc.). These actuators can be up-scaled to exert large activation loads and can be designed to execute ‘unlimited’ displacements. Most conventional actuators and their supplies, however, make noise when activated and consist of complex, heavy and often expensive mechanical structures. With actuators based on active materials on the other hand soft, lightweight structures with low-noise activation capability could be implemented. The major drawbacks of today’s active materials are their limited active deformation potential as well as the elongation-dependent active forces. Nevertheless, for many robotic systems used in human environment (e.g. prosthetic limbs) low-noise and soft actuators are preferred when taking into account the acceptance of the user. For many applications within this category actuators with muscle-like properties are required. So far, however, solely actuators made from electroactive polymers (EAP) exhibit an active performance which corresponds to natural muscles. To demonstrate the potential of the EAP actuator technology in a humanoid robotic system, where the unique properties of this material system comes to a significant design parameter, the first arm wrestling match (AWC) between a human arm and a robotic arm driven by EAP was held at the EAPAD conference in 2005. Certain rules were specified to ensure the impartiality of the competition. As the most important requirements, the robotic arm had to be actuated by an EAP material and should emulate the wrestling action of a human arm. Aside from this, the shape and dimensions of the robotic

Ch27-I047488.indd 279

12/7/2007 3:10:31 PM

280

Chapter 27

arm should match the arm of an average human adult. In addition, the robot is required to execute a reversible motion: the arm had to be capable to return to its starting position after wrestling. The AWC robot developed at the Empa was driven by a system of rolled dielectric elastomer (DE) actuators. Acrylic based elastomer was used as the dielectric layer in the DE actuators, due to its excellent mechanical and electrical properties. By stacking many actuator layers (rolled up structure), the generated activation force was multiplied. In addition, the passive mechanical stiffness of the stack was increased. When several actuators are combined in series, the activation displacement becomes larger. According to (1), the electrode pressure and thus the resulting active strain will increase for pre-stretched films under application of a persistent voltage. In addition, detailed investigations of the used acrylic films VHB 4910 (3M) have shown that pre-stretching increases the dielectric breakdown strength [10] of the film. As a consequence, higher electric fields can be applied to the DE actuator when pre-stretching the film. More than 250 rolled actuators were arranged in two groups according to the human agonist–antagonist muscle configuration in order to achieve an arm-like bi-directional rotation movement. The rotary motion of the arm was activated and deactivated electrically by corresponding actuator groups. The main development process of the robot is presented in this chapter where the design of the DE actuators to produce a work against an external force and to display a working loop by an adequate arrangement of a pre-stressed actuator system is of primary interest. Although the robot has lost the arm wrestling contest against the human opponent, the DE actuators have demonstrated a very promising performance as artificial muscles.

27.2

ROLLED DE ACTUATORS

27.2.1 Actuator design Outstanding actuator performances can be obtained when the acrylic film VHB 4910 is highly mechanically pre-stretched. Typically, the pre-stretching of DE films is characterized by the pre-stretch ratios (ij ) in directions j  x, y, z. The pre-stretch ratio is defined as the ratio of the pre-stretched film length (i) (o) in state (i) to its original length in state (o) ((i) j  Lj /Lj ). However, a support structure is necessary for maintaining the required pre-stretch in the dielectric film. When a high voltage is applied to the planar dielectric film, it expands bi-directionally in the plane. To emulate a natural muscle however, an expansion in only one direction is desired. For obtaining such deformation and in order to gain an electromechanical work output, the support structure must have at least one degree of freedom in which the actuator can deform under activation and thus perform work against an external load. A number of actuator configurations, such as the extender (planar), bimorph, tube, diaphragm, spider and bow-tie actuator, have been designed to support the reaction forces resulting from the pre-stretching of the DE film [11, 12]. Nevertheless, the spring roll actuator [13] displays the most promising design for achieving uni-directional large activation forces and elongations. This actuator is comprised of a bi-directionally pre-stretched and double-side coated dielectric film wrapped around an elastic coil spring (Figs. 27.1 and 27.2). The interface for external fixation of the actuator is made from two threaded rods, which are screwed from both sides into the coil spring. In order to transmit the forces from the dielectric film to the spring, the rolled film is glued to the threaded rods. This actuator elongates in the axial direction when a voltage is applied and contracts back to its original length when deactivated. With rolled actuators, elongations of up to 26% and forces of up to 15 N were achieved [13]. For obtaining the needed wrestling power, rolled actuators consisting of acrylic elastomer VHB 4910 (3M) with elongations of up to 35% were used to drive the arm robot of Empa. Rolled aluminium leaf

Shrinkdown plastic tubing

Spring core

Uncoated dielectricum

Setscrew

Electrode

Figure 27.1 Axial section of the rolled actuator used for the wrestling robot.

Ch27-I047488.indd 280

12/7/2007 3:10:31 PM

Robotic Arm

27.2.2

281

Principle of operation

During the arm wrestling match, the robot must perform mechanical work against the arm of the opponent. Therefore, the electroactive actuators driving the robot must have the capability to exhibit mechanical work for the chosen activation loop. In order to explain the electromechanical principle of operation of the rolled actuators, the force–displacement characteristics of its interacting components namely the coil spring and the deactivated/activated DE film, are plotted qualitatively in Fig. 27.3. The following four states can be identified (Fig. 27.3) from the activation/deactivation of the actuator under its set boundary conditions: (i) Deactivated equilibrium: In the axial direction, the DE film is pre-stretched by the compressed coil spring. In radial direction, the circumferential loads of the pre-stretched DE film are supported by the spring core. (ii) Activated under blocked-strain: Starting from the deactivated force equilibrium state (i), the film relaxes and releases the compressed spring under electrical activation. In order to maintain the length of the actuator, an external compressive force Fcom is required to prevent the expansion of the spring. Coil spring core ϕ z r Pre-stretched DE film Electrodes (a)

(b)

Figure 27.2 (a) Schematic depiction of the rolled actuator and (b) different multilayer rolled actuators made for optimization tests. Force

Coil spring

Deactivated DE film (U  0)

(iv) (i)

Activated DE film (U  0)

(ii) (iii) Displacement Fully compressed coil spring

Free coil spring Free DE film (i) Deactivated equilibrium (U  0)

Fcom

(ii) Activated under blocked-strain (U  0) (iii) Activated under free-strain (U  0)

Ften

(iv) Deactivated under blocked-strain (U  0)

Figure 27.3 Operating states of the DE spring roll actuator.

Ch27-I047488.indd 281

12/7/2007 3:10:31 PM

Chapter 27 Stress difference under activation Δσz (MPa)

282 0.20

Pre-stretch Ratios (λx  λy) 0.15

3.5  3.5 5.0  3.5 3.0  6.5

0.10

0.05

0.00 0

20

40 60 80 Electrical field E (MV/m)

100

Figure 27.4 Axial stress difference in the rolled DE actuator under activation for different pre-stretch ratios x (axial direction) and y (circumferential direction) of the wrapped film.

(iii) Activated under free-strain: Starting from the deactivated force equilibrium state (i), the film relaxes under electrical activation and thus the spring expands until the loads from the spring and the DE film are balanced in a new axial force equilibrium. (iv) Deactivated under blocked-strain: When fixing the length of the actuator in the activated state (iii) and subsequently deactivating the element, an external tensile blocking force Ften is required to prevent the actuator from contracting. High activation performances can be achieved when the two pre-stretch ratios x and y are optimized. Isometric tests with DE spring roll actuators have shown that the best activation stresses are achieved when the pre-stretch ratios are x  3 and y  6.5 (Fig. 27.4). The length of the actuator plays a key role when the absolute displacement under activation is of primary interest. Based on the potential maximum strain of the material, the maximally achievable displacement can be adjusted by the initial length of the actuator. In regard to the actuator’s length, no relevant restrictions with respect to the actuation performance could be found, apart from some specific manufacturing difficulties.

27.2.3

Number of layers

The bi-directionally pre-stretched film which is wrapped around the coil spring core induces an initial compression of the inner film layers by the outer ones. The resulting radial pressure on the inner layers grows for an increasing number of wrappings as well as for stronger circumferentially pre-stretched films. Therefore, the internal stress conditions of rolled actuators have a major impact on the design of the actuators. The following brief computational analysis of the internal stress–strain conditions quantitatively demonstrates the influence of the number of layers. The compressive force Fcom needed to block the elongation of a DE spring roll actuator in th e deactivated equilibrium state (state (U  0) in Fig. 27.3) is estimated as a function of the activation voltage U (Fig. 27.5) . Due to the design of the DE spring roll actuator, the film is initially pre-stretched by planar stresses x(i) and y(i) to planar pre-stretch ratios x(i) and y(i) (Fig. 27.6(a)). The inner film layers in a DE spring roll actuator are strongly compressed in the radial direction by the outer layers. This leads to an increase of the hydrostatic pressure in the inner film layers. The resulting axial stress state of the film is a superposition of the initial tensile stresses from the pre-stretching and the pressure stresses when wrapping the film around the core. According to the radial component of (1), the hydrostatic pressure for the film layers is given by: r:

Ch27-I047488.indd 282

pn(U 0 )   (zi )

w (i ) ⏐ + pr(U, n0 )  z

(27.1)

12/7/2007 3:10:32 PM

Robotic Arm

283

Supporting Load of the core

Load to axially prestretched the film

Load to circumferentially pre-stretched film

Load to compress the spring

Figure 27.5 Passive equilibrium of axial/radial/circumferential loads in the spring roll DE actuator.

Segment of the pre-stretched film

Segment of the wrapped film Deactivated (U  0)

Activated (U  0)

(U  0) Pr,n

σ(i ) y

σ(i )

 0) σ(U ϕ,n

σ(i ) y

z

x

 0) σ(U  0) σ(U ϕ,n ϕ,n

r

 0) σ(U z,n

y

x

(U  0) Pr,n pe

ϕ

 0) σ(U ϕ,n

r

 0) σ(U z,n

ϕ

z

z Layer n

(a)

(b)

Figure 27.6 Theoretical considerations for the modelling of the DE spring roll actuator.

Introducing this expression for the hydrostatic pressure of layer n into the axial (z) and circumferential () components of (1), one obtains: :

w (i ) (i ) w (i ) (U, n0 )   (yi ) ⏐  z ⏐  pr(U, n0 ) :  A  pr(U, n0 )  z  y    

z:

(zU, n0 )

const .

w (i ) (i ) w (i ) ⏐  z ⏐  pr(U, n0 ) :  B  pr(U, n0 )   x  z     (xi )

(27.2)

const .

Obviously, the terms for the strain energy derivations are equal for all layers and are thus defined as constants A and B. In the Yeoh form, the strain energy w depends on the first invariant of the so-called left Cauchy–Green deformation tensor I1: w  C10  C20 ( I1  3)  C30 ( I1  3)2

(27.3)

Thereby, C10, C20 and C30 are material parameters and the first invariant is given by I1   2x   2y   2z

(27.4)

With these equations, one can determine the constants A and B for an experimentally determined set of material parameters for VHB 4910 (C10  7.0  10–2 MPa, C20  –9.0  10–4 MPa and C30  1.7  (U0) (U0) 10–5 MPa) and pre-stretch ratios of the wrapped film ((i)  3, (i)  6.5 and (i) x  z y   z  0) (U0)  1/19.5). By subsequent evaluation of Eq. (27.11), the radial pressure p and the axial (U r r,n (U0) in each film layer n  1, 2, … , N of the deactivated (U  0) and activated (U  0) DE stresses z,n  rolled spring actuator can be calculated.

Ch27-I047488.indd 283

12/7/2007 3:10:33 PM

Chapter 27 2

12

(U0) N  500

6 4

N  200

2

N  20

0 0

100

(a)

200

300

400

Layer number n ()

500

Tensile stress

0

Pressure stress

2

Axial stress σz,n

Radial pressure Pr,n

8

N  20

(MPa)

Deactivated Activated

10

(U0)

(MPa)

284

N  200

4 6

N  500

8

10 (b)

Deactivated Activated

0

100

200 300 400 Layer number n ()

500

Figure 27.7 (a) Distribution of mechanical compressive radial pressure and (b) acting in axial direction on the layers in rolled DE actuators consisting of N film layers.

Figure 27.7 depicts the inner film layers in a DE spring roll actuator with large diameter (several hundreds of film layers) which are strongly compressed by the radial pressure of the outer layers. Under electrical activation, the electrostatic electrode pressure is superimposed on this radial pressure distribution. The dashed curves in Fig. 27.7 represent a blocked-strain activation with U  3.8 kV (electric field of 74.1 MV/m in the dielectric film), which corresponds to an electrode pressure of 0.23 MPa (r  4.7), according to (1). As shown in Fig. 27.7, this compression in the radial direction leads to a reduction of the axial tensile stresses in the inner film layers. The same effect can be observed in the circumferential tensile stresses. When the number of layers N reaches a critical value, the innermost film is fully relaxed in the axial direction (Fig. 27.7(b)). When applying even more DE film layers to the core, the inner layers are no longer under tension but they exert pressure stresses in the axial direction. For certain numbers of film layers and pre-stretch ratios of the film, the axial contraction effect of the outer film layers may be balanced by the axial expansion of the inner layers. In this case, the core would not need to axially introduce any load, but simply has to circumferentially support the wrapped film. Generally, the electrode pressure will be uniformly superimposed to the axial stress distribution across the layers when activating the DE spring roll configuration. Thus under blocked-strain activation, the tensile stresses of all film layers in axial direction will reduce (dashed lines in Fig. 27.7(b)) and an external blocking force is needed to prevent the axial elongation of the actuator. According to this theoretical considerations, DE spring roll actuators with only few film wrappings are especially promising since only weak squeezing of the inner film layers arises with such actuators. As a result, the ‘thick’ DE spring roll actuator with many film layers is not practically useful. Thus, for the activation of the arm wrestling robot, many ‘thin’ DE spring roll actuators switched in parallel were used in order to reach large forces.

27.3 ARM WRESTLING ROBOT For the arm wrestling robot, the uni-directional deformation potential of the rolled actuators had to be transformed into a wrestling motion. In the following sections, the concept and development of the arm wrestling robot and the implementation of the rolled actuators are addressed.

27.3.1

Design

To accomplish the basic rules for the arm wrestling competition, a motion study of the human upper torso was performed. The translatory movement of the upper part of the body is mainly accomplished by the abdominal musculature (Obliquus externus/internus and Rectus abdominis). The upper right arm

Ch27-I047488.indd 284

12/7/2007 3:10:33 PM

Robotic Arm Table 27.1

285

Specification for the arm wrestling robot.

Design Criteria

Design Parameter

Design and principle of operation Orbit of the robot hand Fixation Rotation angle for the robot arm Position of the rotation axis Material for the robot structure Overall mass of the robot Robustness Bi-directional driving unit Uni-directional force for wrestling motion

As human-like as possible (functionality, reversibility) Circular path perpendicular to the table Clamping at the table 90° from the initial position Elbow (upon the table surface) Polymer or composite-based Maximally 15–20 kg Stiff design Electroactive polymers (dielectric elastomer actuators) About 200 N (measured from a female subject)

Figure 27.8 The structure of the arm wrestling robot is made of carbon-fibre composite and contains all actuators to execute the wrestling motion around the rotational axis upon the table. The arm of the wrestling robot is represented by a stiff CFC rod.

of the wrestler has a static force transmission function and does not essentially contribute to the wrestling action. In order to compensate the resulting torsion moment of the wrestling action, the left arm of the wrestler holds on the handle bar of the table. As a result, all activated forces for the arm wrestling motion sequence are based on the superposition of a torsional moment of the upper arm and a lateral displacement force of the upper torso. The arm wrestling motion mainly consists of a single rotary motion of the upper human body, while holding the arm stiff. This observation allowed a considerably simplified functional design for the arm wrestling robot. Thus, the essential motions of the arm wrestler can be well simulated with a simple rotating box equipped with a stiff arm, which is driven by the DE actuators. The box, which carries the artificial muscles, is clamped to the table at the rotating axis. Based on the set of rules and on the observations mentioned above, the Empa robot design specifications are summarized in Table 27.1. Since all actuators are intended to be placed inside the robot body, the available space would be about the size of the upper torso of a human. Despite this limitation, the actuators have to be arranged in such that the needed wrestling force and motion can be produced. This limiting factor determines essentially the design and the arrangement of the driving actuators, which is discussed in detail in the next section. To achieve the needed stiffness and strength of the lightweight robot body, carbon-fibre reinforced composites are used. Since high voltage devices are used, the black CFR box is enwrapped with glassfibre composite material as the insulating layer for safety reasons (Fig. 27.8). The torsion shaft of the robot is provided to fix the robot to the wrestling table and is designed to hold the robots weight and the arising torsion moment. The arm of the robot is made by a simple CFR rod, which is attached to the robot body.

Ch27-I047488.indd 285

12/7/2007 3:10:34 PM

286

Chapter 27 1 0 Wrestling arm

Force Fwrestling

2

Deactivated Actuator Bloc I Activated Deactivated Actuator Bloc II Activated

(a)

Actuator Bloc I

Actuator Bloc II

(b)

2

0

1

Arm position

Figure 27.9 Working principle of the actuator banks and the robot arm.

27.3.2

Principle of operation

In general, a human-like agonist–antagonist operating principle enables the reversible rotational movement of the robot arm (Fig. 27.9). When assembling the robot, the pre-stretch force in the two actuator groups Bloc I and Bloc II are in equilibrium (Fig. 27.13). In the neutral position (position 0), all actuators are deactivated and the robot arm is rotated by 35° from horizontal. By activation of the actuator Bloc I, the robot arm rotates to the upright starting position (position 1) and is thus ready for the wrestling action. During the wrestling match, Bloc I is deactivated and Bloc II is activated simultaneously. Thus, the robot exhibits its maximal wrestling force Fwrestling at position 1. The resulting motion depends on the reaction force of the human opponent. If the opponent produces less than 200 N, the robot arm performs the intended motion from the starting position to winning position 2.

27.3.3 Actuator system The structure of the actuator system essentially depends on the active force–elongation behaviour of the actuators and the wrestling opponent. On the one hand, the length and the number of the actuators are limited by the size of the robot box. On the other hand, the force–elongation characteristic of the actuator must satisfy the requirements based on the estimated wrestling force.

27.3.4

Length of the actuators

According to the size of the natural human body, the maximum length of the pre-stretched and activated actuator must not exceed 400 mm. With an actuator, an elongation of maximally 35% is expected and the length of the rigid end pieces including the length of the unstressed actuators must not exceed 250 mm.

27.3.5

Length of the wrapped dielectric film

The wrestling force of the opponent hand is assumed to be constant during the entire wrestling motion. In contrast to this trend, the activation force of the spring roll actuators decreases when they are elongated. Thus, the arm wrestling force of the robot would continuously slim down when rotating towards winning position 2 (Fig. 27.9). For compensating the decreasing actuator force, we implemented a variable force transmission between the wrestling hand and actuator system. The transmission ratio was determined by the ratio of the actuator force and the force of the wrestler. For the aimed maximal elongation of 80 mm, which corresponds to 35% strain, a maximal transmission ratio of 12 is calculated and must be taken into account. Thus, based on the chosen robot design, a maximal force difference of 12  200 N  2400 N between the agonistic and antagonistic actuator group had to be achieved. Assuming a maximal activation stress of 0.2 MPa for DE films, the needed material for the cross-sectional area for the DE spring roll actuators would amount to at least 1.25  10–2 m2.

Ch27-I047488.indd 286

12/7/2007 3:10:34 PM

Robotic Arm

287

Le Cro ng ss th -se of c wr tion ap al pe ar e d film a

t

Utilization of cross-sectional area (%)

Figure 27.10 Cross-sectional area and length of the wrapped DE film.

100

Axis

80 60 Axis 40 20 0

(a)

0

10

20

30

40

50

60

Outer actuator radius (mm)

70 (b)

Figure 27.11 Utilization of the cross-sectional area of the robot box as a function of the actuator radius (diagram in the centre): (a) arranging many small actuators and (b) arranging a few large actuators.

Regarding an average thickness t of 50 m for the pre-stretched film, this would lead to a considerable total length for the wrapped DE film of 250 m.

27.3.6 Actuator arrangement Next, the number of required actuators and the corresponding diameter for each rolled actuator was determined in order to take the total required length of wrapped dielectric film. Based on the results of the theoretical considerations and on experiments with DE spring roll actuators, the number of wrappings per actuator must be as low as possible to obtain the best specific actuator performance. In addition, the robot’s reliability with many ‘slim’ actuators excels the variant with only few but ‘thick’ actuators (Fig. 27.10). Therefore, the final solution for the actuator system consists of banks equipped with many ‘slim’ DE spring roll actuators. By filling the robot box with DE spring roll actuators, the best utilization of the available space is achieved for ‘slim’ actuators with outer radius of approximately 12 mm (Fig. 27.11). As a result, up to 256 DE spring roll actuators are arranged in two agonist–antagonist groups (Figs. 27.12 and 27.13). Around each actuator, a core of 2 m of pre-stretched acrylic film is wrapped, which results in 35 film wrappings. For practical reasons, both the agonist and the antagonist actuator groups are subdivided into two actuator banks, each equipped with up to 64 actuators. The plates of the banks were dimensioned to distribute the external load to each actuator. The upper plates of the banks were fixated via long threaded rods at the top end of the box. Thus, an adjustment of the initial axial pre-stretch of the actuators was possible. At the lower plate, a chain connected the agonist and antagonist actuator blocs via the axis of rotation. The force transmission was implemented by a specially shaped cam disc, which was interconnected between the actuator banks and the fixing axle.

Ch27-I047488.indd 287

12/7/2007 3:10:34 PM

288

Chapter 27

Figure 27.12 Force transmission mechanism to achieve the required force characteristic from the arm wrestling robot: working principle on upper three pictures and implemented device on lower picture. Top plate of robot box Threaded rod Upper plate Bank of DE roll actuators Lower plate

Figure 27.13 The four actuator banks, each equipped with 64 actuators, placed inside of the arm wrestling robot. The force distribution plates mounted on both ends of the actuator banks also served for the high voltage application.

Since a large number of actuators are used in the robot, the possibility of an electrical breakdown of some actuators must be considered when designing the electrical system. To supply all actuators of one actuator bank with the needed high voltage, they are connected in parallel. However, if one of the actuators electrically break down, all actuators in one bank will instantaneously discharge through the failed element. On one hand, this would lead to a considerable loss of arm wrestling force. On the other hand, an inflammation of the failed actuator caused by the concentrated heat dissipation from the high discharging current might occur. Both effects can be prevented by applying current limiters (fuses) in front of the DE spring roll actuators. However, due to the limited current, the charging time for the actuators of the arm wrestling robot increases distinctively.

27.4

CONCLUSIONS

For the most applications in the field of robotic or prosthetic limbs, the agonist–antagonist actuator arrangement is preferable since it can drive the system in a two way mode. By this robot, it has been shown that it is possible to work against an external force and to display a working loop by an adequate arrangement of a pre-stressed actuator system. Based on the general properties of DE actuators, the preferential fields of applications can be located where active structures with large and complex deformations are needed but only small driving forces are required. Such applications could be found in, for example, humanoid systems (artificial hands or mimic faces), where many small actuators could be

Ch27-I047488.indd 288

12/7/2007 3:10:35 PM

Robotic Arm

289

applied and individually addressed. Furthermore, the actuators of Empa robot have impressively highlighted the potential of the DE, where the unique properties of this material system, as they are noise free when actuated and very lightweight, come to a significant design parameter. Although the Empa robot arm lost the competition, it was the only robot in the competition that could actively return to its upright starting position after the match, as required by the rules of the competition. This bi-directional motion of the robot arm became possible due to the agonist–antagonist actuator configuration. Nevertheless, the scientific experiences during the development process of the arm wrestling robot has pointed out the challenges, which need to be addressed for improving especially the rolled DE actuator configuration.

ACKNOWLEDGEMENTS We would like to thank Mr Claudio Iseli and Mr Lukas Kessler from the Swiss Federal Laboratories for Materials Testing and Research (EMPA) for their contributions to the present work.

References [1] [2]

[3] [4] [5] [6] [7]

[8] [9] [10] [11] [12]

[13] [14]

[15]

Ashley, S. (2003). Artificial muscles. Sci. Am., 289, 52–59. Pelrine, R., Kornbluh, R., Pei, Q., Standford, S., Oh, S. and Eckerle, J. (2002). Dielectric elastomer artificial muscle actuators: toward biomimetic motion. Smart Structures and Materials: Electroactive Polymer Actuators and Devices, San Diego, CA, Proc. SPIE, 4695, 126–137. Bar-Cohen, Y. (2001). Electroactive Polymer (EAP) Actuators as Artificial Muscles – Reality, Potential and Challenges, Vol. 671. SPIE Press, Bellingham, WA. Pelrine, R. E., Kornbluh, R. D. and Joseph, J. P. (1998). Electrostriction of polymer dielectrics with compliant electrodes as a means of actuation. Sens. Act. A Phys., 64(1), 77–85. Pelrine, R., Kornbluh, R., Pei, Q. B. and Joseph, J. (2000). High-speed electrically actuated elastomers with strain greater than 100%. Science, 287(5454), 836–839. Heydt, R., Kornbluh, R., Pelrine, R. and Mason, V. (1998). Design and performance of an electrostrictivepolymer-film acoustic actuator. J. Sound Vibr., 215(2), 297–311. Carpi, F., Chiarelli, P., Mazzoldi, A. and De Rossi, D. (2003). Electromechanical characterisation of dielectric elastomer planar actuators: comparative evaluation of different electrode materials and different counterloads. Sens. Act. A Phys., 107(1), 85–95. Lacour, S. P., Prahlad, H., Pelrine, R. and Wagner, S. (2004). Mechatronic system of dielectric elastomer actuators addressed by thin film photoconductors on plastic. Sens. Act. A Phys., 111(2–3), 288–292. Trujillo, R., Mou, J., Phelan, P. E. and Chau, D. S. (2004). Investigation of electrostrictive polymers as actuators for mesoscale devices. Int. J. Adv. Manufact. Techn., 23(3–4), 176–182. Kofod, G., Sommer-Larsen, P., Kornbluh, R. and Pelrine, R. (2003). Actuation response of polyacrylate dielectric elastomers. J. Intel. Mat. Syst. Struct., 14, 787–793. Pei, Q., Rosenthal, M., Stanford, S., Prahlad, H. and Pelrine, R. (2004). Multiple-degrees-of-freedom electroelastomer roll actuators. Smart Mat. Struct, 13(5), N86–N92. Sommer-Larsen, P., Kofod, G., Benslimane, M. and Gravesen, P. (2002). Performance of dielectric elastomer actuators and materials. Smart Structures and Materials: Electroactive Polymer Actuators and Devices, San Diego, CA, Proc. SPIE, 4695, 158–166. Pei, Q., Pelrine, R., Standford, S., Kornbluh, R. and Rosenthal, M. (2003). Electroelastomer rolls and their application for biomimetic walking robots. Synth. Met., 135–136, 129–131. Ha, S. M., Yuan, W., Pei, Q., Pelrine, M. and Stanford, S. (2006). New high-performance electroelastomer based on interpenetrating polymer networks. Smart Structures and Materials: Electroactive Polymer Actuators and Devices, San Diego, CA, Proc. SPIE, 6168, 08-1–08-12. http://armwrestling.com/rulesandregulations.html.

Ch27-I047488.indd 289

12/7/2007 3:10:36 PM