PreprintPDF Available

Stretchable Electrohydraulic Artificial Muscle for Full Motion Ranges in Musculoskeletal Antagonistic Joints

Authors:
Preprints and early-stage research may not have been peer reviewed yet.

Abstract and Figures

Artificial muscles play a crucial role in musculoskeletal robotics and prosthetics to approximate the force-generating functionality of biological muscle. However, current artificial muscle systems are typically limited to either contraction or extension, not both. This limitation hinders the development of fully functional artificial musculoskeletal systems. We address this challenge by introducing an artificial antagonistic muscle system capable of both contraction and extension. Our design integrates non-stretchable electrohydraulic soft actuators (HASELs) with electrostatic clutches within an antagonistic musculoskeletal framework. This configuration enables an antagonistic joint to achieve a full range of motion without displacement loss due to tendon slack. We implement a synchronization method to coordinate muscle and clutch units, ensuring smooth motion profiles and speeds. This approach facilitates seamless transitions between antagonistic muscles at operational frequencies of up to 3.2 Hz. While our prototype utilizes electrohydraulic actuators, this muscle-clutch concept is adaptable to other non-stretchable artificial muscles, such as McKibben actuators, expanding their capability for extension and full range of motion in antagonistic setups. Our design represents a significant advancement in the development of fundamental components for more functional and efficient artificial musculoskeletal systems, bringing their capabilities closer to those of their biological counterparts.
Content may be subject to copyright.
Stretchable Electrohydraulic Artificial Muscle for Full Motion Ranges
in Musculoskeletal Antagonistic Joints
Amirhossein Kazemipour1, Ronan Hinchet1, Robert K. Katzschmann1
Abstract Artificial muscles play a crucial role in muscu-
loskeletal robotics and prosthetics to approximate the force-
generating functionality of biological muscle. However, cur-
rent artificial muscle systems are typically limited to either
contraction or extension, not both. This limitation hinders
the development of fully functional artificial musculoskeletal
systems. We address this challenge by introducing an artificial
antagonistic muscle system capable of both contraction and ex-
tension. Our design integrates non-stretchable electrohydraulic
soft actuators (HASELs) with electrostatic clutches within
an antagonistic musculoskeletal framework. This configuration
enables an antagonistic joint to achieve a full range of motion
without displacement loss due to tendon slack. We implement a
synchronization method to coordinate muscle and clutch units,
ensuring smooth motion profiles and speeds. This approach
facilitates seamless transitions between antagonistic muscles at
operational frequencies of up to 3.2 Hz. While our prototype
utilizes electrohydraulic actuators, this muscle-clutch concept
is adaptable to other non-stretchable artificial muscles, such as
McKibben actuators, expanding their capability for extension
and full range of motion in antagonistic setups. Our design
represents a significant advancement in the development of
fundamental components for more functional and efficient
artificial musculoskeletal systems, bringing their capabilities
closer to those of their biological counterparts.
I. INTRODUCTION
A. Motivation
Advancements in robotics increasingly draw inspiration from
the versatility and adaptability of natural organisms. Hybrid
rigid-soft robots, which integrate soft materials into rigid
structures [1]–[4], aim to replicate the seamless movements
and dexterity found in nature. Soft elements provide com-
pliance at critical points, such as contact interfaces [5], and
soft muscles attached via tendons to a rigid skeleton generate
motion around passive joints. This design paradigm offers
significant advantages over conventional rigid robots [6]–
[8], including reduced power requirements for equivalent
torque, enhanced joint accuracy, and a more streamlined form
factor [9]. By decoupling actuators from the load-bearing
structure and joints, musculoskeletal robots also achieve
greater payload capacities compared to entirely soft robots.
The performance of musculoskeletal robots hinges on both
skeletal design and the efficacy of their artificial muscles.
Revolute joints, or hinges, are the most prevalent in articu-
lated robots, providing a uniaxial degree of freedom as seen
in human finger and toe joints. More complex joints like
knees and elbows permit slight rotations or shifts, adding
to mechanical complexity. Soft muscles, often arranged in
1Soft Robotics Lab, ETH Zurich, Switzerland
{akazemi, rhinchet, rkk}@ethz.ch
Fig. 1: Overview of the proposed artificial muscle system enabling
full motion range: (A) In antagonistic setups with non-stretchable
muscles, if the tendons are taut, the motion is blocked since
these muscles cannot elongate. (B) Adding slack permits movement
but reduces the motion range. (C) Our solution uses electrostatic
clutches and muscles to enable contraction and extension, restoring
full motion range.
antagonistic pairs akin to biological biceps and triceps,
actuate these joints. To mimic this biological functionality,
musculoskeletal robots employ both soft fluidic [10]–[13]
and electromagnetic [14]–[16] actuators.
Pneumatic Artificial Muscles (PAMs), such as McKibben
muscles, were among the earliest soft artificial muscles.
arXiv:2409.11017v1 [cs.RO] 17 Sep 2024
Comprising an inflatable bladder within a non-stretchable
shell, they contract upon inflation. PAMs are still prevalent
in complex bio-inspired robotic systems like the Shadow
Hand C3 [17], offering excellent strength-to-weight ratios.
However, they are limited by slow response times and, in
untethered systems, the cumbersome requirement for com-
pressors, tanks, and valves.
To overcome these limitations, various types of soft ar-
tificial muscles have been developed [18]. Notably, Hy-
draulically Amplified Self-healing Electrostatic (HASEL)
actuators [19] have emerged as a promising alternative by
combining soft fluidic actuation with electrostatic forces.
HASELs consist of oil-filled pouches partially covered with
electrodes. When activated, the electrodes compress and
redistribute the oil within the pouch, causing it to inflate and
contract similarly to biological muscles. Initially, HASELs
utilized silicone elastomer substrates for flexibility [19], but
challenges like pull-in instabilities compromised control.
This led to the adoption of thin thermoplastic polymer
substrates such as Biaxially Oriented Polyethylene Tereph-
thalate (BOPET) [20], offering improved durability and me-
chanical properties. Using Polyvinylidene Difluoride (PVDF)
terpolymer substrates further reduces actuation voltage by
fivefold while maintaining force and strain performance [21].
However, a significant drawback is that these thermoplastic
substrates render HASELs non-stretchable, akin to McK-
ibben muscles.
B. Problem Statement
An important design limitation of musculoskeletal robots
is the use of non-stretchable artificial muscles, such as
McKibben and HASEL actuators, in antagonistic configu-
rations (e.g., on either side of a hinge). In the human arm,
the biceps (agonist) contracts to move the arm, while the
triceps (antagonist) relaxes to allow movement. However, in
robots, non-stretchable muscles cannot relax and elongate the
antagonist, preventing the agonist’s contraction and blocking
arm motion (Fig. 1A).
One solution is to introduce slack in the antagonist’s
tendon, allowing the agonist to contract and bend the arm
(Fig. 1B). However, to straighten the arm back, the slack
must be limited to a maximum of half of the muscle con-
traction amplitude, restricting effective muscle contraction
to 50%. This significantly limits the motion range and
introduces latency, hindering the robot’s performance.
Alternatively, incorporating linear clutches in series and
synchronized with the muscles can utilize the full muscle
contraction. Clutches dynamically block or couple motion,
enabling variable stiffness systems. When put in series
with muscles, they can disengage the triceps during biceps
contraction to bend the arm and vice versa to straighten it
back, allowing 100% utilization of muscle contraction. This
approach, however, depends on clutch size, weight, holding
force, elasticity, and power efficiency.
Existing clutches [22] are mostly rigid. While pneumatic
and vacuum jamming clutches can be soft, they require bulky
pumps and valves. Electromagnetic clutches, though popular
and efficient, are heavy and power-consuming. Mechanical
latches reduce power usage but are slow and complex. Mag-
netorheological fluid designs [23] are simpler but heavier,
and Piezoelectric clutches [24] are lighter and more efficient
yet rigid and complex. These clutches are unsuitable for
musculoskeletal robots due to their lack of compliance, high
mass, and complexity.
The recent development of electrostatic clutches (ES-
clutches) offers superior force-to-mass (<104) and force-
to-power ratio (<105) [25]. ESclutches are fast, thin (sub-
millimeter), light, compact, and flexible [26]. They consume
minimal current, and their holding force is adjustable via
driving voltage [27]. They have been integrated into tex-
tile [26], [28] in wearable robotics for haptic feedback [27]
and exoskeletons [29]. ESclutches are well-suited to com-
plement HASELs in musculoskeletal robots, but such a
combination has never been evaluated to actuate antagonistic
muscle configurations.
C. Contributions
In this paper, we introduce a novel contractile artificial mus-
cle system capable of stretching by: (1) combining HASEL
actuators with ESclutches in series; (2) integrating this hybrid
system into a musculoskeletal structure with antagonistic
muscles; and (3) demonstrating the synergistic operation of
HASELs and ESclutches to enhance limb motion (Fig. 1C).
The combination of HASEL actuators and ESclutches
is advantageous as both are thin, flexible, and based on
electrostatic, enabling fast and efficient actuation using high
voltages and low currents for safety. This facilitates seamless
integration and control, with driving circuits that can be
combined and miniaturized to a few cubic centimeters [21].
This work presents the first evaluation and demonstration
of such a hybrid actuation system in a musculoskeletal
framework.
II. METHODOLOGY
This section details our solution, covering theoretical anal-
ysis, design and characterization of the clutch and HASEL
actuator, their integration into a unified muscle capable of
both extension and contraction, incorporation into an antag-
onistic joint musculoskeletal system, and control methods for
smooth operation.
A. Theoretical design analysis
We consider a robotic system articulated with a pin joint
and actuated by a pair of HASEL muscle packs in an
antagonistic configuration. The HASEL packs are connected
to the joint via tendons that have slack equal to 50% of the
HASEL strain σat the operating force Fhand voltage. In
this configuration, integrating a stretchable ESclutch in series
with the HASEL muscle packs eliminates the need for tendon
slack. This integration offers two primary options: either a
50% reduction in HASEL length for the same motion or a
doubling of the usable HASEL strain, thereby increasing the
limb’s range of motion.
We consider a HASEL pack with length Lhand width
Whconnected in series with an ESclutch of the same width
Wc=Wh, featuring a surface friction force density Pcat
a given operating voltage. The clutch length Lcrequired to
transmit the HASEL force to the limb is:
Lc=Fh/(PcWh)(1)
If we opt to decrease the HASEL length, the new shorter
length L
hbecomes:
L
h= 0.5×Lh(2)
The ESclutch is packaged with a soft elastic textile sleeve,
adding an elastic component when not actuated. This creates
resistance when the opposite HASEL contracts. Assuming a
textile thickness Ttand an elastic modulus Etthat remains
linear and constant under small deformations, the maximum
resistance force Ftof the opposite clutch is:
Ft=σ(L
h/Lc)×EtWhTt(3)
To compensate for this resistance, the HASEL must be
slightly stronger. A simple approach is to increase its width.
The new HASEL width W
his:
W
h=Wh+FtWh/Fh(4)
Alternatively, if we maintain the original HASEL dimensions
to increase the range of motion, it must exert more force
to compensate for the elastic resistance of the antagonistic
clutch. According to the HASEL stress-strain curve, an
increase in force will reduce the strain. Consequently, the
usable strain will be slightly less than doubled.
A limitation of ESclutches is their theoretical maximum
strain of 100%, which imposes a minimum clutch length and
width Wc:
LcσL
h(5)
Wc=Fh/(PcLc)(6)
Additionally, practical considerations such as attachments
and packaging require extra space, which can impact per-
formance.
B. Electrostatic Clutch Characterization
ESclutches were designed and fabricated following the pro-
cess described in [27]. Each ESclutch consists of two 5 cm
by 5 cm electrodes that can slide on each other (Fig. 2A).
They are composed of 125 µmthick PET films with 50 nm
thick Al electrodes on top. The electrodes are separated by
a6µmthick PVDF terpolymer film deposited on one of
them (Fig. 2B). Electric wires are connected at extremities,
and holes are drilled for attachment. Finally, the ESclutch is
placed between 2 pieces of soft stretchable textile, having
Young’s modulus of around 0.1 MPa [27], for protection
and to ensure return to its original position upon release.
The resulting clutch is 6 cm wide and 9 cm long including
casing and is 1 mm thick in total for a weight of 4.4 g. Inside,
the electrodes overlap is 15 cm2in its released state, and the
clutch can stretch up to 44% or 4 cm.
ESclutches were characterized by applying a symmetric
square voltage at 10 Hz between the electrodes and then
pulling on the attached clutch using a force sensor until the
electrodes slide. This gave us the maximum holding force
Fig. 2: Overview of the ESclutch design and force range. (A)
Structure of the ESclutch describing its different layers and (B)
photo of the ESclutch showing its thin and compact format. (C)
Characterization of the maximum holding force of the ESclutch as a
function of the voltage applied up to 150 V with a 10 Hz AC square
signal, which shows that it can sustain 8.41 kg. (D) Illustration of
the ESclutch behavior under actuation.
of the clutch depending on the applied voltage (Fig. 2C).
It shows that our clutch can hold 4.25 kg when actuated
at 100 V (Fig. 2D) which is equivalent to a shear stress
of 2.8 N cm2. But the clutch can easily block 8.41 kg at
150 V if needed for a bigger robot. This corresponds to a
shear stress of 5.6 N cm2, which is on the same order of
magnitude as [26]. Such a clutch can usually lock in 5 ms
and release in 15 ms.
C. HASEL Characterization
Thermoplastic-based HASEL actuators consist of a non-
stretchable yet flexible shell encapsulating a liquid dielectric,
flanked by electrodes (Fig.3A). When voltage is applied,
the electrodes attract, squeezing the liquid and redistributing
it within the actuator (Fig.3B). As the pouch deforms, the
actuator contracts.
HASELs were fabricated following the process in [30].
Each pouch measures 4.5 cm wide and 2 cm long, with
50% electrode coverage. The shell is a 15 µmthick heat-
sealable Mylar sheet. Black carbon inks serve as electrodes,
with silicone oil as the dielectric fluid. The HASEL muscle
comprises 8 pouches (Fig. 3C), totaling 16 cm in length and
weighing 13.7 g.
Fig. 3: Overview of the HASEL actuator design and force range. (A)
Structure of one HASEL pouch describing its constitutive layers and
(B) zoom-in photo of HASELs at the relaxed and contracted state,
which show the bulging and contraction of the pouch when actuated
at 8 kV with a DC square signal. (C) Photo of one HASEL pack
of 8 pouches in series. (D) The force-displacement characterization
curves of the HASEL depend on the applied voltage and show that
the actuator can generate a force up to 16.3 N or a displacement
up to 18.0 mm when actuated at 8 kV .
HASEL actuators exhibit nonlinear force-displacement
curves (Fig. 3D), obtained by anchoring one end to a load
cell, applying force to the other, and tracking motion with
a laser sensor. HASELs show higher displacement at lower
loads and vice versa. This characteristic is crucial for robotic
system design, requiring consideration of the specific force-
strain curve to meet load and displacement targets. HASEL
performance can be customized by adjusting geometry, struc-
ture, and materials.
D. System Integration
An articulated limb was constructed by connecting a carbon
fiber tube with a 3D-printed PLA joint and ball bearings to
a12 cm-long tube weighing 3.8 g. On the limb, electrostatic
HASEL-clutch units were connected in series and arranged
antagonistically (Fig. 4). Fishing line tendons were attached
to the ends of the electrostatic HASEL-clutch units and were
taut when the limb was in the middle position.
Fig. 4: Musculoskeletal robotic joint actuated with the HASEL-
clutch units. When the bottom HASEL-clutch unit is in the ’Off-
Off’ state, it can elongate, and by setting the top HASEL-clutch
unit to the ’On-On’ state, the limb moves upward. For downward
motion, the operation is reversed.
For design optimization, we considered the fabricated
HASELs of 4.5 cm wide and 16 cm long generating a force
of 1 N at a strain of 8% at 8 kV; and the fabricated clutches
generating a friction force density of 5.5 N cm2at 150 V
and packaged with 1 mm thick stretchable textile (elasticity
modulus 100 kPa).
According to eq. (1), a clutch of 0.18 cm2or 0.4 mm in
length and 4.5 cm in width would theoretically suffice to
transmit the HASEL force. However, this size would limit
stretchability and increase resistance from the antagonistic
clutch’s elastic sleeve. Therefore, we oversized the clutch
to enhance blocking force, reduce strain, and minimize
resistance, which is crucial for high-speed switching and
minimizing leg resonance. Additionally, the packaging was
designed to maintain system modularity.
E. Control Strategy
Fig. 5: State diagram for HASEL-clutch control. Each HASEL-
clutch pair can be in one of four states: State 1 (on/off), State 2
(on/on), State 3 (off/off), and State 4 (off/on). Transitions between
states enable coordinated limb movements and braking.
Our system is actuated using four control signals: two
for HASEL actuators and two for clutches. Achieving a
full range of motion requires synchronizing these signals
through a state machine logic with a feedforward control
strategy. Clutches are driven by an AC symmetric square
wave at 10 Hz to optimize the release speed and the locking
force. For HASELs, we use ramp signals to control their
displacement, as it is proportional to the applied voltage. The
linear lever arm design ensures that HASEL displacement
directly translates to joint angles.
Synchronization between clutches and HASELs is essen-
tial for effective musculoskeletal operation. When a clutch
is ”off, it allows stretching; when ”on, it locks the limb.
Similarly, a HASEL in the ”on” state contracts, and in the
”off” state, it releases contraction. Figure 5 illustrates the
possible states of a HASEL-clutch pair.
To move the limb rightward, the left HASEL-clutch pair
is set to State 3 (off/off) and the right pair to State 2
(on/on), causing the right HASEL to contract and its clutch to
lock, thereby moving the limb right. Leftward movement is
achieved by reversing these settings. Additionally, engaging
the opposite clutch allows for rapid stopping by controlling
the braking force through the ESclutch’s voltage-dependent
friction [27]. This enables variable stiffness and impedance
in the joint, enhancing control speed and responsiveness.
III. RES ULTS AN D DISCUSSIONS
A. Experimental Setup
The setup included four high-voltage amplifiers (one Trek
20/20C, one Trek 610E, two PolyK PK-HVA10005) to power
clutches and HASEL actuators. An RLS-RM08 magnetic
encoder measured joint angles, and a NI DAQ 6343-USB
acquired data. MATLAB with data acquisition toolbox con-
trolled the system, sending commands to amplifiers and
recording voltages and joint angles via the DAQ.
B. Experimental Validation
Fig. 6: Comparison of joint range of motion achieved with (A) a
configuration using only HASEL actuators for actuation without
clutches versus (B) the integrated muscle-clutch mechanism. De-
spite identical actuation of the HASEL actuators, the muscle-clutch
system enables an increased range of motion of ±82over the
muscle-only arrangement of ±52.
Both muscle mechanisms—HASELs with and without ES-
clutches—were evaluated on the artificial limb setup (Fig. 6)
using the same voltage command (a ramp signal with a peak
of 8 kV) and frequency (2.5 Hz), as well as identical tendon
anchor points. With only HASELs and optimized tendon
slack, the limb rotated 52(Fig.6A). In contrast, combining
ESclutches with the same HASELs increased rotation to
82(Fig. 6B), resulting in approximately a 58% increase
in usable strain in an antagonistic muscle configuration.
Fig. 7: Experimental results comparing HASEL-only and HASEL-
clutch systems. Applied voltages to left/right clutches and HASELs
are shown, along with resulting joint angle transitions. The graph
demonstrates smooth motion across increasing actuation frequen-
cies, with the HASEL-clutch solution consistently achieving a
higher range of motion compared to the HASEL-only system while
increasing the actuation frequency up to 3.2 Hz.
The faster actuation speed of ESclutches compared to
HASELs facilitates their synchronization, allowing them to
be engaged simultaneously. Additionally, since their force
is proportional to the applied voltage—and the ESclutch
operates at a lower voltage than the HASEL—it is possible to
control both using a single voltage source (the high voltage
of the HASEL) with a simple voltage divider supplying the
ESclutch. This greatly simplifies the control electronics.
We tested the HASEL-clutch system by actuating it at dif-
ferent frequencies. Figure 7 displays the actuation patterns,
where we alternately activated the right and left HASELs
along with their respective ESclutches. The coordinated
voltage patterns, up to 3.2 Hz, demonstrate smooth limb
movements without abrupt discontinuities. The resulting joint
angle shows seamless transitions between states. Figure 8
presents snapshots of the joint in motion at 2.5 Hz, illustrat-
ing rapid, cyclic movements with a wide range of motion.
t = 0.40s t = 0.45s t = 0.50s t = 0.55s t = 0.60s
Fig. 8: Snapshots of the experiment where the joint is cycling at a freq. of 2.5 Hz, achieving a bidirectional range of motion of 160.
0 0.5 1 1.5 2 2.5 3 3.5
Frequency (Hz)
60
80
100
120
140
160
180
200
Max. Range of Motion (Deg)
Without Clutch
With Clutch
Fig. 9: Maximum range of motion versus actuation frequency for
the HASEL-only and clutch-muscle actuated joint. Range of motion
expands at higher frequencies due to stretchable clutch sleeves
acting as a series elastic component, allowing storage and release of
elastic energy to enhance actuation force and motion range. Smooth
transitions are observed up to 3.2 Hz.
IV. CONCLUSION AND FUTURE WORK
In this paper, we presented a novel approach to create an
artificial muscle that can perform both contraction and ex-
tension movements. Our design combines HASEL actuators
with ESclutches. We integrated these components into a
musculoskeletal system and utilized these HASEL-clutch
units to drive an antagonistic actuated joint. This design
enabled us to achieve complete functionality of the muscles,
ensuring a full range of motion without losing displacement
due to slack.
One limitation of this study is the lightness of the limb.
Investigating the impact of added load on limb motion
would be valuable. Additionally, the current vertical setup
benefits from gravity aiding the limb’s return to the neutral
position. Studying the system horizontally would provide
better insight into the role of the series elastic component in-
troduced by the clutch. Modeling the system could optimize
the clutch’s elasticity for specific resonance frequencies to
maximize amplitude at desired speeds. The released HASEL
could also assist in returning the limb, requiring more
complex synchronization.
Notably, our approach is not restricted to electrostatic
actuators alone. It can also be applied to other types of non-
stretchable muscles, such as McKibben’s muscles [31], [32]
allowing them to extend and achieve full range of motion in
an antagonistic setup. This versatility highlights the potential
broad applicability of our solution in advancing artificial
muscle technology.
V. ACK NOWLEDGMEN TS
We thank Piezotech-Arkema for providing the P(VDF-TrFE-
CTFE) polymer used to fabricate the ESclutches. This work
was supported by the SNSF Project Grant #200021 215489.
Their support was instrumental in the development of this
research.
REFERENCES
[1] E. W. Hawkes, L. H. Blumenschein, J. D. Greer, and A. M. Okamura,
“A soft robot that navigates its environment through growth, Science
Robotics, vol. 2, no. 8, Jul. 2017, publisher: Science Robotics
Section: Research Article.
[2] T. J. Wallin, J. Pikul, and R. F. Shepherd, “3D printing of soft robotic
systems,” Nature Reviews Materials, vol. 3, no. 6, pp. 84–100, Jun.
2018, number: 6 Publisher: Nature Publishing Group.
[3] M. Wehner, R. L. Truby, D. J. Fitzgerald, B. Mosadegh, G. M.
Whitesides, J. A. Lewis, and R. J. Wood, “An integrated design and
fabrication strategy for entirely soft, autonomous robots,” Nature, vol.
536, no. 7617, pp. 451–455, Aug. 2016.
[4] J. Byun, Y. Lee, J. Yoon, B. Lee, E. Oh, S. Chung, T. Lee, K.-J.
Cho, J. Kim, and Y. Hong, “Electronic skins for soft, compact,
reversible assembly of wirelessly activated fully soft robots, Science
Robotics, vol. 3, no. 18, p. eaas9020, May 2018, publisher: American
Association for the Advancement of Science.
[5] K. Kim, P. Spieler, E.-S. Lupu, A. Ramezani, and S.-J. Chung, A
bipedal walking robot that can fly, slackline, and skateboard, Science
Robotics, vol. 6, no. 59, p. eabf8136, Oct. 2021, publisher: American
Association for the Advancement of Science.
[6] M. Hutter, C. Gehring, D. Jud, A. Lauber, C. D. Bellicoso, V. Tsounis,
J. Hwangbo, K. Bodie, P. Fankhauser, M. Bloesch, R. Diethelm,
S. Bachmann, A. Melzer, and M. Hoepflinger, ANYmal - a highly mo-
bile and dynamic quadrupedal robot,” in 2016 IEEE/RSJ International
Conference on Intelligent Robots and Systems (IROS), Oct. 2016, pp.
38–44, iSSN: 2153-0866.
[7] J. Lee, J. Hwangbo, L. Wellhausen, V. Koltun, and M. Hutter,
“Learning quadrupedal locomotion over challenging terrain, Science
Robotics, vol. 5, no. 47, p. eabc5986, Oct. 2020, publisher: American
Association for the Advancement of Science.
[8] T. Miki, J. Lee, J. Hwangbo, L. Wellhausen, V. Koltun, and M. Hutter,
“Learning robust perceptive locomotion for quadrupedal robots in
the wild,” Science Robotics, vol. 7, no. 62, p. eabk2822, Jan. 2022,
publisher: American Association for the Advancement of Science.
[9] T. J. K. Buchner, T. Fukushima, A. Kazemipour, S.-D. Gravert,
M. Prairie, P. Romanescu, P. Arm, Y. Zhang, X. Wang, S. L. Zhang,
J. Walter, C. Keplinger, and R. K. Katzschmann, “Electrohydraulic
musculoskeletal robotic leg for agile, adaptive, yet energy-efficient
locomotion,” Nature Communications, vol. 15, no. 1, p. 7634, Sep.
2024.
[10] A. Hitzmann, H. Masuda, S. Ikemoto, and K. Hosoda, “Anthropo-
morphic musculoskeletal 10 degrees-of-freedom robot arm driven by
pneumatic artificial muscles,” Advanced Robotics, vol. 32, no. 15, pp.
865–878, 2018, publisher: Taylor & Francis.
[11] R. Niiyama, S. Nishikawa, and Y. Kuniyoshi, “Biomechanical ap-
proach to open-loop bipedal running with a musculoskeletal athlete
robot,” Advanced Robotics, vol. 26, no. 3-4, pp. 383–398, 2012,
publisher: Taylor & Francis.
[12] S. Ikemoto, Y. Kimoto, and K. Hosoda, “Shoulder complex linkage
mechanism for humanlike musculoskeletal robot arms,” Bioinspiration
& Biomimetics, vol. 10, no. 6, p. 066009, Nov. 2015.
[13] S. Kurumaya, K. Suzumori, H. Nabae, and S. Wakimoto,
“Musculoskeletal lower-limb robot driven by multifilament muscles,
ROBOMECH Journal, vol. 3, no. 1, p. 18, Dec. 2016.
[14] Y. Asano, T. Kozuki, S. Ookubo, M. Kawamura, S. Nakashima,
T. Katayama, I. Yanokura, T. Hirose, K. Kawaharazuka, S. Makino,
Y. Kakiuchi, K. Okada, and M. Inaba, “Human mimetic musculoskele-
tal humanoid Kengoro toward real world physically interactive ac-
tions,” in 2016 IEEE-RAS 16th International Conference on Humanoid
Robots (Humanoids), Nov. 2016, pp. 876–883, iSSN: 2164-0580.
[15] M. J ¨
antsch, S. Wittmeier, K. Dalamagkidis, A. Panos, F. Volkart,
and A. Knoll, “Anthrob - A printed anthropomimetic robot, in
2013 13th IEEE-RAS International Conference on Humanoid Robots
(Humanoids), Oct. 2013, pp. 342–347, iSSN: 2164-0580.
[16] C. Richter, S. Jentzsch, R. Hostettler, J. A. Garrido, E. Ros, A. Knoll,
F. Rohrbein, P. van der Smagt, and J. Conradt, “Musculoskeletal
Robots: Scalability in Neural Control,” IEEE Robotics & Automation
Magazine, vol. 23, no. 4, pp. 128–137, Dec. 2016.
[17] Y. Cui, T. Matsubara, and K. Sugimoto, “Pneumatic artificial muscle-
driven robot control using local update reinforcement learning,
Advanced Robotics, vol. 31, no. 8, pp. 397–412, Apr. 2017.
[18] N. El-Atab, R. B. Mishra, F. Al-Modaf, L. Joharji, A. A. Alsharif,
H. Alamoudi, M. Diaz, N. Qaiser, and M. M. Hussain, “Soft
Actuators for Soft Robotic Applications: A Review, Advanced
Intelligent Systems, vol. 2, no. 10, p. 2000128, Oct. 2020.
[19] E. Acome, S. K. Mitchell, T. G. Morrissey, M. B. Emmett,
C. Benjamin, M. King, M. Radakovitz, and C. Keplinger,
“Hydraulically amplified self-healing electrostatic actuators with
muscle-like performance,” Science, vol. 359, no. 6371, pp. 61–65,
Jan. 2018.
[20] N. Kellaris, V. Gopaluni Venkata, G. M. Smith, S. K. Mitchell,
and C. Keplinger, “Peano-HASEL actuators: Muscle-mimetic,
electrohydraulic transducers that linearly contract on activation,
Science Robotics, vol. 3, no. 14, p. eaar3276, Jan. 2018.
[21] S.-D. Gravert, E. Varini, A. Kazemipour, M. Y. Michelis, T. Buchner,
R. Hinchet, and R. K. Katzschmann, “Low-voltage electrohydraulic
actuators for untethered robotics,” Science Advances, vol. 10, no. 1,
p. eadi9319, Jan. 2024.
[22] M. Plooij, G. Mathijssen, P. Cherelle, D. Lefeber, and B. Vanderborght,
“Lock Your Robot: A Review of Locking Devices in Robotics, IEEE
Robotics & Automation Magazine, vol. 22, no. 1, pp. 106–117, Mar.
2015.
[23] D. M. Wang, Y. F. Hou, and Z. Z. Tian, A novel high-torque
magnetorheological brake with a water cooling method for heat
dissipation,” Smart Materials and Structures, vol. 22, no. 2, p.
025019, Feb. 2013.
[24] K. Spanner and B. Koc, “Piezoelectric Motors, an Overview,
Actuators, vol. 5, no. 1, p. 6, Feb. 2016.
[25] S. B. Diller, S. H. Collins, and C. Majidi, “The effects
of electroadhesive clutch design parameters on performance
characteristics,” Journal of Intelligent Material Systems and Structures,
vol. 29, no. 19, pp. 3804–3828, Nov. 2018.
[26] R. Hinchet and H. Shea, “High Force Density Tex-
tile Electrostatic Clutch,” Advanced Materials Technolo-
gies, vol. 5, no. 4, p. 1900895, 2020, eprint:
https://onlinelibrary.wiley.com/doi/pdf/10.1002/admt.201900895.
[27] R. J. Hinchet and H. Shea, “Glove- and Sleeve-Format Variable-
Friction Electrostatic Clutches for Kinesthetic Haptics,” Advanced
Intelligent Systems, vol. 4, no. 12, p. 2200174, Dec. 2022.
[28] V. Ramachandran, J. Shintake, and D. Floreano, “All-Fabric Wearable
Electroadhesive Clutch, Advanced Materials Technologies, vol. 4,
no. 2, p. 1800313, Feb. 2019.
[29] S. Diller, C. Majidi, and S. H. Collins, A lightweight, low-power
electroadhesive clutch and spring for exoskeleton actuation, in 2016
IEEE International Conference on Robotics and Automation (ICRA),
May 2016, pp. 682–689.
[30] S. K. Mitchell, X. Wang, E. Acome, T. Martin, K. Ly, N. Kellaris,
V. G. Venkata, and C. Keplinger, “An Easy-to-Implement Toolkit
to Create Versatile and High-Performance HASEL Actuators for
Untethered Soft Robots,” Advanced Science, vol. 6, no. 14, p.
1900178, 2019.
[31] S. Tanaka, H. Nabae, and K. Suzumori, “Back-Stretchable McKibben
Muscles: Expanding the Range of Antagonistic Muscle Driven
Joints,” IEEE Robotics and Automation Letters, vol. 8, no. 9, pp.
5331–5337, Sep. 2023.
[32] S. Tanaka, R. Kobayashi, H. Nabae, and K. Suzumori, “Fiber
Jamming Mechanism for Back-Stretchable McKibben Muscles,” in
2024 IEEE/SICE International Symposium on System Integration
(SII). Ha Long, Vietnam: IEEE, Jan. 2024, pp. 48–53.
ResearchGate has not been able to resolve any citations for this publication.
Article
Full-text available
Robotic locomotion in unstructured terrain demands an agile, adaptive, and energy-efficient architecture. To traverse such terrains, legged robots use rigid electromagnetic motors and sensorized drivetrains to adapt to the environment actively. These systems struggle to compete with animals that excel through their agile and effortless motion in natural environments. We propose a bio-inspired musculoskeletal leg architecture driven by antagonistic pairs of electrohydraulic artificial muscles. Our leg is mounted on a boom arm and can adaptively hop on varying terrain in an energy-efficient yet agile manner. It can also detect obstacles through capacitive self-sensing. The leg performs powerful and agile gait motions beyond 5 Hz and high jumps up to 40 % of the leg height. Our leg’s tunable stiffness and inherent adaptability allow it to hop over grass, sand, gravel, pebbles, and large rocks using only open-loop force control. The electrohydraulic leg features a low cost of transport (0.73), and while squatting, it consumes only a fraction of the energy (1.2 %) compared to its conventional electromagnetic counterpart. Its agile, adaptive, and energy-efficient properties would open a roadmap toward a new class of musculoskeletal robots for versatile locomotion and operation in unstructured natural environments.
Article
Full-text available
Clothing with integrated high‐force actuators enables wearable haptics for immersive virtual reality (VR) and enables soft exoskeletons for rehabilitation or human augmentation. Electrostatic clutches (ESClutches) offer a very‐low‐energy solution to block motion and are mm thin. Challenges for ESClutches in wearables are 1) effective integration in clothing to accurately block body motion while ensuring comfort, 2) well‐controlled sliding for variable stiffness rendering, and 3) adaptation to shoulder or hip joints. Here, the control of sliding friction of soft ESClutches is demonstrated, using materials that enable both integration in textile and efficient force transfer to the user. We present a 1.3 mm‐thick soft glove with five ESClutches, providing up to 50 N of kinesthetic feedback per finger. A clutch in a thin haptic sleeve that controls elbow extension is reported. Eight cable‐format ESClutches on a shoulder are shown, selectively blocking multiple degrees of freedom. VR tests demonstrate that the glove and the sleeve give the user the ability to rank the softness and weight of virtual objects. In a teleoperation scenario, the glove enables the user to remotely feel an object's stiffness. The ESClutches and textile integration pave a path toward socially acceptable and comfortable kinesthetic haptics. Herein, clothing with integrated high‐force electrostatic clutches that enable wearable haptics for more immersive virtual reality (VR) and teleoperation is presented. Variable stiffness rendering, effective clothing integration, and hip joints’ adaptation challenges are presented. This enables to design haptic garments like a 30 g and 1.3 mm‐thick soft glove and a sleeve rendering variable stiffness and weights in VR.
Article
Full-text available
Soft robotics technologies are paving the way toward robotic abilities which are vital for a wide range of applications, including manufacturing, manipulation, gripping, human–machine interaction, locomotion, and more. An essential component in a soft robot is the soft actuator which provides the system with a deformable body and allows it to interact with the environment to achieve a desired actuation pattern, such as locomotion. This Review article aims to provide researchers interested in the soft robotics field with a reference guide about the various state‐of‐the‐art soft actuation methodologies that are developed with a wide range of stimuli including light, heat, applied electric and magnetic fields with a focus on their various applications in soft robotics. The underlying principles of the soft actuators are discussed with a focus on the resulting motion complexities, deformations, and multi‐functionalities. Finally, various promising applications and examples of the different soft actuators are discussed in addition to their further development potential.
Article
Full-text available
Clutches are key elements for blocking or coupling motion in wearable systems such as soft exoskeletons, haptic clothing, and rehabilitation equipment. Electrostatic clutches (ESclutches) are compact and light, making them particularly well‐suited for wearable applications. They are variable capacitors whose electrodes can slide with respect to each other, with a frictional force between electrodes proportional to the square of the applied voltage. A high force–density textile‐based ESclutch is reported here, generating frictional shear stresses of 21 N cm⁻² at only 300 V, a stress level 11 times higher than any other ESclutch, and 88 times better than textile‐based ESclutches. Actuation and release time are inferior to 5 and 15 ms. Power consumption is below 1.2 mW cm⁻². To reach such high frictional stresses, a dielectric material with high permittivity is chosen (P(VDF‐TrFE‐CTFE)), a fabrication process that enables highly planar dielectric and conductive films on textile is developed, and an alternating current waveform is optimized to minimize space charge. The ESclutch is thin, highly flexible, and weighs only 30 mg cm⁻². The device demonstrated here is designed for wearable applications such as kinesthetic haptic feedback for virtual reality or for soft exoskeletons.
Article
Rigid robots can be precise but struggle in environments where compliance, robustness to disturbances, or energy efficiency is crucial. This has led researchers to develop biomimetic robots incorporating soft artificial muscles. Electrohydraulic actuators are promising artificial muscles that perform comparably to mammalian muscles in speed and power density. However, their operation requires several thousand volts. The high voltage leads to bulky and inefficient driving electronics. Here, we present hydraulically amplified low-voltage electrostatic (HALVE) actuators that match mammalian skeletal muscles in average power density (50.5 watts per kilogram) and peak strain rate (971% per second) at a 4.9 times lower driving voltage (1100 volts) compared to the state of the art. HALVE actuators are safe to touch, are waterproof, and exhibit self-clearing properties. We characterize, model, and validate key performance metrics of our actuator. Last, we demonstrate the utility of HALVE actuators on a robotic gripper and a soft robotic swimmer.
Article
McKibben artificial muscles are used in many musculoskeletal robots because they have characteristics similar to those of natural skeletal muscles in terms of the contraction ratio and force generated. However, they differ from natural skeletal muscles in that they cannot perform passive elongation movements (referred to as “back-stretching” in this study), in which the muscle is stretched by an external force from its natural length while deactivated. Therefore, when using McKibben artificial muscles for antagonist muscle drives, it is necessary to use artificial muscles that are longer than the linear distance between the origin and insertion point of the muscle (causing them to sag while deactivated). However, when acting as an agonist muscle, such an artificial muscle does not fully utilize its contractile capacity and reduces the range of motion of the joint. This is different from natural skeletal muscles, which are placed along the bone without sagging under the skin. In this study, we proposed a back-stretchable McKibben muscle that contracts like a conventional McKibben artificial muscle when air pressure is applied and back-stretches like a natural skeletal muscle with no applied pressure. In addition, we developed a design method based on a mechanical model. In comparative experiments on a robotic arm, we achieved a 1.22-fold increase in the antagonist muscle driving range compared with conventional McKibben artificial muscles without causing the muscles to sag significantly.
Article
Legged robots that can operate autonomously in remote and hazardous environments will greatly increase opportunities for exploration into underexplored areas. Exteroceptive perception is crucial for fast and energy-efficient locomotion: Perceiving the terrain before making contact with it enables planning and adaptation of the gait ahead of time to maintain speed and stability. However, using exteroceptive perception robustly for locomotion has remained a grand challenge in robotics. Snow, vegetation, and water visually appear as obstacles on which the robot cannot step or are missing altogether due to high reflectance. In addition, depth perception can degrade due to difficult lighting, dust, fog, reflective or transparent surfaces, sensor occlusion, and more. For this reason, the most robust and general solutions to legged locomotion to date rely solely on proprioception. This severely limits locomotion speed because the robot has to physically feel out the terrain before adapting its gait accordingly. Here, we present a robust and general solution to integrating exteroceptive and proprioceptive perception for legged locomotion. We leverage an attention-based recurrent encoder that integrates proprioceptive and exteroceptive input. The encoder is trained end to end and learns to seamlessly combine the different perception modalities without resorting to heuristics. The result is a legged locomotion controller with high robustness and speed. The controller was tested in a variety of challenging natural and urban environments over multiple seasons and completed an hour-long hike in the Alps in the time recommended for human hikers.
Article
Numerous mobile robots in various forms specialize in either ground or aerial locomotion, whereas very few robots can perform complex locomotion tasks beyond simple walking and flying. We present the design and control of a multimodal locomotion robotic platform called LEONARDO, which bridges the gap between two different locomotion regimes of flying and walking using synchronized control of distributed electric thrusters and a pair of multijoint legs. By combining two distinct locomotion mechanisms, LEONARDO achieves complex maneuvers that require delicate balancing, such as walking on a slackline and skateboarding, which are challenging for existing bipedal robots. LEONARDO also demonstrates agile walking motions, interlaced with flying maneuvers to overcome obstacles using synchronized control of propellers and leg joints. The mechanical design and synchronized control strategy achieve a unique multimodal locomotion capability that could potentially enable robotic missions and operations that would be difficult for single-modal locomotion robots.
Article
Legged locomotion can extend the operational domain of robots to some of the most challenging environments on Earth. However, conventional controllers for legged locomotion are based on elaborate state machines that explicitly trigger the execution of motion primitives and reflexes. These designs have increased in complexity but fallen short of the generality and robustness of animal locomotion. Here, we present a robust controller for blind quadrupedal locomotion in challenging natural environments. Our approach incorporates proprioceptive feedback in locomotion control and demonstrates zero-shot generalization from simulation to natural environments. The controller is trained by reinforcement learning in simulation. The controller is driven by a neural network policy that acts on a stream of proprioceptive signals. The controller retains its robustness under conditions that were never encountered during training: deformable terrains such as mud and snow, dynamic footholds such as rubble, and overground impediments such as thick vegetation and gushing water. The presented work indicates that robust locomotion in natural environments can be achieved by training in simple domains.