Nano Energy 105 (2023) 108013

Contents lists available at ScienceDirect

Nano Energy
journal homepage: www.elsevier.com/locate/nanoen

A space crawling robotic bio-paw (SCRBP) enabled by triboelectric sensors

for surface identification
Xuyan Hou a, 1, Li Zhang a, 1, Yilin Su b, 1, Guowei Gao a, Yuhui Liu a, Zhonglai Na a,
QingZhang Xu c, Tianxiang Ding a, Li Xiao d, *, Long Li e, *, Tao Chen f, *
a
Research Center of Aerospace Mechanism and Control, School of Mechatronics Engineering, Harbin Institute of Technology, Harbin 150080, China
b
College of Biological and Agricultural Engineering, Jilin University, Changchun 130012, China
c
Department of Mechanical Engineering, Faculty oif Engineering, Hong Kong Polytechnic University, Hong Kong Special Administrative Region of China
d
State Key Laboratory of Space Medicine Fundamentals and Application, China Astronaut Research and Training Center, No. 26 Beiqing Road, Haidian District, Beijing
100094, China
e
Shanghai Key Laboratory of Intelligent Manufacturing and Robotics, School of Mechatronic Engineering and Automation, Shanghai University, Shanghai 200444, China
f
Jiangsu Provincial Key Laboratory of Advanced Robotics, School of Mechanical and Electric Engineering, Soochow University, Suzhou 215123, China

A R T I C L E I N F O A B S T R A C T

Keywords: With the rapid development of space crawling robotics technology, tactile perception, a significant source for the
Space environment robot to sense the external environment, has become the preferred solution to gather information in space.
Tactile perception Sensors embedded in the robotic end-effector unit can collect and encode the large tactile data, allowing the
Bio-Paw device
robot to feel and perceive the real surroundings. Therefore, a space crawling robotic bio-paw (SCRBP) inspired by
TENG sensors
Footfall process
a cat paw, which performs as a compliant device and can flat the impulse force during the robot’s contact process
to the target surface, is proposed in this paper. Meanwhile, a touch-sensing system embedded on SCRBP with self-
powered sensors based on the triboelectric nanogenerator (TENG) technology is proposed which can provide the
multi-dimensional sensation information in real time. By combining with machine learning (ML), the sensory
system can be used for surface identification from the footfall process to the robot’s controller hub. In conclusion,
SCRBP device proposed in this paper has obvious advantages in surface information acquisition, space adapt­
ability, power consumption, cost, reliable signal and minimized data. Accordingly, SCRBP system shows fabulous
potential in space robotics.

1. Introduction technological progress. Hence, endowing robots with this ability is a

grand challenge in robotics. The development and research of space
Space exploration, one of the critical areas of scientific research in­ crawling robot technology based on the bionic have been paid signifi­
vestment worldwide, bears the endless imagination of humanity. Owing cant attention worldwide. For instance, the National Aeronautics and
to continuous progress of aerospace technology, space crawling robots, Space Administration (NASA) developed the robot with dog-like trunk
which are critical enabling tools to elevate the automatic level, play an and limbs to demonstrate the autonomy needed to navigate extreme
essential role for major aerospace countries to utilize the space resource environments without human guidance [4], the robot with lemur-like
[1–3]. Space crawling robots can walk over various terrain surfaces with grip that can crawl, walk and even climb rock walls [5] and the robot
high complexity, unstructured and low recognition that may impede that can walk like a llama to be used on rough, uneven terrain which are
wheeled or tracked vehicles of similar size. However, until now, not suited for robots with wheels or tracks [6]. The ETH Zürich proposed
crawling robots could not match the performance of animals in the robot can jump like a cat to investigate the potential of dynamic
real-world terrain. Many animals, such as cats and dogs, can agility stroll walking and jumping to get around in low gravity environments [7]. The
or run in such environments. Nature, the highly efficient biological Robotic Innovation Center in Bremen, Deutsches Forschungszentrum für
mechanisms it has evolved, offers enormous potential to improve our Künstliche Intelligenz (DFKI) presented the Mantis robot and Pan

* Corresponding authors.
E-mail addresses: lixiao_1982@126.com (L. Xiao), lil@shu.edu.cn (L. Li), chent@suda.edu.cn (T. Chen).
1
These authors contributed equally to this work.

https://doi.org/10.1016/j.nanoen.2022.108013
Received 5 September 2022; Received in revised form 25 October 2022; Accepted 11 November 2022
Available online 15 November 2022
2211-2855/© 2022 Elsevier Ltd. All rights reserved.
X. Hou et al. Nano Energy 105 (2023) 108013

troglodytes robot, which can switch walking modes like Mantis and Pan S-TECS. The integrated S-TECS can measure a finger curvature under an
troglodytes according to different terrain [8,9]. ultra-low working frequency. Tao Li et al. [28] proposed a flexible and
As mentioned above, robots working in harsh space environments thin tactile sensor based on dual-mode TENGs. The effective trans­
need to deal with various terrain surfaces with high complexity, un­ duction of touch and pressure stimulus into independent and inter­
structured and low recognition performing the tasks such as detection, pretable electrical signals which can interpret the contact and hardness
assembly, and maintenance [10–12]. As the robot walks with different of objects. Additionally, by combining with the ML techniques can
gaits, the robotic end-effector periodically contacts the surface. The further enhance the functionalities of the sensor while maintaining the
tactile sensors embedded in the unit robotic end-effectors can collect and minimalistic design [37,38]. Rather than the simple addition of each
encode the data, allowing the robot to perceive the real surroundings, unit, ML is actually the collective behavior of a large number of signals,
and playing an important role in sensing system for robotics. However, which has a good adaptability for understanding complex nonlinear
there are several problems existed, such as insufficient perceptual systems such as space environment. Moreover, the ML techniques can
dimension, limited perceptual accuracy, and unreliability under noisy reduce the signal processing complexity and the computing power, and
environment perception. provide the ease of maintenance too. However, TENGs has not get
With respect to the sensing applications on space robots, NASA’s practical application in space missions so far, with reported robot ap­
NeBula-SPOT robot installed onboard sensor suites, including visual plications on ground. Therefore, the research on the application of
sensor, Contact sensor, and Inertial Measurement Unit(IMU) for the TENGs in space environments will be beneficial to expanding the
autonomous mission. Lemur-3 robot used infrared depth sensors and application range of robotics.
force-torque sensors on each gripper to ensure firm contact on the target According to the discussion above, SCRBP is proposed for space
surface [4,13]. The Pan troglodytes-like robot integrates resistive sen­ crawling robot cruising inspired by the ambush characteristics of a cat.
sors to provide the proprio- and exteroceptive information necessary to SCRBP is composed of 8 sets of toes on the outside and paw pads on the
control the system [9]. However, the output signals of those conven­ inside. Each set performs as a compliant component that can indepen­
tional sensors are susceptible to interference under the extreme envi­ dently flat the impulse force during the robot’s contact with the target
ronments in space [14–16]. For instance, the visual sensor requires surface. The achievement of multi-dimensional sensation information
texture to perform stereo matching and consequently struggles with attributes to sensors of two working modes and four kinds arranged on
low-texture surfaces or when parts of the image are under- or over­ SCRBP. The sensors work in vertical contact-separation mode and
exposed [17]. The resistance sensor drifts at the thermal cycle in space. contact-sliding mode. The sensors integrated on SCRBP generate
The detection distance and detection precision of the infrared sensors different signal outputs according to the actual contacts on the footfall
decrease due to the influence of space radiation. Most of these sensors, process and help the robot make logical judgments, as shown in Fig. 1b.
which require extra power consumption, complex circuit arrangement By combining with ML, the spatial array distributed sensor on SCRBP
and corresponding structural protection, will occupy the limited can be used for surface identification from the footfall process to the
launching load of the rocket. In addition, the sensor will be subjected to robot’s controller hub. Generally, SCRBP can assist the space crawling
the harsh working condition of launching, and the complex structure robot in making logical footfall judgments, showing immense applica­
increases the failure probability. Therefore, components with low con­ tion potential.
sumption and high reliability will be more competitive in future launch
missions. Specifically, sensors integrated in space robots should be as 2. Design and working mechanism
lightweight, miniaturized, multifunctional and highly reliable as
possible. 2.1. Design and Integration of SCRBP
Triboelectric nanogenerator sensors(TENGs) based on the friction
effect are a new area of great promise [18]. Due to the requirements and The mechanical part of the SCRBP’s structure is shown in Fig. 1c.
technical challenges of sensor selection in space environments such as Inspired by the cat paw, SCRBP has 8 groups of sub-paw, each sub-paw is
microgravity, extreme temperature variations, and cosmic radiation, composed of a toe pad unit and a palm pad unit that can touch the
TENGs uses the frictional effect to capture the energy in the system’s surface independently. Each pad has 1 degree of freedom in the direction
environment to provide a constant, self-driven energy source for sensors. perpendicular to the touching surface. SCRBP can be installed on the
The self-powered characteristic of the TENGs is beneficial to miniatur­ end-effectors of the robot, and the 16 pads in 2 groups of the SCRBP
ization and compact arrangement while saving energy consumption for enable it adaptable to surface fluctuation.
the robot to extend its working capacity. As the material selection of the As depicted in Fig. 1d, the tactile part of the SCRBP has four types of
sensor is very flexible, a large number of commercial materials can be TENGs in different locations. According to the principle of TENGs, they
used as potential electrode materials. TENGs can choose appropriate can be divided into triboelectric contact sensor (TRCS) that works on
materials to deal with the space environment, such as temperature contact-separation mode and triboelectric sliding sensor (TRSS) that
change, radiation and vacuum. TENGs has four working modes, which works on contact-sliding mode. three TRCS and one TRSS are arranged
are vertical contact-separation mode, contact-sliding mode, on each sub-paw, and there are totally 32 TENGs arranged on SCRBP.
single-electrode mode, and freestanding triboelectric-layer mode that The featured large voltage output of TENGs enables the sensors can be
can measure multiple parameters and obtain comprehensive informa­ further scaled down and placed of a larger number of sensors for better
tion on the target. As a device with tactile, pressure, and motion sensing, resolution or expand the tactile of robot. Meanwhile, large voltage and
it has been focused on by many researchers for its advantages of self-generated output can be clearly distinguished from ambient noise.
self-powered, multimodality, and flexible material selection [19–22]. Fig. 1a shows the space robot performing space missions on large
Presently, TENGs have shown great potential in robotics [23–28], spacecraft or planet with SCRBP mounted on the end. When the robot is
medical facilities [29–31], wearable electronics [32–34], and Internet of in footfall process, SCRBP can adapt to the surface fluctuation inde­
Things [35,36]. pendent movement due to each sub-paw and each pad unit. According to
In relation to robot perception, TENGs has evolved enormous po­ the different contact conditions of each sub-paw, SCRBP generates the
tential to improve technological progress. Xuyan Hou et al. [23] pro­ voltage signals and transmit them to the control hub. After analyzing the
posed a robotic system for attaching and moving on the surface of signals, the robot further understands the surrounding surface and
spacecraft integrated with multifunctional tactile sensors. A continuous reprogrammed, i.e., the robot converts the gait in response to save en­
monitoring system was established to improve the safety and efficiency ergy or redraws the route to raise efficiency of the mission.
of in-orbit assembly of ultra-large-scale spacecraft. Mingzhu Zhu et al.
[24] print a soft robotic finger embedded in a single-electrode TENGs,

2
X. Hou et al. Nano Energy 105 (2023) 108013

Fig. 1. Schematics of space crawling robotic bio-paw (SCRBP) system. (a) Schematic diagram of space crawling robot work in space environment. (b) Schematic
diagram of the robot perceiving through SCRBP. (c) Schematic diagram of SCRBP. (d) Schematic diagram of TRCS and TRSS.

2.2. Mechanical part of SCRBP damage the operating environment. Minimizing the vibration amplitude
and reducing the disruption to the target surface are essential for
For millions of years, Animals have evolved organs and habits to fit completing the whole mission. Therefore, cat paw is an excellent bionic
the Nature environment, which is full of complexity, risk, unpredict­ object of space robot foot.
ability. For example, birds can be fully air-borne owing to the evolution The structure of the cat paw is shown in Fig. 2a. The paw is composed
of the wings, feathers, and flapping modes; mole rats adapted to living of retractable toenails, digital pads, and metacarpal pads [48,49]. The
underground attributed to their sharp claws, strong forelimbs, and forepaw has five forward toenails and digital pads, while the hindpaw
special digging skills. As for the cat, thanks to the migration activities of has four. Each paw has a metacarpal pad in the center. The cat’s digital
ancient humans, they are distributed almost everywhere in the world pads and metacarpal pads are made up of fat and have numerous elastic
like grassland, desert, rainforest, mountain, and other places [39]. As for fibers that are filled with nerves to buffer and sense the vibrations of the
the cat, they are skillful ambush predators. The special structure of cat ground. Tendons in each paw gather the digital pads around metacarpal
paw makes significant contribute to its adaptability advantages. With pad and make the digital pads excess to metacarpal pad. They are used to
the help of delicate physiological structure, soft meat pads and retract­ ensure the cat walks and ambushes silently. The sharp toenails are
able toenails, cats can stalk and pounce their prey silently even in rough retracted inside the paws to prevent them wear, and only stick out when
terrain [40–43]. Correspondingly, endow the robot with the ability to eating, preying, fighting, climbing.
adapt to unstructured terrain is of great significance. The vibration of Fig. 2b present the footfall behavior of laboratory-bred cats under
the target surface is an essential factor affecting the success rate of the unfamiliar terrain by placing the cat on a carpet covered with foam balls.
mission when the space robot operates on high complexity surface like The process of the cat’s footfall can be divided into four steps.
space station, satellite, moon, and Mars [44–47]. The excessive distur­
bance will interfere with the regular operation of the target spacecraft or

3
X. Hou et al. Nano Energy 105 (2023) 108013

Fig. 2. Design and working mechanism of the mechanical part of SCRBP. (a) Anatomy of the cat paw [49]. (b) Four steps of the cat’s footfall process: (i) The cat
prepare to lay its paw on the carpet covered with foam balls. (ii) The cat treads its paw slightly on the carpet. (iii) The cat lay its paw totally on the carpet. (iv) The cat
shifts the weight to the feet. (c) Corresponding structure between the cat paw and the SCRBP’s sub paw proto. (d) Mechanical structure of SCRBP: 1- Cover plate. 2-
Shell. 3- Toe pad spring. 4- Rod spring. 5- Toe pad. 6- Rod. 7- Palm pad. 8- Translational joint. 9- Height (5 mm). 10- Revolute joint. 11- Slot joint. 12- Cylindri­
cal joint.

i. The cat observes the carpet with the toenails retracted naturally. digital pads and metacarpal pad do not touch the ground simul­
The pads are closed together while in the air. The cat’s weight taneously but in sequence. The digital pads feel the ground first
holds still in its original position (Fig. 2b(i)). and make preliminary judgments. The metacarpal pad then
ii. The cat treads its paw slightly on the carpet with its digital pads to touches the ground for further contact and final judgment (Fig. 2b
sense the ground (Fig. 2b(ii)). (iv)).
iii. The cat continues the footfall process. The metacarpal pad
touches the carpet with the total paw on the carpet to judge the Based on the biomechanics of the cat paw analyzed above, the in­
surface stiffness (Fig. 2b(iii)). dependent movement of the pads and their sequential contact are the
iv. After the safety confirmation of the ground, the pads spread apart key factor that the cat can touch the surface silently. The independent
and the cat shifts its weight to this new foot. The whole footfall movement of the pads make it possible for cat adapting to the surface
process is continuous and completed in a short time. The cat’s fluctuation of the target. While, the pads sequential contact to the

4
X. Hou et al. Nano Energy 105 (2023) 108013

surface can extend contact time and reduce impact. Hence, SCRBP is high vacuum, thermal radiation, and other environmental conditions of
designed, as shown in Fig. 2c. Corresponded to the cat paw, this device the space environment, high requirements are proposed for spacecraft
consists of Toe pads, Palm pads, Rod, Shell, Cover plate, and some materials. Conversely, due to the limitations of rocket launch capabil­
springs. The motion pairs between each part are shown in Fig. 2d. The ities, spacecraft materials should satisfy the requirements of low density
cylindrical joint between the Toe pad and the Shell is designed that can and high rigidity. Based on the aforementioned factors, alloy materials
move up and down which mimic the Digital pad of cat paw. Meanwhile, that are commonly used in spacecraft mostly correspond to titanium
the Rod, which is similar to the tendons in the cat paw, connects with the alloys and aluminum alloys.
Shell with the slot joint and connects with the Toe pad with the revolute Titanium is a new type of metal material with good conductivity of
joint, is used to limit the motion of Toe pad and increase the intensity of 2.34 × 10–6 Ω⋅m. The performance of titanium depends on the presence
SCRBP. The translational joint between the Palm pad and the Shell, of impurities, such as carbon, nitrogen, hydrogen, and oxygen. Titanium
which is similar to the metacarpal pad, also provide free sliding in alloys are doped with different substances based on types and perfor­
footfall direction. The Toe pads and Palm pad of SCRBP can move mance requirements. Specifically, α-phase titanium alloy is most widely
independently. In order to improve the standing stability of the space used in aerospace due to its structural rigidity. Therefore, the titanium
robot, the Toe pads and Palm pad are designed into eight groups alloy sample used in this study is Ti-6Al-4 V (TC4), which is commonly
distributed in a circular array. The height difference between the Toe used in aerospace structural parts and elastic parts.
pads and the Palm pads is 5 mm, which can realize the sequential con­
tact to the surface. The springs connected to the Palm pads start to 2.3. Tactile part of SCRBP
compress after the springs connected to the Toe pads compressed the
preset height difference. AS specific analyzed in supplementary Section To achieve the tactile function, two types of TENGs are designed
1, SCRBP have the compliant characteristic that can flat the impulse respectively based on the sequential contact characteristics of cat paw
force during the contact process. and the structures of SCRBP.
In space missions, due to the extreme high and low temperatures,

Fig. 3. Structural design of TENGs system. (a) Triboelectric contact sensors (TRCS): (i) Working principles of TRCS. (ii) Schematics and photos of TRCS’s installation
position. (iii) Triboelectric outputs of TRCS. (b) Triboelectric sliding sensors (TRSS): (i) Working principles of TRSS. (ii) Schematics and photos of TRSS’s installation
position. (iii) Triboelectric outputs of TRSS.

5
X. Hou et al. Nano Energy 105 (2023) 108013

2.3.1. Triboelectric contact sensors (TRCS) by adding contact friction to generate more pulses. Moreover, TRSS is
In order to sense the contact between the SCRBP and the surface fabricated and tested in Fig. 3b(iii), the output signals show some pos­
during the footfall process, TRCS is researched as shown in Fig. 3a. The itive pulses, and then negatives pulses in a sliding loop which is
Nylon strap used for the flexible positive triboelectric material is consistent with the analysis. Fourier transform is used to process the
enclosed into a ring as the positive electrode layer. The PTFE piece used output signal, and the two maximum sub-signals obtained are shown in
for the flexible negative triboelectric material is prepared in the ring as the Fig. S8a and Fig. S8b which can be easily identified from the sliding
the negative electrode layer. These two layers are separated with insu­ and contact effect. Fig. S8c is the sub-signals corresponding to the sliding
lation material. The friction charge generated by the positive and effect, and Fig. S8d is the sub-signals corresponding to the contact effect.
negative poles is collected by aluminum foil arranged on the back of Obviously, the output signal of TRSS is the superposition of the sliding
negative layer. Since the negative electrode layer, the insulation layer and contact effect. TRSS is fabricated and tested in Fig. 3b(iii), the
and aluminum foil have been fixed on one side of the ring enclosed by output signals show some positive pulses, and then negatives pulses in a
the positive electrode, this side of the ring is thicker than the other side. sliding loop which is consistent with the analysis. The number of pulses
The enclosed ring has an uneven thickness. The curvature of the side that is the same as the embossing number on the negative layer. TRSS is
has the negative electrode layer, the insulation layer and aluminum foil arranged in the middle of the Palm pad to feel the sliding and its di­
is smaller than the curvature of the other side, shown as Fig. 3a(i)~(ii). rection (Fig. 1d).
The interfacial charge transfer occurs as the electronegativity differ­
ences of two materials, resulting in a potential difference. When the 2.3.3. Working mechanism of SCRBP
deformed-ring of the TRCS is compressed under external force, the Inspired by the structure and the footfall process of cat paw, the
distance between the two layer decreases. The electric potential between multi-dimensional arrangement of sensors on SCRBP is shown in Fig. 1d
the two electrodes fades away and a reverse current flows from the with the TRCS sensing the contact and TRSS sensing the sliding
negative electrode to the ground. When the two layers are in full contact, conversed by continuous contact. TRCS can be easily installed on the
the potential difference is zero. When the external force is unloaded, the intended location. However, the internal installation of TRSS need to be
deformed-ring of TRCS tends to return to its original position under the well considered. The main difficulty is to fix the electrode pair in the
elastic property of the positive triboelectric material. Based on the an­ corresponding position respectively. The sliding parameters of TRSS,
alyses above, a TRCS sensor is fabricated and tested according to its including sliding distance, sliding track, and the Fixation of each elec­
working mode. The experimental signal agrees with the theoretical trode are constrained by the mechanical structure of SCRBP. Specif­
analysis. As shown in Fig. 3a(iii), when the TRCS is pressed, a positive ically, the positive electrode is installed in the corresponding position of
signal is detected, and when the TRCS is released, a negative signal is the Shell part of SCRBP, and the negative electrode is fixed to the groove
detected. TRCS are arranged at the bottom of the Toe pad, the bottom of of the Palm pad part. The track of TRSS is fixed by a translational joint
the palm pad, and on the Cover plate as the toe pad sensors, the Palm between the two parts which can restrict the movement of the Palm pad
pad sensors, and the Cover plate sensors (Fig. 1d). The deformed-ring part to vertical direction and ensure the two parts slide in parallel mo­
structure of TRCS can take full advantage of the elasticity of flexible tion. Additionally, a limit structure of the Shell part is designed (Fig. S9a
materials to ensure sufficient contact-separation between the interfaces and Fig. S9b) to ensure the TRSS to have the same initial status and the
of the two layers. motion path. As depicted in Fig. 3b(ii), Fig. S9c and Fig. S9d, the bump
located on positive electrode can be installed from external by slotting
2.3.2. Triboelectric sliding sensors (TRSS) the Shell of SCRBP. The slot simplifies the installation of TRSS. In
In order to sense the continuous contact between the SCRBP and the addition, the existence of the slot makes the friction and contact between
surface during the footfall process and eliminate static or noisy signals, the two layers of TRSS electrodes can be easily adjusted by resetting the
TRSS is investigated as shown in Fig. 3b. The flexible positive and height of the bump without uninstalling the SCRBP.
negative triboelectric materials are shaped as positive electrode layer SCRBP realizes the function of autonomous footfall judgment, which
and negative electrode layer via mold fabricated by 3D printing, shown helps the robot to sense the environment better. The judgment during
in Fig. S6 and Fig. S7. The shaped positive electrode layer has a bump the footfall process of SCRBP is as follows:
and the negative has types of embossing. The friction charge is collected
by aluminum foil arranged on the back of positive layer. Fig. 3b(i) shows i. As SCRBP begins to fall toward the target surface at a certain
the sliding process of TRSS. The bump and embossing on the two elec­ speed, the toe pad sensors and palm pad sensors record the con­
trodes contact each other when TRSS start sliding. Therefore, when the tact point in time. Combined with the landing parameters of the
two layers of the TRSS slide relatively to generate the charge transfer robot body, the topography of the target surface can be described.
under the external force, there are two types of friction effect happens, The height difference between the Toe pads and the Palm pad is
one is the sliding effect and the other one is the contact effect. As the set to achieve the contacted sequence similar to that of the digital
positive layer slide out with respect to the negative layer, the relative pads and the metacarpal pad. The digital pads array of each sub-
contact area decreases, resulting the separation of charges and drive the paw is in contact with the target surface before the metacarpal
current from the positive layer to the ground. As the positive layer start pads array, which helps SCRBP maintain balance during the
to slide back, resulting the relative contact area increases, the reverse footfall process. The contact sequence and the spring on the
current flows from ground to the positive layer. For the contact effect, digital pads and metacarpal pads can also achieve a passive
the bump and embossing on the shaped layer can make the electrodes compliance function, increasing the system’s flexibility.
maintaining reliable contact and increase the sliding friction to improve ii. As SCRBP continues to fall towards the target surface, TRSS feels
the output signals of TRSS, avoiding the inappropriately waves caused the slide with the Shell and can detect the continuity of the
by virtual contact between the two electrodes. As the bump approaches contact process between SCRBP and the target surface. When the
to the embossing, the electric potential between the two electrodes de­ target surface collapses, sinks, breaks away or missing due to
creases with the decreasing distance, and drives the current from the unexpected reasons, the sensors can feel the interruption of the
positive layer to the ground. As the bump moves away from the slide signal and pass it to SCRBP for judgment. In addition, this
embossing, the electric potential between the two electrodes reverse design provides an early warning for SCRBP to avoid failure.
with the increasing distance, and drive the current from the ground to iii. As SCRBP falls further towards the target surface. The cover plate
the positive layer. By using the shaped flexible electrodes make it easier sensors feel the contact with the Shell. The contact signal means
for the frictional charges to separate with the increased friction, creating that the Palm pad can resist the reaction of the spring sets and
a larger dipole matrix between the electrodes, and reinforce the signal reach to contact the Cover plate. So, the reaction of the spring sets

6
X. Hou et al. Nano Energy 105 (2023) 108013

is used, which are designed as the final judgment of the footfall robots have been extensively investigated as critical enabling tools to
ability to the target surface. elevate the automatic level and utilize the space resource, which provide
exciting opportunities for tactile perceiving issues. Equipped with tactile
2.4. Analysis of triboelectric materials for space environment perceiving, robots can expand their capabilities of space in-orbit service,
and minimizing the vibration amplitude and reducing the disruption to
The robots are exposed to vacuum, extreme temperature variation, the target surface. From the consumption perspective, the mechanical
and radiation environments when performing the space mission. The part of SCRBP is a passive structure that do not need power supply or
materials of the robot must endure the harsh space environment con­ energy consumption, while the tactile part uses self-powered TENGs. In
ditions and maintain stable performance. Fortunately, benefit from the addition, assembling SCRBP on the end of the robot does not require
broad the variety of alternative triboelectric materials, it is possible to extra joints. Therefore, the consumption of SCRBP for walking is the
select materials applied both in space and TENGs. Based on the limita­ weight of SCRBP itself and the additional footfall force provided by the
tions mentioned above, polyamide-66 (PA66) and aluminum (Al) are robotic joints, as shown in supplementary Section 1 ’Structural Opti­
selected to the potential positive triboelectric material, while kapton mization of SCRBP’. From the cost perspective, the material pairs of
(PI) and Polytetrafluoroethylene (PTFE) are selected to the potential Nylon with PTFE used for SCRBP have already verified and used in space
negative triboelectric material. These four materials have been widely application, resulting the low verification cost. The sensors are self-
used in space applications. Specifically, PA66 is widely used in aero­ powered which can decrease the payloads in the space mission and is
space products, such as power frames, PCB frames and space suits due to beneficial to miniaturization and compact arrangement. The sensors of
the outstanding heat resistance, mechanical properties, dielectric SCRBP works at low frequency, reducing the limited and precious
properties [50]. Moreover, the properties of nylon can be modified by transport bandwidth. In general, it is believed that SCRBP can adapt to
adding polymers to reinforce its mechanical properties, weather resis­ the space mission and throw light on the space robots.
tance, low temperature brittleness, flame retardant properties. Al, as
commonly used materials for spacecraft, is used in spacecraft structural 3. Result and discussion
components such as lightweight honeycomb core panels made of
aluminum foil. PI, as one of the raw materials in making multi-layer 3.1. Selection of triboelectric materials for space environment
insulation(MLI) for coating spacecraft, has the ability of high strength,
high temperature resistance and radiation resistance [51]. PTFE has Without the development and the verification cost of new materials,
strong properties of corrosion resistance and aging resistance, and the selected materials are commonly used in space environment that can
remain elasticity and strength under cryogenic conditions, which has produced in low cost. As depicted in Fig. 4b and Fig. 4c, the selected
been applied in millimeter wave radome, sealing gasket, coating mate­ materials pairs both have the certain voltage output. For the output
rial of cable, etc. PTFE has the property called cold flow that make signals of TRCS shown in Fig. 4b(i)~(iv), Nylon has a higher output
plastic deformation under the long term load. voltage than aluminum for positive electrode, while PTFE has a higher
Space environment conditions such as vacuum, temperature varia­ output voltage than PI for negative electrode. The material pair of Nylon
tion and irradiation may cause significant effects on the performance of with PTFE has the maximum voltage output. After enduring the cyclic
TENGs. The vacuum degree of the space environment is about 10− 5 Pa to loading tests shown in Fig. 4b(v), the material pair of Nylon and PTFE
− 10− 11 Pa. According to the analysis [52], by eliminating the influence shows good reliability for the high elasticity that can maintain the
of air, moisture and other pollutants, the surface charge density of the deformed-ring of TRCS. Fig. S10a presents the status of Nylon electrode
triboelectric material can be strengthened in vacuum and the perfor­ after the cyclic loading tests which nearly remains its original shape. For
mance enhancement of TENGs can be significantly improved. Under the the output signals of TRSS shown in Fig. 4c(i)~(iv), the material pair of
protection of thermal control technology, the design temperature of Nylon with PTFE has the maximum output signals relatively to the other.
space robot components is − 40–40 ℃, while the temperature range of Additionally, the material pairs of Nylon with PTFE also remain pref­
space environment is between − 170 ℃ and 120 ℃. Temperature erably response after sustain the cyclic loading tests shown in Fig. 4c(v).
mainly affects the ability of surface charge density [53] and [54], the As depicted in Fig. S10b, PTFE has the property making the electrode
output of TENGs may decrease, or even disappear under the extreme preserve shape relative to PI.
temperature (less than-50 ℃ or more than 260 ℃). The interaction Furthermore, tests for the effect of the space environment on the
between irradiation and materials is the radiation displacement effect performance of TRCS and TRSS by applying the material pairs of Nylon
and ionization. The radiation displacement effect refers to the defects with PTFE are carried out as verification, and the results are shown in
and performance decline in material caused by atoms lost their original Fig. S11. The result shows that the vacuum pressure condition (102 Pa)
lattice and transferring to others under the irradiation of high-energy has a positive effect on the performance of TRCS and TRSS compared
particle-flow in space. The selected materials are commonly used in with the normal atmospheric (105 Pa) pressure condition (Fig. S11c).
aerospace, which has strong irradiation resistance during lifetime. Secondly, it is verified in Fig. S11d that the effect of the thermal con­
Ionization means the surface charging and internal charge accumulation dition (− 40 ℃ ~+40 ℃) on performance of TRCS and TRSS. The result
under the bombardment by the charged particle-flow in space. Based on shows that the TENGs has well performances under both the lowest
TENGs’ theory, the irradiated surface accumulates charges may increase testing temperature (− 40 ℃) and the highest temperature (+40 ℃). The
the peak of the output signals. irradiation tests shown in Fig. S11 represent the results of TRCS and
TRSS exposure to radiation environments. The results show that SCRBP
2.5. Analysis of SCRBP used in space application maintains good performance in the irradiation environments. Moreover,
as depicted in Fig. S12b, the durability of the TRCS cyclic loading tests
The designed bio-paw SCRBP is composed of the mechanical part are carried out in the sand (Particle Diameter 0.125 mm~0.25 mm) to
which can flat the impulse as the robot contacts to the surface and the simulate the planet surface. The results show that the output perfor­
tactile part which have sensors array that can detect and sense different mance remains stable after 400 working cycles. Therefore, TRCS is
surfaces during the footfall process. The sensors arranged on SCRBP are strong enough to bear the touch cycle.
based on the TENG and use material pairs of Nylon for positive electrode
with PTFE for negative electrode that satisfy the space environment. 3.2. Test for the selection of triboelectric parameters
However, using the qualified materials are insufficient for space appli­
cation. There need to be comprehensive consideration based on demand, The scalabilities influence of 2 cm*cm, 3 cm * 3 cm, and
energy consumption, cost, etc. From the demand perspective, space 4 cm * 4 cm is analyzed in Fig. 4d, the results show the scalabilities has

7
X. Hou et al. Nano Energy 105 (2023) 108013

(caption on next page)

8
X. Hou et al. Nano Energy 105 (2023) 108013

Fig. 4. Sensing test of TENGs system. (a) The test platform used to test the output characteristics. A linear motor is used to cyclically stimulate the sensor, and an
oscilloscope is used to record the sensor output voltage. (b) Test results of TRCS with different electrode materials: (i) Aluminum with PI. (ii) Nylon with PI. (iii)
Aluminum with PTFE. (iv) Nylon with PTFE. (v) Cyclic loading tests of Nylon with PTFE. (c) Test results of TRSS with different electrode materials: (i) Aluminum
with PI. (ii) Nylon with PI. (iii) Aluminum with PTFE. (iv) Nylon with PTFE. (v) Cyclic loading tests of Nylon with PTFE. (d) Test results of TRCS with scalabilities: (i)
Size: 2 cm * 2 cm. (ii) Size: 3 cm * 3 cm. (iii) Size: 4 cm * 4 cm. (e) Test results of TRSS with shape optimization: (i) Schematics and dimensions of TRSS. (ii) Triangle.
(iii) Rectangle. (iv) Semicircle. (iv) Trapezoid. (f) Test results of TRSS with size optimization: (i) Table of height parameters of TRSS’s layers. (ii) Test results of size
optimization. (iii) Variation of the negative electrode height (h1) on the output signals at the positive electrode height (h2). (iv) Variation of the positive electrode
height (h2) on the output signals at the positive electrode height (h1).

an effect on the output and the large size has a certain beneficial effect approach for SCRBP to recognize surfaces, it is found that all pavements
that can generate high voltage. However, too large a size may also with 15 different characteristics can be discerned with a high accuracy
adversely affect output. The size of 3 cm * 3 cm have the highest voltage (Fig. 5c(i)). The identification of different terrains is achieved through a
output in the test. Large size may lead to deformation ring recovery series of statistical data combined with machine learning. Besides, fea­
slowly and affect the output of TRCS. tures like other statistical values, the logical connections behind the data
The influence of TRSS’s negative electrode shape is analyzed in of different pixels, etc. are also the promise for machine learning to work
Fig. 4e, the spacing and shape height are guaranteed to remain the same perfectly. Therefore, SCRBP confluence with machine learning analysis
in the test (Fig. S6c). Four shapes of triangle, rectangle, semicircle, and shows significant ability of surface identification.
trapezoid are tested in Fig. 4e(ii)~(v). The results show that the triangle Supplementary material related to this article can be found online at
has the best effect and can reach the maximum output voltage between doi:10.1016/j.nanoen.2022.108013.
the two electrodes. Moreover, by analyzing the perceptual mechanism of
TRSS, the triangle has the maximum relative separation distance during 4. Conclusion
the sliding of the two electrodes and the most stable structure, which
leads to the maximum output compared to other shapes. The detailed Perception and cognition are essential functions of space robots.
analysis is explained in supplement Section 4 ’Analyzed for the Opti­ Without the sensation system, space robots could hardly accomplish the
mized Shape’. on-orbit operation. In this article, we present SCRBP to assist the space
In order to find the height influence of TRSS’s layer to the output robot in better understanding the target surface. Space robots need to
signals, a group of size optimization tests are analyzed at a certain total minimize the disturbances of the target surface while maximizing the
height of 4.6 mm in Fig. 4e(i). Fig. 4f(i) presents the table of overlaps information get from it when working on a high-complexity, unstruc­
between the height of negative electrode (h1) and the height of positive tured, and low-cognition surface. SCRBP is inspired by the ambush
electrode (h2). The red part in Fig. 4f(ii) indicates that the output voltage characteristics of a cat. With the arrangement of the sensor array, SCRBP
exceeds 180 mV. The maximum output is corresponded to h1 = 3.4 mm can autonomously discern the different terrains with topographic relief,
and h2 = 3 mm, and in this configuration TRSS has the highest output stiffness, et al. that assist the space robot in making logical footfall
voltage 264 mV. Additionally, the height variation is tested in Fig. 4f(iii) judgments to avoid the robot instability problem. The working process
and (iv). The result shows that the optimal output signals increase first, demonstrated with machine learning shows that SCRBP can sense and
and then decrease at the certain height of h1 and h2 respectively. identify the target surface state by using the TRCS and TRSS and the
multi-dimensional array. By combining with ML, SCRBP can realize
3.3. Surface identification of SCRBP surface identification. SCRBP developed in this paper can be used as an
essential complement of the perception system. According to the re­
As mentioned above and analyzed in supplementary Section 1, quirements of different scenes, it can improve the intelligence and
SCRBP can response different signals on surface with variable stiffness automation degree of robots.
through the robot’s footfall process. However, due to the misunderstand It shows prospects for further space station operations, moon and
of the robot, the damping increased by TRCS and TRSS, and the Mars walking, and asteroid in-situ exploration. Future work will be
abnormal charging caused by cosmic ray in space environment, the carried out on quantification surface stiffness recognition and the inte­
practical application ability of theoretical analysis to help decision- gration perception of visual and tactile.
making is limited. It is of great significance to use the robot learning
method to analyze the sensor measurement signal. To establish an 5. Experimental section
optimal machine learning training model, efforts have been made to
examine the appropriate feature extraction method and the optimized 5.1. Fabrication of SCRBP and the sub-paw proto
algorithm. The processing procedure is shown in Fig. 5d(ii). The time-
varying raw data generated by SCRBP is statistically processed into SCRBP and its sub-paw proto are made by the customization with 3D
total 6400 inputs from 8 groups of 32 sensors, and the standard devia­ printing in order to verify the effectiveness in target surface recognition,
tion is extracted. The surfaces are classified into several decision tree as shown in Fig. 4. The sub-paw proto is two times larger than the SCRBP
algorithms for training, and the results of cross validation and new to verify the effect of scalabilities on the output, while the 4.6 mm total
datasets test are obtained. height of the sliding part is the same to ensure the shape optimization of
The 15 different surfaces (Fig. 5b) are chosen to verify the capability TRSS. The Shell, Cover plate, Rod, Palm pad, and Toe pad are all made of
of surface identification. The selected surface can be divided into two photosensitive resin. The Toe pad spring and Rod spring are made of
categories, one is the space surface simulant like materials with PI- stainless steel. The dimension of TRCS arranged on SCRBP is
coated, insulation board, honeycomb board, sand, and Mars soil simu­ 3 cm * 3 cm on Toe pads, 2 cm * 2 cm on Palm pads, and 2 cm * 2 cm
lant, the other is the surface that are commonly touched like rigid plane, on the Cover plate. The dimension of TRSS is 7 mm * 22 mm on SCRBP
block surface, coarse sand, soil, sponge mat, etc. Each surface is touched and 14 mm * 44 mm on the double sized sub-paw proto (Fig. S13).
by SCRBP with 24 times (360 for all). In order to make the neural
network recognize the key features of the terrain, SCRBP with random 5.2. Testing platform for TENG materials selection and optimization
footfall state during each test (Movies S3-S7), verifying the robustness
and generalization of neural network in surface identification. It is The LinMot ZK-DIY-520 linear motor, linear guide and the test rack
essential for robot to build a data base of surface to deal with the sur­ (Fig. 4a) is used as the test platform. This test platform can ensure that
rounding terrains. Hence, by applying the above machine learning the deformation of the electrode materials maintains the same for each

9
X. Hou et al. Nano Energy 105 (2023) 108013

Fig. 5. Schematics and photos of working

mechanism of SCRBP on different terrain
surface. (a) Schematics and photos of
SCRBP: (i) Test process of SCRBP. (ii)
Photos of SCRBP. (b) Schematics of SCRBP
on 15 different terrain surfaces: (i) Rigid
plane. (ii) Block. (iii) Coarse sand. (iv)
Leather board. (v) Water bag. (vi) Insu­
lation board. (vii) Honeycomb board.
(viii) PI-coated glass. (ix) PI-coated board.
(x) Foam ball board. (xi) Plastic board.
(xii) Soil. (xiii) Mars soil simulant. (xiv)
Sponge mat. (xv) Sand. (c) 3D plots of
SCRBP outputs corresponding to different
objects. (d) Process of terrain surface
recognition: (i) The confusion map for
machine learning outcome of 15 different
terrain surfaces with 32 channels. (ii)
Flow diagram of the Neural Network
process.

10
X. Hou et al. Nano Energy 105 (2023) 108013

contact. Furthermore, the linear guide and the rack can also ensure the Acknowledgments
vertical contact of the TENGs. The test rack (Fig. S14) is fixed at the end
of the linear guide to simulate the contact and the sliding effect between The authors thank the funding support from the National Key
SCRBP and the surface. TRCS and TRSS can be installed on the test rack. Research and Development Program of China (2019YFB1309600), the
By adjusting the motion cycle of the linear motor, the cycle of the test National Nature Science Foundation of China (No. 62073229), Jiangsu
signal can be adjusted. Province Robotics and Intelligent Manufacturing Equipment Engineer­
ing Laboratory (Soochow University) Open Fund Project (Grant No.
SDGC2141), EDL fund of Beijing Institute of Space Mechanics and
5.3. Experimental devices for the test of space environmental conditions Electricity (Grant No. EDL19092127).

The high-low temperature test for the TENGs is carried out by using Appendix A. Supporting information
the High-low Temperature Test Box (HTTB) in Harbin Institute of
Technology. The temperature range of the HTTB is − 100 ℃~150 ℃ Supplementary data associated with this article can be found in the
(Fig. S11a(i)). The vacuum test for the TENGs is carried out by using the online version at doi:10.1016/j.nanoen.2022.108013.
Scanning Electron Microscope (SEM) in Soochow University. The pres­
sure range of the SEM is 102 Pa~105 Pa (Fig. S11a(ii)). The irradiation References
test for the TENG sensor is carried out by refitting the SEM with an
irradiation device in Soochow University. In addition, actuators are [1] Yang Gao, Steve Chien, Review on space robotics: toward top-level science through
added in aforementioned devices to realize the contact and separation space exploration, Sci. Robot. 2 (2017) 5074.
[2] Xi.Lun Ding, Ye.Cong Wang, YaoBing Wang, et al., A review of structures,
between the positive materials and the negative materials.
verification, and calibration technologies of space robotic systems for on-orbit
servicing, Sci. China Technol. Sci. 64 (2021) 462–480.
[3] Alex Ellery, Tutorial review on space manipulators for space debris mitigation,
5.4. Experimental surface for intelligent identification Robotics 8 (2019) 34.
[4] A. Agha, K. Otsu, B. Morrell, et al., Nebula: Quest for robotic autonomy in
challenging environments; TEAM CoSTAR at the DARPA Subterranean Challenge,
The test of the obstacle surface is a 60 * 60 * 8 cm obstacle block 2021, arXiv: 2103, 11470.
placed on the rigid body plane. The test of the granular surface is to place [5] A. Parness, N. Abcouwer, C. Fuller, et al. LEMUR 3: A limbed climb robot for
the montmorillonite particles with the size of 1.5 mm in the extreme terrain mobility in space, 2017 IEEE International Conference on Robotics
and Automation(ICRA), 2017, 5467–5473.
36 * 26 * 13 cm basin. The flexible surface is made of a rectangular
[6] J. Nicholson, J. Jasper, A. Kourchians, et al. LLAMA: Design and Control of an
metal frame covered with a layer of Artificial leather. The soft surface is Omnidirectional Human Mission Scale Quadrupedal Robot, 2020 IEEE/RSJ
made of a bag filled with water. By repeatedly placing SCRBP on these International Conference on Intelligent Robots and Systems (IROS), 2020,
3951–3958.
surfaces, the output signals are obtained to carry out the intelligent
[7] N. Rudin, H. Kolvenbach, V. Tsounis, M. Hutter, Cat-like jumping and landing of
identification with ML. legged robots in low gravity using deep reinforcement learning, IEEE Trans. Robot.
38 (2022) 317–328.
[8] M. Marc, B. Sebastian, K. Frank, Mantis - a robot with advanced locomotion and
CRediT authorship contribution statement manipulation abilities, Adv. Space Technol. Robot. Autom. (ASTRA) (2013) 5.
[9] K. Fondahl, D. Kuehn, F. Beinersdorf, et al. An adaptive sensor foot for a bipedal
Xuyan Hou: Conceptualization, Methodology, Validation, Data and quadrupedal robot, 2012 4th IEEE RAS & EMBS International Conference on
Biomedical Robotics and Biomechatronics (BioRob), (2012) 270–275.
curation, Formal analysis, Investigation, Visualization, Writing – orig­ [10] S. Chien, K.L. Wagstaff, Robotic space exploration agents, Sci. Robot. 2 (2017)
inal draft, Writing – review & editing, Funding acquisition. Li Zhang: 4831.
Conceptualization, Methodology, Validation, Data curation, Formal [11] T.F. Tang, X.Y. Hou, et al., Research on motion characteristics of space truss-
crawling robot, Int. J. Adv. Robot. Syst. 16 (2019) 1–17.
analysis, Investigation, Visualization, Writing – original draft, Writing –
[12] Yilin Su, Xuyan Hou, et al., Study on impact energy absorption and adhesion of
review & editing. Yilin Su: Conceptualization, Methodology, Valida­ biomimetic buffer system for space robots, Adv. Space Res. 65 (2020) 1353–1366.
tion, Data curation, Formal analysis, Investigation, Visualization, [13] B. Lindqvist, S. Karlsson, A. Koval, et al., Multimodality robotic systems: integrated
Writing – original draft, Writing – review & editing. Guowei Gao: Data combined legged-aerial mobility for subterranean search-and-rescue, Robot.
Auton. Syst. 154 (2022), 104134.
curation, Formal analysis, Investigation, Visualization, Writing – orig­ [14] H.K. Lee, J. Chung, S.I. Chang, E. Yoon, Real-time measurement of the three-axis
inal draft. Yuhui Liu: Data curation, Formal analysis, Investigation, contact force distribution using a flexible capacitive polymer tactile sensor,
Visualization, Writing – original draft. Zhonglai Na: Data curation, J. Micromech. Microeng. 21 (2011), 035010.
[15] N. Thanh-Vinh, N. Binh-Khiem, H. Takahashi, K. Matsumoto, I. Shimoyama, High
Formal analysis, Investigation, Visualization, Writing – original draft. sensitivity triaxial tactile sensor with elastic microstructures pressing on
QingZhang Xu: Data curation, Formal analysis, Investigation, Visuali­ piezoresistive cantilevers, Sens. Actuators A Phys. 215 (2014) 167–175.
zation, Writing – original draft. Tianxiang Ding: Data curation, Formal [16] X. Chen, J. Shao, H. Tian, X. Li, Y. Tian, C. Wang, Flexible three-axial tactile
sensors with microstructure-enhanced piezoelectric effect and specially-arranged
analysis, Investigation, Visualization, Writing – original draft. Li Xiao: piezoelectric arrays, Smart Mater. Struct. 27 (2018), 025018.
Conceptualization, Methodology, Validation, Investigation, Supervision [17] T. Miki, J. Lee, J. Hwangbo, et al., Learning robust perceptive locomotion for
Validation, Writing – review & editing. Long Li: Conceptualization, quadrupedal robots in the wild, Sci. Robot. 7 (2022) 2822.
[18] Z.L. Wang, J.H. Song, Piezoelectric nanogenerators based on zinc oxide nanowire
Methodology, Investigation, Writing – original draft, Writing – review & arrays, Science 312 (2006) 5771.
editing, Funding acquisition. Tao Chen: Conceptualization, Methodol­ [19] Shaoyu Liu, Yangyang Li, Wei Guo, Triboelectric nanogenerators enabled sensing
ogy, Investigation, Writing – original draft, Writing – review & editing, and actuation for robotics, Nano Energy 65 (2019), 104005.
[20] Chenguo Hu, Making light work with triboelectric energy conversion, J. Mater. Sci.
Funding acquisition, Project administration.
54 (2019) 8829–8830.
[21] T.P. Huynh, H. Haick, Autonomous flexible sensors for health monitoring, Adv.
Mater. 30 (2018), 1802337.
Declaration of Competing Interest [22] Iftach Shlomy, Shay Divald, Keshet Tadmor, Yael Leichtmann-Bardoogo, et al.,
Restoring tactile sensation using a triboelectric nanogenerator, ACS Nano 15
(2021) 11087–11098.
The authors declare that they have no known competing financial [23] Xuyan Hou, Minglu Zhu, Lining Sun, et al., Scalable self-attaching/assembling
interests or personal relationships that could have appeared to influence robotic cluster (S2A2RC) system enabled by triboelectric sensors for in-orbit
the work reported in this paper. spacecraft application, Nano Energy 93 (2022), 106894.
[24] Mingzhu Zhu, Mengying Xie, Xuanming Lu, et al., A soft robotic finger with self-
powered triboelectric curvature sensor based on multi-material 3D printing, Nano
Data Availability Energy 73 (2020), 104772.

All data needed to evaluate the conclusions are present in the paper.

11
X. Hou et al. Nano Energy 105 (2023) 108013

[25] Zuqing Yuan, Guozhen Shen, Caofeng Pan, Zhong Lin Wang, Flexible sliding sensor [39] C. Ottoni, W. Van Neer, B. De Cupere, et al., The palaeogenetics of cat dispersal in
for simultaneous monitoring deformation and displacement on a robotic hand/ the ancient world, Nat. Ecol. Evol. 0139 (2017) 1.
arm, Nano Energy 73 (2020), 104764. [40] B. Vollmerhaus, H. Roos, Principals of construction in the forepaw and hindpaw of
[26] Cheng Zhang, Shaoyu Liu, Xin Huang, A stretchable dual-mode sensor array for the domestic cat (Felis catus). III: musculature, Anat. Histol. Embryol. 30 (2001)
multifunctional robotic electronic skin, Nano Energy 62 (2019) 164–170. 89–105.
[27] Xianjie Pu, Hengyu Guo, Qian Tang, Jie Chen, et al., Rotation sensing and gesture [41] Park, Hangue, Latash, M. Elizaveta, Molkov, I. Yaroslav, et al., Cutaneous sensory
control of a robot joint via triboelectric quantization sensor, Nano Energy 54 feedback from paw pads affects lateral balance control during split-belt locomotion
(2018) 453–460. in the cat, J. Exp. Biol. 222 (2019), 198648.
[28] Tao Li, Jingdian Zou, Fei Xing, et al., From dual-mode triboelectric nanogenerator [42] Y. Mei, G.L. Wang, H.C. Zhou, et al., Investigation on the cushioning characteristic
to smart tactile sensor: a multiplexing design, ACS Nano 11 (2017) 3950–3956. of cat paw pad in motion: a synergy of its histological structure and kinematic
[29] Li. Gui Zhong, Wang, Gui-Gen, Ye, Da-Ming, et al., High-performance transparent mechanical characteristics, Jpn. J. Veterinary Res. 69 (2021) 19–30.
and flexible triboelectric nanogenerators based on PDMS-PTFE composite films, [43] X. Wu, B. Pei, Y. Pei, Y. Hao, et al., Comprehensive biomechanism of impact
Adv. Electron. Mater. 5 (2019), 1800846. resistance in the Cat’s paw pad, Biomed. Res. Int. 2019 (2019), 2183712.
[30] Hao-Yang Mi, Xin Jing, Qifeng Zheng, et al., High-performance flexible [44] D.S. Meng, X.Q. Wang, W.F. Xu, B. Liang, Space robots with flexible appendages:
triboelectric nanogenerator based on porous aerogels and electrospun nanofibers dynamic modeling, coupling measurement, and vibration suppression, J. Sound
for energy harvesting and sensitive self-powered sensing, Nano Energy 48 (2018) Vib. 396 (2017) 30–50.
327–336. [45] C. Wei, H.Y. Gu, Y.Q. Liu, et al., Attitude reactionless and vibration control in space
[31] Shan Yan, Jianwei Lu, Wei Song, Ru Xiao, Flexible triboelectric nanogenerator flexible robot grasping operation, Int. J. Adv. Robot. Syst. 15 (2018) 1.
based on cost-effective thermoplastic polymeric nanofiber membranes for body- [46] D. Ramesh, A. Kathail, D.E. Koditschek, F. Qian, Modulation of robot orientation
motion energy harvesting with high humidity-resistance, Nano Energy 48 (2018) via leg-obstacle contact positions, IEEE Robot. Autom. Lett. 5 (2020) 2054–2061.
248–255. [47] Pshikhopov, Viacheslav Kh, Medvedev, Mikhail Yu, Vasileva, Maria, Neural
[32] Z. Bai, Z. Zhang, J. Li, J. Guo, Textile-based triboelectric nanogenerators with high- network control system of motion of the robot in the environment with obstacles,
performance via optimized functional elastomer composited tribomaterials as Lect. Notes Comput. Sci. 11606 (2019) 173–181.
wearable power source, Nano Energy 65 (2019), 104012. [48] Arlene Coulson, An Atlas of Interpretative Radiographic Anatomy of the Dog & Cat,
[33] S. Chen, T. Huang, H. Zuo, et al., A single integrated 3D-printing process Wiley, 2002.
customizes elastic and sustainable triboelectric nanogenerators for wearable [49] Stanley H. Done, Color Atlas of Veterinary Anatomy: The Dog & Cat, Elsevier,
electronics, Adv. Funct. Mater. 28 (2018), 1805108. 2009.
[34] Mariello Massimo, Fachechi Luca, Guido Francesco, et al., Conformal, ultra-thin [50] M.G. José, C.G. Félix, Felipe Serna, et al. High-performance aromatic polyamides,
skin-contact-actuated hybrid piezo/triboelectric wearable sensor based on aln and Prog. Polym. Sci. 35 (2010) 623–686.
parylene-encapsulated elastomeric blend, Adv. Funct. Mater. 31 (2021), 2101047. [51] Y. Zhao, Progress on research and application of special polymer materials in
[35] X. Peng, K. Dong, C. Ning, et al., All-nanofiber self-powered skin-interfaced real- aerospace industry, Mater. China 32 (2013) 217–228.
time respiratory monitoring system for obstructive sleep apnea-hypopnea [52] X. Yin, D. Liu, L. Zhou, et al., Structure and dimension effects on the performance
syndrome diagnosing, Adv. Funct. Mater. 31 (2021), 2103559. of layered triboelectric nanogenerators in contact-separation mode, ACS Nano 13
[36] S. Lee, W. Ko, Y. Oh, et al., Triboelectric energy harvester based on wearable textile (2019) 698–705.
platforms employing various surface morphologies, Nano Energy 12 (2015) [53] Zhong Lin Wang, Aurelia Chi Wang, On the origin of contact-electrification, Mater.
410–418. Today 30 (2019) 34–51.
[37] F. Wen, Sun Zhongda, He Tianyiyi, et al., Machine learning glove using self- [54] Zhaoqi Liu, Shuyao Li, Shiquan Lin, et al., Crystallization-induced shift in a
powered conductive superhydrophobic triboelectric textile for gesture recognition triboelectric series and even polarity reversal for elastic triboelectric materials,
in VR/AR applications, Adv. Sci. 7 (2020), 2000261. Nano Lett. 22 (2022) 4074–4082.
[38] T. Jin, Zhongda Sun, Long Li, et al., Triboelectric nanogenerator sensors for soft
robotics aiming at digital twinapplications, Nat. Commun. 11 (2020) 5381.

12