Liu 2021

Robotics and Autonomous Systems 141 (2021) 103785
Contents lists available at ScienceDirect
Robotics and Autonomous Systems

journal homepage: www.elsevier.com/locate/robot
Review of snake robots in constrained environments✩

∗
Jindong Liu a , Yuchuang Tong b,c,d , Jinguo Liu b,c ,
a
Hamlyn Centre, Imperial College London, London, UK
b
State Key Laboratory of Robotics, Shenyang Institute of Automation (SIA), Chinese Academy of Sciences (CAS), PR China
c
Institutes for Robotics and Intelligent Manufacturing, Chinese Academy of Sciences (CAS), PR China
d
University of the Chinese Academy of Science, Beijing, PR China
article info a b s t r a c t
Article history: Snake robots have advantages of terrain adaptability over wheeled mobile robots and traditional
Received 30 September 2020 articulated robot arms because of their limbless thin body structure and high flexibility. They
Received in revised form 6 April 2021 have extensive applications in tasks such as rescue, disaster recovery, inspection and minimally
Accepted 7 April 2021
invasive surgery. Current research on snake robots is mainly focused on snake-like locomotion and
Available online 18 April 2021
the embodiment of these motion gaits for different applications. Modular structure and real-time
Keywords: control algorithms are two key aspects for snake robots operating in constrained environments. This
Snake robots review will attempt to address both. First, a review on the snake motion and the body structure
Joint structure is provided, which outlines the biological foundation of all snake robots. This is followed by the
Control algorithms mechanical structure of snake robots, especially the structure of elemental snake modules. Finally,
Constrained environment control algorithms for variant terrain contours and obstacle avoidance are discussed. The review also
outlines emerging application areas and potential future directions of snake robots.
© 2021 Elsevier B.V. All rights reserved.
1. Introduction terrain adaptation of snake robots. For example, the skeleton of

a snake consists of hundreds of vertebrae with limited motion
To overcome the difficulty of wheeled robots in rough terrain, of each vertebra. Snakes can make use of its body for both loco-
variable mechanical structures were proposed in the last few motion and manipulation, and this structure inspired a multiple-
decades. These include, for instance, tracked robots, air cush- joint body design seen in most snake robots. The anisotropic
ion robots [1], bipedal robots [2,3], multi-pedal robots [4,5] and friction of biological snakes is embodied by adding passive wheels
modular reconfigurable robots [6,7]. Although these robots have to the snake robot to reduce the tangential friction between the
a much wider terrain adaptability than a wheeled robot, there contact surface and the body of robots [13,14]. The central pattern
is no one that can fit all types of terrain. The superb mobility generator (CPG) is a neural network found in vertebrates, such
of snakes in various terrains including rocky mountain, desert, as biological snakes, that is responsible for generating rhythmic
water and trees, has attracted researchers to build snake robots patterns and periodic behaviours of animals, such as breathing,
that can mimic the adaptive motion and behaviour of a snake chewing or movement patterns [15,16]. The CPG is a coordi-
in a constrained unstructured environment [8]. Snake robots are nated mode in which the neural circuit only receives simple
low-dimensional input signals and produces high-dimensional
likely to make significant contributions in areas such as res-
rhythmic output signals [17,18]. The CPG neural oscillator in the
cue missions, medical procedures, firefighting and maintenance,
spinal cord stimulates muscles to stretch and contract, which
as the environment may be too narrow or too dangerous for
creates rhythmic body swings [15,16,19,20]. Due to its potential
humans [9–12].
in robotics, this biologically inspired method of motion control
Moreover, many biological mechanisms found in snakes have
has attracted increasing attention [21–23]. Therefore, the self-
been exploited in mechanical design, motor sensory control and
contained feature of CPGs has been used for distributed control
and fault tolerant adaptation in the undulation body motion of
✩ This work was supported in part by National Key R&D Program of China snake robots [21,24]. Some snakes have limited eyesight but
(Grant No. 2018YFB1304600), the Natural Science Foundation of China (Grant on infrared sensing for detecting prey and avoiding predator.
No. 51775541), CAS, PR China Interdisciplinary Innovation Team (Grant No. The underlying mechanism of such a motor-visual system has
JCTD-2018-11).
∗ Corresponding author at: State Key Laboratory of Robotics, Shenyang inspired snake robot researchers to control the robot with limited
Institute of Automation (SIA), Chinese Academy of Sciences (CAS), PR China. sensory information in constrained environments [25,26]. More-
E-mail addresses: [email protected] (J. Liu), [email protected] over, there exists a knowledge gap for snake robots to achieve
(Y. Tong), [email protected] (J. Liu). autonomous locomotion in the same way as snakes. Although
https://doi.org/10.1016/j.robot.2021.103785
0921-8890/© 2021 Elsevier B.V. All rights reserved.
J. Liu, Y. Tong and J. Liu Robotics and Autonomous Systems 141 (2021) 103785
snake robots are inspired by snakes, their motion should not be • Power efficiency: Despite friction against their movements,
restricted to snake-like motions. Therefore, variant locomotion it has proved that snake robots consume relatively small
gait should be specifically developed for snake robots [27,28]. amounts of energy. This can be explained by the fact that the
In this section, we briefly introduce the general background snake robots will not consume too much energy in moving
of snake robots, and summarize their characteristics and classifi- different appendages (such as wheels or legs) [42–44].
cation methods in detail, thus laying a foundation for the subse-
quent research on key technologies and development trends.
1.2. Disadvantages
1.1. Advantages
The most prominent features of snake robots are their adapt- The ultimate goal of many serpentine robots is to better im-
ability, versatility, scalability and robustness [29–31]. Compared itate the real snake in nature. Despite major research efforts in
to other robots, the specific advantages of snake robots are listed recent years, the performance of current generation snake robots
below: is still lagging far behind biological snakes. Snake robots also have
limitations including mechanical design, sensing technology and
• High adaptive ability to constrained environment: The snake
robot has a thin cross-section and a slender flexible body, intelligent control [45]. Compared to the hundreds of vertebrae in
which can help the body to move forward in very nar- biological snakes, snake robots can only have a limited number
row space with minimal interference to the surrounding of joints due to the torque limitation of motors. Snake robots
environment. It can easily adapt to various terrain by chang- cannot carry a heavy load due to their thin body and limited
ing its body shape, such as passing through narrow pipes, power of the single joint. Biological snakes can sense and lo-
small gaps between obstacles, and narrow surfaces [32– cate precise touching of its body to the environment. However,
34]. Moreover, since most snake robots are reconfigurable limited sensors can be equipped on snake robots because of the
modular robots, they can change their shape by rearranging space limitations and sensor miniaturization difficulties. Motion
the connection of the modules to adapt to new environ-
efficiency of snake robots is low because of their redundant multi-
ments or perform new tasks [35]. When a failure occurs,
joint structure. Dynamics modelling and locomotion control of
the system can still maintain some operational capabilities
and quickly detect and resolve problems. With the modu- snake robots are much harder than wheeled robots because their
lar implementation of functional and structural reconfigu- redundant structures. Therefore, it is still a big problem to adapt
ration, snake robots avoid many limitations of traditional a robot body to some complex environments based on onboard
fixed-configuration robots [36,37]. Furthermore, it has been sensors as the complexity of hyper-redundant body control. Some
shown in the literature that snakes can propel their bodies researchers in the past believed that it is difficult for the snake
to move efficiently by using terrain irregularity as scaffold- robot to achieve fully autonomous control because its thin body
ing [32–34,38–40]. Researchers have proposed a variety of can only bear a limited number of sensors. However, in recent
decentralized control methods to achieve autonomous con- years, researchers have proposed a variety of decentralized con-
trol of snake robots. It is shown that autonomous decentral- trol methods to achieve autonomous control of snake robots [32,
ized control mechanisms are the driving force for the devel-
41]. The simulated snake in [32] can automatically change its gait
opment of snake robots with highly adaptive functions [32,
pattern from obstacle-aided locomotion to concertina locomo-
41].
• Versatile ability with uniform body structures: By mimicking tion in response to the environment. Then, hardware realization
body feature of snakes, snake robots can be expected to have of [32] is described in [41] to prove the proposed decentralized
similar versatile abilities. Apart from the most well-known control mechanism. Meanwhile, due to the high surface area and
lateral undulation motion, snake robots can perform variant volume ratio, the temperature control is also difficult [46,47].
locomotion such as swimming, climbing and burrowing. The research field of contemporary robotics has evolved from
They can also manipulate other objects by wrapping their fixed-point operations in a structured environment to autonom -
bodies. These locomotion and manipulation abilities are all ous operations in an unstructured environment. Robots are ur-
realized by their limbless elongated bodies. gently needed to perform tasks in complex, highly dangerous
• Fault tolerance with redundant body structures: Most snake- and inaccessible environments. Snake robots can not only adapt
like robots have many joints that can form a body that is as
to various complex terrain, distribute their own weight equally,
flexible as a snake. Although this redundant structure brings
but also have the characteristics of self-closing structure, which
difficulties regarding motion control, snake robots benefit
from this in that they are tolerant to joint malfunction. has attracted many robotics researchers to carry out research on
When a small number of joints stop working, other joints snake robots. Compared to wheeled and legged mobile mecha-
can take over and maintain the motion of the whole body. nisms, snake robots offer high stability and good terrainability.
• Stable gaits for locomotion: Because the bodies of snake- The exterior can be completely sealed to keep dust and fluids out.
like robots often touch different grounds in many different Due to high redundancy and modularity, the snake robot is robust
places, their movements are very stable. In particular, a to mechanical failure. Despite many advantages, limbless snake
snake robot can keep its body close with the ground dur- robots are also limited by their own configuration. Compared
ing movement in a near 2-D plane, so that it has a lower with wheeled and legged robots, the disadvantages of snake
centre of mass. Moreover, the robot can lift more than half robots mainly include: (1) a large number of redundant degrees
of its body through large obstacles, such as branches or
of freedom bring difficulty to motion planning and control; (2)
rocks. Therefore, the purpose of stable obstacle avoidance
due to the limitations of the structure, the payload capacity is
is achieved. The snake robots have a lot of contact with the
ground during the movement. This large contact area makes limited, and the power efficiency is low; (3) compared with the
the snake robots have good traction properties in variable wheeled mobile robot, the movement velocity of the snake robot
environments [42]. A conventional wheeled or multipedal is slower; (4) due to its mechanical structure, it cannot carry
robot may slip, while the large contact of a snake robot may many sensors, so its analysis and comprehensive sensing ability
reduce the likelihood of this happening [43]. is limited [48,49].
2
Fig. 1. Classification methods of snake robots.
1.3. Classification methodology 2.1. Body structure of snakes
In the past few decades, researches on snake-robots and pub- The skeleton of a snake usually consists of vertebrae, ribs and
lished literature have increased significantly [42,48–50]. These skull (see Fig. 3(a)). The snake has 130 to 500 vertebrae and the
literature is a complete discussion of snake robots, with topics ribs are attached around the vertebrae to protect the internal
ranging from mathematical modelling techniques to mechatron- organs. The vertebrae forms a pillar along the body of the snake,
ics design and implementation to control design strategies. There- and the spinal cord passes through the passageway inside the
fore, they comprehensively evaluated the development prospects vertebrae [48]. The main function of the spinal cord is to transmit
nerve signals between the brain and other parts of the body. The
of various snake robots from various aspects, which are of great
role of the spine is to protect the spinal cord [51,52]. Adjacent
significance for the research on the classification of snake robots.
vertebrae are interconnected to each other and can undergo very
Specific classification methods of snake robots are as shown in
limited relative rotations at an angle of about 10 to 20 around
Fig. 1 in detail, which capture the key aspects of snake robots
the vertical axis and a few degrees around the horizontal axis.
including working environment, locomotion, actuator and joint
However, the sum of all these small angular rotations along the
design, and control methods.
snake body gives it great flexibility in motion [53].
In this survey, we will review the current snake robots ac-
It has been shown in the literature that the musculoskeletal
cording to their ability in working in a constrained environment system of snakes is extremely complex, and there are several
including locomotion styles, joint structures and control methods. muscles in the semispinal-spinal muscles of snakes that span
The remainder of the paper is organized as following. In Section 2, dozens of vertebrae [54–56]. The muscles are arranged diagonally
a brief introduction of snake motion is addressed and the basic along the sides of the snake body, and their ends are attached
principles of snake movement are explained. Furthermore, loco- to the ribs. These muscles are usually connected to distant ribs,
motion gaits and gait control of snake robots are also studied and some are connected to adjacent ribs. The snake uses the
inspired by the locomotion of biological snakes. In Section 3, expansion and contraction of muscle to make periodic change in
the joint structures of snake robots are discussed. Furthermore, body shape, thereby promoting the movement of the snake [48].
different actuators and joint forms of snake robot are introduced Land snakes push themselves through a variety of techniques, in-
in detail. In Section 4, terrain adaptive control methods and cluding lateral undulation through the body, progression through
detailed modelling and control methods are introduced. Section 5 the unilateral contraction/stretching of the abdomen, folding of
will introduce a wide range of application fields of snake robots the body as a concertina for concertina-like motion, and putting
and this is followed by Section 6, which outlines the future devel- the body into a series of spirals [53]. Kano et al. [54] clarified
opment trends and challenges. Section 7 presents the conclusion the mechanism of multi-articular muscles in the movement of
of the paper. biological snakes. Inoue et al. [55] aimed to reveal the under-
lying mechanisms of adaptation of biological snakes by using a
constructive approach. Moreover, biological system models that
2. Snake motion and body structure
reflect biological knowledge are used as tools to analyse the
entire system. Astley et al. [56] used techniques in engineering
In this section, we will briefly introduce the body structure static systems analysis to analyse the consequences of multi-
of the snakes and the classification and features of the snake articular muscle configuration in cantilever gap bridging. These
motions, and explain the biological principle for these motions. results are in good agreement with the biological structure data of
Furthermore, locomotion gaits and gait control of snake robots snakes, and have the potential to greatly improve the mechanics
are also studied inspired by the locomotion of biological snakes. of snake robots.
3
Fig. 2. (a) Lateral undulation [50], (b) concertina gait [50], (c) Sidewinding gait [50], (d) Rectilinear gait (above) and tracks (bottom) [57].
2.2. Locomotion gaits of snakes an obstacle such as a stick or rocks, snake will partly deform the
body nearest that object and exerts force against it. Each of these
Snakes change their locomotion patterns in response to the environmental objects generates a reaction force directed for-
environment. This ability is the driving force for the development ward and towards the midline of the snake, resulting in forward
of snake robots with highly adaptive functions, and autonomous thrust while the lateral components cancel out [70]. Therefore,
decentralized control mechanisms may be the key to under- snake not only avoids the obstacle but also utilizes the counter
standing this movement [32–34,38–40,58–60]. Research on the force to help the forward propagation [71–73]. In this mode of
biological mechanism and propulsion mode in the movement locomotion, obstacles are considered to be moving objects that
of biological snakes laid the foundation for the research on the can help snakes move forward. This movement is defined as
control and structural design of snake robots. obstacle-aided locomotion [74,75]. The study by Heckrotte et al.
Gray et al. [61] initially observed the locomotion of common [76] found that the speed of this movement depends upon the
grass snake (Tropidonotusnatrix) and classified the terrestrial density of push-points in the environment. Moreover, the gait
motion of the snake into four different gaits. Jayne et al. [62–64] of lateral undulation is most efficient in areas of rough ground.
used electromyography (EMG) and cinematography to determine Shorter snakes are unable to align with the desired curve, and
the muscle activity of biological snakes in lateral undulating the heavy pressure will greatly reduce their efficiency, so they
motion on land and water, as well as concertina movement, are not suitable for lateral undulation. The corridors which are
and comprehensively evaluated the locomotion performance of smooth, low-friction surfaces and narrow are also unsuitable for
biological snakes in various movement gaits, which were then this gait.
applied to the design of snake robots. Moon et al. [65] combined
with previous studies on the physical principles and muscle ac- 2.2.2. Concertina
tivity patterns of the lateral undulating gait of snakes, and further Concertina gait is that the parts of the body are stationary,
explored the mechanism of muscle activity producing curvature and the other parts move forward (as shown in Fig. 2(b)). This
and propulsion. Marvi et al. [66–68] respectively studied the movement is repeated periodically to move the body forward.
concertina movement gait and the rectilinear movement gait of This gait is typically applied to snakes that cannot use full range
biological snakes, as well as the body transformation rule and of other gaits through narrow passages such as tunnels [62,64].
control mechanism of its movement on sandy slopes. They specif- As shown in Fig. 2, the torso in front of each contact portion
ically explained the biological mechanism of each movement gait is straightened while being laid down in a curved manner at
of snakes, and discussed and studied its application and inspi- the rear end of the portion. In this process, muscle tissue at or
ration in snake robots. Recently, Gong et al. [69] suggested that near the moving part of the torso needs to be activated. The key
studying how biological snakes coordinate whole-body motion element of the accordion movement is the difference between
to achieve agile behaviours can help improve the performance the larger friction generated by the coefficient of static friction in
of snake robots. They proposed the hypothesis that replaying different directions with the body’s movement direction and the
parameterized shape trajectory data collected from biological smaller force generated by the kinetic coefficient of motion along
snakes can generate equivalent motions in snake robots for snake the body part [66]. Actually, due to momentum changes, static
locomotion that is approximately kinematic. friction and slower speed, concertina locomotion is a relatively
These literature specifically analysed the structure mecha- inefficient mode of movement. However, concertina locomotion
nism and movement mechanism of the various gaits of biological allows biological snakes to move in environments where other
snakes, and discussed in detail that the various gaits of biological locomotion modes cannot advance, such as along pipes, wires
snakes are of great significance to the research of various aspects and cables, and branches. Jayne et al. [77] believed that the
of snake robots. As shown in Fig. 2, various motion modes of the classification of the four moving walks of snakes in the past was
snakes will be described in detail. too conservative, so the concertina locomotion of snakes was
further divided into four types in detail, including: (1) flat-surface
2.2.1. Lateral undulation concertina, (2) tunnel concertina, (3) arboreal concertina with al-
Lateral undulation is the most common gait of terrestrial lo- ternate bends, and (4) arboreal concertina with helical wrapping.
comotion. Moreover, it is also the only form of biological snake Therefore, in some environments, this locomotion mode is easier
motion that does not use static contacts between the snakes to achieve than other locomotion gaits [61,78–80].
and medium [63,65]. In lateral undulation, the entire body of
the snake forms a continuous sinuous movement. At least three 2.2.3. Crotaline/sidewinding
contact points between the snakes and the medium are needed The sidewinding motion (as shown in Fig. 2(c)) is a continuous
to move on: two for force generation and the third for balancing transverse bending oscillation. Since there are only two contact
forces to move in a particular direction. The scales on snake’s skin surfaces with the environment during movement without addi-
contact with the terrain and exploit the consequent friction to tional forces in other directions, this gait can minimize slippage
obtain propagation force (see Fig. 2(a)). As the body contacts with compared to the lateral undulation [62,68]. The researchers have
4
Fig. 3. (a) The skeleton of a snake [48]. (b) The skin of a snakes is completely covered by scales [48]. (c) Dynamic load distribution of snakes during lateral
fluctuations [53].
observed that some flanking snakes can also use this gait to travel change and movement mechanism of the rectilinear locomotion
kilometre-length distances continuously [81]. Marvi et al. [68,82] gait of snakes, and explained the biological mechanism of this
used 3-D high-speed video to characterize the body changes and locomotion gait in detail, and then discussed its application and
movement mechanisms of rattlesnakes as they moved across a inspiration in snake robots. It is indicated that the rectilinear
bed of particles. It further suggests that desert snakes such as rat- locomotion of biological snakes provides propulsion motion by
tlesnakes push themselves towards the sand through sidewinding reciprocating or stretching muscles from the ribs attached to
locomotion, which is a way of movement that relies on spiral elastic skin [89,90]. In rectilinear locomotion, several parts of the
travelling waves. They specifically explained the biological mech- body are in contact with the ground at any moment, and the
anism of sidewinding locomotion gait of snakes and discussed gait uses a symmetrical form rather than a staggered contraction
and studied its application and inspiration in snake robots. Ast- wave. Part of the abdominal skin is pulled forward, so the abdom-
ley et al. [56,83,84] explained that the sidewinding locomotion inal scales grow in bundles. The forwardly drawn portion of the
commonly associated with desert snakes can be represented as body is then pressed downward, and the ventral edge is bonded
a pair of body waves orthogonal in both vertical and horizontal to the surface. The body then slides forward through the skin until
directions, which are independently modulated for high manoeu- it is properly aligned with the surface skin, and the movement
vrability and oblique ascent. Tingle et al. [85] suggested that repeats periodically. Rectilinear locomotion requires only a small
sidewinding is a locomotion gait that only works for a few very vertical movement [91,92].
specific snake species, most of which are venomous snakes from
desert environments. Therefore, it is indicated that the sidewind- 2.2.5. Other gaits
ing locomotion of biological snakes is mainly used by snakes in Snakes also have gaits used in special situations or by certain
desert areas composed of loose soil and sand [68,85,86]. The de- species such as rectilinear crawling, burrowing, jumping, sinus-
velopment and widespread use of the sidewinding motion may be lifting, skidding, swimming, and climbing [50]. Among them, the
related to the need for traction on low shear surfaces (e.g. sand) change of the lateral undulation of the biological snake is called
and to avoid the high temperatures of desert topography. As skidding (also called slidepushing), which is used when passing
shown in Fig. 2(c), the sidewinding movement can be thought through a low-friction surface. In this locomotion gait, the snake
of as a stripping of the body from one track to the next, where rests its head on the ground and sends out bending waves to
the track or line shows the rolling of the body contact during the the body, thus it is a very energy-saving locomotion. Moreover,
movement [87]. the snake can also climb trees through vertical lateral undulation.
In this movement, part of the body of the snake hangs freely in
2.2.4. Rectilinear the air, while the branches are used as support. Ariizumi et al.
Rectilinear motion travels forward by moving the skin of the [110] found that sinus-lifting motion and sidewinding locomotion
body along the direction of motion relative to the bone, which are generally more energy-efficient gaits than is lateral undu-
is a slower peristaltic motion that uses traction in the abdomen lation. Sinus-lifting is an improvement to lateral undulation, in
to provide traction. Researchers have speculated that the lin- which lifting a part of the trunk can avoid lateral slippage and
ear motion is due to rib walking which leading to the active optimize propulsion. This gait is used for high speed movement.
movement of the ribs initially (as shown in Fig. 2(d)). However, More specifically, if the anisotropy in friction is large enough,
this theory has finally proved to be wrong by X-ray observation sinus-lifting motion is the most energy-efficient gait. Toyoshima
of snakes in motion [67,88]. Lissmann et al. [88] described the et al. [111,112] introduced the optimization study of this gait
rectilinear locomotion patterns of biological snakes and discussed in detail, and proposed a locomotion mode suitable for snake
the possible interactions of muscle tissue derived from the me- robots. The researchers also found that most snakes can swim and
chanics and geometry of tissues. Marvi et al. [67] studied the body move forward in the water by undulating laterally like eels [113].
5
Table 1
Joint and motion feature of snake robots.
Snake robot Actuator type/Body structure Single/ Robot motions Operating environment
Total DOF
Modsnake [93] DC motors/(4) 1/16 3D snake gait, climbing, swimming, sidewinding Variant rough terrain
AmphiBot I/II [42] DC motors/(1) 1/8 2D gaits, lateral undulation Smooth surface
PolyBot [29,45] DC motors/(4) 1/32a 3D gaits, rolling, caterpillar turning Variant rough terrain
ACM III [13] DC motors/(1) 1/20 2D gaits, lateral undulation Smooth surface, tunnels
Bayraktaroglu’s snake DC motors/(4) 1/10 2D gaits, lateral undulation, rectilinear Uneven surface, tunnels
robot [94]
ACM-R4 [95] DC motors/(2) 2/9b 3D gaits, climbing, sharp turning Rough terrain
SMA [14] DC motors/(1) 2/9 2D gaits with head lifting, lateral undulation Slope surface
Aiko [96] DC motors/(4) 2/10 2D gaits, lateral undulation 2D surface, loose obstacle
ACM-R5 [97] DC motors, spur gear box/(1) 2/9c 3D gaits, lateral undulation and swimming Rough terrain, water
Snoopy [98] DC motor, bevel and worm gears 2/6 3D manipulation on a mobile robot 3D space
Kulko [99] DC motors, worm gears/(4) 2/10 2D gaits, sidewinding, rolling Smooth surface
Brunete’s micro-robots DC motors, spur gear box/(4) 2/6 Pipe-inspection Pipes
[100]
Medusa [101] DC motors, swivel, bevel gears 3/4 3D manipulation on stationary platform 3D space
GMD-snake2 [102] DC motors/(2) 3/6d 2D gaits with lifting head, Lateral undulation 2D surface
Perambulator-II [103] DC motors, worm gears/(4) 3/7 Rolling, lateral undulation Surface of 3D terrain
SSR [58] Pneumatic/(5) 3/6e Lateral undulation, rectilinear, rolling Smooth surface
Liljeback’s snake robot Pneumatic/(4) 2/6 2D gaits, lateral undulation, sidewinding Smooth surface
[99]
OmniTread-8 Hybrid of DC motors and 2/5 3D gaits, passing holes, crossing gaps, Variant rough terrain
pneumatic joint/(3) rectilinear gait
OmniTread-4 [104] Hybrid of DC motors and 2/7 3D gaits, passing holes, crossing gaps, Variant rough terrain
pneumatic joint/(3) rectilinear gait
f
MOIRA [105]/ MOIRA-II Hybrid of DC motors and 3/4 3D gaits, passing holes, crossing gaps, Variant rough terrain
[106] pneumatic joint/(3) rectilinear gait
CT ARM-1 [107] Tendons 1/7 3D manipulation on stationary platforms Industrial workspace
DDU [108] Tendons 2/2 3D manipulation on stationary platforms Minimally Invasive Surgery
R7 [109] Tendons 2/4 3D manipulation Vision inspection at aircraft
ACM-L2 [35] Tendons 2/9 2D gaits, lateral undulation, sinus-lifting Smooth 2D surface
a
Maximum.
b
1 DOF for bending joint and 1 DOF for active wheel.
c
Typical.
d
2 DOF pitch/yaw, 1 DOF active wheel.
e
2 DOF pitch/yaw, 1 DOF linear expansion.
f
2 DOF pitch/yaw, 1 DOF forward propulsion.
Furthermore, other special patterns are used by some Asian tree promotes movement [119,120]. In [53], experimental studies on
snakes, which glide through the open chest to form a sliding sur- the frictional properties of biological snake skins have shown that
face to motion [114,115]. Flying snakes are special Chrysopelea the coefficient of friction along the tangential direction of the
paradisi species, which can glide without limbs. In the recent body is much smaller than the coefficient of friction in the normal
past, modelling the flight of these snakes has attracted consid- direction. In the movement of a snake, it is shown that the snake
erable attention, and experimental studies have been conducted affected by isotropic friction is uncontrollable [48]. Therefore,
on live flying snakes to understand their abnormal glide mecha- we consider this anisotropic frictional characteristic as the basic
nism [116,117]. Yeaton et al. [118] studied the aerial undulation assumption of snake-shaped robot control design. Experiments
locomotion gait of flying snakes and used high-speed motion cap- on body velocity prediction have shown good agreement with
ture to measure their gliding. Through the reconstruction of the the theoretical hypothesis and proved that snake propulsion on
body that acts like a wing, it is found that the aerial undulation flat ground is mainly dependent on frictional anisotropy of their
gait of flying snakes is composed of horizontal and vertical waves, scale [48,121,122]. The geometry and topological layout of snakes
with phase difference of 90◦ and frequency difference of two advances mitigation of frictional effects both in locomotion and
times. It has been shown that the aerial undulation of snakes acts in general function. Abdel et al. [123] assessed available data
differently from other known animal undulations and provides a on surface metrology and tribological behaviour of snakes and
new control template for dynamic flying robots. defined aspects of that performance. Chango et al. [124] evaluated
the benefits the unique frictional properties conferred by snakes
2.3. Biology principle of snake motion confer to robotic snake locomotion in experimental trials em-
ploying rectilinear, lateral undulation, and sidewinding gaits over
According to the study, the formation principle of snake mo- several distinct surface types: carpet, inhomogeneous concrete
tion is mainly composed of two main reasons: friction anisotropy and homogeneous concrete.
and ground/terrain contact points and weight distribution. These Moreover, in the movement of biological snakes, skin scales
two aspects of the movement of biological snakes will be de- of snakes can use the principle of friction enhancement to make
scribed in detail as follows. them move forward. Marvi et al. [66] measured the friction prop-
erties of biological snakes and found that they use ingenious
2.3.1. Friction anisotropy and friction enhancement friction enhancement techniques to double their coefficient of
The snake body is completely covered by scales, see Fig. 3(b). friction, thus increasing the stability of their motion.
When the snake moves in a rough terrain, the extra corporeal
scales protect the skin from cuts and tears [48]. 2.3.2. Ground/terrain contact points and weight distribution
Skin scales of snakes play a fundamental role in providing In addition to the anisotropic friction coefficients of the snake
anisotropic friction fundamental during movement, which in turn skin, the researchers also proposed a theoretical model of gravity
6
Fig. 4. Snake robot examples for whole body structure classification. (a) ACM-R3 [125], (b) Modsnake [93], (c) OmniTread OT-4 [104], (d) Slim Slime robot [126],
(e) GMD-Snake2 [102].
adjustment to explain the sliding motion of the snake. In [53], environment [135–137]. Enner et al. [138] proposed a novel com-
it is emphasized that the importance of weight distribution in patible controller to estimate the overall state of the robot based
lateral undulating motion, as shown in Fig. 3(c). The ability to on gait parameters. The controller then commands a new gait
redistribute ground/terrain contact points and weight distribu- parameter relative to the estimated state. Kelasidi et al. [139,140]
tion is clearly important when the appendage moves forward in considered both the minimization of the power consumption and
limb movement, with a broader generality, such as its ability to the maximization of the achieved forward velocity in order to in-
improve the movement of the extremity in motion [127]. vestigate the optimal gait parameters for snake robots. Guo et al.
[27,28] proposed a novel guided motion planning method for
2.4. Gaits and gait control of snake robots snake robots. The serpentine gait can be generated by calculating
the connection of the fibre bundle. Compared with the traditional
In recent years, gaits and gait control of snake robots have gait generation method, such as the curve-based method and the
been the focus of research. Researchers have conducted a lot of CPG-based method, this proposed method does not assume that
research on these two aspects. the serpentine gait is a standard sinusoidal signal. Hwang et al.
Researchers have proposed novel gaits for different applica- [141] proposed a novel generative motion model of snake robots
tions of snake robots. Dear et al. [128,129] described the emer- with hierarchical structure representation topology. Alben et al.
gence of a novel gait for snake robots that incorporates the [142] proposed a robust iterative numerical method to study the
system’s singular configurations to produce high locomotive effi- efficiency of a wide range of motions under isotropic Coulomb
friction and found that simple undulatory motions give little net
ciency without incurring unbounded constraint forces. Cao et al.
locomotion in the isotropic regime. Then, they proposed a class of
[130] presented a novel mechanism for slithering locomotion of
smooth body motions that have similarities to concertina locomo-
a soft snake robot. There are two key factors in this slithering
tion (including the involvement of static friction) and can achieve
locomotion: a rotational velocity field and a corresponding local
optimal efficiency for both isotropic and anisotropic friction. Bing
contact region between the beam and ground. Yaqub et al. [131]
et al. [143] presented a smooth slithering gait transition control
proposed a spiral curve gait along with rolling motion to adapt to
based on a lightweight CPG model for snake robots. Manzoor et al.
the changes in the pipe diameter while moving on the pipe. More-
[144,145] proposed a unified neural oscillator model as the CPG
over, due to its special body shape and application requirements,
to achieve different types of locomotion in a snake robot.
the snake robot also has a special movement mode, namely
obstacle-aided locomotion, which was first proposed in [132]. In
3. Mechanical design of snake robot
obstacle-aided locomotion, it has been shown that snakes can
propel their bodies to move efficiently by using terrain irregu- For decades, design and development of snake robots have
larity as scaffolding [32–34,38]. Transeth et al. [132] found that been research hotspots, and researchers from various countries
biological snakes exploit the irregularities of terrain to achieve have designed many novel snake robots. The first recorded snake
faster and more efficient movement. Wang et al. [133,134] pro- robot was introduced by Hirose et al. [36] in 1972. The robot was
posed two novel obstacle-aided locomotion gaits for snake robots, 2 metres long with 20 rotating single degree-of-freedom (DOF)
namely, obstacle-aided concertina gait and obstacle-aided gait joints. It could only be operated on a smooth surface by rotating
through narrow corners. its joints to the left and right to generate forward motion. Later,
Furthermore, it can be shown that the ability in obstacle- Hirose et al. [148] have developed over 10 different types of
aided locomotion is the driving force for the development of snake robots. Some robots were equipped with passive wheels
snake robots with highly adaptive functions, and autonomous around joints, such as ACM-III and ACM-R5 (Fig. 4(a)), while some
decentralized control mechanisms may be the key to understand- robots employed active wheels as the main actuators, such as
ing this movement [32–34,38]. Kano et al. [32–34] developed a ACM-R4. Hirose et al. also applied pneumatically-driven joints
control scheme for snake robots, which exhibits an autonomous to build a snake robot called Slim Slime, which mimics the con-
gait transition. Date et al. [38] proposed a motion control method certina locomotion gait of a snake. Apart from Japan, researchers
for a snake robot inspired by a biological snake, in which the in other countries also developed many snake robots [125,126].
snake robot was supported by obstacles and moved adaptively Borenstein et al. [146] [104] of the University of Michigan took
over a rugged terrain. The research on decentralized control of advantage of active tank treads and pneumatic bellows in de-
snake robots lays the foundation for the development of robots signing joints of snake robots, i.e., OmniTread OT-8 (Fig. 4(b))
that work in uncertain environments and helps to understand the and OmniTread OT-4. These robots were featured with a high
basic principles of animal adaptive motion. propulsion ratio and conformability to different terrain. Another
Many other control methods for locomotion gaits have been major research groups are led by Choset of the Carnegie Mel-
proposed to enable snake robots to be highly adaptive to the lon University. They focused on elaborating joint structures and
7
Fig. 5. Representative snake robots. (a) ACM-R5 robot [97], (b) OmniTread OT-8 robot [146] (c) GreaseFire robot [147].
built several generations of snake robots such as Snoopy [101], Then, we will classify the modular structures according to the
Schmoopie [98] and Modsnake [147] [93]. The latest snake robot DOF and discuss module structures of snake robots.
of the Modsnake series is called Greasefire (Fig. 4(c)) which
demonstrated snake-like 3D locomotion such as crawling, climb- • 1-DOF module: Due to the limited space in this review, we
ing and swimming. Ahmed et al. [149] proposed a flexible snake will only take two examples, Modsnake [93] and Amphi-
robot which has been built using smart material, such as elec- Bot I [152], to explain the design concept of 1-DOF snake
troactive polymers, which can be activated by applying power robot modules. There are many other examples such as
to it. Koopaee et al. [150] designed a wheel-less, modular snake S1 robot designed by Miller et al. [158], modular robot-
robot with series elastic actuators to reliably measure motor PolyBot designed by Yim et al. [45] and ACM III designed by
torque signal. Racioppo et al. [151] presented the design and con- Hirose et al. [125]. Fig. 6(a) shows the CAD drawing of one
trol of a cable-actuated mobile snake robot composed of modular Modsnake robot module which was made with a modified
coupled linkages. Hitec HS-5955TG servo motor. The original servo bottom
In this section, we will introduce the module structures for of HS-5955TG was replaced with a servo back in Fig. 6(a)
each actuation technology and compare them regarding to their so that it allowed extra space for a customized circuit and
performances. specially designed bearing system. Fig. 6(b) illustrates the
module structure of AmphiBot-I snake robot [152]. An ex-
3.1. Modular structures of snake robots ample of two connected modules is shown in Fig. 6. Each
module’s axis of rotation was perpendicular to the operating
In general, snake robots are designed to consist of at least surface and a pair of passive wheels was equipped under
three similar or identical modules which are chained together. each module.
Table 1 listed the joint and motion feature of most snake robot For 1-DOF modules, many planar snake robots were de-
in the world. Snake robots are categorized through certain ba- signed by connecting all modules with all rotation axes
sic properties, including: actuator types, DOF of each module, in parallel. Passive wheels were commonly equipped un-
locomotion types, and operating environment. Regarding whole der modules to reduce the tangential friction. The advan-
body architecture, snake robots are classified into five categories tage of such design is that dynamics modelling and motion
of body structure [42] and summarized in Fig. 5 and Table 1: control are well studied and easy to implement [159,160].
(1) snake robots with passive wheels, such as ACM-R3, ACM-R5, Autonomous planar snake robots [161,162] have been de-
AmphiBot I and AmphiBot II;(2) snake robots with active wheels, veloped for obstacle avoidance and path planning. Although
such as GMD-Snake2, Koryu-II (KR-II) , ACM-R4 and NUTA’s snake planar snake robots can perform some terrain adaptive be-
robots; (3) snake robots with active treads, such as OT-4, Koryu-I, haviour such as going through a tunnel or moving up to a
OmniTread OT-8 and JL-I; (4) snake robots based on sidewind- slope, such snake robots have limited applications in the
ing and inchworm crawling like gait, for instance: Modsnake, real world due to the lack of 3-dimensional motion. Al-
Inchworm robot, snake robots designed by Dowling, PolyBot, M- though 1-DOF modules only have one rotational degree-
TRAN, CONRO and GMD-Snake; and (5) snake robots based on of-freedom, an assembled snake robot can still have 3D
undulation using linear expansion, for instance: Slim Slime Robot, locomotion by arranging the axis of modules to perpendic-
Planar Inchworm robot and Telecubes. ular to each other. An U-shaped frame was normally used
8
Fig. 6. (a) CAD structure of Modsnakejoint [93] (b) Module structure of AmphiBot-I [152]. Left: two connected joints. Right: internal structure of one module. (c)
Actuated universal joint design of Snoopy snake robot [153]. (d) Module of ACM-R5 snake robot [154]. (e) Shammas and Choset’s snake-like manipulator [101]. Left:
assembled manipulator; right: schematic joint structure. (f) Joint structure of Perambulator-II snake robot [155]. (g) Mechanical design of Choset’s angular bevel
joint [156]. (h) Mechanical structure of Slim Slime Robot [95]. (i) Module structure of OmniTread-4 [146]. Left: the internal module structure; Middle: DC Motor and
gear box [157]; right: illustration of pneumatic actuator driving joint.
to hold the motor and to connect two modules together, on Takanashi’s angular swivel joints. Shammas has a good
such as ModSnake [93]. With this design, the snake robot historical view on swivel joints in [153]. Recently, Liljeback
can easily fit its body into variant rough terrain such as et al. [164] proposed an interesting snake robot, Kulko,
a rocky field. Many terrain adaptive behaviours have been which was built by a number of spherical 2-DOF modules.
realized such as climbing trees and the sidewinding gait for Each module consisted of two intersecting half-circle rings
obstacle avoidance. However, many modules are required which are driven by servo motors via worm gear and chain
to have high body flexibility and it is hard to lift many drive. The ACM-R4 [95] snake robot proposed by Yamada
modules. The snake robot modelling and control algorithm and Hirose was built by 2-DOF modules which used one
are much more complex than planar snake robot. For in- motor for bending and another motor for driving a pair of
stance, Chirikjian’s backbone curve [163] is widely used for wheels for propulsion. Fig. 6(d) shows the schematic of the
the kinematic model and Liljeback et al. [164] borrowed modular structure of the ACM-R5 snake robot. The ACM-R5
the dynamics model of classical redundant manipulators for has demonstrated impressive 3D snake motions including
a mobile snake robot. In the past, some researchers have swimming underwater.
The 2-DOF motor-based structure is the second most pop-
argued that due to the complexity of the control, there were
ular structure. Two DC motors are fixed into one module
no autonomous 3D snake robots at that time. However, in
to provide two degrees of freedom. Two types of 2-DOF
recent years, researchers have proposed a variety of decen-
module were reviewed: (1) stacking two revolute motors
tralized control methods to achieve autonomous control of
orthogonally, and (2) parallel arranging two motors with the
snake robots. It is shown that autonomous decentralized
help of worm gears, angular swivel gears or bevel gears. The
control mechanisms are the driving force for the develop-
first solution has simple structure but bulky body while the
ment of snake robots with highly adaptive functions [32, second solution is compact but complex gear box and joint
41]. mechanism. Regarding to modelling and terrain adaptability,
• 2-DOF module: There are mainly two types of mechanical these snake robots are like 3D snake robots with 1-DOF
design of 2-DOF snake robot module: (1) stacking two rev- modules.
olute motors orthogonally, and (2) parallel arranging two • 3-DOF module:
motors with the help of worm gears, angular swivel gears Compared to 1-DOF modules, the 3-DOF structure is more
or bevel gears. Many other snake robots consist of 2-DOF compact so that the snake robot can be shorter [101–103].
modules. For example, Snoopy snake robot designed by Due to the complexity of implementing three rotation axes
Wolf et al. [98] applied two DC motors to build a univer- in one compact module, there are only two of the most
sal actuated joint (see Fig. 6(c)). SHR robot [165] in the typical snake robots which have 3-DOF modules, includ-
Jet Propulsion Laboratory (JPL) and anthropomorphic robot ing: (1) Shammas and Choset’s snake-like hyper-redundant
finger designed by Choi et al. [166] were designed based manipulator, i.e., Medusa [101] (see Fig. 6(e)) and (2) Ye
9
et al. [103] reconfigurable snake robot, i.e., Perambulator-II • Fluid actuator: The main type of fluid actuator in snake
(See Fig. 6(f)). The predecessor of the manipulator is a 2- robots is the pneumatic actuator. The pneumatic actuator
DOF angular swivel joint designed by Choset et al. [156,167] was the type of actuator used in early snake robots, i.e., Slim
(see Fig. 6(g)). In their latest design [101], one more motor Slime robot (SSR) by Ohno et al. [58]. Recently, Liljeback
was added to provide three DOFs, i.e., twisting, and all et al. proposed a modular pneumatic snake robot [132]. Qi
gears were redesigned to use standard off-the-shelf gears. et al. [171] proposed a novel pneumatic soft snake robot that
Ma et al. [168] proposed a 3-DOF snake robot module. Ye uses a travelling wave motion to move in a complex and
et al. [103] replaced the DC motors with FutabaS3305 servo narrow environment. Similar to the structure of pneumatic
motors which were more powerful and accurate. The basic actuators, the hydrodynamic actuator has also been used for
mechanical structures of two modules were similar. The snake robot design. For example, Date et al. [172] proposed
module was called Modular Universal Unit (MUU) and was a snake-like propulsion mechanism driven and controlled
intended to create a reconfigurable snake robot. by fluids. Fig. 6(h) shows the mechanical structure of one
SSR module. Inside a module, three identical metal bellows
Moreover, a pneumatic module can have 2-DOF or 3-DOF were arranged parallel to each other with even spacing. The
motion and the assembled snake robot is able to move in 3D SSR (see Fig. 6(h)) is composed of six modules in series and
space. It has advantages of high force transmission, low cost was successfully able to perform locomotion in an inclined
and linear expansion motion. However, pneumatic modules lack pipe. An upgrade of SSR, SSR2, SSR3, was proposed by Aoki
accurate control and the joint bending angle is limited. Although et al. in 2004 [173], who designed a bridle bellows for
pneumatic snake robots can perform unique concertina snake gait each module in order to produce higher torque. Moreover,
using linear expansion ability, it has low potential for constrained Zhang et al. [174] studied the adaptive stiffness control
environments due to slow propulsion speed and limited 3D mo- and motion optimization of a snake robot with a variable
tion. To overcome these problems, hybrid modules of pneumatic stiffness actuator, and changed its stiffness by controlling
and motor were applied on snake robots in which motors provide the magnetorheological fluid around the actuator.
a propulsion force and pneumatic joints control motion direction. • Hybrid actuator: The main type of hybrid actuator on snake
Successful applications in variant circumstances, such as crossing robots is a combination of DC motors with pneumatic actu-
a gap and moving over cluttered earthquake ruins, have been ators. In each module, a DC motor drives a number of tracks
tested, e.g. OmniTread-R4 [104]. to provide forward propulsion and a pneumatic actuator
connects two modules to control the connecting stiffness
3.2. Actuators of snake robots and bending angle between them. Two research groups
designed hybrid actuators and implemented on snake robot.
Modular structures of snake robots are highly depended on Borenstein et al. [104,146] developed OmniTread-8 and
the type of actuators. In this subsection, we will introduce some OmniTread-4. Osuka et al. [105] at Kyoto University devel-
actuator types and actuation techniques and compare them with oped MOIRA and MOIRA-II. OmniTread and MOIRA robots
the performance. share a similar modular structure. We will take OmniTread
as an example to discuss [106]. Fig. 6(i) shows the modular
3.2.1. DC motors structure of OmniTread-4 (as shown in Fig. 5 for the as-
The most dominating actuator among most snake robots is sembled robot). The module was a rectangular box and four
the direct current (DC) motor. Coupled with a gear train and/or sides of the module were covered by tracks. OmniTread-8
a mechanical linkage, a DC motor can drive a module to bend or and OmniTread-4 demonstrated operation over many ter-
rotate [169,170]. The intuitive design of a snake robot is to control rain types such as gravel, dirt, ramp conditions and smooth
a number of chained segments to rotate relative to each other surfaces.
to mimic undulation motion of a biological snake. A DC motor • Artificial muscle actuator: Designing multifunctional artifi-
is a straight-forward solution for rotating one segment relative cial muscle actuators with diversely complex deformation
to another. Therefore, DC motors are initially and widely used is crucial to the snake robots and their bioinspired ap-
in the module design of snake robots. The early designs with plications. Hu et al. [175] combined the hygroexpansion
DC motors were implemented by coupling a customized gear and thermal expansion commonly observed in nature to
train, an encoder and an electrical circuit board. For example, the fabricate electrically driven paper-based bilayer actuator
first snake robot Active Cord Mechanism ACM-III [95,125] utilized by a simple printing method. This actuator exhibits large
spur gears set with high reduction gear ratio to achieve accurate and anisotropic deformation coupled with features of low-
turning control and rotation torque. To obtain two or three DOFs cost, superior stability, and an additional reversible colour-
on a single module, more motors and complex mechanical struc- change function. Interested readers can refer to [107] for
tures were applied; for instance, Snoopy snake robot [98] used a tendon driven manipulator named Arm-I, and [108] for
two motors in one module and the joint of the JPL robot [165] a tendon driven minimally invasive surgery snake robot
consists of angular bevel gears and angular swivel cup. Apart from named Distal Dexterity Unit. Readers can also refer to [176]
using customized gear trains and control circuits, people also use and [177] for artificial muscle modules, [178] and [179]
commercial RC servos which have a compact size and are already and for shape memory alloy modules, and [180,181] for
equipped with reduction gears, encoders and control circuits. piezoelectric motor modules.
Published examples of using RC servos include Modsnake [93] in • Bioinspired soft actuator: With the development of new
Choset’s laboratory [156] and ACM-R5 [97] in Hirose’s research materials and surface structures, bioinspired soft machines
group. made of highly deformable materials can be used for a vari-
ety of innovative applications, but their movement typically
3.2.2. Other actuators requires multiple independently activated actuators [120].
Apart from DC motor actuators, other technologies of actua- Abdel et al. [123] used the kirigami principle to significantly
tors include pneumatic actuators, artificial muscles, shape mem- enhance the crawling ability of soft actuators. They designed
ory alloys, piezoelectric motors and hybrid actuators using both a highly stretchable kirigami surface in which its mechanical
DC motors and pneumatic actuators. instability caused a transition from a flat to a 3D textured
10
Table 2 modules can only have small bending range ( 30 degrees) and
Comparison of actuators. the potential motion type of the whole robot is limited. Further-
Actuator Advantages Drawbacks more, the control accuracy and response time are worse than DC
DC motors Large bending range; Limited torque; motor-based modules.
Easy to design and control; Miniaturization issues;
Hybrid actuators combining DC motors and pneumatic joints
Turning angle feedback Mechanical friction
have the advantages of stiffness control and a large propulsion
Pneumatic High force transmission; Lack of accurate control;
surface. Applying tracks on the module surface give the robot
Linear expansion motion; Limited joint bending range
Low cost; Simple design maximum propulsion contact in rough terrain. Therefore, motion
efficiency of such snake robots is higher than pneumatic and DC
Hybrid motors Stiffness control; Large Low DOF; Limited joint
propulsion surface bending range; High motion motor modules. However, the hybrid actuators can only bend
efficiency about one axis in a limited range.
4. Terrain adaptive control algorithms
surface, similar to scales on snake skin. In summary, they In general, terrain adaptive snake robots in constrained envi-
have demonstrated that the kirigami principle can be used
ronments remain an ongoing research topic. Some researchers in
to create biologically inspired and deformable skin with
the past considered the 3D obstacle avoidance and autonomous
directional friction properties that can be integrated into
control of the snake robot to be a huge technical challenge, but
soft robots, even with a single extended actuator. Although
the autonomous control of the 3D snake robot can be achieved
researchers have proposed several techniques (including
by developing robot localization and environmental modelling.
rapid prototyping, pop-up and origami) to make deformable
However, in recent years, while some studies have attempted to
structures in recent years, they believe that the proposed
achieve high adaptability of snake robots through robot local-
Kirigami approach offers simpler, faster and cheaper tech-
ization and environment modelling, others have achieved adapt-
niques to create this structure [182,183]. Kirigami skin is
ability by using decentralized control, which does not use the
made by simply inserting a series of incisions into a flat
aforementioned methods. With the gradual deepening of the
sheet. Then the mechanical instability triggered by the uni-
research on snake robots, researchers have proposed a variety of
axial tension is used to create the 3D patterns, which can
also form a permanent folding pattern. decentralized control methods to realize the autonomous control
of snake robots [32–34,38,41]. Regarding terrain adaptive control
3.2.3. Discussion and comparison algorithms, two motion strategies were applied depending to
The choice of actuators for a snake robot module depends on the terrain profile: (1) terrain contour adaptive control and (2)
many factors such as the roughness of the operating surface, the obstacle avoidance control.
minimum dimension of the tunnel to pass, the maximum gap to In this section, we divide this topic into three subtopics: (1)
cross, and other factors such as being waterproof. Therefore, we Modelling Method, (2) terrain contour adaptive control and (3)
generalize these factors into the following criteria: obstacle avoidance control. The first subtopic mainly introduces
the general kinematic and dynamic modelling of snake robots
• Torque capacity operating on an obstacle-free flat surface or in a non-constrained
• Joint bending range space. The second subtopic covers circumstances where snake
• Controllability robots move over dynamic terrain, for example, stumbling on
• Module dimension an irregular surface, changing gait on viscous-variant surface,
moving on a slope surface, climbing on a pole or over obstacles
Table 2 gives a comparison between three types of actuators
and obstacle-aided locomotion. In the third subtopic, we focus on
discussed in this paper.
the situation when snake robots must move around obstacles on
In general, DC motors have the advantage of digital controlla-
their way to a final goal.
bility and large joint bending range; especially off-the-shelf servo
motors, which have integrated gear-box and control circuitry into
one unit. It is relatively easy to design 1-DOF modules using DC 4.1. Modelling method
motors. However, the number of 1-DOF modules needed is high
when the snake robot is designed to operate in a 3D terrain. 4.1.1. Mechanical model
Using 2-DOF and 3-DOF snake robot modules can reduce the total Constructing the right mechanical model is necessary for cor-
number of modules of the whole robot while maintaining the rect design and locomotion control. Researchers applied various
same operating conditions. The main challenge is to minimize approaches to control and construction snakelike robots [181].
the cross-sectional size of the module while ensuring the output First, a simple description of kinematics given by Ma et al. [185]
torque of the joint. The 2-DOF solution of aligning two motors is sufficient if it is only necessary to describe how a serpentine
perpendicular each other had the problem of bulky module be- robot gives a set of joint angles on a plane. This method directly
cause of the small diameter-length ratio. In order to reduce the calculates kinematics based on joint angle and link length. How-
cross-sectional size, two or more motors are placed in parallel and ever, if you want to apply the same calculation method for 3D
their rotation axis is transmitted via worm gears, bevel gears, or snake robots, this approach can be cumbersome. In contrast, the
swivel gears. However, mechanical friction of these gears is much D-H method [186,187] may be more convenient in its application.
higher than spur gears so that the output torque using these gears The D-H method provides a general method of the system to
is reduced. describe the kinematics of a robot based on recursive expressions.
Pneumatic actuators feature high force transmission, simple Both above methods are relatively easy to implement. If the
design of linear deformation and low cost [171]. One module goal is to study various shapes of snake robots and then con-
can achieve 3-DOF motion, for example, SSR [173] can bend vert these shapes into corresponding joint angles, the backbone
about two axes and stretch/shrink about the third axis. The cross- curve method proposed by Chirikjian et al. [163,188] may be
section of the module can be smaller than DC motor modules; more appropriate. However, the cost of clearly understanding
for example, a miniature endoscopy mobile robot named Heart- the relevant theory of the skeleton curve can be quite signifi-
Lander [184], is only 8 mm in diameter. However, pneumatic cant and should be considered when choosing the appropriate
11
modelling techniques. The backbone curve approach allows re- complex. This is partly solved by introducing non-minimum coor-
searchers to focus more on the actual shape of the snake robot dinates for non-smooth 3D models, where the contact force in the
than on its individual joint angle, which is the starting point modelling is set to a rigid body and the velocity can be changed
for modelling using the previously mentioned method [50,189]. instantaneously [50,201]. The model is suitable for simulation,
Nansai et al. [190] elucidated a generalized singularity analysis of but not for analysis considerations. Furthermore, familiarity with
snake robots, which is defined as analysis of the singularity of a the non-smooth dynamics framework requires some work. The
model which defines all designable parameters such as the link discussion above shows that there are many aspects to consider
length and the position of the passive wheel as arbitrary variables. when choosing the right modelling strategy [207–209]. Hrdina
Tanaka et al. [121,191] derived a novel necessary and sufficient et al. [210,211] solved local controllability of a snake robot’s
condition for the singular configurations of the snake robot and position by means of the conformal geometric algebra, more
proposed a shape control method for snake robots, which can
precisely by the Clifford algebra of signature (3, 1). Huang et al.
maintain head position and orientation and avoid joint limits
[212] proposed a novel approach for the dynamic modelling of
and self-collision. Liptak et al. [192] used fibre bundle theory for
a snake robot based on Udwadia-Kalaba theory. This method is
creating a mathematical model of a nonholonomic locomotion
easily processed for many DOFs snake robots. The nonholonomic
system with undulatory movement as a general mathematical
constraints of the passive wheels’ lateral velocity are also easily
model for snake robots.
handled using Udwadia-Kalaba theory.
4.1.2. 2D dynamics model
The Lagrangian formula and the Newton–Eulerian formula 4.2. Terrain contour adaptive control
are similar for rigid body dynamics because the final answer
is the same. The Lagrangian formula differs from the Newton– One of the most attractive features of snake robots is ter-
Eulerian formula in that the model of the snake robot is always rain adaptability in constrained environments. The snake robot
treated as a whole and is modelled using a Lagrangian function can deform its body to follow terrain contours, thereby gaining
that includes system energy [113,197]. Therefore, one might ar- propulsion dependent on the posture and location of the contact
gue that the Newton–Eulerian formula is the most appropriate points. Their long and thin bodies can easily fit into variant
method to obtain the torque required for the desired motion, terrain contours [50,213]. In this strategy, snake robots move over
while the Lagrangian formula is more suitable for describing
various terrain adaptively.
the time evolution of the system’s generalized coordinates [50].
The adaptive control schemes aim to achieve the adaptive
For those familiar with Lagrangian formulas and geometric me-
control of joint angle tracking and direction of locomotion con-
chanics, the 2-dimensional model proposed by Mcisaac et al.
trol [213]. Mukherjee et al. [214,215] presented a sliding mode
[198] may prove to be advantageous, as this method can be
used to control the framework of the momentum. However, for control based approach to address the velocity tracking and
those unfamiliar with the fundamentals of geometric mechanics, head-angle control problem of a planar snake robot. Moham-
a thorough understanding of these methods can be demanding madi et al. [216] investigated a manoeuvring control method for
and time consuming. The advantages of the Newton–Eulerian planar snake robots to make the centre of mass of the snake
formula are the real stick–slip transition and the relatively simple robot converge with a desired path and traverse the path with
contact simulation of the obstacle. Therefore, Sato et al. [199] a desired velocity. Moreover, with the gradual deepening of the
have given the 2D model that is the easiest to implement which research on snake robots, researchers have proposed a variety of
can also be used to develop model-based snake-like robot speed decentralized control methods to realize the autonomous control
and heading control algorithms. This method is based on the of snake robots [39,185,217]. They respectively proposed the
standard Newton–Eulerian formula, where each link is initially local reflexive mechanisms (such as curvature derivative control
calculated independently by describing their respective linear and and Tegotae-based control) of various terrain contour adaptive
angular motions, and the links are connected to each other by control methods for snake robots, which is of great significance
a joint constituting the bilateral binding force. The non-smooth for the study of the high adaptability of snake robots in con-
2D model proposed by Transeth et al. [132,200,201] used the strained environments [32,33,38,218,219]. To generate adaptive
Newton–Euler formula in conjunction with the set-valued force motion, some learning algorithms, such as genetic algorithms or
law. intensive learning, can be used on a large scale [220–222]. They
Excepting the above two methods, Tang et al. [202] proposed can help the robots to collect sensory information, analyse the
the Newton–Euler equations and the convex optimal method to internal state of the robot and the stimuli around it, and learn
approach the problem. Hildenbrand et al. [203] introduced the the appropriate response [96]. After the learning process, the
transcript of direct and differential kinematics of a robotic snake robot will automatically adjust its movement mode or gait to
in terms of compass ruler algebra (CRA). Kelasidi et al. [204,205]
the environment, while under automatic control. Therefore, the
proposed a mathematical model for snake robots in a 2D plane,
control architecture is closed and complete [43,223], as shown in
which includes the extra propulsion forces acting on each link
Fig. 7(a) and (b).
from the added effectors.
We will discuss how existing snake robots can adapt to uneven
surfaces, change gaits regarding to dynamic friction of a surface,
4.1.3. 3D dynamics model
Continuing to observe the 3D model of the snake robot dy- deal with slippery slopes, climb over obstacles, pass through
namics, we can choose many models. Liljeback et al. [164] have narrow passages, avoid getting stuck, and recover from joint
used standard techniques to model robotic manipulators to cre- malfunction in subsequent sections. Moreover, due to its special
ate 3D models of snake robots, and those accustomed to these body shape and application requirements, the snake robot also
general methods will find this method easy. Another benefit has a special movement mode, i.e., obstacle aided locomotion. In
of using these techniques is that many of the methods used obstacle aided locomotion, snake robots can generate thrust to
to precisely control robotic manipulators are now available for support their body movement by touching obstacles and using
snake robots [96,206]. However, the selection of the minimum them as stationary push points [74,132,186,224]. Meanwhile, the
coordinate and flexible contact force model makes the result- obstacle aided locomotion of snake robots will also be introduced
ing equations of motion very cumbersome and computationally in detail.
12
Fig. 7. (a) Adaptive control architecture [43]. (b) Adaptive control architecture using policy gradient reinforcement learning [43]. (c) Illustration of friction adaptive
lateral undulation gait [39,40]. (d) Illustration of backbone as a helical tread around a virtual core cylinder [193]. (e) Modsnake performs sidewinding on a slope [193].
(f) A sequence for raising a fifth segment (dashed plotting is initial position and solid plotting is the end position) [194]. (g) Backbone curve [195]. (h) Path following
of a planar snake robot with obstacles [196]. Left: initial backbone guess, right: final backbone curve of snake robot to follow. (i) Reaction force and body deformation
of a snake according to obstacles [70]. (Circles: pegs as obstacles, arrows with number indicate the direction and magnitude of the forces).
4.2.1. Dynamic friction 4.2.2. Slope

It is well known that snakes change their amplitude so that Sloped terrain represents terrain with a certain incline rate,
they adapt themselves to the surface friction coefficient ratio. For and the friction coefficient of its surface is determined by various
example, the snake increases the angle between the tangential factors such as the slope angle and the type of medium. Due
direction of the motion and the moving surface to increase the to the special terrain of slopes, snake robots always use various
tangential force due to the easy sliding of the surface. Liljeback adaptive control methods to achieve body transformation and
et al. [225] studied the controllability of snake robots affected by movement gait adjustment, thereby achieving maximum slope
viscous ground friction on viscous ground. The analysis shows stability in the process of movement [185,209]. Endo et al. [228]
that the snake robot has limited control when the friction of studied the self-contained terrain adaptability of ACM-I snake
the viscous ground is uniform, but the snake-shaped robot be- robot on a sloping surface. They also proposed an adaptive control
comes very easy to control when the friction of the viscous method to choose the optimum winding angle depending on the
ground is anisotropic. Sato et al. [39,40] proposed a local re- friction coefficient ratio. If the propulsive velocity was increased,
flexive mechanism of autonomous decentralized control system, the robot updated its parameters with the new settings. Ma
and took a two-dimensional snake robot as an example to show et al. [185] explored the relationship between the gradient of a
that the adaptability to the environment changes, as well as slope and winding angle of a lateral undulation gait. They found
the robustness of body segment failure caused by local sensory that the initial winding angle must be set within a range of the
reflexive control, which provides new ideas for the methodology upper and lower limits in order for the snake robot to be able to
of autonomous distributed control systems. To sense the friction move up the slope. The possible range for initial winding angle is
force between the robot and ground, an elastic material was narrow when the slope gradient is large. Wu et al. [229] designed
used to connect two adjacent modules. The experimental results a CPG network with a feedback connection for slope adaptive
showed successful amplitude and wavelength transition of lat- control by altering the winding angle and motion speed. Sato
eral undulation gait between different friction floor surfaces (see et al. [230] proposed a CPG model that can adapt to changes
Fig. 7(c)). It is worth mentioning that this decentralized control in inclination angles. The control scheme proposed in this paper
method has fault tolerance against local malfunction. Nansai et al. enables snake robots to exhibit excellent real-time adaptability
[226,227] designed a head control system capable of adapting to on various friction and slope terrain. Therefore, the proposed
passive side-slipping for snake robots in order to move through CPG model is very useful for the research on adaptive control
the environments which have ground surface conditions with of snake robots on said terrain. Nakajima et al. [231] proposed
nonuniform frictional coefficients. a control method that makes the head of a snake robot follow
13
an arbitrary trajectory of two non-parallel planes. Apart from compliance model in order to intelligently utilize the frictional
lateral undulation, sidewinding gait is also an efficient transla- contact between a robot and its environment to avoid slipping,
tion gait especially on loose and slippery ground. Choset et al. keep balance, and maintain stability. The contact interaction be-
[196] studied the stability of sidewinding gait on a slope and tween the skin made by frictional compliant material and envi-
find a solution for the minimum aspect ratio of the sidewinding ronment was modelled based on a model designed by Walton
gait to maintain stability [193]. They presented a virtual thread et al. [114] to predict the contact force.
sidewinding model in which the backbone of the snake robot was To climb up a cylinder, snake robots have to coil on the
viewed as a helical tread moving around a virtual core elliptical cylinder tightly to gain enough friction force and apply lateral
cylinder (see Fig. 7(d)). Using such a model, the stability problem rolling gait to climb up [241]. Kamegawa et al. [242] studied
of snake robots was converted to look for the local extrema of the relationship between desired coiling angle and friction force.
the potential energy of the virtual ellipse as it rolls along the Experimental results on a snake robot showed that the friction
slope. They calculated the relationship between the maximum force was proportional to the coiling angle. A similar climbing gait
stable gradient and the aspect ratio of two semi-major axes of the has been used on Modsnake as well [93]. Recently, Wang et al.
ellipse. One of ModSnakes was used to evaluate their algorithm as [243,244] analysed the balance principle of snake robots during
shown in Fig. 7(e). Later, Marvi et al. [68] found that the desert- climbing and presented a turning strategy for snake robots based
lived rattlesnake (Crotalus Cerastes) can effectively move through on serpentine curves. Pfotzer et al. [245] presented a navigation
sliding and pitching on sloping granular media such as sand approach for snake robots to autonomously overcome unknown
dunes. Experiments have shown that as the tilt angle increases, and challenging obstacles like stairs or large steps.
the rattlesnake increases the length of contact between the body
and the sand. In the physical model of the snake-shaped robot, 4.2.4. Obstacle-aided locomotion
this movement allows the body to ascend to the maximum sandy In nature, snakes exploit obstacles such as rocks on the ground
slope that is close to the slope stability. Bae et al. [232] proposed to obtain high efficiency locomotion. Snakes have been observed
a driving assistance mechanism that can prevent the snake robot gliding forward by bracing the sides of its body against grass,
from sliding down on a slope and increase its speed. stones or other obstacles on the surface of the ground: the reac-
tions from these obstacles yield reaction force normal to the con-
4.2.3. Climbing tacting surface between the body and obstacles [70] (as shown in
When a snake robot encounters an obstacle that it cannot get Fig. 7(i)). In other words, snakes utilize the stationary obstacles to
around, it has to climb over the obstacle. Moreover, other circum- increase the friction between their bodies and environments [32].
stances also require climbing including through narrow passages In order to carry out such tasks in cluttered environments, snake
such as pipes for internal or external inspection [233]. Nilsson robots must have a high degree of awareness and be capable of
et al. [194] classified climbing into two groups: free climbing and efficient obstacle exploitation to gain propulsion, this locomotion
aided climbing. Free climbing is referred to as the process by is defined as obstacle-aided locomotion [71–73].
which the robot raises some of its modules to a vertical position Regardless of whether the object is stationary or moving,
from a horizontal one, without using any supports or other tools biological snakes or snake-like robots can push their body against
than its own joints. Regarding aided climbing over an obstacle, a the objects to propel themselves forward effectively. Therefore, in
small number of front modules of a snake robot are lifted to lean obstacle aided locomotion, obstacles are considered to be objects
on the obstacle. Aided climbing allows the robot to use supports that can help the snake robot move forward. Furthermore, the
like slopes or poles to climb. In the vertical cylinder climbing snake robot must be highly conscious and able to effectively use
situation, a snake robot can coil its body around the cylinder and obstacles to gain propulsion [74,132,186,224,246]. The idea of uti-
roll its body up and down. Due to limited output torque of snake lizing the reaction force from obstacles for wheeled snake robot
robot joints, the strategy of lifting other modules should be care- was first proposed by Hirose et al. [17]. Shan et al. [247] designed
fully designed to avoid the joints being damaged by the weight a snake robot named MS-1, which allowed contact with obstacles,
of lifted modules. Nilsson et al. [194] also proposed a simple though no reaction force was considered. Bayraktaroglu et al.
recursive climbing algorithm (see Fig. 7(f) for an example) which [248,249] systematically studied the obstacle-aided avoidance on
can minimize the dynamic torque. Brown et al. [234] developed a snake robot and proposed an algorithm to maintain lateral undu-
snake robot named Millibot Train, which can climb vertical steps lation gait with random circle obstacles. Bayraktaroglu et al. [94,
nearly half the robot length using manual configuration. Snake 250] implemented this algorithm on a real snake robot without
robots can move from one position to another on a large scale wheels later. The contact force and friction to obstacle and ground
by applying adaptive motion behaviour [235]. Kon et al. [236] can be modelled for obstacle-aided dynamics of snakes [231].
proposed a control method for semiautonomous step climbing by Nakajima et al. [231] clarified the appropriate contact conditions
a snake robot based on mixed integer quadratic programming to between the snake robot and the contact surface, and designed
generate the reference trajectory of the head of the snake robot. a novel control method to adjust the snake robot to meet these
The features are that the method determines suitable positions contact conditions. Transeth et al. [50,200,251] modelled the
and time durations in which to sense the surroundings. Further- contact force to obstacles in 3D snake robot dynamics. Later,
more, the method takes into account the constraints on velocity Liljeback [99,252] presented a hybrid model of the dynamics of
and acceleration for the snake robot following the generated tra- a planar snake robot interacting with obstacles. Obstacle contact
jectory securely. Fu et al. [237,238] developed a snake robot with forces were calculated by formulating and solving a linear com-
lateral undulation and cantilever gait and snake-like anisotropic plementarity problem. The friction force on the module can be
friction that can quickly and stably climb steps up to one-third of written based on viscous-like model proposed by Transeth et al.
its body length. [50] in this model. The dynamics of each module can then be
In climbing a narrow passage, modules of a snake robot can calculated by Lagrangian [198] and Newton–Euler methods [164].
take advantage of friction to support the whole weight of the Kano et al. [32–34] studied the obstacle-aided locomotion of the
robot if both ends of the module apply enough pressure force snake robot which imitated the actual movement of the biological
on the passage wall. Such self-locking phenomenon was used by snake, and proposed a variety of decentralized control methods
Barazandeh et al. [239] to optimize joint torque in a climbing for this movement. Travers et al. [74] quantified how different
concertina movement. Shapiro et al. [240] developed a frictional levels of coordination affect joint motion in environments with
14
different levels of infrastructure based on kinematic coupling, and an obstacle field from starting point to a goal. Choset et al.
then analysed the control problem of snake robots in obstacle- [196] improved Chirikjian’s method by adapting a roadmap and
aided locomotion. Kamegawa et al. [75] proposed an extended proposed a generalized variance graph (GVG) to simplify the
asymmetric reverse lateral control method for the obstacle-aided search of the virtual tunnel and backbone curve, and this GVG is
locomotion of snake robots. The research on the obstacle-aided an one-dimension map of an arbitrary dimension space. During
locomotion of snake robots lays a foundation for the development this curve growing procedure, the head motion has been deter-
of robots that work in uncertain environments, and helps to mined and the rest of the body just follows its predecessor to
understand the basic principles of animal adaptive motion [32– finish the whole-body path planning, so called follow-the-leader
34,74,75]. method [229]. In Choset’s follow-the-leader approach [156,167],
the robot shape is described by a parametric backbone curve
4.2.5. Other constrained environments (as shown in Fig. 7(g)). The principle of this method is that the
The following describes the adaptive control method of snake inverse kinematic problem of hyper-redundant robot is reduced
robot in other constrained environments, such as horizontal pipe to finding a group of time varying modal participation factors that
and some uneven terrain. Hirose et al. [13] proposed an satisfy task requirements. The solid circle points in Fig. 7(h) are
impedance control method combined posture control of a snake GVG vertices and the lines between them are GVG edges. The
robot operating on uneven terrain. Each module of the robot backbone curve is very close to the GVG edge, which is assured to
can adjust its height to ground as well as the rotation angle be collision free, such as the left of Fig. 7(h). However, such back-
of its joint. Brunete et al. [100] applied a closed loop control bone curves may have sharp corners and exceed physical joint
method to prevent excessive energy consumption by the snake angle limitations. Therefore, they [167] applied a cost function to
robot when around obstacles. The servo can be stopped when the minimize the path with respect to curvature so that the snake
controller detects an unexpectedly high consumption of energy robot can easily follow the path, such as the backbone curve at
due to increased friction when approaching the servo position. the right of Fig. 7(h). The concept of follow-the-leader has been
Recently, Ariizumi et al. [253–255] considered the problem of realized in CPG controllers as well [71,263].
trajectory tracking of a planar snake robot without a lateral con- Other obstacle avoidance algorithms include using mechanical
straint and determined torque control to reduce tracking errors. intelligence [264,265] and behaviour based control [266]. Given
They also presented a multiobjective optimization algorithm for an approximated robot path in a static obstacle environment,
expensive-to-evaluate noisy functions for robotics. Cao et al. [256, many path following algorithms can be used, such as B-spline
257] investigated locomotion efficiency optimization and adap-
path following algorithm proposed by Conkur’s [267]. Hirose
tive path following of snake robots by using approximate dynamic
et al. [13] applied a neural network method based on tactile
programming and neural networks in a complex environment.
sensors to control the joint angles of ACM-III snake robot. Li et al.
[268] proposed a trajectory tracking control law based on the
4.3. Obstacle avoidance control
improved snake-like curve of a multi-joint snake-like robot to
avoid obstacles.
In a cluttered environment, some obstacles are difficult or
unnecessary to climb over for snake robots. Obstacle avoidance
5. Applications of snake robot
control algorithms are applied in such situation. In obstacle avoid-
ance locomotion, obstacles are viewed as passive objects and the
In the past, snake robots were only used in structured en-
robot changes its body shape to avoid contacting them. Snake
vironments due to limitations in technology [43,275,276]. With
robots have many difficulties in controlling their long body to
travel through an obstacle space because of their limited obstacle the development of robotics, especially in the field of sensing
sensing abilities and redundant body structure. Although tactile and control, the use of snake robots has expanded into unstruc-
sensors can be equipped along the whole body, it is still very hard tured environments (see Fig. 8(a)). Section 1 discusses the unique
to sense the position and shape of obstacles. Therefore, many characteristics of snake robots, such as small cross sections, ultra-
solutions are limited to simulations or using an overhead vision redundancy and a high degree of flexibility that can be bent on
system to provide coordinate reference of obstacles and robots. In one or more planes, making snake robots an excellent choice for
this strategy, snake robots must move around obstacles towards a many practical applications, as shown in Fig. 8(b). The applica-
final goal. Obstacle avoidance algorithms for redundant manipu- tion of many snake robots is mainly focused on exploration, fire
lators have been studied for tens of years, though the mobility protection, search and rescue, medical, pipeline inspection and
of snake robot makes it complex or even impossible to apply other industrial applications [277]. This section will specifically
these algorithms because of their assumption of a stationary base introduce the various applications of snake robots in practice.
coordinate system [258,259].
Early study of obstacle avoidance for snake robots can trace 5.1. Medical applications of snake robots
back to the research by Maciejewski et al. [258], Sciavicco et al.
[259] and David et al. [260]. They added Jacobian matrix of an 5.1.1. Invasive surgery
obstacle avoidance point as a second constraint together with Many of other kinds of continuum robots used in the medical
an end-effector Jacobian matrix to determine joint angles. The field are not suitable for use in orthopaedic surgery. They lack
obstacle avoidance point was assigned to the closest point of the the following combination of conditions: (a) inserting the tool
robot to obstacles. However, this method was hard to apply on into a sufficiently large space in the work area, (b) meeting
a snake robot with multiple DOFs especially operating in a 3D the strength and stiffness requirements of the orthopaedic en-
space because of the large dimension of the Jacobian matrix and vironment interaction, and (c) the flexibility to achieve proper
an infinite number of solutions of joint angles. bending [274,278]. However, snake robots may be a potential
To solve this problem, Chirikjian et al. [195,261,262] devel- medical device for assisted surgery, which can overcome the
oped a backbone curve concept to describe the mechanism cen- limitations of traditional surgery. For example, most skin and
treline of a robot (see Fig. 7(g)). The geometric aspects of hyper- tissues need to be opened during surgery, and the robots with
redundant robot motion planning are reduced to the determina- small cross-sections can operate in a much less damaging way to
tion of the proper time varying behaviour of the backbone refer- the body [79,279]. The snake robots can climb into the internal
ence set. To avoid obstacles, a virtual tunnel was built through tissue through a small incision and use the onboard sensors for
15
Fig. 8. (a) The applications of snake robot [48,269]. (b) How to use snake robots to check scaffolding structures and piers in hazardous situations [270,271]. (c–d)
Snake robots applied to firefighting [48] (e) Search and rescue after an earthquake [48]. (f) Snake robot applied in pipeline inspection [272]. (g) Inspection and
maintenance inside a process plant [48]. (h) Snake robot applied in pipeline inspection [273]. (i) The snake robot is deployed on site. The picture above shows the
robot inserting the rainwater pipe and the operator control unit in two positions. The image below shows the robot before deployment (left) and after retrieving
from the rainwater pipe (right) [135,274]. (j) The snake robot applied in the inspection of the storm sewer pipe [49].
further diagnosis. This will significantly reduce the patient’s pain mechanical structure and working principles in practical appli-
and recovery time. In addition, the operation will become faster cations. Li et al. [291] proposed that the workspace and dexterity
and easier with the use of the snake robot, thus reducing medical of these designs are limited due to the lack of control in either
costs [43,280]. In various applications, snake robots are dedicated the length of the bending section or the curvature of the bend-
to medical care, especially invasive surgery. Recently, researchers ing section at the distal end. Therefore, they presented a novel
have tried to describe a new type of continuous snake robot constrained wire-driven flexible mechanism, in which both the
that is specifically designed for orthopaedic surgery [281]. Some length and the curvature of the bending section are controllable.
studies have reported the application of snake robots in heart Neumann et al. [292] considered a tendon-driven snake robot
surgery [282,283]. Jin et al. [284] presented a forward kinematic with extensible sections and presented a method to determine a
model of a wire-driven surgical snake robot with an articulated sequence of robot configurations to achieve motion along a given
joint structure and path generation algorithms with solutions of 3D path in the minimally-invasive surgery.
inverse kinematics.
5.1.3. Clinical applications of virtual fixtures (VF)
5.1.2. Robot assisted minimally invasive surgery In recent years, the clinical application of VF has become more
and more common. There are also some application prospects in
It is well known that due to the complexity of instrument
this field, such as sinus surgery, plastic surgery, cardiac surgery,
operation, the performance of minimally invasive surgery (MIS) is
and ophthalmic surgery [293]. The researchers considered the
combined with the surgeon’s manual flexibility and increased de-
use of snake robots in image-guided surgery [293–298]. Due to
mand for visual motion control [285]. Tissue deformation, limited
the visually constrained and narrow workspace as previously
working space, and limited visibility in a chaotic environment
described, the implementation of VF by snake robots can help
present significant challenges to surgical ergonomics, precision,
the surgeons locate a particular tissue structure or dissect near
and safety. The small cross section of the snake robots and the
a structure with a defined safety margin to prevent unintentional
characteristic of excellent flexibility are also applied to minimally tissue damage. In this way, the robot can be prevented from
invasive surgery [108,286,287]. The use of surgical snake-robots moving beyond the security zone, ensuring the preservation of
avoids many of the shortcomings of traditional minimally inva- key characteristics [297,298]. Recently, Berthet-Rayne et al. [299]
sive surgery, and it also allows surgery to be separated between proposed a new robotic platform: the intuitive imaging sensing
surgeons and patients. Despite the motion scaling enhancement navigated and kinematically enhanced (i2 Snake) robot that aims
principle to precisely move the robot, it achieves the flexibility to improve the field of endoscopic surgery. The proposed robotic
of the surgeon and the need to coordinate the visual move- platform design focuses on ergonomics and intuitive control.
ment [288,289]. For the application of snake robots in minimally
invasive surgery, Hu et al. [290] described the design and fabri- 5.2. Extreme environment applications of snake robots
cation of a new type of flexible joint mechanism that is enabled
by metal powder bed additive manufacturing technique. Shang 5.2.1. Firefighting applications
et al. [279] proposed a snake robot composed of multiple joints In the applications of fire protection, researchers built the
for minimally invasive surgery, and specifically introduced its snake robot fire hose, which is inspired by the biological snake.
16
It can climb into the burning building and extinguish the fire Therefore, the most relevant application of snake-shaped robots
by itself, without putting human firefighters at risk [48,300], in various industrial applications may be pipeline inspection [189,
see Fig. 8(c) and (d). The researchers also proposed a control 332]. The shape and size of the snake robots are extremely
algorithm for the snake robot that can be used for firefighting important, as shown in Fig. 8(f–j). Many researchers [50,189,333]
rescue and maintenance [300,301]. have developed a prototype of snake robots called PIKo, as shown
in Fig. 8(h). Kuwada et al. [135] and Enner et al. [334] proposed
5.2.2. Urban rescue and disaster relief
a method for estimating the motion of a snake robot inside and
For decades, the application of snake robots in urban search
and rescue (USAR) has been an active field of research. In USAR outside a straight tube. This method is achieved by introducing
applications, robots must be able to traverse very small spaces a method to construct a fixed frame of the robot that is aligned
and get the smallest possible support in an unstable environ- with the centreline of the straight tube. Wakimoto et al. [335]
ment [302–304]. Rescue robots require smart devices with learn- and Wang et al. [331] have developed a pipeline navigation
ing capabilities that can adapt to all situations [252,305,306]. snake robot that uses a plane sinusoidal gait to pass through the
The search and rescue mission requires rescuers to find trapped pipeline. Wang et al. [331] have proposed the development of an
survivors in the collapsed structure after an earthquake or other autonomous pipeline robot that performs an on-line ultrasonic
disaster, as shown in Fig. 8(e). The limbless snake robot has inspection of the wall thickness to find the location of a defect.
small cross-sectional areas that allow it to penetrate deep into David et al. [323] proposed a compatible controller for straight
the gravel and the robots can also be equipped with various tubes that allows the robot to automatically adapt to the shape
sensors such as cameras, microphones and infrared detectors to
of its environment as it navigates the pipeline. Bando et al. [336]
obtain information in the gravel. Sensing information can be sent
presented a sound-based online localization method for an in-
to rescuers in real time to help find survivors. This will make
pipe snake robot with an inertial measurement unit. Jiang et al.
the search and rescue missions in the ruins faster and more
efficient [307–309]. The limbless snake robot will be an important [337] performed a neuromorphic vision sensor and a spiking
help for rescuers [310,311]. neural network on a snake robot for the first time to achieve pipe-
After the earthquake in Japan, the application of snake robots like object tracking. Rollinson et al. [136] presented a method
in rescue operations received widespread media attention [305, of achieving whole-body compliant motions with a snake robot
309]. AlonWolf et al. [312] and Wolf et al. [98] considered the which allows the robot to automatically navigate in the pipeline.
design and control of mobile super-redundant serpentine robots
in urban search and rescue. The system can inspect the area
accessible to the mobile base using cameras and sensors mounted 6. Future development trends and challenges
on its end effectors [313]. Moreover, it is also very important
to design snake robots with many uses for this purpose [314– Research on snake robots has increased over the past few
316]. Researchers also provide an interesting job for unstructured
decades, but there are still many challenges related to snake
and uneven terrain [155,317]. The researchers designed a new
robots that can move adaptively through constrained terrain.
flexible modular snake robot that can climb stairs and explored
its complexity [179,269,318]. They also demonstrated that snake Based on the future development trends and challenges of snake
robots can be further designed with better versatility in the USAR robots, the research should be developed in the directions listed
scenario [319]. Ito et al. [320] developed a semi-autonomous in Table 3. As shown in Table 3, snake robots have many chal-
serially connected multi-crawler snake robot for search and res- lenges, such as limited payload capability, poor power efficiency
cue. Zhang et al. [321] proposed an adaptive control scheme for and a high number of DOFs. Nevertheless, the snake robot has
a planar underwater snake robot with model errors and time- the potential of great variant terrain locomotion ability and the
varying noise based on dynamic models, which makes the snake capability of inspecting narrow places. These properties make the
robot able to be applied in underwater disaster rescue. research on snake robots worthwhile.
5.3. Inspection and exploration

6.1. Design of self reconfigurable structure of snake robots
In many cases, humans need to check for environments that
are too small or too dangerous for them. Inspection and ex-
ploration tasks are typically performed in an unstructured and Reconfigurable mechanism means that a mechanical system
compact environment. To complete the inspection and explo- is composed of intelligent modules. Different combinations of
ration tasks, the robots must have a high degree of flexibility and different number of modules can change the shape and size of
sensitivity [319,322–324]. A limbless snake robot with on-board the structure to meet the needs of different work. If a system
sensors is ideal for this type of task. On the one hand, the benefits can change its structure autonomously and dynamically, then it
of a small cross section make it easy for robots to climb in narrow can dynamically adapt to various environments and tasks, then
and compact environments [325,326]. On-board sensors, on the it becomes a self-reconfigurable system [37,179,346]. Different
other hand, help robots collect environmental information and modules can quickly and easily self-check, test and analyse new
perform effective inspections and accurate positioning [280,327]. design ideas and solutions [6,30,45,338]. The snake robot has
Due to its special body structure, snake robots have been used the characteristics of modularization, which is the basis of its
in exploration tasks in recent years. From planetary exploration to
reconfiguration. The reconfigurable function makes the snake
in-depth exploration, any environment that cannot be predicted
robot system have strong versatility, high cost effectiveness and
is a good application area for these robots. Since the average
fast self-recovery ability. The reconfigurable snake robot can not
thickness of lunar soil is very large and the function of the lunar
soil sampling system is inherently limited, it is difficult to collect only produce a variety of gaits, but also respond to environmen-
deep lunar soil samples during lunar exploration. Hou et al. [328] tal stimuli in complex ways with the help of multiple sensors.
proposed the use of serpentine robots to explore lunar craters. Furthermore, the reconfigurable snake robot is able to adjust
Moreover, snake robots can accommodate a variety of diame- different locomotion patterns to adapt to different terrain, and
ter and pipe configurations with a single mechanism [329–331]. this adaptability is very important [7,178,245].
17
Table 3
Directions of the research and future works.
Directions References Future Work
Design of self Kamimura et al. [178]; Wieser et al. [319]; Yim et al. [30,338]; Benavidez Modular structure of snake robot; Adaptive ability of
reconfigurable et al. [339]; Gomez et al. [340]; Li et al. [43]; Liljeback et al. [48,341]; Chen environment; Creating more versatile autonomous
structure et al. [342]; Rollinson et al. [343,344]; Chris et al. [345]; Moubarak et al. robotic systems; Energy self-sufficiency
[346]; Sen et al. [347]; Wang et al. [348,349]; Yang et al. [155]; Date et al.
[38]; Yun et al. [350]
Amphibious snake Crespi et al. [24,152]; Transeth et al. [200,201]; Cohen et al. [351]; Ding et al. Creating snake-robots that can move flexibly on land
robot [352]; Maladen et al. [270,353]; Enner et al. [138,334]; Ishikawa et al. [354]; and in water; sand Swimming Snake Robot;
Liljeback et al. [48]; Kelasidi et al. [139,140] Serpentine Heterogeneous swarm robots
Material selection Crespi et al. [24,152]; Yu et al. [355]; Lichter et al. [271]; Tesch et al. [356]; Binary drivers embedded intelligence material;
Wright et al. [93]; Laschi et al. [357]; Singh et al [358]; Xu et al. [359]; Rubber Spring Damping effect; Combination of rigid
Zawidzki et al. [360] body structure and soft material
Structural design Meyer et al. [137]; Moubarak et al. [346]; Hirose et al. [36]; Liljeback et al. Mechanical simplicity; Weight design; Driving
[48,341]; Hayashi et al. [361]; Atakan et al. [362]; Gomez et al. [340]; Hayashi mechanism; Friction substrate design; Combined
et al. [361]; Huang et al. [363]; Marvi et al. [66–68,82]; Koopaee et al. [364]; joint design; Design of flexible Connection unit
Nishimura et al. [365]; Sturges et al. [366]; Kamegawa et al. [310]; Patil et al.
[367]; Berthet-Rayne et al. [169]
Highly adaptive Tkaczyk et al. [57]; Gomez et al. [340]; Jihoon et a. [264]; Hayashi et al. Path planning and navigation; Optimal control;
functionality [361]; Crespi et al. [152]; Ostrowski et al. [368]; Atakan et al. [362]; Guo Intelligent gait selection; Compatible control based
et al. [316];Gupta et al. [369]; Kuwada et al. [135];Kimura et al. [370]; on gait; Improving robot motions
Garnett et al. [307]; Rollinson et al. [329,344]; Branyan et al. [263]; Sears
et al. [371]; Ouyang et al. [372]; Gang et al. [373]; Xu et al. [359]; Chenming
et al. [374]; Travers et al. [74]; Zhang et al. [375]
Visual system Liljeback et al. [48]; Matsuura et al. [376]; Saito et al. [377]; Hadi et al. [273]; Slam enable mobile robot; Statistical design and
Sato et al. [199]; Jiawen et al. [378] simulated annealing optimization techniques
Synthetic approach Kano et al. [32–34,41,54]; Sato et al. [40,230], Virgala et al. [332], Hwang Biological simulation system; Robotics-inspired
et al. [141], Fu et al. [237,238], Bae et al. [232], Branyan et al. [119], Qi et al. biology
[171], Liao et al. [241], Koopaee et al. [113,197], Dear et al. [129], Kamegawa
et al. [218], Ariizumi et al. [255], Hou et al. [328], Li et al. [219], Zhang et al.
[174], Astley et al. [56,83,84], Chen et al. [222], Tanaka et al. [379], Chang
et al. [90], Qian et al. [246], Bing et al. [220,221], Schiebel et al. [86], Manzoor
et al. [89]; Kotay et al. [380]; Li et al. [43]; Hayashi et al. [361]; Ostrowski
et al. [368]; Voyles et al. [381]; Nishimura et al. [365]; Saito et al. [377];
Voyles et al. [381]; Liptak et al. [382]; Mukherjee et al. [214]; Tanaka et al.
[383–385]; Olatunji et al. [386]; Nansai et al. [190]
6.2. Amphibious snake robots 6.4. Structural design of snake robots
In response to the environmental adaptation needs of snake

The structural design of snake robots is the most important
robots in complex environments such as swamps and shallows,
researchers have extensively analysed the latest developments part in the development of snake robots. The research suggests
in the research of amphibious snake robots [140,204,205,387]. that the ideal condition would be the snake robot has a passive
The advantages of the amphibious snake robot are obvious. If and smooth tactile skin that could slide forward like a biological
the robot has amphibious properties, it can handle unfavourable snake. Mechanical simplicity is necessary for the widespread use
environments including any form of water (such as rain, partially of snake robots, which increases the reliability of locomotion
flooded terrain and mud) [321,388]. The key technologies of and reduces the development cost of the robot [341]. Simul-
amphibious snake robots include: creating snake robots that can taneously, weight is a core issue in the mechanical design of
move flexibly on land and in water, sand swimming snake robots highly redundant snake robots [361]. In order for the snake robot
and snake heterogeneous swarm robots [24,152]. to move in an unstructured and challenging environment, it is
particularly important to establish a strong and robust motion
6.3. Material selection of snake robots
mechanism, which means that there is some lower limit to the
At present, the new materials proposed by researchers for ratio between the strength of the actuator and the weight of
snake robots mainly include: binary drive embedded smart mate- the robot. Therefore, the development of snake robots that max-
rial [120,175,271], variable stiffness smart material [176], rubber imize this ratio is an important design challenge that must be
spring damping material [271,357] and the combination of rigid addressed [48,354]. Furthermore, investigations show that bio-
structures and soft materials [130,171]. The successful develop- logical snakes significantly control the friction and resistance of
ment of various new materials also provides more choices for their substrates by using different mechanisms, thus enabling
driving devices of snake robots. These driving devices have ad- their functional surfaces to effectively interact with the complex
vantages that the traditional motor drive does not have [24,152]. environment. Therefore, functional biologically inspired surfaces
For example, the artificial muscle drive has the advantages of
have been of great interest to researchers and have a wide range
rapid response, silent operation, no need for waterproofing, can
of applications over the years [66,142,240,389]. Moreover, in
be used as a sensor at the same time, and easy to miniaturize.
Shape memory alloy has the characteristics of fast response, order to achieve the stability of the snake robot when it moves on
continuous high frequency response and high integration [49,60]. uneven ground, special flexible joints are needed to connect the
However, due to the characteristics of these new materials, they modules [59,153,284,290]. The design of flexible joints and com-
also add a certain complexity to the establishment of models and bined joints is also one of the key technologies in the structural
motion control of snake robots. design of snake robots [169,219,379].
18
6.5. Highly adaptive functionality of snake robots an approach to understand the essential mechanism of animal lo-
comotion by developing mathematical models and robots [32–34,
Snake robots have been widely used in challenging environ- 41,54]. Through the research on the synthesis method, the snake
ments, and there is an increasing demand for their highly adap- robot can achieve better performance, and then have broader
tive functionality. Maintaining and strengthening this trend is development prospects [40,230].
a necessary condition in order to realize the potential locomo-
tion ability of snake robots. Snake robots have multiple DOF of 7. Conclusion
movement, which can achieve diverse and flexible gaits by using
the flexibility of the body, and can adjust the gait according to Research on snake robots has increased for decades, but many
the state of the environment in a timely manner, so it has a challenges concerning modelling and control of snake robots
good potential to adapt to the environment and a wide range remain before these robots are able to move adaptively through
of application prospects [81,228]. However, due to its complex constrained terrain. Module structures and control algorithms
dynamics, such as under-actuation, high-dimensional nonlinear- are two key elements for snake robots operating in constrained
ity and environmental uncertainty, it is difficult for snake robots environments. This review gave a detailed study of the module
to obtain high efficient locomotion ability and environmental structures and control algorithms, and also attempted to address
adaptability. Therefore, applications of snake robots in the future both. First, a review on the biological snake’s locomotion and its
require more research on adaptive behaviour in locomotion in body structure is provided, which outlines the biological foun-
unknown and chaotic environments [4]. To improve the envi- dation of all snake robots. This is followed by the mechanical
ronment adaptability of the snake robot, it is necessary to solve structure of snake robots, especially the structure of elemental
the three problems, including: gait generation, gait characteristic snake modules. Finally, control algorithms for variant terrain
analysis, and gait application [38,180]. Moreover, in many appli- contours and obstacle avoidance are discussed. The review also
cation fields, the research on the environment sensing of snake outlines emerging application areas and potential future direc-
tions of snake robots. It is hoped that this survey will help
robots is also very important, which indicates that the snake
promote further research on the fascinating topic of snake robots
robot should realize the interaction with the environment and
through the overview given in this paper.
strengthen the environment adaptability [39,41,207].
Declaration of competing interest
6.6. Visual system of snake robots
The authors declare that they have no known competing finan-
Highly redundant snake robots need to work in complex and cial interests or personal relationships that could have appeared
constrained environments, and their goal is to plan motion paths to influence the work reported in this paper.
effectively. Discussions on efficient motion planning strategies
suggest that future snake robots need to be equipped with appro- Acknowledgement
priate visual systems [109,167]. As the research on the vision of
snake robots gradually increases, the vision hardware that can be We gratefully acknowledge Professor Guang-Zhong Yang,
used to realize the efficient movement of the snake robot system Medical Robotics Institute at Shanghai Jiao Tong University, for
is continuously improved. Note that the payload capacity of the his advice and comments to this paper during drafting and re-
snake robot is limited, which means that the vision system should viewing.
be both small in size and limited in weight [48,315]. The robot
generates its own map of the environment and simultaneously References
determines its position in the map is called simultaneous local-
ization and mapping (SLAM). SLAM is an extensive and important [1] R. Hayashi, K. Osuka, T. Ono, Trajectory control of an air cushion
field of research, and this mechanism is often used in situations vehicle, in: IEEE/RSJ/GI International Conference on Intelligent Robots and
Systems - Advanced Robotic Systems and the Real World, vol. 3, 1994,
where maps of the environment are not available in advance, so pp. 1906–1913.
it is essential for autonomous operation of snake robots. Future [2] J. Pratt, C.M. Chew, A. Torres, P. Dilworth, G. Pratt, Virtual model control:
research on SLAM for snake robots can focus on map creation by an intuitive approach for bipedal locomotion, Int. J. Robot. Res. 20 (2)
combining data from vision sensors with data from contact force (2001) 129–143.
[3] E.R. Westervelt, J.W. Grizzle, D.E. Koditschek, Hybrid zero dynamics of
sensors of robots [342]. Among them, the contact force sensing
planar biped walkers, IEEE Trans. Automat. Control 48 (1) (2003) 42–56.
capability of the snake robot can extract information about the [4] Y. Fukuoka, H. Kimura, A.H. Cohen, Adaptive dynamic walking of a
friction coefficient of the surfaces and objects in contact with quadruped robot on irregular terrain based on biological concepts, Int.
itself, as well as important parameters such as elasticity and J. Robot. Res. 22 (3–4) (2003) 187–202.
[5] R. Playter, M. Buehler, M. Raibert, Bigdog, in: Conference on Unmanned
plasticity of external objects [17,218]. Furthermore, it can make
Systems Technology VIII, vol. 6230, 2006, pp. O2302–O2302.
the snake robot interact with the environment better and greatly [6] W.M. Shen, M. Krivokon, H. Chiu, J. Everist, M. Rubenstein, J. Venkatesh,
enhance the adaptability of the snake robot. Multimode locomotion via superbot reconfigurable robots, Auton. Robots
20 (2) (2006) 165–177.
[7] L. Chen, S. Ma, B. Li, C. Wang, Modular universal unit for a snake-like
6.7. Synthetic approach of snake robots robot and reconfigurable robots, Adv. Robot. 23 (7–8) (2009) 865–887.
[8] X.S. Xiao, R. Murphy, A review on snake robot testbeds in granular
Bionic robotics refer to the goal of simultaneously improving and restricted maneuverability spaces, Robot. Auton. Syst. 110 (2018)
the adaptability and learning ability of robots, thereby ‘cultivat- 160–172.
[9] J. Granville, Locomotion capabilities of a modular robot with eight
ing’ a new generation of robots to better survive in complex and pitch-yaw-connecting modules, Gomez 41 (3) (2006) 477–502.
constrained operating environments [176,277]. The bionic snake [10] M. Patil, T. Abukhalil, S. Patel, T. Sobh, Hardware architecture review of
robot mainly imitates some biological characteristics of the bio- swarm robotics system: self reconfigurability, self reassembly and self
logical snake such as structure, mobility, vision, locomotion gait replication, ISRN Robot. 2013 (2013) 433–444.
[11] K.Y. Pettersen, Snake robots, Annu. Rev. Control 44 (2017) 19–44.
and navigation [81,372]. The performance of the natural system is [12] X.Z. Zang, Y.X. Liu, Z.K. Lin, C. Zhang, S. Iqbal, Two multi-linked rescue
very effective, so the research of biological simulation systems is robots: design, construction and field tests, J. Adv. Mech. Des. Syst. Manuf.
urgently needed [183,390]. Moreover, robotics-inspired biology is 10 (6) (2016).
19
[13] B.S. Hirose, Biologically Inspired Robots: Snake-Like Locomotors and [41] T. Kano, A. Ishiguro, Decoding decentralized control mechanism under-
Manipulators, Oxford University Press, 1993. lying adaptive and versatile locomotion of snakes, Integr. Comp. Biol. 60
[14] M. Yamakita, M. Hashimoto, T. Yamada, Control of locomotion and (1) (2020) 232–247.
head configuration of 3D snake robot (SMA), in: 20th IEEE Interna- [42] J.K. Hopkins, B.W. Spranklin, S.K. Gupta, A survey of snake-inspired robot
tional Conference on Robotics and Automation, ICRA, vol. 2, 2003, pp. designs, Bioinspiration Biomim. 4 (2) (2010) 021001.
2055–2060. [43] M. Li, B.H. Kim, A.I. Mourikis, Real-time motion tracking on a cellphone
[15] A.J. Ijspeert, Central pattern generators for locomotion control in animals using inertial sensing and a rolling-shutter camera, in: IEEE International
and robots: a review, Neural Netw. 21 (4) (2008) 642–653. Conference on Robotics and Automation, ICRA, 2013, pp. 4712–4719.
[16] Z.L. Wang, Q. Gao, H.Y. Zhao, CPG-inspired locomotion control for a snake [44] R.J. Webster, A.M. Okamura, N.J. Cowan, Toward active cannulas: minia-
robot basing on nonlinear oscillators, J. Intell. Robot. Syst. 85 (2) (2017) ture snake-like surgical robots, in: IEEE/RSJ International Conference on
209–227. Intelligent Robots and Systems, 2007, pp. 2857–2863.
[17] S. Hirose, Y. Umetani, Kinematic control of active cord mechanism with [45] M. Yim, D.G. Duff, K.D. Roufas, Polybot: a modular reconfigurable robot,
tactile sensors, in: Second CISM/IFTOMM International Symposium on the in: IEEE International Conference on Robotics and Automation, vol. 1,
Theory and Practice of Robots and Manipulators, 1976, pp. 241–252. 2002, pp. 514–520.
[18] R.A. Brooks, A.M. Flynn, Fast, Fast, cheap and out of control: a robot [46] L. Chen, S. Ma, Y. Wang, B. Li, D. Duan, Design and modelling of a snake
invasion of the solar system, J. Br. Interplanet. Soc. 42 (10) (1989) robot in traveling wave locomotion, Mech. Mach. Theory 42 (12) (2007)
478–485. 1632–1642.
[19] X.D. Wu, S.G. Ma, CPG-based control of serpentine locomotion of a [47] M. Walton, B.C. Jayne, A.F. Bennet, The energetic cost of limbless
snake-like robot, Mechatronics 20 (2) (2010) 326–334. locomotion, Science 249 (4968) (1990) 524–527.
[20] X.D. Wu, S.G. Ma, Neurally controlled steering for collision-free behavior [48] P. Liljeback, K. Pettersen, O. Stavdahl, J. Gravdahl, Snake Robots:
of a snake robot, IEEE Trans. Control Syst. Technol. 21 (6) (2013) Modelling, Mechatronics, and Control, Springer, 2013.
2443–2449. [49] D. Rollinson, Control and Design of Snake Robots (Ph.D. thesis), Carnegie
[21] D. Blanchard, K. Aihara, T. Levi, Snake robot controlled by biomimetic Mellon University, Pittsburgh, PA, 2014.
CPGs, in: International Conference on Artificial Life and Robotics, ICAROB, [50] A.A. Transeth, K.Y. Pettersen, P. Liljeback, A survey on snake robot
Japan, 2019. modeling and locomotion, Robotica 27 (7) (2009) 999–1015.
[22] A. Crespi, A.J. Ijspeert, Online optimization of swimming and crawling in [51] C. Wilbur, A lamprey-based undulatory vehicle, Neurotechnol. Biomimetic
an amphibious snake robot, IEEE Trans. Robot. 24 (1) (2008) 75–87. Robots (1973).
[23] Z. Lu, S. Ma, B. Li, Y. Wang, Serpentine locomotion of a snake-like [52] M. Arai, Y. Tanaka, S. Tsukui, Development of ‘Souryu-IV’ and ‘Souryu-V’
robot controlled by cyclic inhibitory CPG model, in: IEEE/RSJ Interna- serially connected crawler vehicles for in-rubble searching operations:
tional Conference on Intelligent Robots and Systems, vol. 1–4, 2005, pp. research articles, J. Field Robotics 25 (1–2) (2008) 31–65.
3019–3024. [53] D.L. Hu, J. Nirody, T. Scott, M.J. Shelley, The mechanics of slithering
[24] A. Crespi, A.J. Ijspeert, Amphibot II: an amphibious snake robot that locomotion, Proc. Natl. Acad. Sci. USA 106 (25) (2009) 10081–10085.
crawls and swims using a central pattern generator, Color Res. Appl. [54] T. Kano, T. Sato, R. Kobayashi, A. Ishiguro, Efficient undulating locomotion
(2006) 19–27. driven by a decentralized control that fully exploits multi-articular
[25] Z.L. Lu, S.G. Ma, B. Li, 3D locomotion of a snake-like robot controlled by muscles, in: Proceedings of the 5th International Symposium on Adaptive
cyclic inhibitory CPG model, Acta Automat. Sinica (2006). Motion in Animals and Machines, AMAM, 2011, pp. 69–70.
[26] J.W. Burdick, J. Radford, G.S. Chirikjian, A sidewinding locomotion gait for [55] K. Inoue, K. Nakamura, M. Suzuki, Y. Mori, Y. Fukuoka, N. Shiroma,
hyper-redundant robots, Adv. Robot. 9 (3) (1995) 195–216. Biological system models reproducing snakes’ musculoskeletal system,
[27] X. Guo, S.G. Ma, B. Li, Y.C. Fang, A novel serpentine gait generation in: IEEE/RSJ International Conference on Intelligent Robots and Systems,
method for snakelike robots based on geometry mechanics, IEEE-ASME IROS, 2010, pp. 2383–2388.
Trans. Mechatron. 23 (3) (2018) 1249–1258. [56] H.C. Astley, The biomechanics of multi-articular muscle–tendon systems
[28] X. Guo, W. Zhu, Y.C. Fang, Guided motion planning for snake-like robots in snakes, Integr. Comp. Biol. 60 (1) (2020) 140–155.
based on geometry mechanics and HJB equation, IEEE Trans. Ind. Electron. [57] F. Tkaczyk, Snake tracks-understanding serpent locomotion, 2010, http:
66 (9) (2019) 7120–7130. //www.wildernesscollege.com/snake-tracks.html.
[29] M. Yim, Y. Zhang, K. Roufas, D. Duff, Connecting and disconnecting for [58] H. Ohno, S. Hirose, Study on slime robot (proposal of slime robot and
chain self-reconfiguration with Polybot, IEEE-ASME Trans. Mechatron 7 design of slim slime robot), in: IEEE/Robotics-Society-of-Japan Interna-
(4) (2002) 442–451. tional Conference on Intelligent Robots and Systems, IROS 2000, 2000,
[30] M. Yim, W.M. Shen, B. Salemi, D. Rus, M. Moll, H. Lipson, Modular self- pp. 2218–2223.
reconfigurable robot systems, in: The Workshop on Robotics-Science & [59] H. Yamada, S. Hirose, Approximations to continuous curves of active
Systems, 2007, pp. 43–52. cord mechanism made of arc-shaped joints or double joints, in: IEEE
[31] B. Yan, D. Shi, J. Wei, Hibot: a generic ROS-based robot-remote-control International Conference on Robotics and Automation, ICRA, 2010, pp.
framework, Intell. Robot Syst. (2017). 703–708.
[32] T. Kano, A. Ishiguro, Obstacles are beneficial to me! Scaffold-based [60] Y.Y. Yang, M. Zhang, D.F. Li, Y.J. Shen, Graphene-based light-driven soft
locomotion of a snake-like robot using decentralized control, in: IEEE/RSJ robot with snake-inspired concertina and serpentine locomotion, Adv.
International Conference on Intelligent Robots and Systems, IROS, 2013, Mater. Technol. 4 (1) (2019).
pp. 3273–3278. [61] J. Gray, The mechanism of locomotion in snakes, J. Exp. Biol. 23 (23)
[33] T. Kano, R. Yoshizawa, A. Ishiguro, Tegotae-based decentralised control (1946) 101–120.
scheme for autonomous gait transition of snake-like robots, Bioinspiration [62] B. Jayne, Muscular mechanisms of snake locomotion-an electromyo-
Biomim. 12 (4) (2017). graphic study of the sidewinding and concertina modes of crotalus-
[34] T. Kano, T. Sato, R. Kobayashi, A. Ishiguro, Local reflexive mechanisms cerastes, nerodia-fasciata and elaphe-obsoleta, J. Exp. Biol. 140 (1988)
essential for snakes’ scaffold-based locomotion, Bioinspiration Biomim. 7 1–33.
(4) (2012). [63] B. Jayne, Muscular mechanisms of snake locomotion-an electromyo-
[35] H. Yamada, S. Hirose, Steering of pedal wave of a snake-like robot graphic study of lateral undulation of the florida banded water snake
by superposition of curvatures, in: IEEE/RSJ International Conference on (nerodia-fasciata) and the yellow rat snake (elaphe-obsoleta), J. Morphol.
Intelligent Robots and Systems, vol. 25, 2010, pp. 419–424. 197 (2) (1988) 159–181.
[36] S. Hirose, Impedance control of articulated body mobile robot Koryu, in: [64] B. Jayne, J. Davis, Kinematics and performance capacity for the concertina
Internaational Conf on Intellgent Robots and Systems: Intelligent Robots locomotion of a snake (coluber-constrictor), J. Exp. Biol. 156 (1991)
for Flexibility, IROS, 1993. 539–556.
[37] J. Liu, X. Zhang, G. Hao, Survey on research and development of [65] B. Moon, C. Gans, Kinematics, muscular activity and propulsion in gopher
reconfigurable modular robots, Adv. Mech. Eng. 8 (8) (2016). snakes, J. Exp. Biol. 201 (19) (1998) 2669–2684.
[38] H. Date, Y. Takita, Adaptive locomotion of a snake like robot based on [66] H. Marvi, D.L. Hu, Friction enhancement in concertina locomotion of
curvature derivatives, in: IEEE/RSJ International Conference on Intelligent snakes, J. R. Soc. Interface 9 (76) (2012) 3067–3080.
Robots and Systems, 2007, pp. 3554–3559. [67] H. Marvi, J. Bridges, D.L. Hu, Snakes mimic earthworms: propulsion using
[39] T. Sato, W. Watanabe, A. Ishiguro, An adaptive decentralized control rectilinear travelling waves, J. R. Soc. Interface 10 (84) (2013).
of a serpentine robot based on the discrepancy between body, brain [68] H. Marvi, C. Gong, N. Gravish, H. Astley, M. Travers, R.L. Hatton, Sidewind-
and environment, in: IEEE International Conference on Robotics and ing with minimal slip: snake and robot ascent of sandy slopes, Science
Automation, ICRA, 2010, pp. 709–714. 346 (6206) (2014) 224.
[40] T. Sato, T. Kano, A. Ishiguro, On the applicability of the decentralized [69] C.H. Gong, M.J. Travers, H.C. Astley, L. Li, J.R. Mendelson, D.I. Goldman, H.
control mechanism extracted from the true slime mold: a robotic case Choset, Kinematic gait synthesis for snake robots, Int. J. Robot. Res. 35
study with a serpentine robot, Bioinspiration Biomim. 6 (2) (2011). (1–3) (2016) 100–113.
20
[70] J. Gray, The kinetics of locomotion of the grass-snake, J. Exp. Biol. 26 (4) [97] H. Yamada, S. Chigisaki, M. Mori, Development of amphibious snake-like
(1950) 354. robot ACM-R5, in: The Proc. of 36th International Symposium on Robotics,
[71] F. Sanfilippo, J. Azpiazu, G. Marafioti, A.A. Transeth, O. Stavdahl, P. vol. 36, 2005, p. 133.
Liljeback, Perception-driven obstacle-aided locomotion for snake robots: [98] A. Wolf, H.B. Brown, R. Casciola, A. Costa, M. Schwerin, E. Shamas, H.
the state of the art, challenges and possibilities, Appl. Sci.-Basel 7 (2017) Choset, A mobile hyper redundant mechanism for search and rescue
4. tasks, in: IEEE/RSJ International Conference on Intelligent Robots and
[72] F. Sanfilippo, E. Helgerud, P.A. Stadheim, S.L. Aronsen, Serpens: a highly Systems, 2003, pp. 2889–2895.
compliant low-cost ROS-based snake robot with series elastic actuators, [99] P. LiljebaCk, K.Y. Pettersen, O. Stavdahl, J.T. Gravdahl, A hybrid model of
stereoscopic vision and a screw-less assembly mechanism, Appl. Sci.-Basel obstacle-aided snake robot locomotion, in: IEEE International Conference
9 (3) (2019). on Robotics and Automation, ICRA, vol. 58, 2010, pp. 675–682.
[73] F. Sanfilippo, O. Stavdahl, G. Marafioti, A.A. Transeth, P. Liljeback, Virtual [100] A. Brunete, E. Gambao, J.E. Torres, M. Hernando, A 2-DOF servomotor-
functional segmentation of snake robots for perception-driven obstacle- based module for pipe inspection modular micro-robots, in: IEEE/RSJ
aided locomotion, in: IEEE International Conference on Robotics and International Conference on Intelligent Robots and Systems, 2006, pp.
Biomimetics, 2016, pp. 1845–1851. 1329–1334.
[74] M. Travers, J. Whitman, H. Choset, Shape-based coordination in [101] E. Shammas, A. Wolf, H. Choset, Three degrees-of-freedom joint for spatial
locomotion control, Int. J. Robot. Res. 37 (10) (2018) 1253–1268. hyper-redundant robots, Mech. Mach. Theory 41 (2) (2006) 170–190.
[75] T. Kamegawa, R. Kuroki, A. Gofuku, Evaluation of snake robot’s behaviour [102] B. Klaassen, K.L. Paap, GMD-snake 2: a snake-like robot driven by wheels
using randomized EARLI in crowded obstacles, in: Proceedings of the and a method for motion control, in: International Conference on Robotics
International Symposium on Safety, Security and Rescue Robotics, 2014. and Automation, ICRA, vol. 4, 1999, pp. 3014–3019.
[76] C. Heckrotte, Relations of body temperature, size, and crawling speed of [103] C. Ye, S. Ma, B. Li, H. Liu, H. Wang, Development of a 3D snake-like robot:
the common garter snake, thamnophis’s sirtalis, Copeia 1967 (4) (1967) perambulator-II, in: IEEE International Conference on Mechatronics and
759–763. Automation, 2007, pp. 117–122.
[77] B.C. Jayne, What defines different modes of snake locomotion? Integr. [104] J. Borenstein, M. Hansen, A. Borrell, The OmniTread OT-4 serpentine robot
Comp. Biol. 60 (1) (2020) 156–170. - design and performance: Field reports, J. Field Robot 24 (7) (2007)
[78] P.S. Schenker, Self-reconfiguration planning for a class of modular robots, 601–621.
in: Proceedings of the Society of Photo-Optical Instrumentation Engineers, [105] K. Osuka, H. Kitajima, Development of mobile inspection robot for rescue
SPIE, vol. 3839, 1999, pp. 246–257. activities: Moira, in: Intelligent Robots and Systems, IROS, vol. 4, 2003,
[79] O. Shmakov, Snakelike robots locomotions control, mechatronics- pp. 3373–3377 vol. 3.
foundations and applications, 2006. [106] R. Haraguchi, K. Osuka, S. Makita, S. Tadokoro, The development of the
[80] B.C. Jayne, Kinematics of terrestrial snake locomotion, Copeia (4) (1986) mobile inspection robot for rescue activity, Moira2, in: 12th International
915–927. Conference on Advanced Robotics, 2005, pp. 498–505.
[81] H.C. Astley, Traversing tight tunnels-implementing an adaptive con-
[107] S. Ma, H. Yoshinada, S. Hirose, CT ARM-I: coupled tendon-driven ma-
certina gait in a biomimetic snake robot, in: Annual Meeting of the
nipulator model I-design and basic experiments, in: IEEE International
Society-for-Integrative-and-Comparative-Biology, SICB, San Francisco, CA,
Conference on Robotics and Automation, vol. 3, 1992, pp. 2094–2100.
2018.
[108] N. Simaan, X. Kai, W. Wei, A. Kapoor, P. Kazanzides, R. Taylor, Design and
[82] H. Marvi, R. Chrystal, J. Shieh, J. Mendelson, R. Hatton, H. Choset, D.
integration of a telerobotic system for minimally invasive surgery of the
Goldman, D. Hu, Sidewinding snakes on sand, Integr. Comp. Biol. 53 (1)
throat, Int. J. Robot. Res. 28 (9) (2009) 1134–1153.
(2013) 139.
[109] M.A. Lewis, D.M. Zehnpfennig, R7: a snake-like robot for 3-D visual
[83] H.C. Astley, Long limbless locomotors over land: the mechanics and
inspection, in: IEEE/RSJ/GI International Conference on Intelligent Robots
biology of elongate, limbless vertebrate locomotion, Integr. Comp. Biol.
and Systems-Advanced Robotic Systems and the Real World, vol. 2, 1994,
60 (1) (2020) 134–139.
pp. 1310–1317.
[84] H.C. Astley, J.M. Rieser, A. Kaba, V.M. Paez, I. Tomkinson, J.R. Mendel-
[110] R. Ariizumi, F. Matsuno, Dynamic analysis of three snake robot gaits, IEEE
son III, D. Goldman I, Side-impact collision: mechanics of obstacle
Trans. Robot. 33 (5) (2017) 1075–1087.
negotiation in sidewinding snakes, Bioinspiration Biomim. 15 (6) (2020).
[111] S. Toyoshima, F. Matsuno, A study on sinus-lifting motion of a snake robot
[85] J.L. Tingle, Facultatively sidewinding snakes and the origins of locomotor
with energetic efficiency, in: IEEE International Conference on Robotics
specialization, Integr. Comp. Biol. 60 (1) (2020) 202–214.
and Automation, ICRA, 2012, pp. 2673–2678.
[86] P.E. Schiebel, H.C. Astley, J.M. Rieser, S. Agarwal, C. Hubicki, A.M. Hubbard,
[112] S. Toyoshima, M. Tanaka, F. Matsuno, A study on sinus-lifting motion of a
K. Diaz, J.R. Mendelson III, K. Kamrin, D. Goldman I, Mitigating memory
snake robot with sequential optimization of a hybrid system, IEEE Trans.
effects during undulatory locomotion on hysteretic materials, ELIFE 9
Autom. Sci. Eng. 11 (1) (2014) 139–144.
(2020).
[87] L. Chen, Y. Wang, S. Ma, B. Li, Studies on lateral rolling locomotion [113] M.J. Koopaee, C. Pretty, K. Classens, X.Q. Chen, Dynamical modelling and
of a snake robot, in: IEEE International Conference on Robotics and control of snake-like motion in vertical plane for locomotion in unstruc-
Automation, 2004, pp. 5070–5074. tured environments, in: Proceedings of the ASME International Design
[88] H.W. Lissmann, Rectilinear locomotion in a snake (boa occidentalis), J. Engineering Technical Conferences and Computers and Information in
Exp. Biol. 26 (4) (1950) 368–379. Engineering Conference, vol. 9, 2020.
[89] S. Manzoor, U. Khan, I. Ullah, Serpentine and rectilinear motion genera- [114] K. Walton, The oblique compression of two elastic spheres, J. Mech. Phys.
tion in snake robot using central pattern generator with gait transition, Solids 26 (3) (1978) 139–150.
Iran. J. Sci. Technol.-Trans. Electr. Eng. 44 (3) (2020) 1093–1103. [115] L. Rai, S.J. Kang, Multi-thread based synchronization of locomotion control
[90] A.H. Chang, P.A. Vela, Shape-centric modeling for control of traveling in snake robots, in: IEEE International Conference on Embedded &
wave rectilinear locomotion on snake-like robots, Robot. Auton. Syst. 124 Real-time Computing Systems & Applications, 2005.
(2020). [116] S.H. Gopi, D.K. Valappil, B. Ramakrishnananda, T.R.S. Kumar, V.S. Balaram-
[91] F. Edwin, L.T. Tung, Starcly robot: a novel compact stair climbing robot, raja, Modeling non-equilibrium glides in flying snakes, in: International
in: International Conference on Mechanical & Electronics Engineering, Conference on Advances in Computing, Communications and Informatics,
2010, pp. V2–75–V2–78. ICACCI, 2017, pp. 91–96.
[92] S. Yu, S. Ma, B. Li, Y. Wang, An amphibious snake-like robot with [117] F. Jafari, S.D. Ross, P.P. Vlachos, J.J. Socha, A theoretical analysis of pitch
terrestrial and aquatic gaits, in: IEEE International Conference on Robotics stability during gliding in flying snakes, Bioinspiration Biomim. 9 (2)
and Automation, 2011, pp. 2960–2961. (2014) 025014.
[93] C. Wright, A. Johnson, A. Peck, Z. McCord, A. Naaktgeboren, P. Gianfortoni, [118] I.J. Yeaton, S.D. Ross, G.A. Baumgardner, J.J. Socha, Undulation enables
M. Gonzalez-Rivero, R. Hatton, H. Choset, Design of a modular snake gliding in flying snakes, Nat. Phys. 16 (9) (2020) 974–982.
robot, in: IEEE/RSJ International Conference on Intelligent Robots and [119] C. Branyan, R.L. Hatton, Y. Menguc, Snake-inspired kirigami skin for lateral
Systems, 2007, pp. 2609–2614. undulation of a soft snake robot, IEEE Robot. Autom. Lett. 5 (2) (2020)
[94] Z.Y. Bayraktaroglu, A. Kilicarslan, A. Kuzucu, Design and control of 1728–1733.
biologically inspired wheel-less snake-like robot, in: IEEE/RAS-EMBS [120] A. Rafsanjani, Y.R. Zhang, B.Y. Liu, S.M. Rubinstein, K. Bertoldi, Kirigami
International Conference on Biomedical Robotics and Biomechatronics, skins make a simple soft actuator crawl, Sci. Robot. 3 (15) (2018).
2006, pp. 1001–1006. [121] T. Takemori, M. Tanaka, F. Matsuno, Gait design for a snake robot by
[95] H. Yamada, Development of practical 3-dimentinal active cord mechanism connecting curve segments and experimental demonstration, IEEE Trans.
ACM-R4, J. Robot. Mechatron. 18 (2006) 305–311. Robot. 34 (5) (2018) 1384–1391.
[96] A.A. Transeth, P. Liljeback, K.Y. Pettersen, Snake robot obstacle aided loco- [122] A. Orthey, O. Roussel, O. Stasse, Motion planning in irreducible path
motion: an experimental validation of a non-smooth modeling approach, spaces, Robot. Auton. Syst. 109 (2018) 97–108.
in: IEEE/RSJ International Conference on Intelligent Robots and Systems, [123] A. Abdel, A. Hisham, Surface structure and tribology of legless squamate
2007, pp. 2582–2589. reptiles, J. Mech. Behav. Biomed. Mater. 79 (2018) 354–398.
21
[124] A.H. Chango, P.A. Vela, Evaluation of bio-inspired scales on locomotion [153] E. Shammas, A. Wolf, H.B. Brown, H. Choset, New joint design for
performance of snake-like robots, Robotica 37 (8) (2019) 1302–1319. three-dimensional hyper redundant robots, in: IEEE/RSJ International
[125] M. Mori, S. Hirose, Three-dimensional serpentine motion and lateral Conference on Intelligent Robots and Systems, vol. 3, 2003, pp.
rolling by active cord mechanism ACM-III, in: IEEE/RSJ International 3594–3599.
Conference on Intelligent Robots and Systems, vol. 1, 2002, pp. 829–834. [154] S. Hirose, H. Yamada, Snake-like robots [tutorial], IEEE Robot. Autom. Mag.
[126] H. Ohno, S. Hirose, Design of slim slime robot and its gait of locomotion, 16 (1) (2009) 88–98.
in: IEEE Conference on Intelligent Robots and Systems, IROS 2001, vol. 2, [155] G.Z. Yang, S.G. Ma, B. Li, M.H. Wang, A hierarchical connectionist CPG
2001, pp. 707–715. controller for controlling the snake-like robot’s 3-dimensional gaits, in:
[127] T. Takemori, M. Tanaka, F. Matsuno, Ladder climbing with a snake robot, The 25th IEEE/RSJ International Conference on Intelligent Robots and
in: 25th IEEE/RSJ International Conference on Intelligent Robots and Systems, IROS, 2012, pp. 822–827.
Systems, IROS, 2018. [156] H.M. Choset, J.E. Luntz, Design and motion planning for serpentine robots,
[128] T. Dear, S.D. Kelly, M. Travers, H. Choset, Locomotive analysis of a single- in: Smart Structures and Materials 2000 Conference, vol. 3990, 2003, pp.
input three-link snake robot, in: IEEE Conference on Decision and Control, 148–155.
2016, pp. 7542–7547. [157] G. Granosik, M.G. Hansen, J. Borenstein, The OmniTread serpentine robot
[129] T. Dear, B. Buchanan, R. Abrajan-Guerrero, S.D. Kelly, M. Travers, H. for industrial inspection and surveillance, Ind. Robot 32 (2) (2005)
Choset, Locomotion of a multi-link non-holonomic snake robot with 139–148.
passive joints, Int. J. Robot. Res. 39 (5) (2020) 598–616. [158] G.S.P. Miller, J. Ayers, J.L. Davis, A. Rudolph, Snake robots for search and
[130] Y.T. Cao, Y.L. Liu, Y.L. Chen, L.L. Zhu, Y. Yan, X. Chen, A novel slithering rescue, 2002.
locomotion mechanism for a snake-like soft robot, J. Mech. Phys. Solids [159] S. Ma, Analysis of creeping locomotion of a snake-like robot, Adv. Robot.
99 (2017) 304–320. 15 (2) (2001) 205–224.
[131] S. Yaqub, A. Ali, M. Usman, K.M. Zuhaib, A.M. Khan, B. An, H. Moon, J.Y. [160] Z.L. Lu, S.G. Ma, B. Li, Y.C. Wang, Gaits-transferable CPG controller for a
Lee, C. Han, A spiral curve gait design for a modular snake robot moving snake-like robot, Sci. China Inf. Sci. 51 (3) (2008) 293–305.
on a pipe, Int. J. Control Autom. Syst. 17 (10) (2019) 2565–2573. [161] K. Ito, Y. Fukumori, Autonomous control of a snake-like robot utilizing
[132] A.A. Transeth, R.I. Leine, C. Glocker, K.Y. Pettersen, P. Liljeback, passive mechanism, in: IEEE International Conference on Robotics and
Snake robot obstacle-aided locomotion: modeling, simulations, and Automation, ICRA, 2006, pp. 381–386.
experiments, IEEE Trans. Robot. 24 (1) (2008) 88–104. [162] K.L. Paap, M. Dehlwisch, B. Klaassen, GMD-snake: a semi-autonomous
[133] K.D. Wang, W.C. Gao, S.G. Ma, Snake-like robot with fusion gait for snake-like robot, Distrib. Auton. Robot. Syst. 2 (1996) 71–77.
high environmental adaptability: design, modeling, and experiment, Appl. [163] G.S. Chirikjian, J.W. Burdick, A modal approach to hyper-redundant
Sci.-Basel 7 (11) (2017). manipulator kinematics, IEEE Trans. Robot. Autom. 10 (3) (1994) 343–354.
[134] K.D. Wang, Y.W. Ma, H.X. Shan, S.G. Ma, A snake-like robot with envelope [164] P. Liljeback, S. Oyvind, K.Y. Pettersen, Modular pneumatic snake robot
wheels and obstacle-aided gaits, Appl. Sci.-Basel 9 (18) (2019). 3D modelling, implementation and control, Model. Identif. Control 38 (1)
[135] A. Kuwada, S. Wakimoto, K. Suzumori, Y. Adomi, Automatic pipe negotia- (2005) 19–24.
tion control for snake-like robot, in: IEEE/ASME International Conference [165] E. Paljug, T. Ohm, S. Hayati, The JPL serpentine robot: a 12-DOF system for
on Advanced Intelligent Mechatronics, 2008, pp. 558–563. inspection, in: IEEE International Conference on Robotics and Automation,
vol. 3, 1995, pp. 3143–3148.
[136] D. Rollinson, H. Choset, Pipe network locomotion with a snake robot, J.
[166] H.R. Choi, S.M. Ryew, Anthropomorphic joint mechanism with two
Field Robotics 33 (3) (2016) 322–336.
degrees of freedom, in: IEEE International Conference on Robotics and
[137] J.A. Meyer, A. Guillot, Biologically inspired robots, in: Springer Handbook
Automation, vol. 2, 2000, pp. 1525–1530.
of Robotics, 2008.
[167] H. Choset, J.Y. Lee, Sensor-based construction of a retract-like structure
[138] F. Enner, D. Rollinson, H. Choset, Motion estimation of snake robots
for a planar rod robot, IEEE Trans. Robot. Autom. 17 (4) (2001) 435–449.
in straight pipes, in: IEEE International Conference on Robotics and
[168] S. Ma, G. Lan, Y. Tanabe, R. Sasaki, A serpentine robot based on 3 DOF
Automation, ICRA, 2013, pp. 5168–5173.
coupled-driven joint, in: IEEE International Conference on Robotics and
[139] E. Kelasidi, M. Jesmani, K.Y. Pettersen, J.T. Gravdahl, Locomotion efficiency
Biomimetics, ROBIO 2004, 2004, pp. 70–75.
optimization of biologically inspired snake robots, Appl. Sci.-Basel 8 (1)
[169] P. Berthet-Rayne, K. Leibrandt, K. Kiyoung, Rolling-joint design opti-
(2018).
mization for tendon driven snake-like surgical robots, in: 25th IEEE/RSJ
[140] E. Kelasidi, P. Liljeback, K.Y. Pettersen, J.T. Gravdahl, Integral line-of-sight
International Conference on Intelligent Robots and Systems, IROS, 2018.
guidance for path following control of underwater snake robots: theory
[170] M.M. Dalvand, S. Nahavandi, R.D. Howe, An analytical loading model for
and experiments, IEEE Trans. Robot. 33 (3) (2017) 610–628.
n-tendon continuum robots, IEEE Trans. Robot. 34 (5) (2018) 1215–1225.
[141] Y. Hwang, M. Ishikawa, Generative locomotion model of snake robot
[171] X. Qi, H. Shi, T. Pinto, X. Tan, A novel pneumatic soft snake robot using
with hierarchical networks for topological representation, in: ICAART:
traveling-wave locomotion in constrained environments, IEEE Robot.
Proceedings of the 12th International Conference on Agents and Artificial
Autom. Lett. 5 (2) (2020) 1610–1617.
Intelligence, vol. 2, 2020, pp. 194–202.
[172] H. Date, Y. Takita, An electricity-free snake-like propulsion mechanism
[142] S. Alben, Efficient sliding locomotion with isotropic friction, Phys. Rev. E driven and controlled by fluids, in: IEEE/RSJ International Conference on
99 (6) (2019). Intelligent Robots and Systems, 2009, pp. 3637–3642.
[143] Z.S. Bing, L. Cheng, G. Chen, F. Rohrbein, K. Huang, A. Knoll, Towards [173] T. Aoki, A. Ochiai, S. Hirose, Study on slime robot: development of the
autonomous locomotion: CPG-based control of smooth 3D slithering gait mobile robot prototype model using bridle bellows, in: IEEE International
transition of a snake-like robot, Bioinspiration Biomim. 12 (3) (2017) Conference on Robotics and Automation, vol. 3, 2004, pp. 2808–2813.
1–16. [174] D. Zhang, H. Yuan, Z. Cao, Environmental adaptive control of a snake-
[144] S. Manzoor, Y. Choi, A unified neural oscillator model for various rhythmic like robot with variable stiffness actuators, IEEE-CAA J. Autom. Sin. 7 (3)
locomotions of snake-like robot, Neurocomputing 173 (2016) 1112–1123. (2020) 745–751.
[145] S. Manzoor, Y.G. Cho, Y. Choi, Neural oscillator based CPG for various [175] Y. Hu, A.F. Xu, J.Q. Liu, L.L. Yang, L.F. Chang, M.J. Huang, W.B. Gu, G.
rhythmic motions of modular snake robot with active joints, J. Intell. Wu, P. Lu, W. Chen, Y.C. Wu, Multifunctional soft actuators based on
Robot. Syst. 94 (3–4) (2019) 641–654. anisotropic paper/polymer bilayer toward bioinspired applications, Adv.
[146] J. Borenstein, G. Granosik, M. Hansen, The omnitread serpentine Mater. Technol. 4 (3) (2019).
robot-design and field performance, Proc. SPIE 5804 (2005) 324–332. [176] H.R. Choi, K. Jung, S. Ryew, J.D. Nam, J. Jeon, J.C. Koo, Biomimetic soft
[147] M. Tesch, K. Lipkin, I. Brown, R. Hatton, A. Peck, J. Rembisz, H. Choset, actuator: design, modeling, control, and applications, IEEE/ASME Trans.
Parameterized and scripted gaits for modular snake robots, Adv. Robot. Mechatronics 10 (5) (2005) 581–593.
23 (9) (2009) 1131–1158. [177] R. Kornbluh, R. Pelrine, J. Eckerle, J. Joseph, Electrostrictive polymer
[148] S. Hirose, H. Yamada, Biologically inspired robots, IEEE Robot. Autom. artificial muscle actuators, in: IEEE International Conference on Robotics
Mag. (2009) 1395–1422. and Automation, vol. 3, 1998, pp. 2147–2154.
[149] M. Ahmed, M.M. Billah, Smart material-actuated flexible tendon-based [178] A. Kamimura, S. Murata, E. Yoshida, H. Kurokawa, Self-reconfigurable
snake robot, Int. J. Adv. Robot. Syst. 13 (2016). modular robot-experiments on reconfiguration and locomotion, in:
[150] M.J. Koopaee, S. Bal, C. Pretty, X.Q. Chen, Design and development of a IEEE/RSJ International Conference on Intelligent Robots and Systems, vol.
wheel-less snake robot with active stiffness control for adaptive pedal 1, 2001, pp. 606–612.
wave locomotion, J. Bionic Eng. 16 (4) (2019) 593–607. [179] J. Liu, Y. Wang, S. Ma, B. Li, Analysis of stairs-climbing ability for a tracked
[151] P. Racioppo, P. Ben-Tzvi, Design and control of a cable-driven articu- reconfigurable modular robot, in: IEEE International Workshop on Safety,
lated modular snake robot, IEEE-ASME Trans. Mechatron. 24 (3) (2019) Security and Rescue Robots, 2005, pp. 36–41.
893–901. [180] I. Tanev, Implications of incorporating learning probabilistic context-
[152] A. Crespi, A. Badertscher, A. Guignard, A.J. Ijspeert, Amphibot I: an sensitive grammar in genetic programming on evolvability of adaptive
amphibious snake-like robot, Robot. Auton. Syst. 50 (4) (2005) 163–175. locomotion gaits of snakebot, Proc. Gecco (2004) 155–166.
22
[181] A.K. Tran, A. Budiyono, K.J. Yoon, Design and analysis of a locomotive [211] J. Hrdina, A. Navrat, P. Vasik, R. Matousek, CGA-based robotic snake
module for snake-mimetic robots using mini-lipcas, in: IEEE International control, Adv. Appl. Clifford Algebr. 27 (1) (2017) 621–632.
Conference on Multisensor Fusion and Integration for Intelligent Systems, [212] K. Huang, K. Shao, S.C. Zhen, H. Sun, A novel approach for modeling and
2008, pp. 131–136. tracking control of a passive-wheel snake robot, Adv. Mech. Eng. 9 (3)
[182] L. Bin, Z. Aoyu, L. John, Design and evaluation of a novel rotatable (2017).
one-element snake bone for notes, J. Med. Dev.-Trans. ASME 12 (2018) [213] L. Tang, J. Huang, L.M. Zhu, X.Y. Zhu, G.Y. Gu, Path tracking of a cable-
2. driven snake robot with a two-level motion planning method, IEEE-ASME
[183] Z. Nana, Z. Hongbin, L. Bing, The effect of different scales on the crawling Trans. Mechatron. 24 (3) (2019) 935–946.
rate of bionic snake robot, in: 1st World Robot Conference (WRC) / [214] J. Mukherjee, S. Roy, I.N. Kar, S. Mukherjee, A double-layered artificial
Symposium on Advanced Robotics and Automation, SARA, 2018. delay-based approach for maneuvering control of planar snake robots,
[184] N.A. Patronik, T. Ota, M.A. Zenati, C.N. Riviere, A miniature mobile robot Trans. ASME, J. Dyn. Syst. Meas. Control 141 (4) (2019).
for navigation and positioning on the beating heart, IEEE Trans. Robot. [215] J. Mukherjee, S. Mukherjee, I.N. Kar, Sliding mode control of planar snake
25 (5) (2009) 1109–1124. robot with uncertainty using virtual holonomic constraints, IEEE Robot.
[185] S. Ma, N. Tadokoro, K. Inoue, Influence of the gradient of a slope on Autom. Lett. 2 (2) (2017) 1077–1084.
optimal locomotion curves of a snake-like robot, Adv. Robot. 20 (4) (2006) [216] A. Mohammadi, E. Rezapour, M. Maggiore, K.Y. Pettersen, Maneuvering
413–428. control of planar snake robots using virtual holonomic constraints, IEEE
[186] P. Liljeback, K.Y. Pettersen, O. Stavdahl, J.T. Gravdahl, Experimental in- Trans. Control Syst. Technol. 24 (3) (2016) 884–899.
vestigation of obstacle-aided locomotion with a snake robot, IEEE Trans. [217] S.G. Ma, N. Tadokoro, K. Inoue, B. Li, Influence of inclining angle of a slope
Robot. 27 (4) (2011) 792–800. to optimal locomotion curves of a snake-like robot, in: IEEE International
[187] B. Siciliano, L. Sciavicco, Modelling and control of robot manipulators, Ind. Conference on Robotics, Intelligent Systems and Signal Processing, vol. 1,
Robot 11 (1) (2000) 1828. 2003, pp. 353–358.
[188] G.S. Chirikjian, Design and analysis of some nonanthropomorphic, [218] T. Kamegawa, T. Akiyama, Y. Suzuki, T. Kishutani, A. Gofuku, Three-
biologically inspired robots: an overview, J. Robot. Syst. (2001). dimensional reflexive behavior by a snake robot with full circumference
[189] S.A. Fjerdingen, P. Liljeback, A.A. Transeth, A snake-like robot for internal pressure sensors, in: EEE/SICE International Symposium on System
inspection of complex pipe structures (PIKO), in: IEEE/RSJ International Integration, SII, 2020, pp. 897–902.
Conference on Intelligent Robots and Systems, 2009, pp. 5665–5671. [219] D. Li, C. Wang, H. Deng, Y. Wei, Motion planning algorithm of a multi-
[190] S. Nansai, M. Iwase, H. Itoh, Generalized singularity analysis of snake-like joint snake-like robot based on improved serpenoid curve, IEEE Access 8
robot, Appl. Sci.-Basel 8 (2018) 10. (2020) 8346–8360.
[191] M. Tanaka, K. Tanaka, Singularity analysis of a snake robot and an [220] Z. Bing, C. Lemke, L. Cheng, K. Huang, A. Knoll, Energy-efficient and
articulated mobile robot with unconstrained links, IEEE Trans. Control damage-recovery slithering gait design for a snake-like robot based on
Syst. Technol. 24 (6) (2016) 2070–2081. reinforcement learning and inverse reinforcement learning, Neural Netw.
[192] T. Liptak, I. Virgala, P. Frankovsky, P. Sarga, A. Gmiterko, L. Balockova, A 129 (2020) 323–333.
geometric approach to modeling of four- and five-link planar snake-like
[221] Z. Bing, C. Lemke, F.O. Morin, Z. Jiang, L. Cheng, K. Huang, A. Knoll,
robot, Int. J. Adv. Robot. Syst. 13 (2016).
Perception-action coupling target tracking control for a snake robot via
[193] R.L. Hatton, H. Choset, Sidewinding on slopes, in: IEEE International
reinforcement learning, Front. Neurorobot. 14 (2020).
Conference on Robotics and Automation, ICRA, 2010, pp. 691–696.
[222] J.-C. Chen, Continual learning for addressing optimization problems with
[194] M. Nilsson, Snake robot-free climbing, IEEE Control Syst. 18 (1) (2002)
a snake-like robot controlled by a self-organizing model, Appl. Sci.-Basel
21–26.
10 (14) (2020).
[195] G.S. Chirikjian, J.W. Burdick, The kinematics of hyper-redundant robot
[223] A.J. Ijspeert, A. Crespi, D. Ryczko, J.M. Cabelguen, From swimming to
locomotion, IEEE Trans. Robot. Autom. 11 (6) (1995) 781–793.
walking with a salamander robot driven by a spinal cord model, Science
[196] H. Choset, W. Henning, A follow-the-leader approach to serpentine robot
315 (5817) (2007) 1416–1420.
motion planning, J. Aerosp. Eng. 12 (2) (1999) 65–73.
[224] E. Muller, P. Kohler, K. Pettersen, F. Allgower, Economic model predictive
[197] M.J. Koopaee, C. Pretty, K. Classens, X. Chen, Dynamical modeling and
control for obstacle-aided snake robot locomotion, in: 21st IFAC World
control of modular snake robots with series elastic actuators for pedal
Congress, 2020.
wave locomotion on uneven terrain, J. Mech. Des. 142 (3, SI) (2020).
[225] P. Liljeback, K.Y. Pettersen, O. Stavdahl, J.T. Gravdahl, Controllability
[198] K. Mcisaac, J.P. Ostrowski, Motion planning for anguilliform locomotion,
analysis of planar snake robots influenced by viscous ground friction,
IEEE Trans. Robot. Autom. 19 (4) (2003) 637–652.
in: IEEE RSJ International Conference on Intelligent Robots and Systems,
[199] M. Sato, M. Fukaya, T. Iwasaki, Serpentine locomotion with robotic snakes,
2009, pp. 3615–3622.
Control Syst. IEEE 22 (1) (2002) 64–81.
[200] A.A. Transeth, R.I. Leine, C. Glocker, K.Y. Pettersen, 3-D snake robot [226] S. Nansai, T. Yamato, M. Iwase, H. Itoh, Locomotion control of snake-like
motion: nonsmooth modeling, simulations, and experiments, IEEE Trans. robot with rotational elastic actuators utilizing observer, Appl. Sci.-Basel
Robot. 24 (2) (2008) 361–376. 9 (19) (2019).
[201] A. Transeth, V.D.W. Nathan, A. Pavlov, Tracking control for snake robot [227] S. Nansai, M.R. Elara, M. Iwase, Dynamic hybrid position force control
joints, in: IEEE/RSJ International Conference on Intelligent Robots & using virtual internal model to realize a cutting task by a snake-like
Systems, 2007. robot, in: IEEE International Conference on Biomedical Robotics and
[202] C.Q. Tang, P. Li, G.B. Zhou, D.Y. Meng, X. Shu, S. Guo, Z.X. Li, Modeling and Biomechatronics, 2016, pp. 151–156.
mechanical analysis of snake robots on cylinders, J. Mech. Robot.-Trans. [228] G. Endo, K. Togawa, S. Hirose, Study on self-contained and terrain
ASME 11 (4) (2019). adaptive active cord mechanism, in: IEEE/RSJ International Conference
[203] D. Hildenbrand, J. Hrdina, A. Navrat, P. Vasik, Local controllability of snake on Intelligent Robots and Systems, 18, 1999, 419–425..
robots based on CRA, theory and practice, Adv. Appl. Clifford Algebr. 30 [229] X. Wu, S. Ma, CPG-controlled asymmetric locomotion of a snake-like robot
(1) (2019). for obstacle avoidance, in: IEEE International Conference on Robotics and
[204] E. Kelasidi, K.Y. Pettersen, P. Liljeback, J.T. Gravdahl, Locomotion effi- Biomimetics, ROBIO, 2009, pp. 69–74.
ciency of underwater snake robots with thrusters, in: IEEE International [230] T. Sato, T. Kano, A. Ishiguro, A decentralized control scheme for an
Symposium on Safety, Security, and Rescue Robotics, 2016, pp. 174–181. effective coordination of phasic and tonic control in a snake-like robot,
[205] E. Kelasidi, K.Y. Pettersen, J.T. Gravdahl, S. Stromsoyen, A.J. Sorensen, Bioinspiration Biomim. 7 (1) (2012).
Modeling and propulsion methods of underwater snake robots, in: IEEE [231] M. Nakajima, M. Tanaka, K. Tanaka, F. Matsuno, Motion control of a
Conference on Control Technology and Applications, 2017, pp. 819–826. snake robot moving between two non-parallel planes, Adv. Robot. 32
[206] M. Li, I. Mourikis, 3-D motion estimation and online temporal calibration (10) (2018) 559–573.
for camera-IMU systems, in: IEEE International Conference on Robotics [232] J. Bae, M. Kim, B. Song, M. Jin, D. Yun, Snake robot with driving assistant
and Automation, 2018, pp. 5709–5716.. mechanism, Appl. Sci.-Basel 10 (21) (2020).
[207] X. Wang, S. Ma, Adaptive creeping locomotion of a CPG-controlled [233] P. Michel, J. Chestnutt, S. Kagami, K. Nishiwaki, GPU-accelerated real-time
snake-like robot to environment change, Auton. Robots 28 (3) (2010) 3D tracking for humanoid locomotion and stair climbing, in: IEEE/RSJ
283–294. International Conference on Intelligent Robots and Systems, 2007, pp.
[208] S. Ma, Y. Ohmameuda, K. Inoue, Dynamic analysis of 3-dimensional snake 463–469.
robots, in: IEEE/RSJ International Conference on Intelligent Robots and [234] H.B. Brown, J.M. Vande Weghe, C.A. Bererton, P.K. Khosla, Millibot trains
Systems, IROS, vol. 1, 2004, pp. 767–772. for enhanced mobility, IEEE/ASME Trans. Mechatronics 7 (4) (2002)
[209] S. Ma, N. Tadokoro, Analysis of creeping locomotion of a snake-like robot 452–461.
on a slope, Adv. Robot. 20 (1) (2006) 15–23. [235] G. Wang, W.X. Yang, Y.T. Shen, H.Y. Shao, C.L. Wang, Adaptive path
[210] J. Hrdina, A. Navrat, P. Vasik, Control of 3-link robotic snake based on following of underactuated snake robot on unknown and varied frictions
conformal geometric algebra, Adv. Appl. Clifford Algebr. 26 (3) (2016) ground: theory and validations, IEEE Robot. Autom. Lett. 3 (4) (2018)
1069–1080. 4273–4280.
23
[236] K. Kon, M. Tanaka, K. Tanaka, Mixed integer programming-based semi- [263] C. Branyan, Y. Menguc, Soft snake robots: investigating the effects of
autonomous step climbing of a snake robot considering sensing strategy, gait parameters on locomotion in complex terrains, in: 25th IEEE/RSJ
IEEE Trans. Control Syst. Technol. 24 (1) (2016) 252–264. International Conference on Intelligent Robots and Systems, IROS, 2018.
[237] Q. Fu, S.W. Gart, T.W. Mitchel, J.S. Kim, G.S. Chirikjian, C. Li, Lateral oscil- [264] K. Jihoon, K. Jongwon, L. Donghun, Mobile robot with passively articulated
lation and body compliance help snakes and snake robots stably traverse driving tracks for high terrainability and maneuverability on unstructured
large, smooth obstacles, Integr. Comp. Biol. 60 (1) (2020) 171–179. rough terrain: design, analysis, and performance evaluation, J. Mech. Sci.
[238] Q.Y. Fu, C. Li, Robotic modelling of snake traversing large, smooth Technol. 32 (11) (2018) 5389–5400.
obstacles reveals stability benefits of body compliance, R. Soc. Open Sci. [265] I.K. Tomkinson, J.M. Rieser, P.E. Schiebel, Improving performance of
7 (2) (2020). a snake-like robot in heterogeneous terrain by managing effects of
[239] F. Barazandeh, B. Bahr, A. Moradi, How self-locking reduces actuators head collisions, in: Annual Meeting of the Society-for-Integrative-and-
torque in climbing snake robots, in: IEEE/ASME International Conference Comparative-Biology, SICB, 2018.
on Advanced Intelligent Mechatronics, 2007, pp. 1–6. [266] E.O. Ari, I. Erkmen, A.M. Erkmen, A FACL controller architecture for a
[240] A. Shapiro, A. Greenfield, H. Choset, Frictional compliance model devel- grasping snake robot, in: IEEE/RSJ International Conference on Intelligent
opment and experiments for snake robot climbing, in: IEEE International Robots and Systems, 2005, pp. 1748–1753.
Conference on Robotics and Automation, 2007, pp. 574–579. [267] E.S. Conkur, Path following algorithm for highly redundant manipulators,
[241] B. Liao, H. Zang, M. Chen, Y. Wang, X. Lang, N. Zhu, Z. Yang, Y. Yi, Soft Robot. Auton. Syst. 45 (1) (2003) 1–22.
rod-climbing robot inspired by winding locomotion of snake, Soft Robot. [268] D.F. Li, Z.H. Pan, H.B. Deng, T. Peng, Trajectory tracking control law of
7 (4) (2020) 500–511. multi-joint snake-like robot based on improved snake-like curve in flow
[242] T. Kamegawa, T. Harada, A. Gofuku, Realization of cylinder climbing field, Int. J. Adv. Robot. Syst. 16 (2) (2019).
locomotion with helical form by a snake robot with passive wheels, in: [269] N. Li, S. Ma, N. Li, M. Wang, Y. Wang, An online stair-climbing control
IEEE International Conference on Robotics and Automation, ICRA, 2009, method for a transformable tracked robot, State Key Lab. Robot. (2012)
pp. 4316–4321. 923–929.
[243] C. Wang, Y.B. Peng, D.F. Li, Z.H. Pan, H.B. Deng, D.G. Li, B. Li, Turning [270] R.D. Maladen, Y. Ding, P.B. Umbanhowar, A. Kamor, D.I. Goldman, Bio-
strategy of snake-like robot based on serpenoid curve under cloud physically inspired development of a sand-swimming robot, Georgia Inst.
assisted smart conditions, Cluster Comput.- J. Netw. Softw. Tools Appl. Technol. (2010).
22 (2019) 13041–13053. [271] D. Lichter, D. Matthew, Concept Development for Lightweight Binary-
[244] C. Wang, H.B. Deng, F. Xia, Y. Li, Study on climbing locomotion mechanism Actuated Robotic Devices, with Application to Space Systems (Master’s
of snake robot with universal unit, in: Proceedings of International thesis), The Massachusetts Institute of Technology, 2001.
Conference on Modelling, Identification & Control, 2016, pp. 460–465. [272] D. Rollinson, H. Choset, Gait-based compliant control for snake robots, in:
[245] L. Pfotzer, S. Klemm, A. Roennau, J.M. Zollner, R. Dillmann, Autonomous IEEE International Conference on Robotics and Automation, ICRA, 2013,
navigation for reconfigurable snake-like robots in challenging, unknown pp. 5138–5143.
environments, Robot. Auton. Syst. 89 (2017) 123–135. [273] K. Hadi, A. Alireza, B. Hossein, Application of statistical techniques in
[246] F. Qian, D.E. Koditschek, An obstacle disturbance selection framework: modeling and optimization of a snake robot, Robotica 31 (4) (2013)
emergent robot steady states under repeated collisions, Int. J. Robot. Res. 623–641.
39 (13) (2020) 1549–1566. [274] R.J. Murphy, M.D.M. Kutzer, S.M. Segreti, B.C. Lucas, M. Armand, Design
[247] Y. Shan, Y. Koren, Design and motion planning of a mechanical snake, and kinematic characterization of a surgical manipulator with a focus on
IEEE Trans. Syst. Man Cybern. 23 (4) (1993) 1091–1100. treating osteolysis, Robotica 32 (6) (2014) 835–850.
[248] Z. Bayraktarouglu, F. Butel, P. Blazevic, V. Pasqui, A geometrical approach [275] C. Vaida, N. Plitea, D. Pisla, B. Gherman, Orientation module for surgical
to the trajectory planning of a snake-like mechanism, in: IEEE/RSJ Inter- instruments-a systematical approach, Meccanica 48 (1) (2013) 145–158.
national Conference on Intelligent Robots and Systems, vol. 3, 1999, pp. [276] Z. Yang, K. Ito, K. Hirotsune, K. Saijo, A. Gofuku, F. Matsuno, A mechanical
1322–1327. intelligence in assisting the navigation by a force feedback steering wheel
[249] Z.Y. Bayraktaroglu, P. Blazevic, Understanding snakelike locomotion for a snake rescue robot, in: 13th IEEE International Workshop on Robot
through a novel push-point approach, J. Dyn. Syst. Meas. Control 127 and Human Interactive Communication, 2006, pp. 113–118.
(1) (2005) 146–152. [277] B. Yang, Z. Mu, T. Liu, Biomimetic snake robot in complex marine medical
[250] Z.Y. Bayraktaroglu, Snake-like locomotion: experimentations with a bi- rescue application, J. Med. Imaging Health Inform. 8 (2) (2018) 232–239.
ologically inspired wheel-less snake robot, Mech. Mach. Theory 44 (3) [278] C.A. Engh, H. Egawa, S.E. Beykirch, The quality of osteolysis grafting with
(2009) 591–602. cementless acetabular component retention, Clin. Orthop. Relat. Res. 465
[251] A.A. Transeth, R.I. Leine, C. Glocker, K.Y. Pettersen, Non-smooth 3D (2007) 150–154.
modeling of a snake robot with external obstacles, in: IEEE International [279] J. Shang, D.P. Noonan, C. Payne, J. Clark, An articulated universal joint
Conference on Robotics and Biomimetics, ROBIO, 2006, pp. 1189–1196. based flexible access robot for minimally invasive surgery, in: IEEE
[252] P. Liljeback, K.Y. Pettersen, Q. Stavdahl, A snake robot with a contact force International Conference on Robotics and Automation, ICRA, 2011, pp.
measurement system for obstacle-aided locomotion, in: IEEE International 1147–1152.
Conference on Robotics and Automation, ICRA, 2010, pp. 683–690. [280] B. Li, W. Yang, C. Li, A snake-like robot for inspection tasks, Klin
[253] R. Ariizumi, M. Tesch, K. Kato, H. Choset, F. Matsuno, Multiobjective opti- Monatsblatter Augenheilkd Augenarztl Fortbild (1) (2008) 201–208.
mization based on expensive robotic experiments under heteroscedastic [281] X. Jin, J. Zhao, M. feng, Snake-like surgical forceps for robot-assisted
noise, IEEE Trans. Robot. 33 (2) (2017) 468–483. minimally invasive surgery, Int. J. Med. Robot. Comput. Assist. Surg. 14
[254] R. Ariizumi, R. Takahashi, M. Tanaka, T. Asai, Head-trajectory-tracking (2018) 4.
control of a snake robot and its robustness under actuator failure, IEEE [282] B.L. Harvey, How snakes work, Math. Sci. (2011).
Trans. Control Syst. Technol. 27 (6) (2019) 2589–2597. [283] C.N. Riviere, N.A. Patronik, M.A. Zenati, Prototype epicardial crawling
[255] R. Ariizumi, M. Tanaka, Manipulability analysis of a snake robot without device for intrapericardial intervention on the beating heart, Heart Surg.
lateral constraint for head position control, Asian J. Control 22 (6) (2020) Forum 7 (6) (2004) 639–643.
2282–2300. [284] S. Jin, S.K. Lee, J. Lee, S. Han, Kinematic model and real-time path
[256] Z.C. Cao, Q. Xiao, R. Huang, M.C. Zhou, Robust neuro-optimal control of generator for a wire-driven surgical robot arm with articulated joint
underactuated snake robots with experience replay, IEEE Trans. Neural structure, Appl. Sci.-Basel 9 (19) (2019).
Netw. Learn. Syst. 29 (1) (2018) 208–217. [285] N.A. Patronik, M.A. Zenati, C.N. Riviere, A miniature cable-driven robot for
[257] Z.C. Cao, D. Zhang, B. Hu, J.G. Liu, Adaptive path following and locomo- crawling on the heart, in: 27th Annual International Conference of the
tion optimization of snake-like robot controlled by the central pattern IEEE Engineering in Medicine and Biology Society, 2005, pp. 5771–5774.
generator, Complexity 2019 (2019). [286] A. Degani, H. Choset, B. Zubiate, T. Ota, M. Zenati, Highly articulated
[258] A.A. Maciejeinski, Obstacle avoidance for kinematically redundant ma- robotic probe for minimally invasive surgery, in: International Conference
nipulators in dynamically varying environments, Int. J. Robot. Res. 4 (3) of the IEEE Engineering in Medicine and Biology Society, 2006, p. 4167.
(1985) 109–117. [287] T. Ota, A. Degani, D. Schwartzman, B. Zubiate, J. Mcgarvey, H. Choset,
[259] L. Sciavicco, On the use of redundancy in robot kinematic control, in: M.A. Zenati, A highly articulated robotic surgical system for minimally
American Control Conference, 1988, 1370–1375.. invasive surgery, Ann. Thorac. Surg. 87 (4) (2009) 1253.
[260] G. David, N. Hukukane, The Jacobian matrix and global univalence of [288] K.W. Kwok, Dynamic active constraints for robot assisted minimally
mappings, Math. Ann. 159 (2) (1965) 81–93. invasive surgery, Imp. Coll. London (2012).
[261] G.S. Chirikjian, Theory and Applications of Hyper-Redundant Robotic [289] S.L. Lee, M. Lerotic, V. Vitiello, S. Giannarou, K.W. Kwok, M. Visentini-
Manipulators (Ph.D. thesis), California Inst. of Tech., Pasadena, 1992. Scarzanella, From medical images to minimally invasive intervention:
[262] G.S. Chirikjian, J.W. Burdick, A geometric approach to hyper-redundant computer assistance for robotic surgery, Computer. Med. Imaging Graph.
manipulator obstacle avoidance, J. Mech. Des. 114 (4) (1992) 580–585. Off. J. Computer. Med. Imaging Soc. 34 (1) (2010) 33.
24
[290] Y. Hu, L. Zhang, W. Li, G.Z. Yang, Design and fabrication of a 3-D printed [317] Y. Yang, H. Qian, X. Wu, G. Xu, Y. Xu, A novel design of tri-star
metallic flexible joint for snake-like surgical robot, IEEE Robot. Autom. wheeled mobile robot for high obstacle climbing, in: The 25th IEEE/RSJ
Lett. 4 (2) (2019) 1557–1563. International Conference on Intelligent Robots and Systems, IROS, 2012,
[291] Z. Li, H.L. Ren, P.W.Y. Chiu, R.X. Du, H.Y. Yu, A novel constrained wire- pp. 920–925.
driven flexible mechanism and its kinematic analysis, Mech. Mach. Theory [318] M. Vim, S. Homans, K. Roufas, Climbing with snake-like robots, IFAC
95 (2016) 59–75. Workshop Mob. Robot Technol. 34 (4) (2001) 7–12.
[292] M. Neumann, J. Burgner-Kahrs, Considerations for follow-the-leader mo- [319] S. Wieser, Locomotion in Modular Robotics, Roombot Modules (Master’s
tion of extensible tendon-driven continuum robots, in: IEEE International thesis), 2008.
Conference on Robotics and Automation, 2016, pp. 917–923. [320] K. Ito, H. Maruyama, Semi-autonomous serially connected multi-crawler
[293] Z. Wang, E. Appleton, The concept and research of a pipe crawling rescue robot for search and rescue, Adv. Robot. 30 (7) (2016) 489–503.
robot, Adv. Robot. 17 (4) (2003) 339–358. [321] A.F. Zhang, S.G. Ma, B. Li, M.H. Wang, X. Guo, Y.C. Wang, Adaptive con-
[294] P. Dario, M.C. Carrozza, M. Marcacci, S. D’Attanasio, A novel mechatronic troller design for underwater snake robot with unmatched uncertainties,
tool for computer-assisted arthroscopy, IEEE Trans. Inf. Technol. Biomed. Sci. China Inf. Sci. 59 (5) (2016).
4 (1) (2000) 15–29. [322] F. Trebuna, I. Virgala, M. Pastor, T. Liptak, L. Mikova, An inspection of
[295] S. Tully, G. Kantor, M.A. Zenati, H. Choset, Shape estimation for image- pipe by snake robot, Int. J. Adv. Robot. Syst. 13 (2016).
guided surgery with a highly articulated snake robot, in: IEEE/RSJ [323] A. Davids, Urban search and rescue robots: from tragedy to technology,
International Conference on Intelligent Robots and Systems, 2011, pp. IEEE Intell. Syst. 17 (2) (2002) 81–83.
1353–1358. [324] Z.P. Weng, T.L. Liu, C.C. Wu, Z.C. Cao, Mechanism design and kinematic
[296] M.A. Zenati, T. Yokota, D. Schwartzman, T. Ota, B. Zubiate, C. Wright, C. performance research of snake-like robot with orthogonal active wheels
Nikou, H. Choset, Model-guided epicardial interventions using a novel in: Advances in Reconfigurable Mechanisms and Robots II, vol. 36, 2016,
snake robot (Cardio-arm) through a subxiphoid port access, in: Proceed- pp. 603–615.
ings of the IASTED International Conference on Robotics and Applications, [325] T. Estier, Y. Crausaz, B. Merminod, M. Lauria, R. Piguet, R. Siegwart, An
2009, pp. 242–246. innovative space rover with extended climbing abilities, Space Robot. 36
[297] A.T. Hillel, A. Kapoor, N. Simaan, R.H. Taylor, P. Flint, Applications of (2000) 333–339.
robotics for laryngeal surgery, Otolaryngol. Clin. North Am. 41 (4) (2008) [326] H. Hong, D. Kim, J. Kim, J. Oh, A locomotive strategy for a stair-climbing
781–791. mobile platform based on a new contact angle estimation, in: IEEE
[298] M.D.M. Kutzer, S.M. Segreti, C.Y. Brown, M. Armand, R.H. Taylor, S.C. International Conference on Robotics and Automation, ICRA, 2013, pp.
Mears, Design of a new cable-driven manipulator with a large open 3819–3824.
lumen: preliminary applications in the minimally-invasive removal of [327] W. Huang, C. Huang, Y. Hung, Sensing pipes of a nuclear power
osteolysis, in: IEEE International Conference on Robotics and Automation, mechanism using low-cost snake robot, Adv. Mech. Eng. 10 (2018) 6.
ICRA, 2011, pp. 2913–2920. [328] X. Hou, Y. Shi, L. Li, Y. Tian, Y. Su, T. Ding, Z. Deng, Revealing the
[299] P. Berthet-Rayne, G. Gras, K. Leibrandt, P. Wisanuvej, A. Schmitz, C.A. mechanical characteristics via kinematic wave model for snake-like robot
Seneci, G.Z. Yang, The i(2) snake robotic platform for endoscopic surgery, executing exploration of lunar craters, IEEE Access 8 (2020) 38368–38379.
Ann. Biomed. Eng. 46 (10) (2018) 1663–1675.
[329] D. Rollinson, A. Buchan, H. Choset, Virtual chassis for snake robots:
[300] P. Liljeback, O. Stavdahl, A. Beitnes, SnakeFighter-development of a water
definition and applications, Adv. Robot. (2012).
hydraulic fire fighting snake robot, International Conference on Control,
[330] S.M. Ryew, S.H. Baik, S.W. Ryu, K.M. Jung, In-pipe inspection robot
Automation, Robotics and Vision, 2007.
system with active steering mechanism, in: IEEE/Robotics-Society-of-
[301] A.A. Transeth, K.Y. Pettersen, Developments in snake robot modeling and
Japan International Conference on Intelligent Robots and Systems, IROS,
locomotion, in: International Conference on Control Automation Robotics
vol. 3, 2000, pp. 1652–1657.
and Vision, vol. 1–5, 2006, p. 339.
[331] Z. Wang, Q. Cao, N. Luan, L. Zhang, Development of an autonomous in-
[302] T. Sato, S. Sakaino, E. Ohashi, K. Ohnishi, Walking trajectory planning on
pipe robot for offshore pipeline maintenance, Ind. Robot 37 (2) (2010)
stairs using virtual slope for biped robots, IEEE Trans. Ind. Electron. 58
177–184.
(4) (2011) 1385–1396.
[332] I. Virgala, M. Kelemen, P. Bozek, Z. Bobovsky, M. Hagara, E. Prada, P.
[303] D. Campbell, M. Buehler, Stair descent in the simple hexapod ‘RHEX’, in:
Oscadal, M. Varga, Investigation of snake robot locomotion possibilities
IEEE International Conference on Robotics and Automation, ICRA, vol. 1,
in a pipe, Symmetry-Basel 12 (6) (2020).
2003, pp. 1380–1385.
[333] E. Kyrkjebo, P. LiljebaCk, A.A. Transeth, A robotic concept for remote
[304] R.C. Luo, H. Ming, T.W. Lin, Erect wheel-legged stair climbing robot for
inspection and maintenance on oil platforms, in: 28th International
indoor service applications, IEEE/RSJ Int. Conf. Intell. Robots Syst. 40 (6)
Conference on Ocean, Offshore and Arctic Engineering, 2009, pp. 667–674.
(2013) 2731–2736.
[334] F. Enner, D. Rollinson, H. Choset, Simplified motion modeling for snake
[305] E. Guizzo, Japan Earthquake: robots help search for survivors, IEEE Spectr.
robots, in: IEEE International Conference on Robotics and Automation,
(2011).
[306] P. Liljeback, K.Y. Pettersen, O. Stavdahl, Hybrid modelling and control of ICRA, 2013, pp. 4216–4221.
obstacle-aided snake robot locomotion, IEEE Trans. Robot. (2010). [335] S. Wakimoto, J. Nakajima, M. Takata, T. Kanda, A micro snake-
[307] R. Garnett, Y. Krishnamurthy, X. Xiong, J. Schneider, R. Mann, Bayesian like robot for small pipe inspection, in: International Symposium on
optimal active search and surveying, in: International Conference on Micromechatronics and Human Science, 2003, pp. 303–308.
Machine Learning, 2012. [336] Y. Bando, H. Suhara, M. Tanaka, T. Kamegawa, K. Itoyama, K. Yoshii,
[308] I. Erkmen, A.M. Erkmen, F. Matsuno, R. Chatterjee, Snake robots to the F. Matsuno, H.G. Okuno, Sound-based online localization for an in-pipe
rescue!, IEEE Robot. Autom. Mag. 9 (3) (2002) 17–25. snake robot, in: IEEE International Symposium on Safety, Security, and
[309] E. Guizzo, Japan earthquake: more robots to the rescue, IEEE Spectr. Rescue Robotics, 2016, pp. 207–213.
(2011). [337] Z.Y. Jiang, Z.S. Bing, K. Huang, A. Knoll, Retina-based pipe-like object
[310] T. Kamegawa, T. Yarnasaki, H. Igarashi, F. Matsuno, Development of tracking implemented through spiking neural network on a snake robot,
the snake-like rescue robot Kohga, in: IEEE International Conference on Front. Neurorobot. 13 (2019).
Robotics and Automation, 2004, pp. 5081–5086. [338] M. Yim, Locomotion with a unit-modular reconfigurable robot, in: Proc
[311] Y. Yamauchi, T. Fujimoto, A. Ishii, A robotic thruster that can handle hairy of the JSME International Conference on Advanced Mechatronics 1994.
flexible cable of serpentine robots for disaster inspection, in: Conference [339] P.J. Benavidez, Design of a Swarm of Autonomous Ground Vehicles for
on Advanced Intelligent Mechatronics, AIM, 2018. Use in Remote Sensing Applications (ProQuest Dissertations and Theses),
[312] H. AlonWolf, Choset, B. Brown, R. Casciola, Design and control of a mobile 2010.
hyper-redundant urban search and rescue robot, Adv. Robot. 19 (3) (2005) [340] J.G. Gomez, Modular Robotics and Locomotion: Application To Limbless
221–248. Robots (Ph.D. thesis), Autonomous University of Madrid, 2008.
[313] A. Ferworn, C. Wright, J. Tran, C. Li, H. Choset, Dog and snake marsupial [341] P. Liljeback, K.Y. Pettersen, O. Stavdahl, A review on modelling, imple-
cooperation for urban search and rescue deployment, in: IEEE Interna- mentation, and control of snake robots, Robot. Auton. Syst. 60 (1) (2012)
tional Symposium on Safety, Security, and Rescue Robotics, 2013, pp. 29–40.
1–5. [342] T.L.T. Chen, S. Liu, J. Yen, A bio-mimetic snake-like robot: sensor based
[314] E.Z. Moore, D. Campbell, F. Grimminger, M. Buehler, Reliable stair climb- gait control, IEEE Workshop on Advanced Robotics and Its Social Impacts,
ing in the simple hexapod ‘RHEX’, in: 19th IEEE International Conference 2008, pp. 1–6.
on Robotics and Automation, ICRA, 2002, pp. 2222–2227. [343] D. Rollinson, A. Buchan, H. Choset, State estimation for snake robots,
[315] R.C. Luo, H. Ming, C.W. Liu, Descending stairs locomotion and somatosen- in: IEEE/RSJ International Conference on Intelligent Robots and Systems,
sory control for an erect wheel-legged service robot, IEEE International 2011, pp. 1075–1080.
Conference on Robotics and Automation, ICRA, 2014, pp. 6356–6361. [344] D. Rollinson, H. Choset, Virtual chassis for snake robots, in: IEEE/RSJ
[316] Z.V. Guo, L. Mahadevan, Limbless undulatory propulsion on land, Proc. International Conference on Intelligent Robots and Systems, 2011, pp.
Natl. Acad. Sci. USA 105 (9) (2008) 3179–3184. 221–226.
25
[345] M. Chris, The New Encyclopedia of Snakes, Princeton University Press, [376] H. Matsuura, M. Mizutame, Y. Moriyama, H. Shimada, S. Hirose, Y.
2007. Umetani, Measuring/grinding system for water turbine runner, in: Proc
[346] P. Moubarak, T.P. Ben, Modular and reconfigurable mobile robotics, Robot. 2nd International Conference on Advanced Research, 1985.
Auton. Syst. 60 (12) (2012) 1648–1663. [377] F. Saito, T. Fukuda, F. Arai, Swing and locomotion control for a two-link
[347] C. Sen, Y. Xiaohu, C. Yuntao, Self-growing and serpentine locomotion of brachiation robot, IEEE Control Syst. Soc. 14 (1) (1994) 5–12.
liquid metal induced by copper ions, ACS Appl. Mater. Interfaces 10 (27) [378] Y. Jiawen, P. Bo, Q. Yunfeng, Prior knowledge snake segmentation of
(2018) 22889–22895. ultrasound images denoised by j-divergence anisotropy diffusion, Int. J.
[348] Z.F. Wang, S.G. Ma, B. Li, Y.C. Wang, A unified dynamic model for Med. Robot. Comput. Assist. Surg. 14 (2018) 5.
locomotion and manipulation of a snake-like robot based on differential [379] M. Tanaka, M. Nakajima, R. Ariizumi, K. Tanaka, Three-dimensional
geometry, Sci. China (Inf. Sci.) 54 (2) (2011) 318–333. steering for an articulated mobile robot with prismatic joints with
[349] A. Wang, M. Deng, Human arm-like robot control using the viscoelasticity consideration of hardware limitations, Adv. Robot. 34 (11, SI) (2020)
of human multi-joint arm, in: International Conference on Computer 767–779.
Sciences & Convergence Information Technology, 2011. [380] K.D. Kotay, D.L. Rus, Navigating 3D steel web structures with an inch-
[350] B. Yun, H. YuanBin, Research of pose control algorithm of coal mine worm robot, in: IEEE/RSJ International Conference on Intelligent Robots
rescue snake robot, Math. Probl. Eng. 4751245 (2018). and Systems-Robotic Intelligence Interacting with Dynamic Worlds, IROS
[351] N. Cohen, J.H. Boyle, Undulatory locomotion, Betascript Publishing 16 (1) 96, vol. 1, 2002, pp. 368–375.
(2009) 321–342. [381] R.M. Voyles, A.C. Larson, Terminatorbot: a novel robot with dual-
[352] Y. Ding, Simulation and Theoretical Study of Swimming and Resistive use mechanism for locomotion and manipulation, IEEE/ASME Trans.
Forces Within Granular Media (Dissertations & Theses - Gradworks), Mechatronics 10 (1) (2005) 17–25.
2011. [382] T. Liptak, I. Virgala, L. Mikova, Modeling and control of two-link snake,
[353] R.D. Maladen, Y. Ding, P.B. Umbanhowar, D.I. Goldman, Undulatory Int. J. Adv. Robot. Syst. 15 (2018) 2.
swimming in sand: experimental and simulation studies of a robotic [383] M. Tanaka, M. Nakajima, Y. Suzuki, Development and control of articu-
sandfish, Int. J. Robot. Res. 30 (7) (2011) 793–805. lated mobile robot for climbing steep stairs, IEEE-ASME Trans. Mechatron.
[354] M. Ishikawa, Y. Minami, T. Sugie, Development and control experiment 23 (2) (2018) 531–541.
of the trident snake robot, IEEE/ASME Trans. Mechatronics (2009). [384] M. Tanaka, K. Tanaka, Shape control of a snake robot with joint limit and
[355] S. Yu, S. Ma, B. Li, Y. Wang, Locomotion control and gaits’ modality of a 3D self-collision avoidance, IEEE Trans. Control Syst. Technol. 25 (4) (2017)
snake-like robot, in: 47th Annual Conference of the Society of Instrument 1441–1448.
and Control Engineers of Japan, SICE, 2008.
[385] M. Tanaka, K. Tadakuma, M. Nakajima, Task-space control of articulated
[356] M. Tesch, Improving Robot Locomotion Through Learning Methods for mobile robots with a soft gripper for operations, IEEE Trans. Robot. 35
Expensive Black-Box Systems, Carnegie Mellon University, 2014. (1) (2019) 135–146.
[357] C. Laschi, M. Cianchetti, Soft robotics: new perspectives for robot
[386] O.M. Omisore, S. Han, L. Ren, A. Elazab, L. Hui, T. Abdelhamid, N.A.
bodyware and control, Front. Bioeng. Biotechnol. 1 (1) (2014) eaah3690.
Azeez, L. Wang, Deeply-learnt damped least-squares (DL-DLS) method for
[358] A. Singh, A. Paigwar, S.T. Manchukanti, Design and motion analysis of
inverse kinematics of snake-like robots, Neural Netw. 107 (2018) 34–47.
compliant omni-directional spherical modular snake robot (COSMOS), in:
[387] E. Kelasidi, A.M. Kohl, K.Y. Pettersen, Experimental investigation of loco-
4th IEEE International Conference on Reconfigurable Mechanisms and
motion efficiency and path-following for underwater snake robots with
Robots, ReMAR, 2018.
and without a caudal fin, Annu. Rev. Control 46 (2018) 281–294.
[359] X. Zhao, L.H. Dou, Z. Su, N. Liu, Study of the navigation method for a
[388] I.L.G. Borlaug, J.T. Gravdahl, J. Sverdrup-Thygeson, K.Y. Pettersen, A. Loria,
snake robot based on the kinematics model with MEMS IMU, Sensors 18
Trajectory tracking for underwater swimming manipulators using a super
(2018) 3.
twisting algorithm, Asian J. Control 21 (1) (2019) 208–223.
[360] M. Zawidzki, J. Szklarski, Transformations of arm-z modular manipulator
[389] H. Marvi, The Role of Functional Surfaces in the Locomotion of
with particle swarm optimization, Adv. Eng. Softw. 126 (2018) 147–160.
Snakes (Ph.D. thesis), Georgia Institute of Technology, 2013.
[361] A. Hayashi, Geometrical Motion Planning for Highly Redundant Ma-
[390] L. Mahadevan, S. Daniel, M.K. Chaudhury, Biomimetic ratcheting motion
nipulators using Acontinuous Model, University of Texas at Austin,
of a soft, slender, sessile gel, Proc. Natl. Acad. Sci. USA 101 (1) (2004)
1991.
23–26.
[362] B. Atakan, 3-D grasping during serpentine motion with a snake-like robot,
in: International Conference on Robotics and Applications, 2005.
[363] H. Huang, G. Yan, G. Ding, Z. Gao, Snake-like robot and analysis for its
motion control, High Technol. Lett. 12 (5) (2002) 90–94.
Jindong Liu is a R&D director at Precision Robotics
[364] M.J. Koopaee, H.B. Van, C. Pretty, Parameters tuning of snake robots
Ltd and an honorary research fellow at the Hamlyn
sidewinding gait using bayesian optimization, in: 4th International
Centre for Robotic Surgery, Imperial College London.
Conference on Control, 2018..
He is interested in fields related to Biologically Inspired
[365] H. Nishimura, K. Funaki, Motion control of three-link brachiation robot
Mobile Robotics, mainly include natural human–robot
by using final-state control with error learning, IEEE/ASME Trans. Mech.
interaction, biomimic robotic fish and compliant ma-
3 (2) (2002) 120–128.
nipulator for healthcare and surgery robotics. He has a
[366] R.H. Sturges, S. Laowattana, A flexible, tendon-controlled device for Ph.D. from the University of Essex where he focused
endoscopy, in: 1991 Internationa Conf On Robotics And Automation, vol. on biologically inspired autonomous robotic fish. He
12, 1991, pp. 121–131. successfully built the first autonomous robotic fish. In
[367] M.D. Patil, Heterogeneous Robot Swarm-Hardware Design and Implemen- 2010, he joined the Hamlyn Centre where he focused
tation (Dissertations & Theses-Gradworks), 2015. on natural human–robot speech interaction, pervasive sensing and surgical
[368] J.P. Ostrowski, The Mechanics and Control of Undulatory Robotic robotics. In 2018, he co-founded Precision Robotics to commercialize a surgical
Locomotion, California Institute of Technology, 1995. robot developed at Hamlyn Centre. His articles have been published in Neuro-
[369] A. Gupta, Lateral undulation of a snake-like robot, in: International computing, Journal of Bionic Engineering, Journal of Neural Network World and
Conference on Advanced Technologies for Communications, ATC, 2007. International Journal of Automation and Computing. He is a senior member of
[370] H. Kimura, S. Hirose, Development of Genbu: active wheel passive IEEE and reviewer of conferences and journals of IEEE and Springer including
joint articulated mobile robot, in: IEEE/RSJ International Conference on IEEE Trans. of Neural Network, IROS and ICRA, etc. He published over 80 papers
Intelligent Robots and Systems, vol. 1, 2002, pp. 823–828. in peer reviewed international conferences and journals. In 2013, his paper had
[371] P. Sears, P. Dupont, A steerable needle technology using curved concentric been evaluated as ’Most cited Article in 2012’ in J. of Bionic Eng.
tubes, in: IEEE/RSJ International Conference on Intelligent Robots and
Systems, 2007, pp. 2850–2856.
[372] W. Ouyang, W. Liang, C. Li, Steering motion control of a snake robot via Yuchuamg Tong graduated from Beijing University of
a biomimetic approach, Front. Inf. Technol. Electron. Eng. 20 (1) (2019) Chemical Technology in 2017 with a bachelor’s de-
32–44. gree in process equipment and control engineering.
[373] G. Wang, W.X. Yang, Y.T. Shen, Adaptive path following of snake robot She is currently a Ph.D. candidate in mechatronic
on ground with unknown and varied friction coefficients, in: IEEE/RSJ engineering from Shenyang Institute of Automation
International Conference on Intelligent Robots and Systems, IROS, 2018. (SIA), Chinese Academy of Sciences (CAS). Her research
[374] C. Li, S. Zhen, H. Sun, Constraints following control of snake robot by interests mainly include redundant sliding manip-
leakage-type adaptive law, in: IEEE International Conference on Advanced ulators, space robots, supernumerary robotic limbs,
Robotics and Mechatronics, ICARM, 2018, pp. 274–279. bio-inspired robots and neural networks.
[375] X.B. Zhang, J.G. Liu, Z.J. Ju, C.G. Yang, Head-raising of snake robots based
on a predefined spiral curve method, Appl. Sci.-Basel 8 (11) (2018) 1–20.
26
Jinguo Liu received his Ph.D. degrees in Mechatronics the Best Paper Award from 2016 China Manned Space Academic Conference, and
from Shenyang Institute of Automation (SIA), Chinese the Outstanding Paper Award from 2017 International Conference on Intelligent
Academy of Sciences (CAS) in 2007. Since January 2011, Robotics and Applications, and the Best Paper Award from 2018 International
he has been a Full Professor with SIA, CAS. He is Conference on Electrical Machines and systems. He is a senior member of the
also holding the Assistant Director position of State IEEE, IEEE Technical Committee on Safety, Security, and Rescue Robotics, IEEE
Key Laboratory of Robotics from 2008 and the Asso- Technical Committee on Marine Robotics, and the Senior Member of Chinese
ciate Director position of Center for Space Automation Mechanical Engineering Society.
Technologies and Systems from 2015. His research He services as the Associate Editor or Technical Editor of several journals
interests include bio-inspired robotics and space robot. such as IEEE/ASME Transactions on Mechatronics, Journal of Field Robotics,
He has authored/coauthored three books, over one Mechanical Sciences, Science China Technological Sciences, Chinese Journal of
hundred papers and fifty patents in above areas. He Mechanical Engineering, and Chinese Journal of Aeronautics.
was awarded the T. J. TARN Best Paper Award in Robotics from 2005 IEEE
International Conference on Robotics and Biomimetics, the Best Paper Award of
the Chinese Mechanical Engineering Society in 2007, the Best Paper Nomination
Award from 2008 International Symposium on Intelligent Unmanned Systems,
27

Liu 2021

Uploaded by

Copyright:

Available Formats

Liu 2021

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Liu 2021

Uploaded by

Copyright:

Available Formats

Robotics and Autonomous Systems 141 (2021) 103785

Contents lists available at ScienceDirect

Robotics and Autonomous Systems

Review of snake robots in constrained environments✩

1. Introduction terrain adaptation of snake robots. For example, the skeleton of

Fig. 1. Classification methods of snake robots.

1.3. Classification methodology 2.1. Body structure of snakes

4. Terrain adaptive control algorithms

4.2.1. Dynamic friction 4.2.2. Slope

5.3. Inspection and exploration

6.2. Amphibious snake robots 6.4. Structural design of snake robots

In response to the environmental adaptation needs of snake

You might also like