biomimetics
Article
Development of a Lizard-Inspired Robot for Mars
Surface Exploration
Guangming Chen 1, *, Long Qiao 1 , Zhenwen Zhou 1 , Lutz Richter 2 and Aihong Ji 1
Citation: Chen, G.; Qiao, L.; Zhou, Z.;
Richter, L.; Ji, A. Development of a
Lizard-Inspired Robot for Mars
Surface Exploration. Biomimetics 2023,
8, 44. https://doi.org/10.3390/
biomimetics8010044
Academic Editor: Hongqiang Wang
Received: 11 December 2022
Revised: 14 January 2023
Accepted: 16 January 2023
Published: 18 January 2023
Copyright: © 2023 by the authors.
Licensee MDPI, Basel, Switzerland.
This article is an open access article
distributed under the terms and
conditions of the Creative Commons
Attribution (CC BY) license (https://
creativecommons.org/licenses/by/
4.0/).
Biomimetics 2023, 8, 44. https://doi.org/10.3390/biomimetics8010044
1 Lab of Locomotion Bioinspiration and Intelligent Robots, College of Mechanical and Electrical Engineering,
Nanjing University of Aeronautics and Astronautics, Nanjing 210016, China
2 Large Space Structures GmbH, Hauptstr. 1e, D-85386 Eching, Germany
* Correspondence: guangming2017@nuaa.edu.cn
Abstract: Exploring Mars is beneficial to increasing our knowledge, understanding the possibility
of ancient microbial life there, and discovering new resources beyond the Earth to prepare for
future human missions to Mars. To assist ambitious uncrewed missions to Mars, specific types
of planetary rovers have been developed for performing tasks on Mars’ surface. Due to the fact
that the surface is composed of granular soils and rocks of various sizes, contemporary rovers can
have difficulties in moving on soft soils and climbing over rocks. To overcome such difficulties, this
research develops a quadruped creeping robot inspired by the locomotion characteristics of the desert
lizard. This biomimetic robot features a flexible spine, which allows swinging movements during
locomotion. The leg structure utilizes a four-linkage mechanism, which ensures a steady lifting
motion. The foot consists of an active ankle and a round pad with four flexible toes that are effective
in grasping soils and rocks. To determine robot motions, kinematic models relating to foot, leg,
and spine are established. Moreover, the coordinated motions between the trunk spine and leg are
numerically verified. In addition, the mobility on granular soils and rocky surface are experimentally
demonstrated, which can imply that this biomimetic robot is suitable for Mars surface terrains.
Keywords: Mars robot; flexible spine; foot trajectory; planetary rover; space exploration
1. Introduction
Space exploration is important for humankind to contribute to comparative planetol-
ogy and discovering new resources out of the Earth, and future human missions to bodies
outside the Earth are being prepared [1,2]. To achieve these goals, a series of activities
have been ongoing by spacefaring actors. After the recent successes in Mars exploration by
NASA with several generations of six-wheeled vehicles of different size ranges, the United
States is currently developing the ‘VIPER’ four-wheeled rover with active suspension for
exploration of the polar regions of the Moon [3]. European nations are ready to send the
‘Exomars’ Rover to Mars which will perform subsurface drilling [4]. China has operated
‘YuTu’ rovers on the moon and ‘ZhuRong’ on Mars [5]. The United Arab Emirates is sending
the ‘Rashid’ rover to the moon [6]. Mars surface exploration currently mainly comprises
investigations of the surface geology and geomorphology [7], study of mineralogy and
chemistry of key material units at the landing sites through sample acquisition and analy-
sis [8], and instrument deployment. In the future, infrastructure construction in support
of human exploration is expected to constitute another group of tasks. In support of such
missions, planetary surface rovers are indispensable. The Mars surface is characterized by
soil-like material and rocks of different sizes [9]. In the attempt to adapt to the Mars surface
terrains, several types of planetary rovers have been proposed.
Contemporarily, wheeled structures are commonly used for planetary surface rovers.
The advantages of such rovers are high agility, high speed, and a simple control system.
Nonetheless, wheeled rovers may encounter severe sinkage and embedding in very soft
https://www.mdpi.com/journal/biomimetics
Biomimetics 2023, 8, 44 2 of 14

weak soils [10]. The ‘Zhurong’ rover possesses the strategy of escaping from sinkage using
the active control suspension system for lifting of wheels [11]. Nonetheless, it requires large
torque to operate an embedded wheel. Leg-wheeled robots are based on the structures and
mechanism of animal legs, so that they can promote higher maneuverability in cornering
and avoiding obstacles (e.g., the NASA’s ‘Athlete’ [12], and ‘SherpaTT’ [13]). Nevertheless,
the wheels can suffer from high slip for the soft soils [14]. Another type of the legged
rover, Spacebok [15], has a simple foot structure and better adaptability on soft soil surface.
However, it is unqualified to adapt to non-flat terrains. The rock-climbing robots which
adopt grippers on the foot toes, such as Lemur 3 [16] and Nagaoka’s robot [17], are capable
of climbing on vertical, and inverted rock surfaces. Nevertheless, their trunks are made of
rigid structures, resulting in a poorer balancing capability and walking speed compared to
real quadruped animals.
Among the legged rovers, the quadrupedal robots are distinguished from other legged
robots in producing effective motions with relatively uncomplicated structure [18]. Our
previous model of a desert Chameleon-inspired robot has successfully demonstrated the
potential of adapting to Mars surface terrains [19]. Moreover, the flexible spine enables
larger strides compared to rigid trunks. However, the stand–walking feature of the robot
structure results in higher center of mass, causing unstable motion as walking speed
increases. Furthermore, the lift height of each leg is not sufficient, which reduces the
ground clearance in traversing rocks. For increasing the adaptability for Mars’ surface
terrains, an innovative structure of the quadruped robot based on other inspiration can
be used.
Toward future applications of Mars surface geomorphology exploration and detection
with loaded scientific instruments, the work described here contributes a biomimetic
quadruped robot based on the structure and mechanism of a creeping animal desert
lizard which moves efficiently on both granular soils and rocky surface. It can produce a
steadier walking motion due to a lower center of mass and updated joint structures. To
ensure coordinated movements among the foot, leg, and spine, the combined kinematics
model are established based on their simultaneous motions during walking. Finally, the
mobility on granular soils and rocks in analogy to the Mars surface are experimentally
demonstrated. The results infer that the biomimetic robot is promotable for adapting to
Mars surface terrains.

2. Biomimetic Structure Design

To propose a bionic creeping quadruped robot based on a desert lizard, the loco-
motion characteristics are illustrated next. Referring to biological characteristics, the
biomimetic structure design of flexible spine, leg, and multi-toe foot is elaborated. Finally,
the biomimetic structure of a lizard-inspired robot is presented.

2.1. Biological Characteristics

Figure 1 shows a desert lizard that can efficiently move on granular sands and rock
surface. The efficient movements on granular sand and rocky surfaces are mainly ascribed
to flexible spine, creeping legs, and graspable toes. The spine contains a series of active
flexible joints. Each leg can be decomposed of swing thigh and shank. During walking,
the shanks of the leg swing in synchronization with motions of the flexible spine [20]. For
adapting to the soft desert surface, they use large contact pads and multiple toes [21]. The
footpad can increase the contact area with respect to granular sands to prevent subsidence.
The grasping toes can solidify the loose sands and thus achieve steady reaction forces.
Biomimetics 2023,8,8,44x FOR PEER REVIEW
Biomimetics2023, 33 of
of14
14

Figure 1. A desert lizard (Photo credit: Thomas C. Brennan) [21],.

2.2. Biomimetic Structure

To design a flexible spine as that of the lizard, four servos are used to generate two
left–right motions and two up–down motions for the symmetric biomimetic flexible
spine [22], as shown in Figure 2a. These four gears are connected by different brackets.
Eight springs are used to connect the hip joints and the middle bracket of the spine,
which increase the loading capability and reduce vibrations. For achieving a lizard-like
Figure
creeping
Figure 1.1.AAmotion,
desertlizard
desert lizard
each (Photo credit:Thomas
leg adopts
(Photo credit: Thomas
two hinges C.Brennan)
C. Brennan)
and each [21],.require one gear to produce a
[21].
swinging movement. The hip joint between the spine and leg is composed of two servos
2.2.
2.2. Biomimetic Structure
and Biomimetic
a four-linkage Structure
mechanism which can promote a stable lift. A servo is also used to join
To design
and theflexible
To design
the thin a ashank.
flexible spine
spineas as
that of the
that lizard,
of the fourfour
lizard, servos are used
servos to generate
are used two left–
to generate two
right motions
left–right
The thigh and
motions
andtwo andup–down
footpadtwoare motions motions
up–down
connected forbytheansymmetric biomimetic
for therotational
active symmetric flexible
that is spine
biomimetic
ankle [22],
flexible
composed
as
ofshown
spine [22],
a servo inand
Figure
as shown 2a. in
a bearing, These four2a.
Figure
as given gears
inThesearefour
Figure connected
2b. A gears
servo byisdifferent
are connected brackets.
placed between Eight
by different
the springs
brackets.
footpad and
are used
Eight to
springsconnect
are the
used hip
to joints
connect and the
the middle
hip jointsbracket
and of
the
the bearing. It connects these four tension springs by a rope and generates active rota-the spine,
middle which
bracket increase
of the the
spine,
loading
which capabilitythe
increase
tional motions toand reduce
loading
adjust vibrations.
to capability
contact Forreduce
and
positions achieving atoes.
lizard-like
vibrations.
of foot The creeping
Forfoot
achieving motion, each
a lizard-like
pad connects four
leg adopts
creepingtoes.
flexible two hinges
motion,
Each each and each require
leg adopts
toe consists of twotwoone gear
hinges
hinges to
and produce
and a swinging
eachatrequire
a claw the tip.one movement.
For gear The
to produce
each toe, hipa
a tension
joint
springbetween
swinging is used the spine
movement.
to connect and
The hip leg is
thejointcomposed
pad between
and the theof two
claw servos
spine
[23].and and a four-linkage
leg is composed
A servo is embedded mechanism
of two servos
inside the
which
and a can
round promote
four-linkage
pad, and when amechanism
stable lift. which
the rope Ais servocanis also
tensioned byused
promote atostable
the servo,jointhethe
lift.thin andgrasp
A servo
four toes the shank.
is also used to join
downward.
the thin and the shank.
The thigh and footpad are connected by an active rotational ankle that is composed
of a servo and a bearing, as given in Figure 2b. A servo is placed between the footpad and
the bearing. It connects these four tension springs by a rope and generates active rota-
tional motions to adjust to contact positions of foot toes. The foot pad connects four
flexible toes. Each toe consists of two hinges and a claw at the tip. For each toe, a tension
spring is used to connect the pad and the claw [23]. A servo is embedded inside the
round pad, and when the rope is tensioned by the servo, the four toes grasp downward.

Figure2.2.Desert
Figure Desertlizard-inspired
lizard-inspiredrobot:
robot:(a)
(a)biomimetic
biomimeticstructure
structureand
and(b)
(b)break
breakout
outsection
sectionofofankle.
ankle.

The thigh and

3. Kinematics footpad are connected by an active rotational ankle that is composed of
Modelling
a servoToand a bearing,
evaluate the as given in Figure
performances of 2b.
the Abiomimetic
servo is placed between
robot, the footpad
kinematics studiesandforthe
the
bearing. It connects these four tension springs by a rope and generates active
spine, leg, and foot are conducted. Moreover, the foot trajectory for forward motion is rotational
motions to adjust to contact positions of foot toes. The foot pad connects four flexible toes.
elaborated and the workspace of the foot is estimated. Finally, the gait planning for robot
Each toe consists of two hinges and a claw at the tip. For each toe, a tension spring is used
forward walking is determined.
to connect the pad and the claw [23]. A servo is embedded inside the round pad, and when
the rope isDesert
Figure tensioned by the servo, the four toes grasp downward.
3.1. Leg2.Motion lizard-inspired
Determination robot: (a) biomimetic structure and (b) break out section of ankle.
According to the biomimetic structure, the spine has four joints, and each leg has
3.3.Kinematics
KinematicsModelling
Modelling
fourTojoints. All the
evaluate joints
thethe of the robot
performances are denoted in
of the the schematic robot diagram of Fig-
To evaluate performances of biomimetic robot,
the biomimetic kinematics
robot, studies
kinematics for the for
studies spine,
the
leg, and foot are conducted. Moreover, the foot trajectory for forward motion is elaborated
spine, leg, and foot are conducted. Moreover, the foot trajectory for forward motion is
and the workspace of the foot is estimated. Finally, the gait planning for robot forward
elaborated and the workspace of the foot is estimated. Finally, the gait planning for robot
walking is determined.
forward walking is determined.
3.1. Leg Motion Determination
3.1. Leg Motion Determination
According to the biomimetic structure, the spine has four joints, and each leg has four
joints.According to the
All the joints biomimetic
of the robot are structure,
denoted inthe
thespine has four
schematic joints,
robot and of
diagram each leg has
Figure 3a.
four joints. All the joints of the robot are denoted in the schematic robot diagram of Fig-
Biomimetics 2023, 8, x FOR PEER REVIEW

Biomimetics 2023, 8, 44 4 of 14
ure 3a. As the robot moves, the ends of the spine rotate with respect to the centr
Figure 3b illustrates the coordinate systems for spine end and whole leg, in which
As the
of the robot
spinemoves,
is thethe
baseendscoordinate
of the spine system.
rotate with respect
The spineto the
endcentral line. Figurewith
is connected 3b the h
illustrates the coordinate systems for spine end and whole leg, in which the end of the
via a four-linkage structure. It can be rotated when driven by a motor, which pr
spine is the base coordinate system. The spine end is connected with the hip joint via a
an
four-linkagedegree
active ofItfreedom
structure. (O1).when
can be rotated Thedriven
otherbyend utilizes
a motor, whicha produces
connecting rod (O2) w
an active
passively rotated(Oin
degree of freedom 1 ).correspondence to the
The other end utilizes steady up–down
a connecting motion
rod (O2 ) which of the leg. Th
is passively
the angles of active degrees of freedom (O1) and passive degrees of angles
rotated in correspondence to the steady up–down motion of the leg. Therefore, the freedom (O1
of active degrees of freedom (O1 ) and passive degrees of freedom (O1 ) are of the same
the same magnitude and in opposite directions. A video of the hip joint motion
magnitude and in opposite directions. A video of the hip joint motion is provided to
vided tothe
illustrate illustrate
hip joint the hip (Supplementary
motions joint motions File).
(Supplementary
The knee and foot,File). The kneehave
respectively, and foot,
tively, have
the active the active
freedoms O3 andfreedoms
O4 . O3 and O4.

Figure
Figure 3.3.Leg
Leg motion
motion analysis:
analysis: (a) of
(a) joints joints of the
the robot robotand
diagram diagram and (b) systems.
(b) coordination coordination systems

Based on the coordinate system shown in Figure 3, the D–H parameters can be obtained
Based on the coordinate system shown in Figure 3, the D–H parameters can
as given in Table 1. By transformations of the pose matrix, the positions of each joint can be
tained
resolvedasas given
the clawin[24].
Table 1. By transformations of the pose matrix, the positions
joint can be resolved as the claw [24].
Table 1. D–H parameter table.
Table 1. D–H parameter table.
Torsional Angle Joint Distance
Link Link Length L Joint Angle θn
α dn
Link Torsional◦ Angle α Link Length L Joint Angle 𝜽𝒏 Joint Distan
1 90 L0 θ0 0
12 90°
0◦ L1 L0 θ1 𝜃 0 0
◦
2 3 − 90
0° L h L1 θ 2 𝜃 0 0
4 0◦ L2 θ3 0
35 −90°
0◦ L3 Lh θ4 𝜃 0 0
4 0° L2 𝜃 0
To5 better control the movement
0° of the robot, L
the 𝜃 corresponding
3 position of each joint 0
to the robot foot trajectory must be predetermined. Assuming that the spine swings at a
constant speed, and denoting θmax the maximum rotational angle of the left trunk end, the
To better control the movement of the robot, the position of each joint corresp
swinging angle θ0 as a function of time t can be expressed by
to the robot foot trajectory must be predetermined. Assuming that the spine swin
𝜃 θmax ∗the
 π
constant speed, and denoting θ0 = sin maximum
t∗ rotational angle of the (1)left tru
4
the swinging angle 𝜃 as a function of time t can be expressed by
For conveniently solving the freedoms of the leg, the hip joint (O1 ) is used as base
π joint
coordinate system. The trajectory that was expressed by the waist  (O0 ), is transformed
θ 0 = θ max * sin  t * 
 4
For conveniently solving the freedoms of the leg, the hip joint (O1) is used
coordinate system. The trajectory that was expressed by the waist joint (O0), is
formed to estimate positions of the three active joints (O1, O3, and O4). The foot p
that was established as referring to the waist joint (O0) can be expressed by hip joi
Biomimetics 2023, 8, 44 5 of 14

to estimate positions of the three active joints (O1 , O3 , and O4 ). The foot position that was
established as referring to the waist joint (O0 ) can be expressed by hip joint (O1 ), which is
0 0
 PX ∗ cos(θ0 ) +0 PY ∗ sin(θ0 ) − L0
 
1P 
 1 0,X  PZ

1
P0 = 01 T −1 ∗ 0 P =  P0,Y  = 0 P ∗ sin( θ ) − 0 P ∗ cos( θ ) (2)
1P 
 X 0 Y 0
0,Z 1


Using the hip joint (O1 ) as the base coordinate system, the position of the foot joint
(O4 ) can be obtained,
  
0 
 L1 ∗ cos(θ1 ) + L2 ∗ cos(θ3 ) + L3 ∗ cos(θ3 + θ4 )
0 L1 ∗ sin(θ1 )

1
P = 15 T ∗ 
 
0 =  (3)

 − L2 ∗ sin(θ3 ) − L3 ∗ sin(θ3 + θ4 )
1 1


Using inverse dynamics, the trajectories of three active joints (O1 , O3 , and O4 ) can be
obtained based on the hip joint (O1 ) [25], which is given by Equation (4).

θ1 = arcsin 1 PY /L1




θ2 = − arcsin 1 PY /L1




  
1 P 2 − L 2 ∗( sin ( θ ))2

Z 3 4
θ3 = arccos (4)

 ( )
L3 ∗sin(θ4 )∗ 1 PX − Lh − L1 ∗cos(θ1 ) −1 PZ ∗( L2 + L3 ∗cos(θ4 ))
1 P − L − L ∗ cos ( θ ) 2 +1 P 2 − L 2 + L 2
  
 ( X h 1 1 ) Z ( 2 3 )
θ4 = arccos


2∗ L2 ∗ L3


Assuming the waist joint (O0 ) as the base coordinate system and substituting joint
functions 1 PX , 1 PY , and 1 PZ in θ1 , θ2 , θ3 , and θ4 with position functions 1 P0,X , 1 P0,Y , and
1P
0,Z of Equation (2), the motion function of each joint can be solved in correspondence to
the designated foot trajectory.
To promote a smooth up–down motion of the foot with respect to the irregular ter-
rains of the Mars surface, the composite non-impact curve is generally recommended to
determine the foot trajectory [26]. Moreover, this type of trajectory can be adjusted by
changing the step length and step height in accordance with the robot motion mechanism.
The designated non-impact composite curve trajectory is given by Equations (5)–(7) [27],
   
s ∗ t − sin(2∗π ∗t/T ) t
≥ 0& t
≤ 1
T 2∗ π  T T 2
PX (t) = (5)
t t t

1
 PX 1 − T 2 ≥ &
T T ≤ 1

PY = ( L0 + L1 + L2 ) (t ≥ 0&t ≤ T ) (6)
  
sin(4∗π ∗ 1t/T
−k )

t/T t t 1− k
2 ∗ h ∗ − T ≥ 0& T ≤




 1− k 4∗ π 2
 
PZ (t) = h Tt ≥ 1− k
& Tt ≤ 1+ k (7)

 2 2

t 
PZ 1 − Tt 1+ k t

≥ & ≤ 1


T 2 T

In which s represents stride length, h represents leg lift height, u is the time variable,
T is the period of movement, and k is the time parameter of forward movement. When
t/T = 1, the swing phase of the foot trajectory is completed.

3.2. Work Space Estimation

To estimate movement space of the legs of the robot, the robot toolbox [28] can be used.
The three lengths, namely L1 , L2 , and L3 of the robot leg are 41 mm, 74 mm, and 150 mm,
respectively. For the quadruped robot with a flexible spine, the foot movement space is
affected by the swinging space of the spine. According to the applied robot leg lengths, the
 T is the period of movement, and k is the time parameter of forward movement. When t/T
= 1, the swing phase of the foot trajectory is completed.

3.2. Work Space Estimation

To estimate movement space of the legs of the robot, the robot toolbox [28] can be
Biomimetics 2023, 8, 44
used. The three lengths, namely 𝐿 , 𝐿 , and 𝐿 of the robot leg are 41 mm, 74 mm,6 and of 14

150 mm, respectively. For the quadruped robot with a flexible spine, the foot movement
space is affected by the swinging space of the spine. According to the applied robot leg
predicted
lengths, motion
the ranges
predicted of theranges
motion legs swinging with
of the legs a flexible
swinging spine
with on the X,
a flexible Y, and
spine Z axis
on the X,
areand
Y, 184 Zmm,
axis250
aremm, and 80
184 mm, mm,
250 mm,respectively,
and 80 mm, as respectively,
shown in Figure 4. For aingiven
as shown Figurerobot foot
4. For a
step size of 40 mm, height of 40 mm, and swing phase of 2 s, the swing phase
given robot foot step size of 40 mm, height of 40 mm, and swing phase of 2 s, the swingtrajectory can
be obtained
phase (Figure
trajectory can 4b) by using (Figure
be obtained Equations
4b)(5)–(7).
by using Equations (5)–(7).

Movement of
Figure 4. Movement of leg and foot: (a)
(a) workspace
workspace of
of leg
leg and
and (b) wobble phase of foot.

3.3. Gait Planning

3.3. Gait Planning
The previous section analyzed the forward and inverse kinematics of part of the spine
The previous section analyzed the forward and inverse kinematics of part of the
to the legs of the robot, which have been solved. In order to enable the whole robot to stably
spine to the legs of the robot, which have been solved. In order to enable the whole robot
move forward, effective gait planning is crucial [27]. Based on the crawling lizard gait at the
to stably move forward, effective gait planning is crucial [27]. Based on the crawling liz-
open space, a triangular gait is implemented. Figure 5 illustrates the gait plan of the robot
ard gait at the open space, a triangular gait is implemented. Figure 5 illustrates the gait
in a straightforward motion. At step 1 , the right front and left hind legs subsequently lift
plan of the robot in a straightforward motion. At step ①, the right front and left hind legs
forward, and meanwhile the spine bends right. Next, the left hind and left front feet move
subsequently lift forward, and meanwhile the spine bends right. Next, the left hind and
forward as the spine returns to the straight line (Step 2 ). Then, the right hind leg moves
left front feet move forward as the spine returns to the straight line (Step ②). Then, the
forward and the spine bends left (Step 3 ). Finally, the robot returns to the initial position
right hind leg moves forward and the spine bends left (Step ③). Finally, the robot returns
while the right front foot moves forward (Step 4 ). For the four steps, the positions of the
to the initial position while the right front foot moves forward ④). For the
(Step between four
same end of the spine are zoomed-in Figure 5b. The advancing
Biomimetics 2023, 8, x FOR PEER REVIEW distance each
7 ofstep
14
steps, the positions of the same end of the spine are zoomed-in Figure 5b. The advancing
is set as s. v is the forward velocity and θ0 is the swinging angle about central line.
distance between each step is set as s. v is the forward velocity and θ0 is the swinging an-
gle about central line.

Figure
Figure 5.
5. Gait
Gaitplanning:
planning: (a)
(a)four
foursteps
stepsof
ofthe
thegait
gaitand
and (b)
(b) illustration
illustration of
of variations
variations in
in the
the trunk
trunk end.
end.

During
During straight
straight walking,
walking, the
the velocity
velocity of
of the spine is maintained as constant. The The
velocity
velocity of
of the
the front
front spine
spine consists
consists of
of two
two components,
components, i.e.,
i.e., the
the swinging
swinging andand the
the forward
forward
velocity
velocity of
of the
the spine.
spine. Therefore,
Therefore,the
themovement
movementof ofeach
eachleg
legmust
mustbebecalculated
calculatedseparately.
separately.
The
The left
left hind
hind leg
leg moves
moves forward
forward when
when the spine bends to the right. Assuming
Assuming that that the
the
position of
position of the
the entire
entire spine
spine remains
remains unchanged,
unchanged, the
the wobble
wobble phase
phase of
of the
the left
left hind
hind foot
foot can
can
be regarded
be regarded as as the
the utilized
utilized non-impact
non-impact motion,
motion, and
and the
the support
support phase
phase can
can bebe aa forward
forward
walking motion. Thus, the motion curve of the foot in the forward direction (𝑃 ) is mod-
ified as

  t sin ( 2 * π * t T )   t t 1
 L3 − s *  −   ≥ 0& ≤ 
 T 2 *π  T T 2
PX ( t ) =  (8)
 t −2  t 1 t 
 L3 − s + s *  ≥ & ≤ 4
 6  T 2 T 
   t sin ( 2 * π * t T )   t t 1
 L3 − s *  −   ≥ 0& ≤ 
 T 2 *π  T T 2
PX ( t ) = 
 t −2  t 1 t 
Biomimetics 2023, 8, 44
 L3 − s + s *  ≥ & ≤ 4
7 of 14
 6  T 2 T 

walking motion. Thus, the motion curve of the foot in the forward direction (PX ) is
4.modified
Verification
as by Simulation
   
 L3 − s ∗ t − sin(2∗π ∗t/T ) t
≥ 0& t
≤ 1
To verify Pthe theoretical results
X (t) =
T 2of
∗π robotT motions,
T 2a simulation approach
(8)
t −2 t 1 t
L
Adams software can be used [25,27].
 3 − s + s ∗
This ≥
6 section
T &
2 T ≤ 4
models the leg motion according
adopted theoretical results of kinematics studies. Moreover, the walking charact
4. Verification by Simulation
by the prescribed gait are predicted. In addition, the grasping motion of the foot
To verify the theoretical results of robot motions, a simulation approach using Adams
simulated.
software can be used [25,27]. This section models the leg motion according to the adopted
theoretical results of kinematics studies. Moreover, the walking characteristics by the
4.1. Leg Motion
prescribed gait are predicted. In addition, the grasping motion of the foot toes is simulated.
To efficiently verify the results of the leg motion by simulation, the structure
4.1. Leg Motion
leg and spine end (Figure 3b) can be simplified, which is shown in Figure 6. D
To efficiently verify the results of the leg motion by simulation, the structure of the leg
colors represent
and spine end (Figure different
3b) can be parts of thewhich
simplified, robot. The corresponding
is shown rotation
in Figure 6. Different colors pairs a
drives
representfordifferent
the joints
partsbetween each
of the robot. Thepart are applied.
corresponding The
rotation motion
pairs and thecurve
drivesof
forEquatio
the joints between each part are applied. The motion curve of Equation
implemented in simulations and the corresponding motion trajectory of the foo (8) is implemented
in simulations and the corresponding motion trajectory of the foot is obtained as shown by
tained as shown by the red curve.
the red curve.

Biomimetics 2023, 8, x FOR PEER REVIEW 8 of 14

Figure 6.6.Foot
Figure Foottrajectory simulation.
trajectory simulation.
Figure 7 shows the obtained trajectory for one cycle period. On the X axis, the foot
centerFigure 7 shows
decreases fromthe56obtained
mm to 16 trajectory
mm afterfor2 one cycle
s, and period.
then Onincreases
linearly the X axis,
to the foot
56 mm.
center
The magnitude is maintained at 140 mm on the Y axis, which is consistent with theThe
decreases from 56 mm to 16 mm after 2 s, and then linearly increases to 56 mm. set
magnitude
foot movement is maintained
trajectory.at 140magnitude
The mm on theofYthe axis, which
motion is consistent
curve on the Z with the set foot
axis increases to
movement trajectory. The magnitude of the motion curve on the Z axis increases to 40 mm
40 mm after 0.75 s, and returns to zero at 2 s. The variations in X and Z axes are consistent
after 0.75 s, and returns to zero at 2 s. The variations in X and Z axes are consistent with the
with the prescribed motion as shown in Figure 4b. Thus, the correctness of the theoretical
prescribed motion as shown in Figure 4b. Thus, the correctness of the theoretical analysis
analysis is verified.
is verified.

Figure 7. Variation
Figure 7. Variation of
of foot
foot positions
positions in
in three
threeaxes
axesin
insimulations.
simulations.

4.2. Gait Simulation

The gait simulation involves material parameters and contact parameters between
the foot and ground surface. The foot structures are printed using resin materials. The
spring material is manganese steel. To estimate the effectiveness of theoretical gait plan-
 Biomimetics 2023, 8, 44 8 of 14

4.2. Gait Simulation

The gait simulation involves material parameters and contact parameters between the
foot and ground surface. The foot structures are printed using resin materials. The spring
material is manganese steel. To estimate the effectiveness of theoretical gait planning, the
ground surface can be modelled as concrete. The contact parameters are referred to the
Adams software manual [29]. The simulation parameters are given in Table 2.

Table 2. Simulation parameters.

Categories Parameters Values

Steel density 7801 Kg/m3
Steel Young’s Modulus 2.07 GPa
Materials Steeel poisson ratio 0.29
Spring stiffness 4.6 N·mm/◦
Concrete 2000 Kg/m3
Stiffness 4855 N/mm
Damping 80 N s/mm
Penetration depth 0.1 mm
Contact Static coefficient 0.7
Dynamic coefficient 0.57
Static transition velocity 0.1 mm/s
Friction transition velocity 10 mm/s

Using the parameters in Table 2 and the predetermined gait (Figure 5), the robot
motions for one cycle are simulated as shown in Figure 8. It demonstrates that the legs and
spine of the robot can cooperate with each other and move forward stably. The period of a
full cycle of the robot movement is 8 s as prescribed in theory. Therefore, the simulation
Biomimetics 2023, 8, x FOR PEER REVIEW
results verify that the theoretical gait plan can be used for the robot to effectively perform
forward walking.

Figure8.8.
Figure Gait
Gait simulation
simulation of forward
of robot robot forward
walking. walking.

To assess the stability of the forward motion, the variations in the robot’s center of
To assess the stability of the forward motion, the variations in the robot’s c
mass for five cycles are shown in Figure 9. It can be seen that the robot generates periodic
mass for
changes fiveit cycles
when areInshown
advances. in the
each cycle, Figure 9. It can
increment of thebe seen
robot in that the robot
the forward generates p
direction
changes
(X when
axis) is 66.7 mm,itwhich
advances. Inthe
is close to each cycle, the
estimations increment
using of The
Equation (8). thechange
robot toinspine
the forwar
swinging motions (Y axis) is 45.2 mm. The deviation in the Y axis direction
tion (X axis) is 66.7 mm, which is close to the estimations using Equation (8). The is 2.04 mm,
which is caused by the slippery aspect of the foot. The position in the vertical direction
to spine swinging motions (Y axis) is 45.2 mm. The deviation in the Y axis directio
(Z axis) is maintained as constant, which is because no up and down movements of the
mm, were
spine which is causedThe
implemented. by average
the slippery
movingaspect
speed isof the foot.
calculated at The position
8.34 mm/s, whichin can
the vertic
tion (Z axis) is maintained
be also estimated by Equation (8). as constant, which is because no up and down movem
the spine were implemented. The average moving speed is calculated at 8.34
which can be also estimated by Equation (8).
 the (X
tion spine
axis) were implemented.
is 66.7 mm, which is closeThe average
to the moving
estimations speed is(8).
using Equation calculated
The chang
which
to spine can be also
swinging estimated
motions (Y axis)by Equation
is 45.2 mm. The (8).
deviation in the Y axis direction is 2.0
mm, which is caused by the slippery aspect of the foot. The position in the vertical direc
tion (Z axis) is maintained as constant, which is because no up and down movements o
Biomimetics 2023, 8, 44 the spine were implemented. The average moving speed is calculated at 9 of8.34
14 mm/s
which can be also estimated by Equation (8).

Figure 9. Variations in the center of mass for five walking cycles.

4.3. Grasping
Figure
Figure 9.9.Variations
VariationsSimulation
in the
in the center
center of mass
of mass forwalking
for five five walking
cycles. cycles.

4.3. Grasping Simulationinitially rest in a relaxed open state, and will bend dow
The foot toes
4.3. Grasping Simulation
driven torques.
The foot Figure
toes initially rest 10
in ashows the simulation
relaxed open state, and will results, in which
bend downward
The foot toes initially rest in a relaxed open state, and will bend downward by th
the red
by the
driven torques. Figure 10 shows the simulation results, in which the red curves are the
trajectories
driven torques. for the bending motion. The maximum rotation angle (θ) of th
trajectories for the Figure
bending10 showsThe
motion. themaximum
simulation results,
rotation in which
angle thetoes
(θ) of the red is curves
about are th
120°,
120 which
trajectories
◦ , which isfor istheexpected
expectedbending
based onbased
motion. on
Thethe
the structure structure
maximum
and and
rotation
geometric geometric
sizes.angle (θ) of
However, sizes.
in the toesHowev
practice, is abou
120°,
the whichangles
thebending
bending is expected
angles based
are
are highly on the
highly
affected structure
byaffected and
by
the stiffness geometric
ofthe
the stiffnesssizes. However,
of the
contact surface. contact in practice
surfac
the bending angles are highly affected by the stiffness of the contact surface.

Figure 10. Grasping behavior simulation: (a) retrieving state and (b) grasping state.
Figure
Figure 10. 10. Grasping
Grasping behavior
behavior simulation:
simulation: (a) retrieving (a)
stateretrieving statestate.
and (b) grasping and (b) grasping state
5. Experimental Tests
5. Experimental Tests
In accordance with
5. Experimental Teststhe dimensions of the structure model in Figure 2, a prototype o
In accordance with the dimensions of the structure model in Figure 2, a prototype of
the biomimetic robot was fabricated. Using a simulated test bed, the grasping behavior
the biomimetic robot was fabricated.
In accordance with theUsing a simulated
dimensions oftest
thebed, the grasping
structure behaviors
model of
in Figure 2,
the foot toes with respect to granular soils were evaluated. Moreover, the mobility on the
the biomimetic
surface analogy to therobot was fabricated.
Mars terrains were tested. Using a simulated test bed, the grasp

5.1. Fabrication
The components for foot, leg, and spine were fabricated using resin materials by
3D printing. A 32-way servo control panel is used to send signals to servos to enable
component motions. A lithium battery (12V) is used to supply electricity to the control
panel. Meanwhile, a voltage regulator module (XL4016E1DC-DC) is used to adjust central
interface and power supply. The type of the 24 servos are PTK 7455MG-D12g which are
commercially purchased from the company of Mayatech. Using high accuracy potentiome-
ters, the deviations of rotation angles from the servos are transferred to the control panel,
where the signals are processed and sent back to servos to adjust the positions and voltages.
The servo adopts asynchronous serial bus communication, so that the port can send and
receive data through different wires. In this manner, simple and long-distance data transfer
can be achieved using less complicated wire configuration [30]. The fabricated biomimetic
robot prototype is shown in Figure 11.
 commercially purchased
commercially purchased from
from thethe company
company of of Mayatech.
Mayatech. Using
Using high
high accuracy
accuracy potenti-
potenti-
ometers, the
ometers, the deviations
deviations of of rotation
rotation angles
angles from
from the
the servos
servos are
are transferred
transferred toto the
the control
control
panel,where
panel, wherethe
thesignals
signalsare
areprocessed
processedandandsent
sentback
backto toservos
servosto
toadjust
adjustthe
thepositions
positionsand and
voltages.The
voltages. Theservo
servoadopts
adoptsasynchronous
asynchronousserialserialbus
buscommunication,
communication,so sothat
thatthe
theport
portcan
can
Biomimetics 2023, 8, 44
sendand
send andreceive
receive data
data through
through different
different wires.
wires. In
Inthis
this manner,
manner,simple
simple and
and long-distance
long-distance
10 of 14
data transfer
data transfer can
can be
be achieved
achieved using
using less
less complicated
complicated wire
wire configuration
configuration [30].
[30]. The
The fabri-
fabri-
catedbiomimetic
cated biomimeticrobot
robotprototype
prototypeisisshown
shownin inFigure
Figure11.11.

Figure11.
Figure
Figure 11.The
11. Thefabricated
fabricatedprototype
prototypeof
prototype ofthe
of thelizard-inspired
the lizard-inspiredquadruped
lizard-inspired quadrupedrobot.
quadruped robot.
robot.

5.2.
5.2.Mobility
5.2. MobilityTest
Mobility Test
Test
Referring
Referringto
Referring tothe
to theMartian
the Martiansoil
Martian soilproperties
soil [31,32],
[31,32],aaatestbed
properties[31,32],
properties testbed was
testbedwas made
wasmade
madeforfor mobility
formobility test.
mobilitytest.
test.
Figure 12
Figure 12
Figure presents
12 presents the
presents the particle
the particlesize distribution,
particle size
size distribution, which
distribution, which shows
which shows that the
shows that majority
that the mass
the majority (75%)
majority mass
mass
are
(75%)theare
(75%) particles
are the whose whose
the particles
particles sizes
whose are smaller
sizes
sizes than
are smaller
are smaller0.1mm. The
than 0.1mm.
than bulk
0.1mm. grain
The
The bulk
bulkdensity
grain is
grain around
density
density isis
2737 kg/m 3 . Figure 13 shows the grasping tests of foot toes with respect to the simulated
around 2737
around 2737 kg/m
kg/m33.. Figure
Figure 1313 shows
shows the the grasping
grasping teststests of
of foot
foot toes
toes with
with respect
respect toto the
the
surface.
simulated
simulated When the When
surface.
surface. motor
Whenthe drives the drives
themotor
motor rope,
drivesthethetoes
the bend
rope,
rope, down
thetoes
the intodown
toesbend
bend soils, into
down and asoils,
grasping
intosoils, andaa
and
angle of higher than 90 ◦ is obtained.
graspingangle
grasping angleof ofhigher
higherthanthan90°
90°isisobtained.
obtained.

Biomimetics 2023, 8, x FOR PEER REVIEW 11 of 14

Figure12.
Figure
Figure 12.Particle
12. Particlesize
Particle sizedistribution
size distributionof
distribution ofthe
of thesoil
the soilused
soil usedin
used inthe
in thetestbed.
the testbed.
testbed.

Figure
Figure 13.
13. Grasping
Grasping tests:
tests: (a)
(a) retrieving
retrieving state
state and
and (b)
(b) grasping
grasping state.
state.

In order to model the rocky environment as that of the Mars surface, a selection of
rocks with
withsizes
sizes60–120
60–120mm mmwerewere embedded
embedded in granular
in the the granular materials
materials [33]. Figure
[33]. Figure 14
14 shows
shows
the fourthe four postures
postures of thewhen
of the robot robotwalking
when walking by a straight
by a straight linesimulated
line on the on the simulated
surface.
surface. It is observed
It is observed that its
that its spine spineinswings
swings in cooperation
cooperation with the with the foot movement.
foot movement. The bending The
motions are
bending weaker
motions arethan the than
weaker simulation results. The
the simulation reason
results. Thefor this isfor
reason that
thisthe
is support
that the
springs on
support the spine
springs counteract
on the some moments
spine counteract of the servos.
some moments of theConsequently, the average
servos. Consequently, the
average speed of the robot is also lower (approximately 5.1 mm/s) than simulations.
Nonetheless, the experiments demonstrated that this robot could maintain forward mo-
tion on the rocky surface by the implemented theoretical gait plan.
 Figure 13. Grasping tests: (a) retrieving state and (b) grasping state.
Figure 13. Grasping tests: (a) retrieving state and (b) grasping state.

In order to model the rocky environment as that of the Mars surface, a selection of
In order to model the rocky environment as that of the Mars surface, a selection of
rocks with sizes 60–120 mm were embedded in the granular materials [33]. Figure 14
rocks with sizes 60–120 mm were embedded in the granular materials [33]. Figure 14
shows the four postures of the robot when walking by a straight line on the simulated
Biomimetics 2023, 8, 44 shows the four postures of the robot when walking by a straight line on the simulated 11 of 14
surface. It is observed that its spine swings in cooperation with the foot movement. The
surface. It is observed that its spine swings in cooperation with the foot movement. The
bending motions are weaker than the simulation results. The reason for this is that the
bending motions are weaker than the simulation results. The reason for this is that the
support springs on the spine counteract some moments of the servos. Consequently, the
support
speed springs
of the roboton the spine counteract some 5.1
moments
mm/s)ofthan
the simulations.
servos. Consequently, the
average speed ofisthe
alsorobot
loweris(approximately
also lower (approximately 5.1 mm/s) thanNonetheless,
simulations.
average
the speed
experiments of the robot
demonstrated is also
that lower
this robot(approximately
could maintain 5.1 mm/s)
forward than on
motion simulations.
the rocky
Nonetheless, the experiments demonstrated that this robot could maintain forward mo-
Nonetheless,
surface by the the experiments
implemented demonstrated
theoretical gait that this robot could maintain forward mo-
plan.
tion on the rocky surface by the implemented theoretical gait plan.
tion on the rocky surface by the implemented theoretical gait plan.

Figure 14.Moving
Moving onsimulated
simulated Marssurface
surface terrains.
Figure 14. Movingon
Figure14. on simulatedMars
Mars surfaceterrains.
terrains.

To furtherdemonstrate
To demonstrate that the robot is suitable for rocky environments, external
Tofurther
further demonstrate that thethe
that robot is suitable
robot for rocky
is suitable environments,
for rocky external
environments, tests
external
tests
were were conducted and are shown in Figure 15. It can be identified that the left hind
tests conducted and are
were conducted andshown in Figure
are shown 15. It can
in Figure 15. be identified
It can that thethat
be identified leftthe
hind
leftleg of
hind
leg
the of
robotthe robot
can can
climb climb
across aacross
rock, a rock, indicating
indicating that thisthat this
improved improved
robot robot
model model
has betterhas
leg of the robot can climb across a rock, indicating that this improved robot model has
better performance
performance comparedcompared with
with thewith the previous
previous work [19].
work [19].
better performance compared the previous work [19].

Figure 15. Climbing over a rock on the simulated Mars surface terrains.
Figure15.
Figure 15.Climbing
Climbingover
overaarock
rockon
onthe
thesimulated
simulatedMars
Marssurface
surfaceterrains.
terrains.

6.Discussion
6.6. Discussion
Discussion
To exploregeomorphology
To geomorphology andconduct
conduct surfaceexploration
exploration onMars
Mars withscientific
scientific
Toexplore
explore geomorphologyand and conductsurface
surface explorationon on Marswith
with scientific
instruments
instruments carried by a robot, this work evaluates a biomimetic creeping robot inspired
instrumentscarried
carriedby
byaarobot,
robot,this
thiswork
workevaluates
evaluatesaabiomimetic
biomimeticcreeping
creepingrobot
robotinspired
inspired
by the desert lizard animal [34]. Its mobility with respect to soft granular soils and rocks as
those of the Mars surface were assessed, which demonstrates that this robot is adaptive
to both granular and rocky terrains, as opposed to the wheeled rovers [11] and legged
rovers [12,13]. The flexible spine promotes larger swinging ranges and thus a larger stride
compared to rigid spine robots [17]. Compared to Lemur [16], fewer and more flexible
toes are used, which reduce the structure complexity and is beneficial to efficient control.
Moreover, the design of creeping legs enables a lower center mass and higher stability
which is more advantageous compared to a stand–walking robot [19]. Specifically, the
leg adopts a four-linkage structure which ensures steady lift of the foot. Therefore, the
structure for this robot can perform a move stable movement than our previous model [19].
To achieve coordinated movement between the foot and spine, the trajectory was
formulated by integrating motions of the foot, leg, and spine. The foot trajectory adopts a
composite curve such that excessive impacts when contacting the ground are avoided. The
foot motion can be regulated by adjusting step length or height as long as the robot motion
mechanism is retained. Moreover, this trajectory has certain adaptability for climbing
different slopes, which promotes higher adaptability to the Mars surface environment [27].
For a faster forward speed, the weight distribution of the spine can be optimized by
Biomimetics 2023, 8, 44 12 of 14

the structure design. To further improve the walking stability for irregular surfaces, gait
planning, by considering lizard locomotion on a complex environment, may be studied [35].
The regolith and terrains of planetary surfaces vary for different locations. Corre-
spondingly, the experimental mobility tests on such terrains is essential to also be tested at
lower gravity levels [9]. Furthermore, it is important to perform the resilient locomotion
of the robot, which would be beneficial to improving the design and control, and thus
the adaptability to Mars surface terrains [36]. On the other hand, numerical simulations
based on MBD-DEM [37] can be used to model the walking characteristics. A full sealing
of the robot drive mechanism and hinges is necessary to protect against particle ingress
from the Mars soil and airborne dust [38]. Considering launch and landing mechanical
mission loads, radiation, and thermal effects, the rigid part of the robot structure for a
flight model design would be replaced by titanium alloys, and the soft part such as that
for electronic components would adopt fiber reinforced composite materials. To achieve
adaptive control, the foot reaction force may be measured and the control strategy would
then be adjusted [39]. Implementation of autonomous control of self-adapting to different
terrains using machine learning is already undergoing [40]. In addition, the requirement
for the continuous power supply in this quadruped robot is to be accounted for in the next
version [41].

7. Conclusions
To adapt to the Mars surface terrains, this research proposed a biomimetic quadruped
robot based on the structure and mechanism of a desert lizard. Kinematic models for
coordinated movements among the foot, leg, and spine were established. To verify these
models, the leg motion and straight walking abilities based on gait planning were analyti-
cally predicted. Furthermore, experimental tests demonstrated that the biomimetic robot
is suitable for granular soils and rocky surfaces, which is of high potential for walking on
the Mars surface terrains. Thus, this work progresses the development of Mars robots for
surface exploration.

Supplementary Materials: The following supporting information can be downloaded at

https://www.mdpi.com/article/10.3390/biomimetics8010044/s1.
Author Contributions: G.C. and L.R. conceptualized, reviewed, and edited the manuscript; the data
processing and original draft preparation were accomplished by L.Q. The experimental methodology
was formulated by A.J.; Z.Z. was responsible for the robot validation tests. All authors have read and
agreed to the published version of the manuscript.
Funding: This research was funded by the Foundation Research Project of Jiangsu Province Natural
Science Fund, grant number BK20190415, and the project of Foreign Culture and Education Experts
(Ministry of Science and Technology), grant number G2021181006L.
Institutional Review Board Statement: Not applicable.
Data Availability Statement: The datasets generated during and/or analyzed during the current
study are available from the corresponding author on reasonable request.
Conflicts of Interest: The authors declare no conflict of interest.

References
1. Malaya Kumar Biswal, M.; Ayyappan, S.; Thomas, A.; Ramesh Kumar, V.; Basanta Das, N. A Baseline Strategy for Human Mars
Exploration. In Proceedings of the Accelerating Space Commerce, Exploration, and New Discovery Conference, Las Vegas, NV,
USA, 15–17 November 2021.
2. Salgado-Rodriguez, M.; Garcia-Luis, U.; Gomez-San-Juan, A.; Ulloa-Sande, C.; Navarro-Medina, F. Conceptual Design and
Research on the Thermal Performance of a Martian Human Base. Acta Astronaut. 2022, 200, 524–538. [CrossRef]
3. Oudart, N.; Ciarletti, V.; Le Gall, A.; Hervé, Y.; Brighi, E. Retrieval of the Ground Dielectric Permittivity by Planetary GPR
Accommodated on a Rover: Application to the Estimation of the Reflectors’ Depth by the WISDOM/ExoMars Radar. Planet.
Space Sci. 2022, 224, 105606. [CrossRef]
Biomimetics 2023, 8, 44 13 of 14

4. Maki, J.N.; Gruel, D.; McKinney, C.; Ravine, M.A.; Morales, M.; Lee, D.; Willson, R.; Copley-Woods, D.; Valvo, M.;
Goodsall, T.; et al. The Mars 2020 Engineering Cameras and Microphone on the Perseverance Rover: A Next-Generation Imaging
System for Mars Exploration. Space Sci. Rev. 2020, 216, 137. [CrossRef]
5. Li, C.; Zheng, Y.; Wang, X.; Zhang, J.; Wang, Y.; Chen, L.; Zhang, L.; Zhao, P.; Liu, Y.; Lv, W.; et al. Layered Subsurface in Utopia
Basin of Mars Revealed by Zhurong Rover Radar. Nature 2022, 610, 308–312. [CrossRef] [PubMed]
6. Almehisni, R.; Garg, P.; Wali, M. Thermal Management System Design and Analysis of Rashid Rover—Emirates Lunar Mission.
In Proceedings of the International Astronautical Congress(IAC), Dubai, United Arab Emirates, 25–29 October 2021.
7. Ceylan, S.; Clinton, J.F.; Giardini, D.; Stähler, S.C.; Horleston, A.; Kawamura, T.; Böse, M.; Charalambous, C.; Dahmen, N.L.; van
Driel, M.; et al. The Marsquake Catalogue from InSight, Sols 0–1011. Phys. Earth Planet. Inter. 2022, 333, 106943. [CrossRef]
8. Zhang, X.; Zhang, G.; Xie, H.; Gao, M.; Wen, Y. A Review of Sampling Exploration and Devices for Extraterrestrial Celestial
Bodies. Space Sci. Rev. 2022, 218, 59. [CrossRef]
9. Medina, F.J.; Manzano, A.; Villacampa, A.; Ciska, M.; Herranz, R. Understanding Reduced Gravity Effects on Early Plant
Development Before Attempting Life-Support Farming in the Moon and Mars. Front. Astron. Space Sci. 2021, 8, 1–7. [CrossRef]
10. Arvidson, R.E.; Bell, J.F.; Bellutta, P.; Cabrol, N.A.; Catalano, J.G.; Cohen, J.; Crumpler, L.S.; Des Marais, D.J.; Estlin, T.A.;
Farrand, W.H.; et al. Spirit Mars Rover Mission: Overview and Selected Results from the Northern Home Plate Winter Haven to
the Side of Scamander Crater. J. Geophys. Res. Planets 2010, 115, 1–19. [CrossRef]
11. Yuan, B.; Wang, C.; Zou, M.; Liu, Y.; Lin, Y.; Jia, Y.; Chen, B.; Jin, J. Experimental Study on the Durability of China’s Mars Rover’s
Mobility System. J. Aerosp. Eng. 2021, 34, 1–8. [CrossRef]
12. Sunspiral, V.; Wheeler, D.W.; Chavez-Clemente, D.; Mittman, D. Development and Field Testing of the FootFall Planning System
for the ATHLETE Robots. J. Field Robot. 2012, 29, 483–505. [CrossRef]
13. Cordes, F.; Kirchner, F.; Babu, A. Design and Field Testing of a Rover with an Actively Articulated Suspension System in a Mars
Analog Terrain. J. Field Robot. 2018, 35, 1149–1181. [CrossRef]
14. Zhang, T.; Peng, S.; Jia, Y.; Tian, H.; Sun, J.; Yan, C. Slip Estimation for Mars Rover Zhurong Based on Data Drive. Appl. Sci. 2022,
12, 1676. [CrossRef]
15. Arm, P.; Zenkl, R.; Barton, P.; Beglinger, L.; Dietsche, A.; Ferrazzini, L.; Hampp, E.; Hinder, J.; Huber, C.; Schaufelberger, D.; et al.
SpaceBok: A Dynamic Legged Robot for Space Exploration. In Proceedings of the IEEE International Conference on Robotics and
Automation 2019, Montreal, QC, Canada, 20–24 May 2019.
16. Parness, A.; Abcouwer, N.; Fuller, C.; Wiltsie, N.; Nash, J.; Kennedy, B. LEMUR 3: A Limbed Climbing Robot for Extreme Terrain
Mobility in Space. In Proceedings of the IEEE International Conference on Robotics and Automation (ICRA) 2017, Singapore, 29
May–3 June 2017.
17. Nagaoka, K.; Minote, H.; Maruya, K.; Shirai, Y.; Yoshida, K.; Hakamada, T.; Sawada, H.; Kubota, T. Passive Spine Gripper for
Free-Climbing Robot in Extreme Terrain. IEEE Robot. Autom. Lett. 2018, 3, 1765–1770. [CrossRef]
18. Biswal, P.; Mohanty, P.K. Development of Quadruped Walking Robots: A Review. Ain Shams Eng. J. 2021, 12, 2017–2031.
[CrossRef]
19. Chen, G.; Qiao, L.; Wang, B.; Richter, L.; Ji, A. Bionic Design of Multi-Toe Quadruped Robot for Planetary Surface Exploration.
Machines 2022, 10, 827. [CrossRef]
20. Wang, W.; Ji, A.; Chen, G.; Ravi, S.; Shen, H.; Gorb, S.N.; Dai, Z. Kinematics of Gecko Climbing: The Lateral Undulation Pattern.
Zoology 2020, 140, 125768. [CrossRef]
21. Li, C.; Hsieh, S.T.; Goldman, D.I. Multi-Functional Foot Use during Running in the Zebra-Tailed Lizard (Callisaurus Draconoides).
J. Exp. Biol. 2012, 215, 3293–3308. [CrossRef]
22. Haomachai, W.; Shao, D.; Wang, W.; Ji, A.; Dai, Z.; Manoonpong, P. Lateral Undulation of the Bendable Body of a Gecko-Inspired
Robot for Energy-Efficient Inclined Surface Climbing. IEEE Robot. Autom. Lett. 2021, 6, 7918–7925. [CrossRef]
23. Uno, K.; Koizumi, Y.; Haji, K.; Kei, M.; Harms, S.; Ribeiro, W.F.R.; Nagaoka, K.; Yoshida, K. Non-Periodic Gait Planning Based on
Salient Region Detection for a Planetary Cave Exploration Robot. In Proceedings of the i-SAIRAS, Virtual Conference, 19–23
October 2020.
24. Chen, M.; Li, Q.; Wang, S.; Zhang, K.; Chen, H.; Zhang, Y. Single-Leg Structural Design and Foot Trajectory Planning for a Novel
Bioinspired Quadruped Robot. Complexity 2021, 2021, 6627043. [CrossRef]
25. Zhang, D.; Fang, T.; Yang, Y.; Fang, T.; Guo, Z. Static Gait Planning of a Quadruped Robot with Four-Bar Shock Absorbing
Mechanism. In Proceedings of the 7th International Conference on Mechatronics and Robotics Engineering (ICMRE), Budapest,
Hungary, 5–9 February 2021.
26. Zhang, C.; An, H.; Wei, Q.; Ma, H. Foot Trajectory Planning Method with Adjustable Parameters for Complex Environment. In
Proceedings of the IEEE International Conference on Robotics and Biomimetics, Dali, China, 27–31 December 2019.
27. Yuan, S.; Zhou, Y.; Luo, C. Crawling Gait Planning Based on Foot Trajectory Optimization for Quadruped Robot. In Proceedings
of the 2019 IEEE International Conference on Mechatronics and Automation, Tianjin, China, 4–7 August 2019.
28. Corke, P.I. Robotics Toolbox. 2002. Available online: Http://www.Petercorke.Com/Robotics%20Toolbox.Html (accessed on 10
December 2022).
29. MSC Software “ADAMS 2013.2 Solver User Manual”. 2013. Available online: https://www.mscsoftware.com/page/adams
(accessed on 10 December 2022).
Biomimetics 2023, 8, 44 14 of 14

30. Fang, S.; Xie, C.; Wu, X.; Wang, X. Design of a Quadruped Wall-Climbing Robot (WCR) with a Three-Row Opposed Gripping
Mechanism. In Proceedings of the 2021 IEEE International Conference on Real-Time Computing and Robotics (RCAR), Xining,
China, 15–19 July 2021.
31. Cannon, K.M.; Britt, D.T.; Smith, T.M.; Fritsche, R.F.; Batcheldor, D. Mars Global Simulant MGS-1: A Rocknest-Based Open
Standard for Basaltic Martian Regolith Simulants. Icarus 2019, 317, 470–478. [CrossRef]
32. Yu, W.; Zeng, X.; Li, X.; Wei, G.; Fang, J. New Martian Dust Simulant JMDS-1 and Applications to Laboratory Thermal Conductivity
Measurements. Earth Space Sci. 2022, 9, 1–15. [CrossRef]
33. Golombek, M.P.; Trussell, A.; Williams, N.; Charalambous, C.; Abarca, H.; Warner, N.H.; Deahn, M.; Trautman, M.; Crocco, R.;
Grant, J.A.; et al. Rock Size-Frequency Distributions at the InSight Landing Site, Mars. Earth Space Sci. 2021, 8, e2021EA001959.
[CrossRef]
34. Schultz, J.T.; Beck, H.K.; Haagensen, T.; Proost, T.; Clemente, C.J. Using a Biologically Mimicking Climbing Robot to Explore the
Performance Landscape of Climbing in Lizards. Proc. Biol. Sci. 2021, 288, 20202576. [CrossRef] [PubMed]
35. Ijspeert, A.J. Amphibious and Sprawling Locomotion: From Biology to Robotics and Back. Annu. Rev. Control Robot. Auton. Syst.
2020, 3, 173–193. [CrossRef]
36. Karakasiliotis, K.; Thandiackal, R.; Melo, K.; Horvat, T.; Mahabadi, N.K.; Tsitkov, S.; Cabelguen, J.M.; Ijspeert, A.J. From
cineradiography to biorobots: An approach for designing robots to emulate and study animal locomotion. J. R. Soc. Interface 2016,
13, 20151089. [CrossRef]
37. Schott, D.; Mohajeri, J.; Jovanova, J.; Lommen, S.; de Kluijver, W. Design Framework for DEM-Supported Prototyping of Grabs
Including Full-Scale Validation. J. Terramechanics 2021, 96, 29–43. [CrossRef]
38. Alshehhi, R.; Gebhardt, C. Detection of Martian Dust Storms Using Mask Regional Convolutional Neural Networks. Prog. Earth
Planet. Sci. 2022, 9, 1–16. [CrossRef]
39. Zhu, X.; Wan, J.; Zhou, C.; Xu, W. A Composite Robust Reactive Control Strategy for Quadruped Robot under External Push
Disturbance. Comput. Electr. Eng. 2021, 91, 107027. [CrossRef]
40. Lopez-Arreguin, A.J.R.; Montenegro, S. Machine Learning in Planetary Rovers: A Survey of Learning versus Classical Estimation
Methods in Terramechanics for in Situ Exploration. J. Terramechanics 2021, 97, 1–17. [CrossRef]
41. Morgan, A.B.; Kumar, J.; Barklay, C.D.; Kramer, D.P. Non-Nuclear Based Thermoelectric + Battery System Concepts for Space
Power Systems. In Proceedings of the 2020 IEEE Aerospace Conference, Big Sky, MT, USA, 21 August 2020.

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual
author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to
people or property resulting from any ideas, methods, instructions or products referred to in the content.