Journal of Bionic Engineering 8 (2011) 429439

Bionic Mechanism and Kinematics Analysis of Hopping Robot

Inspired by Locust Jumping
Diansheng Chen, Junmao Yin, Kai Zhao, Wanjun Zheng, Tianmiao Wang
Robotic Institute, Beihang University, Beijing 100191, P. R. China

Abstract
A flexible-rigid hopping mechanism which is inspired by the locust jumping was proposed, and its kinematic characteristics were analyzed. A series of experiments were conducted to observe locust morphology and jumping process. According to
classic mechanics, the jumping process analysis was conducted to build the relationship of the locust jumping parameters. The
take-off phase was divided into four stages in detail. Based on the biological observation and kinematics analysis, a mechanical
model was proposed to simulate locust jumping. The forces of the flexible-rigid hopping mechanism at each stage were analyzed. The kinematic analysis using pseudo-rigid-body model was described by D-H method. It is confirmed that the proposed
bionic mechanism has the similar performance as the locust hind leg in hopping. Moreover, the jumping angle which decides the
jumping process was discussed, and its relation with other parameters was established. A calculation case analysis corroborated
the method. The results of this paper show that the proposed bionic mechanism which is inspired by the locust hind limb has an
excellent kinematics performance, which can provide a foundation for design and motion planning of the hopping robot.
Keywords: hopping robot, flexible-rigid mechanism, bionic mechanism, kinematics
Copyright 2011, Jilin University. Published by Elsevier Limited and Science Press. All rights reserved.
doi: 10.1016/S1672-6529(11)60048-6

1 Introduction
Hopping robot belong to a kind of ground mobile
robot. Such robot must have the ability of adaption to
unstructured environment, especially to avoid dangers
and overleap obstacles agilely and precisely. Many researchers conducted a variety of studies on ground mobile robots with different forms of movement. At present,
the terrain adaptability of ground mobile robot has already been largely improved, but the obstacle negotiating performance is still unsatisfactory due to the limitation of the locomotion mode. Biologic hopping locomotion can achieve a certain height and distance in a
moment, which enables animals to overcome obstacles
and uneven terrains. Therefore, the study on hopping
robot, which developed from nature observation, has
received much attention in recent years. The exploration
of hopping movement will expand the application fields
of mobile robot.
Raibert and Pratt developed one-legged, biped, and
quadruped running and hopping robots based on
Raiberts pneumatic model, all of which have telescopCorresponding author: Diansheng Chen
E-mail: chends@buaa.edu.cn

ing legs and can achieve different running motions[14].

Fiorini et al. constructed three series of hopping robots
based on single or six bars spring mechanism. Their
robots are characterized by discontinuous motion, i.e.
the robots need to stop after each hop to adjust posture,
get recharged and orientate themselves[5,6]. Brown and
Zeglin proposed a bow leg hopper which has a bow-like
resilient leg and adopt a continuous hopping mechanism[7,8]. All of these robots depend on traditional spring
energy storage mechanism, which is considered to lead
to a low efficiency.
Consequently, a novel model was established by
taking inspirations from animals, whose ingenious
physical structure and locomotion mode have become
gradually adaptive to unstructured and unknown environment after billions of years of evolution. Therefore,
the study of bionic hopping movement mechanism has
been a hot spot in recent years. By incorporating biologically-inspired dynamics into design, Hiraguchi et al.
developed a cricket robot with legs actuated by artificial
muscles[9]. Kovac et al. developed an EPFL jumper
looking like a locust. The versatile jumping mechanism

Journal of Bionic Engineering (2011) Vol.8 No.4

430

can negotiate obstacles with more than 27 times of its

own size (5 cm), and the takeoff angle, jumping height
and force profile are adjustable[1012]. Niiyama et al.
developed a pneumatically actuated bipedal robot named
Mowgli which can hop to a height of over 50% of its size
and land steadily[13].
In view of the overall situation, these studies use
traditional driver and elastic spring to mimic the jumping performance of insect or animal. This may lead to a
low efficiency of energy storage which is caused by the
limitation of traditional mechanism and the linearity of
drivers and springs. Therefore, it is significant to innovate new mechanisms with high efficiency of energy
storage.
In this paper, we simplify and establish a hopping
mechanism with flexible-rigid model inspired by locust
hind leg, and analyze the statics and takeoff characteristics. The proposed mechanism is proved to have a high
efficiency. The results of this paper can provide a
foundation for the design and kinematics plan of hopping robot.

2 Locust observation and analysis

The jumping capabilities of four species are listed in
Table 1[14]. Obviously, locust has excellent hopping performance. By taking experiment condition into account,
we chose locust as the research object. It is not only because the locust has appropriate size, which makes image
capture and observation more easily, but also the hind leg
of locust keeps in vertical plane in hopping locomotion,
which makes the study more feasible.

2.1 Locust morphology

A set of experiments were designed to investigate
the morphology of locust. Three locusts, Locusta migratoria manilensis, were captured from the lawn of our
campus. As Fig. 1 shows, a locust has 3 pairs of limbs,
which are different in size. The fore limb and the mid
limb used for walking are small and weak, while the hind
limb is large and strong enough to achieve jumping
locomotion. A hind leg of locust consists of 4 distinct
sections: coxa, femur, tibiae and tarsus, which are connected by joints. The dimension parameters and mass of
three locusts were measured (Table 2). The centroid is
measured between the base of mid limb and hind limb,
and bilaterally symmetrical from the vertical view.
A hopping process can be decomposed into several
phases including walking (or running), taking-off, flying
certain height and distance, and landing on the ground.
Jumping locomotion depends on the cooperation of
skeleton, muscle, tendon and femur. Skeleton is equal to
lever, and joint is fulcrum. Muscle and tendon offer
power. The folded hind limbs stretch quickly that thrust
against the ground to actuate the locust body in jumping.
In addition, the light and rigid tibia and the flexible
tarsus also have important effects. As Heitler[15] proposed, a remarkable jump should satisfy the requirement
that the legs must thrust against the ground with a lot of
force and develop this force quickly. If that is too slow,
legs only stretch before the force thrusting against the
ground reaches the maximum[15]. The analysis of the
thrust will be presented in Section 4.

Table 1 The jumping capabilities of several species

Frog

Locust

15

23

30

300

Distant (cm)

30

50

70

900

200

12

22

1.2

3.2

Compared to its size (times)

1

Initial velocity (ms )

Kangaroo

Flea

Parameters
Height (cm)

Fig. 1 Locust morphology.

Table 2 Dimension parameters of locusts
Locust 1

Locust 2

Mass (g)

0.22

0.21

0.23

Body length (mm)

33.25

31.29

34.05

Femur

Tibiae

Tarsus

Femur

Hind limb (mm)

13.27

11.97

3.10

13.03

Mid limb (mm)

5.10

4.61

2.34

Fore limb (mm)

4.55

3.98

1.80

Locust 3

Tarsus

Femur

Tibiae

Tarsus

3.64

13.25

11.99

3.55

5.17

Tibiae
12.1
6
4.80

2.91

5.19

4.82

2.85

4.45

3.56

2.49

4.65

4.21

2.41

Chen et al.: Bionic Mechanism and Kinematics Analysis of Hopping Robot Inspired by Locust Jumping

Fig. 2 shows the knee of a locust hind limb. The

coxa is a 2 DOF joint, which can achieve the flexion/extension and adduction/abduction of the femur with
2 pairs of muscle. The tibia extensor and flexor contribute greatly to jumping, which makes knee stretch and
contract respectively. Flexor tendon grows round the
femur lump (equal to beam point). As shown in Fig. 2,
the dense black half-moon shaped region is the external
view of a pair of springs, and Heitler[15] named it as
semi-lunar process. This process is only found in hind
legs[15]. The lever ratio of extensor to tibia terminal is
approximately 1:35. In addition, the flexible multi-tarsus
is attached with flexible suckers which can prevent slip
and grasp the ground. All the structures are specialized.
Therefore, leg structure of locust has jump predominance.

431

Fig. 3 Experimental platform.

200
150
100
50
0

50

100

150

200
x(t) (mm)

250

300

350

400

y (mm)

Fig. 4 The images of the jump phase.

Fig. 2 The knee of locust hind limb[12].

2.2 Observation of jumping

An experimental apparatus is set up to observe the
jumping process, especially for the takeoff phase. As
shown in Fig. 3, the apparatus consists of a high-speed
camera (FASTCAM 1280 PCI, Photron, USA), a light
source, PC, observation table and background (coordinate paper).
The high-speed camera can capture rapid movement with 500 fps (frame per second), so that the whole
jumping process can be recorded and stored in PC.
During experiment, the temperature was kept at about
20 C to ensure the locust in an active status. The images
of takeoff trajectory and locust posture are shown in
Fig. 4. The time interval of each image was 4 ms. The
trajectory curve is shown in Fig. 5.
2.2.1 Jump process analysis
The body center of locust moves as projectile. As
we know, in the analytic form of classical mechanics, the

200
180
160
140
120
100
80
60
40
20
0

50

100

150

200 250
x (mm)

300

350

400

450

Fig. 5 The trajectory curve of jumping.

projectile is well described by law of particle movement

and law of conservation of energy. Thus the locust
jumping performance can be described as Eq. (1).
x v0 t cos D ,

1 2
y v0 t sin D  2 gt ,

(1)

where v0 denotes initial velocity, is takeoff angle, t is

jumping time, x and y denote the height and horizontal
distance respectively, g is gravity. Thus, the maximum
jump height ymax and distance xmax are given by
xmax

ymax

(v02 sin 2D ) g ,
(v0 sin D )2 2 g .

(2)

Journal of Bionic Engineering (2011) Vol.8 No.4

432

Eq. (2) shows that the takeoff speed and angle affect the jump height and distance. With a certain speed,
it can be proved that the jumping achieves xmax when the
takeoff angle is about 45. As the function of projectile
trajectory is given by
y

x tan D 

350

(3)

250
y = y(t)

200
150
100
50

This derivation ignores the air resistance, so it usually applies to the case with a relatively low speed.
However, the takeoff speed of locust may be very high
that it is uncertain whether the locust case adapts the
equations. Therefore, to get a precise result, the derived
curve F(x) = y(x) and second derived curve G(x) = y(x)
of fitting trajectory curve are given in the Fig. 6. It shows
the case that the takeoff angle is 45 and the jumping
process corresponds to the law of particle movement.
Therefore, based on Eq. (3), the conservation of energy
is given by Eq. (4) and energy loss Es reveals the error in
Fig. 6.

(a)

3.0

F(x), G(x)

150
t (ms)

200

250

sqrt(Vx2 , Vy2 )

2.0
Vx = x(t)

1.5
1.0

Vy = y(t)

0.5
(b)
0.0

0.005

Fig. 6 The derived curve F(x) and second derived curve G(x) of
trajectory curve.

100

2.5

(4)

where v is defined to be the velocity when the height is h,

m denotes the body mass. Fig. 7 shows the curves of
jumping height and horizontal distance vs. the time.
According to the takeoff angle, initial velocity and
Eq. (2), the ymax is calculated to be 216.2 mm, and then
the loss of energy Es is also calculated.
Locust jumping is approximately a projectile motion, and the trajectory is a quadratic curve. The energy
loss in jumping process is tested by experiments. The
experimental analysis on the locust jumping parameters
and energy loss can provide reference basis for future
study on kinematics planning of bio-hopping robot.

50

3.5

50

100

150
t (ms)

200

250

300

The acceleration in
the x direction

0.000
Acceleration ( mmms2)

1 2
1
mv0 3 mgh ! mv 2 ! Es ,
2
2

x = x(t)

300
s ( mm )

1 gx 2
.
2 v02 cos 2 D

400

0.005
0.010

The acceleration
in the y direction

0.015
0.020
0.025
0.030
The sum acceleration
in the x and y direction

0.035
0.040

50

100

150
t (ms)

(c)
200

250

300

Fig. 7 The relationship of the jumping height and horizontal

distance vs. time. The takeoff initial speed is 3.2 ms1. The
horizontal speed is reduced slowly, but the vertical velocity decreases rapidly. The air resistance makes the acceleration decreases rapidly besides gravity.

2.2.2 Division of takeoff phase

From rest to takeoff, the joint angles are measured
using the captured images in Fig. 8. The angles data of
each limb are given in Table 3.
The takeoff phase can be disintegrated into 4 stages
with movement coordination and angles of hind legs
joints, as shown in Fig. 9.
Stage 1: Initial contraction and flexion of joints.
The knees of hind legs bend to a location, mainly for
adjusting the leg posture, and the body is up and down
like simplified five linkages slider-crank mechanism.
Full flexion is an essential pre-requisite for jumping, in
which the flexor and extensor muscles contract together.

Chen et al.: Bionic Mechanism and Kinematics Analysis of Hopping Robot Inspired by Locust Jumping

433

Fig. 8 The joint angles calibration on the image.

Table 3 The joints angles in takeoff process
Video time

Body-horizontal

Tarsus-horizontal

Tibiae-tarsus

Femur-Tibiae

Body-femur

0 ms

10.47

0.00

29.39

5.91

146.05

4 ms

10.32

0.00

27.09

3.95

146.55

8 ms

11.10

0.00

26.54

4.16

146.82

12 ms

11.24

0.00

25.89

4.01

146.87

16 ms

11.38

0.00

24.46

3.72

145.88

20 ms

11.37

0.00

26.52

3.87

145.96

24 ms

11.60

0.00

26.45

4.03

145.96

28 ms

12.70

0.00

28.94

5.18

143.53

32 ms

12.70

0.00

40.27

7.09

134.11

36 ms

8.71

0.00

52.22

24.60

143.67

40 ms

3.18

0.00

73.55

71.68

178.06

44 ms

12.44

40.49

147.94

142.63

132.38

48 ms

22.75

46.12

142.75

135.63

118.24

52 ms

11.03

56.03

170.77

143.07

140.64

Fig. 9 The 4 stages of locust takeoff.

Stage 2: Contraction and coordination of the flexor

and extensor muscles. In this stage, the leg extension is
triggered by a sudden relaxation of the flexor muscle, in
which the flexor muscles extend rapidly and forcefully.
The joints of hind legs move coordinately and tarsal limb
thrusts against ground to produce an initial speed of
body with certain direction using the energy stored in
semi-lunar.
Stage 3: Prolonging the thrust and enhancing the
jumping. When the joint is extended to a certain extent,
the body attains a certain speed. Then the flexible tarsal
begin to deformation and stretch, so that keep the limb in

contact with the ground, which makes the hind legs

thrust continually, and enhances the effect of jumping.
Stage 4: Takeoff from the ground. During takeoff,
the initial full fold of hind legs is prerequisite; the triggering of the flexor muscle and coordinated movement
at the first acceleration phase decide whether the jumping is successful or not; while the elastic deformation of
tarsal leg makes the second acceleration and prolongs
the thrust. Finally the body takes off from the ground in
accordance with regular exercise determined by Eq. (1).
Based on the physiological mechanism of locust
movement, the flexible-rigid mechanical model is built
and the force/torque analysis of joints is completed.

3 Modeling and analysis of bionic mechanism inspired by locust

3.1 A mechanical model of locust
Based on the locust morphology and jumping observation, a mechanical model is put forward to simulate
locust jumping. From the viewpoint of jump function,
much importance is attached to the design of hind leg

Journal of Bionic Engineering (2011) Vol.8 No.4

434

bionic mechanism. Fig. 10 shows the model of the hopping robot considered in this paper. Considering that the
semi-lunar process of knee joint is a flexible structure, it
is replaced with a linear spring to store energy. Moreover a flexible link takes the place of the tarsus.
The analysis of kinematic properties will be described in section 4.

Fe

Fs

OL

Ff

lEF
lDE

D
D

Extensor
muscle

Tendon
Femur

Joint

Extensor
muscle for
pretarsus

Trochanter

lCD
x
o

Tibiae
Coxa

Fg

Pretarsus

A
Tarsus

Fm

Ffr

Fig. 11 The free-body diagram of locust knee and tibiae.

Fe

Fs

Ff
M1

M0
Fg

Ffr

Fig. 10 The model of the hopping robot.

3.2 The statics and performance analysis of hind leg

mechanism
Statics studies the relationship between forces
when objects keep balance. Considering there is a
flexible deformation of the tarsal, in this section we
establish a rigid-flexible model, conducts statics and
dynamics analyses, and discusses the mechanics performance. As shown in Fig. 10, a hind leg mechanism is
built in accordance with the structure and function of
locust hind legs.
In Fig. 10, the femur is fixed. And the free-body
diagram of locust knee and tibiae is shown in Fig. 11. In
Fig. 11, points A, B, C, D, E, F, OL denote the front toe,
the contact point with ground, the pastern, the action
point of knee extension force, the connection point of
the energy storage spring and tibia, the action point of
knee flexion force and the supporting point of knee,
respectively; Ffr and Fg are respectively defined as the
friction and reaction forces with ground; lCD, lDE, lEF are
the lengths of CD, DE, EF; D, E are the constant angles between CD and DE, and between EF and DE
correspondingly; and is the angle between EF and

vertical direction. In addition, mt is the mass of

free-body; vf is the speed; F and are the angles between Fe and EF, Ff and DE, respectively. In the stage
of contraction, the knee flexion force Ff is generated by
the contraction of tibia flexion muscle, meanwhile the
tibia extension muscle stretches which causes the extension force Fe. Therefore, Fe is much smaller than Ff.
According to equilibrium relation of force and torque,
the force balance equations of tibia in X and Y directions
are written as
Fe sin(G  M F )  Ff sin(M E  G  T )  mt vfc
Fs  Ffr cos(M E  G  G D ),

(5)

Fe cos(G  M F )
Ff cos(M E  G  T )  Fg cos(M E  G  G D ), (6)
where Fs denote the spring force. The torque balance
equation is given by
Fe lEF sin M F  Ffr (lDE cos(M E  G ) 
lCD cos(M E  G  G D ))  Ff lDE sin T

 M E Z c, (7)

where ME denotes the torque of point E, and is angular

speed. After rearrangement we can derive Ff as
Ff

1
 Fe lEF sin M F  Ffr (lDE cos(M E  G ) 
lDE sin T

lCD cos(M E  G  G D ))  M E Z c  .

(8)

For the convenience of analysis, the above equation

is rewritten as

Chen et al.: Bionic Mechanism and Kinematics Analysis of Hopping Robot Inspired by Locust Jumping

Ff

Fe lEF sin M F Ffr lDE cos(M E  G )



lDE sin T
lDE sin T
Ffr lCD cos(M E  G  G D )
M EZ c
.

lDE sin T
lDE sin T

(9)

At the early stage of contraction, Fe and are

small. In the process of contraction, and E increase
with the decrease in F. What more, lEF is shorter than
half of lDE. Consequently, the first term of Eq. (9) can be
neglected so as to ensure that the flexion muscle is contracted. The last term also can be ignored for the reason
that the contraction phase is very slow and tibia is thin
and light. From the second and the third terms in Eq. (9),
it can be seen that Ff mainly overcomes Ffr, the friction
force between tarsus and the ground. At this stage, locusts body is mainly supported by fore and mid limbs,
so Ffr, which is proportional to Fg, is bit small. In a word,
a small knee flexion force is able to fulfill the motion. In
addition, no external force acts on tarsus. Thus, flexible
deformation is relatively slight enough to be overlooked.
The coxa and femur are connected by a hinge with
2 DOF. In case of > 0, femur move forward, otherwise
it moves backward. Just as the analysis in section 2.2.2,
the initial bending contraction is the precondition for
hopping. After that, the stage of compressing spring is to
store energy. At the beginning, F decreases to the
minimum and increases to the maximum. Then, Fe
increases while F and keep constant. According to
Eq. (5), Fs increases and the spring contracts to afford
force enough to keep balance. At the same time, the energy for hopping is stored. Meanwhile, the angle between
the tibia and femur changes from 25 to approximately 0.
The above analysis is in accordance with the biology and
morphology study of locusts in Ref. [16].
3.3 Analysis of parameters at takeoff phase
As is analyzed in section 2.2.1, the jumping performance directly depends on the takeoff angle and initial velocity. Therefore, joints coordination and takeoff
posture are two key parameters. As shown in Fig. 12, we
define Ci, Ei, Gi, Oi (i = 0, 1, ) as the points of tarsus,

'LGx
't

c
LGx

435

knee, coxa and centroid at position i, and use i, i, i (i =

0, 1, ) to denote the angles of joints C, E and G. C0, E0,
G0, and O0 represent the original posture.

M C0

M C1

Fig. 12 The takeoff posture of the rigid bionic mechanism.

Then the velocity of the centroid VO can be expressed by

JJJJK
VO vOx i  vOy j or VO VG  ZOG u Gi Oi , (10)
where VG denote the velocity of the coxa. The angular
JJJJK
velocity OG is generated by torque of coxa. The Gi Oi
is a vector from coxa point Gi to centroid Oi, which
influences jumping angle of the centroid and the jumping performance. VG can be calculated by
VG

'H Gy
'LGx
i
j,
't
't

(11)

where t is the corresponding change of time; LGx and

HGy are the changes of the distance and height of point
G, which can be respectively derived as
'LGx

'H Gy

lCE M1 sin(

M1

 M0 ) 
2
T T
lEG (T1  T 0 ) sin(G 0  1 0 ),
2
lCE M1 cos(

M1
2

 M0 ) 

lEG (T1  T 0 ) cos(G 0 

T1  T 0
2

),

(13)

where lCE, lEG denote the length of CE, EG correspondingly. When t is small enough, the difference equations
of Eqs. (12) and (13) can be derived as

M
1
 M0 )  lCE M1 M1c cos( 1  M0 ) 
2
2
2
T1  T 0 1
T T
)  lEG (T1  T 0 ) (T1  T 0 )c cos(G 0  1 0 ),
lEG (T1  T 0 )c sin(G 0 
2
2
2

lCE M1c sin(

(12)

M1

(14)

Journal of Bionic Engineering (2011) Vol.8 No.4

436
'H Gy

c
H Gy

't

M
1
 M0 )  lCE M1 M1c sin( 1  M0 ) 
2
2
2
T1  T 0 1
T T
)  lEG (T1  T 0 ) (T1  T 0 )c sin(G 0  1 0 ).
lEG (T1  T 0 )c cos(G 0 
2
2
2

lCE M1c cos(

M1

The control of direction relies on the adjustment of

the reflection between and . So the bearing of velocity at point G results in

DG

actan

c
H Gy

actan

c
LGx

'H Gy
'LGx

k1  k2 sin(G 0 

T1  T 0

)
2
actan
,
T T
k1  k2 cos(G 0  1 0 )
2

lCE M1 sin(

M1

 M0 ), k2 lEG (T1  T 0 ).
2
By Eq. (10), the center of mass off angle is

where k1

W i 3 actan

vOy
vOx

Fa 2
(3l  a ), (0 d a d l ),
6 EI

3.4 Advantages of bionic mechanism

In summary, the rigid-flexible mechanism of hind
leg has advantages in jumping movement. At takeoff
preparation stage, a minor force Ff can finish the posture
adjustment. At the energy storage stage, Ff and Fe compress the spring and keep the mechanism in balance. At
the first acceleration stage, the knee extension torque
generated by Fe keeps increase, meanwhile the torque
generated by elastic force of spring has the same direction, which ensures the fast reaction. At the moment
before jumping, the flexible deformation of tarsus provides a second acceleration, which prolongs contact time
with ground and improve the energy efficiency.

4 Kinematic analysis

(i 3 0,1, 2,...).

All the analysis above is made on the assumption

that tarsus is fixed to the ground. In fact, the tarsal limb
gradually adjusts its location, and the contraction of tibia
causes the tarsus getting close to the body. Then the
tarsal foot gradually begins to play a major role in supporting the body, so as to get prepared for jump.
At second stage of acceleration in takeoff phase, the
torque produced by tarsus and its motion are considered
making the greatest contribution to jump movement, as
shown in Fig. 13. It should be noted that the two stage
acceleration is not strictly separated by time, but only a
matter of the influence on the jump. Based on the athletic
physiology study of locusts, the tarsus can be approximately simplified to a cantilever beam. The deflection Y
can be written as
Y

4.1 Pseudo-rigid body modeling

The model of unilateral hind limb and the body of
locust is simplified as Fig. 14. Focused on motion in the
sagittal plane, this section builds the D-H systems and
analyzes the kinematics of flexible-rigid model with the
method of pseudo-rigid body modeling. First, the model
can be described by D-H method. The homogeneous
transformations of linkages are given as

0
1

cos(T1 (t ))  sin(T1 (t ))
sin(T (t )) cos(T (t ))
1
1

0
0

0
0

0
0
1
0

1
2

cos(T 2 (t ))  sin(T 2 (t ))
sin(T (t )) cos(T (t ))
2
2

0
0

0
0

0 lo12
0
0
,
1
0
0
1

x (t )
y (t )
,
0
1

(16)

where E and I are the elastic modulus and inertia moment; F, a, l denote force, the force position and length
of beam, respectively.

(a)

(b)

(c)

(d)

Fig. 13 The distortion and torque produced by tarsus.

(15)

Fig. 14 The D-H systems of the proposed mechanism.

Chen et al.: Bionic Mechanism and Kinematics Analysis of Hopping Robot Inspired by Locust Jumping

2
3

cos(  T3 (t ))  sin(  T3 (t ))
sin(  T (t )) cos(  T (t ))
3
3

0
0

0
0

0 lo 23
0 0
,
1 0
0 1

3
4

cos(  T 4 (t ))  sin(  T 4 (t ))
sin(  T (t )) cos(  T (t ))
4
4

0
0

0
0

0 lo 34
0 0
,
1 0
0 1

4
5

0
i

T 12T " i 1iT

0
1

1
0

0 0 lo 45
1 0 0
,
0 1 0

0 0 1

nx ox
n o
y y
0 0

0 0

In Eq. (17) (x(t), y(t)) denote the centroid trajectory

and i denotes the rotation angle of joint i. The lo(i)(i+1) is
defined as the distance between two adjacent joints oi
and oi+1. Then, it is significant to reconsider the elastic
stretch of knee joint and the flexible deformation of
tarsus (Fig. 14). The stiffness of the spring is k. Approximately, the tarsus is assumed to be equal cross
section bar. Therefore, according to Eq. (16), the transformation equations should be modified as
cos(  T3 (t ))  sin(  T3 (t ))
sin(  T (t )) cos(  T (t ))
3
3
2

T
=
3

0
0

0
0

0
4
5T =
0

0 0

0 lo 23  k 'x

0
0
, (18)

1
0

0
1

Fa
(3l  a )
.
6 EI

1
lo 45

1 0
0 1
0 0

(Table 3). Therefore, the transformation from the joint

space q(1, 2, 3, 4) to motion space x of model is the
motion equation x = x(q). Then the relationship between
the generalized speed x i (including translational speed
vi and angular velocity i) of each component and the
generalized velocity q of joints is obtained as x 3 J (q)q ,
namely,
vi

i

0 pxi
0 p yi
(i 1, 2,...,5). (17)
1 pzi

0 1

(19)

Finally, the posture of each joint can be derived

according to Eq. (17).
4.2 Description of limb posture and speed

It is easier to ascertain 01T according to the experimental result in section 2. 01T indicates the jumping
performance such as height, horizontal distance and
body posture. The relationship between the rotation
angle of joint and time series has been measured

437

J l1 J l 2
J
a1 J a 2

T1

" J li T2
(i 1, 2,...,5),
" J ai #

Ti

where J(q) denotes the generalized velocity Jacobin.

Based on the vector product method, because the model
only consists of the rotary joints, Ji is calculated by
Ji

zi u  0i R i pn 

.
zi

(20)

4.3 Calculation case analysis

From the experimental data in Table 3, the joint
space can be got from takeoff preparation to takeoff
instance. The trajectory curve in Fig. 5 indicates the
centroid position (x(t), y(t)). Lo12 and law represent the
lengths from centroid to femur and abdomen respectively, Lo12, Lo34, Lo45 represent the lengths of linkages
between two adjacent joints, all of which can be obtained from Table 2. Assume that k is equal to 1 and joint
shrinkage x is 1.5103 m. Moreover, the F, E and I are
estimated to be 0.03 N, 1 MPa and 106, respectively. As
Fig. 13 indicates, the force position is in the middle of
the rod, that is a = 0.5lo45.
The homogeneous coordinate transformation can
be calculated through Eqs. (17) to (19). Then x(q) is
acquired according to the following equation
xi (n) 3 0iT (n, 4) (n 3 1, 2,3; i 3 1, 2,...,5) .

In MATLAB software, the position matrix can be

worked out at any time. At 24 ms, for example, the position matrix x(q(24)) irrespective of flexibility and
x(q(24)) taking flexibility into account are as

x (q(24))

35.5000
31.6765

19.4318

30.1404
27.0404

2.8000 0
2.0311 0
7.1457 0 ,

1.7972 0
1.7983 0

Journal of Bionic Engineering (2011) Vol.8 No.4

438
35.5000
31.6765

19.4332

30.1418
27.0418

x c(q (24))

2.8000 0
2.0311 0
7.1451 0 .

1.7967 0
1.7884 0

In consideration of flexibility, the posture with respect to time is depicted in Fig. 15. The flexible-rigid
mechanism has n = 5 joints, and the i pn0 3 0i R i pn of joint
i (i 3 1, 2,..., n) is calculated in 24 ms as below
p = 1 p50

= 10 R 1 p5

p50
0
2

p50

p50

R 2 p5

0
3

R 3 p5

p50
0
4

R 4 p5

0
5

R 5 p5

34.2523 28.4451 20.5774 30.1423 27.0423

= 9.7422 10.3340 2.2907 1.7877 1.7803 .
0

0
0
0
0

Then, according to Eq. (20), at this time, the velocity Jacobin matrix can be expressed by

1.7803
9.7422 10.3340 2.2907 1.7877
34.2523 28.4451 20.5774 30.1423 27.0423

0
0
0
0

.
0
0
0
0
0

0
0
0
0

1
1
1
1
1

Similarly, the Jacobin matrix at any time can be

calculated and then the linear velocity of each joint can
be computed with the angular velocity.
40

Acknowledgement
This work is financially supported by the National
Natural Science Foundation of China (Grant No.
51075014).

References
[1]

http://www.ai.mit.edu/projects/leglab/robots/robots.html
[2]

Kedzior K (eds). Theory and Practice of Robots and Ma-

t = 52 ms

nipulators, Proceedings of RoManSy'81, 1983, 352367.

[4]

y (mm)

[5]

15

Conference, Snowmass, CO, USA, 1999, 153158.

t = 40 ms

[6]

Fiorini P, Burdick J. The development of hopping capabili-

[7]

Brown B, Zeglin G. The bow leg hopping robot. Proceed-

t = 36 ms

ties for small robots. Autonomous Robot, 2003, 14, 239254.

t = 0, 4, , 32 ms
20

30

40
x (mm)

50

60

Fiorini P, Hayati S, Heverly M, Gensler J. A hopping robot

for planetary exploration. Proceedings of IEEE Aerospace

t = 44 ms

10

Pratt G. Legged robots at MIT Whats new since Raibert.

Clawer News, 2000, 4, 47.

t = 48 ms
20

10

Raibert M H. Dynamic stability and resonance in a

one-legged hopping machine. In: Morecki A, Bianchi G,

25

Raibert M H. Legged Robots that Balance, The MIT Press,

Cambridge, MA, USA, 1985.

30

MIT Leg Laboratory. Running & Hopping Robots,

[2011-3-10],

[3]

Centre of mass
Joint point
Body and limbs
Abdomen

35

ture and speed on jumping performance. A flexible-rigid

mixed model was established to simulate locust jumping.
Then the statics and the motional characteristics of
flexible-rigid mechanism at takeoff phase were analyzed.
At the same time, the kinematics of flexible-rigid
mixed model was described with the method of
pseudo-rigid body modeling, which achieves a unified
description from the joint space to posture space of the
takeoff process. The above analysis provides a foundation for the further design and kinematic planning of
flexible hopping robot.
The highlight of this paper is the rigid-flexible
model and its kinematics analysis, which is the
groundwork for further study of rigid-flexible mixture
dynamics. Further study will concentrate on the relationship between force and the jumping performance.

70

80

Fig. 15 The posture with respect to time.

ings of the 1998 IEEE International Conference on Robotics

and Automation, Leuven, Belgium, 1998, 781786.
[8]

Proceedings of the 1998 IEEE International Conference on

5 Conclusions
By means of experiment, this paper studied the
jumping motion of locust and the effect of takeoff pos-

Zeglin G, Brown B. Control of a bow leg hopping robot.

Robotics & Automation, Leuven, Belgium, 1998, 793798.

[9]

Hiraguchi T, Yamaguchi T, Takahata M. Mechanoreceptors

involved in the hindwing-evoked escape behaviour in

Chen et al.: Bionic Mechanism and Kinematics Analysis of Hopping Robot Inspired by Locust Jumping
cricket, Gryllus bimaculatus. Journal of Experimental Biology, 2003, 206, 523534.

439

and Systems, St. Louis, USA, 2009, 583588.

[13] Niiyama R, Nagakubo A, Kuniyoshi Y. Mowgli: A bipedal

[10] Kovac M. Bio-Inspired Jumping Locomotion for Miniature

jumping and landing robot with an artificial musculoskeletal

Robotics, PhD Thesis, Swiss Federal Institute of Technology,

system. IEEE International Conference on Robotics and

Switzerland, 2010.

Automation, Roma, Italy, 2007, 1014.

[11] Kovac M, Fuchs M, Guignard A. A miniature 7g jumping

robot. Proceedings of the 2008 IEEE International Conference on Robotics and Automation, Pasadena, CA, USA,
2008, 373378.

[14] Qin J D. Animal Locomotion, Tsinghua University Press,

Beijing, China, 2000. (in Chinese)
[15] Heitler W J. How Grasshoppers Jump. [2011-3-10],
http://www.st-andrews.ac.uk/~wjh/ jumping

[12] Kovac M, Schlegel M, Zufferey J-C, Floreano D. A minia-

[16] Heitler W J. The locust jump. III. Structural specializations

ture jumping robot with self-recovery capabilities. The 2009

of the metathoracic tibiae. Journal Experimental Biology,

IEEE/RSJ International Conference on Intelligent Robots

1977, 67, 2936.