Académique Documents
Professionnel Documents
Culture Documents
+
+
=
2 2
(2)
where the value of Tc in (2) is given as in (3).
g
Z
T
c
c
= (3)
The position equation shown in (2) is applicable for the
sagittal and lateral planes.
3.1.2 Velocity
In this sub-section, the derivation of the equation, which
is related to velocity in single linear inverted pendulum
model, is shown. The time derivative of (2), gives an
equation of the velocity, as shown in (4).
( )
c c
T
t
c
i
i
T
t
c
i
i
e
T
x
x
e
T
x
x
t x
+
+
=
2 2
(4)
The velocity equation shown in (4) is applicable for the
sagittal and lateral planes.
3.2 Trajectory planning method of DLLIPM
DLLIPM trajectory planning consists of 11 steps, as
shown in this sub-section. This trajectory planning is
based on an LIPM proposed by Kajita et al. with some
changes [8].
Step 1) Deciding the length of the short pendulum and
long pendulum: Suppose that the biped robot walks on a
flat floor at a certain CoM height. This CoM height on the
flat floor denotes the length of the long pendulum, ZLP.
The height of the higher to lower slope surfaces is a
known value, h. The length of the short pendulum, ZSP is
given by (5). One leg will be on the lower surface and one
leg will be on the higher surface. The pendulum on the
lower surface is denoted by the longer pendulum and the
pendulum on the higher surface is denoted by the shorter
pendulum. The length of ZLP in the sagittal plane is the
same as in the lateral plane. Also the length of ZSP in
sagittal and lateral planes is equal.
h Z Z
LP SP
= (5)
Step 2) Deciding on an odd and even number of landing
points, n
th
: Which foot lands on the higher ground or on
the lower ground needs to be determined based on the
forward moving direction of the biped robot. Also, with
which foot the robot starts to walk must be determined.
For example, the robot is walking along a slope as shown
in Figure 5b) and 6b). The robot moves in such a way that
the left foot always lands on the higher ground, the right
foot always lands on the lower ground and the motion
starts from the right foot. Therefore, the even and odd
numbers of landing points in this case are given as in (6)
and (7).
4 Int. j. adv. robot. syst., 2013, Vol. 10, 377:2013 www.intechopen.com
( ) LP RF n , 4 , 2 , 0 = (6)
( ) SP LF n , 5 , 3 , 1 = (7)
Step 3) Deciding the walking parameters such as
walking cycle, strides, initial CoM positions and initial
landing positions: in the conventional LIPM method,
these parameters should be decided before the robot
starts to walk. In DLLIPM, these parameters should also
be decided in the same manner. The walking cycles for
the short and long pendulum are denoted by tSP and tLP,
correspondingly.
Step 4) Setting the landing points modifications in the
sagittal and lateral planes, Px* and Py*, initial time T and
initial number n, of landing points: T:=0 and n:=0.
Initially Px*=0 and Py*=0.
Step 5) Using LIPM equations to obtain the CoM
trajectory: For the long pendulum, which means n as in
(6) in this example, the CoM is moved from T to T+tLP by
using (8) for the sagittal plane and (9) for the lateral plane.
Here the value of Zc is equal to ZLP. For the short
pendulum, which means n as in (7), the CoM is moved
from T to T+tSP also by using (8) for the sagittal plane
and (9) for the lateral plane. In this case, the value of Zc is
equal to ZSP.
( )
( )
( ) ( )
( )
( )
*
*
/ sinh
/ cosh
x c
n
i c
c x
n
i
P T t x T
T t P x t x
+ +
=
(8)
( )
( )
( ) ( )
( )
( )
*
*
/ sinh
/ cosh
y c
n
i c
c y
n
i
P T t y T
T t P y t y
+ +
=
(9)
where
g
Z
T
c
c
=
Step 6) Increment of number of landing points and time:
1 : + = n n (10)
If n is an even number, T:= T+tLP.
If n is an odd number, T:= T+tSP.
Step 7) Calculation of the next landing points in the
sagittal and lateral planes, Px
(n)
and Py
(n)
: The next
landing point for the sagittal and lateral planes is
calculated by using (11) and (12), respectively.
) ( ) 1 ( ) ( n
x
n
x
n
x
s P P + =
(11)
( )
) ( ) 1 ( ) (
1
n
y
n n
y
n
y
s P P =
(12)
where sx
(n)
and sy
(n)
mean the foot strides of the n
th
step in
the sagittal and lateral directions.
Step 8) Calculation of the final position and final
velocity of the CoM for each single support phase: If n is
an even number, the final CoM position and the final
velocity for the sagittal plane are calculated as in (13) and
(14), respectively. On the other hand, the final CoM
position and the final velocity for the lateral plane are
calculated as in (15) and (16), respectively.
( ) ( ) 1 +
+
=
n
x
SP LP
LP n
LP
s
x x
x
x (13)
( ) ( ) n
LP
c
n
LP x
x
S T
C
v
+
=
1
(14)
( )
( )
( ) 1
1
+
+
=
n
y
SP LP
LP
n n
LP
s
y y
y
y (15)
( ) ( ) n
LP
c
n
LP y
y
S T
C
v
1
(16)
If n is an odd number, the final CoM position and the
final velocity for the sagittal and lateral planes can be
solved in the same way as (13)-(16) but with the respect to
the short pendulum.
The values of C and S in (14) and (16) are given in (17)
and (18) (suppose that the instantaneous time at the end
of each single support phase is denoted by tend).
=
c
end
T
t
C cosh (17)
=
c
end
T
t
S sinh (18)
where the values of Tc in (14), (16), (17) and (18) are the
same values used in Step 5.
Step 9) Calculation of the desired final position and
desired final velocity of the CoM in the sagittal plane ,
d
x and
d
x , and in the lateral plane,
d
y and
d
y , for
each single support phase: If n is an even number, the
calculation of the desired final position of the CoM in the
sagittal plane is calculated from (11) and (13), as shown in
(19). The calculation of the desired final position of the
CoM in the lateral plane is calculated from (12) and (15),
as shown in (20). Furthermore, the desired final velocity
in the sagittal and lateral planes is calculated using (14)
and (16), as shown in (21) and (22), correspondingly.
5 Fariz Ali, Ahmad Zaki Hj. Shukor, Muhammad Fahmi Miskon, Mohd Khairi Mohamed Nor and Sani Irwan Md Salim:
3-D Biped Robot Walking along Slope with Dual Length Linear Inverted Pendulum Method (DLLIPM)
www.intechopen.com
( ) ( ) n
LP
n
x
d
x P x + = (19)
( ) ( ) n
LP
n
y
d
y P y + = (20)
( ) n
LP x
d
v x
= (21)
( ) n
LP y
d
v y
= (22)
If n is an odd number, the desired final position and
desired final velocity of the CoM for the sagittal and
lateral planes can be solved in the same way as (19)-(22)
but with the respect to the short pendulum.
Step 10) Calculation of foot landing modification: The
foot landing modification in the sagittal plane is
calculated by using (23) [8].
( )
( ) ( )
( )
( ) ( )
=
n
i
n
i
c
d
c
n
i c
n
i
d
x
x C x
T
S
x
D T
bS
x S T Cx x
D
C a
P
1
*
(23)
where D = a(C-1)
2
+ b(S/Tc)
2
.
a and b are the weight parameters used in order to
minimize the margin of error for position and velocity.
The values of Tc, C and S in (23) are the same values as in
Step 8. The foot landing modification in the lateral plane
can be solved in the same manner as in (23) but with all
the values from the lateral plane.
Step 11) Go back to Step 5: This step is repeated until all
of the landing points, n, are completed.
4. DLLIPM with Newton Raphson (proposed method)
It is very important for biped robots to have a
symmetrical and smooth motion during walking to
reduce the GRF. In order to achieve this, the position,
velocity and acceleration between single support phases
must be connected smoothly. Furthermore, in 3D walking
pattern generation, the sagittal and lateral planes must be
synchronized properly. Thus, the use of the Newton
Raphson method is proposed and will be presented in
this section.
4.1 Biped modelling by DLLIPM for 3-D walking along a slope
For modelling, the long pendulum and short pendulum
phases are represented by Phase 1 and Phase 2 as shown
in Figure 9 and 10. Phase 1 is defined as the walking cycle
of the biped robot when the robot is supported by the
foot on the lower slope surface, which means the height
of the pendulum in this cycle is referred to as the long
pendulum (LP). On the other hand, Phase 2 is defined as
the walking cycle of the biped robot when the robot is
supported by the foot on the higher slope surface, which
means the height of pendulum in this cycle is referred to
as the short pendulum (SP).
Figure 9. DLLIPM model for walking along slope in the sagittal
plane
Figure 10. DLLIPM model for walking along slope in the lateral
plane
4.1.1 Definitions of parameters
All the parameters involved in Figure 9 and 10 are
explained and defined in Table 1. Where k in Table 1 is
either one or two, which is referred to as Phase 1 or 2,
respectively.
4.1.2 Assumptions
All derivations shown later are based on the following
five assumptions:
Assumption 1: The initial position and initial velocity of
Phase 1 are known.
Assumption 2: The height of the higher slope surface, h,
which the robot will walk on, is a known value.
6 Int. j. adv. robot. syst., 2013, Vol. 10, 377:2013 www.intechopen.com
Parameters Definitions
LP
Z CoM height of long pendulum
SP
Z CoM height of short pendulum
xk
S ,
yk
S Possible maximum stride of phase k
( ) ( ) t y t x
k k
, Instantaneous position of phase k
ki
x ,
ki
y Initial position of phase k
ki
x ,
ki
y Initial velocity of phase k
ki
x ,
ki
y Initial acceleration of phase k
kf k
x x = ,
kf k
y y =
Final position of phase k
kf
x ,
kf
y Final velocity of phase k
kf
x ,
kf
y Final acceleration of phase k
sk
T Instantaneous time at the end of phase k
Table 1. Definitions of parameters, where k is 1 or 2 which is
referred as phase 1 or 2, respectively
Assumption 3: The CoM height of the long pendulum
and the short pendulum are known.
Assumption 4: The trajectories of the long and short
pendulums are symmetrical. Reference of the CoM position
for each phase is the middle of each supporting foot.
Assumption 5: There are six free parameters that need to
be solved which are Sx1, Sx2, Sy1, Sy2, Ts1 and Ts2. Sx1, Sx2, Sy1,
Sy2 are chosen because symmetrical pendulum motion can
be determined when these parameters are made to be half
as shown in the constraints section later. Furthermore, Ts1
and Ts2 are chosen because these walking cycle times are
also important in order to determine symmetrical
pendulum motion when the initial position and velocity
have been decided.
4.1.3 Constraints
In order to ensure smooth motions and symmetrical
pendulums, there are several constraints that must be
enforced. In total, there are six constraints, which result in
six equations, as shown by (24)-(29). The constraints are
listed as follows:
1) By using the position equations, the pendulums for
Phase 1 and 2 are made symmetrical by the following
equations:
2
1
1
x
f
S
x + = (24)
2
2
2
x
f
S
x + = (25)
2
1
1
y
f
S
y + = (26)
2
2
2
y
f
S
y = (27)
2) In order to ensure a smooth trajectory of the velocity
and synchronization between the sagittal and lateral
planes, a constraint as in (28) is enforced.
( ) ( ) ( ) ( )
2
2
2
2
2
1
2
1 f f f f
y x y x + = + (28)
3) In order to ensure a smooth trajectory of the
acceleration and synchronization between the sagittal and
lateral planes, a constraint as in (29) is enforced.
( ) ( ) ( ) ( )
2
2
2
2
2
1
2
1 f f f f
y x y x + = + (29)
4.2 Solution of Sx1, Sx2, Sy1, Sy2, TS1 and TS2 using the Newton-
Raphson method
In order to find a solution for appropriate Sx1, Sx2, Sy1, Sy2, TS1
and TS2 parameters, a numerical method known as Newton-
Raphson is chosen. There are several steps that need to be
taken, which will be discussed further in this section.
4.2.1 Simplification of equations
Before the Newton-Raphson method can be implemented,
long and complex equations must be simplified.
Therefore, (30)-(39) are defined.
1
1
c
s
T
T
e = (30)
2
2
c
s
T
T
e = (31)
+
=
2
1 1 1 i c i
x T x
A
(32)
=
2
1 1 1 i c i
x T x
B
(33)
+
=
2
2 2 2 i c i
x T x
C
(34)
=
2
2 2 2 i c i
x T x
D
(35)
+
=
2
1 1 1 i c i
y T y
E
(36)
=
2
1 1 1 i c i
y T y
F
(37)
7 Fariz Ali, Ahmad Zaki Hj. Shukor, Muhammad Fahmi Miskon, Mohd Khairi Mohamed Nor and Sani Irwan Md Salim:
3-D Biped Robot Walking along Slope with Dual Length Linear Inverted Pendulum Method (DLLIPM)
www.intechopen.com
+
=
2
2 2 2 i c i
y T y
G
(38)
=
2
2 2 2 i c i
y T y
H
(39)
The previous constraint in (24) is realized using (2), as
shown in (40).
2 2
2
1 1 1 1
1 1 1
1
1
1
1
1
x T
T
i c i
T
T
i c i
f
S
e
x T x
e
x T x
x
c
s
c
s
+ =
+
+
(40)
By substituting (30), (32) and (33) into (40), f1(q) is defined
as follows.
( ) B
S
A q f
x
+ =
2
1 2
1
(41)
The previous constraint in (25) is also realized using (2),
as shown in (42).
2 2
2
2 2 2 2
2 2 2
2
2
2
2
2
x T
T
i c i
T
T
i c i
f
S
e
x T x
e
x T x
x
c
s
c
s
+ =
+
+
(42)
By substituting (31), (34) and (35) into (42), f2(q) is
obtained as follows.,
( ) D
S
C q f
x
+ =
2
2 2
2
(41)
The previous constraint in (26) is realized using (2), as
shown in (44).
2 2
2
1
1 1 1
1 1 1
1
1
1
1
1
y T
T
i c i
T
T
i c i
f
S
e
y T y
e
y T y
y
c
s
c
s
+ =
+
+
(44)
By substituting (30), (36) and (37) into (44), f3(q) is defined
as follows.
( ) F
S
E q f
y
+ =
2
1
2
3
(45)
The previous constraint in (27) is also realized using (2),
as shown in (46).
2 2
2
2
2 2 2
2 2 2
2
2
2
2
2
y T
T
i c i
T
T
i c i
f
S
e
y T y
e
y T y
y
c
s
c
s
=
+
+
(46)
By substituting (31), (38) and (39) into (46), f4(q) is defined
as follows.
( ) H
S
G q f
y
+ + =
2
2
2
4
(47)
The previous constraint in (28) is realized using (4), (30)-
(39), as shown in (48). (48) is assigned as f5(q).
2
2
2
2
2
2
2
1
1
2
1
1
2
2
1
2
2
1
2
2
1
2
2
1
y
c
x
c
y
c
x
c
S
G
T
S
C
T
S
E
T
S
A
T
(48)
Then, f5(q) is expanded, as shown in (49).
( )
2
2
2
2
2
2
2
2
2
2 2
2
2
2
2
2
2
2
2
2
2 2
2
1
2
1
2
1
1
2
1
2 2
2
1
2
1
2
1
1
2
1
2 2
5
4
2
4
4
2 4
4
2
4
4
2 4
c
y
c
y
c
c
x
c
x
c
c
y
c
y
c
c
x
c
x
c
T
S
T
S G
T
G
T
S
T
S C
T
C
T
S
T
S E
T
E
T
S
T
S A
T
A
q f
+
+
+ +
=
(49)
8 Int. j. adv. robot. syst., 2013, Vol. 10, 377:2013 www.intechopen.com
The previous constraint in (29) is realized using a time
derivative of (4), (30)-(39), as simplified in (50). (50) is
assigned as f6(q).
2
2
2
2
2
2
2
2
2
2
1
1
2
2
1
1
2 2
2 2
|
|
.
|
\
|
+
|
|
.
|
\
|
=
|
|
.
|
\
|
+
|
|
.
|
\
|
c
y
c
x
c
y
c
x
T
S
T
S
T
S
T
S
(50)
Then, f6(q) is defined by expanding (50) as follows.
( )
4
2
2
2
4
2
2
2
4
1
2
1
4
1
2
1
6
c
y
c
x
c
y
c
x
T
S
T
S
T
S
T
S
q f + = (51)
All the functions are arranged as in (52). In (52) the values
of q are the variables or parameters that need to be solved,
as shown in (53).
( ) ( ) ( ) ( ) ( ) ( ) ( ) [ ]
T
q f q f q f q f q f q f q f
6 5 4 3 2 1
=
(52)
[ ]
T
y y x x
S S S S q
2 1 2 1
=
(53)
4.2.2 Newton Raphson algorithm
The Newton Raphson method involves a process of
iterations. The objective is to obtain the roots from several
complex functions. In this method, the roots are
calculated using (54).
( )
( )
n
n
n n
q f
q f
q q
'
1
=
+
(54)
( )
n
q f
'
in (54) is the derivative of the functions, which is
obtained as in (55).
( )
6 6
6
6
1
6
6
1
1
1
'
(
(
(
(
(
=
q
f
q
f
q
f
q
f
q f
n
(55)
The value of each element in the matrix above is then
obtained.
5. Simulation
5.1 The known parameters and initial guess values
for Newton Raphson calculations
In the Newton Raphson method there are known
parameters and initial guess values for iterations that
must be given. These parameters are shown in Table 2.
The meaning of each parameter in Table 2 can be
determined from Table 1.
Parameters Values Type
LP
Z 0.5376 m Known value
SP
Z 0.4376 m Known value
10 x
S 0.3280 m Initial guess
i
x
1
- Sx10/2 = - 0.164 m Known value
i
x
1
0.7887 m/s Initial guess
20 x
S 0.2800 m Initial guess
i
x
2
- Sx20/2 = - 0.1027 m Known value
i
x
2
0.7887 m/s Initial guess
10 y
S 0.2053 m Initial guess
i
y
1
Sy10/2 = 0.1027 m Known value
i
y
1
- 0.33 m/s Initial guess
20 y
S 0.1616 m Initial guess
i
y
2
- Sy20/2 = - 0.0808 m Known value
i
y
2
0.33 m/s Initial guess
1 s
T 0.9 s Initial guess
2 s
T 0.6 s Initial guess
1 c
T ( ) g Z
LP
/ Known value
2 c
T ( ) g Z
SP
/ Known value
0
e
(Ts1/Tc1)
Initial guess
0
e
(Ts2/Tc2)
Initial guess
Table 2. Parameters for Newton Raphson calculations
5.2 Results of the iterations with the Newton Raphson method
From the values used in Sub-section 5.1, iterations are
done by using (54). The results are shown in Figure 10
and 11. is obtained as 10.5198 at the 5
th
iteration and
is obtained as 10.2831 at the 4
th
iteration. Furthermore,
Sx1, Sx2, Sy1 and Sy2 are obtained as 0.1842m, 0.2438m,
0.2841m and 0.1287m, correspondingly. During the
iterations process of the Newton Raphson method,
different initial settings and different numbers of
iterations may give different solutions. However, the
solutions for all the parameters, , , Sx1, Sx2, Sy1 and Sy2
must always be positive. It should not be negative
because and represent the time, which cannot be
negative. Sx1, Sx2, Sy1 and Sy2 are the distance of the
centre of mass maximum trajectory, which should also
always be positive.
9 Fariz Ali, Ahmad Zaki Hj. Shukor, Muhammad Fahmi Miskon, Mohd Khairi Mohamed Nor and Sani Irwan Md Salim:
3-D Biped Robot Walking along Slope with Dual Length Linear Inverted Pendulum Method (DLLIPM)
www.intechopen.com
Figure 10. Results of and after several iterations by using the
Newton Raphson algorithm
Figure 11. Results of Sx1, Sx2, Sy1 and Sy2 after several iterations by
using the Newton Raphson algorithm
In order to verify all the functions from the Newton
Raphson solutions, Figure 12 is plotted. In this figure, all
functions are obtained as '0' values after all the
parameters are converged. This validates that all the
constraints have been fulfilled.
5.3 Simulation of 3D walking pattern
with DLLIPM (without Newton Raphson)
In this sub-section, a simulation of a biped robot walking
along a slope with the height of 10cm is presented. All
simulations in this paper are done for a slope, which
means the floor is not horizontally flat. Therefore, the foot
should be made parallel to the slope surface and realized
with an orientation command [15]. The strides are
decided as in Table 3.
Figure 12. Results of all functions from f1(q) to f6(q) by using the
Newton Rahpson algorithm
n 1 2 3 4 5 6 7 8 9 10
Sy
(n)
[m] 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2 0.2
Sx
(n)
[m] 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3 0.3
Table 3. Strides table
ZLP, t-LP and t-SP are 0.5376m, 0.9s and 0.6s
correspondingly. Here, the parameter settings are chosen
intuitively without the Newton Raphson method. With these
settings, the trajectory planning method in sub-Section 3.2 is
used. The result for the sagittal plane is obtained as in
Figure 13.The result for lateral plane is shown in Figure 14.
As for the sagittal plane, it is noticed from Figure 13 that
the CoM position trajectory shown by the blue line is
accelerated and decelerated repeatedly. Thus, the impact
force may increase. It is also noticed that there are multiple
levels of maximum and minimum velocity for the long
pendulum and short pendulum cycles. These multiple levels
of velocity will create unsmooth motion during walking.
As for the lateral plane, it is observed from Figure 14 that
the CoM position trajectory becomes deviated from its
initial straight position. The deviated problem is more
critical during the first few steps. It is also observed in
this figure that the shape of the velocity trajectory, shown
by the red line is not symmetrical.
5.4 Simulation of 3D walking pattern
with improved DLLIPM (with Newton Raphson)
By using DLLIPM algorithm shown earlier in Sub-section
3.2, the walking pattern for the environment or situation
in Sub-section 5.3 is simulated again with the new
walking parameters obtained from Sub-section 5.2. The
result for the sagittal and lateral planes is obtained as in
Figure 15 and 16, respectively.
As for the sagittal plane, it is noticed in Figure 15 that the
trajectory of the CoM does not accelerate and decelerate
repeatedly as much as the previous situation, shown in the
Figure 13. Thus, the impact force may be reduced. Besides
that, it is noticed that the minimum velocity for the long
pendulum and short pendulum cycles is now at the same
value, which results in a smoother walking motion.
Figure 13. Walking pattern of ten steps in the sagittal plane
(without proposed method)
10 Int. j. adv. robot. syst., 2013, Vol. 10, 377:2013 www.intechopen.com
Figure 14. Walking pattern of ten steps in the lateral plane
(without proposed method)
Figure 15. Walking pattern of ten steps in the sagittal plane (with
proposed method)
Figure 16. Walking pattern of ten steps in the lateral plane (with
proposed method)
As for the lateral plane, it is observed that the motion of
the CoM does not deviate from its initial straight position
as shown in Figure 16. Furthermore, it is also noticed
from this figure that the problem of the unsymmetrical
shape of velocity trajectory has been solved. Smooth
trajectories will help a bipedal robot to achieve safer and
more stable walking. From all the results illustrated, it is
verified that with the proposed method, the biped
walking pattern for walking along a slope has been
improved.
5.5 Ground reaction force (GRF)
Simulations of biped robot walking with consideration of
the lateral and sagittal planes are conducted for walking
along a slope. The simulations are conducted for walking
with LIPM and DLLIPM approaches, as explained earlier
Figure 17. GRF data measured from force sensors during
walking simulation (without proposed method)
Figure 18. GRF data measured from force sensors during
walking simulation (proposed method)
in Figure 5 and 6. In other words, these simulations are
done for comparison between the conventional method
and the proposed method. In these figures, the RF always
lands on the lower slope surface. Whereas the LF always
lands on the higher slope surface during walking. The
GRF during these walking simulations is measured with
the force sensors located at the RF and LF of the biped
robot. The results are shown in Figure 17 and 18.
The solid lines in Figure 17 and 18 represent forces
measured at the RF of the biped robot. Furthermore, the
dashed lines in these figures are the forces measured at
the LF of the biped robot. It is noticed from the red circles
shown in Figure 17 that the maximum force fluctuation is
415 [N] at 5.4[s] and 4.38[N] at 8.2[s]. The maximum force
fluctuation is reduced to 380[N] at 5.4[s] and 410[N] at
8.3[s] with the proposed method, as shown in Figure 18.
These facts demonstrate the effectiveness of the proposed
method in order to reduce the GRF.
6. Conclusions and future work
In this paper, a generalized method known as DLLIPM
with the Newton Raphson algorithm is proposed in order
to determine parameters of 3D biped walking along a
slope. By using DLLIPM, different heights of pendulum
are applied at the left and right legs.
However, when different heights of pendulum are
applied, difficulty occurred in obtaining symmetrical and
smooth pendulum motions. Furthermore, synchronization
11 Fariz Ali, Ahmad Zaki Hj. Shukor, Muhammad Fahmi Miskon, Mohd Khairi Mohamed Nor and Sani Irwan Md Salim:
3-D Biped Robot Walking along Slope with Dual Length Linear Inverted Pendulum Method (DLLIPM)
www.intechopen.com
between the sagittal and lateral planes is not confirmed.
Therefore, DLLIPM with the Newton Raphson algorithm
is proposed in order to solve these problems. As a result,
smoother walking trajectories and more symmetrical
pendulum motions are achieved. Furthermore, it is
verified that the maximum force fluctuation is reduced
with the proposed method. In the future, the concept of
DLLIPM will be expanded for other possible motions
such as diagonal walking, turning on a slope and stairs.
7. Acknowledgements
The authors would like to thank Universiti Teknikal
Malaysia Melaka (UTeM) for the funding of this research
with grant number PJP/2013/FKE(11C)/S01188 and also to
Professor Atsuo Kawamura from Yokohama National
University for all the great idea.
8. References
[1] S. Panich and N. Afzulpurkar, "Mobile Robot
Integrated with Gyroscope by Using IKF,"
International Journal of Advanced Robotics Systems,
vol.8, no.2, pp. 122-136, 2011.
[2] G. Carbone and M. Ceccarelli, "A Low-Cost Easy-
Operation Hexapod Walking Machine," International
Journal of Advanced Robotics Systems, vol.5, no.2, pp.
161-166, 2008.
[3] U. Asif and J. Iqbal, "Motion Planning Using an
Impact-Based Hybrid Control for Trajectory
Generation in Adaptive Walking," International
Journal of Advanced Robotics Systems, vol.8, no.4, pp.
212-224, 2011.
[4] K. Izumi, M. K. Habib and R. Sato, "Behavior
Selection Based Navigation and Obstacle Avoidance
Approach Using Visual and Ultrasonic Sensory
Information for Quadruped Robots," International
Journal of Advanced Robotics Systems, vol.5, no.4, pp.
379-388, 2008.
[5] M. Vukobratovic and B. Borovac, "Zero-Moment
Point Thirty Five Years of Its Life," Int. J. of
Humanoid Robotics, vol.1, no.1, pp. 157-173, 2004.
[6] N. Aphiratsakun and M. Parnichkun, "Balancing
Control of AIT Leg Exoskeleton Using ZMP based
FLC," International Journal of Advanced Robotics
Systems, vol.6, no.4, pp. 319-328, 2009.
[7] J. Tang, Q. Zhao and R. Yang, "Stability Control for a
Walking-Chair Robot with Human in the Loop,"
International Journal of Advanced Robotics Systems,
vol.6, no.1, pp. 47-52, 2009.
[8] S. Kajita, F. Kanehiro, K. Kaneko, K. Yokoi, and H.
Hirukawa, "The 3-D Linear Inverted Pendulum
Mode: A Simple Modeling for a Biped Walking
Pattern Generation", IEEE/RSJ International Conference
on Intelligent Robots and Systems, pp. 239-246, 2001.
[9] T. Tsuji and K. Ohnishi, "A Control of Biped Robot
which Applies Inverted Pendulum Mode with Virtual
Supporting Point", IEEE International Workshop on
Advanced Motion Control, pp. 478-483, 2002.
[10] J.-Y. Kim, I.-W. Park and J.-H. Oh, "Walking Control
Algorithm of Biped Humanoid Robot on Uneven and
Inclined Floor," Journal of Intelligent and Robotic
Systems, vol. 48, no. 4, pp. 457-484, 2007.
[11] P. Song-hao, L. Ya-qi, Z. Wen and Z. Qiu-bo,
"Application and Research of Humanoid Robot
Based on Second-Order Cone Programming,"
International Journal of Advanced Robotics Systems,
vol.8, no.2, pp. 22-28, 2011.
[12] T. Sato, S. Sakaino, E. Ohashi and K. Ohnishi,
"Trajectory Planning on Stairs using Virtual Slope for
Biped Robots," IEEE Transactions on Industrial
Electronics, vol. 58, no.4, pp. 1385-1392, 2011.
[13] Z. Qiu-bo , P. Song-hao and G. Chao, "Motion
Planning for Humanoid Robot Based on Hybrid
Evolutionary Algorithm," International Journal of
Advanced Robotics Systems, vol.7, no.3, pp. 209-216,
2010.
[14] C. Fu and K. Chen, "Gait Synthesis and Sensory
Control of Stair Climbing for A Humanoid Robot,"
IEEE Transactions on Industrial Electronics, vol. 55,
no.5, pp. 2111-2120, 2008.
[15] F. Ali, A. Che Amran and A. Kawamura, "Bipedal
Robot Walking Strategy on Inclined Surfaces using
Position and Orientation based Inverse Kinematics
Algorithm," Int. Confon Control, Automation, Robotics
and Vision, pp. 181-186, 2010.
[16] T. Hirano, T. Sueyoshi and A. Kawamura,
"Development of ROCOS (Robot Control Simulator)-
Jump of Human-Type Biped Robot by The Adaptive
Impedance Control," 6
th
International Workshop on
Advanced Motion Control, pp. 606-611, 2000.
12 Int. j. adv. robot. syst., 2013, Vol. 10, 377:2013 www.intechopen.com