Stabilization of Six-Legged Robot on Tilt Surface With 9 DOF IMU Based on Invers Kinematic

International Journal of Research in Engineering and Science (IJRES)
ISSN (Online): 2320-9364, ISSN (Print): 2320-9356
www.ijres.org Volume 3 Issue 6 ǁ June 2015 ǁ PP.45-53
www.ijres.org 45 | Page
Stabilization of Six-Legged Robot on Tilt Surface With 9 DOF
IMU Based on Invers Kinematic
Sumardi1
, Muhammad Asrofi1
1
(Department of Electrical Engineering, DiponegoroUniversity, Indonesia)
ABSTRACT: Robot is a tool which is developed very fast. There are several types of robots, one of them is six-
legged robot. One of the problems of this robot is when the robot walks on the tilt surface. This would result the
movement of the robot could be late and the center of gravity is not balanced. In this research, stabilization of
six-legged robot walking on tilt surface using nine degree of freedom (DOF) inertial measurement unit (IMU)
sensor based on invers kinematic is designed. The IMU sensor comprises a gyroscope, a magnetometer, and
three-axis accelerometer. This sensor works as the input of the tilt degree and heading of the robot, therefore
they can be processed in fuzzy-pid controller to balance the body of the robot on tilt surface. The results show
that the robot will move forward when the x-axis translation inverse changed from its original position, move
aside when the y-axis translational modified and move up and down if the translation to the z-axis was
changed. From the testing of IMU get the total of RMSE pitch is 1,73%, roll =1,67% and yaw = 1,24%. In
controller fuzzy-pid get the good respon is on the value Kp have k1=0,5, k2=1 , k3 = 3 , Ki have k1=0,5 ,
k2=0,5, k3=0,5 and Kd have k1=0,25 , k2=0,35 dan k3=0,45.
Keywords -Hexapod, IMU, Fuzzy-PID
I. INTRODUCTION
Robot is one of familiar lessons in control and instrument’s students. The development of robotic is so fast
with so many new system which has founded. One of the types of robots is six-legged robot (hexapod).
Hexapod moves with its legs which is designed to make the body of robot in balance. Legs are built from servo
motor[1].
Some disadvantage of hexapod are if the robot find the tilt surface. In this area, robot’s movement can cause
negative effect on the center of gravity; causing imbalance in the robot. It can be disturbance on the servo
motors which given the biggest load and so the servo motor can be broken faster than before[1].
Because of that, the control of the body of the robot needed to make the body of the hexapod still balance in
one line to reduce error the center of gravity and so it can reduce the broken of servo motor. One solution to
make stabilization of hexapod in tilt surface is with developing invers kinematic. This can give the movement of
hexapod based on X, Y and Z coordinate[2].
Beside on kinematic’s of movement, the sensor needs to give the reference of orientation degrees to
compare with the orientaion degree in robot. The sensor are MPU 6050 and HMC 5883 these sensors built from
accelerometer, gyroscope and magnetometer. The method of control is fuzzy-PID to give the value of SetPoint
and make hexapod still balance in tilt surface.
II. METHOD
2.1 Fuzzy-PID Controller
In this experiment, stabilization of legged robot in tilt surface is made with fuzzy controller for tuning the
parameter of PID. The value of membership function of fuzzy is given by the error of the orientations which
tested before. This membership function can make the parameter of PID. The block diagram of system is show
in figure 1 :

Stabilization of Six-Legged Robot on Tilt Surface With 9 DOF IMU Based on Invers Kinematic
Figure 1.Blok diagramcontroller
The inputs of fuzzy are error and delta error as show in figure of membership function in figure 2 and figure
3.
Figure 2.membership function with input is error.
Figure 3. Membership function with input is delta error
The membership function then make in rule base of fuzzy for tuning Kp in PID as show in table 1:
Table 1 Rule based of fuzzyfor tuning Kp
Small Medium Big
Small 0.5 0.5 0.5
Medium 0.5 1 1
Big 0.5 1 3
Table 2 Rule based of fuzzy for tuning Ki
Small Medium Big
Small 0.5 0.5 0.5
Medium 0.5 0.5 0.5
Big 0.5 0.5 0.5
Table 3 Rule based of fuzzy fortuning Kd
Small Medium Big
Small 0.25 0.25 0.25
Medium 0.25 0.35 0.35
Big 0.25 0.35 0.45
The value of the rule based above is the single tone output which used in the PID controller as the
parameter of controller.
-1 1-0.5 0 0.
5
Small Medium Big
-50 50-20 0 20
Small Medium Big
EΔE
EΔE
EΔE

2.2 Invers Kinematic
Invers Kinematic is defined how to get the position of legs with the calculation both of rotation every
joint. This six legs robot uses invers kinematic with 3 degree of freedom. Degree of freedom is a configuration
of mechanic system which measured how far the system can follow the track which is given in [2].
The representation graphic of legged robot in show in figure 4:
Figure 4. Graphic representation of legged robot
There are two methods to make the solution in invers kinematic, they are algebra solution and
geometric solution. Both of them are same to solve the problem, but the authors just used one of the solutions
that is algebra solution [2].
Which has known, that the calculation of forward kinematic is show in figure 4 above so get the
formula in point x2 and y2 in formula (1) and (2) :
𝑥2 = 𝐿1 cos 𝐴1 + 𝐿2 cos(𝐴1 + 𝐴2) (1)
𝑦2 = 𝐿1 sin 𝐴1 + 𝐿2 sin(𝐴1 + 𝐴2) (2)
To get the output based on distance which needed, so the calculation to search the value of A1 and A2
as shown:
Quadratic of x and y on the formula (1) and (2) :
[𝑥2]2
= [𝐿1 cos 𝐴1 + 𝐿2 cos(𝐴1 + 𝐴2]2
(3)
𝑥2
2
= 𝐿1
2
𝑐𝑜𝑠2
(𝐴1) + 2𝐿1 𝐿2 cos 𝐴1 cos 𝐴1 + 𝐴2 + 𝐿2
2
𝑐𝑜𝑠2
𝐴1 + 𝐴2 (4)
[𝑦2]2
= [𝐿1 sin 𝐴1 + 𝐿2 sin(𝐴1 + 𝐴2)] 2
(5)
𝑦2
2
= 𝐿1
2
𝑠𝑖𝑛2
(𝐴1) + 2𝐿1 𝐿2 sin 𝐴1 sin 𝐴1 + 𝐴2 + 𝐿2
2
𝑠𝑖𝑛2
𝐴1 + 𝐴2 (6)
Sum the x and y which has calculate in formula (4) and (6) :
𝑥2
2
+ 𝑦2
2
= 𝐿1
2
𝑠𝑖𝑛2
𝐴1 + 𝑐𝑜𝑠2
𝐴1 + 𝐿2
2
[𝑠𝑖𝑛2
𝐴1 + 𝐴2 + 𝑐𝑜𝑠2
+2𝐿1 𝐿2 [𝑠𝑖𝑛 𝐴1 𝑠𝑖𝑛 𝐴1 + 𝐴2 + 𝑐𝑜𝑠 𝐴1 𝑐𝑜𝑠(𝐴1 + 𝐴2)]
(7)
As we know that:
𝑠𝑖𝑛2
∝ + 𝑐𝑜𝑠2
∝ = 1 (8)
We can get the new formula from formula (7):
𝑥2
2
+ 𝑦2
2
= 𝐿1
2
+ 𝐿2
2
+ 2𝐿1 𝐿2 [sin 𝐴1 sin 𝐴1 + 𝐴2 + cos 𝐴1 cos 𝐴1 + 𝐴2 ] (9)
Then we have the calculation:
𝑥2
2+𝑦2
2−𝐿1
2−𝐿2
2
2𝐿1 𝐿2
= sin 𝐴1 sin 𝐴1 + 𝐴2 + cos 𝐴1 cos(𝐴1 + 𝐴2) (10)
cos 𝐴2 =
𝑥2
2+𝑦2
2− 𝐿1
2−𝐿2
2
2𝐿1 𝐿2
(11)
From formula (10) and formula (11) we can get the final formula of invers kinematic in formula (12) and
formula (13):
𝐴2 = 𝑐𝑜𝑠−1 𝑥2
2+𝑦2
2− 𝐿1
2−𝐿2
2
2𝐿1 𝐿2
(12)
𝐴1 = 𝑐𝑜𝑠−1 𝐿1+𝐿2 cos (2 ] 𝑥−[−𝐿2 sin (𝐴2)] 𝑦
𝐿1+𝐿2 cos (2 ] 2+ [−𝐿2 sin (𝐴2)] 2 (13)
A1 and A2 is the formula of Invers Kinematic in Femur and Tibia. The calculation of coxa is given by
the simple calculation with used trigonometry theorem and show in figure 5:
Figure 5. representationmovement ofcoxa
Figure 5is representation of coxa joint and we get the formula of A0 in formula (14)

A0 = Atan2 (X, Y) (14)
2.3 MPU 60506 DoF IMU
MPU 6050 is the transformation of 2 sensors that are accelerometer and gyroscope with I2
C
communications. This sensor can detect the velocity in 3 axis (x, y, and z) and detect the angular velocity also in
3 axis. The configuration of pin in MPU 6050 show in figure 6:
Figure 6.Razor 6 DoF IMU.
2. 4 Sensor Magnetometer HMC5883
HMC 5883 is magnetic sensor which make in size of mount 3.0x3.0x0.9 mm 16-pin leadless chip
carrier (LCC). HMC 5883 build with the resistive magnetic sensor with ADC 12-bit resolution to measure the
magnetic of the earth [14].
Figure 7.HMC 5883
2.5 Complementary Filter
Two inputs from 3 sensors have different characteristic. To minimize the noise, the filter which can
filter the low frequency and high frequency is needed. One of methods is used complementary filter.
Complementary filter will filter out the low frequency if the sensor not so good in dynamic system, and will
filtering the high frequency if the sensor not so good in static system. From low frequency and high frequency
the method of complementary filter will compare both of that to one signal as show in figure 8.
Figure8. Principle of Complementary filter.
Accelerometer
Gyroscope
Tapis
Pelewat-tinggi
Perhitungan
Integral
Tapis
Pelewat-rendah




Magnetometer






fz
fy
fx
mz
my
mx
r
q
p
Figure9. Diagram of complementary filter
𝜙 = 𝐾𝑔𝑖𝑟𝑜𝑠𝑘𝑜𝑝 × 𝜙 + 𝜙̇ 𝐺𝑖𝑟𝑜𝑠𝑘𝑜𝑝
× 𝑑𝑡 + 𝐾𝐴𝑘𝑠𝑙𝑀𝑎𝑔 × 𝜙 𝐴𝑘𝑠𝑙𝑀𝑎𝑔 (15)
2.6 Design of Hardware
The block diagram of all of system in stabilization hexapod is shown in figure 10. Figure 10 show
about all hardware which used in this experiment:

Figure 10. Diagram blok of hardware system
Figure 10 shows how the data from sensor will compute in the first microcontroller (ATMega8535) as
master then will send to second microcontroller (ATMega32) as the value of orientation to make the robot’s
movement. Data from the first microcontroller also can send to computer to show in interface and it can be easy
to analyze. The method of sending data from first microcontroller to second microcontroller or from
microcontroller to computer is with serial communications (Tx/Rx).
2.9 Design of Software
The design of software is to realize the algorithm of stabilization hexapod with 9 DOF IMU. The
design in this system is divided in two systems that are design of software in microcontroller and software in
computer.
Software in microcontroller doing to get data, computer and sending the orientation of degree from
sensors and also convert to position of body robot which the form is PWM servo motor. Software in
microcontroller is used in C language and used integrated development environment (IDE) Codevision AVR.
The design is about invers kinematic’s program, for complementary filter and also design about fuzzy-PID in
CV AVR.
The design of software in computer is made to analyze the data of sensor to be easily. The software is
made with Microsoft Visual Studio 2013 in C# language. Design is about to receive serial data, compute data
and showing the data from microcontroller to software.
Figure 11. GUI of Software.
III. RESULT AND ANALYZE
3.1 Motor Servo Testing
Motor servo is test by change the position of body robot based on every coordinate (x, y, z) and
compare with the first coordinate which has given in initial of robot. The result given by table 4:
Table 4.Value PWM Servo at initial position
Servo No Nilai PWM
1 135
2 169
3 167
4 125
5 169
6 167
7 125
8 169
Mikrokontroller
Atmega 32
Servo 1
Servo
10
Servo
2
Servo
3
Servo
4
Servo
5
Servo
6
Servo
7
Servo
8
Servo
9
Servo
11
Servo
12
Servo
13
Servo
14
Servo
15
Servo
16
Servo
17
Servo
18
Mikrokontroller
Atmega 8535
Sensor
MPU
6050
+HMC
5883
Serial
Rx/Tx

9 167
10 124
11 80
12 82
13 124
14 80
15 82
16 124
17 80
18 82
Tabel 5.Value PWM Servo if invers change base in X coordinate
Servo no
Pengujian Posisi body Robot Terhadap Sumbu X
15 20 25 -15 -20 -25
1 118 113 108 155 162 169
2 165 163 161 167 165 163
3 142 133 124 183 187 191
4 99 91 83 151 160 169
5 169 169 169 169 169 169
6 164 162 160 164 162 160
7 104 97 90 141 146 151
8 167 165 163 165 163 161
9 183 187 191 142 133 124
10 145 152 159 108 103 98
11 82 84 86 84 86 88
12 66 62 58 107 116 125
13 150 158 166 98 89 80
14 80 80 80 80 80 80
15 85 87 89 85 87 89
16 141 146 151 104 97 90
17 84 86 88 82 84 86
18 107 116 125 66 62 58
In table 5 is shown the change of PWM servo based of the body in X coordinate. We can look if PWM
change big from the initial position, the PWM servo no.1 is also bigger. It is because the change about X
coordinate will change the value of coxa and femur in system of robot because the movement is translation head
and backward
3.2 Pitch, Roll dan Yaw Testing
The method of analyze is to get the data every 50ms sampling rate. If we have the data of sensor so we
calculate the Root Mean Square of error every degree (pitch, roll and yaw) then we can conclude that the sensor
is good to make in the system of stabilization or no.
Result can be look at table 6, table 7 and table 8
Table 6 pitch data
NO
Degree
0 10 15 20 -10 -15 -20
1 -0.89 10.63 15.17 20.23 -10.60 -15.75 -20.19
2 -0.55 10.49 15.23 20.25 -10.15 -15.52 -20.15
3 -0.89 10.31 15.35 20.37 -10.09 -15.62 -19.68
4 -0.75 10.42 15.34 20.44 -10.17 -15.47 -18.93
5 -0.79 10.73 14.90 20.43 -10.18 -15.89 -18.95
RMSE (%) 0.43 0.28 0.11 0.19 0.13 0.36 0.23

Tabel 7 roll data
NO
Degree
0 10 15 20 -10 -15 -20
1 -0.63 10.51 14.25 20.54 -10.85 -15.49 -20.17
2 -0.66 10.37 14.78 21.03 -10.86 -15.51 -20.23
3 -0.42 9.99 15.38 20.50 -10.44 -15.85 -20.12
4 -0.90 9.64 15.23 20.56 -10.77 -15.94 -19.69
5 -0.79 9.94 14.25 20.43 -10.85 -15.23 -19.74
RMSE (%) 0.37 0.05 0.12 0.34 0.41 0.33 0.05
Tabel 8 yaw data
NO
Degree
0 10 15 20 -10 -15 -20
1 -0.15 10.76 15.06 20.32 -10.71 -15.41 -20.17
2 0.13 10.71 15.01 20.28 -10.64 -15.47 -20.14
3 0.14 10.61 15.04 20.33 -10.54 -15.52 -20.02
4 0.35 10.54 15.34 20.24 -10.43 -15.45 -20.01
5 0.44 10.77 15.40 19.92 -10.29 -15.34 -19.97
RMSE (%) 0.10 0.37 0.09 0.12 0.29 0.24 0.03
From table 6, table 7 and table 8 we know that the total of RMSE of pitch is 1,73%, roll is 1,67% and
yaw 1,24%.
( a ) ( b )
( c) ( d )
( e ) ( f )
Figure 12.Graphic of Degree (a) pitch 0o
, (b) pitch 10o
, (c) roll 0o
, (d) roll 10o
, (e) yaw0o
, (f) yaw 10o
3.3Testing with disturb pitch to up
The first experiment with change the single tone output of fuzzy and get the result in figure 13:
( a ) ( b )

( c ) ( d )
( e )
Figure 13.Testing with disturb of pitch up and with the variation of single tone output fuzzy.
the result show that the good response is 4th
experiment. The experiment is used with Kp has k1=0,5,
k2=1 and k3=3. Parameter Ki is k1=0,5, k2=0,5 and k3=0,5. The Kd is k1=0,25, k2=0,35 and k3=0,45.
3.4The Heading Testing
The heading of robot is test to make the robot can walk to one line. The result is show by figure 14:
Figure14.Response the heading of robot.
The result show that the robot can walk to one line in one heading. Position of body in track show by
figure 15 :
Figure 15.The final position robot in track.
3.5Testing in all of track
In all of track is made by 2 variation of tilt that are 5o
and 7o
. Robot was walking from starting point
with no degree of tilt and finish in tilt to down. The result is show by figure 16:
( a ) ( b )
Figure16.Graphic of Robot (a) First Tilt 5o
(b)First Tilt 7o
The result is good in 5o
but in 7o
that are so many isolation of respond. It is because the distance in 7o
has offset with the maximum movement of robot. In 7o
the offset of distance in maximum point and minimum
point is 4 cm and the maximum of movement robot to move up is just 2.5.

IV. CONCLUSION
Based on the experiment we conclude that if the invers kinematic is change based on X coordinate the
body of robot move to backward and forward translation. If the change in Y coordinate, the body will move in
beside starting point, and if change in Z coordinate the body will move in up and down translational. The single
tone output of fuzzy which very good in this experiment is 4th
variation that are Kp with value k1=0,5 k2=1 and
k3=3, the parameter of Ki is k1=0,5, k2=0,5 and k3=0,5. The Kd is have k1=0,25, k2=0,35 and k3=0,45. The
disturbance up and down or with change the heading is not so change the responds. The respond sill can follow
the SetPoint of degree which has given before.
REFERENCES
[1] Yoo, Tae Suk, Sung Kyung Hong, Hyok Min Yoon, Sungsu Park, Gain-Scheduled Complementary Filter Design for a MEMS Based
Attitude and Heading Reference System, Open Access, Inc, 2011.
[2] Colton, Shane., The Balance Filter: A Simple Solution for Integrating Accelerometer and Gyroscope Measurements for a Balancing
Platform, http://web.mit.edu/scolton/www/filter.pdf, September 2011.
[3] Kusuma, Johan Wijaya, PenerapanInvers KinematicTerhadapPergerakan Kaki pada Robot Hexapod, STMIK GI MDP, 2013
[4] Bejo, Agus., C&AVR RahasiaKemudahanBahasa C dalamMikrokontroler ATMega8535. GrahaIlmu, Yogyakarta, 2008.
[5] Setiawan, Iwan.,Kontrol PID untuk Proses Industri, Elex Media Komputindo, Jakarta, 2008.
[6] Jang, JyhShing Roger, Chuen Tsai Sun, EijiMizutani. Neuro Fuzzy and Soft Computing, Prentice-Hall International, Inc, 1997
[7] Turner, Peter, Mathematic Requirements for Robot Motion. Tribotix
[8] Susilo, Tri Bagus,Slope measurement using accelerometer, Bachelor Thesis, Diponegoro University, Semarang, 2011.
[9] Kurniawan, David, Control of objects followermobile robot based on fuzzy logic, Bachelor Thesis,Diponegoro University, Semarang,
2007.
[10] Ronnback, Sven, “Development of a INS/GPS navigation loop for an UAV”, Masters Thesis Lulea University of Technology, 2000.

Stabilization of Six-Legged Robot on Tilt Surface With 9 DOF IMU Based on Invers Kinematic

More Related Content

What's hot (19)

Similar to Stabilization of Six-Legged Robot on Tilt Surface With 9 DOF IMU Based on Invers Kinematic (20)

More from IJRES Journal (20)

Recently uploaded (20)

Stabilization of Six-Legged Robot on Tilt Surface With 9 DOF IMU Based on Invers Kinematic