International Journal of Smart Home

Vol. 7, No. 3, May, 2013

321

Design of Obstacle Avoidance System for Mobile Robot using Fuzzy
Logic Systems


Xi Li and Byung-Jae Choi
School of Electronic Engineering, Daegu University
Jillyang Gyeongsan-city Gyeonbuk, Korea
[email protected], [email protected]
Abstract
There were many reports about an obstacle avoidance of a mobile robot. In this paper, we
design a fuzzy logic system and propose an obstacle avoidance algorithm for a path planning
in unknown environment for a mobile robot. The ultrasonic sensors are employed for
detecting the distance to obstacles and their positions. An angular velocity control for left and
right wheels is implemented by a fuzzy logic system. We here propose another new rule table
that is induced from the consideration of the distance to obstacles and the angle between the
robot and the goal. Some simulation results show that the proposed method generates a good
path with avoiding obstacles and has faster travelling time.

Keywords: fuzzy logic system, mobile robot, obstacle avoidance, path planning, ultrasonic
sensor

1. Introduction
There were many research results about obstacle avoidance and path tracking of a mobile
robot. The fuzzy logic system has widely used for one of effective means in unknown and
complex industrial environments. In many research results, a fuzzy logic system has usually
implemented for improving the efficiency of obstacle avoidance and path planning of mobile
robot at unknown environments. The fuzzy logic system has well suited for implementing
controllers due to its capabilities of inference and approximate reasoning under uncertainty.
Most fuzzy logic systems have a complex rule table for achieving different control objectives.
And the size of complete rule-bases increases exponentially with the number of input
variables. Many algorithms were addressed in related journals for obstacle avoidance and
path planning of a mobile robot [1-9]. In [1], a hierarchical behavior-based control
architecture was introduced. This structure was motivated by the hierarchical nature of
behavior as hypothesized in ethological models. He proposed a new approach to design a
fuzzy controller for increasing the ability of mobile robot to react to dynamic environment. A
simple structured fuzzy logic system was implemented in [2]. Its main concept was
introduced in [10].
In this paper we propose a fuzzy logic based control system for path tracking of an indoor
mobile robot. The ultrasonic sensor is used for positioning and identifying the recognition of
an obstacle. Here the left and right wheels angular velocities are controlled by a fuzzy logic
system. The methods for fuzzification and reasoning are singleton and Mamdanis method,
respectively. In the case of the conventional fuzzy logic system, the number of control rules is
International Journal of Smart Home
Vol. 7, No. 3, May, 2013


322

forty nine for each wheel [2]. So, we analyze the fuzzy control rules of the conventional
system and then induce another control rules from eight conditions of positions of obstacles
and three parts of the angle between the robot and the target position. The organization of this
paper is as follows: In next section we introduce the controlled process of an indoor mobile
robot. We describe the design process of fuzzy logic based obstacle avoidance algorithm for
an indoor mobile robot in Section 3. Simulation examples are shown in Section 4. We here
use the Matlab/Simulink as a simulation tool. We show that good performance could be
obtained at several simulation results. Finally, we present some concluding remarks.

2. Architecture of Mobile Robot System
A kinematics model of a mobile robot used in this paper is shown in Figure 1.


Figure 1. Kinematics Model of the Robot

Accelerations of left and right wheels are
L
e
and
R
e
, respectively. We assume that the
contact between the wheels and the ground is pure rolling and non-slipping [5]. The
relationship between left and right wheels of velocity and acceleration is as follows:
L L
r V e =
and
R R
r V e =
,
(1)
where r is a radius of the wheel. The linear velocity of the mobile robot is V. Linear velocities
for the left and right wheels are and
R
V
, respectively. The relationship between
L
e
,
R
e
,
L
V

and
R
V
is as follows:
(2)

L
V
L
r
L
V V
L R L R
e e
e

=

=
2 2
L R L R
r
V V
V
e e +
=
+
=
International Journal of Smart Home
Vol. 7, No. 3, May, 2013


323

Now we can summarize a dynamic model for the mobile robot as follows:

u cos V x = '
,
u sin V y = '
,
e u = '
,
(

(
(
(

=
(
(
(

'
'
'
e
u
u
u
V
sin
cos
y
x
1
0
0
0

(3)

3. Design of Fuzzy Logic Systems
The conventional fuzzy logic system mainly includes fuzzification, knowledge base, fuzzy
reasoning and defuzzification. The fuzzification converts the accurate input variables into
input grades named as fuzzy variables. The knowledge base is used to store relevant data and
control rules. The fuzzy reasoning generates fuzzy results from inferencing of the knowledge
base and the inference engine. The defuzzification converts fuzzy variables to accurate output
variables. Its typical architecture is shown in Figure 2.


Figure 2. Typical structure of a fuzzy logic system

3.1. Determination of Input and Output Variables
The inputs for the proposed fuzzy logic system are distances measured from the obstacle to
the sensors and the angle between the robot and the goal. The sensors are located at left, right,
and front sides of the robot. The angle means an angle between robots orientation and target
position. The output variables are velocities of the left and right wheels.

3.2. Fuzzification
Linguistic variables near and far are taken for the distance from the obstacle to the
sensor. Eight conditions are defined by the location of the obstacles like Table 1. The domain
for the angle between the robot and the target position is constructed with {left, front, right}.
Here left means that the goal is located at the left side of the robot. The domain for
velocities of the left and right wheels is constructed with {slow, L-slow, mid, L-fast, fast},
where L and mid stand for Little and middle, respectively. The membership functions are
shown in Figure 3.


International Journal of Smart Home
Vol. 7, No. 3, May, 2013


324

Table 1. Eight conditions according to the detection of obstacles
left-
obstacle
front-
obstacle
right-
obstacle
D1 near near near
D2 near near far
D3 near far near
D4 near far far
D5 far near near
D6 far near far
D7 far far near
D8 far far far



Figure 3. Membership functions for fuzzy logic system

Input and out membership functions were defined as Table 2 and 3, respectively. Values in
the tables present values of x-axis in a kind of singular points. That is, 0 : 80 : 400 : 480 is
values of x-axis in singular points for a fuzzy membership function near of input fuzzy
variables and 0 : 6 : 8 : 10 : 12 is values of x-axis in singular points for a fuzzy
membership function L-Fast of output fuzzy variables.

Table 2. Definition of input membership functions
Terms Meaning Range of Membership
Function
Near Near 0 : 80 : 400 : 480
Far Far 0 : 80 : 400 : 480
Left Left -90 : -30 : 0 : 30 : 90
Front Front -90 : -30 : 0 : 30 : 90
Right Right -90 : -30 : 0 : 30 : 90


International Journal of Smart Home
Vol. 7, No. 3, May, 2013


325

Table 3. Definition of output membership functions
Terms Meaning Range of Membership
Function
Slow Slow 0 : 2 : 4 : 12
L-Slow Little Slow 0 : 2 : 4 : 6 : 12
Mid Middle 0 : 4 : 6 : 8 : 12
L-Fast Little Fast 0 : 6 : 8 : 10 : 12
Fast Fast 0 : 8 : 10 : 12

3.3. Generation of Control Rules
The control rules could be induced by empirical knowledge. They are shown in Table 4
and Table 5 for left and right wheel, respectively. The rules were basically generated by eight
conditions from D1 to D8. For example, the condition D1 means that obstacles are located at
near to the front, left, and right sides. The angle Right means that the goal is located at the
right side of the robot.
Rule base is composed of many fuzzy implication relations, which are obtained based on
lots of experiments, observation and operation experience. Furthermore the actual number of
fuzzy rules should be taken depends on many factors. The general principle is on the
completeness of the premise. In order to simplify the design process of the fuzzy logic system,
a smaller number of rules are better. In [2], the conventional fuzzy logic system had 98 rules,
that is, 49 rules were required for each wheel of left and right.

Table 4. Fuzzy control rules for the left wheel
Angle D1 D2 D3 D4 D5 D6 D7 D8
Left Slow L-Slow Mid L-Slow Slow Slow Mid L-Fast
Front L-Slow L-Slow L-Fast Mid Slow L-Slow L-Slow Fast
Right L-Slow L-Slow Mid L-Fast Slow L-Slow Slow Fast

Table 5. Fuzzy control rules for the right wheel
Angle D1 D2 D3 D4 D5 D6 D7 D8
Left L-Slow Slow Mid Slow L-Slow L-Slow L-Fast Fast
Front Slow Slow L-Fast L-Slow L-Slow Slow Mid Fast
Right Slow Slow Mid Mid L-Slow Slow L-Slow L-Fast

The meaning of several rules of Table 4 and 5 is as follows:
R01:
IF the condition is D1 and the angle is Left,
THEN the left wheel is Slow(
L
V
=Slow) and the right wheel is Little Slow(
R
V
=L-Slow).

R08:
IF the condition is D3 and the angle is Front,
THEN the left wheel is Little Fast(
L
V
=L-Fast) and the right wheel is Little Fast (
R
V
=L-Fast).
International Journal of Smart Home
Vol. 7, No. 3, May, 2013


326

R16:
IF the condition is D6 and the angle is Left,
THEN the left wheel is Slow(
L
V
=Slow) and the right wheel is Little Slow(
R
V
=L-Slow).

R24:
IF the condition is D8 and the angle is Right,
THEN the left wheel is Fast(
L
V
=Fast) and the right wheel is Little Fast (
R
V
=L-Fast).

3.4. Defuzzification
Many defuzzification algorithms have been reported. There are several common methods,
such as maximum membership grade, median clustering, average maximum membership
grade and weighted average method, and etc. We here use the center of gravity, where
( ) z
c

is degree of membership, z is steering angle, and
a
z
is a crisp value.

( )
( )dz z
zdz z
Z
c
c
a
}
}

=

(4)


Figure 4. Center of gravity method for defuzzification

4. Simulation Study
The simulation environment is like Figure 4. It has three obstacles and is the same
environment in [2]. The starting position of the robot and its goal are (0, 0) and (25, 25),
respectively. Then the robot moved along the bold path and line path by the proposed fuzzy
logic system and Ref. [2], respectively. However the conventional fuzzy logic system of [2]
used 49 control rules and the proposed system used only 24 rules.

International Journal of Smart Home
Vol. 7, No. 3, May, 2013


327


Figure 5. Simulation results of obstacle avoidance

The mobile robot could safely avoid some obstacles. The comparison with the
conventional fuzzy logic control system of [2] and the proposed system was shown in Table 6.
From the table, the performance of the conventional fuzzy logic system is better than that of
the proposed system in the case of the travelled length. However the performance of the
proposed fuzzy logic system is better than that of the conventional system in the case of the
travelled time. The proposed system has smaller control rules, then the computational time is
also more fast.

Table 6. Numerical comparison of two methods
Method Traveled Path
Length (m)
Traveled Path
Time (sec)
Proposed
Method
45.6 50.2
Ref.[2] 43.8 55.6

5. Concluding Remarks
We here described the design process of fuzzy logic based obstacle avoidance algorithm
for a mobile robot, and studied an obstacle avoidance system of mobile robot by using fuzzy
logic control systems.
The velocities of the two wheels were independently controlled. Their outputs for the fuzzy
control system were the velocities of the left and right wheels. And their input variables were
positions of obstacles and the angle between the robot and the target position. Here positions
of obstacles were categorized by eight conditions. The angle of input variable was divided by
three part of front, left, and right. Therefore, the number of total control rules was only 24
compared to the case of 49 in [2] for each wheel of the robot. The proposed fuzzy logic
International Journal of Smart Home
Vol. 7, No. 3, May, 2013


328

system showed a good performance with small control rules and fast travelling time. The
simulation results demonstrated the effectiveness of the obstacle avoidance capability with
fast time in an unknown environment.

References

[1] T. Jin, Obstacle Avoidance of Mobile Robot Based on Behavior Hierarchy by Fuzzy Logic, vol. 12, no. 3,
(2012), pp. 245249
[2] S. Jin and B. -J. Choi, Fuzzy Logic System Based Obstacle Avoidance for a Mobile Robot, FGIT-
CA/CES3, (2011).
[3] L. Xu, Y. Chen and H. Ju, Autonomous Obstacle Avoidance for Mobile Robot Based on Dynamic Behavior
Control, Computer Engineering, vol. 33, no. 14, (2007), pp.180-182.
[4] B. Huang and G. Cao, The Path Planning Research for Mobile Robot Based on the Artificial Potential Field,
Computer Engineering and Applications, vol. 27, (2006), pp. 26-28.
[5] C. G. Rusu and I. T. Birou, Obstacle Avoidance Fuzzy System for Mobile Robot with IR Sensors, 10th
International Conference, (2010) May 27-29.
[6] S. K. Pradhan, D. R. Parhi and A. K. Panda, Fuzzy logic techniques for navigation of several mobile robots,
Applied Soft Computing, vol. 9, (2009), pp. 290304.
[7] H. -J. Yeo and M. -H. Sung, Fuzzy Control for the Obstacle Avoidance of Remote Control Mobile Robot,
IEEK 2011, vol. 1, (2011).
[8] K. Jung, J. Kim and T. Jeon, Collision Avoidance of Multiple Path-planning using Fuzzy Inference System,
Proceedings of KIIS Spring Conference, vol. 19, no. 1, (2009).
[9] Q. Zhu, Ant Algorithm for Navigation of Multi-Robot Movement in Unknown Environment, Journal of
Software, vol. 17, no. 9, (2006), pp. 1890-1897.
[10] B. -J. Choi, Absolute Stability of the Simple Fuzzy Logic Controller, Int. J. of Fuzzy Logic and Intelligent
Systems, vol. 11, no. 7, (2001), pp. 574578.
[11] C. C. Lee, Fuzzy logic in control systems: Fuzzy logic controllerPart I & II, IEEE Trans. Syst., Man,
Cybern., vol. 20, no. 2, (1990), pp. 404435.
[12] S. Nazlibilek, Y. Ege and O. Kalender, A Multi-sensor network for direction finding of moving
ferromagnetic objects inside water by magnetic anomaly, Measurement, vol. 6, (2009), pp. 1-17.

Authors

Xi Li
She received her B.S. degree in electronic engineering at Daegu
University. She is currently Master of Science Course in the School of
Electronic Engineering from Daegu University. Her research interests
include intelligent control and systems.


Byung-Jae Choi
He received his B.S. degree in electronic engineering at Kyungbuk
National University, Korea. And he received his M.S. and Ph.D, degrees
received in KAIST. He is currently Professor in the School of Electronic
Engineering from Daegu University. His research interests include
intelligent control theory and its applications.