Research on the mobile robots intelligent path planning based on ant colony algorithm application in manufacturing logistics

RESEARCH ON THE MOBILE ROBOTS
INTELLIGENT PATH PLANNING BASED
ON ANT COLONY ALGORITHM
APPLICATION IN MANUFACTURING
LOGISTICS
GUO Yue , SHEN Xuelian and ZHU Zhanfeng
Management Engineering Institute, Ningbo University of Technology, No.201
Fenghua Road, Ningbo. 315211 Zhejiang, China
guoyue@nbut.edu.cn shenxuelian@nbut.edu.cn
zhuzhanfeng@nbut.edu.cn

ABSTRACT
With the development of robotics and artificial intelligence field unceasingly thorough, path
planning as an important field of robot calculation has been widespread concern. This paper
analyzes the current development of robot and path planning algorithm and focuses on the
advantages and disadvantages of the traditional intelligent path planning as well as the path
planning. The problem of mobile robot path planning is studied by using ant colony algorithm, and
it also provides some solving methods.

KEYWORDS
Manufacturing Logistics; Mobile robots; Path planning; Ant colony algorithm

1. INTRODUCTION
The research of mobile robot started from the late 1960s.The Stanford Institute successfully
developed the autonomous mobile robot—Shakey robot in 1966, The robot has independent
reasoning, planning, control and other functions in complex conditions with the application of
artificial intelligence. At the end of the 1970's, the application of computer and sensor technology
researches on mobile robot reach to a new high tide as a result of the development. The mid
1980's, a large number of world famous company started to develop mobile robot platform. The
mobile robot is mainly used as the mobile robot experiment platform in university laboratories
and research institutions, and promoting the multi-directional learning of the mobile robot. Since
the 1990s, the symbol of environment information sensor and information processing technology
development of high level, high adaptability of mobile robot control technology, programming
technology under the real environment has emerged, and the higher level research of mobile
robotics able to be conducted. In recent years, mobile robots are widely used in space exploration,
ocean development, atomic energy, factory automation, construction, mining, agriculture,
military, and service, etc. Research on mobile robot has become a hot research issue and the
concern of the international robot.
David C. Wyld et al. (Eds) : CST, ITCS, JSE, SIP, ARIA, DMS - 2014
pp. 11–25, 2014. © CS & IT-CSCP 2014

DOI : 10.5121/csit.2014.4102

12

Computer Science & Information Technology (CS & IT)

Intelligent mobile robot is a set of integrated system of multiple functions which consists of
environment perception, dynamic decision-making and planning, behavior controlling and
executing. In recent years, mobile robot has wide application prospect in space exploration, ocean
development, atomic energy, factory automation, construction, mining, agriculture, military, and
service, etc. China started the research on the intelligent robots later than some developed
countries, and there still existed a big gap within China and developed countries. In recent years,
the research theory and method for robot have reached the international advanced level by the
China Robotics Lab, and achieved a number of important scientific research achievements in
robotics frontier exploration and demonstration application etc. Because the state and society paid
much attention on the robot field, which including: all-weather 120Kg suspended wing UAV
system, polar research snow mobile robot, Ling Lizard-anti-terrorism and anti-riot robot, robot
nano operating system, etc.
Mobile robot’s path programming technology is one of the core technology in the field of robot
research, which study of the algorithms is advantageous to the improvement of robot planning to
meet the needs of practical applications. The path programming is that, in the obstacle
environment, according to a certain evaluation standard, finding collision free path from the
initial state to the target state. The main issues include finding the optimal or approximate optimal
one from the initial state to the target state collision through the free path and an algorithm to
built reasonable model by using of the mobile robot environmental information. In the model
which being able to cope with uncertain factors and path tracking errors in the environment,
making the influence of external objects to the robot reduced to a minimum: how to use all the
information known to guide the robot motion, resulting in a better decision 1 .

］［

2. LITERATURE REVIEW
The traditional method of path programming is carried out simulation test based on graph. The
general approach is based on the global path planning. At present domestic and foreign common
methods include grid method, topology, visibility graph, Voronoi graph, method, the artificial
potential field method, A* algorithm etc.
Grid method proposed by Howden in 1968 2 , decomposing the robot planning space into a
number of information network unit working space is divided into unit after the use of heuristic
algorithm to search the safe path in the unit. 3 The search process always uses work space with
four quadtree or octree. Consistency and standards grid makes simple adjacency relation in raster
space. After giving each grid traffic factor, path planning problem turns into a problem of
searching optimal path with two grid nodes in the grid network. Topological method is mainly
divided space with topological feature subspace, and then look for topological path is the starting
point to the target point based on the topology of the network, and finally find the path geometry
by the topological path.[4] The basic idea of the method is to find paths in high dimensional space
transformed into the problem of determining the connectivity of the problem of Low Dimensional
Topology space.[5] The visibility graph method is a kind of configuration space method,[6] it
mainly regards the robot as a particle processing, expanding the boundary of the corresponding
outward obstacles in the work environment, and the boundary is formed with vertices of
polygons, determining its vertices, including the robot starting point and the target point. These
points connect, but each vertex cannot connect across each other, forming a visibility graph.

］［

］［

The Voronoi plot method is first discovered by the Russian mathematician Voronoi which can be
applied to static random environment, which is to say in the process of robot running, the
environment is static.[7] All the obstacles are motionless, but environment is uncertain before the
robot starts its path planning, the size and the location parameter of the obstacles in the
environment is changeable. This method using the path which may be far away from obstacles to


13

show walls arc, that results the path will increase from the initial node to the target node.
Artificial potential field method is a method of local path planning, proposed by Khatib etc.[8]
The basic idea of the method is regarding the motion of the robot in the environment as a virtual
artificial force field motion.[9] Obstacles generate repulsive force on the robot, and attraction on
the target. The joint force of attraction and repulsive force controls robot motion direction, which
will determine the position of the robot[10].
In recent years, with the rapid development of in-depth research and the modern computing
technology in mobile robot path programming, the traditional path programming is hard to meet
its requirements and failure in meeting the need of actual environment changes. Therefore
intelligent path algorithms have been studied and used in robot path programming widely. The
artificial intelligent path programming algorithm improves the accuracy of robot obstacle
avoidance path programming greatly and accelerates the programming speed, all these are to be
met the needs of practical application. Intelligent path programming algorithm includes genetic
algorithm[11][12], particle swarm algorithm[13], fuzzy logic[14][15], neural network[16][17],
artificial immune algorithm[18] and hybrid algorithm[19][20][21]. The above algorithms have
been made certain achievements for the robot protecting the obstacle in known or unknown
circumstances.
Italian scholar Dorigo and Colorni proposed a heuristic optimization algorithm in 1991, which is
biologically inspired.[22] It simulated and reference the behavior of ants in the real world to solve
combinatorial optimization problems under distributed environment.[23] It also solves the
problems of large cost when robot in complex environment contains a large number of irregular
obstacles in the path programming[24].
Ant colony algorithm is produced to simulate the process of ants foraging. Ants release specials
in the search of path when confronted with a no through road, they will randomly select one while
releasing hormone information of path length. When the ants again encountered this intersection,
optimal path on the pheromone concentration increase, while the other pheromone concentration
is cutting with the passage of time.[25] At the same time, the ants can adapt to changes in the
environment when obstacles emerge, they will find an optimal path to go. Ant colony algorithm
has the features of group cooperation, positive feedback and distributed computing. Group
cooperation is a cooperation for better optimization task. Although each artificial ant can build a
solution, but the solution with high quality is always produced by ant colony cooperation. The
feedback mode of the algorithm is used in the optimum solution which leaves more pheromone
on a path, and more pheromone in turn attracts more ants. The positive feedback process guides
the system towards the optimal solution of the evolution direction. Distribute Computing of Ant
colony algorithm can calculate each artificial ant at the multiple points in the problem space, at
the same time, it began to separate the structural problems of solutions. The result will not be
affected only because of one artificial ant cannot successfully obtained the solution Distributed
Computing makes the algorithm easy to be implemented.[26] These characteristics make the ant
colony algorithm suitable for solving complex combinatorial optimization problems.
The path programming is a kind of combinatorial optimization problem, so the ant colony
algorithm is suitable to solve the path planning problems.

14


3. QUESTIONS
3.1 Problems in Design
In a 800 × 800 planar scene graph, there is a robot at the origin of O(0, 0), which can only
activities in the planar scene range. The 12 different shapes of the regions are the obstacles that
the robot cannot collide The description of the mathematical are as shown in the following table:
Table 1. Planar scene graph description

Region
No.
1

Obstacles’
name
square

vertex coordinates at
left corner
(300, 400)

2

circular

3

parallelogram

(360, 240)

4

triangles

(280, 100)

5

square

(80, 60)

6

triangles

(60, 300)

7

rectangular

(0, 470)

8

parallelogram

(150, 600)

9
10
11
12

rectangular
square
square
rectangular

(370, 680)
(540, 600)
(640, 520)
(500, 140)

Other description of the
characteristics
length of 200
center coordinates (550, 450) a
radius of 70
Base length140, the coordinates of
the vertices on the left top(400,
330)
top vertex coordinates (345,
210) the lower right vertex
coordinates (410, 100)
length 150
top vertex coordinates (150,
435) the lower right vertex
coordinates(235, 300)
length 220 width 60
Base length 90 the coordinates of
the vertices on the left top(180, 680)
length60 width120
length130
length80
length300 width60

，

，

，

，

，

，

3.2 Maintaining the Integrity of the Specifications

Specify a point outside the obstacles as the target for the robot to reach (target point and the
distance to the obstacle are at least more than 10 units). Set the rule thar walking path of the robot
should by lines and arcs. The arc parts are robot’s turning path. A robot cannot turn by line. The
turning path consists with straight path tangent to a circle, and can also be composed of two or
more circular arc path, but the minimum radius of each circular arc path is 10 units in order not to
collide with obstacles. It also requires the distance between the robot walking route and obstacle
is no more than 10 units, or a collision will occur and the robot can not complete the straight
walking. 4 point O in the scene graph(0, 0), A(300, 300), B(100, 700), C(700, 640).


×

Fig1. 800

15

800 planar scene graph

The maximum speed of the robot walking straight into 5 unit / sec. and maximum turning speed

v = v(ρ ) =

v0
10 − 0 . 1 ρ

2

1+ e
is
where P is radius of turn . If the speed is higher
than that, the robot will rollover, and is unable to walk.
Question: Start from O (0, 0) ,which is the shortest path of O → A, O → B ?

4. METHODOLOGY
In order to find the shortest path to the target from (0,0) with certain rules walking around
obstacles ,we can draw the envelope of robot walking hazardous area, just around the corner with
a radius of 10 units a quarter arc, by the method of the rope then to find the shortest possible path
(for example, seek the shortest path between O and A, can be connected to the section of rope
between O and A, to the arc of the corner support taut, then the length of this wire is O to a
shortest possible path (A),and then list shortest path possible paths to each target point with
Brute-force method.
Designated O (0,0) after the middle of a number of points around obstacles to reach the target
point according to certain rules in the back of O, which allows us to consider not just obstacles
inflection point, should be considered after the target point in the path at the turn of the
problem.Simple line circle structure can not solve this problem,so we have adopted the form of a
minimum turning radius at the inflection point and the target point on the way.We can also be
appropriate to transform the inflection point of the turning radius, so that the robot can along
straight line through the target point of the way, and then create optimization model to optimize
these two programs, and ultimately obtained the shortest path.
The model assumes and symbol description are analyzed by the following assumptions:
(1)Assuming the width of the robot itself is negligible. Thus, the movement of the robot can be
regarded as a point moves.
(2)Assuming the robot walk straight and turn at maximum speed.
(3)Assuming that the obstacle is always subject to 12 different shapes of the area and the nature
of the location, size, etc. has been the same.

16

Table2 Symbol & Description

Symbol

Description

v0

Maximum speed straight line when walking

ρ

Turning radius
The i-th sub-length of the shortest path from

O → A

L 'i


O → C

t min

The shortest time from

Bi B j

Arc length from

Bi B j

Length of the line from

Si
L

i

O → B

O → A

Bi → B j
Bi → B j

5. FINDINGS AND INTERPRETATIONS
5.1 The Shortest Path From

Optimization Model

O → A

Two points within the plane of the shortest path based on the length of the line segment as the
endpoint,but the connection of these two segments with obstacles intersect, so try to attempt to
bypass the obstacle and its hazardous areas other path.Obstacle is a square, the center of this
square is located in the lower part of the connection, so the robot to bypass the obstacle from the
top of the obstacle path is the shortest path.

Figure 2. O → A path

Shown in Figure 2, the shortest path from O → A is constituted by straight line
tangent

arc A1 A2 wherein

the cut point.

Which A1 A2 as

the cutoff

point.Arc A1 A2 thought

1
circle P (80, 210 ) ,the

the
radius is 10. Set cut-point coordinates
sections of the path length can be calculated:

A2 A = s 3 =
OA1 = s1 =

( x2

( x1

2

2

− 3 0 0 ) + ( y 2 − 3 0 0 ) O A1 = s1
2

− 0) +

( y1

− 0)

2

OA1 and A2 A and

a

the center of

A1 ( x1 , y1 ) A2 ( x2 , y 2 )

(1)

(2)

,these three


( x1

A1 A 2 = s 2 = 2 a r c s i n
Here we have x1 , y1 ,

x2 , y2

− x2

)

2

+

( y1

− y2

20

)

17

(3)

2

⋅r

for the decision variables, the total length of the shortest path as the
3

s

=

∑
s

objective function: Min

s

i

1

2

2

2

O A1 + A1 P = O P , A2 A12 + PA12 = A2 P ,points A1 , A2 on the
Constraints: radius r = 10 ,
arc.In summary, the structure optimization model as follows:
3

s=

Min

∑s

i

=

x1 2 + y 1 2 +

( x1 − 3 0 0 )

2

2

+ ( y2 − 300 ) + 20t

(4)

s1

x1 ≤ 8 0


x1 ≥ 7 0


y1 ≤ 2 2 0

2

y 1 − 1 0 0 − ( x1 − 8 0 ) = 2 1 0

s .t . 
x1 2 + y 1 2 + 1 0 0 = 5 0 5 0 0

 x − 300 2 + y − 300 2 + 100 = 56500
) ( 2
)
( 1
2
2

( x1 − x 2 ) + ( y 1 − y 2 )

sin t =

20


(5)

Solving the above model (see Appendix 1 lingo procedures), the results are as follows:
1) The OA length of the shortest path: s min = 471.0372
A1 = ( 70.50596, 213.1406 )

2) Two arc tangent point coordinates:

A2 = ( 76.60640, 219.4066 )

Robot shortest path from point O to point A is reached can be expressed in the table3:
Table3. the shortest path from O

2

），（

1

Start

0 0

(70.50596,213.1406)

3
(70.6064,219.4066)
Total length

End

(70.50596,213.1406)
(70.6064,219.4066)

），（

No

300 300

→ A

Types of segments

Straight line
(80,210) as the center
of the arc
Straight line

Length

224.4994
9.1105
237.4273
471.0372

18


5.2 The Shortest Path from

O

→

B

Model

In this section, we will simplify the roadmap as an empowered network diagram, and use the ant
colony algorithm to find the approximate route of the shortest path.The ant colony algorithm is a
bionic algorithm derived from the nature of ants routing mode simulation . Ants in the process of
movement will leave a substance called Pheromone on its path through the information
transferred. Ants can perceive this substance in the course of the campaign as their movement
direction.Therefore, a large number of ants showed an information feedback phenomenon: more
ants walking on a path, choose the after the greater the probability.
We put on a plane at some point number, and the relationship of the distance between them
simplified network chart.If the node can directly reach in a straight line rather than an obstacle,
the weights of the edges between them weight of the straight-line distance, otherwise there is no
edge between them. As follows:

Figure4. Nodes Numbers Figure

Figure 5. Empowering network chart

Using ant colony algorithm to select the shortest route from O → B from the simplified network
chart.

1) Ant colony algorithm model
Value the point 1 → 15 ,0-1wether if it on the path ,form15 bit sequence 0,1, thereby calculating the
distance of this path. The distance as a mapping of the pheromone variable, due to the
requirements of the most short-circuit, so you can use the countdown or relative distance as the
pheromone concentration. Then you get each ant transition probability.If transition probability is
greater than the global transfer factor, then the global transfer; otherwise transfer must have step.
So that you can step to the global optimal solution close.

2) Perform steps
The first step to initialize N ants. In fact N road, and calculate the current position of ants.
The second step initialization of operating parameters, start the iteration.
The third step in the iterative complement the range of calculated transition probabilities, less
than the global transition probability for small-scale search, or a wide range of search.
The fourth step is to update the pheromone, records state, ready for the next iteration.
The fifth step is to enter the third step
The Sixth step output and programming.


19

3) Analysis of results
The initial state of 50 ants are disorderly distribution, optimized
polarization, so that we get the optimal solution.

the final position to the

Figure6 are average and optimal curve, from which you can know that the algorithm converges
very fast, the effect is better. The shortest path is 1 → 4 → 8 → 9 → 1 1 → 1 4 → 1 5 . Chromosome:
100100011010011, running time is 0.3910.
Therefore, the shortest path from O → Bas shown below:

Figure 6. Pheromone concentration average value and the optimum value

5.3 The Segmented Path Length from

O→B

Optimization Model for Solving

In the previous section, we have determined to run the route of the shortest node
1 → 4 → 8 → 9 → 1 1 → 1 4 → 1 5 in Figure 3, but this simplified diagram from only consider a
straight line, without regard to the actual deployment of the arc length. Therefore, we put this
route segment, making each piece only route to bypass an obstacle.
Based on the above analysis, from this route O → B is divided into five sections L ( O → B )
L ( B → B ) L ( B → B ) L 4 ( B9 → B12 ) L 5 ( B1 2 → B )
,
,
calculate their length, then the sum thus obtained
the shortest path from O → B .
1

2

3

6

3

6

9

3

20


Figure7. Seeking from

O → B3 shortest path

1) Seeking from O → B3 shortest path
seek the shortest path O → A structure optimization model is similar to when seeking the shortest
path O → B 3 , we will coordinate O B3 Q1 as the route start point ( a, b) , end point (c, d ) and the

、、

arc center (m, n) coordinates variable values.
Namely: a = 0, b = 0; c = 100, d = 378; m = 60, n = 300.
The coordinates of the cut-off point 1 ( 1 1 ) , B2 ( x2 , y2 ) as a decision variable, O → B3 as the
objective function of the length of the shortest structure optimization model as follows:
B x,y

Min

s = ( x1 − m)2 + ( y1 − n)2 +

2

( x1 − c) + ( y2 − d )

2

x1 ≤ m + 1 0


x1 ≥ m


y1 ≤ n + 1 0

2

y1 − 1 0 0 − ( x1 − m ) = n

s .t . 
( x1 − a ) 2 + ( y1 − b ) 2 + 1 0 0 = m 2 + n 2

 x − c 2 + y − d 2 + 100 = m − c 2 + n − d
) ( 2
)
(
) (
( 1
2
2

( x1 − x 2 ) + ( y1 − y 2 )

s in t =

20


Solving the above model, the results are as follows:

s = 397.0986
The length OB3 of the shortest path: min

(6)

+ 20t

(7)

)

2


21

Two arc tangent point coordinates:
B1 = ( 50.1353,301.6396 )
B2 = ( 51.6795,305.547 )

）
）
）
）

2

Seeking

B3 → B6 shortest path model

The optimization model with the same route as the previous paragraph, the coordinates of
B3 B6 Q2 as the beginning (a, b) of the route, the end (c, d ) and the arc center ( m, n) coordinates of the
variable value. Namely:

、、

a = 100, b = 378; c = 185, d = 452.5; m = 150, n = 435.

Using the lingo program solving the optimization model, the length of the shortest path.
Similarly, we can calculate the shortest path of the other sub-routes. In summary, we have come
to the shortest path.
Table4. The result of the length of the shortest path
Segmented

Start

End

Types of segments

Length

1

(0,0)

(50.1353,301.6396)

Straight line

305.7777

2

(50.1353,301.6396)

(51.6795,305.547)

3

(51.6795,305.547)

(141.6795,440.547)

），（

5.88

4

(141.6795,440.547)

(147.9621,444.79.0)

），（

7.7756

5

(147.9621,444.79.0)

(222.0379,460.2099)

6

(222.0379,460.2099)

(230,470)

7

(230,470)

(230,530)

8

(230,530)

(225.5026,538.3538)

9

(225.5026,538.3538)

(144.5033,591.6462)

10

(144.5033,591.6462)

(140.6892,596.3523)

11

(140.6892,596.3523)

(100,700)

162.2498

75.6637

），（

13.6557

Straight line

60

），（

9.8883

220 530 as
the center of the
arc
Straight line
150 600 as
the center of the
arc
Straight line

），（

Total length

60 300 as the
center of the arc
Straight line
150 435 as
the center of the
arc
Straight line
220 470 as
the center of the
arc

96.9536
6.1545
110.377
854.3759

6. CONCLUSIONS
With the continual development of robotic research in the field of artificial intelligence, the use of
ant colony algorithm effectively solves the problem of robot path planning in the practical work
of calculation. Our studies show that in a certain range, the optimized model for ant colony
algorithm can be used to calculate and design the shortest path when a robot moves from a

22


starting point beyond some obstacles and reaches the specified target points opposite the obstacles
without any collision. Nevertheless, further study is necessary in that some limitations still exist
in mobile robot path planning via ant colony algorithm, e.g. the model for the shortest path
planning remains to be optimized, and whether there are other algorithm solutions to mobile robot
path planning etc.

ACKNOWLEDGMENT
We would like to thank the referees very much for their valuable comments and suggestions,
there also got some help from Miss Lu Yan and Ms. Xu Si.

REFERENCES

[4]
[5]

[6]
[7]
[8]
[9]
[10]

[11]
[12]

[13]
[14]
[15]
[16]

[17]
[18]

，

[3]

，

[2]

TAN Min, WANG Shuo.(2013) Research Progress on Robotics. ACTA AUTOMATICA SINICA,
39(7):pp.963-972.
Yahja A., Singh S., Stentz A.(2000) An Efficient Online Path Planner For Outdoor Mobile Robots.
Robotics And Autonomous Systems, 32(2):pp.129-143.
LU Qing.(2007) Research of Path Planning for Car-Like Robot Based on Grid Method. Computer and
Information Technology, 15(6):pp.24-27.
Oommen B., Iyengar S., Rao N.,Kashyap R.(1987) Robot Navigation In Unknown Terrains Using
Learned Visibility Graphs. IEEE Journal of Robotics and Automation, 3(6):pp.672-681.
LI Shan-shou, FANG Qian-sheng, XIAO Ben-xian,QI Dong-liu.(2008) Environment Modeling in
Global Path Planning Based on Modified Visibility Graph. Journal of East China Jiaotong University,
25(6):pp.73-77.
CAI Xiao-hui.(2007)The Path Planning of Mobile Robots Based on Intelligent Algorithms. Zhejiang
University Master Degree Thesis, 5:pp.6-12.
Takahashi O. and R. Scilling.(1989) Motion Planning in a Plane using Generalised Voronoi
Diagrams. IEEE Transactions on Robotics and Automation, 5,(2):pp.169-174.
JIN Lei-ze DU Zhen-jun JIA Kai.(2007) Simulation study on mobile robot path planning based on
potential field. Computer Engineering and Applications, 43(24):pp.226- 229.
Khatib O.(1985) Real Time Obstacle Avoidance For Manipulators And Mobile Robots[J].The
International Journal of Robotics Research, 5(2):pp.500-505.
Koren Y, And Borenstein J.(1991) Potential field methods and their inherent limitations for mobile
robot navigation[C]. Proceedings of the IEEE Conference on Robotics and Automation, Sacramento,
California , April 7-12: pp.1398- 1404.
AI-Taharwa I.,Sheta A.,AI-Weshan M.(2008) A Mobile Robot Path Planning Using Genetic
Algorithm In Static Environment. Journal of Computer Sciences, 4(4):pp.341-344.
YANG Lin-quan, LUO Zhong-wen, TANG Zhong-hua, LV Wei-xian. (2008) Path Planning
Algorithm For Mobile Robot Obstacle Avoidance Adopting Bezier Curve Base on Genetic
Algorithm. 2008 Chinese Control and Decision Conference:pp.3286-3289.
Gondy Leroy, Ann M. Lally and Hsin Chun Chen.(2003) The use of dynamic contexts to improve
casual Internet searching. ACM Transactions on Information System, 21(3):pp.229-253.
St.Preitl, R.E.Precup, J.Fodor, B.Bede.(2006) Feedback Tuning In Fuzzy Control System.Theory and
Applications, 3(3):pp.81-96.
P.Vadakkepat, O.C.Miin, X.Peng, T.H.Lee.(2004) Fuzzy Behavior-based Control of Mobile
Robots[J]. IEEE Transactions on Fuzzy Systems, 12(4):pp.559-564.
CHEN Huahua, DU Xin, GU Weikang.(2004) Neural Network and Avoidance Genetic Algorithm
Based Dynamic Obstacle and Path Planning for A Robot. Journal of Transcluction Technology,
(4):pp.551-555.
NI Bin, CHEN Xiong, LU Gongyu.(2006) A Neural Networks Algorithm for Robot Path Planning in
Unknown Environment. Computer Engineering and Applications, (11):pp.73-76+109.
XU Xin-ying, XIE Jun, XIE Ke-ming.(2008) Path Planning of Mobile Robot Based on Artificial
Immune Potential Field Algorithm.Journal of Beijing University of Technology, 34(10):pp.11161120.

，

[1]


23

[19] CHEN Xi, TAN Guan-zheng, JIANG Bin.(2008) Real-time optimal path planning for mobile robots
based on immune genetic algorithm. Journal of Central South University Science and Technology),
39(3):pp.577-583
[20] DING Wei.(2007) Path Planning Based On Immune Evolution And Chaotic Mutation For Mobile
Robot. Master Degree in Engineering Dissertation, Harbin University of Science and Technology,
March:pp.19-24.
[21] CUI Shi-gang, GONG Jin-feng, PENG Shang-xian, WANG Jun-song.(2004) Hybrid intelligent
algorithm based 3-D of robot path planning. Manufacturing Automation, (2):pp.49-51.
[22] GUO Yu, LI Shi-yong.(2009) Path Planning for Robot Based on Improved Ant Colony Algorithm.
Computer Measurement and Control, 17(1):pp.187-190.
[23] ZHU Qing-bao.(2005) Ant Colony Optimization Parallel Algorithm And Based On Coarse-grain
Model.Computer Engineering, 31(1):pp.157-159.
[24] FAN Lu-qiao YAO Xi-fan BIAN Qing-qing JIANG Liang-zhong.(2008) Ant Colony Algorithm and
the Application on Path Planning For Mobile Robot. Robotics Technology, 23:pp.257-259+261.
[25] XIE Min,GAO Li-xin.(2008) Ant algorithm applied in optimal path planning.Computer
Engineering and Applications. Computer Engineering and Applications, 44(8):pp.245-248.
[26] ZHU Qing-bao.(2005) Ants Predictive Algorithm For Path Planning Of Robot In A Complex
Dynamic Environment. Chinese Journal of Computers, 28(11):pp.1898-1906.

（

：
：
：
：

Appendix

Matlab programming of Ant Colony Algorithm

function shortroad_ant_main
% Ant main program
clear all;close all;clc;%clear all
tic;%time start
Ant=50;Ger=100;%
Running parameter
initialization
power=[0
70
1000

70

1000

276

208

1000

1000

1000

1000

208
1000
1000
0 141
1000
1000
1000
141
100
1000
1000
211
1000
1000
1000
120
120
1000
1000
182
0
1000
1000
1000
60
1000
1000
1000
1000
1000
555
1000
1000

1000

1000

1000
118
1000
1000
0
1000
1000
1000
76
69
1000
1000
170
1000
1000
1000
55
1000
1000
1000
1000
0
1000
1000
1000
82

30
1000
1000
1000
76
1000
1000
1000
0
1000
1000
89
123
1000
1000
1000
128
141
1000
1000
69
82
1000
1000
1000
0];

1000

1000

1000
1000
170

1000
1000
55

1000
1000
123

1000
30
128

1000
1000
0

313
1000
1000

1000
1000
1000

1000
1000
0

1000
555
1000

1000
118
1000

1000
1000
1000

1000
1000
141

276
1000
1000

1000

182
1000
1000

1000

68
1000
1000

168
500
1000

1000

0
131
1000

1000
1000
1000

1000

1000
1000
313

0
1000
1000

1000
1000
1000

120
1000
1000

145
131
89

1000
141
1000

131
0
1000

1000
49
1000

1000
49

1000
0

[PM PN]=size(power);
% Initialization Ant place
v=init_population(Ant,PN);
v(:,1)=1;v(:,PN)=1;% The beginning
and end points in the path
% The distance when the information
factors concentration
fit=short_road_fun(v,power);
% Distance as small as
possible, so and information
factors concentration
corresponding
T0 = max(fit)-fit;
% Draw the picture

1000
1000
1000
211
1000
1000

1000
1000
1000
120
1000
1000

0
132
1000
1000
68
145
1000
1000
168
1000
1000
1000
100
60
1000
1000
132
0
1000
1000
500
131
1000
1000
1000
141


figure(1);grid on;hold on;
plot(fit,'k*');
title('(a) The initial position of ants ');
xlabel('x');ylabel('f(x)');
% Initialization
vmfit=[];vx=[];
P0=0.2; % P0---Global transfer factor
P=0.8; % P ---Pheromone evaporation
coefficient
%C=[];
% Start search the shortest path
for i_ger=1:Ger
lamda=1/i_ger; % Transfer of step
size parameters
[T_Best(i_ger),BestIndex]=ma
x(T0);% The most information
pheromone concentration
for j_g=1:Ant % To obtain the
global transition probability
r=T0(BestIndex)-T0(j_g);% The
best distance with ant
Prob(i_ger,j_g)=r/T0(BestIndex);%
How much speed rate should
be closer to it
end
for j_g_tr=1:Ant
if Prob(i_ger,j_g_tr)<P0
% Local transfer ---- Small step
transfer
M=rand(1,PN)<lamda;
temp=v(j_g_tr,:)2.*(v(j_g_tr,:).*M)+M;
else
%Globle transfer---Big step transfer
M=rand(1,PN)<P0;
temp=v(j_g_tr,:)2.*(v(j_g_tr,:).*M)+M;
end
% Reput the beginning and end
points,which is not in
the moving process of change
temp(:,1)=1;temp(:,end)=1;
if
short_road_fun(temp,power)<s
hort_road_fun(v(j_g_tr,:),powe
r)
%Recorder
v(j_g_tr,:)=temp;

end
end
% Update the information
factors To prepare for the
next iteration
fit=short_road_fun(v,power);
T0 = (1-P)*T0+(max(fit)-fit);%
information factors evaporation
[sol,indb]=min(fit);
v(1,:)=v(indb,:);%
Record the iteration of the state
media=mean(fit);
vx=[vx sol];
vmfit=[vmfit media];
end
%%%%%%%%%%%%%%
%%%%%%%%%%%
%%%%%%%%
%%%% The last result
disp(sprintf('n')); % Blank one line
% Display the optimal solution and
the optimal value
disp(sprintf('Shortroad is
%s',num2str(find(v(indb,:)))));
%num2strData into character.
disp(sprintf('Mininum is %d',sol));
v(indb,:)
% Graphic display optimal results
figure(2);grid on;hold on;
plot(fit,'r*');
title('The last place of Ant');
xlabel('x');
ylabel('f(x)');
% Graphic display optimal and
average function value trend
figure(3);
plot(vx);
title(' optimal, The average function
value trend ');
xlabel('Generations');ylabel('f(x)');
hold on;plot(vmfit,'r');hold off;
runtime=toc%Time end
end
%%
function fit=short_road_fun(v,power)
[vm vn]=size(v);
fit=zeros(vm,1);% Record distance
of each path
for i=1:vm
I=find(v(i,:)==1);% Looking for
the point on the path
[Im,In]=size(I);

，

24


for j=1:In-1
fit(i)=fit(i)+power(I(j),I(j+1));%
Find distance of the path
end
end
end

25

%%
%Function init_population
function v=init_population(n1,s1)
v=round(rand(n1,s1));% Initializes all
the Ants
END

AUTHORS
GUO Yue
PhD in Management Science and Engineering, MAIB, BBA Profersor, Dean of Department
of Enterprising Management ,Ningbo University of Technology In 2010 Dr. Guo award
Chinese Provincial New Century Hundred-Thousand-Ten thousand Engineering Talent, in
2011 he also got Chinese Provincial Backbone Middle-Young Aged Teachers. Dr. Guo has
been published over 50 papers and 8 books in his academic field. In management science
and engineering field he also do some consulting services to solve the actual problems for
local companies and enterprises.
SHEN Xuelian
PhD graduated from Southwest University of Finance and Economics, as a lecturer teaching
in economics field courses at Ningbo University of Technology. He got his master degree
from Swiss University.
ZHU Zhanfeng
Postdoctoral fellow, PhD in Management Science and Engineering, MBA, BA Profersor,
Headof Department of Economics & Management, Ningbo University of Technology He
was visiting scholar of University of magdeburg in Germany; and got advanced logistics
management division, registered senior consultant and Doctoral tutor. Also presently for
ningbo college of engineering management, and director of the institute of the ningbo
college of engineering development of small and medium-sized enterprises. At the same
time, he was also the ministry of education of institutions of higher learning MBA class teaching steering
committee of logistics and electronic commerce, deputy director of the national demonstration on
repository construction project logistics management specialty teaching work committee.

Research on the mobile robots intelligent path planning based on ant colony algorithm application in manufacturing logistics

More Related Content

Viewers also liked (13)

Similar to Research on the mobile robots intelligent path planning based on ant colony algorithm application in manufacturing logistics (20)

Recently uploaded (20)

Research on the mobile robots intelligent path planning based on ant colony algorithm application in manufacturing logistics