PAPER • OPEN ACCESS
Research on Route Planning based on improved Ant Colony Algorithm
To cite this article: Chen Xin et al 2021 J. Phys.: Conf. Ser. 1820 012180
View the article online for updates and enhancements.
Research on Route Planning based on improved Ant Colony
Algorithm
Chen Xin1,a, Luo Qiang1,b*, Wang Changhong1,c, Zheng Yan2,d, Wang Haibao12,e
1Intelligent Mountain Agricultural Machinery Technology Research Center of
Chongqing three Gorges university, Wan Zhou, Chong Qing, China;
2Hefei Institute of Intelligent Machinery, Chinese Academy of Sciences, He Fei, An
Hui, China
aemail:[email protected],cemail:[email protected],demail:1312890632@qq.
com,eemail:[email protected]
*corresponding author: bemail: [email protected]
Abstract: In order to enhance the safe flight performance of UAV and better complete complex tasks, it is extremely necessary to study route planning. Firstly, the feature points of the landform landform are extracted as the flight track points of the UAV, and the track planning problem is transformed into the traveling salesman problem. Secondly, an adaptive pheromone update method is proposed, and an improved ant colony algorithm for optimizing the local pheromone is proposed to avoid the basic ant colony algorithm from falling into the local optimal solution and improve the global search ability. Finally, the proposed algorithm is verified by simulation experiments. The experimental results show that the improved ant colony algorithm is practical and superior, and can effectively solve the problem of route planning and obtain the optimal path of route planning.
1. Introduction
With the rapid development of communication technology and computer technology, UAV has developed rapidly. Now UAV is not only used in military industry, but also uses plant protection UAV to spray and sow seeds in intelligent agriculture. In smart cities, the use of drones for some high-altitude dangerous operations and wonderful UAV[1]performances and other tasks require
complex work, various industries are full of drones. With the increasing variety of UAV, the research on UAV path planning is becoming more and more important to carry out more complex tasks. The trajectory planning of UAV should comprehensively consider the performance of UAV software and hardware. The complexity of mission execution, environmental threats and other comprehensive factors[2]. At present, route planning algorithms can be divided into two categories according to
planning decisions[3], the first is traditional classical algorithms, including artificial potential field
method, simulated annealing algorithm, etc.[4]; the second is heuristic intelligent algorithms, including
genetic algorithm, particle swarm optimization algorithm, ant colony algorithm[5]and so on. Compared
with the traditional classical algorithm, the heuristic intelligent algorithm performs better in the face of complex environment and more track points. Ant colony algorithm plays a great role in solving the problem of track planning. The algorithm has distributed computing, positive feedback adjustment mechanism, good parallelism and robustness[6]. However, due to the distribution of pheromones and
2
the influence of volatilization mechanism, the quality of the solution space is not high, and the time to obtain the track path is too long, resulting in the ant colony algorithm searching for the solution space for too long, easy to fall into the local optimal solution.
Based on it, an adaptive pheromone updating mechanism is proposed in this paper, and the ant colony algorithm optimized by local pheromone avoids the problem of local convergence and prematurity in route selection. The quality and convergence speed of the search solution space of the ant colony algorithm are improved, and the optimal route planning route is obtained.
2. Research basis
Ideally, solving the optimal path planning path in three-dimensional space can be transformed into a traveling salesman problem. Under the condition of environmental priori, the track point is equivalent to the city point of the traveling salesman problem, that is the vertex of the weighted graph. The connection between the two track points is the edge of the weighted graph, and the length of the track point is the weight of the edge. The route planning problem is abstracted into a three-dimensional space of 1000m*1000m*1000m, and the optimal solution of the traveling Salesman problem is the optimal solution of route planning. Remember the weighted graph G = (V, E), V is the vertex set, E is the set of weights or lengths on the path[7]. Set:
others path optimal the on j i if xij , 0 ) , ( , 1 (1) i, j is the adjacent track point
n i n j ij ijx d MinZ 1 1 (2)The mathematical model of the classic traveling Salesman problem is as follows:
1 , 0 5 1 2 , 1 4 , 3 , . 1 1 ij s i j s ij n i ij n j ij x n S V S S x V j x V i x t s ) ( , ) ( ) (In the equation, dij——Distance between vertices
S——All non-empty subsets of vertex set V
|S|——The number of vertices contained in the weighted graph G in the set S
The constraints (3) and (4) have only one edge in and one edge out for each point, and the constraint condition (5) ensures that there are no subloop solutions in the model.
3. Ant colony algorithm
3.1 Basic Ant Colony algorithm
Ant colony algorithm is inspired by the process of ants looking for food in nature[8]. In the process of
looking for food, ants leave pheromones on their path, and other ant colonies around them are aware of this material because of the special substance secreted by ants. Ants leave more pheromones when they pass through a short path. Due to their ability to perceive pheromones[9], ants around them are
urged to approach this path and pass through this path, and with the passage of time. There are more and more pheromones on this path, and the ant population is also increasing on this path. Finally, the whole ant population is under the action of the positive feedback mechanism of information concentration. Choose the best path to find food, corresponding to the optimal path of the problem to be solved. The basic principle of the algorithm is as follows:
In three-dimensional space, there are n track points and m ants exist. Then: 2 2 2 ( ) ( ) ) ( j i j i j i ij x x y y z z d (6)
In the equation,(xi,yi,zi)——Projection coordinates of track point i in three-dimensional coordinate system
(xj,yj,zj)——Projection coordinates of track point j in three-dimensional coordinate system The state transition probability of ant W from track point i to track point j is:
others allowedW j t t t t p allowedW s ij ij ij ij ij , 0 , ) ( ) ( ) ( ) ( (7) ij ij d 1 (8)In the equation, nij——Heuristic function from i to j
τij——Pheromone concentration from track point i to track point j after time t
allowedW——A collection of track points that ants have not yet visited
α——The pheromone factor of the amount of information accumulated by ants in the process of
searching for a path
β——Ants looking for heuristic function factors of path certainty
Each time the ant selects the next track point, it will leave a pheromone on that path, and the pheromone will be updated in the following ways:
1 0 , ) ( ) 1 ( ) 1 ( 1 (ij t ) t ij t ij (9)
m k k ij ij 1 (10) In the equation, τij(t+1)——The information element concentration of ants on the path from trackpoint i to track point j at the t+1 time
ρ——Pheromone volatilization factor
(1-ρ)——The degree of pheromone retention
△τij ——The increment of pheromones in the search path △τk
ij——The k-th ant searches for pheromones left on the track at (t,t+1) time
3.2 Improved Ant Colony algorithm
For the basic ant colony algorithm, due to the influence of positive feedback mechanism and pheromone volatilization, ants give priority to the shortest path when choosing the next path, and when the number of track points increases, ants will ignore the selection of some optimal track points, so that the amount of information is reduced, and some optimal paths can not be selected, causing the algorithm to fall into the local optimal solution. It affects the ability of the track point to search the solution space globally in three-dimensional space. Therefore, an improved ant colony algorithm is proposed.
3.2.1 Adaptive pheromone update mechanism
In order to avoid the search route planning falling into the local optimal solution due to the increase of the track point scale, the pheromone updating mode of the formula (6) is improved.
min max , ) ( ) ( , ) ( ) ( ) 1 ( 1 min max 1 max min t t t t t ij ij S S t ij ij tdt S S t ij t t ij t t ij (11) In the equation, Smin——Ants are searching for the shortest path from track point i to track point j
4
Smax——Ants are searching for the longest path from track point i to track point j
γmax——Maximum factor
γmin——Minimum factor
γmax and γmin introduce the maximum and minimum factors for the assignment at the beginning of
the algorithm, which makes the ant colony algorithm improve the pheromone update mechanism according to the positive feedback mechanism. In the case of the passage of time, the pheromone can be adaptively and dynamically volatilized reasonably, the solution space of the search is more diverse, and the ability of global searching track path is improved.
3.2.2 Local pheromone optimization
When the ant completes the selection of the track point and makes the next path selection, the ant releases the sum of pheromones on the path, updates the pheromone more evenly, and dynamically adjusts the pheromone on each path. The update mode is as follows:
others
S
j
i
W
S
Q
S
j
i
W
S
Q
t
ij,
0
)
,
(
,
)
,
(
,
)
(
min min max max
(12)In the equation, Q——The total amount of pheromones released by ants after completing the track point
Through the way of optimizing local pheromones in this way, ants will not excessively gather on one path or over-disperse on other unreasonable paths every time they search for a path, which makes the pheromones on the path more uniform and avoid falling into precocious and local optimal solutions. The flow chart of the improved ant colony algorithm for solving the route planning problem is shown in figure 1:
Fig.1 Flow chart of improved ant colony algorithm
4. Simulation experiment and result analysis
operating system to verify the effectiveness, rationality and superiority of the improved ant colony algorithm for route planning in three-dimensional space. The landform parameters obtained in three-dimensional space are obtained according to GPS, and the characteristic points are recorded and abstracted into track points in route planning, and the track points are obtained as shown in figure 2:
Fig.2 50 track points
4.1 Parameter optimization and adjustment
At the beginning of the simulation experiment, the parameters need to be initialized. Due to the introduction of two maximum and minimum factors, the comparison experiments are carried out for the information element factor α, the heuristic function factor β, the information element volatilization factor ρ and the total information element Q. And assume that there are no threat points and obstacles between the two track points. The parameter settings of the improved ant colony algorithm are shown in Table 1:
Tab.1 Test parameter setting
Parameters α β ρ γmin γmax
1 0.1 0.1 0.01 0.01 0.1 2 0.3 0.3 0.03 0.03 1 3 0.7 0.7 0.07 0.07 5 4 1 1 0.1 0.1 10 5 2 2 0.2 0.5 20 6 3 3 0.3 1 30 7 4 4 0.4 2 50 8 5 5 0.5 4 100
Through a large number of comparative experiments on parameter adjustment, it is concluded that when the parameters α = 4, β = 5, ρ = 0.3, γ max = 20, γ min = 0.5, and the number of ants is about 1.5 times the number of target points, the improved ant colony algorithm has a good efficiency for route planning. When the number of track points is 30, the route of route planning is shown in figure 3:
6
Fig.3 Path planning of 30 track points
4.2 Comparative Analysis of algorithms
In order to verify the performance of the improved ant colony algorithm, it is compared with the basic ant colony algorithm. Based on the parameters obtained from the above experiments, the experimental comparison of the two algorithms is carried out. When the number of track points is 50, the results are shown in figure 4:
A. Basic ant colony algorithm B. Improved Ant Colony algorithm Fig.4 Route planning of two algorithms for 50 track points
A. Basic ant colony algorithm B. Improved Ant Colony algorithm Fig.5 Route planning of two algorithms for 100 track points
The data obtained from the simulation experiments in route planning according to the two algorithms are shown in Table 2:
Tab.2 Comparison of experimental data of two algorithms
Number of track points
Maximum number of iterations Path length/m
Basic ant colony
algorithm colony algorithm Improved ant Basic ant colony algorithm colony algorithm Improved ant
30 8 7 2219.31 2135.36
50 18 15 3687.96 3571.21
100 154 41 16897.68 15018.46
150 240 59 25766.51 24159.37
200 312 76 33153.20 31069.13
From the data in the table, it can be concluded that when the difference in the number of track points is small, from the number of iterations of the two algorithms, the improved ant colony algorithm is slightly better than the basic ant colony algorithm, and from the track path length, the improved ant colony algorithm is also better than the basic ant colony algorithm. For example, when the number of track points is 50, the maximum number of iterations of the improved ant colony algorithm is 83.33% of that of the basic ant colony algorithm, and the track route length is 96.83% of that of the basic ant colony algorithm. When the number of track points increases, the number of iterations and track route length are far better than the basic ant colony algorithm. For example, when the scale of track points reaches 200. The maximum number of iterations of the improved ant colony algorithm is 24.36% of that of the basic ant colony algorithm, and the track length is 93.71%. This shows the effectiveness and superiority of the improved ant colony algorithm.
5. Conclusion
In this paper, by proposing an adaptive pheromone updating mechanism and an improved ant colony algorithm based on local optimization of pheromones, on the premise of extracting feature points from topography as track points to ensure the safe flight of UAV, for the track planning problem, a large number of simulation experiments are carried out and the following conclusions are drawn:
(1) The improved ant colony algorithm has good performance in solving the route planning problem. When the setting parameters are α = 4, β = 5, ρ = 0.3, γ max = 20, γ min = 0.5, the good solution quality can be obtained.
(2) Compared with the basic ant colony algorithm,when the scale of the track point increases, the improved ant colony algorithm avoids falling into the local optimal solution prematurely, improves
8
the global search solution space of route planning, and makes the solution space diversified, so that a more optimal solution can be obtained.
Through the research on route planning based on the improved ant colony algorithm, this paper provides the possibility for UAV to carry out complex tasks, and to a certain extent, it can reduce fuel consumption and improve the performance of UAV. UAV can better serve intelligent agriculture, build smart cities, and make contributions to drawing a beautiful new China.
Acknowledgments
This paper is a key project of Chongqing Education Commission (Research and Application of Intelligent Citrus picking Technology and equipment) (KJZD-K201901202) and The Natural Science Foundation of Chongqing Science and Technology Commission (Study on Mechanical damage Mechanism and Hyperspectral Imaging Nondestructive testing of Citrus picked by Robot) (cstc2020jcyj-msxmX0143) is one of the phased achievements
References
[1] Liang Hu, Luo Qiang, Deng Zhenghua, et al. Improvement and Simulation Analysis of omnidirectional mobile robot[J]. Mechanical Transmission,2016, v.40;No.230(02):133-136 [2] Li Jun, Zhou Hu, Li Bo. A Local Optimized Ant Colony Algorithm Based on Virtual Ant[J].
Control and Decision-making ,2019.
[3] Hong Xiang, Zhang Xinchi. A Study on the Algorithm of Aerial Point Planning of Agricultural UAV Based on Embedded[J]. Agricultural Mechanization Research, 2019,41(08):45-49. [4] Wang Chong, Li Jun, Li Bo, et al.An Improved Ant Colony and Particle Swarm Hybrid
Algorithm for Solving Traveller's Problem[J]. Computer simulation,2016(11).
[5] Fang Chuncheng, Sun Peiming. Robot Path Planning[J]. Based on Improved Ant Colony Algorithm Measurement and Control Technology: 2018/037(004):28-31.
[6] Zhang Yan, Li Hengfan.A Comparative Study of Path Planning Algorithm for Mobile Robot[J] and MCU and embedded system applications,2020(7).
[7] Xu Yuqiong, Lou Ke, Li Tingting, et al.[J]. Path Planning for Mobile Robot Improved Adaptive Ant Colony Algorithm Journal of Electronic Measurement and Instrumentation, v.33; No.226(10): 94-100.2019
[8] Liang Kai,Mao Jianlin[J]. Path Planning for Indoor Mobile Robot Based on Improved Ant Colony Algorithm Electronic Measurement Technology, 2019(11).
[9] Jiaxu, Ning, Qin, et al. A best-path-updating information-guided ant colony optimization algorithm[J]. Information Sciences An International Journal, 2018.