Proceedings of the 38th Chinese Control Conference

July 27-30, 2019, Guangzhou, China

3D Multi-UAV Collaboration Based on the Hybrid Algorithm of

Artificial Bee Colony and A*
Xin Bai1, Ping Wang2, Ziye Wang1, Lan Zhang1
1. School of Automation & Electronic Engineering, University of Science & Technology Beijing, Beijing 100083
E-mail: [email protected]
2. Space Star Technology Co., LTD, Tianjin 300301, China
E-mail: [email protected]

Abstract: Multi-UAV cooperative operation has more advantages in penetrating enemy defense systems, detecting targets and
attacking missions. Firstly, a 3D multi-UAV track planning method based on the hybrid algorithm of artificial bee colony and
A* is presented. Theoretical and simulation researches have shown that the hybrid algorithm has better performance in
avoiding threats and smaller track cost compared to A* algorithm. Secondly, the adaptive time collaboration under the 3D
multi-UAV collaborative track planning based on the hybrid algorithm is studied. According to the time arrange, the adaptive
time coordination state is divided into simultaneous arrival, sequential arrival, and uncoordinated arrival. In the case of
sequential arrival, we focus on the iterative selection of the arrival time difference instead of specified time intervals, which
can maximize the possibility of the multi-UAV sequential arrival and enhance the ability of coordinated attack.
Key Words: Hybrid algorithm, 3D multi-UAV track planning, Adaptive time collaboration

limitations of a single algorithm, improvement of algorithm

1 Introduction and mix of multiple algorithms become common solutions.
With the rapid development of the aviation technology Reference [10] introduced the covariance to improve the
and artificial intelligence, unmanned aerial vehicles (UAVs) ABC algorithm, and reference [11] proposed a mobile
have broad application prospects in civil fields and military robot track planning algorithm based on the combination of
investigations due to their flexible operation and low cost. ABC and RRT. In this paper, a hybrid algorithm combining
The cooperative combat of multiple UAVs in military ABC and A* is proposed to carry out three-dimensional
missions has more advantages than the single UAV in UAV track planning. First, we determine a few number of
penetrating enemy defense systems, detecting targets, and track points and use the ABC algorithm to generate the
attacking missions [1]. Track planning and multi-UAV initial planning to reduce the computational complexity and
collaboration have become the focus of current research. expect local and global optimality, the initial track is not
The algorithm is the core part of the UAV's autonomous certain due to too few track points; Then the adjacent initial
track planning [2]. Classical algorithms include A* track points are inserted track points which are planned
algorithm, mixed integer linear programming (MILP) [3], with the A* algorithm to obtain the final certain track.
distributed algorithms and so on. In these algorithms, A* Multi-UAV collaboration is also one of the popular
algorithm has many researches and applications [4~6]. It is a researches. Researches on multi-UAV collaboration
classical heuristic search algorithm, which runs fast and include coordinated attacks [1, 5, 12, 13], UAV formation
can quickly solve and get the shortest path, but it is not control [14], cooperative information collection [15], task
ideal for avoiding threats and the track cost is relatively assignment [16] and so on. Time coordination goal is one of
large compared with other algorithms; The intelligent the main tasks in multi-UAV coordinated attack.
algorithms include genetic algorithm (GA) [7, 8], particle Considering the increasing effectiveness and power of
swarm optimization (PSO), and artificial bee colony combat, most of the time coordination focuses on arriving
algorithm (ABC) and so on. Among them, the ABC at the same time. But when the time cannot satisfy the
algorithm simulates bee foraging, and has good global and simultaneous arrival constraint, arriving sequentially which
local optimization ability. After iterative optimization, the has a similar attack effect can be taken into account.
ideal path can be obtained and the obstacle avoidance Sequential arrival can also reduce the chance of exposure
ability becomes strong. Reference [9] solved the UAV to enemy in air defense facilities at the time of
track planning in 3D environment with the ABC algorithm. investigation and evaluation. Reference [12] and reference
The ABC algorithm requires a great number of bees and [13] introduce the multi-UAV coordinated track planning
iterations to generate a reliable path with multiple track of arriving at the same time and arriving sequentially with
points. However, too many track points will complicate the specified time intervals when not meet the simultaneous
calculation, too few makes the track uncertain. Due to the arrival requirements. Reference [17] proposed a method of
track planning based on Laguerre diagram and closed-loop
*
This work is supported by National Natural Science Foundation
(NNSF) of China under Grant 00000000.
speed control to achieve multi-UAV sequential arrival with

3982
Authorized licensed use limited to: NORTHWESTERN POLYTECHNICAL UNIVERSITY. Downloaded on April 29,2023 at 09:16:34 UTC from IEEE Xplore. Restrictions apply.
 specified time intervals. However, it is difficult to achieve of the original honey source. The ABC algorithm has good
the sequential arrival when specify the time intervals. This searching ability and can jump out of the local optimum.
paper considers searching adaptive time intervals by Within a certain range, the more the population of bees and
iterative judgment of the time difference of arrival to the number of iterations, the better the track, but the spatial
maximize the probability of multi-UAV sequential arrival complexity and calculation time are multiplied, moreover,
to improve the multi-UAV attack effectiveness. the local search ability of the ABC algorithm is weakly
Based on the above, the rest of this paper is as follows: convergent.
The second part introduces the track cost model; The third 3.2 A* Algorithm
part studies the ABC algorithm, A* algorithm and the
The A* algorithm [4~6] is a classical heuristic search
process of UAV track planning with the hybrid algorithm
algorithm whose track cost formula is as follows:
based on ABC and A*; The fourth part introduces the
multi-UAV adaptive time coordination with the iteration of ˆሺሻ ൌ ‰ሺሻ ൅ Šሺሻ (2)
time difference; The fifth part is the simulation results and  represents the current track point, ˆሺሻ represents the
discussion; The sixth part is the research summary. track cost, ‰ሺሻ represents the track cost of the current
point to the starting point, and Šሺሻ represents the track
2 Track Cost Model
cost of the current point to the end point, which is taken as
The track cost in UAV track planning is one of the track the length between the two points.
evaluation criteria. Track cost is as follows: The A* algorithm iteratively finds each track point to get
ൌ ሺͳ െ Ⱦሻሺ‫ܬ‬௙௨௟௘ ൅  ‫ܬ‬௛௘௜௚௛௧ Ȁʹሻ ൅ Ⱦ‫ܬ‬௧௛௥௘௔௧ (1) the final path. Each track point has minimum track cost,
Where Ⱦ is the threat cost coefficient, ͳ െ Ⱦ is the fuel which is searched from the nine-square grid points centered
and height cost coefficient, Ͳ ൑ Ⱦ ൑ ͳ; ‫ܬ‬௙௨௟௘ is the fuel on the current track point. Due to the simple selection of
consumption cost, which is proportional to the path length track cost, the A* algorithm runs fast, but it also leads to
in 2D and 3D track planning; ‫ܬ‬௛௘௜௚௛௧ is the height cost insufficient optimization ability of the algorithm.
which is equal to zero in 2D track planning and related to 3.3 Track Planning with Hybrid Algorithm Based on
the height difference between adjacent track points on the ABC and A*
minimum threat surface in 3D track planning; ‫ܬ‬௧௛௥௘௔௧ is the The flow chart of the hybrid algorithm is shown in Fig. 1.
threat cost, the centre points of threatening mountains are First, the initial track is obtained by ABC algorithm with
accounted as the threat points and the threat range is the fewer population of bees and fewer track points. The track
circular domain. The threat cost of a point is sum of the between two adjacent track points is further planned by the
reciprocal of the fourth power of the distance between A* algorithm to obtain more track points so that we can get
threat points and this point. When the track segment within a certain track. Proceed as follows:
the threat range, the track threat cost is the integral of the
1) Combine the actual environment with the threat model
point threat cost along the track segment [18]; when the
as the basic topography, and conduct the terrain smoothing
track segment is out of the threat range, the track threat cost
and curvature limits to meet the UAV’s constraints,
is equal to zero. In order to simplify the calculation, the
including climbing and diving angle, turning angle and
track threat cost is approximately equal to the sum of the
minimum ground clearance to generate the minimum threat
threat cost at ͳȀͳͲǡ ͵ȀͳͲǡ ͷȀͳͲǡ ͹ȀͳͲ and ͻȀͳͲ of the
surface [21, 22]. Since 3D ABC optimization requires a large
track segment instead of the integral of the point threat cost
amount of computation and has a large spatial complexity,
along the track segment.
firstly, we convert the 3D track planning into a 2D track
3 Track Planning Algorithm planning to simplify the operation. Projecting the 3D
3.1 ABC Algorithm fusion topography onto the 2D plane, retaining the threat
information and performing 2D track planning with the
The ABC algorithm [9~11, 19, 20] is a bionic intelligent hybrid algorithm of this paper. The 2D and 3D track costs
algorithm. It divides bees into hiring bees, following bees need to be applied to determine the optimal 2D track, and
and exploring bees. Hiring bees are responsible for finding then the optimal 2D track points are projected back onto
the source of honey and share the information of the found the 3D minimum threat surface to obtain a 3D track. This
honey source with the following bees through their dance. method can reduce the spatial complexity and calculation
Following bees select the appropriate honey source to amount of searching directly in 3D space [21, 23].
collect honey according to “roulette”. When the honey
2) The ABC algorithm optimizes the initial track points:
source cannot be updated to the maximum number of times,
In order to guarantee the operation speed, the fewer number
we need to abandon the original honey source and avoid
of track points are preset (10 in the algorithm comparison
being trapped in local optimum. At this time, the exploring
simulation part and 5 in the multi-UAV collaboration
bees are responsible for finding a new honey source instead
simulation part). To ensure the excellent path and eliminate

3983
Authorized licensed use limited to: NORTHWESTERN POLYTECHNICAL UNIVERSITY. Downloaded on April 29,2023 at 09:16:34 UTC from IEEE Xplore. Restrictions apply.
 the abnormal search situation, we need multiple operations The flow chart of adaptive time coordination of three
to obtain multiple optimal paths and select one path with UAVs is shown in Fig. 2. The iteration condition of ȟ–,
the lowest 3D track cost from them as the initial track. time coordination state and the schematic diagram are
3) A* algorithm optimizes the track: Select the adjacent shown in Table 1.
track points in the initial track as the starting point and the START
end point respectively, and use the A* algorithm to conduct
the second planning to obtain the track points between the Path planning
initial track points, then get the final 2D track points.
4) Restore the height information of the optimal 2D Get track length of each UAV:Li

track, which means project the 2D track points onto the velocity=[100m/s,150m/s]
minimum threat surface to obtain the 3D track points, and
Get flight time range of each
the 3D path is determined after the connection. UAV:Ti=Li/v Arrive at the same time
END

Take the minimum

Calculate the
time of the
START common intersection(Ti) Y speed of each
intersection (Tmin)
UAV
as arrival time
Sequential arrival N
Project Terrain Information
to the 2D plane Sort Timin(The shortest
flight time of each UAV)
Take T1min<T2min<T3min as an example
2D path planning with ABC of
Less population and iteration
UAV1 willl arrive at T1min
Arrival time
N satisfy UAV1:T1min
Number of Select t UAV2:T1min+t
the condition
runs is max? N UAV3:T1min+2t

Y UAV2 will arrive at T2min

Select the optimal track Arrival time
based on the 3D track cost satisfy UAV1:T2min-t
Select t
the condition UAV2:T2min
N UAV3:T2min+t
Optimize the track
with A* algorithm
UAV3 will arrive at T3min
Arrival time
Restore The height information satisfy UAV1:T3min-2t
Select t UAV2:T3min-t
Project the 2D track points onto the condition
the minimum threat surface UAV3:T3min
N

Unable time
END
coordinated arrival

Fig. 1: Track planning with hybrid algorithm flow chart Fig. 2: Adaptive time constraint flow chart

4 Multi-UAV CollaborationüüAdaptive Time Table 1: Adaptive time coordination condition

Collaboration Schematic Diagram Iteration Condition State
In this paper, taking three UAVs as an example, the track 't 0 Simultaneous
Arrival
length of each UAV obtained by the hybrid algorithm is ‹,
‹ ൌ ͳǡʹǡ͵, and the flight speed range of the three UAVs is
˜, ˜ ‫ א‬ሾͳͲͲȀ•ǡ ͳͷͲȀ•ሿ. The flight time range of each UAV 1 will arrive at T 1min
Sequential
­°'t  ['t1 min, 't1 max]
UAV is ‹, ‹ ‫ א‬ሾ‹ȀͳͷͲǡ ‹ȀͳͲͲሿ, ‹ ൌ ͳǡʹǡ͵. ®T 1min  2't  [T 3 min, T 3 max]
°̄'t  [0s,600 s]
Arrival

When these three ‹ have a common intersection, the

three UAVs can arrive at the same time; When there is no UAV 2 will arrive at T 2 min Sequential
­°'t  ['t 2 min, 't 2 max]
intersection, iterate over the time difference of arrival (ȟ–) ®T 2min - 't  [T1 min, T 1 max] Arrival
°̄'t  [0s,600 s]
and select the minimum ȟ– to make the three UAVs
sequential arrival; If there is no ȟ– satisfy the condition, it
UAV 3 will arrive at T 3min Sequential
is considered that the three UAVs cannot cooperate and the ­°T 3min - 't  [T 2 min, T 2 max]
track needs to be re-planned. Finally, the speed of each ®T 3min - 2't  [T1 min, T1 max] Arrival
°̄'t  [0s,600 s]
UAV is calculated according to the arrival time.

3984
Authorized licensed use limited to: NORTHWESTERN POLYTECHNICAL UNIVERSITY. Downloaded on April 29,2023 at 09:16:34 UTC from IEEE Xplore. Restrictions apply.
 The iteration of ȟ– can make the UAV arrive as fast as
possible with a small ȟ– . Compared with arriving
sequentially with a specified ȟ– , the iterative process
increases the possibility that the UAVs can satisfy the
sequential arrival condition. In this paper, ȟ– ‫ א‬ሾͲ•ǡ ͸ͲͲ•ሿ
is required. The range of ȟ– can be set as needed in
practical applications. In the future, we will consider the
event-triggered control to save energy [24, 25].

5 Simulation

5.1 Topography Simulation Fig. 4: Fusion topography

The Daxinganling area is selected from the geospatial

data cloud (http://www.gscloud.cn/), and the Global 5.2 Comparison of Hybrid Algorithm and A*
Mapper software is used to intercept the area of
12.6km×12.6km and convert the topography data into .grd
format for MATLAB reading. Set the interval to 30 meters,
sampling and smoothing to get the basic topography
simulation map as shown in Fig. 3.
Set the threat model and equivalent it to the peak model
as equation (3). The fusion topography of the basic
topography and threat model is as shown in equation (4).
௫ି௫బ೔ ଶ ௬ି௬బ೔ ଶ
‫ݖ‬ଶ ሺšǡ ›ሻ ൌ œ଴ σ௡௜ୀଵ ‫ݖ‬௜ ‡š’ሾെ ቀ ቁ െቀ ቁ  (3)
௫ೞ೔ ௬ೞ೔ Fig. 5: ABCüIterative of different population
œሺšǡ ›ሻ ൌ ƒšሺ‫ݖ‬ଵ ሺšǡ ›ሻǡ ‫ݖ‬ଶ ሺšǡ ›ሻሻ (4)
Where ‫ݖ‬଴ is the topography reference height, šˈ› are
the horizontal coordinate points, ‫ݖ‬ଵ ሺšǡ ›ሻ is the basic
topography height, ‫ݖ‬ଶ ሺšǡ ›ሻ is the height of the threat
model which is equivalent to the peak model,  is the
number of peaks, and ‫ݖ‬௜ is the highest height of the i-th
peak, ‫ݔ‬଴௜ and ‫ݕ‬଴௜ represent the horizontal coordinates at
the highest point of the i-th peak. ‫ݔ‬௦௜ and ‫ݕ‬௦௜ are slope
parameters of the threat model.
Parameters: ‫ݖ‬଴ ൌ ͸,  ൌ ͵. The horizontal coordinates
of the center points of the three threat models are
Fig. 6: ABCüIterative of different track points numbers
ሾͷͲǡͳͲͲǢ ʹͲͲǡͶͲͲǢ ͵ͷͲǡͳͲͲሿ , the vertical coordinates
corresponding the highest points are ሾͶͲͲǢ ͶͷͲǢ ͷ͵Ͳሿ, the
Fig. 5 are the curves of the track cost to the number of
slope parameters of the three threat models are set to
iterations with ͷ track points in population of ͳͲͲ and
ሾʹͲͲǡͳ͵ͲǢ ʹͲͲǡͳ͹ͲǢ ͳͳͲǡͳ͸Ͳሿ.
͵Ͳ respectively. We can observe the change of the
The fusion topography is shown in Fig. 4. population has little effect on the final iteration
convergence value and the number of iterations, but the run
time from the start to the 60th generation is reduced from
͵͵ǤͶ͵• of population ͳͲͲ to ͹ǤͲʹ• of population 30 in
simulation. The result shows the population reduction can
significantly shorten the iterative convergence time.
Fig. 6 are the curves of the track cost to the number of
iterations with population ͳͲͲ in track points of ͷ and 8
respectively. We can observe the change of the number of
track points has significant effect on convergence speed,
 with track points of 8, it requires more iterations to
converge to the optimal value.
Fig. 3: Basic topography Fig. 7 and Fig. 8 are the simulation diagrams of the track
planning by a single UAV using the A* algorithm and the
hybrid algorithm respectively, the horizontal coordinate of

3985
Authorized licensed use limited to: NORTHWESTERN POLYTECHNICAL UNIVERSITY. Downloaded on April 29,2023 at 09:16:34 UTC from IEEE Xplore. Restrictions apply.
 the starting point is ሺͳͲǡͳʹǤ͸ሻ and the horizontal
coordinate of the destination point is ሺͶͲͲǡ͵͹Ͳሻ , the
maximum climbing angle is ͸Ͳι, curvature is ͲǤͳ.

(a) Simultaneous arrival

Fig. 7: Track planning with A*

(b) Sequential arrival

Fig. 9: Multi-UAV adaptive time collaboration result

Fig. 9(a) shows the simultaneous arrival of the three

UAVs. The running result shows there is an intersection of
the three ‹, which is ሾ͹͵͸͵Ǥͻ•ǡ ͻͻͷͲǤʹ•ሿ, ο– is Ͳ•, and
͹͵͸͵Ǥͻ• is selected as the time when the three UAVs
reach the destination point at the same time. The speed of
Fig. 8: Track planning with hybrid algorithm each UAV is ͳ͵͹ǤͲͶȀ•ˈͳ͵ͷǤͳʹȀ• and ͳͷͲȀ•; Fig.
9(b) shows the sequential arrival of the three UAVs, the
running result shows there is no intersection of the three ‹,
Table 2: Track costs comparison of two algorithms
ο– is ͵ͻ͵Ǥ͵•, and the arrival time of the three UAVs is
Algorithm Fuel and Threat 3D ͺͲͷͻǤ͸•ˈͺͶͷʹǤͻ• and ͺͺͶ͸Ǥʹ•, the speed of each UAV
height cost cost total cost
A* 178.1802 581.3259 480.5394 is ͳͲͲǤͲͲͳͶȀ• , ͳʹͲǤʹ͵Ͷ͸Ȁ• and ͳͷͲȀ• ; The
Hybrid 215.1557 124.7718 147.3678 partial view inserted in the upper right corner near the
destination point shows the sequential arrival. Three UAVs
Table 2 shows the track costs comparison of two flew along the pink track until the first UAV arrived at the
algorithms. The fuel consumption and height cost of destination point, and then the two UAVs flew along the
hybrid algorithm related to the total length of the track is blue track. After the ο– time, the second UAV reached the
215.1557, which is slightly larger than the 178.1802 of the destination point. The last UAV flew along the green track
track of the A* algorithm. However, the threat cost to the destination point after the ο– time again, and the
corresponding to the obstacle avoidance effect is reduced three UAVs completed the sequential arrival.
from 581.3259 of the A* algorithm to 124.7718 of the 6 Conclusion
hybrid algorithm, showing a strong obstacle avoidance
This paper present a hybrid algorithm based on the ABC
capability of the hybrid algorithm.
and A* to realize the 3D multi-UAV track planning.
5.3 Multi-UAV Collaboration Based on the Hybrid Simulation shows that the hybrid algorithm has smaller
Algorithm threat cost and smaller total 3D cost, which means the
The multi-UAV collaboration simulation is shown in Fig. hybrid algorithm has better performance than A*. Then, the
9. The population of the ABC is 50, the number of the multi-UAV adaptive time collaboration is studied. UAVs
preset track points is 5, and the number of iterations is 60. can judge the time coordination state by the time ranges,
The UAV parameters refer to the multi-UAV cooperation and the time ranges are calculated by the track lengths and
part of Section 4 in this paper. ܺ଴௜ indicate the coordinates the speed range. When the state is simultaneous arrival, the
of the starting points of the three UAVs, ܻ଴ indicates the multi-UAV should select the shortest arrival time. When
coordinates of the destination points of the three UAVs. the state is sequential arrival, the iteration process of arrival

3986
Authorized licensed use limited to: NORTHWESTERN POLYTECHNICAL UNIVERSITY. Downloaded on April 29,2023 at 09:16:34 UTC from IEEE Xplore. Restrictions apply.
 time difference is adopted, which can maximize the [12] R. Xia, L. Zhao, and SY. Wu, Drone Cooperative Path
possibility of the multi-UAV sequential arrival and Planning Based on Artificial Bee Colony Algorithm,
Wireless Internet Technology, 15(13):13-21, 2018.
enhance the effectiveness of coordinated attack. When the [13] XX. Yang, WW. Zhou, and Y. Zhang, On Collaborative
state is uncoordinated arrival, the re-track planning is Path Planning for Multiple UAVs Based on Pythagorean
required. In the future, we can further study the multi-UAV Hodograph Curve, Chinese Guidance, Navigation and
Control Conference, 2016:971-975.
mission assignment, increase the diversity of missions and
[14] YB, Chen. JQ. Yu, and XL. Su, Path Planning for
strengthen the ability of UAVs to complete tasks. Multi-UAV Formation, Journal of Intelligent & Robotic
Systems, 77(01):229-246, 2015.
References [15] H. Ergezer, and K. Leblebicioglu, 3D Path Planning for
[1] ZX. Yao, M. Li, and ZJ. Chen, Mission Decision-making Multiple UAVs for Maximum Information Collection,
Method of Multi-aircraft Cooperatively Attacking Journal of Intelligent & Robotic Systems, 73(1-4):737-762,
Multi-target Based on Game Theoretic Framework, Chinese 2014.
Journal of Aeronautics, 29(06):1685-1694, 2016. [16] T. Li, J. Jiang, and ZY. Zhen, Mission Planning for Multiple
[2] ZH. Hu, Research on Some Key Techniques of UAV Path UAVs Based on Ant Colony Optimization and Improved
Planning Based on Intelligent optimization algorithm, Dubins Path, Chinese Guidance, Navigation and Control
Nanjing: Nanjing University of Aeronautics and Conference, 2016:954-959.
Astronautics, 2011. [17] XN. Guan, RX. Wei, and Q. Guo, A Cooperative Control
[3] M. Radmanesh, M. Kumar, and A. Nemati, Dynamic Method for Tight Sequencing Arrival of Multiple UAVs,
Optimal UAV Trajectory Planning in The National Airspace Electronics Optics & Control, 21(1):18-22, 2014.
System Via Mixed Integer Linear Programming, [18] Y. Liu, X. Zhang, and X. Guan, Adaptive Sensitivity
Proceedings of the institution of mechanical engineers part Decision Based Path Planning Algorithm for Unmanned
G-Journal of Aerospace engineering, 230(09):1668-1682, Aerial Vehicle with Improved Particle Swarm Optimization,
2016. Aerospace Science & Technology, 58:92-102, 2016.
[4] X. Yang, MY. Ding, and CP. Zhou, Fast On-ship Route [19] S. Li, and J. Wang, Research on Function Optimization
Planning Using Improved Sparse A-star Algorithm for Problem Based on Artificial Bee Colony Algorithm, Control
UAVs, The International Society for Optical Engineering, Conference (CCC), 2015 34th Chinese. IEEE,
7497:749705-749713, 2009. 2015:8428-8433.
[5] YZ. Xi, and DY. Zhou, Trajectory Planning for Coordinated [20] GZ. Tian, L. Zhang, and X. Bai, Real-time Dynamic Track
Attack with Multiple Uninhabited Air Vehicles, Computer Planning of Multi-UAV Formation Based on Improved
Simulation, 27(03):69-72+135, 2010. Artificial Bee Colony Algorithm, Control Conference
[6] HF. Su, DL. Xu, and WJ. Li, Automatic Search of The (CCC), 2018 37th Chinese. IEEE, 2018:6.
Transmission Line Path Based on The Improved Ant Colony [21] PF. Chen, WJ. Liu, and W. Wang, Route Planning of UAV
And A* Algorithm, Journal of Hebei University, Based on Particle Swarm Optimization Algorithm,
37(01):92-100, 2017. Electronic Design Engineering, 21(22):36-39, 2013.
[7] H. Shorakaei, M. Vahdani, and B. Imani, Optimal [22] XS. Chen, WG. Zhai, and WL. Zhao, Route Planning for
Cooperative Path Planning of Unmanned Aerial Vehicles by Unmanned Aerial Vehicles Based on the Particle Swarm
A Parallel Genetic Algorithm, Robotica, 34(4):823-836, Optimization, Modern Computer, 25:8-11+15, 2011.
2016. [23] Z. Lv, LY. Yang, and YQ. He, 3D environment modeling
[8] C. Ramirez-Atencia, G. Bello-Orgaz, and MD. R-Moreno, with height dimension reduction and path planning for UAV,
Solving Complex Multi-UAV Mission Planning Problems 2017 9th International Conference on Modelling ˈ
Using Multi-objective Genetic Algorithms, Soft Computing, Identification and Control (ICMIC), 2017:734-745.
21(17):4883-4900, 2016. [24] J. Liu, Y. Yu, and J. Sun, Distributed eventϋtriggered
[9] ZH. Hu, M Zhao, and M. Yao, Research of 3-D Track fixed ϋ time consensus for leader ϋ follower multiagent
Planning for Unmanned Aerial Vehicles with Sudden systems with nonlinear dynamics and uncertain disturbances,
Threats, Journal of Aeronautics, Astronautics and International Journal of Robust and Nonlinear Control,
Aviation, Series A, 43(3):167-176, 2011. 28(11):3543-3559, 2018.
[10] D. Kumar, and KK. Mishra, Co-variance Guided Artificial [25] J. Liu, YL. Zhang, and Y. Yu, Fixed-time event-triggered
Bee Colony, Applied Soft Computing, 70:86-107, 2018. consensus for nonlinear multiagent systems without
[11] TS. Yue, and HY. Chung, Using ABC and RRT Algorithms continuous communications, IEEE Transactions on Systems,
to Improve Mobile Robot Path Planning with Danger Man, and Cybernetics: Systems, 2018,
Degree, in International Conference on Future Generation doi: 10.1109/TSMC.2018.2876334.
Communication Technologies, 2016:21-26.

3987
Authorized licensed use limited to: NORTHWESTERN POLYTECHNICAL UNIVERSITY. Downloaded on April 29,2023 at 09:16:34 UTC from IEEE Xplore. Restrictions apply.