Review of Path Planning for Autonomous Underwater Vehicles∗

TingTing Yao Tao He† WenLong Zhao Abdou Yahouza M.Sani

School of Mechanical and School of Mechanical and School of Mechanical and School of Mechanical and
Automotive Engineering Automotive Engineering Automotive Engineering Automotive Engineering
Shanghai University Of Shanghai University Of Shanghai University Of Shanghai University Of
Engineering Science Engineering Science Engineering Science Engineering Science
Shanghai China Shanghai China Shanghai China Shanghai China
303058446@qq.com hetao_cn@126.com zwl1664@163.com mahamanesania@ymail.com

ABSTRACT An AUV(autonomous underwater vehicle) [1] is one of the most

important tools for current marine exploration research. To cope
An autonomous underwater vehicle (AUV) is an important tool for with the complex and varied marine environment, intelligent
the exploration and development of marine resources. Path underwater robots must be safe and autonomous. Autonomy is
planning plays an important role in the correct navigation and reflected in the ability of AUV to interact with the underwater
avoidance of obstacles by underwater robots in the sea. Firstly, environment, while real-time control and autonomous path
according to different marine environmental conditions, AUV path planning are one of the important aspects of interaction. The core
planning is classified into two types. Then, the related algorithms technology to ensure AUV security and autonomy is intelligent
of global path planning and local path planning are reviewed, and control technology, including autonomous planning, control and
the key technologies, development status, advantages and status monitoring.
disadvantages of related algorithms are combed. Finally, the
development direction of the AUV in complex environment is
prospected. 2 Overview of autonomous underwater robot
path planning methods
CCS CONCEPTS
• Computer systems organization ~ Robotic autonomy • Software 2.1 Basic tasks of path planning problems
and its engineering ~ Software system structures With the increasing use of underwater robots in civil marine
development and military fields, the demand for AUV's
KEYWORDS intelligence is also increasing. It is expected that the AUV can sense
AUV, Path planning, Global path planning, Local path planning the surrounding environment during the movement and plan the
path of the collision avoidance according to the surrounding
environment conditions, which is the path planning problem.
1 Introduction Generally, land-based robots usually avoid obstacles by stopping
Since the beginning of the 21st century, in the context of the motion and then turning when encountering obstacles. This method
growing scarcity of terrestrial resources, many countries have is not applicable to AUV floating in the ocean. Since the AUV is
accelerated their pace towards the oceans and humans have unable to complete emergency stops, terrain observations, and re-
increasingly attached importance to the research, development, and planning, it is necessary to plan the path in real-time during the
utilization of marine resources. As an important branch in the field navigation process.
of robotics, intelligent underwater robot technology has a very The basic task of path planning is to solve the problem of how
broad application prospect and application value. to effectively use environmental information by AUV. The AUV
† plans the shortest path based on sensor information or
Corresponding author: Tao He. Postal address: 333 Longteng Road, Songjiang
District, Shanghai University Of Engineering Science environmental information, smooths the path and goes to the target
Permission to make digital or hard copies of all or part of this work for personal or location. Path planning is the basis for AUV to perform various
classroom use is granted without fee provided that copies are not made or distributed
for profit or commercial advantage and that copies bear this notice and the full tasks, reflecting the ability of AUV to interact with the environment
citation on the first page. Copyrights for components of this work owned by others during the movement. The path planning working processes is
than ACM must be honored. Abstracting with credit is permitted. To copy otherwise,
or republish, to post on servers or to redistribute to lists, requires prior specific
shown in Figure 1.
permission and/or a fee. Request permissions from Permissions@acm.org.
RICAI '19, September 20–22, 2019, Shanghai, China
© 2019 Association for Computing Machinery.
ACM ISBN 978-1-4503-7298-5/19/09…$15.00
https://doi.org/10.1145/3366194.3366280

483
 Get Sensor Data or
Find Shortest Smooth the
Step along optimal three-dimensional path offline, and comprehensively
Environmental New Smooth
Information
Path Path
Trajectory considered the optimization degree of the path to obtain a planning
method suitable for the underwater three-dimensional environment.
Mahmoudzadeh et al. [3] proposed a path planning method and
No
Is Goal Reached? simulated it. The results show that the genetic algorithm has
Yes
obvious advantages in stability. Zadeh et al. [4] designed a path
End planning method in which the global planning of the path uses the
Figure 1: Path planning working processes GA algorithm.
The disadvantage of the genetic algorithm is that it is easy to fall
2.2 Classification of Path Planning Methods into the local optimal solution, and the calculation time is too long.
Cao et al. [5] aiming at the characteristics of underwater path
According to the planning level, the path planning method can be
planning, using grid to mark the space, and then using genetic
divided into global path planning and local path planning.
algorithm to find the best, adding chamfer operator smoothing path
Global path planning requires a thorough understanding of all
in the algorithm, speeding up the convergence speed of the
the information in the environment in which the robot is located,
algorithm and reducing path planning time. Another advantage of
and an optimal path that meets certain performance requirements
the genetic algorithm is that it can combine other algorithms well,
from the start point to the end point. It is based on a priori
improve the path planning ability of underwater robots, and is
environmental information, and the accuracy of the collection of
excellent in self-organization and self-learning. It can still maintain
environmental information determines the accuracy of the planning.
good global optimization when combined with other algorithms.
The disadvantage is that all the environmental information needs to
PAN et al. [6] combined genetic algorithm with ant colony
be known in advance, which is difficult to achieve in practical
algorithm to improve planning efficiency through reasonable
applications, so the global path planning is static path planning. At
partitioning. Yao et al. [7] improved the GA algorithm and used
present, the commonly used three-dimensional global path
Grey wolf optimization to make the GA algorithm mutate in the
planning algorithms for underwater robots include genetic
ideal direction. The simulation results show that the improved GA
algorithm, ant colony algorithm and rapidly exploring random tree
algorithm can improve the convergence speed of the algorithm and
algorithm. There are also some heuristic algorithms similar to such
reduce the path search time.
algorithms, such as particle swarm algorithm, topological method,
visualization method. Since the improved method of the algorithm
3.2 Ant Colony Optimization Algorithm (ACO)
is similar to the above algorithm, it will not be described here.
Local path planning is used to when the robot cannot obtain a The ant colony optimization algorithm is an algorithm derived from
priori global environmental information. It can only rely on the the path optimization of the simulated ant colony searching for food.
currently obtained information to plan the path of the robot in There are two behaviors of pheromone and perceptual pheromone
online form. It is suitable for the situation where the starting point in the foraging process of ant colony. When the pheromone
is close to each other and can be used as a global path planning quantity and intensity of a certain path are high, it indicates that the
supplement. Local path planning is more practical, but due to the path is a better one and has a greater possibility to be selected,
limitations of the robot to obtain information, it is impossible to presenting positive information feedback. This kind of algorithm is
ensure that the planned path is optimal, and the deadlock characterized by distributed computing capability, and can search
phenomenon may occur, that is, the AUV cannot reach the end for solutions simultaneously in global multi-points, which has
point smoothly. The three-dimensional local path planning strong robustness. The path planning of AUV usually needs to be
methods commonly used in AUV are: artificial potential field carried out in three-dimensional space, and the ant colony
method, fuzzy logic algorithms, artificial neural network method algorithm can realize the 3D path search problem [8].
and A* algorithm. In addition to this, there are extended algorithms However, the ant colony algorithm does not have the
such as D* algorithm and artificial immune method. mathematical theory to solve the problem of function optimization
with constraints. The convergence speed is slow and it is easy to
fall into the local minimum. Therefore, many experts and scholars
3 Technical Methods for Global Path Planning have improved on the basis of traditional ant colony algorithm.
Wang et al. [9] introduced a series of pheromone update rules for
3.1 Genetic Algorithm (GA) ant colony algorithm, designed a new cutting operator and insertion
Genetic algorithm is an algorithm that simulates the biological point operator to realize the smoothing process of simulation
genetics and evolutionary effects in nature. It originated from planning path. The simulation results show that the improved ant
Darwin's biological evolution theory and is characterized by colony algorithm is applicable. In global path planning, you can
random search-ability, adaptability and high parallelism. Moreover, quickly optimize the AUV working path and be safer and more
this algorithm has good robustness and outstanding advantages stable. Lin et al. [10] proposed two kinds of AUV real-time path
compared with traditional algorithms. For AUV path planning, the planners based on quantum ant colony optimization and mixed ant
stealth parallelism of the algorithm is beneficial to global path colony optimization. When static obstacles appear in front of the
search. M. Ataei et al. [2] used the GA algorithm to generate the AUV during the voyage, a new path will be planned to avoid

483
obstacles, and the local path planning and ant colony algorithm will autonomous underwater vehicle (HAUV). The simulation results
be combined to obtain smoother paths. Dong et al. [11] proposed show that the algorithm can achieve stable path planning. Nan [16]
an improved ant colony algorithm to improve the shortcomings of proposed a closed-loop fast exploration random tree (CL-RRT)
traditional ant colony algorithm in practical application. This algorithm, CL-RRT is a closed-loop stochastic dynamic path
algorithm added reward and punishment mechanism in the process planning algorithm based on incremental sampling method, capable
of updating pheromones, which improved the disadvantages of ant of 3D in AUV with six degrees of freedom. Plan feasible paths in
colony algorithm, such as slow search speed and easy to fall into space to quickly complete specific tasks. Li et al. [17] proposed an
local optimal solution. Ant colony algorithm can be combined with improved algorithm for AUV path search Li-RRT. Compared with
other algorithms to improve the low efficiency and time consume the traditional RRT algorithm, the higher the node activity in this
of AUV optimization. PAN et al. [6] proposed a genetic-ant hybrid algorithm, the more easily the random search tree is expanded and
algorithm to simplify the model and design a reasonable the computational efficiency is higher. The comparison of the three
partitioning method to reduce the population search range. An algorithms is shown in Table 1.
optimization strategy to improve efficiency is proposed to
determine the evaluation function. The simulation results show the Table 1: Comparison of the above algorithms
search time and the number of iterations is reduced.
advantages disadvantages
3.3 Rapidly Exploring Random Tree (RRT)
robust, strong search Fall into a local optimal
The rapidly exploring random tree (RRT) algorithm is an
capability, high search solution, poor stability,
incremental forward search algorithm. It starts from the start state GA
efficiency long calculation time
point and looks for the parent node until it reaches the target state
point, and plans a path that satisfies the global constraint. It has highly robust, the Slow convergence rate,
powerful spatial searchability. The RRT algorithm expansion underwater 3D path easily trapped in local
process is shown in Figure 2. Heo et al. [12] used the RRT ACO
search optimum
algorithm in the AUV and found that the underwater environment
can be searched quickly and efficiently. Powerful spatial Not the optimal path,
searchability larger resource
qrand RRT
consumption
qnear qnew

4 Technical Methods for Local Path Planning

qinit 4.1 Artificial Potential Field (APF)

The APF algorithm can find the optimal path by imitating the
motion of the object under the gravitational repulsion and
establishing the potential field function. In the function, the
Figure 2: RRT algorithm expansion process
gravitational force is between the object and the target point, and
the repulsion force is between the object and the obstacle. For
However, the RRT algorithm often has large computational
general terrestrial and aerial robots, the artificial potential field
complexity and falls into local minimum. In the RRT algorithm,
method has been widely developed in path planning, but there is
many improved algorithms are widely used in AUV path planning.
still much room for development of robots in complex underwater
The RRT* algorithm is characterized by relatively uniform
environments. Based on the artificial potential field method, Zhu et
sampling, does not fall into local minimums, and can converge to
al. [18] combined the velocity synthesis algorithm with the artificial
an optimal solution, which can be used to solve problems in a three-
potential field on the basis of the artificial potential field method to
dimensional environment. The RRT* algorithm is used in the AUV,
overcome the influence of current on the underwater navigation of
and the 3D experiment is carried out in the real sea area. The results
AUV, and proposed a new AUV path Planning algorithm. Liu et
show that the path planning effect is better [13]. Jeong et al. [14]
al. [19] mainly used the method of reinforcement learning to
proposed a Q-RRT* algorithm, which has better initial solution and
control and make decisions on AUV. The artificial potential field
convergence speed than RRT* algorithm. Bi-RRT is a dual-tree
algorithm was combined with Q learning algorithm and BP neural
expansion algorithm that is more efficient than singular expansion.
network to avoid collision planning of AUV. AUV adapts to the
The algorithm extends the random tree outward from the two ends
new environment through learning and continuously improves its
of the set path, namely the initial point and the target point. After
autonomous ability, so as to achieve obstacle avoidance task in
selecting a random node, two trees are selected in turn, and then
uncertain environment. Due to the influence of water flow speed,
random expansion is performed on the basis of this, and the
the traditional artificial potential field method needs to be improved
expansion order of the two trees is transformed in the next iteration.
in terms of real-time, dynamic and accuracy. In order to improve
Kim et al. [15] applied this algorithm in the underwater hovering
the dynamic obstacle avoidance ability, Cheng et al. [20] combined

484
the speed synthesis algorithm with the artificial potential field
method to rationally plan the moving path of the AUV in the ocean Fuzzy
Fuzzification Defuzzification
Reasoning
current while avoiding the obstacle. The simulation shows that the AUV
method can effectively improve the moving speed. Das et al. [21]
proposed a new improved algorithm that combines artificial Fuzzy Rules
potential field method and ant colony algorithm to deal with
obstacle avoidance problems, and then combines clonal selection
optimization algorithm to reduce the possibility that artificial Environmental
potential field method is easy to fall into local minimum problem , information
to achieve optimal path selection. Zhang et al. [22] used the motion Figure 3: Fuzzy logic planner
direction as a control variable and applied the chaos optimization
algorithm to the optimized artificial potential field method to solve 4.3 Artificial Neural Network (ANN)
the instability problem in the traditional artificial potential field
method. The neural network algorithm is an algorithm that simulates human
brain thinking. Firstly, the obstacle information is used as the input
4.2 Fuzzy Logic Algorithm (FLA) layer information of the neural network. After the neural network
process, the final sample set is obtained, and then the output layer
Fuzzy logic is an online programming method based on real-time
comprehensive environmental factors guide the robot to navigate
sensor information. It uses the location and motion information of
the obstacle avoidance. Finally, Get the optimal path. The neural
reference objects containing fuzzy concepts to construct a two-
network algorithm has good learning ability and strong robustness.
dimensional membership function based on dynamic environment
However, this algorithm has a large amount of processing data and
model. Using the reflective navigation mechanism, the information
a slow calculation speed. It is not easy to obtain an ideal training
of current environmental obstacles is input into the Fuzzy logic
model when dealing with dynamically changing environments. The
planner, and the desired steering Angle and speed of the robot are
principle of neural network is shown in Figure 4. Lv et al. [29]
output as the inference engine. The fuzzy logic planner flow chart
proposed a bio-stimulus neural network robot path planning
is shown in Figure 3. Gao et al. [23] combined fuzzy control with
method that can realize self-feedback. Zhu et al. [30] introduced the
virtual sub-targets to simulate human behavior in an unknown
underwater environment and proposed a three-dimensional
environment. The method of determining sub-target points in the
dynamic path planning and safe obstacle avoidance method based
detection domain was proposed, and a robot behavior controller
on the bio-inspired model of AUV, but did not consider the
based on fuzzy decision control was designed. Wu et al. [24] based
influence of ocean currents, which are important factors in the
on the fuzzy logic method combined with the principle of repulsion
underwater environment. Then the team [18] corresponding each
in artificial potential field method, designed a fuzzy repulsion
neuron on the neural network with the location unit in the grid map,
algorithm to improve the local path planning algorithm. Sun et al.
and based on the distribution of the active output value of the
[25] combined fuzzy logic with particle swarm optimization to
neuron and combined with the direction reliability algorithm to
propose an effective path planning algorithm for obstacle
realize the path planning of the AUV. The method can be
avoidance for dynamic and static obstacles. Subsequently, Sun et
effectively used in obstacles and current environments.
al. [26] collected the environmental information from sonar on both
Top Layer
horizontal and vertical planes, obtained the acceleration value by f

using the fuzzy system with acceleration module, and compared the
two optimization methods to obtain the optimized fuzzy set, and Middle Layer
found the optimal path. Xiang et al. [27] reduced the design and f f f f
implementation cost of complex dynamic controllers based on the
robust fuzzy control scheme proposed by the current three-
dimensional path algorithm. For the problem that the AUV
underwater navigation path is susceptible to ocean current and sea
wave interference, Li et al. [28] proposed a fuzzy optimization X f Y f Z f

algorithm based on adrc, which introduced adrc to deal with current

and wave disturbances and improve the adaptability of AUV to Bottom Layer

Marine environment. Through simulation and experiment, it can

achieve better control effect in complex sea conditions. Figure 4: Neural Network

4.4 A* Algorithm (A*)

The basis of the A* algorithm is the Dijkstra algorithm, which is a
relatively mature algorithm for minimizing computational paths
and a heuristic algorithm for many other problems. The basic idea
is to determine whether to open or close the list to get the shortest

485
path in the area grid by using the minimum path cost. Therefore, different application scenarios plays an important role in the
another advantage of the A* algorithm is that it can cope with application of AUV path planning.
complex marine environments and can be interrupted and restored
2. Hybrid algorithm. Global path planning and local path
in time during the planning process. Lefebvre et al. [31] added a
planning have their own shortcomings in the application, and
layering technique based on the traditional A* algorithm to ensure
no single algorithm can solve the complex problems in
the computational speed under the premise of minimizing the path.
practical application. However, the integration of multiple
The simulation results show the reliability of the algorithm. In
algorithms can complement each other and produce better
complex submarine environments, especially when ocean currents
path planning methods by learning from each other.
are faster, the A* algorithm can be used for AUV path planning.
The research by Sheng et al. [32] shows that combining the A* 3. Multi-autonomous underwater vehicles cooperation mode.
algorithm with the horizontal set algorithm of ocean current can Through the cooperation and coordination between robots,
reduce the influence of ocean current on AUV path planning. MAUV can not only realize the individual functions but also
Zhang et al. [33] designed the optimized A* algorithm, first reflect more intelligent behaviors in the interaction. MAUV
constructed the model according to the ocean current characteristics, has a good effect on cooperative observation, search,
and then analyzed the speed and force information returned by the information transmission and mechanical operation of
AUV sensor to determine the conditions of the A* algorithm, and unknown environment. However, the path planning of
finally optimized the path line. The comparison of the four MAUV is more complicated, and the difficulty of planning to
algorithms is shown in Table 2. avoid obstacles under interaction is also increased. Therefore,
MAUV has become an important research and development
Table 2: Comparison of the above algorithms direction for improving the working ability of AUV.

advantages disadvantages
6 Conclusion
Simple, security, speed, difficult to design
Autonomous underwater vehicles path planning research is one of
ease of implementation potential field function,
APF the key intelligent robot research under complicated environment,
Insufficient dynamic
this paper summarizes the AUV path planning technology,
simple design, easy to Fuzzy rules are difficult to analyzes the path planning of AUV mission objectives, and then
apply, good robustness establish from two aspects of global path planning and local path planning
FLA combed the various algorithm's working principle, key
Good robustness, Poor generalization technologies and research status. Finally, the development trend of
powerful learning performance, slow AUV path planning is pointed out. With the development of science
ANN and technology, AUV path planning will cause more research hot-
processing speed
spots in the field of autonomous exploration of the ocean.
handle different Slow processing speed
difficulty terrain, be ACKNOWLEDGMENTS
A*
interrupted and resumed
This work is funded by National Key Research and Development
lately
Program of China (Grant No. 2018YFB1307900), Key Research
Program of Shanghai Science and Technology Commission (Grant
No. 16030501200), National Natural Science Foundation of China
5 Future Development Trend of Path Planning (Grant No. 51805314) and Shanghai University of Engineering and
Method Science (Grant No. E3-0501-18-01002). The Robot Functional
Materials Preparation Laboratory in Shanghai University of
With the rapid development of science and technology, the research
Engineering Science is also gratefully acknowledged.
on AUV path planning technology continues to deepen. Computers
and sensors related to path planning have gradually improved.
Various underwater robots applied in ocean exploration have
entered the application stage, but there are still a lot of REFERENCES
[1] F. M. Zhang, G. Marani, R. N. Smith and H. T. Choi(2015). Future Trends in
improvements to be made in the design of specific path planning Marine Robotics. Ieee Robotics & Automation Magazine, 22(1), 14-122.
methods. According to the current development situation, the [2] M. Ataei and A. Yousefi-Koma(2015). Three-dimensional optimal path planning
following directions of underwater robot path planning are worthy for waypoint guidance of an autonomous underwater vehicle. Robotics and
Autonomous Systems, 67, 23-32.
of further study: [3] S. Mahmoudzadeh, D. Powers, K. Sammut, A. Lammas and A. M.
1. Improve the existing algorithm. In the process of the Yazdani(2015). Optimal route planning with prioritized task scheduling for AUV
missions. 2015 IEEE International Symposium on Robotics and Intelligent
theoretical algorithm entering the practical application, each Sensors, 7-14.
algorithm has its advantages and disadvantages, especially [4] S. M. Zadeh, D. M. W. Powers, K. Sammut and A. M. Yazdani(2018). A novel
versatile architecture for autonomous underwater vehicle's motion planning and
the limitations of the algorithm itself. Therefore, the task assignment. Soft Computing, 22(5), 1687-1710.
improvement of algorithm characteristics according to

486
[5] J. A. Cao, Y. Li, S. Q. Zhao, X. S. Bi and Ieee(2016). Genetic-Algorithm-based [31] N. Lefebvre, I. Schjolberg and I. B. Utne(2016). Integration of risk in hierarchical
Global Path Planning for AUV. Proceedings of 2016 9th International path planning of underwater vehicles. Ifac Papersonline, 49(23), 226-231.
Symposium on Computational Intelligence and Design. Ieee, New York, 79-82 [32] L. Sheng, Z. M. Qiu, H. Li and Ieee(2018). A Survey on Route Planning Methods
[6] X. Pan, X. Wu, X. Hou and Y. Feng(2017). Global path planning based on of AUV Considering Influence of Ocean Current. 2018 Ieee 4th International
genetic-ant hybrid algorithm for AUV. Journal of Huazhong University of Conference on Control Science and Systems Engineering. Ieee, New York.
Science and Technology (Natural Science Edition), 45(5), 45-9, 76. [33] H. H. Zhang, L. M. Gong, T. Chen, L. Wang, X. Zhang and Ieee(2016). Global
[7] P. Yao and S. Zhao(2018). Three-Dimensional Path Planning for AUV Based on Path Planning Methods of UUV in Coastal Environment. 2016 Ieee International
Interfered Fluid Dynamical System Under Ocean Current (June 2018). Ieee Conference on Mechatronics and Automation. Ieee, New York.
Access, 6, 42904-42916.
[8] S. Perez-Carabaza, E. Besada-Portas, J. A. Lopez-Orozco and J. M. de la
Cruz(2018). Ant colony optimization for multi-UAV minimum time search in
uncertain domains. Applied Soft Computing, 62, 789-806.
[9] H. J. Hong and W. Xiong(2009). Research on global path planning based on ant
colony optimization for AUV. Journal of Marine Science and Application, 8(1),
58-64.
[10] C. Lin, H. Wang, J. Yuan and M. Fu(2018). An online path planning method
based on hybrid quantum ant colony optimization for AUV. International Journal
of Robotics & Automation, 33(4), 435-444.
[11] L. Y. Dong and H. L. Xu(2017). AUV Path Planning Based on Improved Ant
Colony Algorithm. Automation & Instrumentation, 32(3), 1-4.
[12] Y. J. Heo, W. K. Chung and Ieee(2013). RRT-based Path Planning with
Kinematic Constraints of AUV in Underwater Structured Environment. 2013
10th International Conference on Ubiquitous Robots and Ambient Intelligence,
523-525.
[13] J. D. Hernandez, K. Istenic, N. Gracias, N. Palomeras, R. Campos, E. Vidal, R.
Garcia and M. Carreras(2016). Autonomous Underwater Navigation and Optical
Mapping in Unknown Natural Environments. Sensors, 16(8), 27.
[14] I. B. Jeong, S. J. Lee and J.-H. Kim(2019). Quick-RRT*: Triangular inequality-
based implementation of RRT* with improved initial solution and convergence
rate. Expert Systems with Applications, 123, 82-90.
[15] J. T. Kim, J. H. Li, M. J. Lee, H. J. Kang, W. S. Lee and Ieee(2014). Path Planning
for 3D Coastal Navigation of Underwater Structures. 2014 Oceans - St. John's.
Ieee, New York.
[16] N. Chao, Y. K. Liu, H. Xia, M. J. Peng and A. Ayodeji(2019). DL-RRT*
algorithm for least dose path Re-planning in dynamic radioactive environments.
Nuclear Engineering and Technology, 51(3), 825-836.
[17] Y. Li, F. Zhang, D. Xu and J. Dai(2017). Liveness-Based RRT Algorithm for
Autonomous Underwater Vehicles Motion Planning. Journal of Advanced
Transportation.
[18] D. Q. Zhu, Y. Liu, B. Sun, Q. Q. Liu(2019). Autonomous underwater vehicles
path planning based on autonomous inspired Glasius bio-inspired neural
network algorithm .Control Theory & Applications, 36(02), 183-191.
[19] H. X. Liu, X. Q. Bian, Z. Qin, H. J. Wang (2007). Research on Obstacle
Avoidance Planning for AUV Based on Forward Looking Sonar. Journal of
System Simulation, (24), 5672-5674+5679.
[20] C. L. Cheng, D. Q. Zhu, B. Sun, Z. Z. Chu, J. D. Nie, S. Zhang and Ieee(2015).
Path Planning for Autonomous Underwater Vehicle Based on Artificial Potential
Field and Velocity Synthesis. 2015 Ieee 28th Canadian Conference on Electrical
and Computer Engineering. Ieee, New York.
[21] B. Das, B. Subudhi and B. B. Pati(2016). Co-operative control of a team of
autonomous underwater vehicles in an obstacle-rich environment. Journal of
Marine Engineering and Technology, 15(3), 135-151.
[22] C. Zhang and Iop(2018). Path Planning for Robot based on Chaotic Artificial
Potential Field Method. 4th International Conference on Advanced Engineering
and Technology. Iop Publishing Ltd, Bristol.
[23] J. X. Gao, X. G. Ruan, S. C. Ma (2014). Mobile Robot Active Routing Navigation
Based on Fuzzy Behavior Decision-making . Journal of Beijing University of
Technology, 40(09), 1308-1314.
[24] X. G. Wu, S. P. Gong, P. Xie(2014). Local Path Planning Based on Improved
Artificial Potential Field Using Fuzzy Repulsion Force for Robot. Journal of
Electrical Engineering, 02(3), 48-60.
[25] B. Sun, D. Q. Zhu and Y. Y. Yang(2013) .Fuzzy path planning of autonomous
underwater vehicle based on particle swarm optimization. Chinese High
Tenchnology letters, 23(12), 1284-1291.
[26] B. Sun, D. Q. Zhu and S. X. Yang(2018). An Optimized Fuzzy Control Algorithm
for Three-Dimensional AUV Path Planning. International Journal of Fuzzy
Systems, 20(2), 597-610.
[27] X. B. Xiang, C. Y. Yu and Q. Zhang(2017). Robust fuzzy 3D path following for
autonomous underwater vehicle subject to uncertainties. Computers &
Operations Research, 84, 165-177.
[28] H. J. Li, B. He, Q. Q. Yin, X. K. Mu, J. M. Zhang, J. H. Wan, D. R. Wang and
Y. Shen(2019). Fuzzy Optimized MFAC Based on ADRC in AUV Heading
Control. Electronics, 8(6), 32.
[29] Z. Y. Lv, J. T. Cao (2014). Robot path planning method based on biologically
incentives neural network with self-feedback. Computer Engineering and
Applications . Computer Engineering and Applications, 50(16), 255-258+270.
[30] D. Q. Zhu, B. Sun and L. Li(2015). Algorithm for AUV's 3-D path planning and
safe obstacle avoidance based on biological inspired model. Control and Decision,
30(5), 798-806.

487