Elsevier

Computer Communications

Volume 162, 1 October 2020, Pages 196-203
Computer Communications

Optimized multi-UAV cooperative path planning under the complex confrontation environment

https://doi.org/10.1016/j.comcom.2020.04.050Get rights and content

Abstract

As an emerging technology, multi-UAV collaboration is widely used in military and civil applications, including regional surveillance, remote sensing, target strike, etc. As a key step in the implementation of multi-UAV cooperative missions, path planning aims to generate near-optimal paths that satisfy certain constraints, ensure that each UAV can reach the mission area quickly and reduce the probability of being captured and destroyed by the antagonism side. In this paper, we design an optimized multi-UAV cooperative path planning method under the complex confrontation environment. Firstly, the threat model is designed based on the actual situation. Combining the threat and fuel consumption criteria, under the constraints of time and space, a multi-constraint objective optimization model is established. Following this, an improved grey wolf optimizer algorithm is used to solve the optimization model. Based on the characteristics of the multi-UAV cooperative path planning, the algorithm is improved in three aspects: population initialization, decay factor updating, and individual position updating. The simulation results demonstrate that the proposed algorithm is effective in generating paths for multi-UAV cooperative path planning and has the advantages of a lower path cost and faster convergence speed as compared to the other algorithms tested in this work.

Introduction

In modern warfare, a swarm of UAVs having advantages such as high efficiency, reliability, and adaptability is widely used to carry out a variety of complex military tasks. Path planning is the key to ensuring the successful completion of UAV missions. Multi-UAV cooperative path planning refers to planning paths that can bypass threat areas, satisfy various constraints and cope with cooperative relationships. This is a NP-hard problem, one of the difficulties present in the field of multi-UAV cooperative technologies [1], [2], [3].

The goal of the multi-UAV cooperative path planning is to search for a feasible flight path for each UAV from its starting point to the finishing point under the conditions of minimizing the overall flight cost, satisfying the constraints of the distance between the UAVs, arrival time, and the performance of the UAVs. Presently, the frequently-used research methods are: multi-UAV path planning methods developed from single UAV path planning, such as heuristic method, graph planning method, and artificial potential field method; and swarm intelligence algorithms, such as genetic algorithm (GA), particle swarm optimization (PSO), ant colony (AC) algorithm, and so on.

A* algorithm is the most classic heuristic method for solving the shortest path. The A* uses heuristic information to evaluate the optimal onward node location of the next step, and continuously explores the unknown space and find the optimal path. Both A* and sparse A* algorithms [4], [5] can effectively handle various complicated constraint conditions in environment. Both algorithms have been adopted to solve the single-UAV or multi-UAV collaborative path planning problems under 2D environment. Afterward, researchers extended the sparse A* algorithm to 3D environment [6]. This type of algorithms can ensure to provide the optimal solution for the problem. However, they have high time or space complexity. Therefore, such algorithms are likely to lead to a long run time when applied to complicated path planning in a wide range environment. For the rather large multi-UAV collaborative path planning problem, one of the frequently-used planning methods is using the probability graph to simplify the space representation. This method first reduces the planning space to a simple and safe road network diagram according to the conditions such as system constraints and threat avoiding, and then obtains the optimal flight path on the road network diagram. The typical probability graph algorithm is the Voronoi graph [7]. Although, the Voronoi graph can easily be constructed on 2D surface, it is extremely difficult in 3D space. In addition, the artificial potential field is also an effective method of multi-UAV collaborative path planning. In planning space, it imitates the attraction and repulsion forces to influence the UAV motion, and pulls the UAV to advance along the direction of the fastest decline of the potential function. At that time, the UAV moves along the direction of the resultant force, and can reach the target location point avoiding the barrier and threat [8]. The advantages of artificial potential field method are high speed, high safety, and being fit for collaborative planning; while the disadvantage is that it perhaps results in algorithm stagnating and further planning failure as the repulsion force has the equal stationary point with the attraction force.

The swarm intelligence algorithm is also suitable for solving multi-UAV collaborative path planning problems and is the most widely researched method. It has the advantages of rapid planning speed, good parallelism, easy cooperativity. Earlier, Nokols adopted the genetic algorithm and differential evolution to plane the single-UAV and multi-UAV collaborative flight path [9], [10]. Afterward, a large number of path planning studies were conducted on the basis of the improved evolution algorithms, such as the multi-parallel evolution algorithm [11], joint genetic algorithm and ant colony optimization [12], and a method of combining immunologic mechanism with the genetic algorithm [13]. These approaches have proved that using the evolution mechanism of species and parallel optimization method can rapidly search out the optimal location of solution space, thus obtaining multiple feasible collaborative paths. Additionally, evolutionary theory is not restricted to the space structure, so it is suitable for the path planning problem in 3D and even multi-dimensional space. Since the particle swarm optimization (PSO) has more advantages and higher efficiency than the genetic algorithm, the use of PSO to perform multi-UAV path planning has been increased in recent years [14], [15]. In addition, the number of studies using intelligent algorithms, such as continuous ant colony optimization [16], artificial immune algorithm [17], and neural network method [18], to perform multi-UAV path planning has also increased.

The swarm intelligence algorithms possess a promising future in multi-UAV cooperative path planning because of their preponderance in solving the multi-constraints optimization problem. Grey wolf optimizer (GWO) [19] is a new type of bio-inspired optimization algorithm. This algorithm simulates the grey wolf’s predation strategy and hierarchy in nature. The advantages of the GWO algorithm have a simple structure, flexibility, fewer parameters and the ability to avoid falling into a local minimum to some extent [20]. This paper proposes an improved GWO algorithm, which is effective in generating paths for multi-UAV cooperative path planning.

The main contributions of our work are as follows.

(1) Time and space constraints are considered to simulate the real complex confrontation environment.

(2) A multi-constraint objective optimization model for multi-UAV cooperative path planning is established.

(3) An improved grey wolf optimizer algorithm is designed for solving the objective model.

The remainder of this paper is organized as follows. In Section 2, we describe the problem definitions. In Section 3, we present the details of system model. In Section 4, we highlight an improved grey wolf optimizer algorithm used to solve the optimization model. In Section 5, we evaluate the performance of the proposed algorithm through simulation experiments. Finally, we conclude this paper in Section 6.

Section snippets

Problem formulation

Suppose that in a mission, there are n (n>1) UAVs taking off from different areas, respectively, going to a mission area. Prior to the execution of the mission, constraints such as radar threat, missile threat, and terrain threat should be taken into consideration while planning the flight path for each UAV. In order to improve the success rate of strike missions, the UAVs need to reach the mission area simultaneously. Further, in consideration of the safety of each UAV, a certain safety

Threat modeling

Each UAV must avoid obstacles and threats during its flight in order to meet the safety requirements. Thus, it is necessary to establish the threat model first. There are two types of threat modeling methods: threat probability and threat boundary. The former is suitable for ground radar and missile threats while the latter is suitable for specific objects such as buildings and terrain. Details of the two methods are given below.

(1) Ground radar threat

A radar can find a UAV in its mission space

Grey wolf optimizer

GWO was inspired by the lead and hunt behavior in a pack of grey wolves. Fig. 6 is a hierarchical model of the grey wolf group, where α is called the leader grey wolf who leads the entire grey wolf pack. β is the subordinate grey wolf that assists the leader grey wolf α in making decisions. At the same time, β can command the subordinate ordinary grey wolf δ. ω are the lowest-level grey wolves who help the upper-level grey wolves to make decisions and hunt.

The encircling mathematical model of

Feasibility test for multi-UAV cooperative path planning

Simulation 1: Multi-UAV cooperative path planning under probability threats.

The overall space was set to 200 km × 200 km × 12 km. In this space there were eight threat sources, four ground radars, and four missiles. Four identical UAVs were used. The starting co-ordinates of these 4 UAVs were [149.2537, 23.5880, 7.9200] km, [185.7380, 86.0465, 7.9200] km, [26.5340, 67.1096, 7.9200] km and [149.2537, 23.5880, 7.9200] km respectively. The coordinates of the target point were [56.0531, 168.1063,

Conclusion

In this paper, we propose an algorithm for multi-UAV cooperative path planning. Taking into consideration the threat environment which is encountered in restricted areas and by air defense forces, the models of ground radar threat, missile threat, and terrain threat have been constructed. During mission operation, the goal is to minimize the cost of fuel consumption and threat under multiple constraints. In order to solve this NP-hard problem, we design an improved GWO algorithm, which improves

CRediT authorship contribution statement

Cheng Xu: Writing - original draft. Ming Xu: Methodology, Software. Chanjuan Yin: Writing - review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

References (20)

There are more references available in the full text version of this article.

Cited by (56)

View all citing articles on Scopus
View full text