CN115031736A

CN115031736A - Multi-unmanned aerial vehicle area coverage track planning method and system

Info

Publication number: CN115031736A
Application number: CN202210578008.0A
Authority: CN
Inventors: 董希旺; 石明慧; 化永朝; 任章; 于江龙; 吕金虎
Original assignee: Beihang University
Current assignee: Beihang University
Priority date: 2022-05-25
Filing date: 2022-05-25
Publication date: 2022-09-09

Abstract

The invention relates to a planning method and a system for multi-unmanned aerial vehicle area coverage flight paths, and belongs to the technical field of multi-unmanned aerial vehicle collaborative area search flight path planning. The method comprises the steps of dividing a task area according to the search total capacity of each aggregation point to obtain a search area of each aggregation point, dividing the search area of each aggregation point to obtain a flight area of each unmanned aerial vehicle included in each aggregation point, dividing the flight area by using an area decomposition method to obtain a plurality of units to be merged, recombining the units to be merged by using a dynamic planning method to obtain a plurality of regions after recombination, planning a flight path in each region after recombination to obtain a flight path of each unmanned aerial vehicle, and accordingly completing a region coverage flight path planning process of cooperation of multiple unmanned aerial vehicles.

Description

Multi-unmanned aerial vehicle area coverage track planning method and system

Technical Field

The invention relates to the technical field of multi-unmanned aerial vehicle collaborative area search flight path planning, in particular to a multi-unmanned aerial vehicle area coverage flight path planning method and system based on geometric area segmentation.

Background

Along with the continuous development of science and technology, many unmanned aerial vehicles in the future need adopt the cluster to fight and accomplish the combat mission to improve the ability to resist risk. When the problem of searching the collaborative area of the cluster of the multiple unmanned aerial vehicles is faced, not only the task allocation for different unmanned aerial vehicles needs to be reasonably carried out according to various constraint conditions, but also the unmanned aerial vehicles are required to be mutually matched and collaboratively operated, so that the overall efficiency is maximized.

On the problem of multi-unmanned aerial vehicle collaborative area search, the most common mode is to reasonably divide the area, and the method is efficient, clear and easy to realize. However, the existing unmanned aerial vehicle search flight path planning method is mainly carried out for a single unmanned aerial vehicle, research results in the field of multi-unmanned aerial vehicle collaborative search planning are still thin, the influence of the initial positions of all unmanned aerial vehicles on collaborative search is not considered, and meanwhile, collaborative planning schemes for unmanned aerial vehicles with different search capabilities are lacked, so that the unmanned aerial vehicle search flight path planning method does not have high availability in an actual scene.

Disclosure of Invention

The invention aims to provide a method and a system for planning a multi-unmanned aerial vehicle area coverage track, which can complete the planning process of the multi-unmanned aerial vehicle area coverage track.

In order to achieve the purpose, the invention provides the following scheme:

a multi-UAV area coverage track planning method comprises the following steps:

dividing task areas according to the total search capacity of each rendezvous point to obtain a search area of each rendezvous point; the gathering point comprises a plurality of unmanned aerial vehicles;

for the search area of each gathering point, dividing the search area of the gathering point to obtain the flight area of each unmanned aerial vehicle included in the gathering point;

for the flight area of each unmanned aerial vehicle, dividing the flight area by using an area decomposition method to obtain a plurality of units to be combined; recombining the units to be combined by using a dynamic programming method to obtain a plurality of recombined regions; and planning a flight path in each recombined region to obtain the unmanned flight path.

A multi-drone area coverage track planning system, the track planning system comprising:

the search area dividing module is used for dividing the task area according to the total search capacity of each aggregation point to obtain the search area of each aggregation point; the gathering point comprises a plurality of unmanned aerial vehicles;

the flight area dividing module is used for dividing the search area of each aggregation point to obtain the flight area of each unmanned aerial vehicle included in each aggregation point;

the flight path planning module is used for dividing the flight area of each unmanned aerial vehicle by using an area decomposition method to obtain a plurality of units to be combined; recombining the units to be combined by using a dynamic programming method to obtain a plurality of recombined regions; and planning a flight path in each recombined region to obtain the flight path of the unmanned aerial vehicle.

According to the specific embodiment provided by the invention, the invention discloses the following technical effects:

the invention provides a planning method and a system for multi-unmanned aerial vehicle area coverage tracks, which are used for dividing a task area according to the total searching capacity of each aggregation point to obtain a searching area of each aggregation point, dividing the searching area of each aggregation point to obtain a flight area of each unmanned aerial vehicle included in each aggregation point, finally dividing the flight area by using an area decomposition method to obtain a plurality of units to be combined, recombining the units to be combined by using a dynamic planning method to obtain a plurality of regions after recombination, and planning the tracks in each region after recombination to obtain the tracks of each unmanned aerial vehicle, thereby completing the planning process for multi-unmanned aerial vehicle collaborative area coverage tracks and having guiding significance for realizing the collaborative work of the multi-unmanned aerial vehicles.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required to be used in the embodiments will be briefly described below, and it is obvious that the drawings in the following description are only some embodiments of the present invention, and it is obvious for those skilled in the art that other drawings can be obtained according to the drawings without creative efforts.

Fig. 1 is a flowchart of a method of a flight path planning method according to embodiment 1 of the present invention;

fig. 2 is a schematic diagram of geometric partitioning of a task area when two rendezvous points are adopted according to embodiment 1 of the present invention;

fig. 3 is a schematic diagram of geometric partitioning of a task area when a plurality of rendezvous points are adopted according to embodiment 1 of the present invention;

fig. 4 is a schematic diagram of geometric partitioning of a task area when three aggregation points with different proportions are adopted according to embodiment 1 of the present invention;

FIG. 5 is a schematic partitioning diagram of an example of trapezoidal decomposition provided in embodiment 1 of the present invention;

FIG. 6 is a unit connection diagram of an example of trapezoidal decomposition provided in embodiment 1 of the present invention;

FIG. 7 is a schematic diagram of a rotated division of an exemplary trapezoidal decomposition provided in embodiment 1 of the present invention;

FIG. 8 is a schematic view of a minimum convex polygon of an example of trapezoidal decomposition provided in embodiment 1 of the present invention;

FIG. 9 is a schematic diagram showing the division of a minimum convex polygon of an example of trapezoidal decomposition provided in embodiment 1 of the present invention;

FIG. 10 is a re-organization diagram of a minimum convex polygon of an example of a trapezoidal decomposition provided in embodiment 1 of the present invention;

fig. 11 is a schematic process diagram of a dynamic programming method according to embodiment 1 of the present invention;

fig. 12 is a schematic diagram of division of an example of a polygonal area provided in embodiment 1 of the present invention;

FIG. 13 is a diagram of cell connections for an example of a polygonal area provided in embodiment 1 of the present invention;

fig. 14 is a schematic diagram of a search network of the dynamic programming method according to embodiment 1 of the present invention;

FIG. 15 is a schematic diagram of a reorganization of an example of a polygonal area provided in embodiment 1 of the present invention;

FIG. 16 is a schematic diagram of a linear scanning track provided in embodiment 1 of the present invention;

FIG. 17 is a schematic view of a turn track provided in embodiment 1 of the present invention;

fig. 18 is a schematic diagram of an area to be searched in a simulation process provided in embodiment 1 of the present invention;

fig. 19 is a schematic view of trapezoidal decomposition of an outsourcing region of a region to be searched in a simulation process according to embodiment 1 of the present invention;

fig. 20 is a schematic diagram of reorganization of a region to be searched in a simulation process according to embodiment 1 of the present invention;

fig. 21 is a schematic diagram of a flight path planning of an area to be searched in a simulation process according to embodiment 1 of the present invention;

fig. 22 is a schematic diagram of partitioning a task area of a dual-machine single rendezvous point in a simulation process according to embodiment 1 of the present invention;

fig. 23 is a schematic diagram of a dual-computer single-aggregation-point task area search flight path in a simulation process according to embodiment 1 of the present invention;

fig. 24 is a schematic diagram illustrating division of a dual rendezvous point task area in a simulation process according to embodiment 1 of the present invention;

fig. 25 is a simplified diagram of partitioning a dual rendezvous point task area in a simulation process according to embodiment 1 of the present invention;

fig. 26 is a diagram illustrating a dual rendezvous point task area search track in a simulation process according to embodiment 1 of the present invention;

fig. 27 is a system block diagram of a track planning system according to embodiment 2 of the present invention.

Detailed Description

The technical solutions in the embodiments of the present invention will be described clearly and completely with reference to the accompanying drawings in the embodiments of the present invention, and it is obvious that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be obtained by a person skilled in the art without making any creative effort based on the embodiments in the present invention, belong to the protection scope of the present invention.

In order to make the aforementioned objects, features and advantages of the present invention comprehensible, embodiments accompanied with figures are described in further detail below.

Example 1:

at present, the task planning problem for area coverage search mainly includes the following planning methods: (1) the area distribution of the multiple unmanned aerial vehicles is completed by dividing the polygon area in proportion, and the problem of flight path planning of a single area is solved in the distributed area through a parallel search method. (2) The distributed online heuristic strategy is utilized to solve the problem of cooperative coverage of multiple unmanned aerial vehicles, each unmanned aerial vehicle maintains a probability grid map in the form of a local storage matrix, and two evaluation functions and related technical strategies are designed, so that the unmanned aerial vehicles can make state transition or area transition decisions in an online self-organizing manner. Simulation results show that the strategy has high search efficiency, good robustness and fault tolerance. (3) An optimal full-coverage air self-organizing network construction scheme is provided, and when the maximum network coverage is obtained, the minimum number of unmanned aerial vehicles are deployed to indispensable places. Simulation results show that the scheme is superior to several peer-to-peer algorithms in terms of traversal time and redundant access rate. (4) The multi-unmanned aerial vehicle regional coverage model based on central Voronoi configuration is provided, optimized deployment of searching coverage areas and searching time is achieved, and autonomous cooperative control of unmanned aerial vehicles is achieved based on a cooperative searching strategy of map information updating and fusion. (5) The method comprises the steps of establishing a multi-unmanned-aerial-vehicle collaborative flight model under primary and secondary disasters by fully considering factors such as terrain, inspection time, search area coverage range and the number of unmanned aerial vehicles, obtaining a route plan by combining a fastest descent method and an A-algorithm, completing maximum coverage search by using the unmanned aerial vehicle within specified time, and obtaining a global search path under the secondary disasters by using a tabu search algorithm. (6) A coverage search algorithm based on basic behavior combination and environment mapping is provided, an unmanned aerial vehicle random shape coverage search simulation model is established by using a discrete map, and environment and task changes are updated in time. The comparison between simulation analysis and dynamic planning shows that the method has expandability and can practically change the search strategy. However, the research results are still thin and simple throughout the current collaborative search planning of multiple unmanned aerial vehicles, and the research results still stay in the basic theoretical level. The embodiment integrates the advantages of the methods, expands the advantages to collaborative planning, integrates various basic algorithms, and realizes the optimal flight path planning through mixed application.

The embodiment is used for providing a multi-unmanned aerial vehicle area coverage track planning method, reasonable basic assumption is carried out on the multi-unmanned aerial vehicle collaborative search problem, the unmanned aerial vehicle adopts a model based on mass points, the position of the unmanned aerial vehicle and the position of a task area can be described by coordinates, the sensor coverage area of each unmanned aerial vehicle is clear and can be different, the detection size is far smaller than the area of the task area, and the boundary of the task area can be described by a straight line to be a polygon. As shown in fig. 1, the flight path planning method includes:

s1: dividing the task area according to the total searching capacity of each aggregation point to obtain the searching area of each aggregation point; the gathering point comprises a plurality of unmanned aerial vehicles;

the area coverage search can be divided into two types, i.e. a multi-machine single-aggregation-point problem and a multi-machine multi-aggregation-point problem. For the problem of multiple single rendezvous points, the flight area of each unmanned aerial vehicle is determined by S2 by directly taking the task area as the search area of the single rendezvous point. For the problem of multiple sets of points, the problem can be decomposed into two sub-problems that each set point only comprises one equivalent unmanned aerial vehicle and one set point only comprises multiple unmanned aerial vehicles, the sub-problem that each set point only comprises one equivalent unmanned aerial vehicle is solved by using S1, and the search area of each set point is determined; the sub-problem that one aggregation point comprises a plurality of unmanned aerial vehicles is solved by S2, and the flight area of each unmanned aerial vehicle is determined.

In this embodiment, the irregular region segmentation method based on geometric segmentation is adopted to segment the task region, and then S1 may include:

(1) taking the sum of the searching capacities of all unmanned aerial vehicles included in the rendezvous point as the searching total capacity of the rendezvous point to obtain the searching total capacity of each rendezvous point;

each rendezvous point all has a plurality of unmanned aerial vehicles, to each rendezvous point, can be equivalent to an equivalent unmanned aerial vehicle with all unmanned aerial vehicles that this rendezvous point includes to the total ability of searching that this rendezvous point is represented in the sum of all unmanned aerial vehicle's that this rendezvous point includes search ability. The search capability of the drone is determined by the coverage of the sensor mounted on the drone, and is known in advance.

(2) Dividing the task area according to the proportion of the total searching capacity of each rendezvous point to obtain the searching area of each rendezvous point; the ratio of the total search capability of each rendezvous point is the same as the ratio of the area of the search area of each rendezvous point.

And dividing the task area into the aggregation points according to the ratio of the searching total capacity of each aggregation point, and dividing again in the searching area in which each aggregation point is responsible for searching according to a method of multiple single aggregation points to determine the flight area of each unmanned aerial vehicle.

Specifically, tasks such as reconnaissance and rescue need to be divided as much as possible according to the searching capability of each unmanned aerial vehicle, so that the working time of each unmanned aerial vehicle is consistent and minimum as much as possible. When the speed of the unmanned aerial vehicle is v and the time is t, the voyage s is as follows: s-v t. In many unmanned aerial vehicle clusters, every unmanned aerial vehicle's sensor has different detection range, and when flying the same distance, the size of its detection range has been decided to the performance of sensor. Assuming that the detection radius of a certain unmanned aerial vehicle is r, the detection width is 2r, and the area C of the task region can be detected _i Comprises the following steps: c _i S2 r. The searching capability described in this embodiment is the area of the task area detectable by the unmanned aerial vehicle in unit time.

Area C of total coverage search of all drones _sum Comprises the following steps:

wherein m is the total number of the unmanned aerial vehicles, r _i The detection radius of the ith unmanned aerial vehicle.

The analysis shows that the detection area of each airplane is in direct proportion to the sensor searching capability when the multi-airplane cooperative coverage search is carried out. Suppose there are m airplanes U ₁ ,U ₂ ,···,U _m When a task area is allocated, normalization processing of searching capacity is required to be carried out firstly, and the normalization processing is used as a performance index of the airplane and is defined as K ₁ ,K ₂ ,···,K _m Wherein:

so that the target area is S _R The task area R of (2) needs to be decomposed into m sub-task areas R ₁ ,R ₂ ,···,R _m The area of which is denoted S _R1 ,S _R2 ,···,S _Rm And satisfies the following conditions:

after the areas are divided according to the proportion of the searching capacity of each unmanned aerial vehicle, the sub-areas are correspondingly allocated to the designated unmanned aerial vehicle one by one to execute the searching task of the area, so that the task target of cooperatively completing the area coverage searching by a plurality of aircrafts can be realized. The ratio of the unmanned aerial vehicle detection areas is the ratio of the search capability.

After the total search capability of each rendezvous point is obtained through calculation, the task area can be divided according to the principle, and the search area corresponding to each rendezvous point is obtained.

As shown in fig. 2, the equivalent drones of the two rendezvous points are respectively located at the five-pointed star in fig. 2, the dashed lines represent the connecting line of the two drones and the vertical bisector thereof, and when viewed from the visual distance, the sub-areas 1-5 are closer to the UAV1, and the sub-area 6 is closer to the UAV2, but when the sub-areas are divided according to the search capability ratio, if the search performances are the same, the sub-areas 1-3 are classified as UAV1, and the sub-areas 4-6 are classified as UAV 2. The ratio of the two functions is changed in turn, the dividing line is the intersection line of each subarea in the figure, and through mathematical analysis, each intersection line is a hyperbolic line with the focus at the two airplanes, and the distance difference between the point on the hyperbolic line and the two airplanes is the same.

As shown in fig. 3, when the number of the collection points is greater than three, the task segmentation may generate a graph in a Voronoi diagram method, where the segmentation line is a hyperbola, and the Voronoi diagram is also called a thiessen polygon or a Dirichlet diagram and is composed of a set of continuous polygons composed of perpendicular bisectors connecting two adjacent point straight lines. For the multi-set point problem, the solution can be performed by an incremental method, and the principle is to use the perpendicular bisector of the three set points as the initial dividing line, and then iterate the distribution and moving the dividing line for modification for multiple times, as shown in fig. 4. When the area is divided, determining the intersection line of the area exceeding the performance ratio and the area smaller than the performance ratio as the dividing line to be adjusted, and adjusting by a fixed increment until the division is finished. Fig. 4 shows that three equivalent unmanned planes jointly search for one task area, a dot-dash line is an initial perpendicular bisector, and a final segmentation mode is obtained through incremental adjustment circulation, and the segmentation proportion of the task area of the three-rendezvous-point aircraft is (0.4, 0.2, 0.4).

After the search area responsible for each aggregation point is obtained, the problem of multi-machine single aggregation points needs to be solved next, the problem of multi-machine single aggregation points is regarded as a special case of multi-machine multi-aggregation points, namely aggregation points are overlapped, other conditions are unchanged, the area division method is completely the same as the previous method, only hyperbolas of the divided areas are degenerated into straight lines passing through the aggregation points, and the area ratio of the areas of the divided straight lines is equal to the ratio of the search capabilities of all machines.

S2: for the search area of each aggregation point, dividing the search area of each aggregation point to obtain the flight area of each unmanned aerial vehicle included in each aggregation point;

for the partitioning problem of the search area of the aggregation point, the search area is subjected to rasterization processing, and the search area is partitioned according to the principle that the grid point distributed to a certain unmanned aerial vehicle is as close as possible to the position of the unmanned aerial vehicle. Specifically, S2 may include:

(1) rasterizing the search area to obtain a rasterized area; the rasterized region includes a plurality of raster points;

(2) randomly distributing all grid points to the corresponding sub-sets of all the unmanned planes according to the proportion of the searching capacity of all the unmanned planes, wherein the proportion of the searching capacity of all the unmanned planes is the same as the proportion of the number of the grid points included in the corresponding sub-sets of all the unmanned planes;

(3) circularly exchanging the grid points in each subset according to the cost matrix until the objective function value after exchange is greater than or equal to the objective function value before exchange; the objective function value is the sum of the distance sums corresponding to the subsets, and the distance sum corresponding to each subset is the sum of the distance from each grid point included in the subset to the unmanned aerial vehicle corresponding to the subset.

The construction method of the cost matrix comprises the following steps: and calculating the distance from each grid point to each unmanned aerial vehicle, and constructing a cost matrix, wherein the element of the ith row and the jth column of the cost matrix is the distance from the ith unmanned aerial vehicle to the jth grid point.

Wherein, cyclically exchanging the grid points in each subset according to the cost matrix may include: and performing cyclic exchange in a pairwise exchange mode, calculating the cost difference of each grid point of the two selected subsets according to the cost matrix, taking the cost difference as a standard for judging whether the grid points in the subsets are exchanged, and performing cyclic exchange on the grid points in each subset according to the cost difference.

In the following, the principle that the set point includes two airplanes and the above-mentioned grid point assigned to a certain drone is as close as possible to the position of the drone is described by a formula, assuming that the set of grid scattered points of the search area is K, which includes a point { K } ₁ ，k ₂ ，k ₃ ，...，k _n }。K ₁ 、K ₂ Representing subsets assigned to two unmanned machines, n ₁ And n ₂ Indicating the number of points.

When the coverage search performance of the two drones is the same, then:

the final ideal region segmentation effect is as follows:

wherein, the first and the second end of the pipe are connected with each other,

as a subset K ₁ Each grid point and subset K of ₁ The sum of the distances between the corresponding unmanned aerial vehicles,

as a subset K ₂ Each grid point and subset K of ₂ The distance between the corresponding unmanned aerial vehicles is sum.

In this embodiment, the above allocation target is achieved by establishing a cost matrix, and when two drones allocate, the cost matrix is a 2 × n matrix, and element c _ij Representing the cost of drone i to grid point j. Thus, the cost difference Δ c of a single grid point _j Can be calculated from the following formula: Δ c _j ＝c _1j -c _2j . After the cost difference of each grid point is obtained through calculation, each grid point is randomly allocated to the sub-set K according to the searching capacity ratio of the two unmanned planes ₁ Or subset K ₂ And then circularly exchanging the grid points in the subset until the goal function value after exchange is larger than or equal to the goal function value before exchange.

S3: for the flight area of each unmanned aerial vehicle, dividing the flight area by using an area decomposition method to obtain a plurality of units to be combined; recombining the units to be combined by using a dynamic programming method to obtain a plurality of recombined regions; and planning a flight path in each recombined region to obtain the flight path of the unmanned aerial vehicle.

The method can divide the flight area by using a polygon area division method, if the flight area is a concave polygon, the polygon needs to be firstly decomposed into a plurality of convex polygons, the unmanned aerial vehicle motion planning is realized by decomposing one flight area into a group of geometric figures, and the centers of the geometric figures can be used for searching potential track points of the fastest track, so that the whole track planning process can be rapidly and efficiently operated as far as possible, and the calculation time is reduced. The trapezoidal decomposition is generally selected for the division of the flight area, because the trapezoidal shape is a more effective geometric shape, the cost is lower when S-shaped scanning track coverage is used, the turning number of the unmanned aerial vehicle is obviously less than that of a triangle for the generated long and thin trapezoidal shape, meanwhile, the trapezoidal decomposition is less than the units generated by the triangulation or approximate decomposition technology, the number of the units obtained by decomposition can be kept at a relatively lower limit, and the optimization of area recombination can be accelerated to a certain degree by the aid of the division mode.

Specifically, the present embodiment may perform trapezoidal decomposition of a polygon, and display vertical slice lines generated at each vertex, where each unit formed by each vertical slice line and a boundary of the polygon is a unit obtained after decomposition; a ladder decomposed cell connection graph may also be obtained, where each node represents a cell and each edge represents a cell adjacency. As shown in FIG. 5, which is a partitioning diagram of an example of trapezoidal decomposition, where the vertical slice lines intersecting the polygons define the vertices of all cells, using the trapezoidal decomposition method, a solution can be found in O (nlogn) time, where n is the number of vertices, meaning that a high-vertex polygon will have acceptable computation time. Another important output of the ladder decomposition is the cell connectivity graph, which will be used in the region reorganization stage to determine which cells can be combined, and the cell connectivity graph for the ladder decomposition example is shown in FIG. 6.

Since the slice is always vertical, the final decomposition of a particular polygon depends on its initial rotation, and if the polygon is rotated, decomposed, and then rotated back to its original orientation, there are an infinite number of decomposition possibilities, meaning that there may be different, possibly more desirable decompositions at other angles of rotation. The embodiment can find the optimal flight path only by using the longest axis of the unit and minimizing the number of turns, namely, the vertical slice lines defined by the embodiment are all perpendicular to the long axis of the flight area. As shown in fig. 7, it is a schematic diagram of a division after a trapezoidal decomposition example rotates by 90 degrees.

Each vertex of the polygon must be rotated around the polygon by the angle psi _p Assuming that P is a polygon with n vertices, the rotation of P is shown as follows, where P is _r Is a rotated polygon,. psi _p Is the angle of rotation.

Unlike ground based searches, air flight paths are generally not confined to the mission area (unless obstacles such as no-fly zones exist). They can fly over the non-mission area, directly transfer between multiple convex polygon unit areas, and thus can shorten the total flying cost by filling the outer area. This embodiment proposes a method of extended trapezoidal decomposition including potential external cells that may shorten the overall flight time, and the way to find and generate these external cells is to find the convex hull around the concave polygon. Finding the convex hull around the concave polygon, which is the simplest convex polygon containing all the vertices of the concave polygon, is the way to find these external cells.

As shown in fig. 8, which is a convex hull of the trapezoidal decomposition example, the convex hull minus the original concave polygon generates an outer optional polygon. The external selectable polygon itself can be decomposed into all selectable elements by the trapezoid, such as the areas 10-13 shown in fig. 9, and since the flight area and the external selectable polygon share vertices, the decomposed elements will always be aligned with each other in the vertical slice direction of the trapezoid decomposition, which means that a longer and thinner convex polygon will be generated after the element merging algorithm, which is more favorable for fast search. Fig. 10 shows an example of recombination between cells 1-4 and optional cell 10, which provides a larger single convex area for overlay search, and if there is no optional cell 10, cells 1-4 would incur a larger cost for search.

Based on the above theory, in S3, the dividing the flight area by using the area decomposition method to obtain a plurality of units to be merged may include:

(1) determining a minimum convex polygon around a flight zone; the smallest convex polygon is the convex hull.

(2) For each vertex of the minimum convex polygon, determining a partition line through the vertex and along the direction perpendicular to the long side of the minimum convex polygon; all the dividing lines divide the minimum convex polygon into a plurality of units to be merged. The long side direction is the long axis direction of the minimum convex polygon.

After obtaining a plurality of units to be merged, each unit to be merged can be used as a node, and a connecting edge is added between two adjacent units to be merged to obtain a unit connecting graph.

Dynamic programming is a technique that decomposes complex problems into simple sub-problems. By optimally combining these solutions with smaller problems, a complete solution can be constructed. As with the traversal method, dynamic programming examines each possible solution, thereby ensuring the solution's optimality. The advantage of the idea is that the solution of the simple problem can be stored or memorized, the solution of the same sub-problem can be reused in the iterative computation, if the sub-problems have obvious overlap, a large amount of computation time can be saved, and the purpose of reducing the computation cost is achieved. The embodiment may obtain an optimal solution by merging and decomposing the region units into smaller subproblems and then comparing the optimal solution and the solution of each subproblem with the solution of the whole region by using a dynamic programming recursive algorithm, where the sought optimal target is represented as follows: j (G) ═ min { c (G), min [ J (G1) + J (G2) ] };

where J (G) represents the best cost of the search area G, which is the minimum of the sum of the simple cost of directly searching the entire area c (G) and the best cost of searching any two sub-areas G1 and G2, where G1 and G2 cover the entire search area, and then the algorithm can be applied recursively by solving two sub-problems J (G1) and J (G2) using the same equations. Since there will be redundant flight in the concave region, an increase is chosen to give it a penalty cost, if the polygon is convex, the cost of searching for region G, c (G), is defined here as the time of flight, and if the region is concave, it is given a high penalty cost:

and the flight time t is the ratio of the search cost L to the flight speed of the unmanned aerial vehicle. The search cost L is calculated in the following manner: assuming that the speed of the unmanned aerial vehicle is constant and v, the width of an area which can be covered by the unmanned aerial vehicle linear search is' and the minimum turning radius of the unmanned aerial vehicle is r, the requirement that the constraint d is more than or equal to 2r is met. Since the drone speed is constant, the search cost can be expressed in terms of the overall covered track length:

wherein L is the length of the overall flight path, i is the number of the scanning line,

for the length of the i-th scan line scan track,

the length of the turn track for the ith scan. According to the definition of the Dubins path, the turning track length can be obtained: l _t ＝d+(π-2)r。

In the embodiment, a bottom-up dynamic planning method is selected, and the region merging mode can be roughly divided into two modes, one mode is a bottom-up planning method, namely, the minimum unit is continuously spliced, and the other mode is a splitting method from a whole unit to a local unit. The latter starts from the cost c (G) of global G, searching the whole region first, and then progressively decomposes this region into smaller sub-regions G1 and G2, the former approach starting from a single cell region, progressively building a search map by recombining adjacent cell regions or leaving un-merged, eventually covering the complete search region. The method is the same as the formula in mathematics, but the calculation sequence is slightly different in algorithm, and according to the idea of recursive combination, in order to ensure the result of the sub-problem can be reused, the former method is adopted, so that the lower memory usage is ensured.

As shown in fig. 11, which is a merged concept of the method, it is assumed that five cell regions currently exist, where cells 1-3 are internal cells and cells 4-5 are external supplementary cells, and a corresponding region is shown in fig. 12. The algorithm may start with any initial point and perform operations, for example, cells 1&2, which may be merged with any adjacent cell (e.g., branching on FIG. 11), or may consider that the merging of the current gray region is complete and then take a new initial cell (white) from any other adjacent cell in the search region for merging (e.g., branching off FIG. 11). The algorithm will start recursive work with each new starting point until every internal cell in the search area is covered. The current combined cost is calculated and stored in each step, and the stored costs are reused in the subsequent calculation steps. Subsequent steps such as the lower branch of FIG. 11 combine G {4,5} multiple times, each combination requiring the computation of the cost C (G {4,5 }). If the results of the calculations when this situation is first encountered are stored, they can simply be called from memory at any later time. In this recursive merging method, it is traversed to merge all possible cell regions until a possible solution is found that contains all the required (internal) cells. The flow of the recursive function algorithm described above is summarized as follows:

a) randomly selecting a starting unit area, listing all neighbor areas of the starting unit area, and combining the neighbor areas one by one;

b) calculating and storing the cost of the current unit area after combination;

c) randomly selecting a new initial area from the internal area (not containing the merged unit) and listing the neighbor units of the new initial area;

d) calculating and storing the cost of the current unit area after combination;

e) and returning all the stored cost values when the internal unit areas are completely covered, selecting the minimum cost, and obtaining the optimal combination mode.

For the polygon convexity condition detection problem during cost calculation, a gift wrapping method is adopted in the embodiment, the convexity of the points is evaluated clockwise according to the sequence of the points to finish judgment, and whether the merged area is convex or concave is judged.

Based on the above theory, in S3, the obtaining of the plurality of regions after recombination by recombining the units to be merged using a dynamic programming method may include:

(1) recombining the units to be combined by using a bottom-up dynamic programming method according to the unit connection diagram to obtain a plurality of recombination schemes; each recombination scheme comprises a plurality of alternative regions;

(2) calculating the cost of each recombination scheme, and selecting the recombination scheme with the minimum cost as the optimal recombination scheme; the plurality of candidate regions included in the optimal recombination scheme are a plurality of recombined regions.

Wherein calculating the cost of each reorganization scheme may include:

for each alternative area included by each recombination scheme, judging whether the alternative area is a convex polygon or not by using a gift packing algorithm; if so, the cost of the alternative region is flight time; otherwise, the cost of the alternative region is infinity; and calculating the sum of the costs of all the alternative regions to obtain the cost of the recombination scheme. The calculation mode of the flight time is as follows: and planning the flight path of the alternative area, calculating the searching cost L, and dividing the searching cost L by the flight speed of the unmanned aerial vehicle to obtain the flight time.

The following describes in detail a process of cell reorganization using a dynamic programming-based cell region merging method by using a reorganization example:

fig. 12 shows an exploded flight area consisting of three inner units 1-3 and two

outer units

4,5, and fig. 13 is a unit connection diagram for representing the area adjacency relationship, showing a relationship list merging these unit areas.

The first step in dynamic programming is to build the search network shown in fig. 14, by using a recursive algorithm to branch from node 1 on the left side and move towards the exit node on the right side, thereby building the graph. Once a node is reached where no more forward branches are possible (e.g., an exit node), the process traces back one step and tests alternative solutions, and the recursion continues until all possible branches are exhausted. Any node containing all internal units can branch directly to the egress node, representing a possible final solution. It should be noted that 1|2 indicates that 1 and 2 cells are separate polygons, and 1&2 indicates that they have been recombined. Figure 14 details some of the optimization effects applied in the reorganization process and first it can be seen that in a first step the program checks for reorganization with two

adjacent cells

2 and 4 and a new polygon that extends from cell area 2 (since cell area 4 is an optional external object unit). Second, it can be seen that at the top of the figure, node "1 |2| 3" has a branch through the exit, since this solution already contains all the necessary internal elements and is therefore a solution in itself. Furthermore, although node "1 &2& 3" contains all the necessary cell regions, this node does not branch to the egress node because the reorganization of G {1,2,3} will result in a concave cell, with the cost defined as infinite, so the planning program will choose to continue to branch with further mergers with other cells. Finally, "1 &2&3& 5" has no branch, since in this case the only possible additional cell is cell 4, but the sub-region "1 &2&3&4& 5" has already obtained the result from the "1 &2&3& 4" branch, and therefore no calculation is necessary again. Likewise, there is no branch from "1 &2&4& 5" and "1 &2& 5".

FIG. 15 shows the two final subregions G {1} and G {2,3,5} after decomposition and recombination, with the least costly segmentation scheme containing an optional region # 5. The original outer unit 5 is contained in the combined area, but the area 4 is not, and the introduction of the area 5 combined with the

unit areas

2 and 3 generates a convex sub-area, in which case the search cost of the drone is minimal. By recording the number of the calculation groups, the efficiency of the recursive combination calculation can be compared with that of a direct traversal method, the calculation times are found to be greatly reduced, and the computational complexity can be effectively reduced by the recursive combination calculation method, so that the planning speed is kept at a higher level.

The coverage search track consists of two different flight states: a straight scan path when seeking, and a turn path for transitioning to the next scan. The start and end of each scan can be represented by track points defining the intersection of the scan line and the polygon of the task area, as shown in FIG. 16, the line connecting two track points represents a search scan line with coordinates of

Where i is the scan index:

in the intersection point of the unmanned aerial vehicle and the scanning track on the boundary of the task area, the starting track point is defined as x _f ，y _f ]With a course of scan angle psi _s The direction of (2) defines the track point of the end of the scan as [ x ] ₀ ，y ₀ ]Heading psi _s . And taking the intersection point of the scanning line on the two sides and the boundary of the area as an angular point, and enabling the unmanned aerial vehicle to select one of the four angular points to start searching, so as to realize coverage patrol of the task area according to the correct track point indication sequence.

In fig. 16, an S-shaped cruise track is generated with a scan angle of 45 degrees, which is clearly not optimal, since some straight tracks are very short and the number of turns is high. A simple but effective way to select the scan angle is to represent the traversal order of all scans in a convex polygon by finding the smallest area bounding box that encloses the polygon, aligning the scan with the long axis of the polygon, then taking the angle of the known bounding box long axis as the scan angle, and defining the intersection of the scan line and the task area polygon as a straight course point. The general coverage search starting point is started from one of the polygon corner points, so that the generated cruising distance is not repeated and the complexity of program operation is favorably reduced, thereby ensuring the planning speed.

After obtaining the scan line, the drone needs to transition from one scan track to the next, and also needs to perform a turn operation, which is embodied as alternating between a left turn and a right turn, and involves a course adjustment of 180 degrees. This operation needs to be specifically defined to complete the entire time-of-flight model, which can be achieved in part by using the Dubins path. The Dubins path is the shortest curve connecting two points on a two-dimensional plane, with a limited curvature (in this case, the maximum turn rate of the drone), and as shown in fig. 17, provides a continuous flight path for transitions between adjacent tracks. It can be seen that these tracks are formed by two turning circles (whose radii satisfy the minimum turning radius constraint) and a portion of the common tangent line with respect to the starting and ending points of the tracks. All parts of the flight path are simple geometric shapes, which makes calculating length and time of flight simple and easy.

Based on the above theory, in S3, performing the flight path planning in each post-recombination region, and obtaining the flight path of each drone specifically includes:

(1) for each reorganized area, planning a linear scanning track according to the width of an area covered by the unmanned aerial vehicle during linear search and a scanning angle by taking an included angle between the reorganized area and a horizontal line as the scanning angle to obtain the linear scanning track;

specifically, the width of the covered area is determined by the detection width of the unmanned aerial vehicle sensor during the unmanned aerial vehicle linear search, and the maximum value of the width of the covered area is equal to the detection width during the unmanned aerial vehicle linear search. The values of the scan angles are determined in the following manner: and determining a minimum region boundary box of the reorganized region, and taking an included angle between the minimum region boundary box and a horizontal line as a value of the scanning angle. And planning a first straight line scanning track by starting from any vertex of the recombined region and taking the scanning angle as an extension direction. And then, planning a second linear scanning track by taking the width of the area covered by the linear search of the unmanned aerial vehicle as the distance between the next linear scanning track and the current linear scanning track and taking the scanning angle as the extending direction. And repeating the steps until the next straight line scanning track has no intersection point with the recombined area.

(2) Planning a turning track between two adjacent straight line scanning tracks by using a Dubins path to obtain the turning track; and all the straight line scanning tracks and all the turning tracks form the track of each unmanned aerial vehicle.

Next, the flight path planning method described in this embodiment is simulated, and as shown in fig. 18, the area to be searched of a given single machine is rotated by a predetermined angle, where the predetermined angle is an included angle between a long axis of the area to be searched and a horizontal direction. Since the region is not convex, an outer-wrapped convex polygon is found first, and the rotated region to be searched is trapezoidal-divided, and the division result is shown in fig. 19. The optimal polygon decomposition and combination structure is calculated from bottom to top by using the dynamic programming algorithm and then rotated back to the original angle, and the recombination result is shown in fig. 20. And respectively planning and covering the search tracks according to the combined areas and connecting to obtain the final finished search tracks as follows, as shown in fig. 21.

For the problem of multiple machines, it is generally sufficient to set two aggregation points in real life, so this embodiment only simulates the situation of a dual-machine single aggregation point and a dual-machine dual aggregation point:

(1) dual-machine single aggregation point condition:

two airplanes with the same searching capability are respectively divided into a coordinate (0, 0) point (figure 22(a)) and a coordinate (-134, 351) point (figure 22(b)), according to the searching capability of the unmanned plane, and the division line is a straight line because the two airplanes send out the searching completion from the same point. After the region segmentation is completed, the algorithm of the previous search flight path planning is applied to each region, and the final planned flight path of the region search is completed, wherein the flight path is shown in fig. 23.

(2) The situation of double-machine double-gathering point:

two airplanes with the same searching capability are respectively located at coordinates (-100, 0) and (100, 0), and a searching area is firstly divided by using a hyperbolic curve according to the searching capability ratio, as shown in fig. 24. In order to simplify the curve description, a straight line between the intersections of the hyperbolas and the area boundaries is used instead of the hyperbolas, and the final area division result is shown in fig. 25. The final flight path planning result is obtained as shown in fig. 26 by using the search flight path planning algorithm in the single-machine mission planning technique (if there are multiple airplanes at each rendezvous point, the planning is performed by re-dividing according to the division method of the multiple single rendezvous points). In the cooperative coverage area searching stage, a plurality of unmanned aerial vehicles scan according to a zigzag shape, the number of turns is reduced by preferentially walking long straight line sections, the searching cost is effectively reduced, the use efficiency of the unmanned aerial vehicles is increased, meanwhile, the requirement that the distance between a task point distributed to a certain airplane and the airplane is as short as possible is met in area distribution, and the effect meets the expectation.

Example 2:

this embodiment is used to provide a multi-drone area coverage track planning system, as shown in fig. 27, the track planning system includes:

the search area division module M1 is used for dividing the task area according to the total search capacity of each rendezvous point to obtain the search area of each rendezvous point; the gathering point comprises a plurality of unmanned aerial vehicles;

a flight area dividing module M2, configured to divide the search area of the rendezvous point for the search area of each rendezvous point, so as to obtain a flight area of each unmanned aerial vehicle included in the rendezvous point;

the flight path planning module M3 is used for dividing the flight area of each unmanned aerial vehicle by using an area decomposition method to obtain a plurality of units to be combined; recombining the units to be combined by using a dynamic programming method to obtain a plurality of recombined regions; and planning a flight path in each recombined region to obtain the flight path of the unmanned aerial vehicle.

The emphasis of each embodiment in the present specification is on the difference from the other embodiments, and the same and similar parts between the embodiments may be referred to each other. For the system disclosed by the embodiment, the description is relatively simple because the system corresponds to the method disclosed by the embodiment, and the relevant points can be referred to the description of the method part.

The principles and embodiments of the present invention have been described herein using specific examples, which are provided only to help understand the method and the core concept of the present invention; meanwhile, for a person skilled in the art, according to the idea of the present invention, the specific embodiments and the application range may be changed. In view of the above, the present disclosure should not be construed as limiting the invention.

Claims

1. A multi-unmanned aerial vehicle area coverage flight path planning method is characterized by comprising the following steps:

for the search area of each aggregation point, dividing the search area of each aggregation point to obtain the flight area of each unmanned aerial vehicle included in each aggregation point;

for the flight area of each unmanned aerial vehicle, dividing the flight area by using an area decomposition method to obtain a plurality of units to be combined; recombining the units to be combined by using a dynamic programming method to obtain a plurality of recombined regions; and planning a flight path in each recombined region to obtain the flight path of the unmanned aerial vehicle.

2. The track planning method according to claim 1, wherein the dividing of the task area according to the total search capability of each rendezvous point to obtain the search area of each rendezvous point specifically comprises:

obtaining the total search capacity of each rendezvous point by taking the sum of the search capacities of all unmanned aerial vehicles included in the rendezvous point as the total search capacity of the rendezvous point;

dividing task areas according to the proportion of the total searching capacity of each aggregation point to obtain the searching area of each aggregation point; the ratio of the total search capability of each of the aggregation points is the same as the ratio of the area of the search region of each of the aggregation points.

3. The track planning method according to claim 1, wherein the dividing the search area of the rendezvous point to obtain the flight area of each unmanned aerial vehicle included in the rendezvous point specifically includes:

rasterizing the search area to obtain a rasterized area; the rasterized region comprises a plurality of raster points;

distributing all the grid points to corresponding subsets of all the unmanned aerial vehicles according to the proportion of the searching capacity of all the unmanned aerial vehicles; the proportion of the searching capacity of each unmanned aerial vehicle is the same as the proportion of the number of grid points included in the corresponding subset of each unmanned aerial vehicle;

circularly exchanging the grid points in each subset according to the cost matrix until the objective function value after exchange is greater than or equal to the objective function value before exchange; the objective function value is the sum of the distance sums corresponding to the subsets; the sum of the distances corresponding to the subsets is the sum of the distances from each of the grid points included in the subsets to the drones corresponding to the subsets.

4. The trajectory planning method according to claim 3, wherein before cyclically exchanging the grid points in each of the subsets according to the cost matrix, the trajectory planning method further comprises: calculating the distance from each grid point to each unmanned aerial vehicle, and constructing a cost matrix; and the element of the ith row and the jth column of the cost matrix is the distance from the ith unmanned aerial vehicle to the jth grid point.

5. The flight path planning method according to claim 1, wherein the dividing the flight area by using an area decomposition method to obtain a plurality of units to be merged specifically comprises:

determining a minimum convex polygon around the flight zone;

for each vertex of the minimum convex polygon, determining a partition line through the vertex and along the direction perpendicular to the long side of the minimum convex polygon; all the dividing lines divide the minimum convex polygon into a plurality of units to be merged.

6. The route planning method according to claim 5, wherein before recombining the units to be merged by using the dynamic planning method, the route planning method further comprises: and taking each unit to be merged as a node, and adding a connecting edge between two adjacent units to be merged to obtain a unit connecting graph.

7. The track planning method according to claim 6, wherein the recombining the units to be combined by using the dynamic planning method to obtain the plurality of recombined regions specifically comprises:

recombining the units to be combined by using a dynamic programming method according to the unit connection diagram to obtain a plurality of recombination schemes; each recombination scheme comprises a plurality of alternative regions;

calculating the cost of each recombination scheme, and selecting the recombination scheme with the minimum cost as an optimal recombination scheme; the plurality of candidate regions included in the optimal recombination scheme are a plurality of recombined regions.

8. The trajectory planning method according to claim 7, wherein the calculating the cost of each of the recomposing solutions specifically comprises:

for each alternative area included by each recombination scheme, judging whether the alternative area is a convex polygon by using a gift packing algorithm; if so, the cost of the alternative region is flight time; otherwise, the cost of the alternative region is infinity;

and calculating the sum of the costs of all the alternative regions to obtain the cost of the recombination scheme.

9. The flight path planning method according to claim 1, wherein the performing of the flight path planning in each of the reorganized areas to obtain the flight path of the unmanned aerial vehicle specifically comprises:

planning a linear scanning track according to the width and the scanning angle of the area covered by the unmanned aerial vehicle during linear search to obtain the linear scanning track for each recombined area; the scanning angle is an included angle between the recombined region and a horizontal line;

planning a turning track between two adjacent straight line scanning tracks by using a Dubins path to obtain a turning track; and all the linear scanning tracks and all the turning tracks form the track of each unmanned aerial vehicle.

10. A multi-drone area coverage track planning system, the track planning system comprising:

the flight area dividing module is used for dividing the search area of each gathering point to obtain the flight area of each unmanned aerial vehicle included in the gathering point;