CN116149374B - Multi-unmanned aerial vehicle coverage path planning method - Google Patents
Multi-unmanned aerial vehicle coverage path planning method Download PDFInfo
- Publication number
- CN116149374B CN116149374B CN202310419663.6A CN202310419663A CN116149374B CN 116149374 B CN116149374 B CN 116149374B CN 202310419663 A CN202310419663 A CN 202310419663A CN 116149374 B CN116149374 B CN 116149374B
- Authority
- CN
- China
- Prior art keywords
- path
- cluster
- value
- points
- planning
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
- 238000000034 method Methods 0.000 title claims abstract description 51
- 238000011156 evaluation Methods 0.000 claims description 30
- 230000006870 function Effects 0.000 claims description 21
- 238000009313 farming Methods 0.000 claims description 16
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000004590 computer program Methods 0.000 claims description 7
- 238000005265 energy consumption Methods 0.000 claims description 5
- 101100410811 Mus musculus Pxt1 gene Proteins 0.000 claims description 3
- 238000005516 engineering process Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 3
- 238000003860 storage Methods 0.000 claims description 3
- 238000012364 cultivation method Methods 0.000 claims description 2
- 238000004422 calculation algorithm Methods 0.000 abstract description 22
- 238000002474 experimental method Methods 0.000 description 15
- 238000010586 diagram Methods 0.000 description 11
- 241000283690 Bos taurus Species 0.000 description 10
- 238000000354 decomposition reaction Methods 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 230000009286 beneficial effect Effects 0.000 description 3
- 230000000052 comparative effect Effects 0.000 description 3
- 238000005457 optimization Methods 0.000 description 3
- 230000002349 favourable effect Effects 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 238000013459 approach Methods 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 125000004122 cyclic group Chemical group 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000002068 genetic effect Effects 0.000 description 1
- 235000019580 granularity Nutrition 0.000 description 1
- 238000007689 inspection Methods 0.000 description 1
- 238000007726 management method Methods 0.000 description 1
- 238000002156 mixing Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000002922 simulated annealing Methods 0.000 description 1
- 238000000638 solvent extraction Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05D—SYSTEMS FOR CONTROLLING OR REGULATING NON-ELECTRIC VARIABLES
- G05D1/00—Control of position, course, altitude or attitude of land, water, air or space vehicles, e.g. using automatic pilots
- G05D1/10—Simultaneous control of position or course in three dimensions
- G05D1/101—Simultaneous control of position or course in three dimensions specially adapted for aircraft
- G05D1/106—Change initiated in response to external conditions, e.g. avoidance of elevated terrain or of no-fly zones
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02D—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN INFORMATION AND COMMUNICATION TECHNOLOGIES [ICT], I.E. INFORMATION AND COMMUNICATION TECHNOLOGIES AIMING AT THE REDUCTION OF THEIR OWN ENERGY USE
- Y02D30/00—Reducing energy consumption in communication networks
- Y02D30/70—Reducing energy consumption in communication networks in wireless communication networks
Landscapes
- Engineering & Computer Science (AREA)
- Aviation & Aerospace Engineering (AREA)
- Radar, Positioning & Navigation (AREA)
- Remote Sensing (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a multi-unmanned aerial vehicle coverage path planning method, which comprises the following steps: and importing information of an area needing to be covered by a path planning implementation, determining the area needing to be covered and the area forbidden to fly through, determining a set AS of access points needing to be covered in the area needing to be covered, and then adopting an improved K-center clustering algorithm to carry out multi-unmanned aerial vehicle coverage path planning. The invention uses a weight-adjustable cost function to evaluate the quality of the path planning scheme, wherein the cost function comprises two factors of the average value of the time consumption of each UAV path and the degree of deviation of the time consumption of each UAV path from the average value, and the weights of the two factors can be adjusted. Therefore, by adjusting the weight of the cost function, the invention can plan the path scheme according to the user preference, not only can plan the scheme with small total time consumption of the path, but also can plan the scheme with balanced time consumption of each UAV, and can plan the compromise scheme with small total time consumption and balanced time consumption of each UAV.
Description
Technical Field
The invention belongs to the technical field of unmanned aerial vehicle path planning, and particularly relates to a multi-unmanned aerial vehicle coverage path planning method.
Background
With the development of unmanned aerial vehicle (Unmanned Aerial Vehicle, UAV) technology, UAVs have been widely used in communication, reconnaissance, search, monitoring, inspection, disaster relief, and the like. Among other things, the coverage path planning (CoveragePathPlanning, CPP) problem is an important issue in many unmanned aerial vehicle applications. The task of the CPP is to build a path that ensures that the drone explores each location in a given area while avoiding no-fly zones (NoFlyingZone, NFZ).
Coverage path planning can be divided into two categories, depending on the number of CPP mission UAVs involved: single UAV coverage path planning, multiple UAV coverage path planning.
Currently, many studies are made on single UAV coverage path planning algorithms. Specific methods are Back-and-Forth (Back-and-Forth) and Spiral (Spiral) methods which are applicable to simple-shaped regions and do not require unit decomposition; a trapezoid decomposition method (Trapezoidal Decomposition) and a cow cultivation decomposition method (Boustrophedon Decomposition) for dividing a complex region into subregions and then performing path planning; there are also spanning tree methods (SpanningTreeCoverage, STC), wave front algorithms (Wavefront), genetic algorithms, ant colony algorithms, simulated annealing, etc. that perform path planning after discretizing the region into a set of regular shaped grids.
Common performance indicators for single UAV coverage path planning are: path length, turn angle and turn number, time to complete the path, and energy consumption to complete the path.
In contrast, there is currently less research on the multi-UAV coverage path planning problem, most solutions are performed in two steps:
(1) Dividing the area into a corresponding number of sub-areas according to the number of participating task UAVs, (2) performing single UAV coverage path planning on the sub-areas.
The evaluation indexes of the multi-UAV coverage path planning generally follow the indexes of the single UAV coverage path planning, and commonly used indexes are as follows: total path length, total time consumption, total energy consumption, etc.
Because the current multi-UAV coverage path planning algorithm mutually independent of the two steps of region division and path planning, an overall optimal path planning scheme is difficult to obtain. Region division algorithms tend to divide a region into sub-regions of equal area, regardless of the shape of the region or NFZ distribution, which are important factors affecting UAV path planning. The two steps, independent of each other, can also result in difficulty in optimizing certain overall performance, such as achieving time or energy consuming balances for each UAV.
Disclosure of Invention
The invention aims to: the invention aims to overcome the defects of the existing method, provides a multi-unmanned aerial vehicle coverage path planning method, and solves the problems that in multi-unmanned aerial vehicle coverage path planning, region division and path planning are mutually independent and task distribution is unbalanced.
The technical scheme is as follows: the invention provides a multi-unmanned aerial vehicle coverage path planning method, which comprises the following steps:
s1, importing information of a region to be covered by a path planning implementation, determining a region to be covered and a region to be prohibited from flying over, and determining a set AS of access points to be covered in the region to be covered;
s2, configuring parameter information, comprising: the system comprises a cluster number K, an iteration number constant MI and a strategy selection constant SC, wherein the cluster number K corresponds to the number of unmanned aerial vehicles participating in coverage search, and the strategy selection constant SC is used for adjusting the use proportion of different cluster center point generation strategies and is a number smaller than 1;
s3, setting initial values of iterative calculation variables, including: step of the neighborhood searching strategy and loop variable counter are gradually reduced from MI, step size is adjusted when the step size is reduced to 0, and the step size is reduced from MI again;
s4, judging whether the loop variable counter is larger than an iteration number constant MI, if so, turning to a step S5, otherwise, turning to a step S6;
s5, randomly selecting K points from a set AS of the access points to be accessed AS initial center points of the clusters, and then turning to step S9;
s6, judging whether the counter is larger thanIf yes, go to step S7, otherwise go to step S8;
s7, generating new center points of K clusters by adopting a random search strategy, and then turning to a step S9;
s8, generating new center points of K clusters by adopting a neighborhood searching strategy;
s9, distributing all points in the AS set to clusters closest to the AS set;
and S10, carrying out path planning on each cluster by adopting a cow farming method to obtain a new path newPaths.
Further, the method further comprises:
s11, calculating an evaluation value newF of a new path newPaths;
s12, judging whether the evaluation value newF is smaller than the evaluation value minF of the best scheme bestPaths at present; if newPaths is better than bestPaths, updating bestPaths to newPaths, updating minF to newF, setting counter to MI, otherwise, reducing counter by 1;
s13, judging whether the counter is reduced to 0; if counter is less than or equal to 0, the search step is adjusted toWherein, C2 is a step adjustment coefficient smaller than 1;
s14, judging whether step is smaller than MinStep, if step is not smaller than MinStep, turning to step S4 to carry out iterative computation, otherwise, ending the iterative computation, and outputting an optimal path planning scheme bestPaths.
Further, the method comprises the steps of:
the step S1 specifically includes:
the method comprises the steps of dispersing a region to be covered by a grid-based technology into grids, dividing the grids into A, N, B types, wherein the A type grids correspond to the region to be covered, and recording the center points of the grids as follows: a is that s (x, y), wherein A is the category of the point, s is the serial number of the point, and (x, y) is the coordinates of the point; n types of grids correspond to the no-fly zones, adjacent no-fly zone grids are combined into a rectangle as large as possible, and the rectangle of the no-fly zone is recorded as follows: NR (NR) t (N 0 ,N 1 ) Wherein t is the serial number of rectangle, N 0 Is the upper left corner point of rectangle, N 1 Is the right lower corner of the rectangle; class B grids are other grids that do not need to cover searches nor prohibit UAV fly-through, and thus the entire area information can be represented AS g=g (AS, NRS), where AS is the set of class a grid center points and NRS is the set of corresponding rectangles for no-fly zones.
Further, the method comprises the steps of:
in the step S4, setting an initial value of the iterative computation variable specifically includes:
；
step is the step length of the neighborhood searching strategy, and the larger the step value is, the larger the neighborhood searching range is; DN is the number of attribute-value-range divisions, C1 is a constant less than 1, used to set the initial step size; bestPaths is a list, and records the current optimal path planning scheme; the minF records an evaluation value of the cost function corresponding to bestPaths.
Further, the method comprises the steps of:
in step S7, generating new center points of K clusters by adopting a random search strategy, which specifically includes:
randomly selecting a value as a new central point of each cluster in the range of the value ranges of all points of each cluster, wherein the value is expressed as follows:
；
wherein, newCP i,j Is the value of the j attribute of the i-th cluster new center point; minP i,j Is the minimum value of the j-th attribute of all points in the i-th cluster; maxP i,j Is the maximum value of the j-th attribute of all points in the i-th cluster; rand is a random integer between 0 and DN, DN is the attribute value range division number;is the j attribute value of the class A grid center point with the sequence number of k; CS (circuit switching) i To be assigned to all points in the ith cluster.
Further, the method comprises the steps of:
the step S8 specifically includes:
searching and generating a new central point of the cluster in the neighborhood range of the central point of each current cluster, wherein the step length corresponds to the size of the neighborhood searching range, and the calculation formula of the new central point of the cluster is as follows:
；
wherein, newCP i,j Is the value of the j attribute of the i-th cluster new center point; oldCP i,j Is the value of the j attribute of the current center point of the i cluster; stepP j Is newCP i,j Increment on the j-th attribute; rand is the interval [ -C3, C3]Random integers in the range, C3 being a constant; minP j Is the minimum value of the j-th attribute of all points in the AS set; maxP j Is the maximum value of the j-th attribute of all points in the AS set.
Further, the method comprises the steps of:
the step S10 includes the steps of:
s10-1 in order from { CS 1 ，CS 2 ，…，CS k Extraction of CS i Is the current cluster;
s10-2 pair of current cluster CS i All points in (1) adopt depth-first searching partyDrawing a plurality of paths to cow cultivation rules different from the starting point; the depth-first search direction is divided into: horizontal, i.e., horizontal, vertical, diagonal, i.e., diagonal, starting points are divided into: xminYmin, i.e. upper left corner, xminYmax, i.e. lower left corner, xmaxYmin, i.e. upper right corner, xmaxYmax, i.e. lower right corner, different depth-first search directions and starting points are combined into different cow-farming clusters CS i Performing path planning;
s10-3 calculation of the current Cluster CS i Selecting the path with the smallest evaluation value as the planning path of the current cluster as the path according to the evaluation values of different path cost functions i And pass path i Adding to a newPaths list;
the path evaluation comprehensively considers the length and the rotation angle of the path, and the adopted cost function has flight time and energy consumption;
s10-4 whether all clusters CS have been completed i If not, go to step S10-1.
Further, the method comprises the steps of:
in the step S11, the evaluation value newF is obtained according to a formulated evaluation function, where the evaluation function F is expressed as:
；
wherein F is a cost function of a newPaths of the path planning scheme, and the smaller the value is, the better the corresponding scheme is; f (F) 0 Time consuming CT for each path in newPaths i Is the average value of (2); f (F) 1 Time-consuming deviations from mean F for paths in newPaths 0 The extent of (3); c4 is constant, F in F is adjusted 0 And F 1 The larger the duty cycle of C4, F 1 The heavier the duty ratio, the more balanced the time consumption of each unmanned aerial vehicle in the obtained planning scheme.
In another aspect, the invention provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the method described above when executing the program.
Finally, the present invention provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the method described above.
The beneficial effects are that: (1) The invention merges task allocation and path planning into one algorithm, and can seek the overall optimal solution of the multi-UAV coverage path planning problem.
(2) The invention uses a weight-adjustable cost function to evaluate the quality of the path planning scheme. The cost function includes a mean value of each UAV path time consumption and a degree to which each UAV path time consumption deviates from the mean value by two factors, and the two factor weights can be adjusted. By adjusting the weight of the cost function, a path scheme can be planned according to user preference, a scheme with small total time consumption of the path can be planned, a scheme with balanced time consumption of each UAV can be planned, and a compromise scheme with small total time consumption and balanced time consumption of each UAV can be planned.
(3) The present invention employs a hybrid search strategy to generate a new center point for a cluster. The hybrid search strategy uses two strategies, one strategy searches new center points randomly in the cluster value domain to obtain a globally optimal center point, and the other strategy searches new center points in the neighborhood of the cluster center point, and the strategy is favorable for finding out local optimal points. The hybrid search strategy can obtain more excellent planning paths, and the more complex the model is, the more obvious the advantages are.
(4) The invention expands the point taking range of the center point. The traditional K center clustering algorithm only searches for a new center point in the sample points, and the invention searches for the new center point in the range of the value range of the sample points. The expansion of the point taking range is beneficial to finding more and more suitable center points, so that a more excellent planning path is obtained.
(5) The invention adopts a variable step length neighborhood searching strategy. The algorithm starts to use a larger step length, searches a new center point in a neighborhood with a larger range so as to avoid local optimization and find a global optimization point; then, the step size is gradually reduced, and the search range is narrowed, so that the search is expanded in the local area, and the local optimal point is found. The variable step length neighborhood searching strategy realizes the searching modes of different thickness granularities, has the global optimizing and local optimizing capabilities, and is favorable for finding out more excellent planning paths.
(6) The present invention adopts a plurality of different element cow farming methods to plan regional paths. The cattle farming method of the invention uses different depth-first search directions and starting points, and the depth-first search directions are divided into: horizontal/Horizontal, vertical/Vertical, diagonal/Diagonal, starting points are divided into: xminYmin/upper left corner, xminYmax/lower left corner, xmaxYmin/upper right corner, xmaxYmax/lower right corner. The adoption of a plurality of different element cow farming methods is beneficial to finding a better path planning scheme.
Drawings
Fig. 1 is a flowchart of a multi-unmanned aerial vehicle coverage path planning method according to an embodiment of the present invention;
FIG. 2 is a region diagram and an NFZ diagram of a CPP that is required to be implemented in an embodiment of the present invention;
FIG. 3 is a region diagram after clustering in an embodiment of the invention;
FIG. 4 is a path layout diagram of a cow-farming method of different elements according to an embodiment of the present invention, wherein (a) the cow-farming depth-first search direction is Horizontal, the starting point is XminYmin, (b) the cow-farming depth-first search direction is Horizontal, the starting point is XmaxYmax, (c) the cow-farming depth-first search direction is Vertical, the starting point is XminYmin, (d) the cow-farming depth-first search direction is Vertical, the starting point is XminYmax, (e) the cow-farming depth-first search direction is Diaginal, the starting point is XminYmin, and (f) the cow-farming depth-first search direction is Diaginal, the starting point is XmaxYmax;
FIG. 5 is a diagram of a multi-unmanned plane path layout corresponding to CPP model1 in an embodiment of the present invention;
FIG. 6 is a schematic diagram of an interface of a development platform in an embodiment of the invention;
FIG. 7 is a schematic diagram of a result save file of a platform according to an embodiment of the present invention;
FIG. 8 is a scene graph corresponding to CPP model2 in an embodiment of the present invention;
FIG. 9 is a scene graph corresponding to CPP model3 in an embodiment of the present invention;
FIG. 10 is a scene graph corresponding to CPP model4 in an embodiment of the present invention;
FIG. 11 is a diagram of a multi-unmanned plane path layout corresponding to CPP model2 in an embodiment of the present invention;
FIG. 12 is a diagram of a multi-unmanned plane path layout corresponding to CPP model3 in an embodiment of the present invention;
FIG. 13 is a diagram of a multi-unmanned aerial vehicle path layout corresponding to CPP model4 in an embodiment of the present invention.
Detailed Description
The invention is further described below with reference to the drawings and the detailed description.
The invention relates to a multi-unmanned aerial vehicle coverage path planning method, wherein a flow chart is shown in fig. 1, and the method specifically comprises the following steps:
step 1: information of the CPP area to be implemented is imported. The method mainly comprises a set AS of required access points and NFZ rectangular information for prohibiting fly-by.
The area is discretized into grids using grid-based techniques and the grids are classified into three categories A, N, B. The class A grid corresponds to an area to be covered, and the center point of the record grid is as follows: a is that i (x, y), wherein A is the category of the point, i is the serial number of the point, and (x, y) is the coordinates of the point. N types of grids correspond to NFZ, adjacent NFZ grids are combined into a rectangle as large as possible, and the record NFZ rectangle is: NR (NR) i (N 0 ,N 1 ) Wherein i is the serial number of rectangle, N 0 Is the upper left corner point of rectangle, N 1 Is the lower right corner of the rectangle. Class B grids are other grids that do not require an overlay search nor prohibit UAV fly-by. The entire area information may be expressed AS g=g (AS, NRS), where AS is a set of class a grid center points and NRS is a set of NFZ rectangles.
In FIG. 2, the whole area has a size of 160m120m, dividing the region into a plurality of grids by solid lines parallel to the x and y axes, the size of the grids being 20m +.>20m, and the cross-point of the two dashed lines in the grid is the center point of the grid. Marking with dotsCenter points of class A grids, and recording coordinate information of the center points, e.g. A 1 (10,10)，A 2 (30,10),…，A 40 (150,110). Filling N kinds of grids into gray, combining adjacent N kinds of grids into NFZ rectangular, and recording coordinate information of upper left corner and lower right corner of the rectangular, such as N marked with 'X' in the figure 0 (60, 40) and N 1 (100, 80) NFZ rectangular NR respectively 1 Upper left corner and lower right corner of (c). The information of the whole area is as follows:
G={{A 1 (10,10)， A 2 (30,10), …， A 40 (150,110)},{[N 0 (60,40),N 1 (100,80)]}}
wherein,,
AS={A 1 (10,10)，A 2 (30,10), …，A 40 (150,110)}，NRS={[N 0 (60,40),N 1 (100,80)]}。
step 2: and configuring global parameter information.
Wherein the global parameter information includes: including the number of clusters K, the iteration number constant MI, the number of attribute value range divisions DN, the minimum step size MinStep, the policy selection constant SC, etc. The cluster number K corresponds to the number of UAVs participating in coverage search and is also the number of clusters required to be divided by a clustering algorithm; and dividing the value domain of a certain attribute into intervals with the same width, wherein DN corresponds to the number of the intervals.
Step 3: setting an initial value of the iterative computation variable. And (3) making:
；
step is the step length of the neighborhood searching strategy, and the larger the step value is, the larger the neighborhood searching range is; c1 is a constant less than 1, e.g., 0.5, for setting the initial step size; counter is a cyclic variable, the counter is gradually reduced from MI, the step size is adjusted when the counter is reduced to 0, and the counter is gradually reduced from MI again; bestPaths is a list, and records the current optimal path planning scheme; the minF records an evaluation value of the cost function corresponding to bestPaths.
Step 4: judging whether the counter is larger than the MI, if so, turning to the step 5, otherwise turning to the step 6. Step 5 is used to generate the initial center points of the K clusters, which is performed only once at the beginning of the iterative calculation.
Step 5: an initial center point of K clusters is generated and then goes to step 9. In the class a grid center point set AS of the region, K points are randomly selected AS initial center points of the cluster. In the example of fig. 2, let k=3, a is randomly selected in AS 2 、A 20 、A 36 Is the initial center point of the cluster, denoted as C 1 (30,10)、C 2 (130,50)、C 3 (70,110)。
Step 6: judging whether the counter is larger thanIf so, go to step 7, otherwise go to step 8. The algorithm updates the cluster center point by adopting a mixed search strategy, and the SC is a number smaller than 1 and is used for adjusting the use proportion of different cluster center point generation strategies.
Step 7: a random search strategy is used to generate new center points for K clusters and then go to step 9. Randomly selecting a value as a new central point of each cluster in the range of the value ranges of all points of each cluster, wherein the value is expressed as follows:
；
wherein, newCP i,j Is the value of the j attribute of the i-th cluster new center point; minP i,j Is the minimum value of the j-th attribute of all points in the i-th cluster; maxP i,j Is the maximum value of the j-th attribute of all points in the i-th cluster; rand is a random integer between 0 and DN; DN is the attribute value range division number;is the j attribute value of the class A grid center point with the sequence number of k; CS (circuit switching) i To be assigned to all points in the ith cluster.
In FIG. 3, all dots are formed with C 1 For the central cluster, CS can be found 1 The value ranges of the various attributes are:
；
assuming dn=100, rand1=65, rand2=30, the new center point of cluster 1 is newC 1 (62,28) the calculation is as follows:
；
step 8: a neighborhood search strategy is employed to generate new center points for K clusters. Searching and generating new central points of the clusters in the neighborhood range of the central point of each current cluster, wherein the step length corresponds to the size of the neighborhood searching range. The clustering new center point calculation formula is as follows:
；
wherein, newCP i,j Is the value of the j attribute of the i-th cluster new center point; oldCP i,j Is the value of the j attribute of the current center point of the i cluster; stepP j Is newCP i,j Increment on the j-th attribute; rand is the interval [ -C3, C3]Random integers in the range, C3 is constant, such as c3=5; step is the current step; minP j Is the minimum value of the j-th attribute of all points in the AS set; maxP j Is the maximum value of the j-th attribute of all points in the AS set; DN is the attribute value range division number.
In fig. 3, the value ranges of the attributes of all points in the AS set can be found AS follows:
；
assuming dn=100, step=2, rand1= -1, rand2=3, the center point of the current 1 st cluster is C 1 (30, 10), the new center point is newC 1 (27.2, 16) the calculation procedure is as follows:
；
step 9: all points in the AS set are assigned to clusters nearest to the cluster. For each point in the AS set, its distance to the center point of the K clusters is calculated separately and this point is assigned to the cluster with the smallest distance. In point A 11 (70, 30) for example, calculate it and C 1 (30,10)、C 2 (130,50)、C 3 The Manhattan distance of (70, 110) is as follows:
；
as can be seen from the above description,minimum, so will A 11 Assigned to C 1 Is in a cluster of center points. Similarly, other points in the AS set may be assigned to clusters nearest to it. As shown in FIG. 3 as C 1 、C 2 、C 3 For the clustering result of the cluster center, all dots form a cluster C 1 As a central cluster, all square points constitute a cluster with C 2 Is a central cluster, and all triangle points form a cluster with C 3 Is a central cluster.
Step 10: and (5) carrying out path planning on each cluster by adopting a cow farming method to obtain newPaths. The calculation steps are as follows:
step 10-1: in order from { CS 1 ，CS 2 ，…，CS k Extraction of CS i Is the current cluster.
Step 10-2: for the current cluster CS i And (3) marking out a plurality of paths by adopting cow cultivation rules with different depth-first search directions and starting points. The depth-first search direction is divided into: horizontal/Horizontal, vertical/Vertical, diagonal/Diagonal, starting points are divided into: xminYmin/upper left corner, xminYmax/lower left corner, xmaxYmin/upper right corner, xmaxYmax/lower right corner, different depth-first search directions and starting points are combined into different cow-farming pairs CS i And (5) path planning is implemented.
Fig. 4 shows a cluster CS 1 And (3) implementing various path plans. Taking the search direction and the starting point as (a) in fig. 4For example, horizontal_xminymin has a starting point of the cow cultivation method of upper left corner A 1 (10, 10) performing a depth-first search in the horizontal direction, then A in that direction 2 (30,10)、 A 3 (50,10)、…、A 5 (90, 10) all join paths Path1, A 5 Is CS 1 The search goes to the next line, i.e. the line with increased Y continues to perform the depth-first search in the horizontal direction, at the edge of the region, repeating the above-mentioned process until the CS is traversed 1 Is included in the above list.
Step 10-3: calculating the current cluster CS i Selecting the path with the smallest evaluation value as the planning path of the current cluster as the path according to the evaluation values of different path cost functions i And pass path i Added to the newpath list. Path evaluation comprehensively considers the length and the rotation angle of the path, and available cost functions include flight time and energy consumption. Taking the time of flight as an example, the calculation formula is as follows:
；
the path length is the total length of the path, namely the sum of the distances between adjacent points of the path; the path angle is the total rotation angle of the path, namely the sum of rotation angles of all points in the path; UV is unmanned aerial vehicle flight line speed; UW is unmanned aerial vehicle turning angular velocity; the cosnfz is a penalty term for a path crossing NFZ, and is a very large constant if any line segment that makes up the path intersects any NFZ interior region, or is 0 otherwise.
In fig. 4, assuming uv=10 m/s, uw=30°/s, for the cluster CS 1 The pathLen, pathAngle and CT values of the different paths planned are shown in the following table:
；
as can be seen from the table, the path planned by the cow farming method using the horizontal_XmaxYmax method is the least time consuming, and thus the path is the cluster CS 1 Is provided for the path planning.
Step 10-4: whether all clusters CS have been completed i Path planning tasks of (a)If not, go to step 10-1.
Step 11: an evaluation value newF of the path planning scheme newpath is calculated. For path planning of multiple UAVs, the minimum total UAV consumption is considered, and the time consumption of each UAV is considered to be balanced as much as possible, so that an evaluation function is formulated as follows:
；
wherein F is a cost function of a newPaths of the path planning scheme, and the smaller the value is, the better the scheme is; f (F) 0 Time consuming CT for each path in newPaths i Is the average value of (2); f (F) 1 Time-consuming deviations from mean F for paths in newPaths 0 The extent of (3); c4 is constant, F in F is adjusted 0 And F 1 The larger the duty cycle of C4, F 1 The heavier the duty cycle, the more time consuming each UAV is in the resulting planning scheme.
Step 12: and judging whether the evaluation value newF of the new path planning scheme newPaths is smaller than the evaluation value minF of the best scheme bestPaths at present. If newF is less than minF, i.e., newPaths is better than bestPaths, then bestPaths is updated to newPaths, minF is updated to newF, and the counter is set to MI. If newF is not less than minF, i.e., newPaths are not better than bestPaths, then counter is decremented by 1.
Step 13: it is determined whether counter has decreased to 0. If counter is less than or equal to 0, the search step is adjusted toWherein C2 is a step size adjustment factor of less than 1.
Step 14: judging whether step is smaller than MinStep, if step is not smaller than MinStep, turning to step 4 to perform iterative computation, otherwise, ending the iterative computation, and outputting an optimal path planning scheme bestPaths.
Fig. 5 shows an optimal path planning scheme for one experiment of the algorithm, where the total path length pathLen is 748.28 meters, the total path angle is 1080.00 degrees, and the total path time CT is 110.82 seconds.
In order to verify the performance of the algorithm provided by the invention, an experimental platform is developed by using PyCharm as a development tool and python as a development language, and a main interface of the platform is shown in FIG. 6. The experimental result is saved into two types of files, as shown in fig. 7, one type is a text file for recording information such as the total length of a scheme path, the total rotation angle of the path, the total time consumption, the evaluation value of a cost function, the center point of a cluster, the cluster containing a point set and the like, and the other type is a picture file for drawing a path plan. Experiments on 3 types of multi-unmanned aerial vehicle CPPs were performed by using a platform: the class 1 experiment is used for verifying the validity of the hybrid search strategy, the class 2 experiment is used for verifying the validity of the variable step length neighborhood search strategy, and the class 3 experiment is used for comparing the performance of the algorithm with other literature algorithms. Other documents referred to in the present invention include: comparative document 1 PPS Energy-Aware Grid-Based Coverage Path Planning for UAVs Using Area Partitioning in the Presence of NFZs,
comparative document 2. Aerial Coverage Optimization in Precision Agriculture Management: A Musical Harmony Inspired Approach.
4 different scenarios were used in the experiments, named CPP model1, CPP model2, CPP model3 and CPP model4, respectively, and the corresponding scenario diagrams are FIG. 2, FIG. 8, FIG. 9 and FIG. 10, respectively, where CPP model1 and CPP model4 are relatively simple, and CPP model2 and CPP model3 are relatively complex.
(1) And (5) verifying experiments by using a mixed search strategy. The experiment compares the performance of four different strategies: the method comprises an M1 strategy for randomly searching in a cluster point set, an M2 strategy for randomly searching in a cluster value domain, an M3 strategy for searching in a cluster center point neighborhood according to step length and an HM strategy for mixing M2 and M3. 10 experiments were performed with each strategy on CPP model1 and CPP model2, respectively, and Table 2 lists the total time-consuming CT of the four strategy planning path schemes, with Best, mean, worst being the best, mean and worst of the 10 experiments, respectively. Experimental results prove that the mixed search strategy can obtain a more excellent planning path, and the more complex the model is, the more obvious the advantages are.
；
(2) And (5) verifying experiments by using a variable step length neighborhood search strategy. Through experiments, the performances of two schemes of manually setting the Step length and automatically adjusting the Step length by an algorithm are compared. The experimental results are shown in table 3, and the experimental results show that different CPP models need to adopt different step sizes to obtain excellent performance, and the variable step size neighborhood search strategy has wider applicability, so that more excellent planning paths are obtained on different CPP models.
；
(3) Comparative experiments with other documents. The algorithm of the invention is compared with some existing documents with large influence. The experiment is reduced to 0.02, namely the minimum total time consumption scheme is searched to the maximum extent, specific data of a comparison experiment is given in table 4 and part of existing literature algorithms, and the optimal path planning obtained by the algorithm is given in fig. 11, 12 and 13. Experimental results prove that the algorithm provided by the invention can obtain a more excellent path.
；
In summary, the improved center clustering algorithm provided by the invention can effectively solve the problem of multi-unmanned aerial vehicle coverage path planning.
In another aspect, the invention provides an electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, the processor implementing the steps of the method described above when the program is executed.
Finally, the invention provides a computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the steps of the method described above.
The foregoing is merely a preferred embodiment of the present invention, and it should be noted that modifications and variations could be made by those skilled in the art without departing from the technical principles of the present invention, and such modifications and variations should also be regarded as being within the scope of the invention.
Claims (9)
1. The multi-unmanned aerial vehicle coverage path planning method is characterized by comprising the following steps of:
s1, importing information of a region to be covered by a path planning implementation, determining a region to be covered and a region to be prohibited from flying over, and determining a set AS of access points to be covered in the region to be covered;
s2, configuring parameter information, comprising: the system comprises a cluster number K, an iteration number constant MI and a strategy selection constant SC, wherein the cluster number K corresponds to the number of unmanned aerial vehicles participating in coverage search, and the strategy selection constant SC is used for adjusting the use proportion of different cluster center point generation strategies and is a number smaller than 1;
s3, setting initial values of iterative calculation variables, including: step of the neighborhood searching strategy and loop variable counter are gradually reduced from MI, step size is adjusted when the step size is reduced to 0, and the step size is reduced from MI again;
s4, judging whether the loop variable counter is larger than an iteration number constant MI, if so, turning to a step S5, otherwise, turning to a step S6;
s5, randomly selecting K points from a set AS of the access points to be accessed AS initial center points of the clusters, and then turning to step S9;
s6, judging whether the counter is larger thanIf yes, go to step S7, otherwise go to step S8;
s7, generating new center points of K clusters by adopting a random search strategy, and then turning to a step S9;
s8, generating new center points of K clusters by adopting a neighborhood searching strategy;
s9, distributing all points in the AS set to clusters closest to the AS set;
s10, carrying out path planning on each cluster by adopting a cow cultivation method to obtain a new path newPaths;
s11, calculating an evaluation value newF of a new path newPaths;
s12, judging whether the evaluation value newF is smaller than the evaluation value minF of the best scheme bestPaths at present; if newPaths is better than bestPaths, updating bestPaths to newPaths, updating minF to newF, and setting counter to MI, otherwise, reducing counter by 1;
s13, judging whether the counter is reduced to 0; if counter is less than or equal to 0, the search step is adjusted toWherein, C2 is a step adjustment coefficient smaller than 1;
s14, judging whether step is smaller than MinStep, if step is not smaller than MinStep, turning to step S4 to carry out iterative computation, otherwise, ending the iterative computation, and outputting an optimal path planning scheme bestPaths, wherein MinStep is the minimum step length of a neighborhood searching strategy.
2. The method for planning a coverage path of a multi-unmanned aerial vehicle according to claim 1, wherein the step S1 specifically comprises:
the method comprises the steps of dispersing a region to be covered by a grid-based technology into grids, dividing the grids into A, N, B types, wherein the A type grids correspond to the region to be covered, and recording the center points of the grids as follows: a is that s (x, y), wherein A is the category of the point, s is the serial number of the point, and (x, y) is the coordinates of the point; n types of grids correspond to the no-fly zones, adjacent no-fly zone grids are combined into a rectangle as large as possible, and the rectangle of the no-fly zone is recorded as follows: NR (NR) t (N 0 ,N 1 ) Wherein t is the serial number of rectangle, N 0 Is the upper left corner point of rectangle, N 1 Is the right lower corner of the rectangle; class B grids are other grids that do not need to cover searches nor prohibit UAV fly-through, and thus the entire area information can be represented AS g=g (AS, NRS), where AS is the set of class a grid center points and NRS is the set of corresponding rectangles for no-fly zones.
3. The method for planning a coverage path of multiple unmanned aerial vehicles according to claim 2, wherein in step S3, an initial value of an iterative computation variable is set, specifically comprising:
；
step is the step length of the neighborhood searching strategy, and the larger the step value is, the larger the neighborhood searching range is; DN is the number of attribute-value-range divisions, C1 is a constant less than 1, used to set the initial step size; bestPaths is a list, and records the current optimal path planning scheme; the minF records an evaluation value of the cost function corresponding to bestPaths.
4. The method for planning a coverage path of multiple unmanned aerial vehicles according to claim 2, wherein in step S7, a random search strategy is adopted to generate new center points of K clusters, and the method specifically comprises:
randomly selecting a value as a new central point of each cluster in the range of the value ranges of all points of each cluster, wherein the value is expressed as follows:
；
wherein, newCP i,j Is the value of the j attribute of the i-th cluster new center point; minP i,j Is the minimum value of the j-th attribute of all points in the i-th cluster; maxP i,j Is the maximum value of the j-th attribute of all points in the i-th cluster; rand is a random integer between 0 and DN, DN is the attribute value range division number;is the j attribute value of the class A grid center point with the sequence number of k; CS (circuit switching) i To be assigned to all points in the ith cluster.
5. The method for planning a coverage path of a multi-unmanned aerial vehicle according to claim 4, wherein step S8 specifically comprises:
searching and generating a new central point of the cluster in the neighborhood range of the central point of each current cluster, wherein the step length corresponds to the size of the neighborhood searching range, and the calculation formula of the new central point of the cluster is as follows:
；
wherein, newCP i,j Is the value of the j attribute of the i-th cluster new center point; oldCP i,j Is the value of the j attribute of the current center point of the i cluster; stepP j Is newCP i,j Increment on the j-th attribute; rand is the interval [ -C3, C3]Random integers in the range, C3 being a constant; minP j Is the minimum value of the j-th attribute of all points in the AS set; maxP j Is the maximum value of the j-th attribute of all points in the AS set.
6. The multi-unmanned aerial vehicle coverage path planning method according to claim 5, wherein the step S10 comprises the steps of:
s10-1 in order from { CS 1 ，CS 2 ，…，CS k Extraction of CS i Is the current cluster;
s10-2 pair of current cluster CS i Drawing a plurality of paths by adopting cow cultivation rules with different depth-first search directions and starting points; the depth-first search direction is divided into: horizontal, i.e., horizontal, vertical, diagonal, i.e., diagonal, starting points are divided into: xminYmin, i.e. upper left corner, xminYmax, i.e. lower left corner, xmaxYmin, i.e. upper right corner, xmaxYmax, i.e. lower right corner, different depth-first search directions and starting points are combined into different cow-farming clusters CS i Performing path planning;
s10-3 calculation of the current Cluster CS i Selecting the path with the smallest evaluation value as the planning path of the current cluster as the path according to the evaluation values of different path cost functions i And pass path i Adding to a newPaths list;
the path evaluation comprehensively considers the length and the rotation angle of the path, and the adopted cost function has flight time and energy consumption;
s10-4 whether all clusters CS have been completed i If not, go to step S10-1.
7. The method of planning a path for multiple unmanned aerial vehicles according to claim 6, wherein in step S11, the evaluation value newF is obtained according to a formulated evaluation function, and the evaluation function F is expressed as:
；
wherein F is a cost function of a newPaths of the path planning scheme, and the smaller the value is, the better the corresponding scheme is; f (F) 0 Time consuming CT for each path in newPaths i Is the average value of (2); f (F) 1 Time-consuming deviations from mean F for paths in newPaths 0 The extent of (3); c4 is constant, F in F is adjusted 0 And F 1 The larger the duty cycle of C4, F 1 The heavier the duty ratio, the more balanced the time consumption of each unmanned aerial vehicle in the obtained planning scheme.
8. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, characterized in that the processor implements the method of any of the preceding claims 1-7 when executing the program.
9. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the method of any of the preceding claims 1-7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310419663.6A CN116149374B (en) | 2023-04-19 | 2023-04-19 | Multi-unmanned aerial vehicle coverage path planning method |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310419663.6A CN116149374B (en) | 2023-04-19 | 2023-04-19 | Multi-unmanned aerial vehicle coverage path planning method |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116149374A CN116149374A (en) | 2023-05-23 |
| CN116149374B true CN116149374B (en) | 2023-07-21 |
Family
ID=86339191
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310419663.6A Active CN116149374B (en) | 2023-04-19 | 2023-04-19 | Multi-unmanned aerial vehicle coverage path planning method |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116149374B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116774733B (en) * | 2023-08-21 | 2023-10-31 | 南京航空航天大学 | Multi-unmanned aerial vehicle coverage path planning method |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114911270A (en) * | 2022-06-20 | 2022-08-16 | 南京信息工程大学 | Unmanned aerial vehicle coverage path planning method based on parallel self-adaptive ant colony algorithm |
| CN115202394A (en) * | 2022-07-13 | 2022-10-18 | 安徽成方智能科技有限公司 | Unmanned aerial vehicle full-coverage path planning method based on improved genetic algorithm |
| CN115755940A (en) * | 2022-12-09 | 2023-03-07 | 华中科技大学 | Underwater interesting area coverage path planning method and device of unmanned vehicle |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US10488865B2 (en) * | 2014-12-16 | 2019-11-26 | Al Incorporated | Methods and systems for robotic surface coverage |
| US9418560B1 (en) * | 2015-04-07 | 2016-08-16 | Raytheon Company | Automated sensor platform routing and tracking for observing a region of interest while avoiding obstacles |
| CN108459616B (en) * | 2018-03-07 | 2021-08-03 | 西安电子科技大学 | Unmanned aerial vehicle group collaborative coverage route planning method based on artificial bee colony algorithm |
| CN111580549A (en) * | 2020-04-20 | 2020-08-25 | 王方正 | Unmanned aerial vehicle full-coverage type path planning method |
| CN111984033B (en) * | 2020-08-19 | 2022-12-30 | 天津(滨海)人工智能军民融合创新中心 | Multi-unmanned aerial vehicle coverage task path planning method and device |
| CN112230675B (en) * | 2020-10-19 | 2022-06-24 | 南京航空航天大学 | Unmanned aerial vehicle task allocation method considering operation environment and performance in collaborative search and rescue |
| CN113741425B (en) * | 2021-08-04 | 2023-11-24 | 中山大学 | Full-coverage path planning method and navigation system |
| CN115951699A (en) * | 2022-10-24 | 2023-04-11 | 国网江苏省电力有限公司 | Collaborative path planning method and system for 5G power inspection unmanned aerial vehicle |
-
2023
- 2023-04-19 CN CN202310419663.6A patent/CN116149374B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN114911270A (en) * | 2022-06-20 | 2022-08-16 | 南京信息工程大学 | Unmanned aerial vehicle coverage path planning method based on parallel self-adaptive ant colony algorithm |
| CN115202394A (en) * | 2022-07-13 | 2022-10-18 | 安徽成方智能科技有限公司 | Unmanned aerial vehicle full-coverage path planning method based on improved genetic algorithm |
| CN115755940A (en) * | 2022-12-09 | 2023-03-07 | 华中科技大学 | Underwater interesting area coverage path planning method and device of unmanned vehicle |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116149374A (en) | 2023-05-23 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Deb et al. | A taxonomy for metamodeling frameworks for evolutionary multiobjective optimization | |
| Sharifzadeh et al. | The optimal sequenced route query | |
| Zhang et al. | Particle swarm optimization for construction site unequal-area layout | |
| US20130151535A1 (en) | Distributed indexing of data | |
| CN116149374B (en) | Multi-unmanned aerial vehicle coverage path planning method | |
| Zhang et al. | Simulation optimization using the particle swarm optimization with optimal computing budget allocation | |
| Agnesina et al. | Autodmp: Automated dreamplace-based macro placement | |
| CN115755954B (en) | Routing inspection path planning method, system, computer equipment and storage medium | |
| Jia et al. | Pareto-optimal solutions based multi-objective particle swarm optimization control for batch processes | |
| Yang et al. | Multigranulation rough set: a multiset based strategy | |
| Gilles et al. | Evasive path planning under surveillance uncertainty | |
| Suppakitpaisarn et al. | A Voronoi-based method for land-use optimization using semidefinite programming and gradient descent algorithm | |
| Liu et al. | DAACO: adaptive dynamic quantity of ant ACO algorithm to solve the traveling salesman problem | |
| Huang et al. | Grid interpolation algorithm based on nearest neighbor fast search | |
| Karakostas | Land-use planning via enhanced multi-objective evolutionary algorithms: optimizing the land value of major Greenfield initiatives | |
| Shetiya et al. | A unified optimization algorithm for solving" regret-minimizing representative" problems | |
| Shahabi et al. | Indexing land surface for efficient knn query | |
| Airan et al. | Linear machine solution to point location problem | |
| Zhang et al. | Optimization Models | |
| Wang et al. | A mixed integer programming formulation and solution for traffic analysis zone delineation considering zone amount decision | |
| Quan et al. | New Harmony Search Algorithm for Mobile Robot Path Planning | |
| Deng et al. | Utility distribution matters: Enabling fast belief propagation for multi-agent optimization with dense local utility function | |
| Shimamura et al. | Bayesian sparse convex clustering via global-local shrinkage priors | |
| Huang et al. | ACR-Tree: Constructing R-Trees Using Deep Reinforcement Learning | |
| Gavagsaz | Weighted spatial skyline queries with distributed dominance tests |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240110 Address after: Room 401, 4th Floor, Building A, No. 188 Tongsheng Avenue, Xinkai Street, Development Zone, Nantong City, Jiangsu Province, 226000 Patentee after: Nantong Senchuan Information Technology Co.,Ltd. Address before: 210044 No. 219 Ning six road, Jiangbei new district, Nanjing, Jiangsu Patentee before: Nanjing University of Information Science and Technology |
|
| TR01 | Transfer of patent right |