A containerized area partitioning method suitable for air-ground cooperative systems

Abstract

This invention proposes a bin-packing region partitioning method suitable for air-ground cooperative systems, comprising two main steps: bin initialization and bin-packing operation. Bin initialization aims to determine the bin's location by selecting the first target point to be binned. The target point closest to the centroid of all already binned points is selected to be binned into a new bin. The first binning point for the first bin is selected as the point furthest from the centroid of all target points. After bin initialization, the bin-packing operation will binned other target points into the newly generated bins. The first binning point of a bin is set as the starting point for the autonomous vehicle's (RV) stop. Generated bins prioritize packing target points closest to the starting point, thus packing as many target points as possible. Bin initialization and bin-packing operation iterate alternately until all target points are binned. Compared to traditional methods that utilize clustering to partition target regions, this invention reduces additional charging and transportation time, improving the task execution efficiency of air-ground cooperative systems.

Description

Technical Field

[0001] This invention relates to the field of path planning for air-ground cooperative systems, and in particular to a boxed area division method applicable to air-ground cooperative systems. Background Technology

[0002] Unmanned aerial vehicles (UAVs) and unmanned vehicles (UAVs) possess vastly different kinematic and dynamic models due to their distinct functions, shapes, payloads, and mobility capabilities, making them typical heterogeneous intelligent agents. Air-ground cooperative systems, composed of UAVs and UAVs, can fully leverage the characteristics of both and achieve complementary advantages, resulting in significantly higher work efficiency than single intelligent agents. This allows them to handle more complex tasks and has wide applications in both military and civilian fields. In the military sphere, air-ground cooperative systems have become an important tool for battlefield reconnaissance in countries worldwide. In the civilian sphere, air-ground cooperative systems can replace manual labor in completing long-duration, large-scale patrol and surveillance missions.

[0003] For large-scale reconnaissance missions, drones have limited battery power, allowing them to traverse only a limited number of target points in a single flight. Therefore, the target area needs to be divided into several sub-regions, transforming a single observation mission into multiple drone flights to extend battery life. An unmanned vehicle (UAV) transports the drone to the starting point of each sub-region, charging it during the transport. Each drone flight traverses all target points within a sub-region. While the drone is visiting target points in a sub-region, the UAV remains at the starting point to provide communication services. After visiting all target points in a sub-region, the drone flies back to the starting point.

[0004] Typically, K-means divides the target region into several sub-regions. However, clustering algorithms like K-means only divide the target region based on the distribution of target points and a preset number of clusters. If there are too many target points in a sub-region, the drone may not have enough power to traverse them and will need more time to fly back to charge. If a target point is outside the vehicle-to-drone communication range, the drone will need additional time to transport it to that point. This additional charging and transportation time prolongs the overall mission execution time, thus reducing mission efficiency.

[0005] The bin packing problem is a typical combinatorial optimization problem, defined as packing a set of items of different volumes into a set of bins of the same capacity, with the total volume of the items packed into the bins not exceeding the capacity of the bins and the number of bins being minimized. Dividing the target region can be transformed into a bin packing problem, where all target points correspond to items to be packed into bins, and all sub-regions correspond to bins. Summary of the Invention

[0006] To address the problems existing in the prior art, this invention proposes a bin-based region partitioning method, which provides a faster and more reasonable collaborative path planning method for air-ground collaborative systems, thereby solving the aforementioned defects of clustering algorithms when partitioning target regions.

[0007] This invention transforms the division of the target area into a packing problem, where all target points correspond to items to be packed into boxes, and all sub-regions correspond to boxes.

[0008] Technical solution of the present invention:

[0009] A bin-packing method for region partitioning suitable for air-ground cooperative systems is proposed, summarized as follows: the partitioning of target regions is transformed into a variant of the bin-packing problem: each sub-region corresponds to a virtual bin. Unlike the classic bin-packing problem, the capacity of the bin is defined as the number of target points that the UAV's battery can access in a loop within the communication range of the UAV and the unmanned vehicle. The capacity of the bin depends on the length of the loop path formed by all target points within the bin and the distribution of the target points.

[0010] A target point can be placed in the box if and only if the drone has enough remaining power to access the point and the point is within the communication range of the drone and the unmanned vehicle.

[0011] Each box should contain as many target points as possible to minimize the number of sub-regions, thereby maximizing the efficiency of the drone in a single flight.

[0012] A boxed area partitioning method suitable for air-ground cooperative systems, the specific implementation steps are as follows:

[0013] Step 1.1: Set the detection target area V = {v1,...,v} containing n detection target points. i ,...,v n}, Initialize algorithm parameters: autonomous vehicle speed Drone flight speed Vehicle-to-machine communication range d, maximum drone operating time T c Drone power consumption rate Autonomous vehicle charging rate

[0014] Initialize the set of unboxed target points U = V, and the set of boxed target points... The starting point set of each sub-region is Θ = φ, and the number of sub-regions is l = 1.

[0015] Step 1.2: Initialize the drone path set P a =φ and total flight time T a =0, the set of autonomous vehicle paths P g =φ and total travel time T g =0.

[0016] Step 1.3: Determine the starting point of the first sub-region and add it to the set of starting points for all sub-regions;

[0017] Calculate the centroid of the set of unpacked target points The formula is as follows:

[0018]

[0019] will with The furthest unpacked target point is set as the starting point θ of the first sub-region. l The formula is as follows:

[0020]

[0021] and θ l Add the starting point set Θ={θ l}∪Θ.

[0022] Step 1.4: Load other target points into the first sub-region and determine the shortest flight path and required time for the UAV within that sub-region;

[0023] First, the binning process is executed: the binning conditions are set as follows: the target point is within the communication range of the unmanned vehicle and the drone has enough power to travel to the traversal point. All unbinded target points that meet the conditions are grouped into θ in the binning process, ordered by distance. l Subregion b l This continues until there are no unpacked target points that meet the criteria within the set of unpacked target points.

[0024] Finally, a path planning algorithm is used to determine the closed-loop path P for the UAV to traverse all target points within the sub-region. l Calculate the flight time T of the drone within this sub-region. l The formula is as follows:

[0025]

[0026] in This indicates that the drone traversed the last target point within the sub-region. Update the drone path set P. a ={P l}∪P a Total flight time T of the drone a =T l +T a .

[0027] Step 1.5: Update the sub-region count and determine the starting point of the next sub-region.

[0028] Update the number of sub-regions l = l + 1. Calculate the centroid of the set of binned target points. The formula is as follows:

[0029]

[0030] will with The nearest unpacked target point is set as the starting point θ of the l-th sub-region. l The formula is as follows:

[0031]

[0032] and θ l Add the starting point set Θ={θ l}∪Θ.

[0033] Step 1.6: Load other target points into the newly generated sub-region and determine the shortest flight path and required time for the UAV within that sub-region.

[0034] First, the packing process is executed: all eligible unpacked target points are assigned to θ in order of distance, using a packing method. l Subregion B l This continues until no unboxed target points in the unboxed target point set meet the conditions. Finally, a path planning algorithm is used to determine the closed-loop path P for the UAV to traverse all target points in this sub-region. l and the required time T l Update the drone path set P a ={P l}∪P a Total flight time T of the drone a =T l +T a The operation method is the same as step 1.4.

[0035] Step 1.7: If the unboxed target point set is not empty, proceed to step 1.5; otherwise, proceed to step 1.8.

[0036] Step 1.8: Use a path planning algorithm to determine the closed-loop path P for the autonomous vehicle to traverse all sub-region starting point sets Θ. g Calculate the driving time T of the autonomous vehicle. g The formula is as follows:

[0037]

[0038] Where p Θ| This indicates the starting point of the last sub-region traversed by the drone.

[0039] Step 1.9: Calculate the waiting time T for the drone to charge. wThe formula is as follows:

[0040]

[0041] Step 1.10: Calculate the total time T required to complete the detection task, using the following formula:

[0042] T = T a +T g +T w

[0043] Step 1.11: Output the UAV flight path P a The autonomous vehicle's driving path P g And the total time required, T.

[0044] Specifically:

[0045] In steps 1.4 and 1.6 above, the specific process of packing is as follows:

[0046] Step (1): Initialize a set of candidate target points C = U and a sub-region B. l =φ.

[0047] Step (2): Set the starting point θ l Add the already packed target point set U and sub-region B l In, and never packed target point set. Remove from the candidate target point set C.

[0048] Step (3): Find the distance θ from the starting point from the candidate target point set C. l nearest point The formula is as follows:

[0049]

[0050] Will Remove from C.

[0051] Step (4): Use path planning algorithms to determine the traversal path of the UAV. The shortest time T' required to close the loop for all target points within the area. If and θ l The distance between them is less than the communication distance d and T' is less than the maximum operating time of the UAV T. c Then Add the already packed target point set U and sub-region B l The target point set was never packed. Remove.

[0052] Step (5): If the set of candidate target points C is not empty, then execute step (3); otherwise, execute step (6).

[0053] Step (6): Output the sub-region set B l .

[0054] Beneficial effects:

[0055] Compared to traditional clustering methods, this invention ensures that the UAV remains within the communication range of the unmanned vehicle during area detection missions in the air-ground cooperative system, and guarantees that the UAV has sufficient battery power to complete the traversal of each sub-region's target point. Compared to similar methods, this invention requires less mission execution time, thus improving mission efficiency. Attached Figure Description

[0056] Figure 1 This is a flowchart of the containerized area division method applicable to air-ground cooperative systems according to the present invention;

[0057] Figure 2 This describes the process of generating a sub-region in an embodiment of the present invention;

[0058] Figure 3 Box diagrams showing the total time used by the bin packing method (BPM) and the K-means-based partitioning method (ATMA) under different target point distribution settings in embodiments of the present invention;

[0059] Figure 4 The figures show the fitting curves of the total time used by the bin-based region partitioning method (BPM) and the K-means-based partitioning method (ATMA) under different communication distance settings in the embodiments of the present invention.

[0060] Figure 5 The figures show the fitting curves of the total time used by the boxed area partitioning method (BPM) and the K-means partitioning method (ATMA) under different UAV battery settings in this embodiment of the invention. Detailed Implementation

[0061] The technical solutions provided in this application will be further described below with reference to specific embodiments and accompanying drawings. The advantages and features of this application will become clearer from the following description.

[0062] Application of the boxed area partitioning method for air-ground cooperative systems in path planning of air-ground cooperative systems:

[0063] The division of the target region is transformed into a variant of the bin packing problem, with each sub-region corresponding to a virtual bin. Unlike the classic bin packing problem, the bin capacity is defined as the number of target points that the drone's battery can reach in a loop within the communication range of the drone and the unmanned vehicle. The bin capacity depends on the length of the loop path formed by all target points within the bin and the distribution of the target points. A target point can be put into a bin if and only if the drone has enough remaining battery power to reach the point and the point is within the communication range of the drone and the unmanned vehicle. Each bin should hold as many target points as possible to minimize the number of sub-regions, thereby maximizing the efficiency of the drone in a single flight. The specific implementation process includes two main steps: bin packing initialization and bin packing operation. The purpose of bin packing initialization is to determine the bin's position by selecting the first target point to be packed into the bin. The target point closest to the centroid of all already packed points is selected to be packed into the new bin, so that the generated bins can be closely packed together. Specifically, the first bin of the first bin is selected as the point farthest from the centroid of all target points, thus ensuring that all already packed points are on the same side as the unpacked points throughout the entire bin packing process. After bin packing initialization, the bin packing operation will load other target points into the newly generated bins. The first bin packing point of a bin is set as the starting point for the autonomous vehicle's (RV) stop. Generated bins prioritize packing target points closest to the starting point to reduce the distance between the RV and the drone. This maximizes the efficiency of the drone's single flight by loading as many target points as possible. Bin packing initialization and bin packing operations iterate alternately until all target points have been binned. Finally, path planning methods are used to plan the RV's driving path and the drone's flight path within each sub-region.

[0064] The implementation scheme adopted in this invention is as follows:

[0065] like Figure 1 A boxed area partitioning method suitable for air-ground cooperative systems, the specific implementation steps are as follows:

[0066] Step 1.1: Set the detection target area V = {v1,...,v} containing n detection target points. i ,...,v n}, Initialize algorithm parameters: autonomous vehicle speed Drone flight speed Vehicle-to-machine communication range d, maximum drone operating time T c Drone power consumption rate Autonomous vehicle charging rate

[0067] Initialize the set of unboxed target points U = V, and the set of boxed target points... The starting point set of each sub-region is Θ = φ, and the number of sub-regions is l = 1.

[0068] Step 1.2: Initialize the drone path set P a =φ and total flight time T a =0, the set of autonomous vehicle paths P g =φ and total travel time T g =0.

[0069] Step 1.3: Calculate the centroid of the set of unpacked target points. The formula is as follows:

[0070]

[0071] will with The furthest unpacked target point is set as the starting point θ of the first sub-region. l The formula is as follows:

[0072]

[0073] and θ l Add the starting point set Θ={θ l}∪Θ.

[0074] Step 1.4: Load other target points into the first sub-region and determine the shortest flight path and required time for the UAV within that sub-region;

[0075] First, the binning process is executed: the binning conditions are set as follows: the target point is within the communication range of the unmanned vehicle and the drone has enough power to travel to the traversal point. All unbinded target points that meet the conditions are grouped into θ in the binning process, ordered by distance. l Subregion b l This continues until no unboxed target points in the unboxed target point set meet the conditions. Finally, using path planning algorithms such as evolutionary algorithms and ant colony algorithms, a closed-loop path P is determined for the UAV to traverse all target points within the sub-region. l Calculate the flight time T of the drone within this sub-region. l The formula is as follows:

[0076]

[0077] Update the drone path set P a ={P l}∪P a Total flight time T of the drone a =T l +T a .

[0078] Step 1.5: Update the number of sub-regions l = l + 1. Calculate the centroid of the set of binned target points. The formula is as follows:

[0079]

[0080] will with The nearest unpacked target point is set as the starting point θ of the l-th sub-region. l The formula is as follows:

[0081]

[0082] and θ l Add the starting point set Θ={θ l}∪Θ.

[0083] Step 1.6: Load other target points into the newly generated sub-region and determine the shortest flight path and required time for the UAV within that sub-region.

[0084] First, the binning process is executed: all unbinded target points that meet the conditions (i.e., the target point is within the communication range of the unmanned vehicle and the drone has enough power to travel to the traversal point) are assigned to θ in the binning process according to their distance. l Subregion B l This continues until there are no unpacked target points that meet the criteria within the set of unpacked target points.

[0085] Finally, path planning algorithms such as evolutionary algorithms and ant colony algorithms are used to determine the closed-loop path P for the UAV to traverse all target points in the sub-region. l and the required time T l Update the drone path set P a ={P l}∪P a Total flight time T of the drone a =T l +T a .

[0086] Step 1.7: If the unboxed target point set is not empty, proceed to step 1.5; otherwise, proceed to step 1.8.

[0087] Step 1.8: Use a path planning algorithm to determine the closed-loop path P for the autonomous vehicle to traverse all sub-region starting point sets Θ. g Calculate the driving time T of the autonomous vehicle. g The formula is as follows:

[0088]

[0089] Step 1.9: Calculate the waiting time T for the drone to charge. w The formula is as follows:

[0090]

[0091] Step 1.10: Calculate the total time T required to complete the detection task, using the following formula:

[0092] T = T a +T g +T w

[0093] Step 1.11: Output the UAV flight path P a The autonomous vehicle's driving path P g And the total time required, T.

[0094] Specifically:

[0095] In steps 1.4 and 1.6 above, the specific process of packing is as follows:

[0096] Step (1): Initialize a set of candidate target points C = U and a sub-region B. l =φ.

[0097] Step (2): Set the starting point θ l Add the already packed target point set U and sub-region B l In, and never packed target point set. Remove from the candidate target point set C.

[0098] Step (3): Find the distance θ from the starting point from the candidate target point set C. l nearest point The formula is as follows:

[0099]

[0100] Will Remove from C.

[0101] Step (4): Use path planning algorithms to determine the traversal path of the UAV. The shortest time T' required to close the loop for all target points within the area. If and θ l The distance between them is less than the communication distance d and T' is less than the maximum operating time of the UAV T. c Then Add the already packed target point set U and sub-region B l The target point set was never packed. Remove.

[0102] Step (5): If the set of candidate target points C is not empty, then execute step (3); otherwise, execute step (6).

[0103] Step (6): Output the sub-region set B l .

[0104] To more intuitively verify the performance of this invention in path planning of an air-ground cooperative system, different target point distributions, UAV battery levels, and communication distance settings were used to fully demonstrate the performance of this invention.

[0105] The binning-based region partitioning method (BPM) proposed in this invention is compared with the current state-of-the-art air-ground cooperative path planning method (ATMA, J. Li, T. Sun, X. Huang, L. Ma, Q. Lin, J. Chen, and VCM Leung, “A memetic pathplanning algorithm for unmanned air / ground vehicle cooperative detection systems,” IEEE Transactions on Automation Science and Engineering, vol. 19, no. 4, pp. 2724–2737, 2022.). The results are as follows: Figures 3-5 As shown. Figure 3 The diagram illustrates the total time required for the two methods to complete the task under different target point distributions. Five different target point distributions are shown in the diagram: dispersion of 600, 800, 1000, random, and uniform. Under all distributions, BPM completes the task in less total time and exhibits better algorithmic stability than ATMA. Figure 4 The fitted curves of the total time used by the two methods are shown under different communication distance settings, ranging from 1000 to 3000 meters. Under different communication distance settings, BPM requires less time to complete the task compared to ATMA. Figure 5 The fitted curves of the total time used by the two methods are shown under different drone battery settings, with the longest drone working time ranging from 10 to 30 minutes. Under different drone battery constraints, BPM also requires less time to complete the task compared to ATMA. The above results indicate that BPM requires less time to visit all target points than ATMA and has stable algorithmic performance for different target point distributions.

[0106] The above description represents the preferred embodiments of the present invention. It should be noted that those skilled in the art can make various improvements and modifications without departing from the principles of the present invention, and these improvements and modifications are also considered to be within the scope of protection of the present invention.

Claims

1. A binning area division method suitable for air-ground cooperative systems, the specific implementation steps are as follows: Step 1.1: Set the detection target area V = {v1,...,v} containing n detection target points. i ,...,v n }, Initialize algorithm parameters: autonomous vehicle speed Drone flight speed Vehicle-to-machine communication range d, maximum drone operating time T c Drone power consumption rate Autonomous vehicle charging rate Initialize the unboxed target point set U = V, the boxed target point set Initialize the sub-region start point set Θ = φ, the sub-region number l = 1; Step 1.2: Initialize the set of drone paths P a = φ and total flight time T a = 0, the set of drone paths P g = φ and total flight time T g = 0; Step 1.3: Determine the start point θ of the first sub-region l and add it to the set of start points of the respective sub-region; Step 1.4: loading other target points into the first sub-region and determining the shortest flight path and required time of the unmanned aerial vehicle in the sub-region; First, the binning process is executed: the binning conditions are set as follows: the target point is within the communication range of the unmanned vehicle and the drone has enough power to travel to the traversal point. All unbinded target points that meet the conditions are grouped into θ in the binning process, ordered by distance. l Subregion b l The process continues until no unpacked target points meet the criteria within the set of unpacked target points; finally, a path planning algorithm is used to determine the closed-loop path P for the UAV to traverse all target points within this sub-region. l and the required time T l Update the drone path set P a ={P l }∪P a Total flight time T of the drone a =T l +T a ; Step 1.5: updating the sub-region number count and determining the starting point of the next sub-region; Update the sub-region number l = l + 1; calculate the centroid of the packed target point set The formula is as follows: The closest unboxed target point is set as the starting point θ of the lth sub-region The closest unboxed target point is set as the starting point θ of the lth sub-region l The formula is as follows: and Θ l Add each sub-region start set Θ = {θ l}∪Θ; Step 1.6: loading other target points into the newly generated sub-region and determining the shortest flight path and required time of the unmanned aerial vehicle in the sub-region; First, the binning process is performed: all eligible un-binned target points are binned into θ in order of distance l The sub-region B where the target point is located l , until there are no eligible un-binned target points in the set of un-binned target points; finally, the path planning algorithm is used to determine the path loop P of the UAV traversing all target points in the sub-region l ; the flight T of the UAV in the sub-region is calculated l , as follows: updating the set of paths P for the drone a = {P l}∪P a and the total flight time T for the drone a = T l + T a ; Step 1.7: if the un-binned target point set is not empty, execute step 1.5, otherwise execute step 1.8; Step 1.8: Determine the path loop P of the unmanned vehicle traversing all the starting point sets Θ of the sub-regions using a path planning algorithm g ; Calculate the travel time T of the unmanned vehicle g , as follows: where p Θ represents the starting point of the last sub-region traversed by the UAV; Step 1.9: Calculate the waiting time T for the drone to charge w The formula is as follows: Step 1.10: calculate the total time T required to complete the detection task, the formula is as follows: T = T a + T g + T w Step 1.11: Outputting the drone flight path P a unmanned vehicle travel path P g and the total time T required.

2. The binning-based region division method suitable for air-ground collaborative system according to claim 1, wherein, In step 1.3, the starting point of the first sub-region is determined as follows: calculating a centroid of the un-boxed target point set The formula is as follows: The target point with the farthest distance is set as the starting point θ of the first sub-region The target point with the farthest distance is set as the starting point θ of the first sub-region l The target point with the farthest distance is set as the starting point θ of the first sub-region and Θ l Add each sub-region start set Θ = {θ l}∪Θ.

3. The binning-based region division method suitable for air-ground collaborative system according to claim 1, wherein, In steps 1.4 and 1.6, the binning process is as follows: Step (1): initialize a candidate target point set C = U and a sub-region B l = φ; Step (2): Set the starting point θ l Add the boxed target point set U and the sub-region B l to the unboxed target point set U and remove from the candidate target point set C and the candidate target point set C Step (3): Find the candidate target point set C from the set of candidate target points C l the nearest point The formula is as follows: Remove from C Remove from C Step (4): determining the shortest time T' required for the UAV to traverse the path closed loop of all target points using a path planning algorithm; if the distance between the target point and the UAV is less than the communication distance d and T' is less than the longest working time T of the UAV, then the target point is added to the set of packed target points U and the sub-region B and removed from the set of unpacked target points U . Step (5): repeating steps (1) to (4) until all target points are packed. l Step (6): packing the target points in the set of unpacked target points U using the path planning algorithm. c Step (7): repeating steps (1) to (6) until all target points are packed. l Step (8): packing the target points in the set of unpacked target points U using the path planning algorithm. l Step (9): repeating steps (1) to (8) until all target points are packed. Step (5): if the candidate target point set C is not empty, execute step (3), otherwise execute step (6); Step (6): outputting the sub-region set B l .

Images