Path planning method for multi-unmanned-aerial-vehicle collaborative maritime search

Abstract

A path planning method for a multi-unmanned-aerial-vehicle collaborative maritime search. The method comprises: acquiring and analyzing related ocean current information of a region to be searched, and establishing an ocean current model on the basis of the obtained ocean current information (step 1); on the basis of the region to be searched, a command and control center establishing a grid map by taking a grid cell as a unit, and then establishing a target probability map on the basis of the related ocean current information of the region to be searched, prior target information and the grid map, wherein the target probability map provides the probability of a prior target being present in any grid cell (step 2); if there is an obstacle in the region to be searched, the obstacle being represented by an irregular polygon, generating a segmentation line for region division by taking a vertex of the irregular polygon as one end point of the segmentation line and also meeting a limiting constraint condition for region division, so as to divide the region to be searched into Nr sub-regions, and the command and control center allocating a sub-region to each unmanned aerial vehicle on the basis of the performance of the unmanned aerial vehicle by using a static allocation strategy (step 3); after the sub-region is allocated to each unmanned aerial vehicle, the command and control center planning a cross-regional shortest path for each unmanned aerial vehicle and using a path planning method to plan an unmanned aerial vehicle coverage path, so as to implement a single-region shortest-time coverage path (step 4); and monitoring the endurance state of each unmanned aerial vehicle during the process of executing a search task, and if it is found that the endurance capability is insufficient, recalling the unmanned aerial vehicle, and if a target is found or the search task is completed, recalling the unmanned aerial vehicle and returning to target coordinates (step 5).

Description

A multi-UAV collaborative maritime search path planning method Technical Field

[0001] The present invention belongs to the technical field of unmanned aerial vehicle (UAV) path planning, and in particular, is a method for collaborative maritime search path planning for multiple UAVs. Background Art

[0002] Currently, maritime search missions primarily rely on manned patrol vessels and manned aircraft, which are inefficient and costly. However, with the advancement of drone technology, unmanned equipment, such as unmanned vessels and drones, is increasingly being used in maritime emergencies. Drones can cover large areas in a short period of time, offering greater speed and rapid target location compared to traditional searches conducted by humans or manned aircraft. Drones can fly over difficult-to-reach or dangerous areas, enabling extensive target searches, including remote, inaccessible areas and at sea. Equipped with a variety of sensors, such as high-resolution cameras, infrared sensors, and multispectral sensors, drones collect and analyze images and data of the target area in real time. Equipped with advanced positioning systems, drones provide highly accurate target location information, facilitating accurate target location. Multiple drones can work collaboratively to search a single area, improving search efficiency and success rates through data sharing and collaborative decision-making.

[0003] In a multi-UAV collaborative search, UAVs need to work together to avoid overlapping searches or missed areas. Collaborative path planning algorithms can ensure that each UAV's search path is interconnected, forming an efficient search team. Path planning can be optimized based on sensor data from multiple UAVs. Sensor data fusion can provide more accurate target location information, which influences path planning decisions. During a search mission, the target may change over time, requiring dynamic path planning adjustments to adapt to changing conditions. Dynamic path planning algorithms can adjust UAV flight paths based on real-time data to maximize search efficiency. UAV search teams may need to distribute search targets within the search area. Path planning technology can help determine which area each UAV should search to avoid overlapping or missed targets. One of the goals of path planning technology is to improve search efficiency and reduce search time.

[0004] By properly planning paths, drones can cover more areas in the shortest possible time, improving search success rates. Currently, path planning technology is widely used in multi-UAV collaborative search, particularly in search and rescue, disaster response, and agricultural monitoring. However, several challenges remain, such as the complexity of large-scale multi-UAV collaboration, real-time data processing, and differences in flight performance, which require continued research and innovation.

[0005] Most current multi-UAV collaborative search path planning methods fail to consider the impact of real-world sea conditions on target probability maps, making it difficult to accurately predict how the target's location may shift with ocean currents. Many traditional methods fail to base their method selection strategies on prior target information, resulting in inefficiency and wasted resources. Furthermore, traditional coverage search methods struggle to simultaneously maintain optimal time and optimal path repetition rate. Summary of the Invention

[0006] High path duplication in traditional multi-UAV search is a common problem, especially in large-scale search missions, where multiple UAVs may search the same area at different times or repeatedly. This leads to wasted resources, reduced search efficiency, and lower mission success rates.

[0007] Traditional multi-drone maritime searches fail to consider real-world sea conditions, which can lead to problems and challenges that affect search efficiency and success rates. If drones fail to consider sea conditions, they may be forced to slow down or even stop flying in adverse sea conditions, resulting in reduced search efficiency and increased search time.

[0008] In response to the above-mentioned problems in the prior art, the present invention provides a method for collaborative maritime search path planning for multiple UAVs.

[0009] The purpose of this invention is to address the problems of existing multi-UAV collaborative search technology that does not incorporate real sea conditions and has high path duplication, and to propose a more efficient multi-UAV collaborative maritime search path planning method. The invention includes two entities:

[0010] Drone platforms include various types of drones, such as fixed-wing and multi-rotor drones. Choosing the appropriate drone type depends on mission requirements, ensuring it has appropriate flight performance, endurance, and payload capacity. Drones are also equipped with various sensors, such as high-resolution cameras, infrared sensors, multispectral sensors, and radar, to detect and locate targets and obtain information about sea conditions and the environment.

[0011] The Command and Control Center is responsible for developing search missions, planning routes, and assigning tasks to different drones. It also receives and analyzes data from drones. This is the core of the entire system, providing real-time monitoring and scheduling of tasks.

[0012] The method comprises the following steps:

[0013] Step 1: Obtain relevant ocean current information of the search area and analyze it, and establish an ocean current model based on the obtained ocean current information;

[0014] Step 2: The command and control center creates a grid map based on the search area, with grid cells as the unit. Then, based on the relevant ocean current information and prior target information of the search area, a target probability map is created based on the grid map. The target probability map gives the prior probability of the target existing in any grid cell.

[0015] Step 3: If there are obstacles in the search area, the obstacles are represented by irregular polygons, and the vertices of the irregular polygons are used as endpoints of the segmentation line. At the same time, the constraint conditions of the area division are met, and the segmentation line of the segmentation area is generated to divide the search area into N r The command and control center allocates a sub-area to each UAV using a static allocation strategy based on the UAV’s performance.

[0016] Step 4: After allocating sub-regions to each drone, the command and control center plans the shortest path across the region for each drone and uses path planning methods to plan the drone coverage path to achieve the shortest time coverage path for a single region;

[0017] Step 5: Monitor the endurance of each drone during the search mission. If the endurance is insufficient, recall the drone. If the target is found or the search mission is completed, recall the drone and return the target coordinates.

[0018] Furthermore, the specific steps of step 1 are as follows:

[0019] Step 1.1: Obtain and analyze ocean current information based on the search area. Ocean current information includes historical ocean current data, satellite observation data, ocean buoy data, and ocean weather station data.

[0020] Step 1.2: Based on the acquired ocean current information, a mathematical model of ocean current is established. The generated ocean current model is used to analyze the impact of ocean current on the probability of target existence in each grid cell. Assume that the direction and magnitude of ocean current are [lb cd ,ub cd ] and [lb cm ,ub cm ], the ocean current model is as follows:

[0021]

[0022]

[0023] Where: c dN is the predicted value of the current direction; c mN is the predicted value of the ocean current size; υ is the uncertainty factor of the ocean current direction; σ is the uncertainty factor of the ocean current size.

[0024] Furthermore, the specific steps of step 2 are as follows:

[0025] Step 2.1: The command and control center defines the working area L*W with a length of L and a width of W as the search area where the prior target exists. A grid map with square grid cells is established based on the search area. The search area is decomposed into N square grid cells. The set of all grid cells in the search area is represented by C. all ={C1,C2…,C N}, where C i ,i∈{1,2,…,N} represents the i-th grid cell;

[0026] Step 2.2: The command and control center calculates the C of each grid cell based on the ocean current data analyzed by the ocean current model and the prior target information. i The probability of the existence of the prior target in the grid map is used to establish a target probability map; any grid cell C in the search area i The probability of the internal target existing is expressed as P i According to the prior target information and ocean current data, the initial target existence probability P in each grid cell is given i (0); the sum of the initial probabilities of all grid cells in the target probability map is:

[0027]

[0028] In the target probability map, the probability that the target does not exist is 1-P i ; For ease of processing, l(C i ) is recorded as the probability of the target existing in the grid cell:

[0029]

[0030] Furthermore, when the UAV performs a search mission in the search area, it feeds back search information and ocean current change information. The command and control center updates the target probability map in real time based on the information fed back by the UAV. i , at time t, grid cell C i The observed value of the internal target existence probability is z; the target probability map update formula is:

[0031]

[0032] In the formula, l0(C i ) is the grid cell C at the initial moment i Target existence probability value; l t (C i ) is the time when the UAV completes searching grid cell C at time t. iThe probability value updated later; α is the probability weight factor of the influence of the current direction; β is the probability weight factor of the influence of the current speed; ub cd is the maximum estimated value of the current direction; lb cd is the minimum estimated value of the current direction; ub cm is the maximum estimated value of the current speed; lb cm is the minimum estimated value of the ocean current velocity.

[0033] The area partitioning problem aims to divide the task area into N r sub-areas, following two basic principles: workload balancing and avoiding splitting possible target areas, while effectively reducing the path duplication rate.

[0034] Furthermore, in step 3, the search area is allocated to R drones to perform the search task; the division constraint of the search area is:

[0035]

[0036] Where: N z Indicates the number of possible target areas before region division, the default is N z The value of N is 1; z Indicates the number of possible target areas after area division; S r Represents the area of ​​the possible target area before region division; S′ r Indicates the area of ​​the possible target area after area division; N r Indicates the number of sub-areas that the area is divided into; A indicates the area to be searched; A r Indicates the area allocated to the drone; A r′ represents the area not assigned to drones; r′ represents the number of drones that have not been assigned a sub-area; r represents the number of drones that have been assigned a sub-area;

[0037] In the equation, the value of the first objective function f1 represents the division coefficient of the search area; by minimizing the value of f1, a smaller number of drones are allocated to the possible target area;

[0038] The value of the second objective function f2 represents the deviation of the UAV's workload; by calculating the minimum value of f2, the optimal workload is assigned to the UAV so that the workload is balanced among multiple UAVs; the union of all sub-areas should cover the search area, and the sub-areas should not overlap with each other.

[0039] Furthermore, in step 3, the performance characteristics of each UAV are considered, including endurance, flight speed, and carrying capacity. After the area division is completed, a static allocation strategy is used to allocate sub-areas to each UAV according to the performance of the UAV. When the UAV performs the search mission, the sub-area allocation scheme is dynamically adjusted according to the changes in real-time search information. For each sub-area C area , calculate the drone evaluation function:

[0040]

[0041] Where: S i represents the search benefit of assigning the sub-region to the k-th UAV; represents the weight factor of the sub-region; k represents the UAV performance factor; R represents the number of UAVs; d k represents the Euclidean distance between the kth UAV and the sub-region center;

[0042] The search benefit of each drone is calculated for each sub-area, and the sub-area is allocated to the drone with the benefit, and the drone is responsible for performing the search mission in the sub-area.

[0043] Furthermore, in step 4, the command and control center plans the shortest cross-region path for each drone as follows:

[0044] The command and control center will divide the N r The adjacent relationship between sub-regions is modeled into a directed graph, in which each sub-region is identified as a node; the ocean current information of the sub-region and the current probability of the prior target are mapped into the weight of the directed edge; based on the comparison of the weights, the shortest path across the region is planned for each drone.

[0045] Furthermore, in step 4, a path planning method is used to plan the UAV coverage path to achieve the shortest time coverage path for a single area, specifically:

[0046] Step 4.1: The command and control center conducts a comprehensive analysis of the target probability map and the area division and allocation plan, issues a search mission to the UAV, and initializes the grid map and the UAV search starting point;

[0047] Step 4.2: The command and control center uses an intelligent optimization algorithm to plan the path of each UAV; considering the ocean current speed C at each location d and direction C m , the range of ocean current speed is (lb cm ,ub cm ), the range of the current direction is (lb cd ,ub cd), the Euclidean distance d between the current grid cell and the candidate grid cell, considering the flight performance of the UAV, the distance threshold is d m , the probability p of the target existing in the candidate grid cell i , calculate the cost of the path and construct the conditional function G;

[0048]

[0049] S is the set of candidate grid cells; i is the candidate grid cell number;

[0050] Step 4.3: Consider the ocean current speed and direction in each grid cell, the UAV’s own flight performance, the Euclidean distance between the UAV’s current position and the candidate grid cell, and the probability of the target existing in the candidate grid cell. Calculate the UAV’s search benefit for the next grid cell and construct the target optimization function. is the weighting factor of the probability of target existence in the candidate grid cell, γ is the weighting factor of the Euclidean distance between the current grid cell and the candidate grid cell, x is the weighting factor of the current speed, and δ is the weighting factor of the current direction. The size of the weighting factor is measured by multiple experiments, which represents the influence of the current speed, current direction, the Euclidean distance between the current grid cell and the candidate grid cell, and the probability of target existence in the candidate grid cell on the target optimization function; based on the constructed conditional function, the sub-area search path is generated by calculating the target optimization function.

[0051] Furthermore, the specific steps of step 5 are as follows:

[0052] Step 5.1: Real-time monitoring of battery charge, battery voltage, and battery temperature to estimate the remaining flight time of the drone, and to issue a warning or switch drones in advance when the battery is close to depletion.

[0053] Step 5.2: If the target is found or the search mission is completed, recall the drone and return the target coordinates.

[0054] The present invention has the following beneficial effects: 1. It constructs a current model based on the ocean current information in the search area. The current data analyzed by the current model is used to plan the drone's path, including both cross-regional and individual sub-regional paths. Real-time updates of sea conditions and reasonable division of the search area allow the drone to select a flight path that best suits the current sea conditions. This helps prevent the drone from entering areas with adverse sea conditions and improves flight stability and safety.

[0055] 2. The present invention adopts regional division in the collaborative search of multiple UAVs at sea, avoiding the problem of repeated search paths and repeated search areas of multiple UAVs in traditional methods, thereby improving search efficiency.

[0056] 3. The present invention updates the target probability map in real time based on the information fed back by the drones, accurately determines the location of the target in the search area, and considers balancing the drone load when allocating sub-areas to the drones. This solves the problem of low efficiency caused by multiple round trips of drones due to insufficient endurance in traditional methods, while minimizing the number of drones performing search tasks to reduce costs.

[0057] 4. Traditionally, drone search paths are often manually pre-planned, potentially leading to duplicate searches or missed areas. Task assignments are typically static and pre-planned, unable to dynamically adapt to changing circumstances and mission requirements. Static assignments and planned paths result in some drones being underutilized during the mission, wasting their resources and capabilities. By dividing the search area into multiple sub-areas, each drone is responsible for searching its own specific sub-area, improving search parallelism and efficiency. This regional division allows multiple drones to work simultaneously, effectively reducing the time required to complete the entire search mission. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] FIG1 is a diagram of a system model of the present invention.

[0059] FIG2 is a grid map of the area to be searched in the present invention.

[0060] FIG3 is a diagram of an ocean current model in the present invention.

[0061] FIG4 is a flow chart of the area division and cross-domain path planning solution in the present invention.

[0062] FIG5 is a flow chart of neutron region path planning according to the present invention.

[0063] Figure 6 shows the target probability map related data.

[0064] FIG7 is a simulation diagram of the division of the area to be searched.

[0065] Figure 8 shows the simulation results of cross-regional path planning. DETAILED DESCRIPTION

[0066] In order to deepen the understanding of the present invention, the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. The embodiments are only used to explain the present invention and do not limit the scope of protection of the present invention.

[0067] As shown in FIG1 , the solution described in the present invention includes the following two entities:

[0068] Drone platforms include various types of drones, such as fixed-wing and multi-rotor drones. Choosing the appropriate drone type depends on mission requirements, ensuring it has appropriate flight performance, endurance, and payload capacity. Drones are also equipped with various sensors, such as high-resolution cameras, infrared sensors, multispectral sensors, and radar, to detect and locate targets and obtain information about sea conditions and the environment.

[0069] The Command and Control Center is responsible for developing search missions, planning routes, and assigning tasks to different drones. It also receives and analyzes data from drones. This is the core of the entire system, providing real-time monitoring and scheduling of tasks.

[0070] Embodiment: A method for collaborative maritime search path planning by multiple UAVs specifically includes the following steps:

[0071] Step 1: Initialize the command and control center, set the mission scenario to an open sea area with islands, obtain and analyze relevant information about the search area, and build a current model based on the obtained current information;

[0072] The specific implementation steps of step 1 are as follows:

[0073] Step 1.1 The command and control center confirms the sea area for the search mission and obtains and analyzes ocean current information based on the search area. Ocean current information includes historical ocean current data, satellite observation data, ocean buoy data, marine weather station data, etc.

[0074] Step 1.2: Build a mathematical model of ocean currents based on the collected data. Ocean currents are often influenced by a variety of factors, including the Earth's rotation, wind, and tides. Using this mathematical model, simulate and generate ocean current data. This data can represent the speed and direction of ocean currents, as well as how they change over time. Because ocean currents are dynamic, the model can be compared with actual measured data in real time and filter technology can be used to update the ocean current field information in real time to reflect current changes.

[0075] Assume that the direction and magnitude of the current are [lb cd ,ub cd ] and [lb cm ,ub cm ]

[0076]

[0077]

[0078] Where: c dN is the predicted value of the current direction; c mNis the predicted value of the ocean current size; υ is the uncertainty factor of the ocean current direction; σ is the uncertainty factor of the ocean current size; the ocean current model is shown in Figure 3.

[0079] Step 2: The command and control center creates a grid map based on the search area, with grid cells as the unit, as shown in Figure 2. Based on the relevant ocean current information and prior target information for the search area, a target probability map is created based on the grid map, as shown in Figure 6. The target probability map calculates the probability of the target existing in each grid cell based on the grid map. The target probability map gives the probability of the target existing in any grid cell under the influence of ocean currents.

[0080] The specific implementation steps of step 2 are as follows:

[0081] Step 2.1 The command and control center defines the working area L*W with a length of L and a width of W as the search area with a priori targets. A grid map with square grid cells is established based on the search area. The search area is decomposed into N square grid cells. The set of all grid cells in the search area is represented by C. all ={C1,C2…,C N}, where C i ,i∈{1,2,…,N} represents the i-th grid cell;

[0082] Step 2.2 The command and control center calculates the C of each grid cell based on the ocean current data analyzed by the ocean current model and the prior target information. i The probability of the existence of the prior target in the grid map is used to establish a target probability map; any grid cell C in the search area i The probability of the internal target existing is expressed as P i According to the prior target information and ocean current data, the initial target existence probability P in each grid cell is given i (0); the sum of the initial probabilities of all grid cells in the target probability map is:

[0083]

[0084] In the target probability map, the probability that the target does not exist is 1-P i ; For ease of processing, l(C i ) is recorded as the probability of the target existing in the grid cell:

[0085]

[0086] Step 2.3: During the search mission, the drone detects changes in ocean currents in real time and feeds back search information and real-time ocean current information to the command and control center. The command and control center calculates real-time updated ocean current data based on the ocean current model from Step 1 and updates the target probability map.

[0087] The drone has searched a grid cell C at time t. i , at time t, grid cell C i The observed value of the internal target existence probability is z; the target probability map update formula is:

[0088]

[0089] In the formula, l0(C i ) is the grid cell C at the initial moment i Target existence probability value; l t (C i ) is the time when the UAV completes searching grid cell C at time t. i The probability value updated later; α is the probability weight factor of the influence of the current direction; β is the probability weight factor of the influence of the current speed; ub cd is the maximum estimated value of the current direction; lb cd is the minimum estimated value of the current direction; ub cm is the maximum estimated value of the current speed; lb cm is the minimum estimated value of the ocean current velocity.

[0090] Step 3: There are obstacles such as islands in the search area. The obstacles are represented by irregular polygons. In order to reduce the number of turns the drone has to make when searching and to improve the search efficiency, the vertex of the irregular polygon is used as an endpoint of the dividing line. At the same time, the limiting constraints of the area division are met and the dividing line of the area is generated to divide the search area into N r The command and control centerline uses a static allocation strategy to allocate sub-areas for each UAV based on its performance, as shown in Figure 7.

[0091] The specific implementation steps of step 3 are as follows:

[0092] Step 3.1: The region partitioning problem aims to divide the mission area into multiple sub-regions. Two basic principles are followed: balancing the workload of UAVs and avoiding segmenting possible target areas. In step 3, the search area is assigned to R UAVs to perform the search mission; the constraints for the search area partitioning are:

[0093]

[0094] Where: N z Indicates the number of possible target areas before region division, the default is N z The value of is 1;

[0095] N′ z Indicates the number of possible target areas after area division;

[0096] S r Indicates the area of ​​the possible target area before area division;

[0097] S′ r Indicates the area of ​​the possible target area after area division;

[0098] N r Indicates the number of sub-regions that the region is divided into;

[0099] A represents the area to be searched;

[0100] A r Indicates the area allocated to the drone;

[0101] A r′ Indicates the area not allocated to drones;

[0102] r′ represents the number of the drone that is not assigned a sub-area;

[0103] r represents the number of the drone that has been assigned a sub-area;

[0104] In the equation, the value of the first objective function f1 represents the division coefficient of the search area; by minimizing the value of f1, a smaller number of drones are allocated to the possible target area;

[0105] The value of the second objective function f2 represents the deviation of the UAV's workload; by calculating the minimum value of f2, the optimal workload is assigned to the UAV, so that the workload is balanced among multiple UAVs; the union of all sub-areas should cover the area to be searched, and secondly, the sub-areas should not overlap with each other; the multi-UAV collaborative search path planning of the entire sea area is converted into R single-UAV coverage path planning, effectively reducing the multi-UAV path repetition rate.

[0106] Step 3.2 Consider the performance characteristics of each drone, including endurance, flight speed, carrying capacity, etc. These characteristics will affect the size and location of the sub-area that the drone can be responsible for. After the area division is completed, a static allocation strategy is used to allocate sub-areas to each drone based on the performance of the drone. When the drone performs the search mission, the sub-area allocation plan is dynamically adjusted according to the changes in real-time search information. For each sub-area C area , calculate the drone evaluation function;

[0107]

[0108] Where: S i represents the search payoff of assigning the subregion to the kth UAV.

[0109] Represents the weight factor of the sub-region.

[0110] ξk Represents the UAV performance factor.

[0111] R represents the number of drones.

[0112] d k represents the Euclidean distance between the kth UAV and the center of the sub-region.

[0113] The search benefit of each drone is calculated for each sub-area, and the sub-area is allocated to the drone with the benefit, and the drone is responsible for performing the search mission in the sub-area.

[0114] Step 4: After assigning each UAV to a sub-region, the command and control center plans the shortest cross-region path for each UAV and uses path planning methods to plan the UAV coverage path to achieve the shortest time coverage route for a single region. The cross-region path planning process is shown in Figure 4; the single sub-region path planning process is shown in Figure 5.

[0115] The command and control center plans the shortest path across regions for each drone as follows: The command and control center divides the N r The adjacent relationship between sub-regions is modeled into a directed graph, in which each sub-region is identified as a node; the ocean current information of the sub-region and the current probability of the prior target are mapped into the weight of the directed edge; based on the comparison of the weights, the shortest path across the region is planned for each drone, as shown in Figure 8. By analyzing the relationship between the nodes and edges in the graph, the corresponding drone that can most effectively perform the task is determined. Communication links are established between drones to help determine which drones need to communicate to work together. This helps ensure timely information transmission and better coordination of drone search tasks.

[0116] Use the path planning method to plan the UAV coverage path to achieve the shortest time coverage path for a single area. The specific implementation steps are as follows:

[0117] Step 4.1 The command and control center conducts a comprehensive analysis of the target probability map and the area division and allocation plan, issues a search mission to the UAV, and initializes the grid map and the UAV search starting point;

[0118] Step 4.2 The command and control center uses an intelligent optimization algorithm to plan the path of each drone; considering the ocean current speed C at each location d and direction C m , the range of ocean current speed is (lb cm ,ub cm ), the range of the current direction is (lb cd ,ub cd ), the Euclidean distance d between the current grid cell and the candidate grid cell, considering the flight performance of the UAV, the distance threshold is d m, the probability p of the target existing in the candidate grid cell i , calculate the cost of the path and construct the conditional function G;

[0119]

[0120] S is the set of candidate grid cells; i is the candidate grid cell number;

[0121] Step 4.3 Considering the ocean current speed and direction in each grid cell, the UAV’s own flight performance, the Euclidean distance between the UAV’s current position and the candidate grid cell, and the probability of the target existing in the candidate grid cell, calculate the UAV’s search benefit for the next grid cell and construct the target optimization function. is the weighting factor of the probability of target existence in the candidate grid cell, γ is the weighting factor of the Euclidean distance between the current grid cell and the candidate grid cell, x is the weighting factor of the current speed, and δ is the weighting factor of the current direction. The size of the weighting factor is measured by multiple experiments, which represents the influence of the current speed, current direction, the Euclidean distance between the current grid cell and the candidate grid cell, and the probability of target existence in the candidate grid cell on the target optimization function; based on the constructed conditional function, the sub-area search path is generated by calculating the target optimization function.

[0122] Step 5: During the search mission, monitor the endurance of each drone. If the endurance is insufficient, recall the drone. If the target is found or the search mission is completed, recall the drone and return the target coordinates.

[0123] Step 5.1: During the search mission, the UAV detects that the current speed and direction have changed significantly, exceeding a threshold, taking into account the real-time changes in the ocean current. The UAV then provides feedback on the search and current changes. Command and Control then recalculates the target probability map and recalculates the path cost to ensure that each search path is optimal.

[0124] Step 5.2 The command and control center monitors the flight status of each drone in real time. If the flight time is insufficient, the drone is recalled and dynamically adjusted based on the remaining energy consumption of each drone, the distance between drones, and the workload, generating a new area allocation plan. At the same time, the command and control center calculates and updates the coverage rate of the entire search mission area in real time;

[0125] Step 5.3 If the target is found or the search mission is completed, recall the drone and return the target coordinates.

[0126] The above description is an exemplary embodiment of the present invention and does not limit the scope of patent protection of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the description and drawings of the present invention, or directly or indirectly applied in other related technical fields, are also included in the scope of patent protection of the present invention.

Claims

1. A multi-UAV collaborative maritime search path planning method uses a UAV platform and a command and control center. The UAV platform includes different types of UAVs. The command and control center is responsible for formulating search tasks, planning paths, and assigning tasks to different UAVs. At the same time, it can also receive and analyze data from the UAVs; it is characterized in that, The method includes the following steps: Step 1: Obtain and analyze the relevant ocean current information of the area to be searched, and establish an ocean current model based on the obtained ocean current information. Step 2: The command and control center establishes a grid map in units of grid cells according to the area to be searched, and then based on the relevant ocean current information and prior target information of the area to be searched, establishes a target probability map based on the grid map; the target probability map gives the probability of the existence of the prior target in any grid cell. Step 3: There are obstacles in the area to be searched. The obstacles are represented by irregular polygons. One endpoint of the dividing line is taken as the vertex of the irregular polygon, and at the same time, the dividing line for dividing the area is generated while satisfying the restricted constraint conditions of area division, dividing the area to be searched into N r sub-areas; The command and control center assigns sub-areas to each UAV according to the performance of the UAVs using a static allocation strategy; Step 4: After allocating sub-regions to each UAV, the command and control center plans the shortest cross-regional path for each UAV and uses a path planning method to plan the UAV coverage path to achieve the shortest time coverage path for a single region. Step 5: Monitor the battery life status of each UAV during the search mission execution. If the battery life is insufficient, recall the UAV. If the target is found or the search mission is completed, recall the UAV and return the target coordinates.

2. The multi-UAV collaborative maritime search path planning method according to claim 1, wherein The specific steps of Step 1 are as follows: Step 1.1: According to the area to be searched, obtain and analyze the ocean current information, and the ocean current information includes historical ocean current data, satellite observation data, ocean buoy data, and ocean weather station data. Step 1.

2. Based on the obtained ocean current information, establish a mathematical model of the ocean current; assume that the ocean current direction and magnitude intervals are [lb cd , ub cd and [lb cm , ub cm , and the ocean current model is as follows: where: c dN is the predicted value of the ocean current direction; c mN is the predicted value of the ocean current magnitude; υ is the uncertainty factor of the ocean current direction; σ is the uncertainty factor of the ocean current magnitude.

3. The multi-UAV collaborative maritime search path planning method according to claim 1, wherein, The specific steps of Step 2 are as follows: Step 2.1: The command and control center defines the working area of L*W with length L and width W as the search area with prior targets, and establishes a grid map with square grid cells as units according to the search area; the search area is decomposed into N square grid cells, and the set of all grid cells in the search area is represented as C all ={C1, C2…, C N}, where C i , i ∈ {1, 2,..., N} represents the i-th grid cell; Step 2.2: The command and control center calculates the probability of the existence of a prior target in each grid cell C based on the ocean current data analyzed by the ocean current model and the prior target information, and establishes a target probability map based on the grid map; the probability of the existence of a target in any grid cell C in the area to be searched i is represented by P i ; According to the prior target information and the ocean current data, the initial probability of the existence of a target in each grid cell, P i , is given; the sum of the initial probabilities of all grid cells in the target probability map is: i (0); In the target probability map, the probability that the target does not exist is 1 - P i ; For ease of processing, denote l(C i ) as the probability that the target exists in the grid cell:

4. The multi-UAV collaborative maritime search path planning method according to claim 3, wherein, When the UAV performs a search mission in the area to be searched, it feeds back search information and ocean current change information. The command and control center updates the target probability map in real time according to the information fed back by the UAV. The UAV has searched a certain grid cell C at time t i , and at time t, the grid cell C i has an observed target presence probability value of z; The target probability map update formula is as follows: In the formula, l0(C i ) is the grid cell C at the initial moment i Target existence probability value; l t (C i ) is the time when the drone completes searching grid cell C at time t. i The probability value updated later; α is the probability weight factor of the current direction; β is the probability weight factor of the current speed; ub cd is the maximum estimated value of the current direction; lb cd is the minimum estimated value of the current direction; ub cm is the maximum estimated value of the current speed; lb cm is the minimum estimated value of the ocean current velocity.

5. The multi-UAV collaborative maritime search path planning method according to claim 1, wherein In step 3, the area to be searched is assigned to R drones to perform the search task; the constraint conditions for the area division of the area to be searched are as follows: Where: N Z represents the number of possible target regions before region division, and by default, the value of N Z is 1; N′ Z Indicates the number of possible target regions after area division; S r Represents the area of the possible target area before area division; S′ r represents the area of the possible target region after the area division; N r Indicates the number of sub-regions into which the region is divided; A represents the area to be searched; A r Indicates the area allocated to the drone; A r′ Indicates the area not allocated to the drone; r′ represents the UAV number of the unallocated sub-region; r represents the UAV number of the allocated sub-region; In the equation, the value of the first objective function f1 represents the area to be searched segmentation coefficient; by minimizing the value of f1, fewer UAVs are allocated to the possible target areas. The value of the second objective function f2 represents the deviation of the workload of the UAVs; by calculating and minimizing the value of f2, the best workload is allocated to the UAVs to balance the workload among multiple UAVs; the union of all sub-regions should cover the area to be searched, and the sub-regions do not overlap with each other.

6. The multi-UAV collaborative maritime search path planning method according to claim 1, characterized in that In step 3, the performance characteristics of each UAV are considered, including endurance, flight speed, and payload capacity; after the area division is completed, a static allocation strategy is adopted to allocate sub-areas to each UAV according to the performance of the UAV, and during the search mission of the UAV, the sub-area allocation plan is dynamically adjusted according to the changes in real-time search information; for each sub-area C area , calculate the UAV evaluation function: Where: S i represents the search gain of allocating the sub-region to the k-th UAV; Represents the weight factor of the sub-region; ξ k represents the performance factor of the drone; R represents the number of UAVs; d k represents the Euclidean distance between the k-th UAV and the center of the sub-region; Calculate the search benefit of each UAV for each sub-region, and allocate the sub-region to the UAV with the highest benefit, and the UAV is responsible for performing the search mission for the sub-region.

7. The multi-UAV collaborative maritime search path planning method according to claim 1, wherein In Step 4, the command and control center plans the shortest cross-regional path for each UAV specifically as follows: The command and control center models the adjacency relationship between the N r sub-regions after division into a directed graph. In the directed graph, each sub-region is identified as a node; the ocean current information of the sub-region and the probability of the prior target being present are mapped into the weights of the directed edges; and the shortest cross-region path for each UAV is planned based on the comparison of the weights.

8. The multi-UAV collaborative maritime search path planning method according to claim 1, wherein In Step 4, the path planning method is used to plan the UAV coverage path to achieve the shortest time coverage path for a single region, specifically as follows: Step 4.1: The command and control center comprehensively analyzes the target probability map and the regional division allocation plan, issues a search mission to the UAVs, and initializes the grid map and the UAV search starting point. Step 4.2: The command and control center uses an intelligent optimization algorithm to plan the path of a single UAV; considering the sea current speed C d and direction C m , the range of the sea current speed is (lb cm , ub cm ), the range of the sea current direction is (lb cd , ub cd ), the Euclidean distance d between the current grid cell and the candidate grid cell. Considering the flight performance of the UAV, the distance threshold is d m , the probability p of the target existing in the candidate grid cell i , calculate the cost of the path and construct the conditional function G; S is the set of candidate grid cells; i is the candidate grid cell number; Step 4.3: Considering the sea current velocity and direction in each grid cell, the flight performance of the UAV itself, the Euclidean distance between the current position of the UAV and the candidate grid cell, and combining the probability of the target existing in the candidate grid cell, calculate the benefit of the UAV searching for the next grid cell, and construct an objective optimization function Is the weighted factor of the target existence probability in the candidate grid cell, γ is the weighted factor of the Euclidean distance between the current grid cell and the candidate grid cell, x is the weighted factor of the ocean current velocity, and δ is the weighted factor of the ocean current direction; based on the constructed conditional function, the sub-region search path is generated by calculating the target optimization function.

9. The multi-UAV collaborative maritime search path planning method according to claim 1, characterized in that The specific steps of Step 5 are as follows: Step 5.1: Real-time monitor the battery power, battery voltage, and battery temperature to estimate the remaining flight time of the UAV, and give an early warning or switch the UAV when the battery power is close to exhaustion. Step 5.2, if the target is found or the search task is completed, recall the UAV and return the target coordinates.

Images