A task planning method for manned / unmanned collaborative detection
Patent Information
- Application Number
- CN202610771697.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-06-01
- Publication Date
- 2026-09-15
Smart Images
Figure CN122759622A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the field of detection and sensing technology, specifically relating to a mission planning method for manned / unmanned collaborative detection, which is particularly suitable for collaborative mission planning of heterogeneous detection resources including UAVs, unmanned vehicles, and manned platforms. Background Technology
[0002] In modern complex land-based combat environments, the environmental situation is highly dynamic, and targets possess concealment, mobility, and deception capabilities. Comprehensive detection is crucial for gaining global information superiority. Multi-domain collaborative missions require the integration of air / ground and manned / unmanned detection resources to build flexible multi-platform collaborative situational awareness capabilities, rapidly and accurately acquire target and environmental information, and achieve comprehensive perception of the overall environmental situation. Currently, air / ground and manned / unmanned detection platforms are diverse, with heterogeneous resources operating independently and lacking collaborative utilization. It is necessary to coordinate the heterogeneous detection resources of various platforms within the detection system, flexibly combine equipment according to mission requirements, and form a highly distributed, flexible, and robust detection action plan.
[0003] Existing task planning methods have the following main shortcomings:
[0004] 1. The equipment selection, target allocation, node deployment, and path planning are all independent of each other and lack integrated coupling optimization, resulting in a suboptimal overall solution;
[0005] 2. Target allocation mostly adopts conventional clustering methods such as K-means or is based on simple distance cost, without considering the feedback impact of heterogeneous equipment capabilities, such as differences in detection accuracy and endurance time, on the allocation results;
[0006] 3. Moreover, traditional node deployment optimization methods such as particle swarm optimization and genetic algorithms have slow convergence speed and are difficult to handle complex scenarios with irregular regions and communication constraints.
[0007] 4. Existing solutions lack the ability to comprehensively model mission time windows and dynamic constraints on equipment endurance.
[0008] Therefore, there is an urgent need to provide an integrated mission planning method for manned / unmanned collaborative exploration, to overcome the above-mentioned shortcomings, achieve multi-platform complementarity and collaboration, effectively make up for the limitations of single-platform exploration, and improve the accuracy and real-time performance of global environmental situation perception. Summary of the Invention
[0009] To address the problems existing in the prior art, this invention proposes a task planning method for manned / unmanned collaborative exploration. By constructing a multi-stage coupled optimization model, introducing an improved fuzzy clustering-Hungarian allocation mechanism, an improved fruit fly optimization node deployment algorithm, and dynamic path planning under time window constraints, the method achieves precise scheduling and collaborative utilization of exploration resources, thereby improving task completion efficiency and global perception capabilities.
[0010] The technical solution of this invention is as follows:
[0011] A mission planning method for manned / unmanned collaborative exploration includes the following steps:
[0012] Step 1: Set equipment resource information and detection mission information, wherein the detection mission includes target detection mission and area detection mission;
[0013] Step 2: Construct an equipment selection model based on adaptive multi-attribute decision-making, dynamically adjust the weight factors according to the mission type, calculate the comprehensive suitability of each piece of equipment, establish an objective function that includes an endurance penalty term, solve it using the binary particle swarm optimization algorithm, and obtain the equipment combination participating in the mission.
[0014] Step 3: If it is a target detection mission, perform weighted constrained FCM clustering on the target points, introduce the target priority and equipment capability correction factor into the membership calculation to obtain several target groups; construct the cost matrix from the equipment to the center of the target group, and introduce the Lagrange relaxation penalty term of the endurance constraint, and use the extended Hungarian algorithm to solve the allocation scheme between the equipment and the target group.
[0015] Step 4: If it is a regional reconnaissance mission, the mission area is adaptively divided into grids, and a multi-objective optimization model including coverage, number of equipment used, average movement distance and communication connectivity penalty terms is established. The improved fruit fly optimization algorithm is used to solve the model and obtain the deployment location of each piece of equipment.
[0016] Step 5: Perform path planning for the selected equipment: For vehicles or personnel, adopt improved methods. The algorithm uses energy consumption as the actual cost and adds a time window violation penalty to generate a path that satisfies the endurance and time limit constraints; for drones, a three-dimensional approach is used in target detection scenarios. The algorithm avoids no-fly zones and obtains the regional detection mission path by combining sub-region segmentation with coverage path planning in the regional detection scenario.
[0017] Step 6: Integrate the outputs of Steps 2 to 5 to generate an action plan, which includes a task ID, equipment list, assignment relationship, and path list.
[0018] A further preferred embodiment, the formula for calculating the overall fitness in step 2 is:
[0019]
[0020] in For equipment Maximum detection radius; ; For equipment The number of detection methods carried; ; For equipment Positioning accuracy, ; For equipment The current remaining energy, ; For equipment The Euclidean distance from the initial position to the center of the mission area. ; For adaptive weights, satisfying It is dynamically adjusted according to the task type.
[0021] A further preferred embodiment, the objective function in step 2 is:
[0022]
[0023] in For decision variables; This is the penalty coefficient; For equipment The remaining battery life; The estimated duration of the mission.
[0024] A further preferred embodiment is that the membership degree calculation formula for the weighted constrained FCM clustering in step 3 is as follows:
[0025]
[0026] in To correct for distance:
[0027]
[0028] For target point To the The Euclidean distance between the cluster centers; This is the priority influence coefficient; For the goal priority, For equipment capability correction factors, For fuzzy index, This represents the number of clusters.
[0029] Further optimization scheme, equipment capability correction factor Defined as:
[0030]
[0031] in and These are the reciprocals of the average detection radius and average accuracy of the selected equipment, respectively. This is the adjustment coefficient.
[0032] A further preferred approach is to use the overall cost function obtained by the extended Hungarian algorithm in step 3 as follows:
[0033]
[0034] in For equipment To the target group The sum of the time cost and energy penalty at the center. As decision variables, Let Lagrange be the penalty coefficient. Indicates equipment Execution target group Total estimated time required For equipment Battery life.
[0035] A further preferred embodiment, the multi-objective optimization model in step 4, is as follows:
[0036]
[0037] in For joint coverage, This refers to the actual number of equipment deployed. The total number of selected equipment. The average distance the equipment travels from its initial position to its deployment position. This is a communication connectivity penalty term. As a weighting factor, satisfying .
[0038] A further preferred embodiment of the improved fruit fly optimization algorithm includes: using Logistic chaotic mapping for population initialization, adaptive decay of the olfactory search step size and introduction of the Levy flight strategy, and employing an elite retention strategy in the visual search phase.
[0039] A further preferred embodiment is the improvement described in step 5. The cost function of the algorithm is:
[0040]
[0041] in From the starting point to the node The actual energy consumption, This is a heuristic function used to estimate the number of nodes. The minimum residual cost to reach the target point. The penalty coefficient for violating the time window. To reach the node from the starting point The cumulative time, This is the deadline.
[0042] A further preferred embodiment, in step 5, specifically includes the path planning in the UAV area detection scenario:
[0043] The task area is uniformly discretized to generate a point set. The K-means algorithm is then used to cluster these points into subsets equal to the number of UAVs. For each subset, the minimum convex hull is calculated. For each convex hull, the scanning direction is determined to determine the detection radius. Scan lines are generated with a spacing of 10 times, and the endpoints of the scan line segments are connected in a zigzag order to form a coverage path. Finally, transition paths from the current position of the UAV to the endpoint of the first scan line and from the endpoint of the last scan line to the return point are added at the start and end of the path to obtain the UAV area detection mission path.
[0044] Beneficial effects
[0045] This invention achieves integrated coordination of equipment selection, target allocation / node deployment, and path planning by constructing a multi-stage coupled optimization model: In the equipment selection stage, an adaptive multi-attribute decision model is introduced, dynamically adjusting the weights of detection range, accuracy, and energy according to the mission type, and adding an endurance penalty term to make the selected equipment combination more closely match the actual mission requirements; in the target allocation stage, a weighted constraint FCM and extended Hungarian algorithm are used to integrate target priority and heterogeneous equipment capabilities (detection radius, positioning accuracy, remaining energy) into the soft clustering and assignment process, while introducing endurance constraints, significantly improving the feasibility of the allocation scheme and the mission success rate; in the regional detection node deployment stage, a fruit fly optimization algorithm improved by chaotic initialization, adaptive step size, and Levy flight, combined with a communication connectivity penalty term, is used to effectively improve convergence speed and coverage quality; in the path planning stage, through... The algorithm incorporates an energy consumption model and a time window penalty term to ensure the accessibility of vehicles / personnel under endurance and time constraints. The UAV employs sub-region segmentation and scanline coverage path generation methods to achieve full area coverage. Compared with existing technologies, this invention achieves advancements in overall mission planning optimality, handling of heterogeneous equipment constraints, and energy-time coupling optimization.
[0046] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description
[0047] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the description of the embodiments taken in conjunction with the following drawings, in which:
[0048] Figure 1 This is a schematic diagram of a planning method for unmanned collaborative exploration missions according to the present invention.
[0049] Figure 2 This is a schematic diagram of the regional detection mission of the present invention.
[0050] Figure 3 This is a schematic diagram of the target detection mission of the present invention.
[0051] Figure 4 This is a schematic diagram of target allocation in this invention.
[0052] Figure 5 This is a schematic diagram of the node layout of the present invention.
[0053] Figure 6 This is a schematic diagram illustrating the task planning results for regional exploration missions according to the present invention.
[0054] Figure 7 This is a schematic diagram of the task planning results for target detection tasks according to the present invention. Detailed Implementation
[0055] With the increase in the types and quantities of detection equipment, the growing demand for collaborative use of detection resources, and the improvement of intelligent technologies, there is an urgent need to optimize the scheduling of detection resources, ensure the most comprehensive coverage of relevant mission areas according to the mission, and rationally allocate detection resources to target locations to achieve all-round perception of the mission area.
[0056] To this end, this invention proposes a mission planning method for manned / unmanned collaborative exploration, including equipment resource and mission settings, equipment selection, target allocation, node deployment, path planning, and scheme generation. The specific steps are as follows:
[0057] Step 1: Equipment Resources and Task Settings:
[0058] The system acquires equipment database data and user-input mission data. The equipment database contains information such as equipment ID, type, detection range, movement speed, positioning accuracy, endurance, remaining energy, outline dimensions, and initial position. Mission data includes mission type, such as target detection or area detection, and corresponding mission parameters.
[0059] The specific content includes:
[0060] Step 1.1: Equipment Resource Settings: Initialize the set of available equipment. , This represents the total number of available equipment. For each piece of equipment... ,Record:
[0061] Equipment ID and type, such as vehicle, personnel, drone, etc.;
[0062] Detection range Unit: meters;
[0063] movement speed Unit: meters per second;
[0064] Positioning accuracy Unit: meters; the smaller the value, the higher the precision.
[0065] Battery life Unit: seconds, remaining working time from the current moment;
[0066] Remaining energy , is a dimensionless normalized value, and its full energy is 1;
[0067] Outline dimensions, including length, width, and height, are used for path planning and obstacle avoidance;
[0068] Initial position coordinates .
[0069] Step 1.2: Task settings: Enables target detection and area detection tasks.
[0070] Target detection mission: The mission parameters set include:
[0071] Task ID; Target Set Each target Includes target ID, type, and location coordinates. Priority (Set by the user based on the importance of the target); Task time window .
[0072] Area exploration mission: The mission parameters set include:
[0073] Task ID; Polygon region vertex set Task Time Window Minimum coverage requirement .
[0074] This step ultimately outputs a structured collection of equipment. And task parameters.
[0075] Step 2: Equipment selection based on adaptive multi-attribute decision-making:
[0076] The input for this step is the equipment set output from step 1. And mission parameters; mission type, such as target detection mission or area detection mission; minimum number of equipment set by the user. The equipment combination for this mission is obtained through adaptive multi-attribute decision-making. Equipment quantity And the overall suitability of each selected piece of equipment. The specific content includes:
[0077] Step 2.1: Calculation of overall equipment suitability
[0078] equipment Overall suitability for the current task Calculated by the following formula:
[0079]
[0080] in For equipment Maximum detection radius; ; For equipment The number of detection methods carried, such as visible light, infrared, radar, etc.; ; For equipment The positioning accuracy is such that the smaller the value, the higher the accuracy. Reflecting the advantage of accuracy, ; For equipment The current remaining energy is normalized and ranges from 0 to 1; , usually 1; For equipment The Euclidean distance from the initial position to the center of the mission region; for target detection missions, the center of the mission region is the geometric center of the target set; for region detection missions, the center of the mission region is the centroid of the region polygon. .
[0081] For adaptive weights, satisfying Furthermore, the weight vector is dynamically adjusted based on the task type. :
[0082] For regional exploration missions, the focus is on covering a wide area, therefore, increasing the weight of the exploration range is crucial. ,For example , , , , .
[0083] For target detection tasks, the focus is on accurate localization and identification of high-value targets; therefore, increasing the weight of accuracy is crucial. And the weight of the number of detection methods ,For example , , , , .
[0084] If the task time is tight, the distance penalty weight can be appropriately reduced. Increase energy weight To ensure equipment availability.
[0085] The specific weight values can be preset by the user based on experience, or obtained by combining the Analytic Hierarchy Process (AHP) with expert scoring, and fine-tuned according to the user's intentions before each task.
[0086] Step 2.2: Equipment Selection Optimization Model:
[0087] Construct the following 0-1 integer programming model:
[0088]
[0089] The constraints are as follows:
[0090]
[0091] In the model, As decision variables, Indicates selecting equipment Otherwise, it is 0; For equipment Overall suitability; This is a penalty coefficient, taken as a positive number, used to penalize equipment with insufficient battery life. In this embodiment, it is taken as... ; For equipment The remaining battery life; The estimated duration of the task, either input by the user or estimated based on the task size; This indicates that if the battery life is less than the mission duration, a positive penalty is applied; otherwise, it is 0. The minimum number of equipment required to complete the mission is set by the user based on the mission's complexity.
[0092] In this optimization model, the objective function is... The optimization model maximizes the overall suitability of all selected equipment while deducting penalties for insufficient endurance. This approach selects powerful equipment while avoiding equipment unable to complete the entire mission. In this embodiment, an improved binary particle swarm optimization (BPSO) algorithm is used to solve this optimization model, yielding the equipment combination. BPSO is a discrete version of the standard particle swarm optimization algorithm and is a well-known algorithm in this field.
[0093] Step 3: For the target detection task, target allocation is performed using the improved FCM-Hungarian algorithm:
[0094] The input for this step is the equipment combination obtained in step 2. The number of equipment is ; Target set in step 1 The target quantity is The battery life of each piece of equipment ,speed ; Target Priority Task time window. Using the above input, target allocation is performed using the improved FCM-Hungarian algorithm to obtain the equipment-target group allocation scheme, that is, which equipment each target is assigned to, and the list of target groups assigned to each equipment. The specific process is as follows:
[0095] Step 3.1: Weighted Constraint FCM Clustering:
[0096] First, perform soft clustering on the target points to obtain... There are n target groups, and the number of cluster centers is 1. ,in The maximum number of clusters set for the system.
[0097] The membership formula for traditional FCM is:
[0098]
[0099] This invention introduces target weight, i.e. target priority. and equipment capability modification factor First, prioritize the target. As a correction factor for distance, the corrected distance is defined as follows:
[0100]
[0101] in For target point To the The Euclidean distance between the cluster centers; As the priority influence coefficient, the correction distance of high-priority targets is compressed, making them more easily attracted to cluster centers. Then, using the equipment capability correction factor, the improved membership calculation formula is obtained as follows:
[0102]
[0103] in For cluster center index, ; Index the target point. ; To obtain the summation index, traverse all cluster centers; For target point Belonging to the The membership degree of each cluster center satisfies ; The fuzzy index is taken in this embodiment. To control the degree of fuzziness in clustering; The equipment capability correction factor is defined as follows:
[0104]
[0105] in and These are the reciprocals of the average detection radius and average accuracy of the selected equipment, respectively. As the adjustment coefficient, in this embodiment, we take... , The larger the value, the stronger the equipment's capabilities, and the more targets its corresponding cluster center will attract during clustering.
[0106] Iterative calculation of membership matrix The cluster centers are updated until convergence. Finally, based on the principle of maximum membership, each target point is assigned to its most joined class, forming a cluster. target groups Each target group contains several target points.
[0107] It should be noted that the number of cluster centers Usually not greater than the number of selected equipment. To adjust the equipment capability correction factor Introducing a clustering process, Selected equipment is categorized by overall suitability. Sort in descending order and take the first few. The equipment capability modification factor for each piece of equipment is assigned sequentially to... There are several cluster centers, so that the cluster centers corresponding to high-capability equipment will attract more high-priority targets, making the clustering results better match the subsequent allocation.
[0108] Step 3.2: Construct the equipment-target group cost matrix:
[0109] For each target group , Calculate its center position:
[0110]
[0111] in Indicates the target point Location coordinates. Equipment , The cost of reaching the center is:
[0112]
[0113] in For equipment Current location to target group center The Euclidean distance; This is an energy penalty factor used to encourage the selection of equipment with high remaining energy. In this embodiment... , The energy sensitivity coefficient is used in this embodiment. ;
[0114] Step 3.3: Solve the many-to-many assignment using the extended Hungarian algorithm:
[0115] Since a single piece of equipment can execute multiple target groups, but in the optimal state a target group is executed by only one piece of equipment (to avoid duplicate probing), this is a typical multi-assignment problem. Therefore, the equipment set is copied. Next, construct the extended cost matrix. for dimension:
[0116]
[0117] Since the expanded cost matrix is OK, Columns, usually ,like Then add The virtual rows are used to transform the matrix into a square matrix. Furthermore, this is because for any equipment... If it is assigned to multiple target groups, the total detection time for each target group (including travel time to each center and detection time spent in each target group's area) must not exceed its remaining endurance. Therefore, a Lagrangian relaxation penalty term is introduced for the range constraint, thus defining the penalized overall cost function. as follows:
[0118]
[0119] in For matrix The elements in As a decision variable, when Indicate target group Assigned to equipment ; The Lagrange penalty coefficient is applied when the total mission time of a piece of equipment exceeds its endurance. Indicates equipment Execution target group The total estimated time required includes:
[0120]
[0121] Multiply the dwell time at each target point by the target group. The target number.
[0122] exist Minimize the overall cost function under the constraints. The Hungarian algorithm is used to solve the problem, and the final distribution matrix is obtained. If some target groups cannot be assigned under the constraints, return to step 2 to select additional equipment.
[0123] Step 4: For the region exploration task, deploy nodes based on the improved fruit fly optimization algorithm:
[0124] The input for this step is the equipment combination obtained in step 2. The number of equipment is The set of vertices of the region polygon in step 1 Minimum coverage requirement Detection range of each piece of equipment and initial position The deployment location of each selected piece of equipment was calculated using an improved fruit fly optimization algorithm. and estimated coverage The specific process is as follows:
[0125] Step 4.1: Adaptive Mesh Generation:
[0126] Task area The bounding rectangle is divided into a grid. Grid size. This ensures that each mesh cell is completely covered by at least one piece of equipment. The set of mesh nodes is denoted as... .
[0127] Step 4.2: Construct a multi-objective optimization model:
[0128] With each piece of equipment Deployment location As a decision variable, and with its deployment location continuously changing, while the z-coordinate is determined by the terrain elevation, the optimization objective is to minimize the following weighting function:
[0129]
[0130] in For combined coverage:
[0131]
[0132] In the formula A grid node represents a discrete sampling location within the task region after adaptive grid partitioning. From mesh nodes to equipment Deployment location The Euclidean distance. Joint coverage rate represents the number of grid nodes that satisfy the condition that "there is at least one piece of equipment that can cover the grid node", that is, the proportion of the total number of grid nodes covered by at least one piece of equipment to the total number of grid nodes.
[0133] This refers to the actual number of equipment deployed. The total number of selected equipment. .
[0134] The average distance the equipment travels from its initial position to its deployment position:
[0135]
[0136] The length of the diagonal of the task region, used for normalization;
[0137] This is a communication connectivity penalty. It requires the network consisting of all deployed nodes to be connected; if not connected, then... Otherwise, it is 0.
[0138] As a weighting factor, satisfying In this embodiment, coverage is prioritized, therefore , , , .
[0139] Step 4.3: For the constructed multi-objective optimization model, the improved fruit fly optimization algorithm is used for solution:
[0140] The standard fruit fly optimization algorithm was proposed in 2012 for continuous function optimization. This embodiment mainly incorporates three improvements:
[0141] Step 4.3.1: Chaotic Initialization: An initial population is generated using a Logistic chaotic mapping. Chaotic sequence formula:
[0142]
[0143] In this embodiment , Random. Mapping chaotic variables to a defined region:
[0144]
[0145] Similarly, each individual fruit fly corresponds to a group. Each deployment location.
[0146] Step 4.3.2: Olfactory Search (Improved Step Size): Each individual randomly searches near its current location, using an adaptive step size:
[0147]
[0148] in As the initial step size, in this embodiment, it is taken as 0.1 times the width of the region. This represents the current iteration number. The maximum number of iterations, The decay exponent is used; at the same time, the Levy flight strategy is introduced to make large step jumps with a certain probability to avoid getting trapped in local optima.
[0149] Step 4.3.3: Visual Search (Elite Preservation): Calculate the fitness of each individual (lower is better), select the best individual in the population as the visual target, and move the remaining individuals toward the best individual. At the same time, retain the top E elite individuals unchanged.
[0150] Iterate until the maximum number of iterations is reached or the fitness change is less than a threshold. Output the set of placement locations corresponding to the optimal individual. and verify If the conditions are not met, the weights will be adjusted. After increasing the coverage ratio, we re-optimized.
[0151] Step 5: Path planning that couples time windows and energy constraints:
[0152] For target detection missions, the input for this step is the equipment. Current location The list of assigned target points or the center of the target group, the task time window, and the digital elevation map of the task area.
[0153] For regional exploration missions, the input for this step is the equipment. Current location Step 4 output deployment location Task area polygon, task time window.
[0154] Path planning yields the movement path for each piece of equipment and the estimated arrival time at each key point. Specific details include:
[0155] Step 5.1: Based on the improvement The algorithm performs path planning for vehicles or people:
[0156] This embodiment is based on the traditional... The algorithm's cost function is improved by adding a time window constraint and an energy consumption model. The improved cost function is as follows:
[0157] For nodes in the search space The cost function is defined as follows:
[0158]
[0159] in To go from the starting point through the node The estimated total cost to reach the target point.
[0160] From the starting point to the node The actual energy consumption is calculated and accumulated from the length, gradient, and unit energy consumption of each segment of the route:
[0161]
[0162] From the starting point to the node The number of edges traversed For the first Energy consumption along the edge. For vehicles, ,in Energy consumption per unit distance in flat terrain. The slope energy consumption coefficient, This represents the slope of the road section (a positive value indicates an uphill slope). A similar definition applies to personnel, but the coefficients differ. In practice, pre-calculated elevation data can be used to obtain the slope.
[0163] This is a heuristic function used to estimate the number of nodes. The minimum remaining cost to reach the target point. This invention uses the Euclidean distance divided by the maximum speed as an estimate of the remaining time, then multiplies it by the baseline energy consumption rate:
[0164]
[0165] To equip maximum speed, This represents the baseline energy consumption per unit time in flat terrain.
[0166] The time window violation penalty coefficient is set to 100 in this embodiment, which is used to constrain the arrival of the target point before the deadline.
[0167] To reach the node from the starting point The cumulative time is obtained by adding up the travel times of each side of the path:
[0168]
[0169] For this section of road The average speed on the surface.
[0170] The deadline is specified, where if a node... If it is the mission objective point (i.e., the location where probing needs to be performed), then That is, the task deadline; if the node If it is an intermediate node, then That is, there are no hard time constraints.
[0171] Using the above cost function, according to the standard The algorithm's iterative framework is used for searching:
[0172] 1: Initialize the open set (containing only the starting point) and the closed set (empty).
[0173] 2: Repeat the following steps until the endpoint is found or the open set is empty:
[0174] 2.1: Selecting from open sets The smallest node is selected as the current node.
[0175] 2.2: If the current node is the target point, backtrack the parent node to generate the path; otherwise, generate the neighbor nodes of the current node, calculate g, arrival time and penalty term for each neighbor to obtain the f value, and update the open set.
[0176] 3. During the search process, if the current cumulative energy consumption... If the total energy of the equipment exceeds 80%, the nearest supply point will be automatically inserted into the path as an intermediate target point.
[0177] The final result is a sequence of path points from the starting point to the target point.
[0178] Step 5.2: Perform UAV path planning:
[0179] Drone path planning is divided into two scenarios based on mission type: target detection missions and area detection missions. These are as follows:
[0180] Step 5.2.1: Target Detection Task:
[0181] The drones need to fly sequentially to one or more assigned target points (or the center of a target group). A three-dimensional [system / mechanism] is used. Algorithms, in traditional Based on this, elevation dimension and no-fly zone constraints are added.
[0182] The cost function is defined as:
[0183]
[0184] in From the starting point to the node The actual flight distance; since the energy consumption of a drone is mainly related to distance, distance is used here instead of energy consumption. The heuristic function is represented by a three-dimensional Euclidean distance.
[0185]
[0186] in For nodes Three-dimensional spatial coordinates, The three-dimensional spatial coordinates of the target point; This refers to the penalty coefficient for no-fly zones. For a binary function, if the node If the object is located inside any no-fly zone polygon or at a distance less than the safety margin from the no-fly zone boundary, the value is 1; otherwise, the value is 0.
[0187] Then, using the aforementioned cost function, and also according to the standard... The algorithm's iterative framework searches and ultimately obtains a sequence of path points from the starting point to the target point.
[0188] Step 5.2.2: Regional Exploration Mission
[0189] When a drone needs to perform full-coverage reconnaissance of a designated area, a sub-region segmentation + coverage path planning method is used, with the following steps:
[0190] 1. Discretize the points inside the polygon of the task region uniformly to generate a dense point set. The K-means algorithm is used to divide the point set Clustering A subset, of which Number of available drones. Number of clusters. Each subset This corresponds to the area of responsibility of a single drone.
[0191] 2. For each subset Calculate its minimum convex hull The vertices of the convex hull are arranged in counter-clockwise order. The convex hull is defined as the smallest convex polygon containing all points of that subset. The purpose of the convex hull is to simplify the boundaries of covered paths and avoid redundant coverage caused by concave regions.
[0192] 3. For each convex hull The scanline coverage algorithm is used to generate the coverage path. This is an existing method, and its main process is as follows:
[0193] Determine the scanning direction: Calculate the minimum bounding rectangle of the convex hull, and use the major axis of the rectangle as the scanning direction; use the probe radius... The scan line spacing is calculated, and a set of parallel lines is generated along a direction perpendicular to the scanning direction, with the scan line spacing as the step size. Each line intersects the convex hull boundary to obtain line segments. These line segments are then connected in a zigzag order to form a sequence of covered path points. Simultaneously, it is ensured that straight lines between adjacent path points do not cross no-fly zones. Finally, transition paths are added at the start and end points of the path, from the UAV's current position to the endpoint of the first scan line, and from the endpoint of the last scan line to the return point, thus obtaining the UAV's area reconnaissance mission path. If the total flight time exceeds the UAV's endurance, the path is divided into multiple segments, and the UAV returns to the resupply point after each segment.
[0194] Step 6: Integrate all outputs from Steps 2 to 5 to generate an action plan:
[0195]
[0196] in For task ID; The list of equipment IDs selected for step 2; For target detection missions, the assignment relationship between equipment and its assigned tasks or locations is defined as follows: For area reconnaissance missions, the mapping from equipment IDs to target group lists determined based on the target allocation results obtained in step 3 is used. The mapping from equipment ID to deployment location is determined based on the node deployment results obtained in step 4. This is the path point sequence for each piece of equipment obtained in step 5.
[0197] The following is in conjunction with the appendix Figure 6 and attached Figure 7 Two examples are given: a target detection task and a region detection task.
[0198] Target detection mission:
[0199] A task area is defined with 10 target points to be detected, with uneven priority distribution. Available equipment includes 3 UAVs (detection radius 200m, speed 20m / s, remaining endurance 1200s) and 2 unmanned vehicles (UAVs) (detection radius 100m, speed 10m / s, remaining endurance 1800s). The task time window is one hour after the current moment. Step 1 sets the task; Step 2 calculates the overall fitness and selects 2 UAVs and 2 UAVs. Step 3 uses weighted FCM clustering to obtain 4 target groups; after allocation using the extended Hungarian algorithm, the UAVs and UAVs are responsible for their respective target groups. Step 5 plans the paths for each piece of equipment, considering the time window constraint: the UAVs directly hover over the target points for detection, while the UAVs travel along roads. Step 6 generates a plan, such as... Figure 7 As shown.
[0200] Regional exploration mission:
[0201] A flat hexagonal region is defined, requiring full coverage detection. Equipment used for the same target detection mission can be employed. Step 2 employs region detection weighting, selecting all 3 UAVs and 2 unmanned vehicles. Step 4 uses adaptive grid partitioning with a Δ=50m grid and approximately 400 grid points. An improved fruit fly optimization population size of 30 is used, with 100 iterations, chaotic initialization, and an initial step size of 0.1 times the region width. The final deployment position achieves 96% coverage, an average movement distance of 125m, and communication connectivity. Step 5 plans coverage paths for the UAVs (K-means++ clustering into 3 sub-regions, generating scanline paths for each), and plans the shortest path from the starting position to the deployment point for the unmanned vehicles. Step 6 generates the scheme, such as... Figure 6 As shown.
[0202] Although embodiments of the present invention have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting the present invention. Those skilled in the art can make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention without departing from the principles and spirit of the present invention.
Claims
1. A mission planning method for manned / unmanned collaborative exploration, characterized in that: Includes the following steps: Step 1: Set equipment resource information and detection mission information, wherein the detection mission includes target detection mission and area detection mission; Step 2: Construct an equipment selection model based on adaptive multi-attribute decision-making, dynamically adjust the weight factors according to the mission type, calculate the comprehensive suitability of each piece of equipment, establish an objective function that includes an endurance penalty term, solve it using the binary particle swarm optimization algorithm, and obtain the equipment combination participating in the mission. Step 3: If it is a target detection mission, perform weighted constrained FCM clustering on the target points, introduce the target priority and equipment capability correction factor into the membership calculation to obtain several target groups; construct the cost matrix from the equipment to the center of the target group, and introduce the Lagrange relaxation penalty term of the endurance constraint, and use the extended Hungarian algorithm to solve the allocation scheme between the equipment and the target group. Step 4: If it is a regional reconnaissance mission, the mission area is adaptively divided into grids, and a multi-objective optimization model including coverage, number of equipment used, average movement distance and communication connectivity penalty terms is established. The improved fruit fly optimization algorithm is used to solve the model and obtain the deployment location of each piece of equipment. Step 5: Perform path planning for the selected equipment: For vehicles or personnel, adopt improved methods. The algorithm uses energy consumption as the actual cost and adds a time window violation penalty to generate a path that satisfies the endurance and time limit constraints; for drones, a three-dimensional approach is used in target detection scenarios. The algorithm avoids no-fly zones and obtains the regional detection mission path by combining sub-region segmentation with coverage path planning in the regional detection scenario. Step 6: Integrate the outputs of Steps 2 to 5 to generate an action plan, which includes a task ID, equipment list, assignment relationship, and path list.
2. The mission planning method for manned / unmanned collaborative exploration according to claim 1, characterized in that: The formula for calculating the overall fitness in step 2 is as follows: in For equipment Maximum detection radius; ; For equipment The number of detection methods carried; ; For equipment Positioning accuracy, ; For equipment The current remaining energy, ; For equipment The Euclidean distance from the initial position to the center of the mission area. ; For adaptive weights, satisfying It is dynamically adjusted according to the task type.
3. The mission planning method for manned / unmanned collaborative exploration as described in claim 2, characterized in that: The objective function mentioned in step 2 is: in For decision variables; This is the penalty coefficient; For equipment The remaining battery life; The estimated duration of the mission.
4. The mission planning method for manned / unmanned collaborative exploration according to claim 1, characterized in that: The membership calculation formula for the weighted constrained FCM clustering mentioned in step 3 is as follows: in To correct for distance: For target point To the The Euclidean distance between the cluster centers; This is the priority influence coefficient; For the goal priority, For equipment capability correction factors, For fuzzy index, This represents the number of clusters.
5. The mission planning method for manned / unmanned collaborative exploration according to claim 4, characterized in that: Equipment capability modification factor Defined as: in and These are the reciprocals of the average detection radius and average accuracy of the selected equipment, respectively. This is the adjustment coefficient.
6. The mission planning method for manned / unmanned collaborative exploration according to claim 1, characterized in that: The overall cost function obtained by the extended Hungarian algorithm in step 3 is: in For equipment To the target group The sum of the time cost and energy penalty at the center. As decision variables, Let Lagrange be the penalty coefficient. Indicates equipment Execution target group Total estimated time required For equipment Battery life.
7. The mission planning method for manned / unmanned collaborative exploration according to claim 1, characterized in that: The multi-objective optimization model mentioned in step 4 is: in For joint coverage, This refers to the actual number of equipment deployed. The total number of selected equipment. The average distance the equipment travels from its initial position to its deployment position. This is a communication connectivity penalty term. As a weighting factor, satisfying .
8. The mission planning method for manned / unmanned collaborative exploration according to claim 1, characterized in that: The improved fruit fly optimization algorithm includes: using Logistic chaotic mapping for population initialization, adaptive decay of the olfactory search step size and introduction of the Levy flight strategy, and using an elite retention strategy in the visual search phase.
9. The mission planning method for manned / unmanned collaborative exploration according to claim 1, characterized in that: The improvement described in step 5 The cost function of the algorithm is: in From the starting point to the node The actual energy consumption, This is a heuristic function used to estimate the number of nodes. The minimum residual cost to reach the target point. The penalty coefficient for violating the time window. To reach the node from the starting point The cumulative time, This is the deadline.
10. The mission planning method for manned / unmanned collaborative exploration according to claim 1, characterized in that: Step 5, the path planning in the UAV area detection scenario, specifically includes: The task area is uniformly discretized to generate a point set. The K-means algorithm is then used to cluster these points into subsets equal to the number of UAVs. For each subset, the minimum convex hull is calculated. For each convex hull, the scanning direction is determined to determine the detection radius. Scan lines are generated with a spacing of 10 times, and the endpoints of the scan line segments are connected in a zigzag order to form a coverage path. Finally, transition paths from the current position of the UAV to the endpoint of the first scan line and from the endpoint of the last scan line to the return point are added at the start and end of the path to obtain the UAV area detection mission path.