A Multi-UAV Multi-Task Allocation Method Based on an Improved Chemical Reaction Optimization Algorithm
By combining improved chemical reaction optimization algorithms with Dubins curves, the problems of premature convergence and path infeasibility in UAV target allocation were solved, achieving efficient and reliable allocation of multi-UAV missions and improving mission execution efficiency in battlefield environments.
Patent Information
- Application Number
- CN202611062690.2
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-17
- Publication Date
- 2026-11-17
AI Technical Summary
Existing UAV target allocation and path planning methods suffer from problems such as premature convergence, loss of population diversity, high computational complexity, and path infeasibility in complex battlefield environments, making it difficult to meet real-time decision-making requirements. Furthermore, traditional methods do not consider the flight dynamics characteristics of UAVs, resulting in poor engineering feasibility of the solutions.
An improved chemical reaction optimization algorithm is adopted. By pre-setting a unified cost function with multiple constraints, and combining Dubins curves and PSO search operators, segmented trajectories that conform to the flight dynamics characteristics of UAVs are generated. A unified cost function containing path cost and multi-dimensional constraint penalty terms is constructed to achieve reliable allocation of multiple UAVs and multiple tasks.
It improves the feasibility and adaptability of multi-UAV task allocation, ensures the actual feasibility of paths, reduces computational complexity, and enhances dynamic response capabilities and task execution efficiency.
Smart Images

Figure FT_1 
Figure FT_2 
Figure FT_3
Abstract
Description
Technical Field
[0001] This invention relates to the field of path allocation technology, and more specifically to a multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm. Background Technology
[0002] With the rapid development of UAV technology, fixed-wing UAV swarms, due to their long endurance, high speed, and strong collaborative operation capabilities, have been widely used in complex combat scenarios such as battlefield reconnaissance, target tracking, and precision strikes. Multi-UAV, multi-task allocation is the core of UAV swarm command and control. Essentially, it involves finding the optimal pairing scheme between UAVs and mission targets under multiple constraints, including UAV flight performance, mission matching, and environmental limitations. This directly determines the swarm's mission execution efficiency and overall combat effectiveness.
[0003] Current UAV target allocation and path planning methods have shortcomings in complex battlefield environments. First, traditional intelligent optimization algorithms are prone to premature convergence and loss of population diversity in high-dimensional allocation scenarios involving multiple UAVs and multiple targets, resulting in low solution efficiency and numerous iterations. Especially when the UAV-to-target ratio increases and the target size expands, the computational complexity increases exponentially, making it difficult to meet the real-time decision-making requirements in battlefield environments. Moreover, most existing task allocation methods use Euclidean distance as the basis for calculating range cost, without considering the flight dynamics characteristics of UAVs, such as minimum turning radius, heading constraints, and climb rate limits. The theoretical shortest path planned does not conform to the actual flight rules of fixed-wing UAVs, leading to large deviations in cost calculation and problems such as exceeding range limits and unreachable headings during actual execution, resulting in poor engineering feasibility of the solutions.
[0004] Therefore, there is an urgent need for a multi-UAV multi-task allocation method that can take into account the feasibility of the scheme, multi-task adaptability and dynamic response capability, so as to solve the above-mentioned defects of the existing technology. Summary of the Invention
[0005] To overcome the problems in the prior art, the present invention aims to provide a multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm. This method can reliably allocate multiple UAVs and tasks in complex environments by pre-setting a unified cost function with multiple constraints, pre-processing multi-objective clustering, and calculating path distance based on the UAV's turning flight parameters.
[0006] To achieve the above objectives, this invention provides a multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm, comprising the following steps:
[0007] Step S1: Classify and model various types of UAV missions, and preset UAV flight performance constraints, payload matching constraints, and environmental constraints;
[0008] Step S2: Perform spatial clustering on multi-task objectives to generate clusters and Standoff tracking reference points, and set feasible domain constraints for task allocation.
[0009] Step S3: Based on the minimum turning radius constraint of the UAV, use Dubins curves to generate segmented feasible tracks and calculate the shortest path length from each UAV to each target;
[0010] Step S4: Construct a unified cost function that includes path cost and multi-dimensional constraint penalty terms, and use an improved CRO algorithm that integrates PSO search operators to iteratively solve the optimal UAV-target task allocation scheme under feasible region constraints;
[0011] Step S5: Generate a task allocation scheme that satisfies the full coverage constraint and the full participation constraint, and simultaneously output the Dubins track corresponding to each UAV.
[0012] Furthermore, in step S2, the K-means algorithm is used to generate clusters, specifically by: sorting the N targets according to their spatial location. The direction of motion is divided into several clusters, with the cluster center... This is determined by minimizing the sum of squared distances, as expressed in the formula:
[0013] ;
[0014] in, This represents the total number of task objectives. For the first The location coordinates of the target The first The x and y coordinates of each target in the reference coordinate system Let the coordinates of the cluster center of the k-th cluster be . This represents the clustering loss value.
[0015] Furthermore, in step S2, the Standoff tracking center is calculated using a weighted distance matrix and a density index, specifically including:
[0016] Step S21: Represent the relationship between targets using a weighted matrix, expressed by the formula:
[0017] ;
[0018] in, For the first The first goal and the first Distance weights between targets For the first The first goal and the first Euclidean distance function between targets and The first The first goal and the first The x-coordinate of each target; and The first The first goal and the first The vertical coordinate of each target;
[0019] Step S22: By calculating the first The sum of the distances of a target to all other targets represents the density index of each target, expressed by the formula:
[0020] ;
[0021] in, For the first The density index of each target The total number of task objectives;
[0022] S23: The Standoff tracking center is calculated using the target location weighted by average density, expressed by the following formula:
[0023]
[0024] in, For density-weighted target centers, For the first The density index of each target For the first The planar position coordinates of the target.
[0025] Furthermore, in step S2, the feasible domain constraints for setting task allocation include full coverage constraints and full participation constraints.
[0026] The full coverage constraint specifically means that any task target must be within the effective coverage radius of at least one drone, ensuring that all tasks are executed by drones without any tasks being missed. This can be expressed as: ;
[0027] The full participation constraint specifically states that at least one mission objective exists within the effective coverage radius of any UAV, ensuring that all UAVs in the cluster participate in mission execution with no resource idleness. This can be expressed as: ;
[0028] R represents the effective coverage radius of the drone's mission. For the first The location coordinates of the target Let Z be the position coordinates of the z-th UAV. This represents the total number of task objectives.
[0029] Furthermore, in step S3, the segmented feasible track generated by using Dubins curves is composed of two circular arc segments and one straight line segment. The path types include four types: LSL, LSR, RSL, and RSR, where L represents a left-turn circular arc segment, S represents a straight line segment, and R represents a right-turn circular arc segment.
[0030] Furthermore, the specific method for generating a single Dubins path is as follows:
[0031] Step S31: Model the Dubins curve path. Input parameters include the UAV's starting point coordinates, starting point heading angle, target point coordinates, ending point heading angle, and the UAV's minimum turning radius.
[0032] Step S32: Calculate the coordinates of the center of the arc segment's trajectory;
[0033] For the starting position, the coordinates of the center of the clockwise turning arc segment The coordinates of the center of the counterclockwise turning arc segment The calculation formulas are expressed as follows:
[0034]
[0035] in, The coordinates of the drone's starting plane; R is the starting heading angle of the drone; R is the minimum turning radius of the drone.
[0036] Similarly, for the endpoint, the coordinates of the center of the clockwise turning arc segment... The coordinates of the center of the counterclockwise turning arc segment The formulas are expressed as follows:
[0037]
[0038] in, The endpoint plane coordinates of the mission objective; R is the final heading angle for the UAV to reach the target point; R is the minimum turning radius of the UAV.
[0039] Step S33: Calculate the straight line segment path using the common tangent of the two arcs. The calculation method for the common tangent parameter is as follows: Define the center of the starting arc as... The center of the arc at the endpoint is Let D be the distance between the center of the starting arc and the center of the ending arc, and let the angle between the line connecting the two centers and the horizontal axis of the reference coordinate system be θ. ,but:
[0040]
[0041]
[0042] in, The coordinates of the center of the starting side arc are given. The coordinates of the center of the arc on the endpoint side;
[0043] Step S34: Determine the path type. When two arcs turn in the same direction, use the external common tangent. The angle between the line connecting the center of the circle to the point of tangency and the tangent of the line connecting the two centers is... satisfy ;
[0044] When two circular arcs turn in opposite directions, the internal common tangent is used, and the angle between the line connecting the center of the arc to the point of tangency and the tangent to the line connecting the two centers is used. satisfy: ;
[0045] Where R is the minimum turning radius of the drone; D is the distance between the centers of the two arcs;
[0046] Step S35: Calculate the coordinates of the tangent point of the arc segment using the polar angle of tangency, specifically as follows:
[0047] Coordinates of the tangent point of the starting side arc The calculation formula is expressed as follows:
[0048] ;
[0049] Coordinates of the tangent point of the arc at the endpoint The calculation formula is expressed as:
[0050] ;
[0051] in, When the polar angle of the tangent point is the path of the external common tangent, Conversely, if the two trajectories are in opposite directions, then the path is the internal common tangent. ; , This represents the minimum turning radius of the drone.
[0052] Step S36: Calculate the total length of a single Dubins path based on the starting point coordinates, the ending point coordinates, the tangent point coordinates of the starting point side arc and the ending point side arc. The total length of a single Dubins path is the sum of the length of the starting point arc segment, the length of the straight line segment, and the length of the ending point arc segment.
[0053] Furthermore, the unified cost function is defined as a weighted sum of multi-dimensional constraint penalty terms, which include the Dubins path length for the UAV to perform the assigned task, a range exceeding the limit penalty term, a task type mismatch penalty term, and an obstacle avoidance penalty term, expressed as:
[0054] ;
[0055] in, The total path length for assigning tasks to all drones. Path length weight; This is a penalty for exceeding the flight range limit. The penalty is triggered when the total flight distance of the drone exceeds its maximum flight range. This represents the corresponding penalty coefficient; A penalty is set for matching the task type incorrectly. When the payload carried by the drone does not match the assigned task type, the corresponding penalty is triggered. Match error penalty coefficients to task types; As an obstacle avoidance penalty, a corresponding penalty is triggered when the planned flight path crosses an obstacle or a no-fly zone. This is the obstacle avoidance penalty coefficient.
[0056] Furthermore, the improved chemical reaction optimization algorithm in step S4 uses a molecular population as the set of task allocation solutions. The spatial position of each molecule corresponds to a UAV-target allocation scheme, and the energy value of the molecule corresponds to the cost value of a unified cost function F. The solution space search is achieved through three basic reactions between molecules: collision, decomposition, and synthesis. A PSO velocity update operator is introduced to dynamically switch the direction of global search and local search until the algorithm termination condition is reached.
[0057] Furthermore, the introduction of the PSO search operator specifically involves: in each iteration, introducing a guiding term for each molecule, which includes both the individual's historical best position and the population's global best position, to update the molecule's position. The update formula is expressed as:
[0058] ;
[0059] in, Let be the search speed of the k-th molecule in the t-th iteration; Let be the spatial position of the k-th molecule in the t-th iteration, corresponding to a UAV-target task allocation scheme; Inertial weights are used to balance global and local search capabilities; , Here, represents the learning factor, and represents the individual cognitive coefficient and the social cognitive coefficient, respectively. , It is a random number that takes a value in the interval [0,1]. This represents the historical best position since the k-th molecule iteration; This represents the globally optimal position for the entire population to date.
[0060] Furthermore, the termination algorithm is determined as follows: when the number of iterations reaches the maximum number of algebras or consecutive iterations... The algorithm terminates if the global optimal cost is not improved.
[0061] This invention employs Dubins curves for segmented trajectory modeling, fully considering the minimum turning radius and start / end heading constraints of the UAV. The generated trajectory perfectly conforms to the flight dynamics characteristics of the UAV, avoiding the path infeasibility problem caused by traditional Euclidean distance modeling. Furthermore, by constructing a unified cost function that includes flight path cost and multi-dimensional constraint penalty terms, this invention can achieve accurate allocation of various mixed tasks. It also employs an improved CRO algorithm that integrates the PSO search operator, achieving a balance between global and local exploration through guidance from individual historical optima and population global optima. Attached Figure Description
[0062] Figure 1 A flowchart of a multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm is provided for an embodiment of the present invention;
[0063] Figure 2 A schematic diagram of multi-target clustering based on distance and azimuth angle provided in an embodiment of the present invention;
[0064] Figure 3 A schematic diagram of the turning radius of a drone provided in an embodiment of the present invention;
[0065] Figure 4 This is a schematic diagram of Dubins curve path modeling provided in an embodiment of the present invention;
[0066] Figure 5 This is a schematic diagram of a UAV performing two-segment circular arc turning and reversing flight paths, provided in an embodiment of the present invention.
[0067] Figure 6 A semi-physical simulation diagram of a drone and a target during the cooperative search phase, provided for an embodiment of the present invention;
[0068] Figure 7 This is a semi-physical simulation diagram of the drone and the target being tracked by their respective tasks, provided for an embodiment of the present invention. Detailed Implementation
[0069] This embodiment provides a multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm, such as... Figure 1 As shown, it includes the following steps:
[0070] Step S1: Classify and model various types of UAV missions, combining UAV flight performance constraints, payload matching constraints, and environmental constraints. In this embodiment, five typical mission types are defined: reconnaissance, relay, tracking, strike, and assessment. Different payloads, dwell times, coverage radii, etc., are set for UAVs performing each type of mission. Environmental constraints include known terrain, obstacle locations, and no-fly zones, etc.
[0071] Step S2: Perform spatial clustering on the multi-task objectives to generate clusters and Standoff tracking reference points, and set feasible domain constraints for task allocation.
[0072] like Figure 2 As shown, in this embodiment, the K-means algorithm is used to generate clusters, specifically: N targets are sorted according to their spatial location. The direction of motion is divided into several clusters, with the cluster center... This is determined by minimizing the sum of squared distances, as expressed in the formula:
[0073] ;
[0074] in, This represents the total number of task objectives. For the first The location coordinates of the target The first The x and y coordinates of each target in the reference coordinate system Let the coordinates of the cluster center of the k-th cluster be . The clustering loss value is used to determine the optimal cluster center location for each cluster by minimizing the clustering loss.
[0075] The Standoff tracking center is calculated using a weighted distance matrix and a density index, specifically including:
[0076] Step S21: Represent the relationship between targets using a weighted matrix, expressed by the formula:
[0077] ;
[0078] in, For the first The first goal and the first Distance weights between targets For the first The first goal and the first Euclidean distance function between targets and The first The first goal and the first The x-coordinate of each target; and The first The first goal and the first The vertical coordinate of each target.
[0079] Step S22: Calculate the density index for each target, expressed by the formula:
[0080] ;
[0081] in, For the first The density index of the nth target indicates that the nth target The sum of the distances between each target and all other targets. This represents the total number of task objectives.
[0082] S23: To determine the standoff tracking center of the UAV for a dynamic target, a density-weighted average of the target positions is used to calculate the standoff tracking center, expressed by the formula:
[0083]
[0084] in, For density-weighted target centers, For the first The density index of each target For the first The planar position coordinates of the target.
[0085] In the allocation phase, in order to ensure that no task is missed and that UAV resources are fully utilized, the feasible domain constraints for task allocation include full coverage constraints and full participation constraints.
[0086] The full coverage constraint specifically means that any task target must be within the effective coverage radius of at least one drone, ensuring that all tasks are executed by drones without any tasks being missed. This can be expressed as: The full participation constraint specifically states that at least one mission objective exists within the effective coverage radius of any UAV, ensuring that all UAVs in the cluster participate in mission execution with no resource idleness. This is expressed by the formula: Where, 𝑅 represents the effective coverage radius of the drone's mission. For the first The location coordinates of the target Let Z be the position coordinates of the z-th UAV. This represents the total number of task objectives.
[0087] Step S3: Based on the minimum turning radius constraint of the UAV, use Dubins curves to generate segmented feasible tracks and calculate the shortest path length from each UAV to each target.
[0088] like Figure 3 , Figure 4 , Figure 5 As shown, the segmented feasible track generated by the Dubins curve consists of two circular arc segments and one straight line segment. The path types include four types: LSL, LSR, RSL, and RSR, where L represents a left-turn circular arc segment, S represents a straight line segment, and R represents a right-turn circular arc segment.
[0089] The specific method for generating a single Dubins path is as follows:
[0090] Step S31: Model the Dubins curve path. Input parameters include the UAV's starting point coordinates, starting point heading angle, target point coordinates, ending point heading angle, and the UAV's minimum turning radius.
[0091] Step S32: Calculate the coordinates of the center of the arc segment path.
[0092] Given the starting coordinates, the coordinates of the center of the clockwise turning arc segment are... The coordinates of the center of the counterclockwise turning arc segment The calculation formulas are expressed as follows:
[0093] ;
[0094] in, The coordinates of the drone's starting plane; R is the starting heading angle of the drone; R is the minimum turning radius of the drone.
[0095] Similarly, for the endpoint coordinates, the coordinates of the center of the clockwise turning arc segment are... The coordinates of the center of the counterclockwise turning arc segment The calculation formulas are expressed as follows:
[0096] ;
[0097] in, The endpoint plane coordinates of the mission objective; R is the final heading angle for the UAV to reach the target point; R is the minimum turning radius of the UAV.
[0098] Step S33: Calculate the straight line segment path using the common tangent of the two arcs: Define the center of the starting arc as... The center of the arc at the endpoint is Let D be the distance between the center of the starting arc and the center of the ending arc, and let the angle between the line connecting the two centers and the horizontal axis of the reference coordinate system be θ. The formula is then expressed as:
[0099] ;
[0100] ;
[0101] in, The coordinates of the center of the starting side arc are given. The coordinates of the center of the arc on the endpoint side.
[0102] Step S34: Determine the path type. When the two arcs turn in the same direction, i.e., both turn left or both turn right, use the external common tangent. The angle between the line connecting the center of the circle to the point of tangency and the tangent of the line connecting the two centers is used. satisfy .
[0103] When the two arcs turn in opposite directions, i.e., one to the left and one to the right, the internal common tangent is used, and the angle between the line connecting the center of the arc to the point of tangency and the tangent of the line connecting the two centers is used. satisfy: ;
[0104] Where R is the minimum turning radius of the drone; D is the distance between the centers of the two arcs.
[0105] Step S35: Calculate the coordinates of the tangent point of the arc segment using the polar angle of tangency, specifically as follows:
[0106] Calculate the coordinates of the tangent point of the starting arc. The calculation formula is expressed as follows:
[0107] ;
[0108] Calculate the coordinates of the tangent point of the arc at the endpoint. The calculation formula is expressed as:
[0109] ;
[0110] in, The polar angle of tangency refers to the angle between the vector pointing from the center of the starting arc to the point of tangency and the horizontal axis of the reference coordinate system. This angle is defined when the path is the external common tangent. Conversely, if the two trajectories are in opposite directions, then the path is the internal common tangent. ; , This represents the minimum turning radius of the drone.
[0111] Step S36: Calculate the total length of a single Dubins path based on the starting point coordinates, the ending point coordinates, the tangent point coordinates of the starting point side arc and the ending point side arc. The total length of a single Dubins path is the sum of the length of the starting point arc segment, the length of the straight line segment, and the length of the ending point arc segment.
[0112] Step S4: Construct a unified cost function that includes path cost and multi-dimensional constraint penalty terms, and use an improved CRO algorithm that integrates the PSO search operator to iteratively solve the optimal UAV-target task allocation scheme under feasible region constraints.
[0113] The unified cost function is defined as a weighted sum of multi-dimensional constraint penalty terms, which include the total length of the Dubins path for the UAV to perform the assigned task, a range exceeding the limit penalty, a task type mismatch penalty, and an obstacle avoidance penalty. The expression is as follows:
[0114] ;
[0115] in, The total path length for assigning tasks to all drones. Path length weight; This is a penalty for exceeding the flight range limit. The penalty is triggered when the total flight distance of the drone exceeds its maximum flight range. This represents the corresponding penalty coefficient; A penalty is set for matching the task type incorrectly. When the payload carried by the drone does not match the assigned task type, the corresponding penalty is triggered. Match error penalty coefficients to task types; As an obstacle avoidance penalty, a corresponding penalty is triggered when the planned flight path crosses an obstacle or a no-fly zone. This is the obstacle avoidance penalty coefficient.
[0116] The improved chemical reaction optimization algorithm in step S4 uses a molecular population as the set of task allocation solutions. The spatial position of each molecule corresponds to a UAV-target allocation scheme, and the energy value of the molecule corresponds to the cost value of a unified cost function F. Solution space search is achieved through three basic types of reactions: collision, decomposition, and synthesis between molecules. In each iteration of the CRO algorithm, molecules undergo ineffective collisions, decomposition, synthesis, and mutual collisions. To improve search efficiency and escape local optima, a PSO velocity-position update mechanism is introduced during molecule position updates. In each iteration, a guiding term combining the individual's historical optimal position and the population's global optimal position is introduced for each molecule, updating its position. The update formula is expressed as:
[0117] ;
[0118] in, Let be the search speed of the k-th molecule in the t-th iteration; Let be the spatial position of the k-th molecule in the t-th iteration, corresponding to a UAV-target task allocation scheme; Inertial weights are used to balance global and local search capabilities; , Here, represents the learning factor, and represents the individual cognitive coefficient and the social cognitive coefficient, respectively. , It is a random number that takes a value in the interval [0,1]. This represents the historical best position since the k-th molecule iteration; This represents the globally optimal position for the entire population to date.
[0119] When the number of iterations reaches the maximum number of algebras or continuous The algorithm terminates if the global optimal cost is not improved.
[0120] Step S5: Generate a task allocation scheme that satisfies the full coverage constraint and the full participation constraint, and simultaneously output the Dubins track corresponding to each UAV.
[0121] The above methods are applied to simulate the scenario, such as... Figure 6 , Figure 7 As shown, Figure 6 The image shows three targets moving at different speeds and in different directions, with multiple circular no-fly zones set up in the environment. Figure 7 The final allocation results and path planning diagram are shown. It can be seen that the three drones were reasonably allocated to form a tracking and reconnaissance configuration for the three targets, and the paths all avoided the no-fly zone.
Claims
1. A multi-task allocation method for multiple unmanned aerial vehicles (UAVs) based on an improved chemical reaction optimization algorithm, characterized in that, Includes the following steps: Step S1: Classify and model various types of UAV missions, and preset UAV flight performance constraints, payload matching constraints, and environmental constraints; Step S2: Perform spatial clustering on multi-task objectives to generate clusters and Standoff tracking reference points, and set feasible domain constraints for task allocation. Step S3: Based on the minimum turning radius constraint of the UAV, use Dubins curves to generate segmented feasible tracks and calculate the shortest path length from each UAV to each target; Step S4: Construct a unified cost function that includes path cost and multi-dimensional constraint penalty terms, and use an improved CRO algorithm that integrates PSO search operators to iteratively solve the optimal UAV-target task allocation scheme under feasible region constraints; Step S5: Generate a task allocation scheme that satisfies the full coverage constraint and the full participation constraint, and simultaneously output the Dubins track corresponding to each UAV.
2. The multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 1, characterized in that, In step S2, the clusters are generated using the K-means algorithm, specifically by: sorting the N targets according to their spatial positions. The direction of motion is divided into several clusters, with the cluster center... This is determined by minimizing the sum of squared distances, as expressed in the formula: ; in, This represents the total number of task objectives. For the first The location coordinates of the target The first The x and y coordinates of each target in the reference coordinate system Let the coordinates of the cluster center of the k-th cluster be . This represents the clustering loss value.
3. The multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 1, characterized in that, In step S2, the Standoff tracking center is calculated using a weighted distance matrix and a density index, specifically including: Step S21: Represent the relationship between targets using a weighted matrix, expressed by the formula: ; in, For the first The first goal and the first Distance weights between targets For the first The first goal and the first Euclidean distance function between targets and The first The first goal and the first The x-coordinate of each target; and The first The first goal and the first The vertical coordinate of each target; Step S22: By calculating the first The sum of the distances of a target to all other targets represents the density index of each target, expressed by the formula: ; in, For the first The density index of each target The total number of task objectives; S23: The Standoff tracking center is calculated using the target location weighted by average density, expressed by the following formula: ; in, For density-weighted target centers, For the first The density index of each target For the first The planar position coordinates of the target.
4. The multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 1, characterized in that, In step S2, the feasible domain constraints for setting task allocation include full coverage constraints and full participation constraints. The full coverage constraint specifically means that any task target must be within the effective coverage radius of at least one drone, ensuring that all tasks are executed by drones without any tasks being missed. This can be expressed as: ; The full participation constraint specifically states that at least one mission objective exists within the effective coverage radius of any UAV, ensuring that all UAVs in the cluster participate in mission execution with no resource idleness. This can be expressed as: ; R represents the effective coverage radius of the drone's mission. For the first The location coordinates of the target Let Z be the position coordinates of the z-th UAV. This represents the total number of task objectives.
5. The multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 1, characterized in that, In step S3, the segmented feasible track generated by using Dubins curves is composed of two circular arc segments and one straight line segment. The path types include four types: LSL, LSR, RSL, and RSR, where L represents a left-turn circular arc segment, S represents a straight line segment, and R represents a right-turn circular arc segment.
6. The multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 5, characterized in that, The specific method for generating a single Dubins path is as follows: Step S31: Model the Dubins curve path. Input parameters include the UAV's starting point coordinates, starting point heading angle, target point coordinates, ending point heading angle, and the UAV's minimum turning radius. Step S32: Calculate the coordinates of the center of the arc segment's trajectory; For the starting position, the coordinates of the center of the clockwise turning arc segment The coordinates of the center of the counterclockwise turning arc segment The calculation formulas are expressed as follows: ; in, The starting plane coordinates of the drone; R is the starting heading angle of the drone; R is the minimum turning radius of the drone. Similarly, for the endpoint, the coordinates of the center of the clockwise turning arc segment... The coordinates of the center of the counterclockwise turning arc segment The formulas are expressed as follows: ; in, The endpoint plane coordinates of the mission objective; R is the final heading angle for the drone to reach the target point; R is the minimum turning radius of the drone. Step S33: Calculate the straight line segment path using the common tangent of the two arcs. The calculation method for the common tangent parameter is as follows: define the center of the starting arc as... The center of the arc at the endpoint is Let D be the distance between the center of the starting arc and the center of the ending arc, and let the angle between the line connecting the two centers and the horizontal axis of the reference coordinate system be θ. ,but: ; ; in, The coordinates of the center of the starting side arc are given. The coordinates of the center of the arc on the endpoint side; Step S34: Determine the path type. When two arcs turn in the same direction, use the external common tangent. The angle between the line connecting the center of the circle to the point of tangency and the tangent of the line connecting the two centers is... satisfy ; When two circular arcs turn in opposite directions, the internal common tangent is used, and the angle between the line connecting the center of the arc to the point of tangency and the tangent to the line connecting the two centers is used. satisfy: ; Where R is the minimum turning radius of the drone; D is the distance between the centers of the two arcs; Step S35: Calculate the coordinates of the tangent point of the arc segment using the polar angle of tangency, specifically as follows: Coordinates of the tangent point of the starting side arc The calculation formula is expressed as follows: ; Coordinates of the tangent point of the arc at the endpoint The calculation formula is expressed as: ; in, When the polar angle of the tangent point is the path of the external common tangent, Conversely, if the two trajectories are in opposite directions, then the path is the internal common tangent. ; , This represents the minimum turning radius of the drone. Step S36: Calculate the total length of a single Dubins path based on the starting point coordinates, the ending point coordinates, the tangent point coordinates of the starting point side arc and the ending point side arc. The total length of a single Dubins path is the sum of the length of the starting point arc segment, the length of the straight line segment, and the length of the ending point arc segment.
7. A multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 6, characterized in that, The unified cost function is defined as a weighted sum of multi-dimensional constraint penalty terms, which include the Dubins path length for the UAV to perform the assigned task, a range exceeding the limit penalty, a task type mismatch penalty, and an obstacle avoidance penalty. The expression is as follows: ; in, The total path length for assigning tasks to all drones. Path length weight; This is a penalty for exceeding the flight range limit. The penalty is triggered when the total flight distance of the drone exceeds its maximum flight range. This represents the corresponding penalty coefficient; A penalty is set for matching the task type incorrectly. When the payload carried by the drone does not match the assigned task type, the corresponding penalty is triggered. Match error penalty coefficients to task types; As an obstacle avoidance penalty, a corresponding penalty is triggered when the planned flight path crosses an obstacle or a no-fly zone. This is the obstacle avoidance penalty coefficient.
8. A multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 7, characterized in that, The improved chemical reaction optimization algorithm in step S4 uses a molecular population as the set of task allocation solutions. The spatial position of each molecule corresponds to a UAV-target allocation scheme, and the energy value of the molecule corresponds to the cost value of a unified cost function F. The solution space search is achieved through three basic reactions between molecules: collision, decomposition, and synthesis. A PSO velocity update operator is introduced to dynamically switch the direction of global search and local search until the algorithm termination condition is met.
9. A multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 8, characterized in that, The introduction of the PSO search operator specifically involves: in each iteration, introducing a guiding term for each molecule, which includes both the individual's historical best position and the population's global best position, to update the molecule's position. The update formula is expressed as: ; in, Let be the search speed of the k-th molecule in the t-th iteration; Let be the spatial position of the k-th molecule in the t-th iteration, corresponding to a UAV-target task allocation scheme; Inertial weights are used to balance global and local search capabilities; , Here, represents the learning factor, and represents the individual cognitive coefficient and the social cognitive coefficient, respectively. , It is a random number that takes a value in the interval [0,1]. This represents the historical best position since the k-th molecule iteration; This represents the globally optimal position for the entire population to date.
10. A multi-UAV multi-task allocation method based on an improved chemical reaction optimization algorithm according to claim 8, characterized in that, The termination algorithm is determined as follows: when the number of iterations reaches the maximum number of algebras or is consecutive... The algorithm terminates if the global optimal cost is not improved.
