Multi-unmanned vehicle field cooperative task allocation method based on auction-ant colony algorithm

Abstract

The invention relates to the field of unmanned system cooperative control and intelligent task planning, in particular to a multi-unmanned vehicle field cooperative task allocation method based on an auction-ant colony algorithm, comprising the following steps: S1, constructing a capability matrix, a task demand matrix and an environment constraint matrix of multiple unmanned vehicles, and coupling to generate a comprehensive feature matrix; s2, on the basis of the comprehensive feature matrix, task pre-allocation and global optimization are realized through a three-dimensional linkage model of energy order precision-monopole precision research-collaborative aggregation optimization; s3, adopting an auction algorithm to complete pre-distribution of an energy level fine feed stage and a single-pole fine research stage, and screening a high-adaptation unmanned vehicle-task pair; according to the multi-unmanned vehicle field cooperative task allocation method based on the auction-ant colony algorithm, tasks with high adaptation degree are quickly screened and allocated through two-stage auction of energy level fine feed and monopole fine research, the initial solution space is greatly reduced, and the early-stage allocation efficiency is improved.

Description

Technical Field

[0001] This invention belongs to the field of unmanned system cooperative control and intelligent task planning, specifically involving a method for multi-unmanned vehicle field cooperative task allocation based on auction-ant colony algorithm. Background Technology

[0002] Multi-vehicle collaborative operations are a key means to improve the efficiency of field missions. Compared to single-vehicle operations, multi-vehicle swarms can cover a wider operational range and handle more complex mission requirements through task splitting and collaboration, while significantly enhancing system robustness and reducing the impact of single-point failures on the overall mission. Currently, this technology has shown great application potential in fields such as field resource exploration, emergency rescue, and border patrol. However, efficient and accurate task allocation is the core prerequisite for the effectiveness of multi-vehicle collaborative operations. This requires comprehensive consideration of the differences in the capabilities of unmanned vehicles, the dynamic constraints of the field environment, and the diverse needs of the mission, ensuring that "the right unmanned vehicle performs the right task at the right time," while simultaneously optimizing the overall mission completion rate, total system energy consumption, and load balancing among units.

[0003] In the field of multi-vehicle task allocation technology, existing research and practical solutions can be mainly divided into three categories, but they all have obvious limitations when dealing with complex field scenarios:

[0004] 1. Rule-based allocation methods (such as greedy algorithms and priority allocation): These methods make rapid decisions based on pre-defined fixed rules (such as allocation based on proximity or prioritizing the most capable autonomous vehicle), resulting in fast response times. However, they are essentially local optimizations and are prone to getting trapped in local optima. For example, continuously assigning tasks to the most capable autonomous vehicle will overload it while other units remain idle, disrupting the system's load balance. More importantly, their rigid rules cannot adapt to dynamic changes in parameters such as communication quality and terrain mobility in the field, leading to a disconnect between the allocation scheme and actual execution conditions, resulting in low overall task completion rates and resource utilization efficiency.

[0005] 2. Allocation methods based on traditional optimization algorithms (such as linear programming and integer programming): These methods model the allocation problem as a precise mathematical model to seek the theoretically optimal solution. However, they heavily rely on precise problem modeling, while the field environment is highly uncertain and dynamic. The autonomous vehicle's state, task requirements, and environmental parameters may all change in real time, making it difficult to establish a precise mathematical model and unable to quickly reconstruct it. Furthermore, as the scale of the autonomous vehicle and the task increases, the computational complexity of the algorithm grows exponentially, failing to meet the stringent real-time requirements of field tasks.

[0006] 3. Allocation methods based on a single intelligent algorithm (such as using ant colony optimization or genetic algorithms alone): Intelligent algorithms possess the ability to perform global searches in nonlinear, multi-constraint spaces. However, in large-scale field scenarios, they have inherent limitations: First, the initial solution space is too large, leading to slow convergence speed, and a feasible solution may not be found within the critical task window; second, a single algorithm struggles to balance "local adaptability" and "global synergy." For example, if the ant colony optimization algorithm only aims at the shortest path, it may ignore the capabilities and current load of different autonomous vehicles, resulting in excessively high overall system energy consumption or premature energy depletion in some units, affecting the overall task progress.

[0007] In addition to the common limitations at the algorithm level mentioned above, the unique characteristics of unstructured environments in the wild pose even more severe challenges to task allocation techniques:

[0008] Dynamic nature of environmental constraints: The complexity of field terrain, communication quality, weather conditions and other factors change drastically with time and space, requiring the distribution system to have online real-time perception and dynamic response capabilities. However, existing methods mostly treat the environment as a static parameter, which is not adaptable enough.

[0009] Ambiguity in capability-demand matching: Existing methods often use a single indicator for matching, ignoring the multidimensional and heterogeneous attributes of autonomous vehicles (such as range, terrain passability, and reconnaissance accuracy) and tasks (such as requiring high-precision sensors and high mobility), resulting in "capability mismatch" and failing to fully leverage the professional advantages of autonomous vehicles;

[0010] The complexity of multi-task and multi-autonomous vehicle collaboration: Many solutions are limited to pairwise matching of autonomous vehicles and tasks, failing to consider the spatiotemporal correlation between tasks (such as execution along the same route) and complex tasks that require multi-vehicle collaboration, resulting in unreasonable task sequences, low overall execution efficiency and waste of resources;

[0011] To address the aforementioned issues, this invention proposes a multi-unmanned vehicle (UAV) collaborative task allocation method in the field based on an auction-ant colony algorithm. This method achieves precise quantification of the UAV's suitability for the task by constructing a comprehensive feature matrix that couples "capability-requirement-environment" in three dimensions. A three-dimensional linkage model of "capability-refinement-single-level optimization" is constructed. The auction algorithm completes "capability-refinement" and "single-level optimization" for pre-allocation, and an improved ant colony algorithm achieves "collaborative optimization," balancing task completion rate, total energy consumption, and load balance. Simultaneously, a real-time feedback mechanism is introduced to dynamically correct the impact of environmental changes and state deviations on the allocation scheme. Ultimately, this achieves efficient and robust collaborative task allocation for multiple UAVs in complex field scenarios, overcoming the insufficient adaptability of existing technologies in field environments. Summary of the Invention

[0012] The purpose of this invention is to overcome the shortcomings of existing task allocation methods in the background art, such as poor environmental adaptability, insufficient global coordination, and difficulty in balancing real-time performance and optimization, resulting in low task completion rate and resource waste. The invention aims to realize a multi-unmanned vehicle field collaborative task allocation method based on auction-ant colony algorithm.

[0013] To achieve the above-mentioned objectives, the technical solution of this invention is: a method for multi-unmanned vehicle field collaborative task allocation based on auction-ant colony algorithm, comprising the following steps:

[0014] S1. Construct the capability matrix, task requirement matrix, and environmental constraint matrix of multiple unmanned vehicles, and couple them to generate a comprehensive feature matrix;

[0015] S2. Based on the comprehensive feature matrix, task pre-allocation and global optimization are achieved through a three-dimensional linkage model of "energy level refinement - single-level refinement - collaborative optimization".

[0016] S3. Use an auction algorithm to pre-allocate the energy level refinement and single-level refinement stages, and select highly adaptable unmanned vehicle-task pairs;

[0017] S4. For unassigned tasks and idle unmanned vehicles, an improved ant colony algorithm is used for collaborative optimization to generate the globally optimal allocation scheme.

[0018] S5. Introduce a real-time feedback mechanism during execution to dynamically adjust the allocation scheme to adapt to environmental changes.

[0019] Specifically, the formula for calculating the environmental impact factor is as follows:

[0020]

[0021] in, For the first The first driverless car to perform Environmental impact factor for the task; For communication quality; The complexity of the terrain.

[0022] Specifically, the formula for calculating the coupling value of capability matching degree and environmental constraints for the comprehensive feature matrix elements is as follows:

[0023]

[0024] Where Ci,l is the capability value of the i-th autonomous vehicle in the l-th dimension; Dj,l is the requirement value of the j-th task in the l-th dimension; and Ei,j is the environmental constraint coefficient. This is a capability dimension.

[0025] Specifically, the energy level refinement stage includes the following steps:

[0026] S10. Calculate the average fitness of the task according to the type of unmanned vehicle;

[0027]

[0028] in, For the first Autonomous vehicles for the first Average valuation of the task It is a driverless car type. This refers to the number of this type of driverless vehicle;

[0029] S11. Select the type of driverless vehicle with the highest bid in the first round of auction as the candidate type;

[0030] The formula for submitting a quote is:

[0031]

[0032] in, To address the submission of quotes for the task, This is the quotation coefficient;

[0033] The formula for the highest bid is:

[0034]

[0035] in, Choose the type with the highest price.

[0036] Specifically, the single-pole refining stage includes the following steps:

[0037] S20. Calculate the individual fitness of each autonomous vehicle in the candidate types, and fuse the capability matching degree and location cost;

[0038]

[0039] in, For the first Autonomous vehicles for the first Average valuation of the task; The maximum task distance within the scene; This is the position attenuation factor; The distance from the autonomous vehicle to the mission point;

[0040] S21. High-adaptability tasks are directly allocated through the second round of auction, while low-adaptability tasks are included in the collaborative optimization pool.

[0041] ;

[0042] in, Submit a bid for the autonomous vehicle; β is the individual bid coefficient, if the highest bid... , The auction threshold is set to 0.6 to avoid low-fitness assignments; in this case, the task is directly assigned to the corresponding autonomous vehicle. Then the task will be included in the collaborative optimization pool; For the first Autonomous vehicles for the first The average estimate of the task.

[0043] Specifically, the collaborative optimization includes the following steps:

[0044] S30. Map the tasks and idle unmanned vehicles in the collaborative optimization pool to the ant colony algorithm solution space.

[0045] S31. Define the heuristic function and integrate the pre-allocated guiding factor;

[0046]

[0047] Where ηi,j represents the heuristic degree of selecting the “autonomous vehicle-task” matching pair; For pre-allocated guiding coefficients, To pre-allocate association factors; For the first Autonomous vehicles for the first Average valuation of the task;

[0048] S32. Through iterative optimization of pheromone updates and path selection, the fitness targets are task completion rate, total energy consumption, and load balancing.

[0049] The formula for path selection:

[0050]

[0051] Where ηi,j represents the heuristic degree of selecting the “autonomous vehicle-task” matching pair; For pheromone importance factors; This is the importance factor of the heuristic function; for The autonomous vehicle's mission is to match pheromone concentrations with potential partners. For the first The probability of an ant choosing the autonomous vehicle task at the number of iterations; This represents the number of ants.

[0052] Specifically, the information update includes partial updates and global updates;

[0053] Formula for partial update:

[0054]

[0055] in, for Pheromones concentration in autonomous vehicles and task matching pairs at all times; for Pheromones concentration in autonomous vehicles and task matching pairs at all times; The pheromone evaporation coefficient; This represents the initial value of the pheromone concentration.

[0056] The formula for global update:

[0057]

[0058] in, for Pheromones concentration in autonomous vehicles and task matching pairs at all times; for Pheromones concentration in autonomous vehicles and task matching pairs at all times; The pheromone evaporation coefficient; The pheromone increment for the optimal path;

[0059]

[0060] in, This represents the fitness value of the optimal path.

[0061] The fitness function is:

[0062]

[0063] Where R is the task completion rate; Etotal is the total energy consumption; and σ is the load balancing coefficient. The weighting coefficients are satisfied. .

[0064] A multi-unmanned vehicle field collaborative task allocation system includes:

[0065] The central control platform is used to execute a multi-unmanned vehicle field collaborative task allocation method based on the auction-ant colony algorithm;

[0066] Multiple driverless vehicles equipped with condition monitoring modules;

[0067] The environmental perception module is used to collect information on terrain complexity and communication quality.

[0068] The communication module is used for issuing commands and providing status feedback.

[0069] Compared with existing technologies, the multi-unmanned vehicle field cooperative task allocation method based on auction-ant colony algorithm of the present invention has at least the following beneficial effects:

[0070] 1. This invention combines the auction algorithm with the improved ant colony algorithm. Through a two-stage auction of energy level refinement and single-level refinement, it quickly filters and allocates tasks with high fitness, which greatly reduces the initial solution space and improves the efficiency of early allocation. For the remaining complex tasks, the ant colony algorithm is used to perform a refined global search, which effectively avoids the problem of greedy algorithms getting trapped in local optima, thus achieving a balance between efficiency and optimality overall.

[0071] 2. This invention constructs an environmental constraint matrix to quantify key field environmental factors such as communication quality and terrain complexity and integrate them into the allocation model. At the same time, it introduces a real-time feedback and dynamic correction mechanism, enabling the system to perceive environmental changes and re-plan task allocation in real time. This overcomes the shortcomings of traditional optimization algorithms that rely on static models and are difficult to cope with dynamic changes, ensuring the continuity and success rate of task execution.

[0072] 3. This invention employs a three-dimensional linkage allocation model—energy level refinement, single-level refinement, and collaborative optimization—to construct a progressive allocation process from vehicle selection to individual optimization and then to global collaborative optimization. This model has a clear hierarchy, considering both the typological capabilities of autonomous vehicles and refining the matching of individual vehicles with tasks. Finally, collaborative optimization resolves complex coupling relationships across vehicles and tasks, achieving accurate and efficient matching of massive vehicle-task combinations in large-scale scenarios.

[0073] 4. This invention provides a high-quality initial solution guide for the ant colony algorithm by introducing an auction algorithm, effectively avoiding the slow convergence speed caused by blind searching in the initial stage. The improved ant colony algorithm is further optimized to compensate for the auction algorithm's shortcomings in global coordination. The deep integration of the two algorithms leverages their respective advantages, resulting in an overall method that outperforms either the single algorithm in terms of convergence speed and the quality of the final solution. Attached Figure Description

[0074] Figure 1 This is a flowchart illustrating the present invention. Detailed Implementation

[0075] The following description, in conjunction with the accompanying drawings and specific embodiments, provides a more detailed account of the multi-unmanned vehicle field collaborative task allocation method based on the auction-ant colony algorithm of the present invention.

[0076] In the description of this invention, it should be understood that the terms "upper", "lower", "front", "rear", "left", "right", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on this invention.

[0077] This embodiment discloses a method for multi-unmanned vehicle field collaborative task allocation based on auction-ant colony algorithm, such as... Figure 1 As shown, a method for multi-unmanned vehicle field cooperative task allocation based on auction-ant colony algorithm includes the following steps:

[0078] S1. Construct the capability matrix, task requirement matrix, and environmental constraint matrix of multiple unmanned vehicles, and couple them to generate a comprehensive feature matrix;

[0079] S2. Based on the comprehensive feature matrix, task pre-allocation and global optimization are achieved through a three-dimensional linkage model of "energy level refinement - single-level refinement - collaborative optimization".

[0080] S3. Use an auction algorithm to pre-allocate the energy level refinement and single-level refinement stages, and select highly adaptable unmanned vehicle-task pairs;

[0081] S4. For unassigned tasks and idle unmanned vehicles, an improved ant colony algorithm is used for collaborative optimization to generate the globally optimal allocation scheme.

[0082] S5. Introduce a real-time feedback mechanism during execution to dynamically adjust the allocation scheme to adapt to environmental changes.

[0083] Autonomous vehicle capability parameter collection: Collect data for each autonomous vehicle (denoted as...). The core capability parameters of ) are used to construct a capability matrix. ,in In terms of capabilities, including battery life. Terrain passability Task execution efficiency , Indicates the first The number of driverless cars in the The capability value of the dimension, where the range is measured in km and the terrain passability is a probability value in the range of [0,1].

[0084] Task requirement parameter definition: The field task to be assigned (denoted as...) Quantify the demand and construct a demand matrix. , Indicates the first The task in the Minimum requirement for a dimension (e.g., task) Requires a range of ≥50km, corresponding );

[0085] Environmental constraint parameter embedding: Introducing field environmental constraint factors, including communication quality. Terrain complexity Construct the environmental constraint matrix ;

[0086] The formula for calculating the environmental impact factor is:

[0087]

[0088] in, For the first The first driverless car to perform Environmental impact factor for the task; For communication quality, The better the communication quality, the closer the value is to 1; Due to the complexity of the terrain, The more complex the terrain, the larger the value, ultimately forming a comprehensive feature matrix of "unmanned vehicle-mission";

[0089] The comprehensive feature matrix elements are the coupling values ​​of capability matching degree and environmental constraints. The formula for calculating these values ​​is as follows:

[0090]

[0091] Where Ci,l is the capability value of the i-th autonomous vehicle in the l-th dimension; Dj,l is the requirement value of the j-th task in the l-th dimension; and Ei,j is the environmental constraint coefficient. This is a capability dimension.

[0092] The energy level refinement stage includes the following steps:

[0093] S10. Calculate the average fitness of the task according to the type of unmanned vehicle;

[0094] Capability adaptability calculation: Unmanned vehicles are grouped by type, including reconnaissance, combat, and transport vehicles. The average capability adaptability of each unmanned vehicle type to each mission is calculated and used as the valuation for the first round of auctions.

[0095]

[0096] in, For the first Autonomous vehicles for the first Average valuation of the task It is a driverless car type. This refers to the number of this type of driverless vehicle;

[0097] S11. Select the type of driverless vehicle with the highest bid in the first round of auction as the candidate type;

[0098] The formula for submitting a quote is:

[0099]

[0100] in, To address the submission of quotes for the task, This is the pricing coefficient; for each task submission, the type with the highest price is selected as the candidate type.

[0101] The formula for the highest bid is:

[0102]

[0103] in, Choose the type with the highest price.

[0104] The single-pole refining stage includes the following steps:

[0105] S20. Calculate the individual fitness of each autonomous vehicle in the candidate types, and fuse the capability matching degree and location cost;

[0106]

[0107] in, For the first Autonomous vehicles for the first Average valuation of the task; The maximum task distance within the scene; This is the location attenuation factor; the closer the distance, the closer the factor is to 1. The distance from the autonomous vehicle to the mission point;

[0108] S21. High-adaptability tasks are directly allocated through the second round of auction, while low-adaptability tasks are included in the collaborative optimization pool.

[0109]

[0110] in, Submit a bid for the autonomous vehicle; β is the individual bid coefficient, set to 1.1, if the highest bid... , The auction threshold is set to 0.6 to avoid low-fitness assignments; in this case, the task is directly assigned to the corresponding autonomous vehicle. Then the task will be included in the collaborative optimization pool; For the first Autonomous vehicles for the first The average estimate of the task.

[0111] The collaborative optimization includes the following steps:

[0112] S30. Map the tasks and idle unmanned vehicles in the collaborative optimization pool to the ant colony algorithm solution space.

[0113] S31. Define a heuristic function that incorporates pre-assigned guiding factors. The heuristic function represents the degree of inspiration for selecting autonomous vehicle-task matching pairs, based on individual fitness correction during the pre-assignment phase.

[0114]

[0115] Where ηi,j represents the heuristic degree of selecting the "autonomous vehicle-task" matching pair. The pre-allocated guiding coefficient is set to 0.3. For pre-assigned association factors, if they are marked as candidates in the pre-assignment... Otherwise, it is 0; For the first Autonomous vehicles for the first The average cost of each task is calculated; for tasks that have not been assigned, tasks and idle unmanned vehicles in the collaborative optimization pool are marked as collaborative optimization objects, and the pre-assignment results are output, including the assigned "task-unmanned vehicle" pairs and the objects to be optimized, which serve as the initial input for the ant colony algorithm; the idle unmanned vehicles and tasks to be assigned in the collaborative optimization pool are mapped to the solution space of the ant colony algorithm, and the path of each ant corresponds to a set of "unmanned vehicle-task" allocation schemes, and the path nodes are the matching pairs of "unmanned vehicle-task".

[0116] S32. Through iterative optimization of pheromone updates and path selection, the fitness targets are task completion rate, total energy consumption, and load balancing.

[0117] The formula for path selection:

[0118]

[0119] Where ηi,j represents the heuristic degree of selecting the “autonomous vehicle-task” matching pair; The pheromone importance factor is set to 1.8. Let be the importance factor of the heuristic function, set to 2.2; for The pheromone concentration of the "driverless vehicle - mission" matching pair at any given moment;

[0120] For the first The probability of an ant choosing the autonomous vehicle task at the number of iterations; Let the number of ants be set as follows: To ensure solution space coverage.

[0121] After the iteration is completed, the information update includes local update and global update;

[0122] Formula for partial update:

[0123]

[0124] in, for Pheromones concentration in autonomous vehicles and task matching pairs at all times; for Pheromones concentration in autonomous vehicles and task matching pairs at all times; The pheromone evaporation coefficient; This represents the initial value of the pheromone concentration.

[0125] The formula for global update:

[0126]

[0127] in, for Pheromones concentration in autonomous vehicles and task matching pairs at all times; for Pheromones concentration in autonomous vehicles and task matching pairs at all times; The pheromone evaporation coefficient; The pheromone increment for the optimal path;

[0128]

[0129] in, This represents the fitness value of the optimal path.

[0130] The fitness function is:

[0131]

[0132] Where R is the task completion rate; Etotal is the total energy consumption; and σ is the load balancing coefficient. The weighting coefficients are satisfied. ; , , ;

[0133] Total energy consumption; This refers to the energy consumption per unit distance for autonomous vehicles.

[0134] This is the load balancing coefficient. ,in, The number of tasks for the autonomous vehicle. This represents the average number of tasks.

[0135] The algorithm terminates and outputs the final task allocation scheme when the number of iterations reaches the maximum number of iterations (set to 100) or the fluctuation of the optimal fitness value in 10 consecutive iterations is less than (set to 0.001).

[0136] A multi-unmanned vehicle field collaborative task allocation system includes:

[0137] Central control platform; The field electronic map and algorithm parameters are pre-stored in the storage unit of the central control platform. Before the data acquisition and preprocessing tasks begin, the central control platform completes the basic data acquisition and preprocessing.

[0138] Multiple unmanned vehicles are equipped with a status monitoring module. The status monitoring module collects parameters such as the vehicle's range, terrain passability, and execution efficiency, and generates a capability matrix after normalization.

[0139] The status monitoring module includes a battery life sensor and an execution progress recorder;

[0140] The environmental perception module, including lidar and communication quality detectors, is used to collect information on terrain complexity and communication quality.

[0141] The communication module is used for issuing commands and providing status feedback.

[0142] During the autonomous vehicle's mission, status data (remaining range, mission progress, and actual environmental parameters) is uploaded every 5 minutes. If deviations occur, including actual energy consumption exceeding the budget by 10% or a sudden drop in communication quality, a local reallocation is triggered: the capability matrix and environment matrix are updated based on the current status, and the second round of auction in step two is repeated to achieve dynamic adjustment. After the mission is completed, the data from this allocation is calculated and archived to the historical database for subsequent algorithm parameter optimization (including adjustments to the ant colony algorithm). , This approach ensures that the allocation of tasks for multiple unmanned vehicles in the field always aligns with the dynamic matching requirements of "vehicle-task-environment": In complex terrain and communication-limited field scenarios, the combination of highly adaptable "vehicle-task" pairs selected through auction pre-allocation and the global optimization of the ant colony algorithm can avoid single-vehicle overload and resource waste; when tasks are dynamically adjusted or the environment changes abruptly, real-time feedback triggers local reallocation to quickly correct the solution. It leverages the rapid selection advantage of the auction algorithm while relying on the improved ant colony algorithm to achieve global optimization, eliminating the adverse effects of uncertainties in the field environment and differences in unmanned vehicle capabilities on task execution, thus ensuring the efficiency, stability, and robustness of multi-unmanned vehicle collaboration.

[0143] When in use, this invention ensures that all units of the system are connected normally by activating the central control platform, all unmanned vehicles, environmental perception module, and communication module. The system inputs the tasks to be executed and their requirements into the central control platform. Each unmanned vehicle (UAV) reports its initial status through the status monitoring module, including current location, battery level, load capacity, and sensor type, forming a capability matrix. The environmental perception module then begins operation, collecting and updating global terrain complexity and communication quality data in real time through satellite maps, terrain sensors, and communication signal strength detection, constructing an environmental constraint matrix. Task allocation is automatically executed. The central control platform automatically constructs a task requirement matrix based on the input task list and UAV status, coupling it with the capability matrix and environmental constraint matrix to generate a comprehensive feature matrix. The system then immediately initiates an auction-ant colony hybrid algorithm: through energy level refinement, the system first groups UAVs by type (e.g., reconnaissance, transportation), calculates the average fitness of each category for each task, and selects the UAV type with the highest bid for each task as a candidate through the first round of auction algorithm. Through single-level refinement, the individual fitness of each specific UAV in the candidate types is calculated, comprehensively considering its capability matching degree and distance to the task point. Then, a second round of auction is conducted, selecting those with extremely high matching degrees (highest bid P2_max ≥ 0.5). Tasks with a threshold θ are directly assigned to the corresponding optimal autonomous vehicle. Through collaborative optimization, remaining tasks that are difficult to assign or have low matching degrees are placed in a collaborative optimization pool. The system calls an improved ant colony algorithm to map these tasks to idle autonomous vehicles into a solution space. The algorithm continuously updates pheromones through iterative search, using task completion rate, total energy consumption, and load balancing as comprehensive objectives, and finally generates a globally optimized allocation scheme. For task execution and real-time dynamic adjustment, the central control platform distributes the final allocation scheme to each autonomous vehicle through the communication module, and the autonomous vehicles begin to execute tasks. Throughout the process: the status monitoring module continuously reports the real-time status of the autonomous vehicles, including power consumption, task progress, and unexpected faults; the environmental perception module continuously monitors environmental changes, including sudden fog in a certain area leading to increased terrain complexity and communication interruptions. Once the system determines through real-time feedback data that the current allocation scheme is no longer optimal or cannot be executed, including vehicle malfunctions, the insertion of new tasks, and environmental deterioration, a dynamic correction mechanism will be automatically triggered. The system will immediately re-execute the rapid allocation process in step S2 based on the latest matrix data, generate a new plan and issue it to ensure that the entire system can adapt to the complex and ever-changing environment in the field. For task completion and performance evaluation, after all tasks are completed, the system will automatically summarize and output a performance report of this collaborative task, including key indicators such as total task completion rate, total system energy consumption, and load balancing coefficient, for subsequent analysis and optimization.

[0144] It should be noted that, in actual implementation, the structure depicted in the accompanying drawings is not a fixed or unchanging embodiment. The components of the embodiments of the invention described and shown in these drawings can typically be arranged and designed in various different configurations. Furthermore, the accompanying drawings and abstract drawings are merely illustrative and do not represent the specific structure or actual quantity in a concrete implementation.

[0145] Unless otherwise defined, the technical or scientific terms used herein should be understood in their ordinary sense as understood by one of ordinary skill in the art to which this invention pertains. The use of terms such as "a" or "an" in this specification and claims is not necessarily a limitation of quantity. Terms such as "comprising" or "including" mean that the element or component preceding the word encompasses the element or component listed following the word and its equivalents, without excluding other elements or components. Terms such as "connected" or "linked" are not limited to physical or mechanical connections, but can include electrical connections, whether direct or indirect.

[0146] The exemplary embodiments of the present invention have been described in detail above with reference to preferred embodiments. However, those skilled in the art will understand that various modifications and alterations can be made to the above specific embodiments without departing from the concept of the present invention, and various combinations can be made to the various technical features and structures proposed in the present invention without exceeding the protection scope of the present invention.

Claims

1. A method for allocating collaborative tasks among multiple unmanned vehicles in the field based on an auction-ant colony algorithm, characterized in that: Includes the following steps: S1. Construct the capability matrix, task requirement matrix, and environmental constraint matrix of multiple unmanned vehicles, and couple them to generate a comprehensive feature matrix; S2. Based on the comprehensive feature matrix, task pre-allocation and global optimization are achieved through a three-dimensional linkage model of "energy level improvement - single-level refinement - collaborative optimization". S3. Use an auction algorithm to pre-allocate the energy level refinement and single-level refinement stages, and select highly adaptable unmanned vehicle-task pairs; S4. For unassigned tasks and idle unmanned vehicles, an improved ant colony algorithm is used for collaborative optimization to generate the globally optimal allocation scheme. S5. Introduce a real-time feedback mechanism during execution to dynamically adjust the allocation scheme to adapt to environmental changes.

2. The multi-unmanned vehicle field collaborative task allocation method based on auction-ant colony algorithm according to claim 1, characterized in that: The formula for calculating the environmental impact factor is: in, For the first The first driverless car to perform Environmental impact factor for the task; For communication quality; The complexity of the terrain.

3. The multi-unmanned vehicle field collaborative task allocation method based on auction-ant colony algorithm according to claim 1, characterized in that: The comprehensive feature matrix elements are the coupling values ​​of capability matching degree and environmental constraints. The formula for calculating these values ​​is as follows: Where Ci,l is the capability value of the i-th autonomous vehicle in the l-th dimension; Dj,l is the requirement value of the j-th task in the l-th dimension; and Ei,j is the environmental constraint coefficient. This is a capability dimension.

4. The multi-unmanned vehicle field collaborative task allocation method based on auction-ant colony algorithm according to claim 3, characterized in that: The energy level refinement stage includes the following steps: S10. Calculate the average fitness of the task according to the type of unmanned vehicle; in, For the first Autonomous vehicles for the first Average valuation of the task It is a driverless car type. This refers to the number of this type of driverless vehicle; S11. Select the type of driverless vehicle with the highest bid in the first round of auction as the candidate type; The formula for submitting a quote is: in, To address the submission of quotes for the task, This is the quotation coefficient; The formula for the highest bid is: in, Choose the type with the highest price.

5. The multi-unmanned vehicle field collaborative task allocation method based on auction-ant colony algorithm according to claim 4, characterized in that: The single-pole refining stage includes the following steps: S20. Calculate the individual fitness of each autonomous vehicle in the candidate types, and fuse the capability matching degree and location cost; in, For the first Autonomous vehicles for the first Average valuation of the task; The maximum task distance within the scene; This is the position attenuation factor; The distance from the autonomous vehicle to the mission point; S21. High-adaptability tasks are directly allocated through the second round of auction, while low-adaptability tasks are included in the collaborative optimization pool. in, Submit a bid for the autonomous vehicle; β is the individual bid coefficient, if the highest bid... , The auction threshold is set to 0.6 to avoid low-fitness assignments; in this case, the task is directly assigned to the corresponding autonomous vehicle. Then the task will be included in the collaborative optimization pool; For the first Autonomous vehicles for the first The average estimate of the task.

6. The multi-unmanned vehicle field collaborative task allocation method based on auction-ant colony algorithm according to claim 5, characterized in that: The collaborative optimization includes the following steps: S30. Map the tasks and idle unmanned vehicles in the collaborative optimization pool to the ant colony algorithm solution space. S31. Define the heuristic function and integrate the pre-allocated guiding factor; Where ηi,j represents the heuristic degree of selecting the "autonomous vehicle-task" matching pair; For pre-allocated guiding coefficients, To pre-allocate association factors; For the first Autonomous vehicles for the first Average valuation of the task; S32. Through iterative optimization of pheromone updates and path selection, the fitness targets are task completion rate, total energy consumption, and load balancing. The formula for path selection: Where ηi,j represents the heuristic degree of selecting the "autonomous vehicle-task" matching pair; For pheromone importance factors; This is the importance factor of the heuristic function; for The autonomous vehicle's mission is to match pheromone concentrations between vehicles. For the first The probability of an ant choosing the autonomous vehicle task at the number of iterations; This represents the number of ants.

7. The multi-unmanned vehicle field collaborative task allocation method based on auction-ant colony algorithm according to claim 6, characterized in that: The information update includes partial updates and global updates; Formula for partial update: in, for Pheromones concentration in autonomous vehicles and task matching pairs at all times; for Pheromones concentration in autonomous vehicles and task matching pairs at all times; The pheromone evaporation coefficient; This represents the initial value of the pheromone concentration. The formula for global update: in, for Pheromones concentration in autonomous vehicles and task matching pairs at all times; for Pheromones concentration in autonomous vehicles and task matching pairs at all times; The pheromone evaporation coefficient; The pheromone increment for the optimal path; in, This represents the fitness value of the optimal path. The fitness function is: Where R is the task completion rate; Etotal is the total energy consumption; and σ is the load balancing coefficient. The weighting coefficients are satisfied. .

8. A multi-unmanned vehicle field collaborative task allocation system, characterized in that: include: The central control platform is used to execute the multi-unmanned vehicle field collaborative task allocation method based on the auction-ant colony algorithm as described in any one of claims 1-7; Multiple driverless vehicles equipped with condition monitoring modules; The environmental perception module is used to collect information on terrain complexity and communication quality. The communication module is used for issuing commands and providing status feedback.

Images