An unmanned tractor cooperative seeding operation scheduling control system

Abstract

The application discloses an unmanned tractor cooperative seeding operation scheduling control system and relates to the technical field of task scheduling. A target seeding area is acquired, discrete points are determined according to the contour of an obstacle-free area, and the task equivalent points of the obstacle-free area are calculated to obtain a task equivalent point set. Random sample points are generated for the obstacle-free area, all sample points of the obstacle area are acquired, and a sample point set is formed. Adjacent node connection operation is performed on the task equivalent point set and the sample point set to obtain a target path graph. The target path graph is input into a target optimal model to obtain an optimal task sequence. Target path planning operation is performed on the optimal task sequence to obtain a globally optimal task path. Each unmanned tractor is distributed with different required operation tasks according to the globally optimal task path. The unmanned tractor operation scheduling efficiency is improved through the optimal task sequence.

Description

Technical Field

[0001] This invention relates to the field of task scheduling technology, specifically to a scheduling and control system for collaborative sowing operations using unmanned tractors. Background Technology

[0002] In modern distributed computing and intelligent control systems, the task scheduling system plays a crucial "last mile" command role. It precisely delivers macro-level operational instructions generated by the upper-level task planning layer or the local real-time decision-making module to each unmanned tractor. Through a series of scheduling facilities—task decomposition algorithms, dynamic priority queues, and the construction of a multi-time-period, multi-constraint collaborative network—it ensures that each unit maintains consistent rhythm, balanced load, and zero conflicts across time and space, thereby providing a stable, reliable, and flexible operational order guarantee for large-scale unmanned seeding.

[0003] The prior art (publication number CN114370081A) discloses a method and system for scheduling multiple underground loader collaborative operations, which can reduce the number of control platforms and drivers. The driver only needs to use a remote operation control platform to manually control another vehicle to scoop ore while one vehicle is automatically traveling to the destination. After scooping the ore, the vehicle is allowed to automatically travel to the destination again. This process is repeated, and multiple vehicles can operate simultaneously. One driver can control multiple vehicles, increasing operating efficiency and saving material and labor costs.

[0004] The aforementioned patent enables one driver to control multiple vehicles, but in actual operation, collision defects or waiting situations may occur between multiple vehicles, thereby reducing work efficiency, increasing task completion time, and reducing job scheduling efficiency. Summary of the Invention

[0005] The purpose of this invention is to solve the problem of reduced work scheduling efficiency by proposing an unmanned tractor collaborative seeding operation scheduling and control system.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] An unmanned tractor collaborative seeding operation scheduling and control system includes the following modules:

[0008] Seeding area module: Obtain the target seeding area, which includes an obstacle area and an obstacle-free area;

[0009] Task equivalent point module: Determine each discrete point based on the outline of the barrier-free area, calculate the task equivalent point of the barrier-free area for each discrete point, and obtain the task equivalent point set by acquiring the task equivalent point of the barrier-free area;

[0010] Target path graph module: Randomly generate sample points in the obstacle-free area, obtain all sample points in the obstacle area and form a sample point set, and perform adjacent node connection operation on the task equivalent point set and the sample point set to obtain the target path graph.

[0011] Optimal Task Sequence Module: Inputs the target path graph into the target optimal model to obtain the optimal task sequence;

[0012] Global path planning module: Performs target path planning operations on the optimal task sequence to obtain the globally optimal task path;

[0013] Central control module: The central control unit is used to receive the globally optimal task path and distribute different required work tasks to each unmanned tractor according to the globally optimal task path. The work task is the target sowing area that the unmanned tractor needs to perform.

[0014] Optionally, the task equivalent point module includes: a discrete point set module and a region point set acquisition module.

[0015] The discrete point set module is used to acquire each discrete point of the target task region and determine the discrete point set of the target task region based on the discrete points; the target task region is any task region of the target seeding region.

[0016] The region point set acquisition module is used to determine the task equivalent points of the task region based on the discrete point set, and obtain the task equivalent points of all task regions to obtain the task equivalent point set.

[0017] The formula for calculating the task equivalent point includes:

[0018]

[0019] in, Indicates the equivalent center point of the target task region. These are the three-dimensional coordinates of the equivalent center point, and h represents the total number of discrete points in the target task region. , , These are the three-dimensional coordinates of the m-th discrete point, where m = 1, 2, ..., h.

[0020] Optionally, the target path graph module includes: a path graph node set module and a dynamic radius module.

[0021] The path graph node set module is used to combine the task equivalent point set and the sample point set to obtain the path graph node set;

[0022] The dynamic radius module is used to filter the set of adjacent nodes according to the dynamic radius and determine whether there is a collision between the target node and the target adjacent node. The target node is any node in the path graph node set, and the target adjacent node is any adjacent node in the adjacent node set. If there is no collision between the target node and the target adjacent node, an edge is constructed. All edges of all task regions are obtained, and the target path graph is constructed based on all edges.

[0023] Optionally, the calculation process for the dynamic radius includes:

[0024] The formula for calculating the dynamic radius is:

[0025]

[0026] Where R represents the dynamic radius and d represents the spatial dimension. Lebesgue measure representing an unobstructed region Indicates an area free of obstacles. Let n represent the volume of a d-dimensional unit sphere, and n represent the total number of nodes corresponding to the node set in the path graph.

[0027] Optionally, the principle and process of the target optimal model include:

[0028] The collision-free path cost between each node is calculated based on the target path graph, and the target cost matrix is ​​constructed using the collision-free path cost between each node.

[0029] An initial task sequence is obtained by randomly generating the target path graph;

[0030] The optimal task sequence is obtained by optimizing the initial task sequence using the LKH algorithm.

[0031] Optionally, the initial task sequence can be randomly generated from the target path graph, including:

[0032] A1: Select the two nodes corresponding to the path cost with the minimum cost in the cost matrix and add them to the target task sequence;

[0033] A2: From the remaining nodes, select the node with the lowest path cost to the last node in the target task sequence and add it to the target task sequence;

[0034] A3: Repeat step A2 until all nodes are incorporated into the target task sequence. After the traversal is complete, record the target task sequence as the initial task sequence.

[0035] Optionally, the optimal task sequence can be obtained by optimizing the initial task sequence using the LKH algorithm, including:

[0036] S1: Generate an initial solution by performing the nearest neighbor method on the initial task sequence, and calculate the total cost of the initial solution based on the target cost matrix;

[0037] S2: Randomly select an edge from the initial solution as the starting edge, and apply the λ-opt rule to the starting edge and the other edges of the initial solution to obtain the target alternating path set;

[0038] S3: Calculate the total cost corresponding to all target alternation paths in the target alternation path set, and the change in the total cost corresponding to each target alternation path before and after the calculation of alternation;

[0039] When the change in total cost is less than zero and the absolute value of the change in total cost is the largest, it is considered the optimal alternation path;

[0040] S4: The latest solution is determined through the optimal alternating path, the total cost of the latest solution is calculated, and the initial solution is replaced with the latest solution;

[0041] S5: Repeat steps S2-S4 until the preset maximum number of iterations is met or the total cost decrease is less than the preset cost threshold, then the optimal solution is obtained, and the optimal solution is denoted as the optimal task sequence.

[0042] Optionally, the global path planning module includes: an initialization module, a minimum path module, and an optimal task sequence module.

[0043] The initialization module is used to initialize the entry point of all tasks, where the entry point is the specific location where the unmanned tractor begins to execute a certain task.

[0044] The minimized path module is used to obtain the entry point of the target task, the entry point of the preceding task, and the entry point of the following task; and to calculate the entry point of the target task with the minimized path cost based on the entry point of the target task, the entry point of the preceding task, and the entry point of the following task, wherein the target task is any task in the optimal task sequence.

[0045] The optimal task sequence module is used to repeatedly adjust the entry point of the optimal task sequence until the total path cost tends to a stable cost threshold or reaches the maximum number of iterations. Then, the optimized optimal task sequence is output, which is denoted as the global optimal task path.

[0046] Optionally, the entry point of the target task that minimizes the path cost, calculated based on the entry point of the target task, the entry point of the preceding task, and the entry point of the succeeding task, includes:

[0047] The formula for calculating the entry point of the objective task that minimizes path cost is as follows:

[0048]

[0049] in, This represents the entry point of the objective task that minimizes path cost. , and These represent the entry points of the preceding task, the target task, and the subsequent task, respectively; COST() represents the collision-free path cost function. {} represents the independent variable of the function that obtains the minimum value, and j represents the current index of the target task.

[0050] Optionally, the entry point for initializing all tasks includes:

[0051] Obtain the target task, its preceding task, and its succeeding task; connect the equivalent points of the preceding and succeeding tasks to obtain the straight-line distance to the target.

[0052] The points corresponding to the shortest distances between the target task and its predecessor and successor tasks are determined by the straight-line distance to the target, and are denoted as the entry points of the target task.

[0053] The beneficial effects of this invention are:

[0054] This invention transforms the target sowing area into a target path map, and then processes the target path map using a target optimization model to output the optimal task sequence, thereby saving the completion time of sowing operations, improving the scheduling efficiency of unmanned tractors, and reducing the probability of unmanned tractors colliding or waiting. Attached Figure Description

[0055] Figure 1 This is a framework diagram of a collaborative seeding operation scheduling and control system for unmanned tractors;

[0056] Figure 2 A flowchart illustrating the principle of a target optimal model provided in this embodiment of the invention.

[0057] Figure 3 The flowchart shows how the LKH algorithm in this invention optimizes the initial task sequence to obtain the optimal task sequence. Detailed Implementation

[0058] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0059] See Figure 1This invention provides a scheduling and control system for collaborative seeding operations using unmanned tractors, which includes the following modules:

[0060] Seeding Area Module: Obtains the target seeding area, which includes obstacle areas and obstacle-free areas;

[0061] Task equivalent point module: Determine each discrete point based on the outline of the barrier-free area, calculate the task equivalent point of the barrier-free area for each discrete point, and obtain the task equivalent point set by acquiring the task equivalent point of the barrier-free area;

[0062] Target path graph module: Randomly generate sample points in the obstacle-free area, obtain all sample points in the obstacle area, and form a sample point set. Perform adjacent node connection operations on the task equivalent point set and the sample point set to obtain the target path graph.

[0063] Optimal Task Sequence Module: Inputs the target path graph into the target optimal model to obtain the optimal task sequence;

[0064] Global path planning module: Performs target path planning operations on the optimal task sequence to obtain the globally optimal task path;

[0065] Central control module: The central control unit is used to receive the globally optimal task path and distribute different required work tasks to each unmanned tractor according to the globally optimal task path. The work task is the target sowing area that the unmanned tractor needs to perform.

[0066] The unmanned tractor collaborative sowing operation scheduling and control system provided by the present invention transforms the target sowing area into a target path map, and then processes the target path map through the target optimal model to output the optimal task sequence, thereby saving the completion time of sowing operations, improving the scheduling efficiency of unmanned tractor operations, and reducing the probability of unmanned tractors colliding or waiting.

[0067] Specifically, the target sowing area can be irregular farmland, etc.; the specific methods for randomly generating each sample point in the unobstructed area can be pseudo-random sequences, low-difference sequences, etc.

[0068] In one implementation, discrete points are determined based on the outline of the barrier-free region. The shape of the outline of the barrier-free region includes regular shapes (polygonal shapes, circular shapes, rectangular shapes, etc.) and irregular shapes (curved surfaces, etc.).

[0069] In one implementation, the task equivalent point module includes: a discrete point set module and a region point set acquisition module.

[0070] The discrete point set module is used to acquire each discrete point in the target task region and determine the discrete point set of the target task region based on the discrete points; the target task region is any task region of the target seeding region;

[0071] The region point set acquisition module is used to determine the task equivalent points of the task region based on the discrete point set, and to obtain the task equivalent point set of all task regions.

[0072] The formulas for calculating the task equivalent point include:

[0073]

[0074] in, Indicates the equivalent center point of the target task region. These are the three-dimensional coordinates of the equivalent center point, and h represents the total number of discrete points in the target task region. , , These are the three-dimensional coordinates of the m-th discrete point, where m = 1, 2, ..., h.

[0075] One implementation addresses the challenge of using irregular task regions (such as points, lines, and polygons) directly as "nodes" by simplifying each task region into a single equivalent center point. This transforms complex multi-region tasks into point set problems, clearing the way for subsequent task sequence optimization. Furthermore, this simplification is based on the mean of all discrete points within the region, preserving the spatial characteristics of the task region to a large extent and ensuring smooth subsequent path planning and other operations based on the equivalent center point.

[0076] In one implementation, the target path graph module includes: a path graph node set module and a dynamic radius module.

[0077] The path graph node set module is used to combine the task equivalent point set and the sample point set to obtain the path graph node set;

[0078] The dynamic radius module is used to filter the set of adjacent nodes based on the dynamic radius and determine whether there is a collision between the target node and the target adjacent node. The target node is any node in the path graph node set, and the target adjacent node is any adjacent node in the adjacent node set. If there is no collision between the target node and the target adjacent node, an edge is constructed. All edges of all task regions are obtained, and the target path graph is constructed based on all edges.

[0079] In one implementation, the calculation process for the dynamic radius includes:

[0080] The formula for calculating the dynamic radius is:

[0081]

[0082] Where R represents the dynamic radius and d represents the spatial dimension. Lebesgue measure representing an unobstructed region Indicates an area free of obstacles. Let n represent the volume of a d-dimensional unit sphere, and n represent the total number of nodes corresponding to the node set in the path graph.

[0083] In one implementation, by randomly generating sampling points in an obstacle-free area and combining them with the task's equivalent center point to form a node set, the planning space can be fully covered, providing abundant node resources for subsequent path planning and ensuring sufficient space support for path search. Secondly, during the neighbor node connection process, neighbors are selected based on the dynamic radius and collision-free detection is performed. This ensures both the rationality and safety of the connections between nodes, and by utilizing the dynamic radius calculation formula, the asymptotic optimality of the path graph is ensured, making path planning based on this graph closer to the optimal solution. This approach effectively balances the efficiency and quality of path planning, laying a solid and efficient foundation for subsequently calculating the collision-free path cost between any two task equivalent center points. It allows complex path planning problems to be solved more conveniently and accurately within an ordered graph structure, thereby improving the efficiency of subsequent collaborative sowing by unmanned tractors.

[0084] See Figure 2 In one implementation, the principle and process of the target optimal model include:

[0085] The collision-free path cost between each node is calculated based on the target path graph, and the target cost matrix is ​​constructed using the collision-free path cost between each node.

[0086] An initial task sequence is obtained by randomly generating the target path graph;

[0087] The optimal task sequence is obtained by optimizing the initial task sequence using the LKH algorithm.

[0088] In one implementation, the initial task sequence is obtained by randomly generating the target path graph, including:

[0089] A1: Select the two nodes corresponding to the path cost with the minimum cost in the cost matrix and add them to the target task sequence;

[0090] A2: From the remaining nodes, select the node with the lowest path cost to the last node in the target task sequence and add it to the target task sequence;

[0091] A3: Repeat step A2 until all nodes are incorporated into the target task sequence. After the traversal is complete, record the target task sequence as the initial task sequence.

[0092] See Figure 3In one implementation, the optimal task sequence is obtained by optimizing the initial task sequence using the LKH algorithm, including:

[0093] S1: Generate an initial solution by using the nearest neighbor method on the initial task sequence, and calculate the total cost of the initial solution based on the target cost matrix;

[0094] S2: Randomly select an edge from the initial solution as the starting edge, and apply the λ-opt rule to the starting edge and the other edges of the initial solution to obtain the target alternating path set;

[0095] S3: Calculate the total cost corresponding to all target alternation paths in the target alternation path set, and the change in the total cost of each target alternation path before and after the calculation of alternation;

[0096] When the change in total cost is less than zero and the absolute value of the change in total cost is the largest, it is considered the optimal alternation path.

[0097] S4: Determine the latest solution through the optimal alternating path, calculate the total cost of the latest solution, and replace the initial solution with the latest solution;

[0098] S5: Repeat steps S2-S4 until the preset maximum number of iterations is met or the total cost decrease is less than the preset cost threshold. Then, the optimal solution is obtained and the optimal solution is denoted as the optimal task sequence.

[0099] Specifically, the λ-opt rule includes:

[0100] 2-opt: The simplest local search; eliminates path intersections by swapping two pairs of edges.

[0101] Specific steps: Select two non-adjacent edges (such as AB and CD), disconnect them, and then reconnect them as AC and BD (or AD and BC) to form a new path. If the new path is shorter, accept the change.

[0102] 3-opt: Disconnect three edges and then reconnect them in another way. There are 7 possible reconnection methods (including the original path).

[0103] λ-opt: A generalization of the above operation, it disconnects λ edges at once and then reconnects them. The larger λ is, the stronger the search capability, but the computational complexity also increases dramatically.

[0104] In one implementation, the global path planning module includes: an initialization module, a path minimization module, and an optimal task sequence module.

[0105] The initialization module is used to initialize the entry point for all tasks. The entry point is the specific position where the unmanned tractor begins to execute a task.

[0106] The minimized path module is used to obtain the entry point of the target task, the entry point of the preceding task, and the entry point of the following task; and calculates the entry point of the target task with the minimum path cost based on the entry point of the target task, the entry point of the preceding task, and the entry point of the following task. The target task is any task in the optimal task sequence.

[0107] The optimal task sequence module is used to repeatedly adjust the entry point of the optimal task sequence until the total cost of the path tends to a stable cost threshold or the maximum number of iterations is reached. Then, the optimized optimal task sequence is output, which is denoted as the global optimal task path.

[0108] In one implementation, the entry point of each field and each sowing task is uniformly encoded to prevent tractors from repeatedly compacting already sown areas due to initial positional deviations, thus preserving soil moisture. Subsequently, the minimum path module performs dynamic triangulation on the entry points of the current task and its preceding and succeeding tasks, calculating the shortest collision-free path in real time. This reduces the travel distance for a single task, minimizing fuel and tire wear. The optimal task sequence module uses total cost as the convergence objective, repeatedly swapping task orders and synchronously adjusting corresponding entry points. While ensuring no reseeding or missed sowing, this shortens the scheduling time for multiple tractors to collaboratively sow tens of thousands of acres, significantly reducing seed, fuel, and labor costs, and improving operational efficiency and sowing quality consistency.

[0109] In one implementation, the entry point of the target task that minimizes the path cost is calculated based on the entry point of the target task, the entry point of the preceding task, and the entry point of the succeeding task, including:

[0110] The formula for calculating the entry point of the objective task that minimizes path cost:

[0111]

[0112] in, This represents the entry point of the objective task that minimizes path cost. , and These represent the entry points of the preceding task, the target task, and the subsequent task, respectively; COST() represents the collision-free path cost function. {} represents the independent variable of the function that obtains the minimum value, and j represents the current index of the target task.

[0113] In one implementation, the entry point for initializing all tasks includes:

[0114] Obtain the target task, its preceding task, and its succeeding task. Connect the equivalent points of the preceding and succeeding tasks to obtain the straight-line distance to the target.

[0115] The points corresponding to the shortest distances between the target task and its predecessor and successor tasks are determined by the straight-line distance to the target, and are denoted as the entry point of the target task.

[0116] It should be noted that, in this document, terms such as “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0117] Although embodiments of the invention have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made to these embodiments without departing from the principles and spirit of the invention.

Claims

1. A scheduling and control system for cooperative seeding operations using unmanned tractors, characterized in that, The system includes: Seeding area module: Obtain the target seeding area, which includes an obstacle area and an obstacle-free area; Task equivalent point module: Determine each discrete point based on the outline of the barrier-free area, calculate the task equivalent point of the barrier-free area for each discrete point, and obtain the task equivalent point set by acquiring the task equivalent point of the barrier-free area; Target path graph module: Randomly generate sample points in the obstacle-free area, obtain all sample points in the obstacle area and form a sample point set, and perform adjacent node connection operation on the task equivalent point set and the sample point set to obtain the target path graph. Optimal Task Sequence Module: Inputs the target path graph into the target optimal model to obtain the optimal task sequence; Global path planning module: Performs target path planning operations on the optimal task sequence to obtain the globally optimal task path; Central control module: The central control terminal is used to receive the globally optimal task path and distribute different required work tasks to each unmanned tractor according to the globally optimal task path. The work task is the target sowing area that the unmanned tractor needs to perform. The task equivalent point module includes: a discrete point set module and a region point set acquisition module. The discrete point set module is used to acquire each discrete point of the target task region and determine the discrete point set of the target task region based on the discrete points; the target task region is any task region of the target seeding region. The region point set acquisition module is used to determine the task equivalent points of the task region based on the discrete point set, and obtain the task equivalent points of all task regions to obtain the task equivalent point set. The formula for calculating the task equivalent point includes: ； in, Indicates the equivalent center point of the target task region. These are the three-dimensional coordinates of the equivalent center point, and h represents the total number of discrete points in the target task region. , , These are the three-dimensional coordinates of the m-th discrete point, where m = 1, 2, ..., h; The target path graph module includes: a path graph node set module and a dynamic radius module. The path graph node set module is used to combine the task equivalent point set and the sample point set to obtain the path graph node set; The dynamic radius module is used to filter the set of adjacent nodes according to the dynamic radius and determine whether there is a collision between the target node and the target adjacent node. The target node is any node in the path graph node set, and the target adjacent node is any adjacent node in the adjacent node set. If there is no collision between the target node and the target adjacent node, an edge is constructed. All edges of all task regions are obtained, and the target path graph is constructed based on all edges. The calculation process of the dynamic radius includes: The formula for calculating the dynamic radius is: ； Where R represents the dynamic radius and d represents the spatial dimension. Lebesgue measure representing an unobstructed region Indicates an area free of obstacles. Let n represent the volume of a d-dimensional unit sphere, and n represent the total number of nodes corresponding to the node set of the path graph. The principle and process of the target optimal model include: The collision-free path cost between each node is calculated based on the target path graph, and the target cost matrix is ​​constructed using the collision-free path cost between each node. An initial task sequence is obtained by randomly generating the target path graph; The optimal task sequence is obtained by optimizing the initial task sequence using the LKH algorithm; The global path planning module includes: an initialization module, a path minimization module, and an optimal task sequence module. The initialization module is used to initialize the entry point of all tasks, where the entry point is the specific location where the unmanned tractor begins to execute a certain task. The minimized path module is used to obtain the entry point of the target task, the entry point of the preceding task, and the entry point of the following task; and to calculate the entry point of the target task with the minimized path cost based on the entry point of the target task, the entry point of the preceding task, and the entry point of the following task, wherein the target task is any task in the optimal task sequence. The optimal task sequence module is used to repeatedly adjust the entry point of the optimal task sequence until the total path cost tends to a stable cost threshold or reaches the maximum number of iterations. Then, the optimized optimal task sequence is output, which is denoted as the global optimal task path.

2. The unmanned tractor collaborative seeding operation scheduling and control system according to claim 1, characterized in that, The initial task sequence is obtained by randomly generating the target path graph, including: A1: Select the two nodes corresponding to the path cost with the minimum cost in the cost matrix and add them to the target task sequence; A2: From the remaining nodes, select the node with the lowest path cost to the last node in the target task sequence and add it to the target task sequence; A3: Repeat step A2 until all nodes are incorporated into the target task sequence. After the traversal is complete, record the target task sequence as the initial task sequence.

3. The unmanned tractor collaborative seeding operation scheduling and control system according to claim 1, characterized in that, The process of optimizing the initial task sequence using the LKH algorithm to obtain the optimal task sequence includes: S1: Generate an initial solution by performing the nearest neighbor method on the initial task sequence, and calculate the total cost of the initial solution based on the target cost matrix; S2: Randomly select an edge from the initial solution as the starting edge, and apply the λ-opt rule to the starting edge and the other edges of the initial solution to obtain the target alternating path set; S3: Calculate the total cost corresponding to all target alternation paths in the target alternation path set, and the change in the total cost corresponding to each target alternation path before and after the calculation of alternation; When the change in total cost is less than zero and the absolute value of the change in total cost is the largest, it is considered the optimal alternation path; S4: The latest solution is determined through the optimal alternating path, the total cost of the latest solution is calculated, and the initial solution is replaced with the latest solution; S5: Repeat steps S2-S4 until the preset maximum number of iterations is met or the total cost decrease is less than the preset cost threshold, then the optimal solution is obtained, and the optimal solution is denoted as the optimal task sequence.

4. The unmanned tractor collaborative seeding operation scheduling and control system according to claim 1, characterized in that, The step of calculating the entry point of the target task that minimizes the path cost based on the entry point of the target task, the entry point of the preceding task, and the entry point of the succeeding task includes: The formula for calculating the entry point of the objective task that minimizes path cost is as follows: ； in, This represents the entry point of the objective task that minimizes path cost. , and These represent the entry points of the preceding task, the target task, and the subsequent task, respectively; COST() represents the collision-free path cost function. {} represents the independent variable of the function that obtains the minimum value, and j represents the current index of the target task.

5. The unmanned tractor collaborative sowing operation scheduling and control system according to claim 1, characterized in that, The entry points for initializing all tasks include: Obtain the target task, its preceding task, and its succeeding task; connect the equivalent points of the preceding and succeeding tasks to obtain the straight-line distance to the target. The points corresponding to the shortest distances between the target task and its predecessor and successor tasks are determined by the straight-line distance to the target, and are denoted as the entry points of the target task.

Images