A method and system for planning agricultural machinery operations for intelligent land management

Abstract

The present invention discloses a method and system for planning agricultural machinery operations for intelligent plot management, which relates to the field of precision agriculture technology, including using a drone to obtain plot data, generating a plot image according to the posture information of the drone and performing 3D mapping of the plot, using a semantic segmentation network to extract plot image features and refine the operation area; generating an operation parameter distribution map based on the semantic segmentation result, calculating the optimal operation path of the agricultural machinery according to the operation parameter distribution map and the 3D mapping of the plot; detecting obstacles on the forward path of the agricultural machinery in real time and dynamically adjusting the agricultural machinery operation path, storing data generated during the agricultural machinery operation planning process and implementing access control. The present invention can effectively adapt to complex terrain and instantly changing environmental conditions, dynamically respond to changes in the operation site, and improve the optimization level of the agricultural machinery operation path and the pertinence of the agricultural machinery operation planning.

Description

Technical Field

[0001] The present invention relates to the field of precision agriculture technology, and in particular to an agricultural machinery operation planning method and system for intelligent plot management. Background Art

[0002] In the development of modern precision agriculture, agricultural operation planning and management technology has made significant progress, especially in plot management. Technological advances have made agricultural operations more efficient and accurate. Early agricultural operation planning mainly relied on manual experience for design. This method is not only time-consuming but also inefficient and difficult to cope with large-scale or changing agricultural environments. With the application of remote sensing technology and geographic information system (GIS), the acquisition and processing of plot data has become more automated and systematic. The introduction of unmanned aerial vehicle (UAV) technology has revolutionized the accuracy and convenience of data collection. UAVs can cover vast plots in a short time and obtain high-resolution image data. In addition, advances in computer vision, especially the development of image processing and machine learning algorithms, such as semantic segmentation networks, have further improved the accuracy of plot data processing and the level of automation of operations. Traditional methods often cannot accurately identify and classify various crops and soil conditions in the processing and analysis of plot images, resulting in a lack of personalization and accuracy in operation planning. Secondly, although drones can efficiently collect plot data, their data integration and real-time path planning capabilities are still insufficient and cannot dynamically respond to changes on the scene. Summary of the invention

[0003] In view of the problems existing in the above-mentioned existing agricultural machinery operation planning methods and systems for intelligent plot management, the present invention is proposed.

[0004] Therefore, the problem to be solved by the present invention is that traditional methods often cannot accurately identify and classify various crops and soil conditions in the processing and analysis of plot images, resulting in a lack of personalization and precision in operation planning. Secondly, although drones can efficiently collect plot data, their capabilities for data integration and real-time path planning are still insufficient, and they are unable to dynamically respond to changes on site.

[0005] To solve the above technical problems, the present invention provides the following technical solutions: a method for agricultural machinery operation planning for intelligent plot management, which includes: using a drone to obtain plot data, generating a plot image according to the posture information of the drone and performing 3D mapping of the plot, and using a semantic segmentation network to extract plot image features to refine the operation area; generating an operation parameter distribution map based on the semantic segmentation result, and calculating the optimal operation path of the agricultural machinery according to the operation parameter distribution map and the 3D mapping of the plot; real-time detection of obstacles on the forward path of the agricultural machinery and dynamic adjustment of the agricultural machinery operation path, storing data generated during the agricultural machinery operation planning process and implementing access control.

[0006] As a preferred solution of the agricultural machinery operation planning method for intelligent plot management of the present invention, wherein: the use of a drone to obtain plot data, generate a plot image according to the position information of the drone and perform 3D mapping of the plot refers to using a drone equipped with a high-resolution camera and a multi-spectral sensor, installing an RTK-GPS mobile receiver on the drone, using flight planning software to design a flight route covering the entire plot, using the drone to capture plot images and spectral data in real time, and recording the position coordinates of each image in real time through the RTK-GPS system;

[0007] Preprocess the collected plot images, including adjusting the contrast and brightness of the plot images, and automatically correcting image color aberration and uneven illumination using image processing software;

[0008] Construct the scale space of the image, define the DoG function D(x,y,σ), and detect the local extreme points in the image:

[0009] D(x,y,σ)=L(x,y,p'σ)-L(x,y,σ),

[0010] Where L(x,y,σ) is the convolution result of image I with Gaussian kernel G at scale σ, p' is the constant for adjusting the scale, L(x,y,p'σ) is the image obtained after Gaussian filtering with scale p'σ at position (x,y), and D represents the Gaussian difference image at position (x,y) and scale σ;

[0011] Find local maximum and minimum values ​​in the DoG function as potential key feature points, and determine the position of each potential key feature point;

[0012] Assign multiple directions to each key feature point, and in the area around each key feature point, select the image gradient and direction statistics of the corresponding area according to the scale of the key feature point to generate a feature descriptor;

[0013] Use the fast nearest neighbor search algorithm to match SIFT feature descriptors, calculate the geometric transformation matrix between the plot images through the matched SIFT feature descriptors, and align the corresponding feature points between the plot images;

[0014] The positional relationship between images is calculated based on the RTK-GPS coordinate data of each plot image and the flight posture information of the drone, and the image stitching is automatically completed to generate a seamlessly stitched plot panorama;

[0015] The disparity d of the land image is calculated based on the matched SIFT feature descriptors using stereo matching technology, and the depth of each pixel is calculated based on the disparity d and the focal length of the drone camera and the distance between the drone cameras;

[0016] Convert the position of each pixel and the calculated depth value into coordinates in three-dimensional space to generate a point cloud, and use pixel alignment feature fusion technology to integrate point cloud data from different perspectives;

[0017] Use convolutional neural networks to embed semantic information into the updated point cloud to generate point cloud data rich in conceptual information;

[0018] The point cloud data is processed by Pix4D to generate a 3D map of the plot and perform geometric correction on the generated 3D map.

[0019] As a preferred solution of the agricultural machinery operation planning method for intelligent plot management of the present invention, wherein: the use of a semantic segmentation network to extract plot image features and refine the operation area refers to using a pre-trained DeepLab v3+ model as a semantic segmentation tool, inputting the pre-processed plot image data into the DeepLab v3+ model for forward propagation, obtaining the category probability distribution of each pixel in the plot image output by the model, selecting the category with the highest probability as the label of the pixel point to generate a preliminary classification map of the entire plot;

[0020] According to the converted category label map, different work areas are identified and defined, and each work area is clustered according to the category to form a continuous pixel block;

[0021] The Canny edge detection algorithm is used to extract edges from the image obtained from semantic segmentation, and morphological operations are applied to the detected boundaries to eliminate noise and fill boundary discontinuities, and merge close and similar operating areas.

[0022] As a preferred solution of the agricultural machinery operation planning method for intelligent plot management of the present invention, wherein: the generation of the operation parameter distribution map based on the semantic segmentation result refers to collecting relevant operation area data from the GIS system, defining the fertilization amount based on soil nutrient demand and crop absorption rate, defining the spraying amount based on pest and disease monitoring data and crop resistance, and defining the irrigation water amount based on soil moisture and rainfall forecast;

[0023] The calculated fertilizer amount, spray amount and irrigation water amount are mapped to the specific plot location through GIS software, different colors are used to identify different operation requirements, and the operation parameter distribution map is generated using GIS tools.

[0024] As a preferred solution of the agricultural machinery operation planning method for intelligent plot management of the present invention, wherein: the calculation of the optimal operation path of the agricultural machinery according to the operation parameter distribution map and the 3D mapping of the plot refers to importing 3D mapping data and the operation parameter distribution map into the GIS software, superimposing the terrain layer of the 3D mapping with the operation parameter distribution layer, analyzing the terrain data, identifying the influence of terrain slope and height changes on the operation parameters and marking the areas requiring special treatment, adjusting the operation parameters according to the terrain analysis results and updating the operation parameter distribution map, superimposing the adjusted operation parameters on the 3D mapping of the plot to generate a comprehensive data layer and marking the specific operation attributes and terrain attributes corresponding to each area in the comprehensive data layer;

[0025] The initial operation path is randomly generated based on the comprehensive data layer, the search space is composed according to the geographical boundaries of the plot and the operation path points, the particle swarm size of the PSO algorithm is determined, the initial position of each particle is randomly generated, the particle speed is initialized as a random vector, and two optimization goals are defined: minimizing the total length of the path and maximizing the coverage of the key operation area. The key parameters of the PSO algorithm are set, including the maximum number of iterations, inertia weight, individual factor and social learning factor;

[0026] Run the PSO algorithm for multiple iterations. For each iteration, calculate the path cost of each particle and evaluate the effectiveness of the path using the data in the comprehensive data layer.

[0027] Update the personal best position of each particle and the global best position in the group, and update the speed and position of each particle;

[0028] After each iteration, the quality of the newly generated path is evaluated, and the path with the lowest cost is selected as the optimal path. The final path data is formatted and imported into the navigation system of the agricultural machinery to guide the operation of the agricultural machinery.

[0029] As a preferred solution of the agricultural machinery operation planning method for intelligent plot management of the present invention, wherein: the real-time detection of obstacles on the forward path of the agricultural machinery and the dynamic adjustment of the agricultural machinery operation path are to guide the agricultural machinery equipped with high-precision laser radar and visual camera to operate on the calculated optimal path, and the configured equipment is used to monitor the surrounding environment of the agricultural machinery in real time to collect image data, and the image data collected by the two devices are integrated through the Kalman filter to generate an obstacle map;

[0030] Use a density-based fast clustering algorithm to identify and classify laser mine data, analyze the identified obstacle clusters, determine the core area and boundary points of the obstacle clusters, use the convex hull algorithm to calculate the minimum outer shape of each obstacle cluster, sort the points by x-coordinate, if the x-coordinates are the same, sort them by y-coordinate, traverse the sorted points from left to right to build the upper convex hull, traverse from right to left to build the lower convex hull, for each convex hull, determine and adjust the point order through the cross product to determine the minimum outer shape of each obstacle cluster;

[0031] For obstacles with complex shapes, the rotating caliper algorithm is further used to extract the minimum circumscribed rectangle and measure the length and width of the obstacle;

[0032] Combined with the image data obtained from the visual camera, the visual features of the obstacle are recognized by color and texture, and the shape, scale and visual features of the obstacle are combined to fully characterize each obstacle;

[0033] By calculating the Euclidean distance from the position of the agricultural machinery to the convex hull of the obstacle, the distance from the current position of the agricultural machinery to each obstacle is obtained. The calculation result is updated to the obstacle map, and the position and dynamics of the obstacles are tracked in real time. A cost map is generated based on the distance and direction of the obstacles. The cost of each direction is adjusted according to the distance and size of the obstacle. The VFH+ algorithm is used to analyze the cost map and select the direction with the lowest cost as the new direction of travel:

[0034] The original polar coordinate obstacle histogram is constructed from the lidar data. Each sector in the histogram represents the cumulative density of obstacles in a specific direction. The cumulative cost of each sector is calculated using the following formula:

[0035]

[0036] Where C(μ) is the total cost of direction μ, g j is the distance from the jth obstacle to the agricultural machinery, μ j is the direction of the jth obstacle, N is the total number of obstacles detected, a is the distance weight factor, s(μ-μ j ) is a Gaussian function;

[0037] A new travel path is calculated based on the travel direction combined with the current speed and position of the agricultural machinery, and the newly calculated travel path is imported into the navigation system of the agricultural machinery to guide the operation of the agricultural machinery.

[0038] As a preferred solution of the agricultural machinery operation planning method for intelligent plot management described in the present invention, wherein: the storage of data generated in the agricultural machinery operation planning process and the implementation of access control refers to the collection of data from the sensors and navigation systems of the agricultural machinery and the storage in the database, setting security access measures, the database will store the data in the cloud, and regularly perform integrity checks on the stored data and the backup data. After the test is completed, an integrity test record is generated and synchronously stored in the database, allowing authorized users to view the data.

[0039] Another object of the present invention is to provide an agricultural machinery operation planning system for intelligent plot management, which comprises:

[0040] The data collection module is used to collect relevant data of the plot using drones and analyze and extract features of the data to perform 3D mapping of the plot;

[0041] The semantic segmentation module is used to perform semantic segmentation and recognition on the plot image and define different operation areas, and analyze different operation areas to generate operation parameter distribution maps;

[0042] The path planning module is used to generate a comprehensive data layer based on the operation parameter distribution map and the 3D mapping of the plot and calculate the optimal operation path of the agricultural machinery using the PSO algorithm;

[0043] The obstacle recognition module is used to operate agricultural machinery on the optimal operation path, identify obstacles on the operation path in real time, and dynamically adjust the operation path of the agricultural machinery;

[0044] The data storage module is used to collect data generated in agricultural machinery operation planning and to back up and control access to the data.

[0045] A computer device comprises: a memory and a processor; the memory stores a computer program, and the processor implements the steps of an agricultural machinery operation planning method for intelligent land plot management when executing the computer program.

[0046] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of a method for planning agricultural machinery operations for intelligent plot management.

[0047] The beneficial effects of the present invention are as follows: the present invention obtains plot data by using a drone and generates plot images for 3D mapping, extracts plot image features through a semantic segmentation network to refine the operation area and generate an operation parameter distribution map, calculates the optimal operation path of the agricultural machinery according to the operation parameter distribution map and 3D mapping, and monitors obstacles on the way of the agricultural machinery in real time to dynamically adjust the operation direction of the agricultural machinery. The present invention can effectively adapt to complex terrain and instantly changing environmental conditions, dynamically respond to changes in the operation site, and improve the optimization level of the agricultural machinery operation path and the pertinence of the agricultural machinery operation planning. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0049] Figure 1 Schematic diagram of the process of agricultural machinery operation planning method for intelligent field management.

[0050] Figure 2 Schematic diagram of the structure of the agricultural machinery operation planning system for intelligent plot management. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific embodiments of the present invention are described in detail below in conjunction with the accompanying drawings.

[0052] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0053] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The term "in one embodiment" that appears in different places in this specification does not necessarily refer to the same embodiment, nor is it a separate or selective embodiment that is mutually exclusive with other embodiments.

[0054] Example 1, reference Figure 1 , which is the first embodiment of the present invention, and which provides an agricultural machinery operation planning method for intelligent plot management. The agricultural machinery operation planning method for intelligent plot management includes:

[0055] S1. Use drones to obtain plot data, generate plot images based on the drone’s position information and perform 3D mapping of the plots, and use semantic segmentation networks to extract plot image features and refine the operation area;

[0056] Specifically, using drones to obtain plot data, generating plot images based on the drone's position information and performing 3D mapping of the plots means using drones equipped with high-resolution cameras and multispectral sensors, installing RTK-GPS mobile receivers on the drones, using flight planning software to design a flight route covering the entire plot, using drones to capture plot images and spectral data in real time, and recording the position coordinates of each image in real time through the RTK-GPS system;

[0057] Preprocess the collected plot images, including adjusting the contrast and brightness of the plot images, and automatically correcting image color aberration and uneven illumination using image processing software;

[0058] Construct the scale space of the image:

[0059] L(x,y,σ)=G(x,y,σ)*I(x,y),

[0060] Where L(x,y,σ) is the convolution result of image I with Gaussian kernel G at scale σ, G(x,y,σ) is the Gaussian function used to smooth the image I(x,y), I(x,y) is the pixel intensity value of the original input image at position (x,y), and x,y are the pixel coordinates;

[0061] Define the DoG function D(x,y,σ) to detect local extreme points in the image:

[0062] D(x,y,σ)=L(x,y,p'σ)-L(x,y,σ),

[0063] Where L(x,y,σ) is the convolution result of image I at scale σ and Gaussian kernel G, p' is a constant for adjusting the scale, and a value greater than 1 is selected to generate an image at a slightly higher scale, L(x,y,p'σ) is the image obtained after Gaussian filtering of scale p'σ at position (x,y), and D represents the Gaussian difference image at position (x,y) and scale σ;

[0064] Find local maxima and minima in the DoG function as potential key feature points, and determine the location of each potential key feature point:

[0065]

[0066] Where D(X) is the derivative in scale space, X = (x, y, σ) T is the offset of the key feature point, which is used to refine the position and scale of the key feature point. D is the value of the DoG function in the scale space at the position and scale of the key point. is the first-order derivative of the DoG function D with respect to the vector X, indicating the sensitivity of D to the position x, y and scale σ near the key feature point, calculated by numerical differentiation method, It is the second-order derivative of the DoG function D with respect to the vector X, which is used to evaluate the curvature of D at the key point position and help determine whether the point is a true local extreme point. It is calculated by numerical differentiation method, and T is the transposition operation;

[0067] Assign multiple directions to each key feature point, and in the area around each key feature point, select the image gradient and direction statistics of the corresponding area according to the scale of the key feature point to generate a feature descriptor;

[0068] Use the fast nearest neighbor search algorithm to match SIFT feature descriptors:

[0069]

[0070] Where O(f1,f2) is the Euclidean distance between two feature descriptors f1 and f2, each descriptor consists of a 128-dimensional SIFT feature vector, l is the index, and f 1l and f 2l are the values ​​of the first and second feature descriptors in the lth dimension respectively;

[0071] Through the matched SIFT feature descriptors, the geometric transformation matrix between the plot images is calculated to align the corresponding feature points between the plot images:

[0072]

[0073] In the formula, (x1, y1) is the coordinate of the point in the original image, (x1′, y1′) is the coordinate after transformation, a, b, c, d in the matrix control rotation and scaling, t x , t y Control translation;

[0074] The positional relationship between images is calculated based on the RTK-GPS coordinate data of each plot image and the flight posture information of the drone, and the image stitching is automatically completed to generate a seamlessly stitched plot panorama;

[0075] The disparity d of the plot image is calculated based on the matched SIFT feature descriptors using stereo matching techniques:

[0076] d = x l -x r ,

[0077] In the formula, x l and x r These are the horizontal coordinates of the same scene point on the left and right images respectively;

[0078] The depth of each pixel is calculated based on the parallax d, the focal length of the drone camera, and the distance between the drone cameras;

[0079] Convert the position of each pixel and the calculated depth value into coordinates in three-dimensional space to generate a point cloud, and use pixel alignment feature fusion technology to integrate point cloud data from different perspectives;

[0080] Use convolutional neural networks to embed semantic information into the updated point cloud to generate point cloud data rich in conceptual information:

[0081] v k =Conv(f P,k ),

[0082] In the formula, v k is the conceptual vector of the point cloud vertex, Conv is the convolution operation, and f P,k is the pixel-aligned feature, p is the pixel in the convolution operation, and k is the kth point cloud vertex;

[0083] The point cloud data is processed by Pix4D to generate a 3D map of the plot and perform geometric correction on the generated 3D map.

[0084] The RTK-GPS mobile receiver is used to accurately record the exact position of the drone when each image is taken, which is crucial for subsequent image processing and 3D mapping. This step allows the precise geolocation of plot images, providing data support for high-precision 3D mapping, significantly improving the practicality of plot data and the accuracy of operation planning. The SIFT algorithm is used to detect key feature points in the image and generate feature descriptors. These feature points and descriptors are used for accurate matching between images, which improves the accuracy of image processing, especially in plots with complex terrain and diverse vegetation coverage. It can effectively identify and match key land features, thereby optimizing the quality of image stitching and 3D modeling. The geometric transformation matrix between the matched SIFT feature descriptors is calculated to align the corresponding feature points between plot images, allowing accurate alignment and stitching of plot images captured from multiple perspectives and time points, greatly enhancing the information integrity and application value of plot images. Obstacles on the path of agricultural machinery are detected in real time and the operation path is dynamically adjusted to ensure the continuity and safety of agricultural machinery operations. The flexibility and responsiveness of agricultural machinery operations are improved, which is particularly important for coping with complex and changing farmland environments, ensuring operation efficiency and safety.

[0085] Furthermore, using a semantic segmentation network to extract plot image features and refine the work area means using a pre-trained DeepLab v3+ model as a semantic segmentation tool, inputting the pre-processed plot image data into the DeepLab v3+ model for forward propagation, obtaining the category probability distribution of each pixel in the plot image output by the model, and selecting the category with the highest probability as the label of the pixel point to generate a preliminary classification map of the entire plot;

[0086] According to the converted category label map, different work areas are identified and defined, and each work area is clustered according to the category to form a continuous pixel block;

[0087] The Canny edge detection algorithm is used to extract edges from the image obtained from semantic segmentation, and morphological operations are applied to the detected boundaries to eliminate noise and fill boundary discontinuities, and merge close and similar operating areas.

[0088] By using the DeepLab v3+ model to perform precise semantic segmentation on plot images, the present invention can effectively distinguish soil types, crop growth conditions, and other key geographic features. The automated processing of this step significantly improves the speed and accuracy of data processing and reduces the need for manual intervention. The converted category label map can clearly distinguish different operating areas, such as cultivated areas, protected areas, etc. This helps to achieve more sophisticated agricultural management and operation execution. The use of the Canny algorithm improves the recognition accuracy of image edges, and morphological operations further optimize the performance of these edges, eliminating noise and filling gaps, which is very important for the overall image quality and subsequent processing.

[0089] S2, generating an operation parameter distribution map based on the semantic segmentation results, and calculating the optimal operation path of the agricultural machinery according to the operation parameter distribution map and the 3D mapping of the plot;

[0090] Specifically, generating an operation parameter distribution map based on semantic segmentation results means collecting relevant operation area data from the GIS system, including crop types, growth stages, soil nutrient conditions, and historical operation data, etc., defining the amount of fertilizer based on soil nutrient requirements and crop absorption rate, defining the amount of spraying based on pest and disease monitoring data and crop resistance, and defining the amount of irrigation water based on soil moisture and rainfall forecasts;

[0091] The calculated fertilizer amount, spray amount and irrigation water amount are mapped to the specific plot location through GIS software, different colors are used to identify different operation requirements, and the operation parameter distribution map is generated using GIS tools.

[0092] The GIS system integrates a variety of data related to the plot, including crop types, growth stages, soil nutrients, and historical operation data. These data provide comprehensive information on the plot status and help to formulate accurate operation plans. The GIS system can update the plot data in real time to ensure that the operation plan is based on the latest information and improve the accuracy and timeliness of decision-making. The fertilizer machine rate is defined according to the soil nutrient requirements and crop absorption rate, which can ensure that each plot receives the right amount of nutrients and avoid over-fertilization or under-fertilization. The spraying amount is defined according to the pest monitoring data and crop resistance, which can effectively control pests and diseases, while reducing the use of pesticides and reducing the risk of environmental pollution. The irrigation water volume is defined according to soil moisture and rainfall forecasts to ensure that crops receive the right amount of water support and avoid water resource waste and soil erosion caused by excessive irrigation. The calculated fertilizer amount, spraying amount, and irrigation water amount are mapped to the specific plot location through GIS software, and different colors are used to identify different operation requirements, which can intuitively display the specific operation requirements of each area. The operation parameter distribution map provides an important reference for the optimization of the operation path of agricultural machinery, which can guide agricultural machinery to operate along the best path and improve operation efficiency.

[0093] Further, calculating the optimal operation path of the agricultural machinery according to the operation parameter distribution map and the 3D mapping of the plot refers to importing the 3D mapping data and the operation parameter distribution map into the GIS software, superimposing the terrain layer of the 3D mapping with the operation parameter distribution layer, analyzing the terrain data, identifying the influence of terrain slope and height changes on the operation parameters and marking the areas requiring special treatment, adjusting the operation parameters according to the terrain analysis results and updating the operation parameter distribution map, superimposing the adjusted operation parameters on the 3D mapping of the plot again to generate a comprehensive data layer and marking the specific operation attributes and terrain attributes corresponding to each area in the comprehensive data layer;

[0094] The initial operation path is randomly generated based on the comprehensive data layer. The search space is formed according to the geographic boundaries of the plot and the operation path points (the entrance, exit, key operation area and detour points around obstacles of the plot). The particle swarm size of the PSO algorithm is determined, and the initial position of each particle is randomly generated. The particle speed is initialized as a random vector. Two optimization goals are defined: minimizing the total length of the path and maximizing the coverage of the key operation area. The key parameters of the PSO algorithm are set, including the maximum number of iterations, inertia weight, individual factor and social learning factor.

[0095] Run the PSO algorithm for multiple iterations. For each iteration, calculate the path cost of each particle and use the data in the comprehensive data layer to evaluate the effectiveness of the path. The path cost formula is expressed as:

[0096]

[0097] Where A is the path cost, (xt ,y t ) is the coordinate of the tth point in the path, M is the total number of points in the path, x t+1 ,y t+1 The horizontal and vertical coordinates of the t+1th point on the path;

[0098] Update the personal best position of each particle and the global best position in the group, and update the velocity v of each particle i+1 and position u' i+11 :

[0099] v i+1 =w·v i +c1·r1·(p best -u' i )+c2·r2·(g best -u' i ),

[0100] u' i+1 =u' i +v i+1 ,

[0101] In the formula, v i is the speed of the particle at iteration i, w is the inertia weight, which is used to control the retention of the particle speed, c1 and c2 are individual learning factors and social learning factors, r1 and r2 are random numbers generated in the interval [0,1], which are used to introduce randomness so that the search process can jump out of the local optimum and increase exploration, p best is the particle's personal best historical position, that is, the best solution found by the particle in the search process so far, u' i is the position of the particle at iteration i, g best is the global optimal position, i.e., the optimal solution found by all particles in the search process so far;

[0102] After each iteration, the quality of the newly generated path is evaluated, and the path with the lowest cost is selected as the optimal path. The final path data is formatted and imported into the navigation system of the agricultural machinery to guide the operation of the agricultural machinery.

[0103] Overlaying the 3D terrain layer with the operation parameter distribution layer can fully understand the geography and operation requirements of the plot, provide a comprehensive view, and help to plan operations more accurately. By analyzing terrain data, the impact of terrain slope and height changes on operation parameters can be identified, and areas that require special treatment can be marked in advance. This preprocessing improves the safety and effectiveness of operations. The initial operation path is optimized multiple times through the PSO algorithm, which can effectively find the shortest and optimal path covering the key operation area, greatly improving the operation efficiency. The PSO algorithm combines global search and local search to avoid falling into the local optimum and ensure that the global optimal solution is found. Each iteration evaluates the effectiveness of the path based on the path cost and the comprehensive data layer data to ensure dynamic adjustment and optimization of path planning. By adjusting the values ​​of inertia weights, individual factors, and social factors, the PSO algorithm can find the best balance between exploring new paths and developing existing paths, improving search efficiency. The path cost formula can accurately evaluate the total cost of each path to ensure that the lowest cost path is selected as the optimal path. The path cost evaluation provides a quantitative basis for decision-making, which helps to select the optimal solution among multiple candidate paths and ensure the economy of the operation.

[0104] S3, real-time detection of obstacles on the path of agricultural machinery and dynamic adjustment of the agricultural machinery operation path, storage of data generated during the agricultural machinery operation planning process and implementation of access control;

[0105] Specifically, the obstacles on the forward path of agricultural machinery are detected in real time and the operation path of agricultural machinery is adjusted dynamically. The agricultural machinery equipped with high-precision laser radar and visual camera operates on the calculated optimal path. The configured equipment monitors the surrounding environment of agricultural machinery in real time to collect image data, and the image data collected by the two devices are integrated through the Kalman filter to generate an obstacle map.

[0106] Use a density-based fast clustering algorithm to identify and classify laser mine data, analyze the identified obstacle clusters, determine the core area and boundary points of the obstacle clusters, use the convex hull algorithm to calculate the minimum outer shape of each obstacle cluster, sort the points by x-coordinate, if the x-coordinates are the same, sort them by y-coordinate, traverse the sorted points from left to right to build the upper convex hull, traverse from right to left to build the lower convex hull, for each convex hull, determine and adjust the point order through the cross product to determine the minimum outer shape of each obstacle cluster;

[0107] For obstacles with complex shapes, the rotating caliper algorithm is further used to extract the minimum circumscribed rectangle and measure the length and width of the obstacle;

[0108] Combined with the image data obtained from the visual camera, the visual features of the obstacle are recognized by color and texture, and the shape, scale and visual features of the obstacle are combined to fully characterize each obstacle;

[0109] By calculating the Euclidean distance from the position of the agricultural machinery to the convex hull of the obstacle, the distance from the current position of the agricultural machinery to each obstacle is obtained. The calculation result is updated to the obstacle map, and the position and dynamics of the obstacles are tracked in real time. A cost map is generated based on the distance and direction of the obstacles. The cost of each direction is adjusted according to the distance and size of the obstacle. The VFH+ algorithm is used to analyze the cost map and select the direction with the lowest cost as the new direction of travel:

[0110] The original polar coordinate obstacle histogram is constructed from the lidar data. Each sector in the histogram represents the cumulative density of obstacles in a specific direction. The cumulative cost of each sector is calculated using the following formula:

[0111]

[0112] Where C(μ) is the total cost of direction μ, g j is the distance from the jth obstacle to the agricultural machinery, μ j is the direction of the jth obstacle, N is the total number of obstacles detected, a is the distance weight factor calculated by the optimization algorithm, and s(μ-μ j ) is a Gaussian function used to increase or decrease the cost according to the angle difference;

[0113] A new travel path is calculated based on the travel direction combined with the current speed and position of the agricultural machinery, and the newly calculated travel path is imported into the navigation system of the agricultural machinery to guide the operation of the agricultural machinery.

[0114] High-precision LiDAR and visual cameras are used to monitor the surrounding environment of agricultural machinery in real time, and obstacles can be accurately detected and located, which improves the safety and efficiency of agricultural machinery operations. The Kalman filter integrates the data of LiDAR and visual cameras to generate an accurate obstacle map, which enhances the accuracy of environmental perception. The density-based fast clustering algorithm is used to identify and classify LiDAR data, which can accurately identify and distinguish different types of obstacles. The convex hull algorithm and rotating caliper algorithm can accurately calculate the minimum outer shape and minimum circumscribed rectangle of obstacles, providing accurate obstacle size and shape information for path planning. Combined with the image data of the visual camera, the Kalman filter can accurately identify and distinguish different types of obstacles. It can provide more comprehensive obstacle information and improve the accuracy of obstacle recognition by performing color recognition and texture analysis. It combines the shape, scale and visual features of the obstacle to fully characterize each obstacle, thereby improving the accuracy of recognition and processing. It calculates the Euclidean distance from the position of the agricultural machinery to the convex hull of the obstacle, tracks the position and dynamics of the obstacle in real time, generates a cost map based on the distance and direction of the obstacle, uses the VFH+ algorithm to select the direction of travel with the lowest cost, and realizes dynamic path optimization. It formats the optimized path data and imports it into the navigation system of the agricultural machinery, which can guide the precise operation of agricultural machinery, reduce human intervention, and improve the level of operation automation.

[0115] Furthermore, storing the data generated during the agricultural machinery operation planning process and implementing access control means collecting data from the sensors and navigation systems of the agricultural machinery and storing them in the database, setting up security access measures, the database will back up the stored data in the cloud, and regularly perform integrity checks on the stored data and backup data. After the test is completed, an integrity test record is generated and synchronously stored in the database, allowing authorized users to view the data.

[0116] By collecting data in real time from the sensors and navigation systems of agricultural machinery, it is possible to ensure that key information in all operations is accurately recorded. These data include location, speed, path, obstacle information, etc. All collected data are stored in a centralized database for subsequent management, analysis and call, avoiding the inconvenience caused by decentralized data management and the risk of data loss. By setting up multi-level security access measures (such as identity authentication, permission management, encrypted transmission, etc.), the security of stored data can be effectively protected to prevent unauthorized access and data leakage.

[0117] Example 2, reference Figure 2 , which is the second embodiment of the present invention, and which is different from the previous embodiment, provides an agricultural machinery operation planning system for intelligent land management, which includes:

[0118] The data collection module is used to collect relevant data of the plot using drones and analyze and extract features of the data to perform 3D mapping of the plot;

[0119] The semantic segmentation module is used to perform semantic segmentation and recognition on the plot image and define different operation areas, and analyze different operation areas to generate operation parameter distribution maps;

[0120] The path planning module is used to generate a comprehensive data layer based on the operation parameter distribution map and the 3D mapping of the plot and calculate the optimal operation path of the agricultural machinery using the PSO algorithm;

[0121] The obstacle recognition module is used to operate agricultural machinery on the optimal operation path, identify obstacles on the operation path in real time, and dynamically adjust the operation path of the agricultural machinery;

[0122] The data storage module is used to collect data generated in agricultural machinery operation planning and to back up and control access to the data.

[0123] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk, etc., which can store program codes.

[0124] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0125] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0126] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, any one of the following technologies known in the art or a combination thereof can be used to implement: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

Claims

1. A method for planning agricultural machinery operations for intelligent plot management, characterized in that: include, Use drones to obtain land data, generate land images based on the drone's position information and perform 3D mapping of the land, and use semantic segmentation networks to extract land image features and refine the work area; Generate an operation parameter distribution map based on the semantic segmentation results, and calculate the optimal operation path of the agricultural machinery based on the operation parameter distribution map and the 3D mapping of the plot; Real-time detection of obstacles on the path of agricultural machinery and dynamic adjustment of the agricultural machinery operation path, storage of data generated during agricultural machinery operation planning and implementation of access control; The calculating of the optimal operation path of the agricultural machinery according to the operation parameter distribution map and the 3D mapping of the plot refers to importing the 3D mapping data and the operation parameter distribution map into the GIS software, superimposing the terrain layer of the 3D mapping with the operation parameter distribution layer, analyzing the terrain data, identifying the influence of terrain slope and height changes on the operation parameters and marking the areas requiring special treatment, adjusting the operation parameters according to the terrain analysis results and updating the operation parameter distribution map, superimposing the adjusted operation parameters on the 3D mapping of the plot again to generate a comprehensive data layer and marking the specific operation attributes and terrain attributes corresponding to each area in the comprehensive data layer; The initial operation path is randomly generated based on the comprehensive data layer, the search space is composed according to the geographical boundaries of the plot and the operation path points, the particle swarm size of the PSO algorithm is determined, the initial position of each particle is randomly generated, the particle speed is initialized as a random vector, and two optimization goals are defined: minimizing the total length of the path and maximizing the coverage of the key operation area. The key parameters of the PSO algorithm are set, including the maximum number of iterations, inertia weight, individual factor and social learning factor; Run the PSO algorithm for multiple iterations. For each iteration, calculate the path cost of each particle and evaluate the effectiveness of the path using the data in the comprehensive data layer. Update the personal best position of each particle and the global best position in the group, and update the speed and position of each particle; After each iteration, the quality of the newly generated path is evaluated, and the path with the lowest cost is selected as the optimal path. The final path data is formatted and imported into the navigation system of the agricultural machinery to guide the operation of the agricultural machinery.

2. The method for planning agricultural machinery operations for intelligent plot management according to claim 1, characterized in that: The method of using a drone to obtain land parcel data, generating land parcel images according to the position information of the drone and performing 3D mapping of the land parcel refers to using a drone equipped with a high-resolution camera and a multi-spectral sensor, installing an RTK-GPS mobile receiver on the drone, using flight planning software to design a flight route covering the entire land parcel, using the drone to capture land parcel images and spectral data in real time, and recording the position coordinates of each image in real time through an RTK-GPS system; Preprocess the collected plot images, including adjusting the contrast and brightness of the plot images, and automatically correcting image color aberration and uneven illumination using image processing software; Construct the scale space of the image, define the DoG function, and detect local extreme points in the image: ， In the formula, is the convolution result of image I with Gaussian kernel G at scale σ, p' is the constant for adjusting the scale, is the image obtained after Gaussian filtering of scale p'σ at position (x, y), and D represents the Gaussian difference image at position (x, y) and scale σ; Find local maximum and minimum values ​​in the DoG function as potential key feature points, and determine the position of each potential key feature point; Assign multiple directions to each key feature point, and in the area around each key feature point, select the image gradient and direction statistics of the corresponding area according to the scale of the key feature point to generate a feature descriptor; Use the fast nearest neighbor search algorithm to match SIFT feature descriptors, calculate the geometric transformation matrix between the plot images through the matched SIFT feature descriptors, and align the corresponding feature points between the plot images; The positional relationship between images is calculated based on the RTK-GPS coordinate data of each plot image and the flight posture information of the drone, and the image stitching is automatically completed to generate a seamlessly stitched plot panorama; The disparity d of the land image is calculated based on the matched SIFT feature descriptors using stereo matching technology, and the depth of each pixel is calculated based on the disparity d and the focal length of the drone camera and the distance between the drone cameras; Convert the position of each pixel and the calculated depth value into coordinates in three-dimensional space to generate a point cloud, and use pixel alignment feature fusion technology to integrate point cloud data from different perspectives; Use convolutional neural networks to embed semantic information into the updated point cloud to generate point cloud data rich in conceptual information; The point cloud data is processed by Pix4D to generate a 3D map of the plot and perform geometric correction on the generated 3D map.

3. The method for planning agricultural machinery operations for intelligent plot management according to claim 2, characterized in that: The method of using a semantic segmentation network to extract plot image features and refine the work area refers to using a pre-trained DeepLab v3+ model as a semantic segmentation tool, inputting the pre-processed plot image data into the DeepLab v3+ model for forward propagation, obtaining the category probability distribution of each pixel in the plot image output by the model, and selecting the category with the highest probability as the label of the pixel point to generate a preliminary classification map of the entire plot; According to the converted category label map, different work areas are identified and defined, and each work area is clustered according to the category to form a continuous pixel block; The Canny edge detection algorithm is used to extract edges from the image obtained from semantic segmentation, and morphological operations are applied to the detected boundaries to eliminate noise and fill boundary discontinuities, and merge close and similar operating areas.

4. The method for planning agricultural machinery operations for intelligent plot management according to claim 3, characterized in that: Generating the operation parameter distribution map based on the semantic segmentation result refers to collecting relevant operation area data from the GIS system, defining the amount of fertilizer based on soil nutrient demand and crop absorption rate, defining the amount of spraying based on pest and disease monitoring data and crop resistance, and defining the amount of irrigation water based on soil moisture and rainfall forecast; The calculated fertilizer amount, spray amount and irrigation water amount are mapped to the specific plot location through GIS software, different colors are used to identify different operation requirements, and the operation parameter distribution map is generated using GIS tools.

5. The method for planning agricultural machinery operations for intelligent plot management according to claim 4, characterized in that: The real-time detection of obstacles on the forward path of the agricultural machinery and the dynamic adjustment of the agricultural machinery operation path are to guide the agricultural machinery equipped with high-precision laser radar and visual camera to operate on the calculated optimal path, and the configured equipment is used to monitor the surrounding environment of the agricultural machinery in real time to collect image data, and the image data collected by the two devices are integrated through the Kalman filter to generate an obstacle map; Use a density-based fast clustering algorithm to identify and classify laser mine data, analyze the identified obstacle clusters, determine the core area and boundary points of the obstacle clusters, use the convex hull algorithm to calculate the minimum outer shape of each obstacle cluster, sort the points by x-coordinate, if the x-coordinates are the same, sort them by y-coordinate, traverse the sorted points from left to right to build the upper convex hull, traverse from right to left to build the lower convex hull, for each convex hull, determine and adjust the point order through the cross product to determine the minimum outer shape of each obstacle cluster; For obstacles with complex shapes, the rotating caliper algorithm is further used to extract the minimum circumscribed rectangle and measure the length and width of the obstacle; Combined with the image data obtained from the visual camera, the visual features of the obstacle are recognized by color and texture, and the shape, scale and visual features of the obstacle are combined to fully characterize each obstacle; By calculating the Euclidean distance from the position of the agricultural machinery to the convex hull of the obstacle, the distance from the current position of the agricultural machinery to each obstacle is obtained. The calculation result is updated to the obstacle map, and the position and dynamics of the obstacles are tracked in real time. A cost map is generated based on the distance and direction of the obstacles. The cost of each direction is adjusted according to the distance and size of the obstacle. The VFH+ algorithm is used to analyze the cost map and select the direction with the lowest cost as the new direction of travel: The original polar coordinate obstacle histogram is constructed from the lidar data. Each sector in the histogram represents the cumulative density of obstacles in a specific direction. The cumulative cost of each sector is calculated using the following formula: ， In the formula, It is the direction The total cost, g j is the distance from the jth obstacle to the agricultural machinery, is the direction of the jth obstacle, N is the total number of obstacles detected, a is the distance weight factor, is a Gaussian function; A new travel path is calculated based on the travel direction combined with the current speed and position of the agricultural machinery, and the newly calculated travel path is imported into the navigation system of the agricultural machinery to guide the operation of the agricultural machinery.

6. The method for planning agricultural machinery operations for intelligent plot management according to claim 5, characterized in that: The storage of data generated during the agricultural machinery operation planning process and the implementation of access control refers to the collection of data from the sensors and navigation systems of the agricultural machinery and the storage in the database, the setting of security access measures, the database will store the data in the cloud, and regularly perform integrity checks on the stored data and the backup data. After the test is completed, an integrity test record is generated and synchronously stored in the database, allowing authorized users to view the data.

7. An agricultural machinery operation planning system for intelligent plot management based on the agricultural machinery operation planning method for intelligent plot management according to any one of claims 1 to 6, characterized in that: include, The data collection module is used to collect relevant data of the plot using drones and analyze and extract features of the data to perform 3D mapping of the plot; The semantic segmentation module is used to perform semantic segmentation and recognition on the plot image and define different operation areas, and analyze different operation areas to generate operation parameter distribution maps; The path planning module is used to generate a comprehensive data layer based on the operation parameter distribution map and the 3D mapping of the plot and calculate the optimal operation path of the agricultural machinery using the PSO algorithm; The obstacle recognition module is used to operate agricultural machinery on the optimal operation path, identify obstacles on the operation path in real time, and dynamically adjust the operation path of the agricultural machinery; The data storage module is used to collect data generated in agricultural machinery operation planning and to back up and control access to the data.

8. A computer device comprising: Memory and processor; The memory stores a computer program, characterized in that when the processor executes the computer program, the steps of the agricultural machinery operation planning method for intelligent plot management according to any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the agricultural machinery operation planning method for intelligent plot management according to any one of claims 1 to 6 are implemented.

Images