Intelligent crop maintenance method, server and medium

Abstract

The invention discloses an intelligent crop maintenance method, a server and a medium, and relates to the technical field of electric digital data processing. The method comprises the following steps: acquiring a real-time robot attitude and a real-time sensor position in a target planting area; based on the real-time robot posture, the real-time sensor position and the telescopic length range of the telescopic device, determining a position set which the soil data sensor can reach; determining a predicted distance between each position in the position set and the corresponding crop root system; determining an acquisition position based on the predicted distance, a moving route between the soil data sensor and each position and a soil density distribution condition of the target planting area; sending the acquisition position to the follow-up robot; acquiring soil data acquired by a soil data sensor at the acquisition position; evaluating the growth condition of crops in the target planting area through a preset evaluation model; and controlling the preset intelligent maintenance equipment to execute the corresponding maintenance task. By implementing the technical scheme, the data acquisition accuracy of the soil data sensor is improved.

Description

Technical Field

[0001] This application relates to the field of electronic digital data processing, and in particular, to a method, a server, and a medium for intelligent crop maintenance. Background Art

[0002] With the acceleration of the agricultural modernization process, the technology of intelligent crop maintenance has become a key means to improve the efficiency and quality of agricultural production. By real-time monitoring of environmental data such as soil and meteorology, and precisely regulating maintenance measures such as irrigation and fertilization, it can effectively achieve cost reduction, efficiency increase, and green sustainable development.

[0003] Currently, intelligent agricultural systems often adopt Internet of Things technology. By deploying preset sensors in farmland, key parameters such as soil temperature and humidity, pH value, and nutrient content are real-time monitored, and decision optimization for crop maintenance is carried out by combining big data analysis and machine learning algorithms. Some systems also introduce edge computing and cloud computing architectures to achieve local data processing and cloud collaborative analysis.

[0004] However, the metabolic activities of microorganisms in the soil and the drilling and movement of organisms such as earthworms will gradually shift the spatial positions of the preset sensors in the soil. This shift will not only cause the data collected by the sensors to deviate from the target monitoring area, resulting in data distortion of soil temperature and humidity, nutrient concentration, etc., but also interfere with the accurate judgment of the machine learning model on the changing trend of the soil environment. The generated maintenance decisions, such as irrigation and fertilization plans, will be out of touch with the actual needs, which may lead to problems such as water resource waste and soil compaction, and reduce the reliability of the intelligent agricultural system. Summary of the Invention

[0005] This application provides a method, a server, and a medium for intelligent crop maintenance, which can improve the accuracy of soil data sensor data collection.

[0006] In a first aspect, the present application provides a method for intelligent crop maintenance, which is applied to a server of an intelligent agriculture system. The intelligent agriculture system further includes a follow-up robot, a soil data sensor, and a communication gateway. The follow-up robot is disposed in the soil below the crop roots, the soil data sensor is disposed above the follow-up robot, and the soil data sensor is used to collect soil data below the crop roots. The follow-up robot is used to control the built-in telescopic device to move up and down to adjust the position of the soil data sensor. The server is connected to the soil data sensor and the follow-up robot respectively through the communication gateway. The method includes: obtaining the real-time robot posture of the follow-up robot and the real-time sensor position of the soil data sensor in the target planting area, where the follow-up robot cannot autonomously adjust its own position and posture; determining the set of positions that the soil data sensor can reach based on the real-time robot posture, the real-time sensor position, and the telescopic length range of the telescopic device; determining the predicted distance between each position and the corresponding crop roots based on the predicted root growth of the crops corresponding to each position in the position set; determining the acquisition positions corresponding to the movement routes that do not affect crop growth based on the predicted distance, the movement routes between the soil data sensor and each position, and the soil density distribution of the target planting area; sending the acquisition positions to the follow-up robot; after the follow-up robot moves the soil data sensor to the acquisition positions, obtaining the soil data collected by the soil data sensor at the acquisition positions; evaluating the growth condition of the crops in the target planting area through a preset evaluation model based on the soil data, the crop leaf images, and the growth trend videos; and controlling a preset intelligent maintenance device to perform corresponding maintenance tasks based on the growth condition.

[0007] By adopting the above technical solution, the position of the soil data sensor can be dynamically adjusted by the follow-up robot, and the sensor can be dynamically adjusted to move along, so as to better collect sensing data as the plant roots change seasonally. By obtaining the real-time position and real-time posture of the follow-up robot, determining the set of positions that the soil data sensor can reach, and combining the prediction of crop root growth, the movement route, and the soil density distribution to select the acquisition positions, the optimal acquisition positions that the sensor can reach are selected to collect soil data, making the collected data more in line with the data collected at the preset positions, reducing the influence of the position deviation of the follow-up robot (i.e., the position deviation of the sensor) on the accuracy of data collection, and improving the accuracy of soil data collection. At the same time, the growth condition of the crops is evaluated by comprehensively considering the soil data, the crop leaf images, and the growth trend videos, generating maintenance tasks, and controlling the intelligent maintenance device to perform the tasks, realizing the automatic maintenance of the crops and improving the efficiency and quality of agricultural production.

[0008] In some embodiments in combination with some embodiments of the first aspect, based on the predicted distance, the movement route between the soil data sensor and each position, and the soil density distribution of the target planting area, the acquisition positions corresponding to the movement route that does not affect crop growth are determined, specifically including: obtaining the soil density distribution of the target planting area, where the soil density distribution includes the first soil density distribution in the crop root area, the second soil density distribution in the robot movement area, and the third soil density distribution in the biological activity area; based on the movement route between the soil data sensor and each position, the soil density distribution, and the preset density impact on the soil within the preset area range when the soil data sensor moves, determining the target soil density distribution when the soil data sensor reaches each position; based on the soil density distribution and the target soil density distribution, determining the target impact degree of the soil density change on crop growth; obtaining a set of target positions where the target impact degree is within the preset impact degree range; based on the predicted distance and the corresponding preset ideal distance of each target position in the set of target positions, taking the target position corresponding to the smallest distance difference as the acquisition position, where the smallest distance difference is the smallest among all the distance differences of the target positions, and the distance difference is the difference between the predicted distance and the corresponding ideal distance.

[0009] With the above technical solutions, by obtaining the soil density distribution of the target planting area and combining the sensor movement route, the target soil density distribution when the sensor reaches each position is estimated, so as to quantify the soil density change before and after the sensor moves. Furthermore, according to the change situation, the impact degree of the sensor movement on crop growth is determined. A set of target positions with an impact degree within a reasonable range is selected, and the final acquisition position is determined based on the difference between the predicted distance and the ideal distance. This not only ensures that the sensor movement will not cause significant negative impacts on the soil environment and crop growth, but also ensures that the acquisition position maintains the best distance from the crop roots, enabling the collected data to more truly reflect the soil environment information required for crop root growth, further improving the accuracy and effectiveness of soil data acquisition, providing more reliable data support for subsequent maintenance decision-making, and ensuring the stability and sustainability of the crop growth environment.

[0010] In some embodiments in combination with some embodiments of the first aspect, obtaining the soil density distribution of the target planting area specifically includes: dividing the target planting area into a crop root area, a robot moving area, and a biological activity area based on the predicted root growth of each crop in the target planting area, the historical movement trajectories of the follow-up robot and the soil sensor, and the range of soil biological activities; determining the first soil density distribution of the crop root area according to the predicted root growth; determining the number of movements and the movement position deviation of the follow-up robot within a preset time period according to the historical movement trajectory; predicting the biological activity of the biological activity area based on the number of movements and the movement position deviation; determining the second soil density distribution of the robot moving area and the third soil density distribution of the biological activity area based on the biological activity, the historical movement trajectory, and the influence of the soil data sensor on the preset density of the soil within the preset area range during movement.

[0011] By adopting the above technical solution, by comprehensively considering crop root growth, robot movement trajectories, and the range of soil biological activities, the target planting area is partitioned, and the influencing factors of soil density distribution in different areas are defined. Determining the soil density of the crop root area based on the predicted root growth, predicting the biological activity in combination with the historical movement data of the robot, and considering the influence of sensor movement on soil density can more accurately determine the soil density distribution of the robot moving area and the biological activity area, avoiding soil density data deviation caused by fuzzy area division and incomplete consideration of influencing factors, and providing more reliable soil environment data for subsequent determination of the optimal collection position.

[0012] In some embodiments in combination with some embodiments of the first aspect, after the follow-up robot moves the soil data sensor to the collection position and after the step of obtaining the soil data collected by the soil data sensor at the collection position, the method further includes: obtaining the first ideal collection position of the target planting area; when the soil data sensor fails, screening out a second position set within a preset distance range from the first ideal collection position in the first position set that the first soil data sensor can reach, and the first soil data sensor can reach the target planting area; determining the degree of influence of the change in soil density corresponding to each first movement route on crop growth based on the first movement route of the first soil data sensor, the soil density distribution, and the fourth soil density distribution of the first planting area of the first soil data sensor; when there is a degree of influence within the preset degree of influence range, controlling the first soil data sensor to reach the new collection position corresponding to the second movement route with the minimum degree of influence to collect new soil data based on the new collection position; updating the collection position and the soil data based on the new collection position and the new soil data.

[0013] With the above technical solution, when the soil data sensor fails, since the soil data sensor is located below the crop, repairing or replacing the sensor will affect the crop growing above the sensor. At this time, by screening the target planting areas set in other planting areas around the faulty sensor that can reach the faulty sensor, and the first soil data sensor whose arrival position in the target planting area is within a preset range of the ideal acquisition position in the target planting area (i.e., the data collected at this arrival position can reflect the soil data in the target planting area), and evaluating the degree of influence of the movement route of the first soil data sensor on the soil environment, the optimal replacement plan is selected. The soil data of the target planting area is collected according to the optimal replacement plan, avoiding the damage to the crop caused by repairing or replacing the sensor, ensuring the accuracy and continuity of soil data collection, and improving the fault response ability and overall operation efficiency of the intelligent agriculture system.

[0014] Combined with some embodiments of the first aspect, in some embodiments, obtaining the first ideal acquisition position of the target planting area specifically includes: determining the crop growth difference area and the boundary area between different crop types according to the growth conditions and crop types of each crop in the target planting area; determining the root-intensive area based on the predicted root growth conditions of each crop in the target planting area; screening out the target horizontal coordinate positions whose horizontal coordinate positions are in the growth difference area, the boundary area, and the root-intensive area; calculating the target vertical coordinate position based on the ideal distance between the crop roots corresponding to the target horizontal coordinate position and the sensor and the crop root depth; and obtaining the first ideal acquisition position based on the target horizontal coordinate position and the target vertical coordinate position.

[0015] With the above technical solution, based on the crop growth conditions, types, and predicted root growth, the growth difference, boundary, and root-intensive areas are determined to screen out the target horizontal coordinate positions. And by combining the ideal distance and root depth, the target vertical coordinates corresponding to each target horizontal coordinate position are calculated to obtain the first ideal acquisition position, so that the data collected by the sensor at the first ideal acquisition position can better reflect the soil conditions in the key crop growth areas, providing a basis for subsequent evaluation of crop growth and formulation of reasonable maintenance strategies.

[0016] In combination with some embodiments of the first aspect, in some embodiments, after the step of screening out a second set of positions within a preset distance range from the first ideal acquisition position in the first set of positions reachable by the first soil data sensor when the soil data sensor fails, the method further includes: obtaining a third set of positions within the first planting area of the first soil data sensor in the first set of positions; calculating the data confidence of each third position based on the second ideal acquisition position of the first planting area; screening out a fourth set of positions with a data confidence higher than a preset confidence threshold in the third set of positions; planning a first movement route for the first soil data sensor from the current position to the second position and then from the second position to the fourth position.

[0017] With the above technical solution, after screening out a suitable second set of positions from the first ideal acquisition position, the third set of positions within the planting area where the first sensor is located is further locked, and the data confidence is calculated based on the second ideal acquisition position, and the fourth set of positions with high confidence is screened out to ensure the reliability of data acquisition. At the same time, planning the movement route of "current position - second position - fourth position" not only ensures the emergency replacement effect of data acquisition by the faulty sensor, but also takes into account the soil data acquisition requirements in the area where the first sensor is located.

[0018] In combination with some embodiments of the first aspect, in some embodiments, after the step of updating the acquisition position and the soil data based on the new acquisition position and the new soil data, the method further includes: if the first ideal acquisition position and the acquisition position are not within a preset deviation range, determining correction values for each soil parameter in the soil data according to the deviation between the first ideal position and the acquisition position and the soil density distribution; updating the parameter values of each soil parameter in the soil data based on the correction values.

[0019] With the above technical solution, by comparing the first ideal acquisition position with the actual acquisition position, when the deviation between the two exceeds the preset range, the correction values of the soil parameters are determined in combination with the soil density distribution, and the soil data parameters are updated, compensating for the soil data error caused by the deviation of the acquisition position and making the soil data more accurately reflect the true soil conditions of the target planting area.

[0020] In some embodiments in combination with some embodiments of the first aspect, the intelligent agriculture system further includes a drone. The drone is equipped with a camera for collecting images and videos of crops. The drone is used to move the camera to the collection position and perform maintenance tasks. The server is connected to the drone through the communication gateway, and based on the soil data, crop leaf images, and growth trend videos, the growth conditions of the crops in the target planting area are evaluated through a preset evaluation model, specifically including: obtaining the crop leaf images and growth trend videos collected by the camera carried by the drone in the target planting area; inputting the crop leaf images, growth trend videos, and the soil data into the preset evaluation model to obtain the growth conditions of the crops in the target planting area.

[0021] By adopting the above technical solution, the drone is equipped with a camera to collect crop leaf images and growth trend videos, which are jointly input into the preset evaluation model together with the accurately collected soil data, realizing the in-depth fusion analysis of multi-source data. This method breaks through the limitations of a single data dimension, can comprehensively and three-dimensionally evaluate the growth conditions of crops from multiple aspects such as the soil environment and crop phenotypes, and avoids evaluation biases caused by one-sided information. It not only improves the accuracy and reliability of crop growth evaluation, but also provides a more scientific and comprehensive decision-making basis for formulating maintenance tasks, promoting the intelligent and refined development of agricultural production.

[0022] In a second aspect, an embodiment of the present application provides a server, including: one or more processors and a memory; the memory is coupled to the one or more processors, and the memory is used to store computer program code, and the computer program code includes computer instructions. The one or more processors call the computer instructions to cause the server to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0023] In a third aspect, an embodiment of the present application provides a computer-readable storage medium, including instructions, which when running on the server, cause the server to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0024] In a fourth aspect, the present application provides a computer program product, which when running on the server, causes the server to execute the method described in the first aspect and any possible implementation manner in the first aspect.

[0025] It can be understood that the server provided in the second aspect, the storage medium provided in the third aspect, and the computer program product provided in the fourth aspect are all used to execute the method provided in the present application. Therefore, the beneficial effects that can be achieved can refer to the beneficial effects in the corresponding method, which will not be elaborated here.

[0026] One or more technical solutions provided in the embodiments of the present application have at least the following technical effects or advantages: 1. By obtaining the real-time position and real-time attitude of the follow-up robot, the present application determines the set of positions that the soil data sensor can reach, and selects the collection position by combining crop root growth prediction, movement route, and soil density distribution, and selects the optimal collection position that the sensor can reach to collect soil data, making the collected data more in line with the data collected at the preset position, reducing the influence of the position deviation of the follow-up robot (i.e., the position deviation of the sensor) on the accuracy of data collection, and improving the accuracy of soil data collection.

[0027] 2. By screening the target planting areas where other planting areas around the faulty sensor can reach the faulty sensor, and the first soil data sensor whose arrival position in the target planting area is within the preset range from the ideal collection position in the target planting area, and evaluating the influence degree of the movement route of the first soil data sensor on the soil environment, the optimal alternative solution is selected. The soil data of the target planting area is collected according to the optimal replacement solution, avoiding the damage to the crops caused by repairing or replacing the sensor, ensuring the accuracy and continuity of soil data collection, and enhancing the fault response ability and overall operation efficiency of the intelligent agriculture system.

[0028] 3. By comparing the first ideal collection position with the actual collection position, when the deviation between the two exceeds the preset range, the soil parameter correction value is determined in combination with the soil density distribution, and the soil data parameter is updated, making up for the soil data error caused by the collection position deviation, and making the soil data more accurately reflect the real soil condition of the target planting area. Description of the Drawings

[0029] Figure 1 is a schematic structural diagram of a system architecture to which the crop intelligent maintenance method in the embodiments of the present application can be applied; Figure 2 is a schematic flowchart of a crop intelligent maintenance method in the embodiments of the present application; Figure 3 is another schematic flowchart of a crop intelligent maintenance method in the embodiments of the present application; Figure 4 is another schematic structural diagram of a system architecture to which the crop intelligent maintenance method in the embodiments of the present application can be applied; Figure 5 is an exemplary hardware structural diagram of a server in the embodiments of the present application. Detailed Embodiments

[0030] The terms used in the following embodiments of this application are only for the purpose of describing specific embodiments and are not intended to limit this application. As used in the specification and appended claims of this application, the singular forms "a", "an", "the", "above-mentioned", "said", and "this" are also intended to include the plural forms, unless there is a clear indication to the contrary in the context. It should also be understood that the term "and / or" used in this application refers to any or all possible combinations including one or more of the listed items.

[0031] Hereinafter, the terms "first" and "second" are only for descriptive purposes and cannot be construed as implying or suggesting relative importance or implicitly indicating the quantity of the indicated technical features. Thus, the features defined with "first" and "second" may explicitly or implicitly include one or more of such features. In the description of the embodiments of this application, unless otherwise specified, the meaning of "a plurality" is two or more.

[0032] Figure 1 It is a structural schematic diagram of a system architecture to which the crop intelligent maintenance method in the embodiments of this application can be applied.

[0033] Please refer to Figure 1 , the intelligent agriculture system includes a follow-up robot, a soil data sensor, a communication gateway, a drone, a camera, and a server.

[0034] The server, as the core component of the system, is used to analyze and process the data collected by the soil data sensor and the camera, and transmit control instructions to the follow-up robot and the drone through the communication gateway. The communication gateway is used to convert the protocols of the data collected by the soil data sensor and the camera and then transmit them to the server, and convert the format of the control instructions of the server and transmit them to the follow-up robot and the drone. The follow-up robot is arranged in the soil under the crop roots and is used to control the built-in telescopic device to move up and down to adjust the position of the soil data sensor in the soil. The soil data sensor is arranged above the follow-up robot and is used to collect the soil data under the crop roots and transmit the collected soil data to the server through the communication gateway. The drone is used to move the carried camera to a preset collection position to collect data and perform maintenance tasks. The camera is carried on the drone and is used to collect the image data and video data of the crop and transmit the collected data to the server through the communication gateway.

[0035] Through the above system architecture, the intelligent agriculture system can identify the growth conditions of crops based on the soil data collected by the soil data sensor and the image data and video data collected by the camera, and perform corresponding maintenance tasks according to the growth conditions of the crops to ensure the healthy growth of the crops and achieve intelligent and efficient agricultural production.

[0036] In related technologies, the intelligent agriculture system often adopts Internet of Things technology. By deploying preset sensors in farmland, it can monitor key parameters such as soil temperature and humidity, pH value, and nutrient content in real time, and combine big data analysis and machine learning algorithms to optimize the decision-making for crop maintenance. However, the metabolic activities of microorganisms in the soil and the drilling and movement of organisms such as earthworms will gradually shift the spatial positions of the preset sensors in the soil. This shift will not only cause the data collected by the sensors to deviate from the target monitoring area, resulting in the distortion of data such as soil temperature and humidity and nutrient concentration, but also interfere with the accurate judgment of the machine learning model on the changing trend of the soil environment. The generated maintenance decisions, such as irrigation and fertilization plans, will be out of touch with the actual needs, which may cause problems such as water resource waste and soil compaction, and reduce the reliability of the intelligent agriculture system.

[0037] However, by adopting the crop intelligent maintenance method in the embodiments of the present application, by obtaining the real-time position and real-time attitude of the follow-up robot, determining the set of positions that the soil data sensor can reach, and combining crop root growth prediction, movement route and soil density distribution to select the collection position, and selecting the optimal collection position that the sensor can reach to collect soil data, the collected data is more in line with the data collected at the preset position, reducing the influence of the position deviation of the follow-up robot (i.e., the sensor position deviation) on the accuracy of data collection, and improving the accuracy of soil data collection.

[0038] The following will be combined with Figure 2 to illustrate the method of the embodiments of the present application.

[0039] Please refer to Figure 2 , which is a schematic flow chart of the crop intelligent maintenance method in the embodiments of the present application.

[0040] S201. Obtain the real-time robot attitude of the follow-up robot and the real-time sensor position of the soil data sensor in the target planting area.

[0041] Among them, the follow-up robot cannot autonomously adjust its own position and attitude.

[0042] Specifically, the server first controls a plurality of preset ultra-wideband (UWB) positioning base stations to send ultra-wideband wireless signals to the UWB tag module. The UWB tag module is installed on the follow-up robot and is adjacent to the soil data sensor. After receiving the radio wave signals emitted by each UWB base station, the UWB tag module determines the distance information from each base station by measuring the round-trip propagation time of the radio wave. Subsequently, the UWB tag module will upload the measured distance information to the server in real time through the wireless communication module. After that, based on the received distance information between the UWB tag module and multiple base stations and combined with the known coordinates of the base stations, the server uses a three-dimensional space positioning algorithm (such as a three-dimensional multilateration algorithm or the least squares method) to calculate the horizontal coordinates and vertical coordinates of the UWB tag module in the target planting area (i.e., the position of the robot on the two-dimensional plane and its depth in the soil), and obtains the current position of the UWB tag module.

[0043] Among them, the ultra-wideband (UWB) positioning base stations are pre-installed in the soil of the crop planting area. The positions of these UWB base stations have been accurately measured and recorded in the server database in advance, forming a known and fixed spatial coordinate system. The layout of the positions of the UWB base stations covers the entire target planting area and can achieve cross-coverage of the spatial positioning signals to ensure accurate measurement of the position of the follow-up robot.

[0044] For the real-time robot attitude, obtain the real-time attitude angles (such as pitch angle, roll angle, yaw angle) of the follow-up robot measured by the built-in inertial measurement unit (IMU) of the follow-up robot. The inertial measurement unit includes sensors such as a three-axis accelerometer, a three-axis gyroscope, and a magnetometer. The accelerometer is used to measure the acceleration of the follow-up robot in three axial directions (X, Y, Z axes). By integrating the acceleration, speed and displacement information can be obtained. The gyroscope is used to measure the rotational angular velocity of the follow-up robot around three axes. By integrating the angular velocity, the angle change can be obtained, thereby determining the attitude change of the robot. The magnetometer is used to measure the direction of the earth's magnetic field, providing an absolute azimuth reference for the robot to help calibrate the attitude data.

[0045] In some embodiments, the real-time robot attitude can also be determined based on the preset position of the UWB tag module when the follow-up robot is vertically placed and the current position of the UWB tag module. Based on the preset position and the current position, calculate the difference between the coordinate values of the current position in the X, Y, and Z-axis directions and the corresponding coordinate values of the preset position to obtain the offset values of the UWB tag module in each direction. According to the calculated offset values in the X, Y, and Z-axis directions, combined with a certain attitude calculation model or algorithm, determine the real-time attitude of the follow-up robot. For example, by calculating the proportional relationship, angular relationship, etc. between the offset values, infer the attitude information such as the tilt angle and rotation angle of the robot in space. The specific calculation model or algorithm can be selected and optimized according to the actual robot design and application scenario. For example, mathematical methods such as vector operations and trigonometric functions are used to determine the attitude angle, such as calculating parameters such as pitch angle, roll angle, and yaw angle.

[0046] For the real-time sensor position, first obtain the preset relative position vector of the UWB tag module relative to the soil data sensor (i.e., the fixed offset distance on the X, Y, and Z axes), then calculate the rotation matrix according to the pitch angle, roll angle, and yaw angle of the robot, rotate the relative position vector, and obtain the relative position vector considering the attitude in the global coordinate system. Finally, add the relative position vector to the current position coordinates of the UWB tag module to obtain the real-time sensor position of the soil data sensor.

[0047] S202. Determine the set of positions that the soil data sensor can reach based on the real-time robot attitude, the real-time sensor position, and the telescopic length range of the telescopic device.

[0048] Specifically, first establish a local coordinate system based on the real-time sensor position and the real-time robot attitude, with the center point of the follow-up robot as the reference. In this local coordinate system, use the real-time sensor position as the origin and determine the direction of the coordinate axes according to the real-time attitude of the robot. Among them, the Z-axis represents the telescopic direction of the telescopic device.

[0049] Then, determine the range of coordinate positions reachable by the soil data sensor in the local coordinate system. Since the telescopic device can only move up and down, in the local coordinate system, the movement of the telescopic device mainly affects the coordinates in the Z-axis direction. According to the maximum extension length and the minimum retraction length in the telescopic length range of the telescopic device, determine the reachable range of the soil data sensor in the Z-axis direction (that is, from the minimum retraction length minus the current telescopic length of the telescopic rod to the maximum retraction length minus the current telescopic length of the telescopic rod). For the horizontal direction (X-axis and Y-axis), since the slave robot itself cannot autonomously adjust its position, the reachable position of the soil data sensor in the horizontal direction is fixed and unchanged, which is the same as the horizontal coordinate position of the origin in the horizontal direction. Integrate the reachable range in the Z-axis direction and the horizontal coordinate position in the horizontal direction to obtain the range of coordinate positions reachable by the soil data sensor in the local coordinate system.

[0050] Next, based on the attitude of the slave robot, determine the set of positions reachable by the soil data sensor in the global coordinate system. Convert each coordinate position in the reachable coordinate position range in the local coordinate system to the global coordinate system through coordinate transformation, and integrate the multiple coordinate positions in the global coordinate system after conversion to obtain the set of positions reachable by the soil data sensor. The specific coordinate transformation process requires rotation and translation operations according to the real-time attitude of the robot. For example, if the robot has a certain pitch angle and roll angle, then corresponding rotation transformations need to be performed on the coordinates in the local coordinate system during the conversion process; at the same time, translate the coordinates after conversion according to the current position of the robot to align them with the global coordinate system. Among them, the global coordinate system is a unified space coordinate system established within the entire farmland, which is the same as the coordinate system of the current position of the robot obtained in step S201. The farmland is divided into multiple planting areas.

[0051] S203. Based on the predicted root growth conditions of the crops corresponding to each position in the position set, determine the predicted distance between each position and the corresponding crop roots.

[0052] Specifically, first obtain the pre-constructed crop root growth prediction model. This model is constructed based on machine learning algorithms such as neural networks and decision trees, and is trained using a large amount of crop growth-related data, so that the model can learn the root growth laws of different types of crops under different growth environment conditions. During the training process, the model will continuously adjust its own parameters according to a large amount of historical data, such as the weights and biases in the neural network, and the division rules in the decision tree, etc., to minimize the error between the prediction result and the actual root growth situation. Among them, the crop growth-related data covers crop variety information, planting time, soil characteristics, meteorological data, and past root growth monitoring data.

[0053] Next, input the relevant information of each crop in the target planting area into the crop root growth prediction model. This information includes crop variety, planting time, current soil characteristics, meteorological data, and historical root growth data, etc. Based on these inputs, the model uses the learned rules to predict the current and future root growth of the crop, including root topological structure data, etc. The root topological structure data includes the type of each root of the crop (such as main root, primary lateral root, secondary lateral root, etc.), the growth path in three-dimensional space (a series of connected node coordinates, and the coordinate system of the coordinates is the global coordinate system), and the connection relationship with other root segments (parent-child relationship), etc.

[0054] Subsequently, based on the three-dimensional space coordinate system, the target planting area is gridded, and the area is divided into multiple small spatial unit grids according to the horizontal position division rule. For each grid, determine the root distribution within the grid. Traverse the growth path of each root in the root topological structure data of the crops within the preset range of the grid, compare the three-dimensional space node coordinates in the growth path with the coordinate range of the grid in the global coordinate system, and filter out the root segments that are completely or partially located within the grid. For the filtered root segments, the server classifies and counts them according to their type fields (main root, primary lateral root, etc.). If it is of the "main root" type, increment the main root quantity counter within the grid by 1; if it is a lateral root type, accumulate the lateral root counters of the corresponding levels according to its hierarchical relationship (such as traced through the parent-child connection relationship field). At the same time, extract all the node coordinates of the root segment located within the grid, sort them in ascending order according to the Z-axis coordinate value, and determine the minimum value as the starting depth and the maximum value as the ending depth. During the processing, the server will perform special processing on overlapping or intersecting root segments. When multiple root segments share some nodes, they are distinguished by the unique identifier and the connection relationship field to avoid double counting. If a root segment spans multiple grids, the server only counts the node part located within the current grid and generates continuous depth data according to the node connection order. After processing all root segments, the server stores the statistically obtained main root / lateral root quantity, the starting depth, ending depth, and growth path within the grid of each root segment in the root distribution data of the corresponding grid in the form of structured data.

[0055] After that, for each position in the position set, by comparing the three-dimensional coordinates of the position with the coordinate range of each grid in the global coordinate system, determine the target grid where the position is located. Then, extract the ending depth of each root segment within the grid from the root distribution data table of the target grid. Filter out the deepest depth among all the ending depths. Subtract the coordinate value of the position in the vertical direction (Z-axis direction) from the deepest depth to obtain the predicted distance of the position.

[0056] S204. Based on the predicted distance, the movement routes between the soil data sensor and various positions, and the soil density distribution in the target planting area, determine the collection positions corresponding to the movement routes that do not affect crop growth.

[0057] Specifically, first, based on the predicted root growth of each crop in the target planting area, the historical movement trajectories of the soil data sensor and the follow-up robot, and the range of soil biological activities, divide the target planting area into a crop root area, a robot movement area, and a biological activity area.

[0058] For the division of the crop root area, obtain all the termination depths in the root distribution data of each grid in the target planting area. For each grid, screen out the deepest depth among the termination depths. Then, based on the global coordinate system, identify the three-dimensional space area above the deepest depth in each grid, and then superimpose and merge these areas to obtain the crop root area.

[0059] For the division of the robot movement area, first obtain the first historical movement trajectory data of the follow-up robot (the trajectory points determined by the center point of the robot) and the second historical movement trajectory data of the soil data sensor (the trajectory points determined by the center point of the sensor) within a preset time period. Based on the mechanical parameters of the robot and the soil data sensor and the preset influence radius, generate cylindrical or hemispherical influence range buffers centered on each trajectory point in the trajectory data. Use the spatial overlay analysis function of the Geographic Information System (GIS) to merge and overlap all the buffers to obtain the first robot movement area. Exclude the part of the first robot movement area that belongs to the crop root area to obtain the robot movement area.

[0060] For the division of the biological activity area, first extract the habit data of various soil organisms such as earthworms and microorganisms from the soil biological information database. Then, combine the environmental data such as humidity, temperature, and pH value and meteorological information collected in real time by the soil sensor, and use spatial interpolation algorithms such as Kriging to generate a continuous spatial distribution layer of soil environmental parameters. Match the soil biological habit data with the layer to obtain their respective potential activity areas. For the overlapping parts in the activity areas, adjust the boundaries of the activity areas according to the preset biological symbiotic or competitive relationships. Then, use the GIS spatial analysis function to integrate the processed areas, remove duplicates, and merge adjacent areas to determine the first biological activity area within the entire target planting area. Exclude the parts of the first biological activity area that belong to the crop root area and the robot movement area to obtain the biological activity area.

[0061] Secondly, according to the predicted root growth situation, determine the first soil density distribution in the crop root area. Obtain the root distribution data of each grid in the target planting area. For the root distribution data within each grid, extract the type of each root within the grid and the growth nodes located in each sub-grid within the three-dimensional space growth path of the root. The sub-grid is obtained by dividing the grid according to a preset depth. For each growth node in all sub-grids, classify and count according to the type of root. For the main root, each main root is counted as one unit; for the lateral roots, according to their hierarchical relationship, the lateral roots of the same level are grouped into one category for counting. Then, according to the volume of the sub-grid (determined by the length, width of the grid, and the above-mentioned preset depth), calculate the number of different types of roots per unit volume, that is, the root number density. Screen out the soil density corresponding to the root number density of different types of roots in each sub-grid in the preset soil density correspondence table to obtain the soil density of each sub-grid. Integrate the soil densities of each sub-grid to obtain the soil density at different positions within the crop root area (i.e., the first soil density distribution).

[0062] Thirdly, according to the historical movement trajectory, determine the number of movements and the movement position offset of the follow-up robot within a preset time period. Obtain the first historical movement trajectory of the follow-up robot within the preset time period. Count the number of trajectory points in the first historical movement trajectory to obtain the number of movements of the follow-up robot. For each trajectory point, calculate the offset values in the vertical and horizontal directions between this trajectory point and the previous trajectory point before this trajectory point to obtain the offset situation of the follow-up robot during each movement.

[0063] Fourthly, based on the number of movements and the movement position offset, predict the biological activity situation in the biological activity area. First, analyze the data of the number of movements and the movement position offset of the follow-up robot. Divide the number of movements into multiple time windows according to the time series, and count the average value of the number of movements and the standard deviation of the offset values within each time window to construct the robot movement activity index and the movement stability index. Then, obtain the preset biological activity corresponding to the movement activity index and the movement stability index to get the biological activity situation in the biological activity area.

[0064] Fifthly, based on the biological activity situation, the historical movement trajectory, and the influence of the movement of the soil data sensor on the preset density of the soil within the preset area range, determine the second soil density distribution in the robot movement area and the third soil density distribution in the biological activity area.

[0065] For the robot moving area, in combination with the movement of the soil data sensor and the preset density influence coefficient of the robot's movement on the soil (such as the empirical value of the soil density change caused by the single movement of the sensor), calculate the soil density change amount generated by the sensor movement or the robot movement at each position in this area. For each trajectory point passed by the soil data sensor, obtain the first telescopic length of the telescopic device corresponding to this trajectory point and the second telescopic length of the previous trajectory point of this trajectory point. If the first telescopic length is greater than the second telescopic length, obtain the first preset density influence coefficient table. Add the first preset density influence coefficient corresponding to the distance between this trajectory point and each position point in the first preset density influence coefficient table to the soil density stored at each position point within the preset range of this trajectory point. If the first telescopic length is less than the second telescopic length, obtain the second preset density influence coefficient table. Subtract the second preset density influence coefficient corresponding to the distance between this trajectory point and each position point in the second preset density influence coefficient table from the soil density stored at each position point within the preset range of this trajectory point. For each trajectory point passed by the follow-up robot, obtain the corresponding offset situation. Based on the offset values in the vertical and horizontal directions in the offset situation, obtain the first preset range and the third preset density influence coefficient table, the second preset range and the fourth preset density influence coefficient table corresponding to the offset values. Add the third preset density influence coefficient corresponding to the distance between this trajectory point and each position point in the third preset density influence coefficient table to the soil density stored at each position point within the first preset range of this trajectory point, and subtract the fourth preset density influence coefficient corresponding to the distance between this trajectory point and each position point in the fourth preset density influence coefficient table from the soil density stored at each position point within the first preset range of this trajectory point.

[0066] At the same time, consider the loosening or compaction effect of biological activities on the soil (such as earthworm activities can reduce the soil density), and correct the soil density at the positions where biological activities exist according to the biological activity level. The positions where biological activities exist are determined based on the biological activity range. Finally, integrate the soil density at each position to obtain the second soil density distribution of the robot moving area.

[0067] For the biological activity area, calculate the change amount of soil density caused by biological activities in each sub-area according to the preset density influence coefficient of biological activities on the soil density in different sub-areas (divided according to the activity areas of different organisms). Finally, integrate the density change amounts of each sub-area and the corresponding stored soil density data to generate the third soil density distribution of the biological activity area.

[0068] Sixth, integrate the first soil density distribution of the crop root area, the second soil density distribution of the robot moving area, and the third soil density distribution of the biological activity area to obtain the soil density distribution of the target planting area.

[0069] Seventh, based on the movement route between the soil data sensor and each position, the soil density distribution, and the influence of the soil data sensor on the preset density of the soil within the preset area during movement, determine the target soil density distribution when the soil data sensor reaches each position. Obtain the predicted distance corresponding to each position. Determine whether there is a target predicted distance less than zero among all the predicted distances. If so, delete the positions passed by the soil data sensor in the position set after reaching the target position. Based on the current position of the soil data sensor and the movement rules, generate the movement route corresponding to each position. The movement route includes multiple trajectory points. For each position, first generate the target movement trajectory data according to the movement route and the second movement trajectory data. Then, based on the target movement trajectory data, re-divide the area to obtain a new robot movement area and a new biological activity area. Redetermine the soil density distribution of the new robot movement area and the new biological activity area according to the above determination method to obtain the new second soil density distribution and the new third soil density distribution. Integrate the first soil density distribution, the new second soil density distribution, and the new third soil density distribution to obtain the target soil density distribution when the soil data sensor reaches this position.

[0070] Eighth, based on the soil density distribution and the target soil density distribution, determine the target influence degree of soil density change on crop growth. Match the positions of the soil density distribution of the target planting area and the target soil density distribution, and calculate the soil density change value at each position point. Then, for the positions where the density change amount is not zero, obtain the corresponding collapse soil density change threshold from the table according to the variety and growth stage of the planted crops corresponding to the position above. Compare the density change amount of each grid with the collapse soil density change threshold, and determine the number of positions where the density change amount exceeds the collapse soil density change threshold. Divide this number by the total number of position points in the target planting area to obtain the target influence degree value.

[0071] Ninth, obtain the set of target positions where the target influence degree is within the preset influence degree range. Screen out the set of target influence degree values within the preset influence degree range among all the target influence degree values. Obtain the positions corresponding to each target influence degree in the target influence degree set to obtain the set of target positions.

[0072] Finally, based on the predicted distances of each target position in the target position set and the corresponding preset ideal distances, the target position corresponding to the minimum distance difference is taken as the collection position. The minimum distance difference is the smallest among the distance differences of all target positions, and the distance difference is the difference between the predicted distance and the corresponding ideal distance. For each target position, obtain the root system distribution data of the grid where the target position is located and the total root density of different types of roots in the grid. The total root density is obtained by adding the root number densities corresponding to each sub-grid in the grid. Obtain the root type corresponding to the deepest depth in the grid. Look up the ideal distance corresponding to the root type and the total root density of different types of roots in the preset distance table, and calculate the difference between the predicted distance and the ideal distance to obtain the distance difference. Take the target position corresponding to the minimum distance difference as the collection position.

[0073] S205. Send the collection position to the follow-up robot.

[0074] Specifically, obtain the telescopic length of the telescopic device corresponding to the collection position. Based on the telescopic length, generate a control instruction. Transmit the control instruction to the follow-up robot through the communication gateway. After receiving the control instruction, the follow-up robot adjusts the actual telescopic length of the telescopic device to the telescopic length in the instruction.

[0075] S206. After the follow-up robot moves the soil data sensor to the collection position, obtain the soil data collected by the soil data sensor at the collection position.

[0076] After receiving the feedback information that the operation execution is completed transmitted by the follow-up robot, obtain the soil data collected by the soil data sensor at the collection position.

[0077] Among them, the soil data includes parameters such as temperature, humidity, pH value, and conductivity.

[0078] S207. Based on the soil data, crop leaf images, and growth trend videos, evaluate the growth conditions of the crops in the target planting area through a preset evaluation model.

[0079] Among them, the growth conditions include the current health status, future growth potential, and potential growth risks.

[0080] Specifically, for crop leaf images, methods such as edge detection and morphological operations are used for leaf segmentation to accurately extract the leaf area. Then, features are extracted from the leaf images, such as leaf color features (by calculating statistical quantities such as the mean and standard deviation of different color channels), shape features (the aspect ratio, perimeter, area, etc. of the leaf), and texture features (by calculating texture parameters through methods such as gray-level co-occurrence matrices).

[0081] For the growth trend video, the video is decomposed into consecutive image frames, and then object detection algorithms, such as the YOLO algorithm, are used to identify the crop plants in the video and track the growth dynamics of the plants. By analyzing the image frames at different time points, growth indicators such as the change in plant height, the change in the number of leaves, and the change in the number of tillers are calculated.

[0082] The soil data, leaf image features, and the analysis results of the growth trend video are input into a preset evaluation model. This evaluation model is based on machine learning or deep learning models, such as neural network models. In the model training stage, a large amount of data including soil data, crop leaf images (annotated with healthy leaves, leaves with different pest and disease symptoms, nutrient deficiency symptom leaves, etc.), growth trend videos (covering the growth process records of crops at different growth stages and different growth conditions), and crop actual growth status labels (grading of the current health status, such as healthy, mildly abnormal, severely abnormal; actual yield data and growth rate data in a future period; types and degrees of actual pest and diseases occurred, growth risk events caused by environmental factors, etc.) are used to train the model, enabling the model to learn the complex mapping relationship between soil data, leaf image features, growth trend information and the current health status, future growth potential, and potential growth risks of the crops.

[0083] For the evaluation of the current health status, the model will comprehensively consider the soil nutrient status, leaf health status, and crop growth dynamics according to the input data to evaluate the current health status of the crops. For example, if the soil nutrient level is low, there are pest and disease symptoms on the leaves, and the growth indicators increase slowly, it is judged that the current health status of the crops is poor.

[0084] For the evaluation of future growth potential, the model will make predictions by combining historical data and current growth conditions. According to the reserve situation of soil nutrients, the growth trend of the crops, and the predicted information of environmental factors, the growth rate, yield potential, etc. of the crops in a future period are predicted. For example, if the phosphorus content in the soil is high and the current growth trend of the crops is good, the model predicts that the root development and fruit development of the crops in the future may be better and the growth potential is greater.

[0085] For the evaluation of potential growth risks, the model will consider various factors. On the one hand, according to the pest and disease symptoms identified from the leaf images, combined with the spread rules of pests and diseases and environmental conditions, the spread risk of pests and diseases is predicted. On the other hand, the abnormal situations in the soil data, such as abnormal soil pH value and excessive heavy metal content, are analyzed to evaluate the potential harm of these factors to crop growth. At the same time, external factors such as meteorological data are considered. For example, if it is predicted that extreme weather such as heavy rain or drought is about to occur, its impact on crop growth is evaluated.

[0086] Finally, the model outputs the growth situation, and the evaluation results include the current health status, future growth potential, and potential growth risks of the crops.

[0087] S208. Control a preset intelligent maintenance device to perform corresponding maintenance tasks based on the growth situation.

[0088] Specifically, if the growth situation shows that the crops are abnormal (such as the health status being "slightly abnormal" or "severely abnormal") or the potential growth risks are relatively high (such as the significant risk of pest and disease spread, insufficient soil nutrients, etc.), it is determined that the crops in the target planting area need maintenance. Based on the abnormal parameters corresponding to the current health status and the risk problems corresponding to the potential growth risks, the type of maintenance to be performed (such as irrigation, fertilization, spraying pesticides, etc.) is determined. If the type of maintenance is irrigation, first retrieve the soil moisture data of the target planting area, and combine it with the water demand characteristics of the crops at the current growth stage to calculate the difference between the actual soil water content and the ideal water content, and determine the amount of water to be supplemented. At the same time, refer to the weather forecast data to adjust the irrigation amount. If the type of maintenance is fertilization, determine the type, ratio, and dosage of the required fertilizers based on the actual content of nutrients such as nitrogen, phosphorus, and potassium in the soil test data and the nutrient requirement standards of the crops at the current growth stage. If the type of maintenance is spraying pesticides, first determine the types and infection degrees of pests and diseases through image recognition and data analysis. Then, screen out the effective pesticides against the pests and diseases from the pesticide database, and calculate the optimal dilution ratio and spraying dosage according to the characteristics and usage instructions of the pesticides. At the same time, refer to the weather conditions and select an appropriate pesticide spraying time. Based on the type of maintenance and the data of each maintenance parameter corresponding to the type of maintenance, generate corresponding maintenance tasks.

[0089] Then, according to the current positions and working states of each maintenance device, use a task allocation algorithm to match the most suitable intelligent maintenance device for each task. After the allocation is completed, the server sends a task instruction to the selected intelligent maintenance device, and the instruction includes detailed information such as operation path planning (generating the shortest and most efficient driving route by using a path optimization algorithm and combining information such as farmland terrain and obstacle distribution), execution parameters, and task deadline.

[0090] In the embodiments of the present application, by dynamically adjusting the position of the soil data sensor with a follow-up robot, the sensor can be dynamically adjusted to move in a follow-up manner, so as to better collect sensing data as the plant roots change seasonally. By obtaining the real-time position and real-time attitude of the follow-up robot, the set of positions that the soil data sensor can reach is determined, and the collection position is selected by combining the prediction of crop root growth, the movement route, and the soil density distribution. The optimal collection position that the sensor can reach is selected to collect soil data, making the collected data more in line with the data collected at the preset position, reducing the impact of the position deviation of the follow-up robot (i.e., the position deviation of the sensor) on the accuracy of data collection, and improving the accuracy of soil data collection. At the same time, by comprehensively evaluating the crop growth situation based on soil data, crop leaf images, and growth trend videos, a maintenance task is generated, and an intelligent maintenance device is controlled to execute the task, realizing the automatic maintenance of crops and improving the efficiency and quality of agricultural production.

[0091] Next, in combination with Figure 3 to further illustrate the method of the embodiments of the present application.

[0092] Please refer to Figure 3 , which is another schematic flowchart of the crop intelligent maintenance method in the embodiments of the present application.

[0093] S301. Obtain the real-time robot attitude and real-time sensor position in the target planting area.

[0094] S302. Determine the set of positions that the soil data sensor can reach.

[0095] S303. Based on the predicted root growth of the crops corresponding to each position in the position set, determine the predicted distance between each position and the corresponding crop roots.

[0096] S304. Obtain the soil density distribution in the target planting area S305. Determine the target soil density distribution when the soil data sensor reaches each position.

[0097] S306. Based on the soil density distribution and the target soil density distribution, determine the target influence degree of soil density change on crop growth.

[0098] S307. Obtain the set of target positions within the preset influence degree range of the target influence degree.

[0099] S308. Based on the predicted distance of each target position in the set of target positions and the corresponding preset ideal distance, use the target position corresponding to the minimum distance difference as the collection position.

[0100] S309. Send the collection position to the follow-up robot.

[0101] S310. After the servo robot moves the soil data sensor to the collection position, obtain the soil data collected by the soil data sensor at the collection position.

[0102] Steps S301 - S310 are similar to Figure 2 Steps S201 - S206 in the illustrated embodiment. Refer to the descriptions in Steps S201 - S206, and details are not elaborated here.

[0103] S311. According to the growth conditions and crop types of each crop in the target planting area, determine the crop growth difference areas and the boundary areas between different crop types.

[0104] Specifically, first divide the target planting area into multiple grids according to certain rules. For each grid, retrieve the growth condition data and crop type information of the crops within the grid. By comparing the growth condition data of the crops between adjacent grids, calculate the difference degree of each growth index. If the difference degree exceeds the preset threshold, mark these grids as crop growth difference areas.

[0105] For the identification of the boundary areas between different crop types, traverse all grids. Based on the crop type information within the grids, determine whether there are multiple crop types in the grid. If so, mark the grid as a boundary area.

[0106] S312. Based on the predicted root growth conditions of each crop in the target planting area, determine the root - dense areas.

[0107] Specifically, obtain the root quantity density of different types of roots in each sub - grid calculated in Step S204. Add up the root quantity densities of different types of roots in the sub - grids within each grid to obtain the total root density of different types of roots in the grid. Then, traverse all grids and mark the grids with a total root density greater than the threshold as candidate root - dense areas.

[0108] S313. Screen out the target horizontal coordinate positions whose horizontal coordinate positions are within the growth difference areas, boundary areas, and root - dense areas.

[0109] Specifically, traverse the horizontal position coordinates of each grid in the growth difference area, and check whether there is a first horizontal position coordinate of a first grid in the boundary area within a preset range of this horizontal position coordinate. If so, continue to check whether there is a second horizontal position coordinate of a second grid in the root - dense area within a preset range of this horizontal position coordinate. If there is also a second horizontal position coordinate of a second grid in the root - dense area within a preset range of this horizontal position coordinate, take this horizontal position coordinate as the target horizontal coordinate position. After the traversal is completed, obtain the set of target horizontal coordinate positions.

[0110] S314, calculating the target vertical coordinate position based on the ideal distance between the crop root system and the sensor and the crop root system depth corresponding to the target horizontal coordinate position.

[0111] Specifically, for each target horizontal coordinate position, obtain the root distribution data of the grid where the target horizontal coordinate position is located and the total root density of different types of roots in the grid. Obtain the root type corresponding to the deepest depth in the grid. Find the ideal distance corresponding to the root type and the total root density of different types of roots in the preset distance table. Add the deepest depth to the ideal distance to obtain the target vertical coordinate position corresponding to the target horizontal coordinate position.

[0112] S315: Obtain a first ideal acquisition position based on the target horizontal coordinate position and the target vertical coordinate position.

[0113] The target horizontal coordinate position and the corresponding target vertical coordinate position are integrated into three-dimensional coordinates to obtain a first ideal acquisition position.

[0114] S316. When a soil data sensor fails, a second position set is selected from the first position set reachable by the first soil data sensor and whose distance from the first ideal collection position is within a preset distance range.

[0115] Among them, the first soil data sensor can reach the target planting area.

[0116] Specifically, we first construct data features for each soil parameter in the soil data, such as soil moisture, temperature, pH, nutrient content, etc. For each parameter, we calculate the mean deviation and standard deviation of the current data and the historical data of the same period. If the deviation exceeds the preset deviation value, we use statistical methods and machine learning models to detect anomalies. We use unsupervised learning algorithms such as Isolation Forest and One-Class SVM to train the model on historical soil data, learn the normal data distribution, input the current soil data into the model, calculate the anomaly score, and if the score exceeds the set threshold, we determine that there is a fault. At the same time, considering the correlation between soil parameters, we use dimensionality reduction methods such as principal component analysis (PCA) to explore the potential relationship between data. If the anomalies of multiple parameters violate the laws of soil physical and chemical changes at the same time (such as a sudden increase in soil temperature but no corresponding decrease in humidity), we confirm that there is a fault.

[0117] If the soil data sensor fails, first obtain the set of soil data sensors with normal operating status within the preset range of the soil data sensor. Then, based on the sensor positions of the soil data sensors in the set of soil data sensors, the robot attitude, and the root distribution of the corresponding first planting area, determine the first set of positions that each soil data sensor can reach and will not damage the crop roots (the distance from the crop roots is greater than the preset value). Next, screen out the first soil data sensor in the set of soil data sensors. There is a first position in the first set of positions of the first soil data sensor within the target planting area. After that, obtain the first set of positions of the first soil data sensor and calculate the distance between each first position and the first ideal acquisition position. Finally, screen out the second set of positions in the first set of positions whose distance from the first ideal acquisition position is within the preset distance range.

[0118] If the soil data sensor does not fail, determine whether the first ideal acquisition position and the acquisition position are within the preset deviation range. If so, execute the steps of S322; if not, execute the steps of S320.

[0119] S317. Based on the first movement route of the first soil data sensor, the soil density distribution, and the fourth soil density distribution of the first planting area, determine the degree of influence of the soil density change corresponding to each first movement route on crop growth.

[0120] Specifically, first, obtain the third set of positions within the first planting area of the first soil data sensor in the first set of positions. For each first position, determine whether the first position is within the first planting area of the corresponding first soil data sensor. If so, add the position to the third set of positions.

[0121] Secondly, based on the second ideal acquisition position of the first planting area, calculate the data confidence of each third position. Obtain the second ideal acquisition position of the first planting area. Calculate the horizontal deviation value and the vertical deviation value between the horizontal position and the vertical position of each third position and the second ideal acquisition position. Obtain the preset first score value corresponding to the horizontal deviation value and the preset second score value corresponding to the vertical deviation value. Based on the preset horizontal weight and vertical weight, the first score value and the second score value, calculate the data confidence of each third position through weighted calculation.

[0122] Thirdly, screen out the fourth set of positions in the third set of positions whose data confidence is higher than the preset confidence threshold. For each third position, compare the data confidence of the third position with the preset confidence threshold. If the data confidence is higher than the preset confidence threshold, add the third position to the fourth set of positions.

[0123] Fourth, plan the first moving route of the first soil data sensor from the current position to the second position and then from the second position to the fourth position. For each second position, based on the current position of the first soil data sensor and the preset moving rules, generate the second moving route of the sensor from the current position to the second position. Then, based on the current position of the first soil data sensor and the preset moving rules, generate the set of third moving routes of the sensor from the second position to each fourth position. Merge the second moving route with each third moving route in the set of third moving routes to obtain the set of first moving routes.

[0124] Fifth, divide the first moving route into a fourth moving route and a fifth moving route. Among them, each trajectory point in the fourth moving route is within the target planting area, and each trajectory point in the fifth moving route is within the first planting area.

[0125] Sixth, based on each fourth moving route, the soil density distribution, and the influence of the soil data sensor on the preset density of the soil within the preset area during movement, determine the first target soil density distribution when the first soil data sensor reaches each fourth position. At the same time, based on each fifth moving route, the fourth soil density distribution of the first planting area, and the influence of the soil data sensor on the preset density of the soil within the preset area during movement, determine the second target soil density distribution when the first soil data sensor reaches each fourth position.

[0126] Finally, based on the soil density distribution, the first target soil density distribution, and the second target soil density distribution, determine the target influence degree of the soil density change on crop growth.

[0127] The implementation steps for determining the influence degree in step S317 are similar to Figure 2 the implementation steps for determining the target influence degree in step S204 in the illustrated embodiment, and reference can be made to the description in step S204, which will not be elaborated here.

[0128] S318. When there is an influence degree within the preset influence degree range, based on the new acquisition position corresponding to the second moving route with the minimum influence degree, control the first soil data sensor to reach the new acquisition position to collect new soil data.

[0129] Specifically, when there is an influence degree within the preset influence degree range, use the second position in the second moving route corresponding to the minimum influence degree as the new acquisition position. Send the new acquisition position to the target follow-up robot corresponding to the second moving route. After the target follow-up robot moves the first soil data sensor to the new acquisition position, obtain the new soil data collected by the first soil data sensor at the new acquisition position.

[0130] S319. Update the collection location and soil data based on the newly collected location and new soil data.

[0131] Replace the data content in the collection location and soil data with the data content in the newly collected location and new soil data.

[0132] In some embodiments, when comparing and analyzing the soil data with historical soil data, if it is detected that only the parameter values of some parameters are abnormal, only replace the parameter values of the parameters with abnormal collection in the soil data, and replace the content of the collection location corresponding to the parameter with abnormal collection with the newly collected location.

[0133] S320. If the first ideal collection location and the collection location are not within the preset deviation range, determine the correction values of each soil parameter in the soil data.

[0134] If the first ideal collection location and the collection location are not within the preset deviation range, determine the correction values of each soil parameter in the soil data according to the deviation situation between the first ideal location and the collection location and the soil density distribution.

[0135] Specifically, first calculate parameters such as the straight-line distance between the first ideal collection location and the collection location, the offset values of each direction coordinate, etc., and determine whether they are within the preset deviation range.

[0136] If not, construct a multi-dimensional analysis model to conduct in-depth coupling analysis of the deviation situation and soil density distribution. First, based on Geographic Information System (GIS) data, project the first ideal collection location and the collection location onto a three-dimensional terrain model, and combine information such as terrain slope and altitude to refine the spatial characteristics of the deviation.

[0137] Next, retrieve the soil density distribution situation, and extract the soil density data within a certain range around the collection location and its vicinity. These data are stored in a grid form, and each grid corresponds to a specific soil density value and related attributes. The server will use an interpolation algorithm to convert the discrete soil density data into a continuous density field, visualize the soil density distribution, and locate the soil density gradient change situation of the collection location.

[0138] Subsequently, for each soil parameter, construct a corresponding deviation-density correction function based on historical experimental data and expert experience. Taking the soil moisture parameter as an example, the correction function takes the straight-line distance, the offset values of each direction coordinate, and the soil density gradient change as input variables and outputs a correction coefficient. When constructing the function, the server uses machine learning algorithms to train a large number of labeled soil parameter measurement data to optimize the parameter weights in the function to ensure that the correction function can accurately reflect the impact of actual environmental changes on soil parameters.

[0139] When calculating the correction value, multiply the original measured value of each soil parameter by the corresponding correction coefficient to obtain a preliminary correction value. To improve the correction accuracy, the server also introduces a dynamic compensation mechanism to perform a secondary adjustment on the preliminary correction result in combination with real-time meteorological data (such as wind speed, rainfall, etc.). For example, if it rains during the collection, the server will perform a compensation calculation on the soil moisture correction value according to the amount of rainfall and the duration of rainfall to avoid correction errors caused by environmental factors.

[0140] S321. Update the parameter values of each soil parameter in the soil data based on the correction value.

[0141] Add the corresponding correction value to the parameter value of each soil parameter in the soil data.

[0142] S322. Obtain the crop leaf images and growth trend videos collected by the camera carried by the drone in the target planting area.

[0143] Specifically, send a control instruction to the drone. After receiving the instruction, the drone first plans the movement route from the current position to the target planting area according to the preset path planning algorithm, and then moves to above the target planting area along the movement route. When the drone reaches above the target planting area, send a control instruction to the camera carried by the drone. After receiving the instruction, the camera collects the crop leaf images and growth trend videos of the target planting area, and transmits the collected data to the server through the communication gateway.

[0144] S323. Input the crop leaf images, growth trend videos, and soil data into a preset evaluation model to obtain the growth conditions of the crops in the target planting area.

[0145] S324. Based on the growth conditions, control the preset intelligent maintenance equipment to perform corresponding maintenance tasks.

[0146] Steps S323 - S324 are similar to Figure 2 Steps S207 - S208 in the illustrated embodiment, and reference can be made to the description in Steps S207 - S208, which will not be elaborated here.

[0147] In the embodiments of the present application, when a soil data sensor fails, by screening other planting areas around the faulty sensor to set a target planting area that can reach the faulty sensor, and a first soil data sensor whose arrival position in the target planting area is within a preset range of the ideal acquisition position of the target planting area (i.e., the data collected at this arrival position can reflect the soil data within the target planting area), and evaluating the impact degree of the movement route of the first soil data sensor on the soil environment, so as to select the optimal alternative solution. Collect the soil data of the target planting area according to the optimal replacement plan, avoiding the damage to the crops caused by repairing or replacing the sensor, ensuring the accuracy and continuity of soil data collection, and improving the fault response ability and overall operation efficiency of the intelligent agriculture system. At the same time, by comparing the first ideal acquisition position with the actual acquisition position, when the deviation between the two exceeds the preset range, determine the soil parameter correction value in combination with the soil density distribution, and update the soil data parameters, making up for the soil data error caused by the deviation of the acquisition position, and making the soil data more accurately reflect the true soil conditions of the target planting area.

[0148] The crop intelligent maintenance method in the embodiments of the present application is described above. Next, the data transmission process in the crop intelligent maintenance method will be described in detail.

[0149] Please refer to Figure 4 , which is another structural schematic diagram of the applicable system framework of the crop intelligent maintenance method in the embodiments of the present application.

[0150] Figure 4 It includes a server, a Beidou ground station, Beidou satellites, a Beidou gateway, soil data sensors and a follow-up robot.

[0151] The data collected by the soil data sensors is first transmitted to the Beidou gateway. The Beidou gateway plays the role of data transfer, preliminarily processing and aggregating the information collected by the soil data sensors. After receiving the data, the Beidou gateway sends the data to the Beidou satellites. As an air communication hub, the Beidou satellites receive data from multiple Beidou gateways and transmit it to the Beidou ground station. The Beidou ground station receives the data from the satellite, further processes and converts it, and transmits the data to the server.

[0152] The above describes the transmission process of data to the server. The transmission process of the server transmitting data to the follow-up robot or the soil data sensor is opposite to the above transmission process.

[0153] Next, in combination with the above crop intelligent maintenance method, the server in the embodiments of the present application will be described in detail.

[0154] Please refer to Figure 5 , which is an exemplary hardware structural schematic diagram of the server in the embodiments of the present application.

[0155] In some embodiments, the server 500 includes a computer device, which may be a terminal device. The computer device includes a processor 501, a memory 502, a communication module 503, an input device 504, and an output device 505 connected by a system bus. Among them, the processor 501 of the computer device is used to provide computing and control capabilities. The memory 502 of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database is used to store data. The communication module 503 of the computer device is used to transmit data such as collected soil data, crop leaf images, and growth trend videos to the server and transmit control instructions to follow-up robots and drones, etc. The input device 504 of the computer device is used to receive collected soil data, crop leaf images, and growth trend videos, etc. The output device 505 of the computer device is used to display the growth situation of crops, etc. When the computer program is executed by the processor 501, it implements the crop intelligent maintenance method in the embodiments of the present application.

[0156] Those skilled in the art can understand that Figure 5 the structure shown in

[0157] is only a block diagram of some structures related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0158] In some embodiments of the present application, a computer-readable storage medium is provided, including instructions, which when running on the server 500, can cause the server 500 to execute the crop intelligent maintenance method in the embodiments of the present application.

[0159] As described above, the above embodiments are only used to illustrate the technical solutions of the present application, rather than to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present application.

[0160] As used in the foregoing embodiments, depending on the context, the term "when" may be construed to mean "if", "after", "in response to determining", or "in response to detecting". Similarly, depending on the context, the phrase "when determining" or "if (the stated condition or event) is detected" may be construed to mean "if determined", "in response to determining", "when (the stated condition or event) is detected", or "in response to detecting (the stated condition or event)".

[0161] In the foregoing embodiments, it may be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, it may be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions. When the computer program instructions are loaded and executed on a computer, the processes or functions described in the embodiments of the present application are generated in whole or in part. The computer may be a general-purpose computer, a special-purpose computer, a computer network, or other programmable devices. The computer instructions may be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions may be transmitted from one website, computer, server, or data center to another website, computer, server, or data center by wire (such as coaxial cable, fiber optic, digital subscriber line) or wirelessly (such as infrared, wireless, microwave, etc.). The computer-readable storage medium may be any available medium that the computer can access or a data storage device such as a server or data center that includes one or more integrated available media. The available media may be magnetic media (such as floppy disks, hard disks, magnetic tapes), optical media (such as DVDs), or semiconductor media (such as solid state drives), etc.

[0162] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the foregoing embodiments can be completed by hardware instructed by a computer program. The program can be stored in a computer-readable storage medium. When the program is executed, it may include the processes of the foregoing method embodiments. The foregoing storage medium includes various media that can store program codes, such as ROM or random access memory RAM, magnetic disks, or optical discs.

Claims

1. A method for intelligent maintenance of crops, characterized in that, A server applied to a smart agriculture system, the smart agriculture system further includes a follow-up robot, a soil data sensor, and a communication gateway. The follow-up robot is disposed in the soil below the crop roots, the soil data sensor is disposed above the follow-up robot, the soil data sensor is used to collect soil data below the crop roots, the follow-up robot is used to control the built-in telescopic device to move up and down to adjust the position of the soil data sensor, and the server is connected to the soil data sensor and the follow-up robot respectively through the communication gateway. The method includes: Obtain the real-time robot posture of the follow-up robot and the real-time sensor position of the soil data sensor in the target planting area, and the follow-up robot cannot autonomously adjust its own position and posture; Based on the real-time robot posture, the real-time sensor position, and the telescopic length range of the telescopic device, determine the set of positions that the soil data sensor can reach; Based on the predicted root growth of the crops corresponding to each position in the position set, determine the predicted distance between each position and the corresponding crop roots; Based on the predicted distance, the movement route between the soil data sensor and each position, and the soil density distribution of the target planting area, determine the collection positions corresponding to the movement routes that do not affect crop growth; Send the collection positions to the follow-up robot; After the follow-up robot moves the soil data sensor to the collection position, obtain the soil data collected by the soil data sensor at the collection position; Based on the soil data, crop leaf images, and growth trend videos, evaluate the growth of crops in the target planting area through a preset evaluation model; Based on the growth situation, control the preset intelligent maintenance equipment to perform corresponding maintenance tasks.

2. The method according to claim 1, wherein The determining the collection positions corresponding to the movement routes that do not affect crop growth based on the predicted distance, the movement route between the soil data sensor and each position, and the soil density distribution of the target planting area specifically includes: Obtain the soil density distribution of the target planting area, and the soil density distribution includes the first soil density distribution in the crop root area, the second soil density distribution in the robot movement area, and the third soil density distribution in the biological activity area; Based on the movement route between the soil data sensor and each position, the soil density distribution, and the preset density influence on the soil within the preset area range when the soil data sensor moves, determine the target soil density distribution when the soil data sensor reaches each position; Based on the soil density distribution and the target soil density distribution, determine the target influence degree of the soil density change on crop growth; Obtain the set of target positions where the target influence degree is within the preset influence degree range; Based on the predicted distances of the target positions in the set of target positions and the corresponding preset ideal distances, the target position corresponding to the minimum distance difference is used as the acquisition position, where the minimum distance difference is the smallest among the distance differences of all target positions, and the distance difference is the difference between the predicted distance and the corresponding ideal distance.

3. The method according to claim 2, wherein The obtaining of the soil density distribution of the target planting area specifically includes: Based on the predicted root growth of each crop in the target planting area, the historical movement trajectories of the follow-up robot and the soil sensor, and the soil biological activity range, the target planting area is divided into a crop root area, a robot movement area, and a biological activity area; According to the predicted root growth, determine the first soil density distribution of the crop root area; According to the historical movement trajectory, determine the number of movements and the movement position offset of the follow-up robot within a preset time period; Based on the number of movements and the movement position offset, predict the biological activity in the biological activity area; Based on the biological activity, the historical movement trajectory, and the influence of the preset density of the soil within a preset area range when the soil data sensor moves, determine the second soil density distribution of the robot movement area and the third soil density distribution of the biological activity area.

4. The method according to claim 1, characterized in that, After the follow-up robot moves the soil data sensor to the acquisition position and after the step of obtaining the soil data collected by the soil data sensor at the acquisition position, the method further includes: Obtain the first ideal acquisition position of the target planting area; When the soil data sensor fails, screen out a second position set within a preset distance range from the first ideal acquisition position in the first position set that the first soil data sensor can reach, where the first soil data sensor can reach the target planting area; Based on the first movement route of the first soil data sensor, the soil density distribution, and the fourth soil density distribution of the first planting area of the first soil data sensor, determine the influence degree of the change in soil density corresponding to each first movement route on crop growth; When there is an influence degree within the preset influence degree range, based on the new acquisition position corresponding to the second movement route with the minimum influence degree, control the first soil data sensor to reach the new acquisition position to collect new soil data; Based on the new acquisition position and the new soil data, update the acquisition position and the soil data.

5. The method according to claim 4, wherein The obtaining of the first ideal acquisition position of the target planting area specifically includes: According to the growth conditions and crop types of each crop in the target planting area, determine the crop growth difference area and the junction area of different crop types; Based on the predicted root growth of each crop in the target planting area, determine the root dense area; Screen out the target horizontal coordinate positions whose horizontal coordinate positions are within the growth difference area, the junction area, and the root dense area; Calculate a target vertical coordinate position based on an ideal distance between a crop root system corresponding to the target horizontal coordinate position and a sensor and a crop root system depth. Obtain a first ideal acquisition position based on the target horizontal coordinate position and the target vertical coordinate position.

6. The method according to claim 4, characterized in that, After the step of, when the soil data sensor has a fault, screening out a second position set within a preset distance range from the first ideal acquisition position in a first position set reachable by a first soil data sensor, the method further includes: Obtain a third position set within a first planting area of the first soil data sensor in the first position set; Calculate data confidence levels of each third position based on a second ideal acquisition position of the first planting area; Screen out a fourth position set with data confidence levels higher than a preset confidence level threshold in the third position set; Plan a first movement route of the first soil data sensor from a current position to a second position and then from the second position to the fourth position.

7. The method according to claim 4, wherein After the step of updating the acquisition position and the soil data based on the new acquisition position and the new soil data, the method further includes: If the first ideal acquisition position and the acquisition position are not within a preset deviation range, determine correction values of each soil parameter in the soil data according to a deviation condition between the first ideal position and the acquisition position and a soil density distribution condition; Update parameter values of each soil parameter in the soil data based on the correction values.

8. The method according to claim 1, wherein The intelligent agriculture system further includes a drone, the drone is equipped with a camera, the camera is used to collect images and videos of crops, the drone is used to move the camera to an acquisition position and perform maintenance tasks, the server is connected to the drone through the communication gateway, and based on the soil data, crop leaf images, and growth trend videos, evaluate the growth conditions of crops in the target planting area through a preset evaluation model, specifically including: Obtain crop leaf images and growth trend videos collected by the camera carried by the drone in the target planting area; Input the crop leaf images, growth trend videos, and the soil data into the preset evaluation model to obtain the growth conditions of crops in the target planting area.

9. A server, characterized in that, Include: One or more processors and a memory; The memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to enable the server to execute the method according to any one of claims 1-7.

10. A computer-readable storage medium storing computer instructions, characterized in that, When the computer instructions run on the server, enable the server to execute the method according to any one of claims 1-7.

Images