Agricultural planting system

Abstract

The embodiment of the invention provides an agricultural planting system. The system comprises a task management module, an execution control module, a data acquisition module and an algorithm processing module, wherein the task management module is used for generating an inspection task; the execution control module communicates with the task management module through a first standardized interface and is used for receiving the inspection task from the task management module and generating an acquisition control instruction according to the inspection task; the data acquisition module communicates with the execution control module through a second standardized interface and is used for receiving the acquisition control instruction from the execution control module, acquiring planting data of a target planting area according to the acquisition control instruction and feeding back the planting data to the algorithm processing module; and the algorithm processing module communicates with the data acquisition module through a third standardized interface and is used for receiving the planting data from the data acquisition module and analyzing the crop growth state of the target planting area according to the planting data by adopting a preset data processing algorithm. All the function modules are decoupled independently, and replacement and upgrading of the modules are facilitated.

Description

Technical Field

[0001] This invention relates to the field of smart agriculture technology, and more specifically to an agricultural planting system. Background Technology

[0002] With the rapid development of agricultural modernization and intelligentization, traditional planting methods that rely on manual inspection and experience-based fertilization can no longer meet the demands for high-yield, stable-yield, and high-quality planting. Smart agriculture systems are gradually introducing technologies such as the Internet of Things, image recognition, and mobile robots to achieve data-driven and intelligent management of crops throughout the entire process from sowing and growth to harvesting.

[0003] Existing agricultural planting systems are typically static and complex structures with chaotic architecture, unclear module functions, and are not conducive to migration and application. Summary of the Invention

[0004] The present invention was proposed in view of the above-mentioned problems. The present invention provides an agricultural planting system.

[0005] According to one aspect of the present invention, an agricultural planting system is provided, comprising a task management module, an execution control module, a data acquisition module, and an algorithm processing module. The task management module generates inspection tasks; the execution control module communicates with the task management module through a first standardized interface to receive inspection tasks from the task management module and generate acquisition control instructions based on the inspection tasks; the data acquisition module communicates with the execution control module through a second standardized interface to receive acquisition control instructions from the execution control module, acquire planting data of a target planting area according to the acquisition control instructions, and feed the planting data back to the algorithm processing module; the algorithm processing module communicates with the data acquisition module through a third standardized interface to receive planting data from the data acquisition module, and uses a preset data processing algorithm to analyze the crop growth status of the target planting area based on the planting data to obtain analysis results.

[0006] For example, the planting data includes crop data of the target crop and / or environmental data of the target planting area. The crop data is image data. The agricultural planting system also includes: a data storage module, which communicates with the algorithm processing module through a fourth standardized interface to receive planting data and analysis results, and stores the image data in local storage. The metadata of the image data, environmental data and analysis results are stored in the local database in the form of structured data.

[0007] For example, the preset data processing algorithm is implemented through a first artificial intelligence model. The agricultural planting system also includes a model deployment module. The model deployment module communicates with the data storage module through a fifth standardized interface to obtain planting data and analysis results from the data storage module, and trains a second artificial intelligence model based on the planting data and analysis results. The first artificial intelligence model is at least a part of the second artificial intelligence model. The algorithm processing module includes the model deployment module, or communicates with the model deployment module through a sixth standardized interface to obtain the first artificial intelligence model from the model deployment module.

[0008] For example, the task management module is specifically used to: receive user-defined tasks issued by the remote task system in response to the user's task setting operation, and generate inspection tasks according to the user-defined tasks; the algorithm processing module is also used to feed back the analysis results to the remote task system.

[0009] For example, the agricultural planting system also includes a voice interaction module and an instruction execution module. The voice interaction module communicates with the instruction execution module through a seventh standardized interface to receive voice questions input by the user, generate decision instructions based on the voice questions, and / or generate and output question-and-answer results based on the voice questions. The decision instructions include inspection instructions. The instruction execution module communicates with the task management module through an eighth standardized interface to generate corresponding control signals based on the inspection instructions and send the control signals to the task management module to generate inspection tasks.

[0010] For example, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the decision-making instructions also include one or more of status query instructions, data query instructions, and function control instructions. The agricultural planting system also includes an instruction interface module, which communicates with the instruction execution module through a ninth standardized interface. The instruction interface module is used to: feed back the status information of the inspection robot to the instruction execution module according to the status query instructions, so that the instruction execution module can send the status information to the voice interaction module for output; and / or, feed back planting data and / or analysis results to the instruction execution module according to the data query instructions, so that the instruction execution module can send the planting data and / or analysis results to the voice interaction module for output; and control the motion function and / or data acquisition function of the inspection robot according to the function control instructions.

[0011] For example, the agricultural planting system also includes an instruction interface module, which communicates with the task management module through a tenth standardized interface. This module receives user-defined tasks from the remote task system and transmits these user-defined tasks to the task management module. The data acquisition module includes cameras and / or environmental sensors on the inspection robot. The instruction interface module also receives debugging instructions from the remote debugging system, debugs the inspection robot according to the instructions, and feeds back the debugging results to the remote debugging system. The instruction interface module supports one or more of the following functions: real-time status monitoring, parameter configuration, user-defined task issuance, robot coordinate query, preset data query, video stream acquisition, remote self-test, firmware upgrade, and version information viewing.

[0012] For example, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the task management module includes a task entry unit, a priority management unit, a pre-task self-check unit, and a task scheduling unit. The task entry unit is used to generate a task queue containing inspection tasks; the priority management unit is used to sort the inspection tasks in the task queue; the pre-task self-check unit is used to perform at least one of the following on the inspection robot: power self-check, hardware self-check, and algorithm service self-check, to obtain a self-check result; and the task scheduling unit is used to send the sorted inspection tasks to the execution control module when the self-check result indicates that the self-check has passed.

[0013] For example, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the task management module includes a status monitoring unit, an anomaly handling unit, and a task recovery unit. The status monitoring unit is used to monitor the status information of the inspection robot, including motion status, health status, and task execution status. The anomaly handling unit is used to determine whether the inspection robot has encountered an anomaly during the execution of the inspection task based on the status information, and to record the breakpoint information of the inspection task when an anomaly occurs during the execution of any inspection task. The task recovery unit is used to resume the execution of the inspection task based on the breakpoint status information when the preset recovery conditions are detected based on the status information.

[0014] For example, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the algorithm processing module includes an algorithm preprocessing unit and an algorithm postprocessing unit; the algorithm preprocessing unit is used to preprocess the planting data, and the preprocessing includes one or more of noise suppression, geometric correction, radiometric correction, format standardization, and multi-source data synchronization; the algorithm postprocessing unit is used to use a preset data processing algorithm to analyze the crop growth status based on the preprocessed planting data and obtain the analysis results.

[0015] For example, the data acquisition module includes a camera mounted on the gimbal of the inspection robot, the gimbal being mounted on the chassis of the inspection robot; the algorithm preprocessing unit is further configured to: optimize the acquisition parameters of the camera based on the planting data to obtain target acquisition parameters, and / or correct the pitch angle of the gimbal of the inspection robot to obtain a target pitch angle, and / or correct the acquisition distance of the inspection robot to obtain a target azimuth angle; the execution control module is specifically configured to: control the camera to adjust the acquisition parameters based on the target acquisition parameters, and / or control the gimbal movement based on the target pitch angle, and / or control the chassis movement of the inspection robot based on the target azimuth angle.

[0016] For example, the data acquisition module includes a first environmental sensor on the inspection robot, and the crop data includes first environmental data collected by the first environmental sensor; the algorithm processing module is also used to acquire second environmental data collected by a second environmental sensor within the target planting area, and to correct the first environmental sensor according to the deviation between the second environmental data and the first environmental data.

[0017] According to the agricultural planting system of the present invention, the main functional modules of the agricultural planting system, including the task management module, the execution control module, the data acquisition module and the algorithm processing module, can communicate with each other through standardized interfaces, so that each functional module is independent and decoupled from each other, and the functions implemented by each functional module are independent of each other. This makes it easy to replace and upgrade any functional module without affecting other functional modules, and can greatly facilitate the migration and expansion of the entire agricultural planting system in different application scenarios. Attached Figure Description

[0018] The above and other objects, features, and advantages of the present invention will become more apparent from the more detailed description of the embodiments of the invention in conjunction with the accompanying drawings. The drawings are provided to further illustrate the embodiments of the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings, the same reference numerals generally represent the same parts or steps.

[0019] Figure 1 A schematic block diagram of an agricultural planting system according to an embodiment of the present invention is shown;

[0020] Figure 2 A schematic block diagram of an agricultural planting system and related remote task system and remote debugging system according to an embodiment of the present invention is shown.

[0021] Figure 3 A schematic diagram of the system architecture of an agricultural planting system according to an embodiment of the present invention is shown. Detailed Implementation

[0022] To make the objectives, technical solutions, and advantages of the present invention more apparent, exemplary embodiments according to the present invention will be described in detail below with reference to the accompanying drawings. Obviously, the described embodiments are merely a part of the embodiments of the present invention, and not all of the embodiments of the present invention. It should be understood that the present invention is not limited to the exemplary embodiments described herein. Based on the embodiments of the present invention described herein, all other embodiments obtained by those skilled in the art without inventive effort should fall within the protection scope of the present invention.

[0023] To at least partially address the aforementioned technical problems, this invention discloses an agricultural planting system. Employing a design philosophy of "modularization + standardized interfaces + intelligent sensing," it integrates a task management module, an execution control module, a data acquisition (i.e., sensing) module, an algorithm processing module, and standardized interfaces to construct a full-process, closed-loop planting management platform for facility agriculture. Each functional module of this agricultural planting system (data acquisition, task management, execution control, and algorithm processing) is independently decoupled, facilitating module replacement and upgrades. In different planting scenarios, the system can be quickly migrated and applied through simple replacement and upgrades of its functional modules. Simultaneously, this agricultural planting system supports automatic task management, automatic data sensing, and real-time processing, effectively improving the intelligence level and management efficiency of agricultural production.

[0024] Figure 1 A schematic block diagram of an agricultural planting system 100 according to an embodiment of the present invention is shown. Figure 1 As shown, the agricultural planting system 100 includes a task management module 110, an execution control module 120, a data acquisition module 130, and an algorithm processing module 140.

[0025] The task management module 110 is used to generate inspection tasks. The inspection tasks described herein instruct the inspection robot to collect data to analyze the growth status of the target crop and / or the environmental conditions surrounding the target crop (e.g., light intensity). The target crop is the crop to be inspected, which can be any type of crop, such as grapes, cucumbers, cabbage, apples, etc. The target crop can include only one type of crop or multiple different types of crops. The target crop is located within a target planting area. The target planting area can be any area, including but not limited to open fields, greenhouses, plastic tunnels, and orchards. The target planting area can include one or more plots, and each plot can include one or more points. The "plot" described herein is a logical plot; each plot can be defined by a set of boundary coordinates and has unique identification information. Each plot can be further divided into several points, and each point can also have unique identification information and be bound to a single crop. The crop types within the same plot can be the same or different. It is preferable that the crop types within the same plot are the same. The growth stages of the same crop located in different plots or points can be the same or different. The growth stage refers to the growth cycle, such as the budding stage, seedling stage, fruiting stage, flowering stage, and ripening stage. Each inspection task is a task instance, which can be represented by unique identification information (e.g., task ID). Each inspection task can correspond to a location to be inspected (referred to as the inspection area in this document). The inspection area of ​​each inspection task can include at least a portion of the plots in the target planting area, i.e., one or more plots. The inspection area of ​​each inspection task can also include at least a portion of the points within any plot, i.e., one or more locations within that plot. Each inspection task can correspond to a task type. The task type can be distinguished according to the inspection purpose of the inspection task. For example, the task type can include one or more of the following: pest and disease identification, weed identification, flower count, fruit count, nitrogen content measurement, water content measurement, chlorophyll content measurement, leaf area index measurement, leaf temperature measurement, fruit sweetness measurement, etc. Each inspection task can also correspond to an execution time, which can be a specific point in time or a time period. Different inspection tasks may correspond to different inspection areas and / or task types and / or execution times. The number of inspection tasks can be one or more, depending on the actual situation.

[0026] The planting data collected by the data acquisition module may include crop data of the target crop and / or environmental data of the target planting area. For example, crop data may include one or more of color (RGB) images, depth images, thermal infrared images, and hyperspectral images. Each of the RGB, depth, thermal infrared, and hyperspectral images can be a static image or a continuous dynamic image, i.e., a video stream. Environmental data may include meteorological data reflecting the air conditions inside and / or outside the target planting area. For example, environmental data may include one or more of air temperature, air humidity, light intensity, and air carbon dioxide concentration.

[0027] The inspection robot can be equipped with one or more cameras and / or environmental sensors to collect data. The camera of the inspection robot can include one or more of the following: color (RGB) camera, color depth (RGBD) camera, thermal infrared camera, hyperspectral camera, etc. Correspondingly, the crop data can include one or more of the following: RGB images acquired by an RGB camera, RGB images acquired by an RGBD camera, depth images acquired by an RGBD camera, thermal infrared images acquired by a thermal infrared camera, hyperspectral images acquired by a hyperspectral camera, etc. The RGB camera can be, for example, a high-resolution RGB camera with at least 12 million effective pixels. Details of the target crop, such as color, texture, and small lesions, can be clearly identified from the RGB images acquired by the high-resolution RGB camera. The hyperspectral camera of the inspection robot can dynamically adjust imaging parameters through an adaptive spectral selection algorithm to optimize the quality of crop data acquisition. The environmental sensors of the inspection robot can collect environmental data. Environmental sensors can include one or more of the following: air temperature sensor, air humidity sensor, light sensor, carbon dioxide concentration sensor, etc. Each of the environmental sensors can exist independently or multiple sensors can be integrated together; for example, the air temperature sensor and air humidity sensor can be implemented using an integrated air temperature and humidity sensor. For example, environmental data may include one or more of the following: air temperature collected by an air temperature sensor, air humidity collected by an air humidity sensor, light intensity collected by a light sensor, and air carbon dioxide concentration collected by a carbon dioxide concentration sensor. Preferably, planting data may include multi-source data (or multimodal data), i.e., multiple different types of data. In existing technologies, data acquisition methods primarily use a single image recognition model (e.g., Faster R-CNN) to process single-type data, lacking the ability to fuse multiple types of data (e.g., hyperspectral images, thermal infrared images). Simultaneous acquisition of multi-source data and unified processing through a pre-set data processing algorithm can yield more accurate analysis results. The pre-set data processing algorithm can be implemented using a pre-trained neural network model, such as YOLOv5 or Segment-Anything. The input to the neural network model is the planting data, and the output is the corresponding analysis result.

[0028] The inspection robot described in this article can be any type of robot capable of performing inspection tasks, including but not limited to one or more of ground inspection robots, aerial inspection robots (such as drones), and underwater inspection robots. The inspection robots corresponding to any two inspection tasks, i.e., the inspection robots used to perform these two inspection tasks, can be the same or different.

[0029] Inspection tasks can include routine inspection tasks and / or anomaly review tasks. For example, the task management module 110 can receive user-defined tasks from a remote task system and / or inspection instructions output by a voice interaction module, and generate inspection tasks (routine inspection tasks) based on the user-defined tasks and / or inspection instructions. For example, the task management module 110 can also automatically generate inspection tasks, such as generating routine inspection tasks according to preset rules and / or generating anomaly review tasks when anomalies are identified in the execution results of an inspection task. Routine inspection tasks are periodic inspection tasks performed according to a preset inspection cycle. For example, routine inspection tasks for the inspection robot can be automatically generated by combining the crop type of the target crop, the growth stage of the target crop, the Geographic Information System (GIS) of the target planting area, the plot calendar, and inspection templates. The inspection template can store preset inspection cycles and task types corresponding to at least one crop type at different growth stages. For example, the inspection template can store information such as "Grapes - Flowering stage - Daily inspection - (Pest and disease identification + flower count statistics)" and "Cucumber - Seedling stage - Inspection every three days - Weed identification". Therefore, by querying the inspection template based on the crop type and growth stage of the target crop to be inspected, the preset inspection cycle and task type for the target crop at its current growth stage can be determined. Simultaneously, by combining a geographic information system and a plot calendar, the required inspection location (i.e., inspection area) and specific execution time for each routine inspection task can be determined, thereby generating the corresponding routine inspection task. The task type and preset inspection cycle for each routine inspection task are based on the crop type and growth stage of the corresponding crop within the inspection area. The task management module 110 can be deployed on the inspection robot, or on an edge computing node and / or a cloud server. For example, the model deployment module 110 can also be partially deployed on the inspection robot and partially deployed on an edge computing node and / or a cloud server.

[0030] The execution control module 120 communicates with the task management module 110 through the first standardized interface to receive inspection tasks from the task management module 110 and generate acquisition control instructions based on the inspection tasks.

[0031] The various standardized interfaces described in this document (including the first standardized interface, the second standardized interface, the third standardized interface, etc.) can be RESTful API interfaces and / or WebSocket interfaces. These standardized interfaces can support the transmission of video streams (Real-Time Streaming Protocol, RTSP) and data in standard formats such as JSON, Protobuf, and XML, thus facilitating third-party integration. For example, an inspection task may include: task objectives, inspection area, task parameters, and constraints. Task objectives may be, for example, "detecting pests and diseases and identifying developmental progress." Inspection areas may be represented by, for example, "GPS boundaries or polygon coordinates." Task parameters may be, for example, "image resolution, acquisition frequency, quality requirements," etc. Constraints may be, for example, "time, energy, priority," etc. The execution control module 120 can extract from the inspection task: inspection area (map coordinates or logical area name), issues of concern (e.g., pest and disease detection), data type requirements (determining which cameras and / or environmental sensors to use), and coverage strategies (line-to-line traversal, fixed-point hovering, etc.). The execution control module 120 can perform task parsing, i.e., understand the task requirements and constraints. The execution control module 120 can maintain a navigation map of the target planting area and perform path planning based on the analysis results and the navigation map to generate the optimal inspection path. The execution control module 120 can also configure cameras and / or environmental sensors based on the analysis results to determine the cameras and / or environmental sensors required for performing the inspection task, as well as the acquisition parameters of each camera and / or environmental sensor. The execution control module 120 can generate a specific sequence of control commands, i.e., acquisition control commands, based on the path planning results and the camera and / or environmental sensor configuration results. The execution control module 120 can be deployed on the inspection robot, or on an edge computing node and / or a cloud server. Exemplarily, the execution control module 120 can also be partially deployed on the inspection robot and partially deployed on an edge computing node and / or a cloud server.

[0032] The data acquisition module 130 communicates with the execution control module 120 through the second standardized interface. It receives acquisition control instructions from the execution control module 120, acquires planting data of the target planting area according to the acquisition control instructions, and feeds the planting data back to the algorithm processing module 140.

[0033] The data acquisition module 130 may include a camera and / or environmental sensors of the inspection robot. The content of the planting data and the exemplary methods of its acquisition have already been described above and will not be repeated here.

[0034] The algorithm processing module 140 communicates with the data acquisition module 130 through a third standardized interface. It receives planting data from the data acquisition module 130 and uses a preset data processing algorithm to analyze the crop growth status of the target planting area based on the planting data, obtaining analysis results. Crop growth status includes one or more of the following: number of flowers, number of fruits, leaf area index, leaf temperature, pest and disease detection results, and fruit maturity. For example, crop growth status detection may include: accurately analyzing growth indicators such as leaf area index, fruit maturity, and number of flowers based on high-resolution images, hyperspectral images, and environmental data of the target crop. These growth indicators can reflect the growth status of the target crop. For example, the preset data processing algorithm may include one or more of the following: hyperspectral analysis algorithm, target detection algorithm, image recognition algorithm, and image segmentation / reconstruction algorithm, and may be implemented using one or more neural network models. The algorithm processing module 140 can be deployed on an inspection robot, or on an edge computing node and / or a cloud server. For example, the algorithm processing module 140 can also be partially deployed on an inspection robot and partially deployed on an edge computing node and / or a cloud server.

[0035] By adopting the above technical solution, the main functional modules of the agricultural planting system, including the task management module, execution control module, data acquisition module, and algorithm processing module, can communicate with each other through standardized interfaces. This allows each functional module to be independent and decoupled from the others, and the functions implemented by each module are independent of each other. This makes it easy to replace and upgrade any functional module without affecting other functional modules, and greatly facilitates the migration and expansion of the entire agricultural planting system in different application scenarios.

[0036] According to an embodiment of the present invention, the planting data includes crop data of the target crop and / or environmental data of the target planting area. The crop data is image data. The agricultural planting system further includes a data storage module, which communicates with the algorithm processing module through a fourth standardized interface to receive planting data and analysis results, and stores the image data in local storage. The metadata, environmental data, and analysis results of the image data are stored in a local database in the form of structured data. For example, the image data stored in the local storage and the data stored in the local database can be synchronized to a cloud storage server for storage. This allows the model deployment module 160 to access this data for model training when it is also deployed in the cloud. Of course, storing data in a cloud storage server also enables rapid data recovery in case of local server failure, ensuring data security and business continuity.

[0037] Figure 2 A schematic block diagram of an agricultural planting system 100 and related remote task system 210 and remote debugging system 220 according to an embodiment of the present invention is shown. Note that... Figure 2 The specific structure of the agricultural planting system 100 shown is merely an example and not a limitation of the invention. For example, Figure 2 The task management module 110, execution control module 120, and algorithm processing module 140 are shown as being located within the inspection robot system (i.e., the inspection robot). However, this is only an example; these modules can be deployed, in whole or in part, in other locations, such as edge nodes or cloud servers. Figure 2 As shown, the agricultural planting system 100 may include a task management module 110, an execution control module 120, a data acquisition module 130, and an algorithm processing module 140. Figure 2 In the diagram, the task management module 110, execution control module 120, and data acquisition module 130 are indicated by solid boxes, while the algorithm processing module 140 is indicated by a dashed box. See also Figure 2 The agricultural planting system 100 may also include a data storage module 150. The data storage module 150 may be deployed on an inspection robot, or on an edge computing node and / or a cloud server. Exemplarily, the data storage module 150 may also be partially deployed on the inspection robot and partially deployed on an edge computing node and / or a cloud server. The data storage module 150 may include local storage and a local database. The local storage may be a file server for storing still images / videos. The local database may be a database such as MySQL, PostgreSQL, or SQLite for storing structured data. The data storage module 150 may store image data (i.e., crop data) from the planting data in the local storage. The image data may include one or more of the aforementioned RGB images, depth images, thermal infrared images, and hyperspectral images. RGB images and depth images may optionally be acquired by the same RGBD camera, referred to as RGBD images. Figure 2 In this example, RGBD images are displayed as RGBD data, thermal infrared images as thermal infrared data, and hyperspectral images as hyperspectral data. For instance, a date tag can be created for each day to distinguish different dates, such as... Figure 2The labels are in the format "YYYY-MM-DD". Daily inspection tasks can store the inspection results (i.e., the collected data) in the corresponding locations under the respective labels. For example, during storage, a separate folder can be created for each date, and each folder can contain folders for different image data, such as folders corresponding to RGBD images, thermal infrared images, and hyperspectral images, respectively. Each folder contains the corresponding images collected at each inspection point on that day. For example, the RGBD image folder contains RGBD images collected at each inspection point on that day, the thermal infrared image folder contains thermal infrared images collected at each inspection point on that day, and the hyperspectral image folder contains hyperspectral images collected at each inspection point on that day. Figure 2 The image data stored in the local memory is shown to consist of multiple sets, each corresponding to an inspection point in a single inspection task, and includes RGBD data, thermal infrared data, and hyperspectral data. However, it should be noted that this is only an example; the data types contained in the image data are not limited to these three types and can include fewer or more. The data storage module 150 can also store environmental data and analysis results from the planting data (in... Figure 2 The metadata of the image data (shown as "algorithm results") is stored in a local database. Figure 2 As shown, for example, the metadata of image data can include background information, acquisition parameters (i.e., camera acquisition parameters), data tags, image storage location, and device information (i.e., the identification information of the inspection robot). Analysis results can include plant number (i.e., crop number), task type, detection results, and image path. Environmental data can include detection type, acquisition time, detection value, and acquisition location. Structured data resides in the database, while unstructured data resides in the file system. They can be linked through the metadata of the image data. For external access, a search can be performed in the local database based on the task ID to retrieve all structured data, such as the data acquisition location (i.e., inspection point), air temperature and humidity, and the metadata of the image data. Then, based on the metadata of the image data, the storage location of the image data can be obtained, that is, different types of images under the date tag. This is how the image data and environmental data are associated.

[0038] By adopting the above technical solution, the data storage module 150 supports collaborative storage of structured databases and unstructured file systems. This facilitates the classification and storage of data by task / time / crop, improving data management efficiency. Furthermore, the data storage module 150 also communicates with other modules through standardized interfaces, meaning it is independently decoupled, facilitating replacement and upgrades.

[0039] According to an embodiment of the present invention, the preset data processing algorithm is implemented through a first artificial intelligence (AI) model. The agricultural planting system also includes a model deployment module, which communicates with the data storage module through a fifth standardized interface to obtain planting data and analysis results from the data storage module, and trains a second AI model based on the planting data and analysis results. The first AI model is at least a part of the second AI model. The algorithm processing module includes the model deployment module, or communicates with the model deployment module through a sixth standardized interface to obtain the first AI model from the model deployment module.

[0040] like Figure 2As shown, the agricultural planting system also includes a model deployment module 160. The model deployment module 160 can communicate with the data storage module 150 through a fifth standardized interface to obtain a dataset from the data storage module 150. The dataset may include the aforementioned image data and structured data, i.e., planting data and corresponding analysis results. The model deployment module 160 can train a second AI model based on the planting data and analysis results. For example, the second AI model can be constructed using one or more algorithms such as Convolutional Neural Network (CNN), Long Short-Term Memory (LSTM), and Random Forest. Preferably, CNN, LSTM, and Random Forest algorithms are integrated to construct a multi-dimensional AI model. For example, the AI ​​model may include one or more of the following: a crop growth status monitoring model, a pest and disease prediction model, a water and fertilizer requirement model, and an environmental adaptation model. The crop growth status monitoring model is used to analyze the growth status of the target crop based on at least a portion of the planting data to determine crop growth status information. For example, crop growth status monitoring models can accurately analyze growth indicators such as leaf area index, fruit development progress, and chlorophyll content based on high-resolution images, hyperspectral images, and environmental data of the target crop. These growth indicators can reflect the growth status of the target crop (e.g., growth stage, presence of abnormalities such as yellowing disease, fruit cracking, and powdery mildew). Pest and disease prediction models are used to predict pests and diseases based on at least some data from the planting data to determine pest and disease prediction information. For example, pest and disease prediction models can combine hyperspectral images of the target crop, environmental data (e.g., air temperature, air humidity, air carbon dioxide concentration), weather conditions, and historical disease data of the target planting area to construct a time-series prediction model, providing early warnings of pest and disease risk and spread at a predetermined time (e.g., 3-7 days). Water and fertilizer requirement models are used to analyze water and fertilizer requirements based on at least some data from the planting data to determine water and fertilizer requirement information. For example, water and fertilizer demand models can calculate precise irrigation amounts, fertilizer types (nitrogen, phosphorus, and potassium ratios), and fertilizer application times based on soil data (such as soil moisture and nutrient content), the target crop's growth stage (i.e., growth cycle, such as flowering period), and meteorological data (such as light intensity, weather conditions, and precipitation). Environmental adaptation models are used to perform environmental adaptation analysis based on at least a portion of the planting data to determine environmental adaptation information. For example, an environmental adaptation model can analyze the target crop's growth stage based on crop data and dynamically match preset suitable environmental data (i.e., environmental adaptation information) for the target crop at different growth stages, such as suitable air temperature, air humidity, light intensity, and air carbon dioxide concentration at the current growth stage. Environmental adaptation information can provide a basis for decision-making in environmental control in planting scenarios such as greenhouses.

[0041] For example, the model deployment module 160 can be deployed on the inspection robot, or on an edge computing node and / or a cloud server. For example, the model deployment module 160 can also be partially deployed on the inspection robot and partially on an edge computing node and / or a cloud server. For instance, in the event of a sudden situation in the field, such as a pest outbreak or extreme weather, the corresponding detection and / or prediction model can be deployed on the inspection robot, while other models can be deployed on edge computing nodes or in the cloud. The detection and / or prediction model for a pest outbreak can be the aforementioned pest detection and / or prediction model, and the detection and / or prediction model for extreme weather can be based on weather data in the environmental data to determine whether the environmental data has changed or is about to change drastically within a preset time period, thereby determining whether extreme weather has occurred or is about to occur.

[0042] The first AI model can be all or part of the second AI model. The model deployment module 160 can train the second AI model, and the trained all or part of the AI ​​model (i.e., the first AI model) can be deployed on the inspection robot as a preset data processing algorithm to analyze planting data. In one embodiment, the second AI model is the same as the first AI model, and the algorithm processing module 140 includes the model deployment module 160, meaning they are the same module. In another embodiment, the algorithm processing module 140 and the model deployment module are independent of each other, but they can also communicate through a standardized interface. The algorithm processing module 140 can call the trained first AI model from the model deployment module 160 through methods such as Representational State Transfer (REST).

[0043] By adopting the above technical solution, the model deployment module can obtain planting data and analysis results from the data storage module to train the second AI model. The algorithm processing module can then extract all or part of the model from the second AI model as the first AI model. In this way, the collected planting data and corresponding analysis results can be used to train the AI ​​model, and the trained AI model can be applied to analyze the collected planting data, completing a closed-loop feedback optimization of data processing. This optimization scheme can continuously iterate and optimize the AI ​​model parameters, forming a fully adaptive closed loop that is suitable for various planting scenarios such as grains, fruits and vegetables, and cash crops.

[0044] According to an embodiment of the present invention, the task management module is specifically used to: receive user-defined tasks issued by the remote task system in response to the user's task setting operation, and generate inspection tasks according to the user-defined tasks; the algorithm processing module is also used to feed back the analysis results to the remote task system.

[0045] See Figure 2 The diagram shows a remote task system 210. The remote task system 210 can be deployed at a remote control center and / or on a user's terminal device, etc. Figure 2 The remote task system 210 is shown to include a digital twin module 212 and a planting management software module 214. The digital twin module 212 provides a human-machine interface and displays robot data through this interface. Exemplarily, the displayed robot data may include one or more of the following: robot status information, task information, real-time detection results, robot trajectory, etc. Exemplarily, status information may include battery level, location, and estimated range. Exemplarily, task information may include the current inspection task and historical inspection tasks. Exemplarily, real-time detection results may include all or part of the planting data currently collected by the data acquisition module 130, such as crop data (various image data) from the planting data; real-time detection results may also include the analysis results currently obtained by the algorithm processing module 140. The planting management software module 214 is used to determine the user-defined task in response to user input. For example, the planting management software module 214 may provide the user with an interface for operating the planting management software. In the user interface, users can use the controls provided by the planting management software to perform task setting operations, such as setting the target area, crop type, inspection target, and execution time window for the user-defined task. The target area can be selected by methods such as manually drawing a polygon; for example, users can click / select the target area (e.g., "Greenhouse No. 5") on the electronic map displayed on the user interface. The target area can be the total inspection area of ​​one or more inspection robots. The crop type can be selected from drop-down menu controls, such as selecting tomato as the crop type to be tested from a drop-down menu control containing "tomato, rice, citrus, etc." The inspection target can be selected from drop-down menu controls, such as selecting "pest and disease detection" as the detection target from a drop-down menu control containing "pest and disease detection, growth assessment, flowering / fruit setting statistics, water stress detection, etc." The execution time window can be set via checkboxes and / or text boxes, for example, to execute immediately, execute by scheduled date range, or execute according to a fixed cycle (e.g., 06:00-09:00 every day for 7 consecutive days).

[0046] On the user side, "task setting" involves the user specifying "where to check, what to check, and how to check." The planting management software module 214 translates the user-defined task into a structured task description file (e.g., JSON / YAML) that the inspection robot can execute, and then distributes it to the task management module 110 via cloud / local download. The task management module 110 can then generate, sort, and schedule more specific tasks based on the user-defined task, determining the inspection tasks to be performed by the inspection robot.

[0047] After obtaining the analysis results, the algorithm processing module 140 can feed them back to the remote task system 210 in real time, so that the remote task system 210 can output them for user viewing. For example, the twin digital module 212 can display planting data and / or analysis results for user viewing, allowing the user to independently understand the crop growth status. For example, the algorithm processing module 140 can feed the analysis results back to the remote task system 210 through the following instruction interface module.

[0048] By adopting the above technical solution, the execution of inspection tasks can be controlled through a remote task system, which allows users to manage and control the inspection tasks of the planting area through any device such as a mobile terminal, facilitating remote operation and improving management efficiency.

[0049] According to an embodiment of the present invention, the agricultural planting system further includes a voice interaction module and an instruction execution module. The voice interaction module communicates with the instruction execution module through a seventh standardized interface, and is used to receive voice question information input by the user, generate decision instructions based on the voice question information, and / or generate and output question-and-answer results based on the voice question information. The decision instructions include inspection instructions. The instruction execution module communicates with the task management module through an eighth standardized interface, and is used to generate corresponding control signals based on the inspection instructions, and send the control signals to the task management module to generate inspection tasks.

[0050] like Figure 2 The diagram illustrates a voice interaction module 170 and an instruction execution module 180. Exemplarily, the voice interaction module 170 can receive voice questions input by users through multiple devices such as mobile terminals (e.g., mobile phone apps), inspection robots, and dedicated voice terminals. For example, it may contain voice questions about the growth status of a target crop. Exemplarily, the voice interaction module 170 may include a voice recognition unit and an intent recognition unit. After voice recognition by the voice recognition unit, the voice recognition result can be input into the intent recognition unit to obtain the user's intent. The voice recognition unit can perform voice recognition on the voice questions. For example, the voice questions can be in Mandarin Chinese, and a Mandarin-specific Automatic Speech Recognition (ASR) model can be used for voice recognition. This model can filter environmental noise generated by ventilation equipment, irrigation equipment, etc., with a recognition accuracy of ≥95%. The voice recognition unit can convert the voice recognition result into structured text to obtain a structured voice recognition result.

[0051] For example, the second AI model described above may further include a pre-trained Natural Language Processing (NLP) model. The voice interaction module 170 can communicate with the model deployment module 160 through the eleventh standardized interface to call the NLP model in the model deployment module 160 to perform intent recognition on the speech recognition results. Specifically, the intent recognition unit can perform intent recognition on the speech recognition results to determine the user's intent. Alternatively, the NLP model can also be deployed in the voice interaction module 170. For example, the voice interaction module 170 can obtain and store a trained NLP model in advance from the model deployment module 160. Of course, the voice interaction module 170 can also obtain the NLP model from other sources. For example, the NLP model may be, for example, a Bidirectional Encoder Representations from Transformers (BERT) model or a similar model. Through the pre-trained natural language model, the user's intent can be recognized and parsed to obtain the parsing result. For example, the parsing result may be "Anomaly cause query; Target object: grapes; Location: Greenhouse No. 1; Symptom description: Leaves are yellow".

[0052] For example, see Figure 2 The voice interaction module 170 may further include a Retrieval-Augmented Generation (RAG) model unit and an instruction generation unit. The RAG model unit can use the RAG model to search a preset knowledge base based on the user's intent output by the intent recognition unit, obtaining a solution corresponding to the user's intent. For example, in an inspection robot system, if a user asks, "How to prevent and control early blight in tomatoes?", the RAG model can first search the preset knowledge base for "methods to prevent and control early blight in tomatoes," and then generate a professional and reliable solution based on the search results. The instruction generation unit can generate decision instructions and / or question-and-answer results based on the solution to control the operation of the inspection robot or other execution devices. The instruction execution module 180 may include a task execution unit, used to generate corresponding control signals based on the decision instructions and send the decision instructions to the corresponding devices for execution. For example, the decision instructions may include inspection instructions, and the control signals corresponding to the inspection instructions can be sent to the task management module to generate inspection tasks. For example, the decision instructions may also include device execution instructions, which can be sent to other execution devices to generate work tasks. The execution equipment may include one or more of the following: mobile execution terminals, irrigation equipment, fertilization equipment, supplemental lighting equipment, ventilation equipment, shading equipment, pesticide application equipment, and plant protection robots. Furthermore, the question-and-answer results can be output via the voice interaction module 170.

[0053] Those skilled in the art will understand that the main functions of an inspection robot include perception, reconnaissance, and diagnosis (problem detection), and its output is data and information. The main functions of a plant protection robot include execution, intervention, and handling (problem resolution), and its output is physical actions. Inspection robots may have sensors for data collection, including cameras and / or environmental sensors. Inspection robots possess autonomous movement and path planning capabilities, and can trigger close-range data collection on crop growth status and pest and disease conditions according to preset routes or instructions, and can simultaneously acquire environmental data at the collection points. Plant protection robots may have actuators, such as pumps, nozzles, and robotic arms, to perform one or more agricultural operations such as spraying (i.e., applying pesticides), weeding, pruning, and harvesting. Supplemental lighting equipment, ventilation equipment, and shading equipment can be referred to as environmental control equipment. Irrigation equipment, fertilization equipment, and pesticide application equipment can be implemented independently, or any two or more can be integrated together; for example, irrigation equipment and fertilization equipment can be integrated into a single water and fertilizer machine. Irrigation equipment, fertilization equipment, supplemental lighting equipment, ventilation equipment, shading equipment, and pesticide application equipment can be devices that are fixedly installed in the target planting area. Mobile execution terminals can include one or more of the following: drones, underwater robots, and rail robots. They can function as inspection robots and / or plant protection robots, possessing sensors for data collection and / or actuators for performing agricultural operations. Mobile execution terminals can supplement the insufficient coverage of ground inspection robots, fixed environmental sensors, and fixed execution equipment, enabling integrated aerial inspection, high-altitude plant protection operations, and large-scale data collection. They are suitable for planting areas such as large-scale open-field planting and complex terrain (e.g., mountains and hills). The command execution module 180 can convert standardized decision commands into control signals recognizable by the equipment and distribute them to the corresponding inspection robot or other execution equipment to control the inspection robot or execution equipment to perform corresponding tasks. These tasks can include data collection and / or agricultural operations. Through control signals, the inspection robot and / or mobile execution terminal can be controlled to collect data from target plots and / or target locations, and / or, one or more of the following—mobile execution terminals, irrigation and fertilization equipment, supplemental lighting equipment, ventilation equipment, shading equipment, pesticide application equipment, and plant protection robots—can be controlled to perform corresponding agricultural operations.

[0054] Exemplarily, the voice interaction module 170 may also include an offline wake-up unit. The offline wake-up unit can initiate the operation of other parts of the voice interaction module 170 besides the offline wake-up unit in response to a user's voice wake-up command. Either the voice interaction module 170 or the instruction execution module 180 may be deployed on the inspection robot, or on an edge computing node and / or a cloud server. Exemplarily, either the voice interaction module 170 or the instruction execution module 180 may also be partially deployed on the inspection robot and partially deployed on an edge computing node and / or a cloud server.

[0055] By adopting the above technical solution, the output of question and answer results and / or the execution of inspection tasks can be automatically realized based on the user's voice question information. This enables the automated linkage of voice information, question and answer and / or automatic inspection, which can effectively improve the management efficiency of agricultural planting system.

[0056] According to an embodiment of the present invention, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the decision-making instructions include one or more of status query instructions, data query instructions, and function control instructions. The agricultural planting system also includes an instruction interface module, which communicates with the instruction execution module through a ninth standardized interface, and is used to: feed back the status information of the inspection robot to the instruction execution module according to the status query instructions, so that the instruction execution module can send the status information to the voice interaction module for output; and / or, feed back planting data and / or analysis results to the instruction execution module according to the data query instructions, so that the instruction execution module can send the planting data and / or analysis results to the voice interaction module for output; and control the motion function and / or data acquisition function of the inspection robot according to the function control instructions.

[0057] like Figure 2 As shown, the instruction execution module 180 may include one or more of a status query unit, a data query unit, and a function control unit, respectively used to receive one or more of a status query instruction, a data query instruction, and a function control instruction, and send the received instruction to the instruction interface module. Figure 2As shown, the agricultural planting system 100 may also include an instruction interface module 190. The instruction interface module 190 is the core communication component in the agricultural planting system, responsible for receiving, parsing, verifying, distributing, and feeding back external control instructions. All modules of the inspection robot can interact with the outside world through the instruction interface module 190. Through the instruction interface module 190, status queries, data queries, and function control of the inspection robot can be achieved. The instruction interface module 190 can feed back the inspection robot's status information to the instruction execution module 180 based on the status query instruction. The status information may include motion status (e.g., whether navigation is abnormal), health status (e.g., whether hardware or charging is abnormal), and task execution status (e.g., execution completed, execution failed, execution interrupted, execution skipped, etc.). The execution result of the inspection task may include the data collected by the inspection robot. If the execution result includes all the data required for this inspection task, it indicates that the inspection task has been completed. If the execution result is that no data was collected or only a portion of the data required for this inspection task was collected, it indicates that the inspection task failed, was interrupted, or was skipped. Execution failure refers to an error occurring at a certain step of the inspection task, making it impossible to continue (e.g., equipment malfunction of the inspection robot). In this case, the execution log can be recorded directly and / or an alarm message can be issued. Alternatively, the task can be retried once or multiple times. If consecutive failures occur, the execution log can be recorded again and / or an alarm message can be issued. Execution interruption refers to interruption caused by external factors (e.g., insufficient remaining power, temporary restriction on the inspection area, manual pause, etc.). In this case, the execution log can be recorded, containing breakpoint information. Execution can be resumed when the external factors are eliminated. Skipping execution means explicitly not executing within the current inspection time window, usually manually skipped by the user, and will not be resumed later. The instruction interface module 190 can feed back planting data and / or analysis results to the instruction execution module 180 based on data query instructions. The instruction interface module 190 can control the inspection robot's motion function and / or data acquisition function based on function control instructions. Motion control can include controlling whether the inspection robot moves and the movement mode (e.g., movement trajectory). Data acquisition control can include controlling whether the inspection robot collects data, the type of data collected, and the collection parameters of the camera and / or environmental sensors.

[0058] By adopting the above technical solution, the status query and / or data query and / or function control of the inspection robot can be realized based on the command interface module. It can realize the automatic status query of the inspection robot by voice questioning, which can improve the interactivity of the agricultural planting system and make it convenient for users to query and / or control the inspection robot by voice at any time.

[0059] According to an embodiment of the present invention, the agricultural planting system further includes an instruction interface module, which communicates with the task management module through a tenth standardized interface, for receiving user-defined tasks issued by the remote task system and transmitting the user-defined tasks to the task management module; the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the instruction interface module is also used to receive debugging instructions from the remote debugging system, debug the inspection robot according to the debugging instructions, and feed back the debugging results to the remote debugging system; the instruction interface module supports one or more of the following functions: real-time status monitoring, parameter configuration, user-defined task issuance, robot coordinate query, preset data query, video stream acquisition, remote self-test, firmware upgrade, and version information viewing.

[0060] like Figure 2As shown, the debugging of the remote debugging system can include control of the Automated Guided Vehicle (AGV) of the inspection robot, power control, gimbal control, abnormal task management, camera parameter settings (i.e., setting the camera's acquisition parameters), log display (i.e., displaying the execution log of the inspection task), and manual algorithm execution (i.e., manually controlling the execution of the aforementioned preset data processing algorithms). The execution log is a time-series detailed record of various key operations, state changes, environmental data, abnormal events, and intermediate results throughout the entire process of the inspection task from generation to completion. For example, the execution log can include one or more of the following layers: task metadata layer, process trajectory layer, job details layer, and system and event layer. The task metadata layer can include one or more of the following: the identification information of the inspection task (e.g., task ID), the identification information of the inspection robot (e.g., robot ID), task type, planned execution time, actual execution time, execution status, etc. The process trajectory layer can include one or more of the following: high-frequency pose flow, path key points, environmental context, etc. The high-frequency pose flow can include, for example, timestamps, GPS / laser SLAM coordinates, speed, heading angle, etc. Key path points can include, for example, path planning results, replanning events and their causes. Environmental context can include, for example, light intensity, air temperature, and air humidity corresponding to a timestamp. The job details layer can include one or more of the following: point access sequence, action execution record, etc. Point access sequence can include, for example, arrival at a point, departure from a point, and dwell time. Action execution record can include, for example, detection type (e.g., infrared thermometry), detection parameters (e.g., camera parameters, environmental sensor parameters), and detection results (e.g., raw planting data, processed planting data, planting data analysis results). The above execution results can include detection results. The system and event layer can include one or more of the following: robot health, critical system events, etc. Robot health can include, for example, battery level corresponding to a timestamp, core temperature, and motor drive current, etc. Critical system events can include, for example, heartbeat reporting, emergency obstacle avoidance triggering, and communication interruption / recovery, etc.

[0061] Real-time status monitoring Figure 2 The "real-time status" (shown as "real-time status") refers to the real-time monitoring of the inspection robot's status information. As mentioned above, the status information can include motion status, health status, and task execution status. Through the real-time status monitoring function, the remote task system 210 and the instruction execution module 180 can query the inspection robot's status information. Parameter configuration refers to configuring the robot's specific operating parameters, such as whether it automatically returns to charging after the inspection task is completed, what level of battery level triggers an alarm, and the time for updating map information. User-defined task assignment (…) Figure 2 The text, labeled "Task Assignment," refers to assigning user-defined tasks to the inspection robot. Robot coordinate query (...) Figure 2The option shown as "Coordinate Query" refers to querying the current coordinates of the inspection robot. Preset data query ( Figure 2 The data that can be queried (shown as "Data Query") can include the current inspection task of the inspection robot, historical inspection tasks, execution results of historical inspection tasks, real-time crop data and analysis results of the current inspection task, running trajectory, etc. Video stream acquisition involves acquiring real-time video streams from the inspection robot. For example, it can be pushed via Fast Forward Moving Picture Experts Group (FFmpeg) to send the real-time images captured by the inspection robot to a remote task system. Remote self-test refers to the user sending commands through a remote task system or remote scheduling system to instruct the inspection robot to perform a self-test to check if its hardware and software functions are normal. Firmware upgrade refers to controlling the inspection robot to upgrade its internal firmware. View version information ( Figure 2 The text displayed as "Version Information" refers to viewing the current software and hardware version of the inspection robot.

[0062] By adopting the above technical solution, information can be received from the remote task system and / or remote debugging system through the modular instruction interface module, and task scheduling or remote debugging can be performed based on the received information, which facilitates users to remotely manage and debug the agricultural planting system.

[0063] According to an embodiment of the present invention, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the task management module includes a task enqueue unit, a priority management unit, a pre-task self-check unit, and a task scheduling unit. The task enqueue unit is used to generate a task queue containing inspection tasks; the priority management unit is used to sort the inspection tasks in the task queue; the pre-task self-check unit is used to perform at least one of the following on the inspection robot: power self-check, hardware self-check, and algorithm service self-check, to obtain a self-check result; and the task scheduling unit is used to send the sorted inspection tasks to the execution control module when the self-check result indicates that the self-check has passed.

[0064] like Figure 2As shown, the task management module 110 may include a task enqueue unit, a priority management unit, a pre-task self-check unit, and a task scheduling unit. The task enqueue unit can generate inspection tasks based on user-defined tasks and / or control signals from the instruction execution module 180, and / or based on a preset inspection template, and add these inspection tasks to the task queue, i.e., enqueue the inspection tasks. For example, regular inspection tasks can be generated within each inspection time window. For instance, the inspection time window can be 24 hours (0:00 to 24:00 every day). The task enqueue unit can adjust the inspection tasks in the task queue in real time based on the execution results of the regular inspection tasks within the current inspection time window to obtain a new task queue. The subsequent priority management unit can sort the inspection tasks in the adjusted task queue. For example, the adjustment may include adding, deleting, and updating the execution time of inspection tasks in the task queue. That is, the task enqueue unit can automatically add, delete, or update subsequent inspection tasks based on the execution results. For example, if an anomaly is identified based on the execution results (such as yellowing leaves or lesion coverage), an anomaly review task can be automatically generated. Both anomaly review tasks and routine inspection tasks are types of inspection tasks. As another example, if the same group of routine inspection tasks (routine inspection tasks with the same inspection area and task type belong to one group) fails to identify anomalies in three consecutive executions (each execution result corresponds to one routine inspection task), its inspection frequency can be reduced or the group of routine inspection tasks can be merged with other inspection tasks. Furthermore, if the inspection area of ​​a routine inspection task is interrupted due to inaccessibility or temporary access restrictions, the routine inspection task can be temporarily frozen and the breakpoint information recorded. The execution time of an inspection task can be represented in the task queue by the execution timestamp associated with the inspection task.

[0065] The priority management unit can sort the inspection tasks in the task queue. For example, the inspection tasks in the task queue can be sorted according to a target sorting strategy. The target sorting strategy is one of at least one preset sorting strategy. When there is only one preset sorting strategy, that preset sorting strategy can be directly determined as the target sorting strategy. When there are multiple preset sorting strategies, they can be switched as needed, with each switch to the desired preset sorting strategy serving as the target sorting strategy. The sorting strategy can be switched at any time during the sorting process of the inspection tasks in the task queue. At least one preset sorting strategy can include one or more of the following: an emergency priority strategy, an energy consumption priority strategy, and a rating priority strategy. An emergency priority strategy is a decision principle that sorts inspection tasks according to the response requirements corresponding to their urgency level. It can be understood that a higher urgency level requires a faster response speed, i.e., a shorter response time. An energy consumption priority strategy is a decision principle that minimizes the total system energy consumption as the highest optimization objective in task planning and execution. A rating priority strategy is a decision principle that sorts inspection tasks according to their priority rating results. Priority scoring results can be obtained as follows: For each inspection task in the task queue, according to preset priority scoring rules, priority scores are assigned to various data points corresponding to the inspection task, obtaining priority scores for each data point; the priority scores for each data point are then weighted and summed according to preset weights to obtain the priority score result for the inspection task. The various data points include at least two of the following: the pest and disease level of the inspection area, the urgency of the crop growth stage in the inspection area, the historical frequency of anomalies in the inspection area, the distance between the inspection area and the current travel path, the difference between the estimated power required to perform the inspection task and the remaining power of the inspection robot, and the test level of the inspection area. The preset priority scoring rules include at least two of the following: the higher the pest and disease level, the higher the priority score; the higher the urgency of the growth stage, the higher the priority score; the higher the historical frequency of anomalies, the higher the priority score; the closer the distance, the higher the priority score; the greater the power difference, the lower the priority score; and the higher the test level, the higher the priority score.

[0066] Multiple data sources refer to multi-dimensional data. For example, multiple data sources may include the pest and disease level of the inspection area in an inspection task. The pest and disease level can be identified and determined based on the execution results of historical inspection tasks within a previous preset time period (which can be called the first preset time period). The first preset time period can be set as needed, such as 7 days, 14 days, 30 days, or 90 days. For example, 7 days can be used as the default historical statistical window for scoring (i.e., the first preset time period). The first preset time period can be customized by the user or adjusted by the user based on the default settings. After each inspection task is executed, the task management module can save its corresponding execution log in a preset database. For example, the preset database can save the execution log in a first-in, first-out manner, for example, retaining the execution log for more than or equal to 180 days. The execution log includes the execution results of the inspection task, such as the collected data. Based on the data collected by the inspection task, the pest and disease level of the inspection area can be identified. The classification method for the pest and disease level can be set as needed, and this document does not limit it. It can be understood that the higher the pest and disease level, the more severe the pest and disease, and in this case, its priority score can be set higher. For example, various data may include the urgency of crop growth stages in the inspection area of ​​the inspection task. Each growth stage of the crop may correspond to a preset growth stage urgency. Based on the current growth stage of the crop in the inspection area, the urgency of the growth stage corresponding to that current growth stage (i.e., the crop growth stage urgency of the inspection area) can be determined. The urgency of each growth stage of the crop can be set as needed, for example, the flowering stage has the highest urgency, followed by the seedling stage, and then the fruiting stage, and so on. The higher the urgency of the growth stage, the higher the corresponding priority score can be. For example, various data may include the historical frequency of anomalies in the inspection area of ​​the inspection task. The historical frequency of anomalies can be determined based on the execution results of historical inspection tasks within a previous preset time period (which can be called the second preset time period). The second preset time period can be set as needed, such as 7 days, 14 days, 30 days, or 90 days. For example, 7 days can be used as the default historical statistics window for scoring (i.e., the second preset time period). The second preset time period can be customized by the user or allowed to be adjusted based on the default settings. The second preset time period can be the same as or different from the first preset time period mentioned above. The higher the frequency of historical anomalies, the greater the corresponding priority score can be. For example, various data may include the distance between the inspection area of ​​the inspection task and the current travel path. The task management module can store a navigation map of the target planting area. Based on the map, path planning can be performed for each inspection task, and the distance between the inspection area of ​​each inspection task and the current travel path of the inspection robot performing that task can be determined in real time. The closer the distance, the greater the corresponding priority score can be.For example, the various data may include the difference between the estimated power required to perform the inspection task and the remaining power of the inspection robot. The task management module can monitor the status of the inspection robot in real time, including its remaining power. The task management module can compare the remaining power with the estimated power required to perform the inspection task. When the remaining power is insufficient to support the execution of the inspection task, the execution of the inspection task can be delayed. That is, the larger the power difference, the lower the corresponding priority score. For example, the various data may include the trial level to which the inspection area of ​​the inspection task belongs. Trial levels can be divided as needed. For example, areas used for variety comparison trials, cultivation technology trials, etc., have a higher priority than ordinary non-trial areas. That is, the higher the trial level, the higher the corresponding priority score. The data types of the various data corresponding to any two different inspection tasks can be all the same, all different, or partially the same and partially different. The number of data types of the various data corresponding to any two different inspection tasks can be the same or different.

[0067] For example, the target sorting strategy can be implemented using a deep learning model. The inspection tasks are input into the deep learning model, which can automatically sort them to obtain the sorted inspection tasks. This approach requires significant training costs. Alternatively, the order of inspection tasks can be manually adjusted by the user. The pre-task self-check unit can perform at least one of the following self-checks: power level self-check, hardware self-check, and algorithm service self-check (i.e., the software algorithm built into the inspection robot). The task scheduling unit can schedule the sorted inspection tasks, assigning execution time and inspection area to specific tasks, and send the information to the execution control module 120 for execution.

[0068] By adopting the above technical solution, inspection tasks can be automatically entered into the queue and sorted, and self-checks can be performed before task scheduling, which helps to improve the orderliness and accuracy of inspection tasks and reduce the failure rate during task execution.

[0069] According to an embodiment of the present invention, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the task management module includes a status monitoring unit, an anomaly handling unit, and a task recovery unit. The status monitoring unit is used to monitor the status information of the inspection robot, including motion status, health status, and task execution status. The anomaly handling unit is used to determine whether the inspection robot has encountered an anomaly during the execution of the inspection task based on the status information, and to record the breakpoint information of the inspection task when an anomaly occurs during the execution of any inspection task. The task recovery unit is used to resume the execution of the inspection task based on the breakpoint status information when a preset recovery condition is detected based on the status information.

[0070] When the inspection robot's status is problematic, the inspection task may be interrupted. For example, navigation failure (such as getting stuck in mud or being blocked by obstacles) or malfunction (such as hardware failure or navigation failure) will both lead to the interruption of the inspection task. Of course, there are other reasons that may cause the inspection task to be interrupted, such as restricted access to the inspection area or manual suspension of the inspection task. When the inspection task is interrupted, the inspection task can be temporarily frozen and an execution log can be recorded. The execution log includes breakpoint information, and execution can be resumed later when preset recovery conditions are met. Freezing means keeping the inspection task in the task queue but not participating in sorting and distribution. Recovery means allowing the inspection task to participate in sorting and distribution. The trigger for recovery may include, for example, the travel path in the inspection area becoming reachable again, the restricted access period in the inspection area ending, sufficient resources of the inspection robot being restored (such as remaining power greater than or equal to a preset power threshold), or manual unfreezing of the inspection task. For example, breakpoint information may include one or more of the following: interruption time, interruption location, interruption reason, task progress at the time of interruption, and on-site snapshot. Interruption time is the moment the interruption occurs. The interruption location refers to the specific location where the interruption occurred, such as a specific plot of land and / or point. Reasons for interruption may include, for example, sudden thunderstorms, drone disconnection, insufficient battery power, or the discovery of a major epidemic requiring urgent reporting. Task progress may include collected blocks and / or points, collected data, uncollected blocks and / or points, and uncollected data. On-site snapshots may include photos or videos of the on-site environment (such as weather conditions and crop status) at the time of interruption. Using this embodiment, when an inspection task is interrupted, execution can automatically resume from the breakpoint, achieving intelligent breakpoint continuation inspection, i.e., an intelligent fault recovery mechanism, thereby improving task scheduling efficiency.

[0071] If the inspection robot has a problem with its status, the inspection task may fail. For example, the task enqueue unit can be used to: if any inspection task fails to execute, and the number of consecutive failures of the inspection task within the third preset time period from the current time does not reach a preset threshold, then calculate the next execution time of the inspection task and re-add the inspection task to the task queue; if any inspection task fails to execute, and the number of consecutive failures of the inspection task within the third preset time period reaches the preset threshold, then record the execution log of the inspection task and / or output alarm information.

[0072] The preset threshold number of attempts can be set to any suitable value as needed, such as 3, 4, or 5 times. The third preset time period can also be set to any suitable value as needed, such as 3 days, 7 days, or 10 days. The third preset time period can be equal to or different from either the first or second preset time period. As mentioned above, an error may occur at a certain step of the inspection task, making it impossible to continue (e.g., equipment failure of the inspection robot), resulting in task failure. It can be understood that when counting the number of failures of any inspection task within the third preset time period preceding the current moment after any inspection task fails, the failure of this execution is also included in the count. For example, if any inspection task has not failed 3 times consecutively in the most recent 7 days (i.e., preceding the current moment), the next execution time can be calculated according to the exponential backoff algorithm, and the inspection task can be added back to the task queue according to the next execution time, thereby retrying the inspection task. If any inspection task fails three times consecutively within the last seven days, it can be confirmed that the inspection task has indeed failed. The execution log of the inspection task can be saved to a preset database, and / or an alarm message can be output. For example, the execution log may include one or more of the following: the inspection task's identification information (e.g., task ID), failure time, failure reason, failure location, number of retries, task progress for each execution, and the identification information of the inspection robot executing the inspection task (e.g., robot ID). The failure location is the specific location at the time of failure, such as a specific plot of land and / or point. Task progress may include collected blocks and / or points, collected data, uncollected blocks and / or points, and uncollected data. Saving the inspection task's execution log to a preset database allows users to view and understand the task's execution status later. Outputting alarm messages may include pushing alarm messages to the user's mobile terminal (e.g., mobile phone), or outputting alarm messages through locally set flashlights, speakers, etc. The alarm message can prompt manual intervention. According to this embodiment, for inspection tasks that fail to execute, they can be retried until the number of consecutive failures within a third preset time period reaches a preset number. This approach can improve the fault tolerance of task scheduling.

[0073] According to an embodiment of the present invention, the data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the algorithm processing module includes an algorithm preprocessing unit and an algorithm postprocessing unit; the algorithm preprocessing unit is used to preprocess the planting data, and the preprocessing includes one or more of noise suppression, geometric correction, radiometric correction, format standardization, and multi-source data synchronization; the algorithm postprocessing unit is used to use a preset data processing algorithm to analyze the crop growth status based on the preprocessed planting data and obtain the analysis results.

[0074] like Figure 2As shown, the algorithm processing module 140 may include an algorithm preprocessing unit and an algorithm postprocessing unit. The algorithm preprocessing unit can perform various preprocessing operations on the initially collected planting data, including noise suppression. Preprocessing can be implemented using existing data preprocessing algorithms. The algorithm postprocessing unit is used to analyze the crop growth status based on the preprocessed planting data and obtain analysis results.

[0075] By adopting the above technical solution, planting data can be preprocessed through a separate algorithm preprocessing unit, which can save time for subsequent formal analysis, reduce the computational pressure of the preset data processing algorithm, and improve processing efficiency.

[0076] According to an embodiment of the present invention, the data acquisition module includes a camera mounted on a gimbal of an inspection robot, the gimbal being mounted on the chassis of the inspection robot; the algorithm preprocessing unit is further configured to: optimize the acquisition parameters of the camera based on planting data to obtain target acquisition parameters, and / or correct the pitch angle of the gimbal of the inspection robot to obtain a target pitch angle, and / or correct the acquisition distance of the inspection robot to obtain a target azimuth angle; the execution control module is specifically configured to: control the camera to adjust the acquisition parameters based on the target acquisition parameters, and / or control the gimbal movement based on the target pitch angle, and / or control the chassis movement of the inspection robot based on the target azimuth angle.

[0077] See Figure 2 The execution control module 120 can communicate with the algorithm preprocessing unit. The algorithm preprocessing unit can optimize the camera's acquisition parameters based on the planting data received from the data acquisition module 130. Figure 2 The image shows one or more of the following: "camera parameter optimization," pitch angle correction, and shooting distance (i.e., acquisition distance) correction. This corresponds to obtaining one or more of the target acquisition parameters, target pitch angle, and target azimuth angle. The execution control module 120 can control the chassis movement of the inspection robot based on one or more of the target acquisition parameters, target pitch angle, and target azimuth angle determined by the algorithm preprocessing unit. Figure 2 (shown as "AGV control") and / or gimbal movement ( Figure 2 (As shown in the diagram as "PTZ control") and / or controlling the camera to change acquisition parameters ( Figure 2(Not shown in the image) to acquire images that better meet preset requirements. The chassis movement control scheme can be adapted to different brands of AGV chassis, only requiring matching the communication protocol between the chassis and the execution control module 120. The above control process can be repeated until the algorithm preprocessing unit detects that the planting data acquired by the data acquisition module 130 meets the preset requirements. For example, when the inspection robot runs to a designated point, the gimbal angle can be adjusted up and down. Based on the image recognition algorithm, the best object to be measured (such as unobstructed leaves, grapes that can be seen in their entirety, etc.) can be selected. The distance to the object to be measured can also be calculated using an RGBD camera to achieve real-time adjustment of the acquisition distance. When the planting data acquired by the data acquisition module 130 meets the preset requirements, the algorithm preprocessing unit can preprocess the planting data at this time and transmit the preprocessed planting data to the algorithm postprocessing unit for further processing to obtain the above analysis results. Figure 2 As shown, the execution control module 120 also includes a map management unit and a point management unit, which can manage the navigation map and each point in the navigation map, respectively. The execution control module 120 can plan the travel path of the inspection robot based on the coordinates of the navigation map and the points, and control the chassis movement of the inspection robot to reach the corresponding inspection area.

[0078] By adopting the above technical solution, the camera acquisition parameters, pitch angle, acquisition distance, etc. can be automatically optimized and corrected by the algorithm preprocessing unit to obtain better data acquisition results.

[0079] According to an embodiment of the present invention, the data acquisition module includes a first environmental sensor on the inspection robot, and the crop data includes first environmental data collected by the first environmental sensor; the algorithm processing module is further used to acquire second environmental data collected by a second environmental sensor within the target planting area, and to correct the first environmental sensor according to the deviation between the second environmental data and the first environmental data.

[0080] The inspection robot can be equipped with a first environmental sensor, and a second environmental sensor can also be installed externally. For example, the inspection robot can be equipped with temperature and / or humidity sensors, and fixed temperature and / or humidity sensors can also be installed inside the greenhouse. The indoor air temperature and humidity collected by the first environmental sensor of the inspection robot can be acquired separately and compared with the indoor air temperature and humidity collected by the second environmental sensor inside the greenhouse. The deviation between the two can be used to correct (i.e., calibrate) the environmental sensor of the inspection robot.

[0081] This approach can automatically calibrate the first environmental sensor on the inspection robot by comparing environmental data, eliminate or compensate for the inherent system errors of the first environmental sensor, and unify the spatiotemporal reference between multiple sensors, thereby obtaining accurate, consistent, and reliable environmental data.

[0082] According to an embodiment of the present invention, the execution control module is further configured to: if the remaining power of the inspection robot is less than or equal to the power safety threshold, stop the operation of generating acquisition control instructions according to the inspection task, and control the inspection robot to return to the preset charging position for charging.

[0083] The preset charging location can be, for example, a pre-set charging station or charging pile location. These locations can be pre-set in the inspection robot's navigation map, allowing the robot to automatically navigate to them for autonomous charging when its battery is low. This approach enables the inspection robot to automatically return to its charging station, enhancing its overall intelligence.

[0084] Figure 3 A schematic diagram illustrating the system architecture (or technology stack hierarchy) of an agricultural planting system according to an embodiment of the present invention is provided. Note that... Figure 3 The system architecture shown is merely an example and not a limitation of the invention. Figure 3 As shown, the system architecture of an agricultural planting system can include a perception layer, an access layer, a transmission layer, and a data layer. The interaction platform can be part of the agricultural planting system or independent of it.

[0085] like Figure 3As shown, the perception layer may include an inspection robot (i.e., the part of the inspection robot excluding the camera and environmental sensors), a hyperspectral imaging system, a visible light imaging system, and environmental sensors. The inspection robot may include an AGV chassis and an upper structure, the upper structure including a gimbal, on which the hyperspectral imaging system, visible light imaging system, and environmental sensors can be mounted. The hyperspectral imaging system may include a camera and a supplementary light. The visible light imaging system may include an RGBD camera and an RGB & thermal infrared camera. Environmental sensors may include a temperature / humidity sensor (shown as "temperature / humidity" in the figure), a light intensity sensor (shown as "illuminance" in the figure), and a carbon dioxide concentration sensor (shown as "carbon dioxide concentration" in the figure). The access layer may include intelligent gateway modules such as WebSocket, RTSP, HTTP / HTTPS, and Modbus. The transmission layer may include 4G / 5G / WIFI communication interfaces. The data layer may include a local database for storing temperature, humidity, illuminance, and carbon dioxide concentration, and a local storage device for storing spectral information (i.e., the aforementioned hyperspectral images or hyperspectral data), images, and videos. The interactive platform, namely the aforementioned remote task system and / or remote scheduling system, may include a robot inspection module, a hyperspectral module, a video monitoring module, and an environmental monitoring module. The robot inspection module has a robot overview function, allowing users to browse the robot's ID and / or model, for example, through the aforementioned digital twin module 212, a realistic visualization of the inspection robot and the target planting area can be displayed. The robot inspection module has an alarm recording function, allowing users to view historical alarm records. The robot inspection module has an inspection recording function, allowing users to view information about historical inspection tasks, such as the execution time, inspection area, execution status, collected planting data, and corresponding analysis results. The robot inspection module has a task template (i.e., the aforementioned inspection template) function, allowing users to customize task templates. The robot inspection module has a robot monitoring function, which can monitor and output the inspection robot's status information in real time. The robot inspection module has an inspection data function, allowing users to query the currently collected planting data and / or corresponding analysis results of the inspection robot. The robot inspection module features task management, allowing users to manage inspection tasks themselves, such as adding, deleting, and adjusting execution times. It also includes robot management functionality, enabling users to manage inspection robots, such as pausing tasks and moving them to designated locations (e.g., charging stations). The hyperspectral module manages the hyperspectral camera, allowing users to enable or disable it and adjust its acquisition parameters. Finally, it configures alarm rules, allowing users to set thresholds for alarms.For example, an alarm can be triggered when the area of ​​a lesion detected based on hyperspectral imagery exceeds a preset area threshold, which is the threshold corresponding to the alarm. The hyperspectral module has an alarm logging function, allowing users to view historical alarm records. The video monitoring module has RGBD camera management functions, allowing users to manage RGBD cameras through this function, such as controlling whether RGBD cameras are enabled and adjusting their acquisition parameters. The video monitoring module also has PTZ camera management functions, allowing users to manage RGBD camera PTZ cameras through this function, such as adjusting their physical orientation, optical parameters, and operating modes. The video monitoring module has alarm rule configuration and alarm logging functions; these two functions can be understood by referring to similar functions of the hyperspectral camera and will not be elaborated further. The environmental detection module has real-time monitoring functions, allowing users to monitor the working status of environmental sensors in real time. The environmental detection module has sensor management functions, allowing users to manage environmental sensors through this function, such as controlling the enabling and disabling of each environmental sensor. The environmental detection module also has alarm rule configuration and alarm logging functions; these two functions can be understood by referring to similar functions of the hyperspectral camera and will not be elaborated further. In addition, it provides algorithmic support for agricultural planting systems, which may include the aforementioned hyperspectral analysis algorithms, target detection algorithms, image recognition algorithms, image segmentation / reconstruction algorithms, and navigation algorithms.

[0086] Although exemplary embodiments have been described herein with reference to the accompanying drawings, it should be understood that the above exemplary embodiments are merely illustrative and are not intended to limit the scope of the invention. Various changes and modifications can be made therein by those skilled in the art without departing from the scope and spirit of the invention. All such changes and modifications are intended to be included within the scope of the invention as claimed in the appended claims.

[0087] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0088] In the several embodiments provided in this application, it should be understood that the disclosed devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed.

[0089] Numerous specific details are set forth in the specification provided herein. However, it will be understood that embodiments of the invention may be practiced without these specific details. In some instances, well-known methods, structures, and techniques have not been shown in detail so as not to obscure the understanding of this specification.

[0090] Similarly, it should be understood that, in order to streamline the invention and aid in understanding one or more of the various aspects of the invention, features of the invention are sometimes grouped together in a single embodiment, figure, or description thereof in the description of exemplary embodiments of the invention. However, this approach should not be construed as reflecting an intention that the claimed invention requires more features than are expressly recited in each claim. Rather, as reflected in the corresponding claims, its inventive point lies in solving the corresponding technical problem with fewer features than all of those in a single disclosed embodiment. Therefore, the claims following the detailed description are hereby expressly incorporated into that detailed description, wherein each claim itself is a separate embodiment of the invention.

[0091] Those skilled in the art will understand that, apart from the mutual exclusion of features, all features disclosed in this specification (including the accompanying claims, abstract, and drawings) and all processes or units of any method or apparatus so disclosed can be combined in any combination. Unless otherwise expressly stated, each feature disclosed in this specification (including the accompanying claims, abstract, and drawings) may be replaced by an alternative feature that serves the same, equivalent, or similar purpose.

[0092] Furthermore, those skilled in the art will understand that although some embodiments described herein include certain features but not others included in other embodiments, combinations of features from different embodiments are intended to be within the scope of the invention and form different embodiments. For example, in the claims, any of the claimed embodiments can be used in any combination.

[0093] The various component embodiments of the present invention can be implemented in hardware, or as software modules running on one or more processors, or a combination thereof. Those skilled in the art will understand that microprocessors or digital signal processors (DSPs) can be used in practice to implement some or all of the functions of some modules in the agricultural planting system according to embodiments of the present invention. The present invention can also be implemented as an apparatus program (e.g., a computer program and computer program product) for performing part or all of the methods described herein. Such programs implementing the present invention can be stored on a computer-readable medium, or can be in the form of one or more signals. Such signals can be downloaded from an Internet website, provided on a carrier signal, or provided in any other form.

[0094] It should be noted that the above embodiments are illustrative of the invention and not restrictive, and that those skilled in the art can devise alternative embodiments without departing from the scope of the appended claims. In the claims, any reference signs placed between parentheses should not be construed as limiting the claims. The word "comprising" does not exclude the presence of elements or steps not listed in the claims. The word "a" or "an" preceding an element does not exclude the presence of a plurality of such elements. The invention can be implemented by means of hardware comprising several different elements and by means of a suitably programmed computer. In the unit claims enumerating several means, several of these means may be embodied by the same item of hardware. The use of the words first, second, and third, etc., does not indicate any order. These words can be interpreted as names.

[0095] The above description is merely a specific embodiment of the present invention or an explanation of that embodiment. The scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. The scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An agricultural planting system, characterized in that, It includes a task management module, an execution control module, a data acquisition module, and an algorithm processing module, among which, The task management module is used to generate inspection tasks; The execution control module communicates with the task management module through a first standardized interface, and is used to receive the inspection task from the task management module and generate acquisition control instructions according to the inspection task. The data acquisition module communicates with the execution control module through a second standardized interface, and is used to receive the acquisition control command from the execution control module, acquire planting data of the target planting area according to the acquisition control command, and feed the planting data back to the algorithm processing module; The algorithm processing module communicates with the data acquisition module through a third standardized interface to receive the planting data from the data acquisition module and uses a preset data processing algorithm to analyze the crop growth status of the target planting area based on the planting data to obtain analysis results.

2. The agricultural planting system according to claim 1, characterized in that, The planting data includes crop data of the target crop and / or environmental data of the target planting area, wherein the crop data is image data, and the agricultural planting system further includes: The data storage module communicates with the algorithm processing module through the fourth standardized interface. It is used to receive the planting data and the analysis results, store the image data in the local memory, and store the metadata of the image data, the environmental data, and the analysis results in the local database in the form of structured data.

3. The agricultural planting system according to claim 2, characterized in that, The preset data processing algorithm is implemented through a first artificial intelligence model, and the agricultural planting system also includes a model deployment module. The model deployment module communicates with the data storage module through the fifth standardized interface to obtain the planting data and the analysis results from the data storage module, and to train the second artificial intelligence model based on the planting data and the analysis results. The first artificial intelligence model is at least a part of the second artificial intelligence model. The algorithm processing module includes the model deployment module, or communicates with the model deployment module through a sixth standardized interface to obtain the first artificial intelligence model from the model deployment module.

4. The agricultural planting system according to any one of claims 1-3, characterized in that, The task management module is specifically used to: receive user-defined tasks issued by the remote task system in response to the user's task setting operation, and generate the inspection task according to the user-defined tasks; The algorithm processing module is also used to feed the analysis results back to the remote task system.

5. The agricultural planting system according to any one of claims 1-3, characterized in that, The agricultural planting system also includes a voice interaction module and a command execution module. The voice interaction module communicates with the instruction execution module through the seventh standardized interface, and is used to receive voice question information input by the user, generate decision instructions based on the voice question information, and / or generate and output question and answer results based on the voice question information, wherein the decision instructions include inspection instructions; The instruction execution module communicates with the task management module through the eighth standardized interface, and is used to generate corresponding control signals according to the inspection instructions, and send the control signals to the task management module to generate the inspection task.

6. The agricultural planting system according to claim 5, characterized in that, The data acquisition module includes a camera and / or environmental sensors on the inspection robot; the decision-making instructions also include one or more of the following: status query instructions, data query instructions, and function control instructions; and the agricultural planting system also includes an instruction interface module. The instruction interface module communicates with the instruction execution module through the ninth standardized interface, for the following purposes: The status information of the inspection robot is fed back to the instruction execution module according to the status query instruction, so that the instruction execution module can send the status information to the voice interaction module for output; And / or, The planting data and / or the analysis results are fed back to the instruction execution module according to the data query instruction, so that the instruction execution module can send the planting data and / or the analysis results to the voice interaction module for output; The inspection robot's motion function and / or data acquisition function are controlled according to the function control instructions.

7. The agricultural planting system according to any one of claims 1-3, characterized in that, The agricultural planting system also includes an instruction interface module, which communicates with the task management module through a tenth standardized interface, for receiving user-defined tasks issued by the remote task system and transmitting the user-defined tasks to the task management module; The data acquisition module includes a camera and / or environmental sensors on the inspection robot. The command interface module is also used to receive debugging commands from the remote debugging system, debug the inspection robot according to the debugging commands, and feed back the debugging results to the remote debugging system. The instruction interface module supports one or more of the following functions: real-time status monitoring, parameter configuration, user-defined task issuance, robot coordinate query, preset data query, video stream acquisition, remote self-test, firmware upgrade, and version information viewing.

8. The agricultural planting system according to any one of claims 1-3, characterized in that, The data acquisition module includes cameras and / or environmental sensors on the inspection robot, and the task management module includes a task entry unit, a priority management unit, a pre-task self-check unit, and a task scheduling unit. The task enqueueing unit is used to generate a task queue containing the inspection tasks; The priority management unit is used to sort the inspection tasks in the task queue; The pre-task self-test unit is used to perform at least one of the following on the inspection robot: power self-test, hardware self-test, and algorithm service self-test, in order to obtain the self-test result; The task scheduling unit is used to send the sorted inspection tasks to the execution control module when the self-test result indicates that the self-test has passed.

9. The agricultural planting system according to any one of claims 1-3, characterized in that, The data acquisition module includes cameras and / or environmental sensors on the inspection robot, and the task management module includes a status monitoring unit, an anomaly handling unit, and a task recovery unit. The status monitoring unit is used to monitor the status information of the inspection robot, including motion status, health status, and task execution status. The exception handling unit is used to determine whether the inspection robot encounters an exception during the execution of the inspection task based on the status information, and to record the breakpoint information of the inspection task when the inspection robot encounters an exception during the execution of any inspection task. The task recovery unit is used to resume the execution of the inspection task based on the breakpoint status information when the preset recovery conditions are detected according to the status information.

10. The agricultural planting system according to any one of claims 1-3, characterized in that, The data acquisition module includes a camera and / or environmental sensors on the inspection robot, and the algorithm processing module includes an algorithm preprocessing unit and an algorithm postprocessing unit. The algorithm preprocessing unit is used to preprocess the planting data, and the preprocessing includes one or more of the following: noise suppression, geometric correction, radiometric correction, format standardization, and multi-source data synchronization. The algorithm post-processing unit is used to analyze the crop growth status based on the pre-processed planting data using a preset data processing algorithm, and obtain the analysis results.

11. The agricultural planting system according to claim 10, characterized in that, The data acquisition module includes a camera mounted on the gimbal of the inspection robot, and the gimbal is mounted on the chassis of the inspection robot. The algorithm preprocessing unit is also used to: optimize the acquisition parameters of the camera according to the planting data to obtain target acquisition parameters, and / or correct the gimbal pitch angle of the inspection robot to obtain target pitch angle, and / or correct the acquisition distance of the inspection robot to obtain target azimuth angle. The execution control module is specifically used to: control the camera to adjust the acquisition parameters according to the target acquisition parameters, and / or control the gimbal movement according to the target pitch angle, and / or control the chassis movement of the inspection robot according to the target azimuth angle.

12. The agricultural planting system according to any one of claims 1-3, characterized in that, The data acquisition module includes a first environmental sensor on the inspection robot, and the crop data includes first environmental data collected by the first environmental sensor. The algorithm processing module is also used to acquire second environmental data collected by a second environmental sensor within the target planting area, and to correct the first environmental sensor based on the deviation between the second environmental data and the first environmental data.

Images