Spraying control method, electronic device, and machine-readable storage medium

Abstract

This application discloses a spraying control method, electronic device, and machine-readable storage medium, relating to the field of agricultural production technology. The method includes: acquiring environmental parameters and image information of a target area; based on the environmental parameters, obtaining first growth state prediction information of the object to be sprayed through a first prediction model; based on the image information, obtaining second growth state prediction information of the object to be sprayed through a second prediction model; fusing the first and second growth state prediction information to obtain co-growth state prediction information of the object to be sprayed; identifying the distribution information of the object to be sprayed in the target area based on the image information; and controlling the spraying equipment to perform spraying operations according to the co-growth state prediction information and the distribution information. The accuracy of growth state prediction is improved through dual-model prediction and fusion, and differentiated spraying control is achieved by combining distribution information, thereby improving the precision and efficiency of agricultural spraying operations.

Description

Technical Field

[0001] This application relates to the field of agricultural production technology, and more specifically to a spraying control method, electronic equipment, and machine-readable storage medium. Background Technology

[0002] With the rapid development of precision agriculture technology, utilizing drone remote sensing, environmental perception, and intelligent decision-making to achieve precise crop management has become an important development direction in the field of agricultural plant protection. In agricultural production, the control of field weeds is a key link in ensuring crop yield and quality.

[0003] In existing technologies, weed control mainly relies on manual inspection and spot spraying or uniform spraying across the entire field. The former is inefficient, while the latter easily leads to pesticide waste and environmental pollution. Some solutions attempt to use drone aerial photography combined with image recognition for targeted weed spraying, but due to the limitations of a single information source, the accuracy and reliability of spraying decisions still need improvement. Furthermore, the operation process lacks effective responses to changes in field conditions, making it difficult to guarantee the consistency and stability of the application effect. Summary of the Invention

[0004] In view of the above-mentioned shortcomings of the prior art, the purpose of this application is to provide a spraying control method, an electronic device, and a machine-readable storage medium.

[0005] To achieve the above objectives, the first aspect of this application provides a spraying control method, comprising: Acquire environmental parameters and image information of the target area, wherein the environmental parameters include at least temperature parameters; Based on environmental parameters, the first growth state prediction information of the object to be sprayed is obtained through the first prediction model; Based on image information, the second growth state prediction information of the object to be sprayed is obtained through a second prediction model, wherein the second prediction model is different from the first prediction model; The first growth state prediction information and the second growth state prediction information are fused together to obtain the synergistic growth state prediction information of the object to be sprayed. Based on image information, identify the distribution information of the objects to be sprayed in the target area; Based on the predicted information and distribution information of the synergistic growth status, the spraying equipment is controlled to perform spraying operations.

[0006] In this embodiment of the application, the first growth state prediction information and the second growth state prediction information are fused to obtain the synergistic growth state prediction information of the object to be sprayed, including: Obtain the first prediction reliability of the first prediction model and the second prediction reliability of the second prediction model; Based on the first prediction reliability and the second prediction reliability, determine the first weight corresponding to the first growth state prediction information and the second weight corresponding to the second growth state prediction information. Based on the first weight and the second weight, the first growth state prediction information and the second growth state prediction information are weighted and fused to obtain the collaborative growth state prediction information.

[0007] In this embodiment of the application, based on image information, the distribution information of the object to be sprayed in the target area is identified, including: The image information is input into the target recognition model to obtain the location and category information of the object to be sprayed, which serves as the distribution information. The target recognition model includes a lightweight feature extraction network to reduce computational cost, an attention mechanism module to enhance the expression of key features, and a multi-scale feature fusion module to fuse features at different scales.

[0008] In this embodiment of the application, the spraying equipment is controlled to perform spraying operations based on synergistic growth state prediction information and distribution information, including: Determine the timing of spraying based on the information on synergistic growth status prediction; Based on the distribution information, the target area is divided into multiple units; The baseline spraying amount for each unit is determined based on the distribution information of the objects to be sprayed within each unit. A spraying plan is generated based on the spraying timing and the baseline spraying amount for each unit; Control the spraying equipment to perform spraying operations based on the spraying plan.

[0009] In this embodiment of the application, controlling the spraying equipment to perform spraying operations based on the spraying plan includes: Acquire real-time image information of each unit collected in real time during the spraying operation by the spraying equipment; For each unit: The first correction coefficient corresponding to the unit is determined based on the difference between the first quantity and the second quantity, wherein the first quantity is the number of objects to be sprayed in the unit based on the real-time image information, and the second quantity is the number of objects to be sprayed in the unit based on the distribution information. The corrected spraying amount of the unit is determined based on the base spraying amount of the unit and the first correction factor. When the spraying equipment flies to the unit, the output parameters of the spraying equipment are adjusted according to the corrected spraying amount corresponding to the unit, so that the spraying equipment can perform spraying operations based on the output parameters.

[0010] In this embodiment of the application, the corrected spraying amount of the unit is determined based on the base spraying amount of the unit and the first correction coefficient, including: Acquire the environmental parameters of the unit in real time during the spraying operation of the spraying equipment; The second correction coefficient corresponding to the unit is determined based on the implementation environment parameters; The corrected application rate of the unit is determined based on the unit's baseline application rate, the first correction factor, and the second correction factor.

[0011] In this embodiment of the application, controlling the spraying equipment to perform spraying operations based on the spraying plan includes: Based on the location information of multiple units, an initial flight path covering multiple units is generated through a path planning algorithm; The initial flight path is optimized based on the trajectory optimization algorithm to generate a smooth flight trajectory that conforms to the motion characteristics of the spraying equipment; The spraying equipment is controlled to fly along a smooth flight trajectory, and when it flies to each unit, it performs spraying operations according to the spraying plan corresponding to that unit.

[0012] In this embodiment of the application, controlling the spraying equipment to fly along a smooth flight trajectory includes: Obstacle information is acquired when obstacles are detected during the flight of the spraying equipment; Based on obstacle information and preset safety rules, the flight path of the spraying equipment is replanned.

[0013] A second aspect of this application provides an electronic device, comprising: The memory is configured to store instructions; The processor is configured to retrieve instructions from memory and, when executing the instructions, to implement the spraying control method as described in the above embodiments.

[0014] A third aspect of this application provides a machine-readable storage medium storing instructions for causing a machine to perform the spraying control method as described in the above embodiments.

[0015] The above technical solution acquires environmental parameters and image information of the target area; based on the environmental parameters, a first prediction model is used to obtain the first growth state prediction information of the object to be sprayed; based on the image information, a second prediction model is used to obtain the second growth state prediction information of the object to be sprayed; the first and second growth state prediction information are fused to obtain the co-growth state prediction information of the object to be sprayed; based on the image information, the distribution information of the object to be sprayed in the target area is identified; and based on the co-growth state prediction information and distribution information, the spraying equipment is controlled to perform spraying operations. Multi-dimensional data acquisition achieves comprehensive perception of the target area and the object to be sprayed. The dual-model prediction and fusion using different modeling principles improves the accuracy and robustness of growth state prediction. Combined with the distribution information of the object to be sprayed, differentiated spraying control is achieved, effectively avoiding problems such as improper timing, pesticide waste, and poor spraying effects in traditional spraying operations. This improves the accuracy and efficiency of agricultural spraying operations, while reducing the amount of agricultural inputs used and lowering the environmental impact. It is suitable for spraying operations in various agricultural planting scenarios and has broad application value.

[0016] Other features and advantages of the embodiments of this application will be described in detail in the following detailed description section. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the embodiments of this application and form part of the specification. They are used together with the following detailed description to explain the embodiments of this application, but do not constitute a limitation on the embodiments of this application. In the drawings: Figure 1 The schematic diagram illustrates a flow chart of a spraying control method according to an embodiment of this application; Figure 2 A schematic block diagram of an electronic device according to an embodiment of this application is shown. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are only for illustration and explanation of the embodiments of this application and are not intended to limit the embodiments of this application. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0019] It should be noted that the acquisition, transmission, storage, use, and processing of data in the technical solution of this application all comply with relevant laws and regulations. In the embodiments of this application, certain existing industry solutions such as software, components, and models may be mentioned. These should be considered exemplary, intended only to illustrate the feasibility of implementing the technical solution of this application, and do not imply that the applicant has already used or necessarily used such solutions.

[0020] It should be noted that if the embodiments of this application involve directional indicators (such as up, down, left, right, front, back, etc.), the directional indicators are only used to explain the relative positional relationship and movement of each component in a certain specific posture (as shown in the figure). If the specific posture changes, the directional indicators will also change accordingly.

[0021] Furthermore, if the embodiments of this application involve descriptions such as "first" or "second," these descriptions are for descriptive purposes only and should not be construed as indicating or implying their relative importance or implicitly specifying the number of technical features indicated. Therefore, features defined with "first" or "second" may explicitly or implicitly include at least one of those features. Additionally, the technical solutions of various embodiments can be combined with each other, but this must be based on the ability of those skilled in the art to implement them. If the combination of technical solutions is contradictory or impossible to implement, it should be considered that such a combination of technical solutions does not exist and is not within the scope of protection claimed in this application.

[0022] Figure 1 A schematic flowchart illustrating a spraying control method according to an embodiment of this application is shown. Figure 1 As shown in the figure, this application provides a spraying control method, which may include the following steps: Step 100: Obtain environmental parameters and image information of the target area, wherein the environmental parameters include at least temperature parameters; It should be noted that in agricultural spraying operations, traditional uniform spraying methods do not take into account the differences in the growth state and distribution of the target organisms, which can easily lead to problems such as improper spraying timing, waste of pesticides, or poor results. In this embodiment, multi-dimensional data acquisition and multi-model fusion analysis are used to achieve precise control of spraying operations.

[0023] Specifically, the target area can refer to areas such as farmland, orchards, and tea gardens where agricultural spraying operations are required. Environmental parameters are various environmental indicators that affect the growth and development of the objects to be sprayed. In addition to temperature parameters, these may include humidity parameters, light intensity parameters, wind speed parameters, and soil temperature and humidity parameters. Temperature parameters can be obtained through devices such as contact temperature sensors and infrared temperature sensors, while humidity parameters can be obtained through devices such as capacitive humidity sensors and resistive humidity sensors. Image information is visual data obtained by image acquisition of the target area. This can be obtained through visible light cameras, multispectral cameras, satellite remote sensing equipment, and ground-based fixed camera equipment mounted on drones. For example, in rice paddy spraying operations, a quadcopter drone equipped with a multispectral camera can be used to take aerial photos of the rice paddy to obtain image information. Environmental parameters such as temperature and humidity can be obtained through sensor arrays deployed in the rice paddy. In orchard spraying operations, image information of the fruit tree area can be obtained through rail-mounted camera equipment, and environmental parameters such as temperature and light intensity within the orchard can be obtained through small weather stations.

[0024] Step 200: Based on environmental parameters, obtain the first growth state prediction information of the object to be sprayed through the first prediction model; It should be noted that this embodiment predicts the growth status of the target object at the mechanistic level based on environmental parameters. The first prediction model is a prediction model constructed based on environmental parameters and the growth and development law of the target object. The modeling principle of the first prediction model is mostly mathematical fitting or mechanistic analysis combined with agricultural production laws. The target object may include farmland weeds, crop diseases, and the crop itself that needs fertilization. The growth status prediction information reflects the growth stage, growth status, and development degree of the target object, such as the leaf age and plant height of weeds, the growth period of crops, and the spread stage of diseases. For example, for barnyard grass in paddy fields, the accumulated temperature can be calculated based on temperature parameters. The leaf age information of barnyard grass can be obtained as the first growth status prediction information by fitting the accumulated temperature with the leaf age of barnyard grass. For the fertilization needs of fruit trees in orchards, a prediction model of the vegetative growth stage of fruit trees can be constructed based on temperature and light parameters to obtain the vegetative growth stage of fruit trees as the first growth status prediction information.

[0025] Specifically, the first prediction model is a mechanistic prediction model constructed based on environmental parameters and the growth and development patterns of the target organism. In the rice weed control scenario, the first prediction model can adopt the accumulated temperature growth model, which is calculated using the following formula: Accumulated temperature =Σ[(Tmax+Tmin) / 2-Tb] The temperature and humidity coefficient is calculated using the following formula: Temperature and humidity coefficient = Σ[(Tmax-Tmin) / 2] Where Tmax represents the highest temperature of the day; Tmin represents the lowest temperature of the day; Tb represents the biological lower limit temperature of the target organism to be sprayed, and for barnyard grass, Tb is taken as 10℃; the temperature-humidity coefficient reflects the effect of diurnal temperature difference on weed growth.

[0026] The growth model for barnyard grass can be calculated using the following formula: Y = 0.0032X² + 0.1845X - 1.6875 The following formula can be used to calculate the growth model of a baby girl: Y = 0.0028X² + 0.2105X - 2.3125 Where Y represents leaf age and X represents accumulated temperature; by substituting the calculated accumulated temperature value into the corresponding formula, weed leaf age can be obtained as the prediction information for the first growth state.

[0027] Step 300: Based on image information, obtain the second growth state prediction information of the object to be sprayed through the second prediction model, wherein the second prediction model is different from the first prediction model; It should be noted that this embodiment predicts the growth status of the target object based on image information at the visual level. The second prediction model differs from the first prediction model in terms of modeling principles, data input types, and model structure. The second prediction model is often a machine learning or deep learning model trained on image data. It can extract and analyze features from the image information of the target area to obtain the growth status information of the target object. For example, by processing images of weeds in paddy fields using a convolutional neural network, the number of weed leaves can be identified to obtain leaf age information as the second growth status prediction information. Similarly, by analyzing images of fruit trees using a deep learning model, the growth length of new shoots can be identified to obtain the fruit tree's growth pattern as the second growth status prediction information. This embodiment achieves growth status prediction through visual data, which can intuitively reflect the actual growth of the target object and overcome the limitations of prediction based on a single environmental parameter.

[0028] In one embodiment, the second prediction model can employ an improved Mask R-CNN model. This model uses Mask R-CNN as a baseline, selects MobileNetV3 as the backbone network, and incorporates an SE attention mechanism for improvement. Before training, field plant images must be collected using a mobile phenotyping platform or fixed equipment, and pixel-level masks for each plant and leaf in the images must be meticulously annotated. The actual leaf age is also recorded. The model is then trained using an enhanced dataset and an optimized loss function, enabling it to learn how to accurately segment both the plant and leaves simultaneously. After applying the trained second prediction model to new images, a post-processing algorithm counts the number of leaves per plant based on the leaf masks, thus achieving automatic, high-throughput estimation of leaf age. The resulting leaf age information is the second growth state prediction information, and the recognition accuracy of the second prediction model must meet the requirement of ≥90%.

[0029] Step 400: The first growth state prediction information and the second growth state prediction information are fused together to obtain the synergistic growth state prediction information of the object to be sprayed. This embodiment fuses two types of growth state prediction information, combining prediction results from both the mechanistic and visual levels. Fusion methods may include weighted fusion, feature splicing fusion, and voting fusion. The collaborative growth state prediction information is a more accurate growth state information that integrates multi-dimensional prediction results. For example, weighted fusion of weed leaf age obtained based on accumulated temperature and weed leaf age obtained based on image recognition yields more accurate collaborative prediction information for weed leaf age. Similarly, voting fusion of fruit tree growth stage obtained based on environmental parameters and fruit tree growth stage obtained based on image analysis determines the final growth state prediction information for the fruit tree. By fusing the prediction results of different models, the prediction error caused by data bias or model limitations of a single model is reduced.

[0030] Step 500: Based on image information, identify the distribution information of the objects to be sprayed in the target area; It should be noted that the distribution information can include the location, quantity, density, and regional distribution range of the target object in the target area. This can be achieved through image processing techniques such as target detection, image segmentation, and feature recognition. For example, by processing aerial images of paddy fields using target detection algorithms, the location and quantity of each weed can be identified, thus obtaining the distribution information of weeds in the paddy fields. Similarly, by processing images of orchard diseases using image segmentation techniques, the distribution area and area of ​​diseased organisms on fruit trees can be identified, thus obtaining the distribution information of diseased organisms. This step provides a spatial dimension basis for subsequent differentiated spraying, avoiding resource waste caused by indiscriminate spraying.

[0031] Step 600: Based on the synergistic growth status prediction information and distribution information, control the spraying equipment to perform spraying operations.

[0032] It should be noted that spraying equipment can include plant protection drones, ground-based self-propelled sprayers, suspended sprayers, orchard wind-assisted sprayers, etc. The type and concentration of the pesticide to be sprayed can be determined according to the growth status of the target object, and the spraying amount and range for different areas can be determined according to the distribution information. For example, for weeds in paddy fields, the optimal herbicide and spraying concentration can be determined according to the leaf age of the weeds, and the amount of herbicide to be sprayed in different areas of the paddy field can be determined according to the location and density of the weeds. Plant protection drones can be controlled to carry out targeted spraying. For fertilizing fruit trees in orchards, the type and concentration of fertilizer can be determined according to the growth stage of the fruit trees, and the amount of fertilizer to be applied to different fruit trees can be determined according to the distribution location and growth status of the fruit trees. Ground sprayers can be controlled to carry out precise fertilization.

[0033] In this embodiment, environmental parameters and image information of the target area are acquired; based on the environmental parameters, a first prediction model is used to obtain the first growth state prediction information of the object to be sprayed; based on the image information, a second prediction model is used to obtain the second growth state prediction information of the object to be sprayed; the first and second growth state prediction information are fused to obtain the co-growth state prediction information of the object to be sprayed; based on the image information, the distribution information of the object to be sprayed in the target area is identified; and the spraying equipment is controlled to perform spraying operations according to the co-growth state prediction information and the distribution information. Comprehensive perception of the target area and the object to be sprayed is achieved through multi-dimensional data acquisition. The accuracy and robustness of growth state prediction are improved through the dual-model prediction and fusion using different modeling principles. Differentiated spraying control is achieved by combining the distribution information of the object to be sprayed, effectively avoiding problems such as improper timing, pesticide waste, and poor spraying effects in traditional spraying operations. This improves the accuracy and efficiency of agricultural spraying operations, while reducing the amount of agricultural inputs used and lowering the environmental impact. It is suitable for spraying operations in various agricultural planting scenarios and has broad application value.

[0034] In one embodiment, the first growth state prediction information and the second growth state prediction information are fused to obtain the synergistic growth state prediction information of the object to be sprayed, including: Obtain the first prediction reliability of the first prediction model and the second prediction reliability of the second prediction model; Based on the first prediction reliability and the second prediction reliability, determine the first weight corresponding to the first growth state prediction information and the second weight corresponding to the second growth state prediction information. Based on the first weight and the second weight, the first growth state prediction information and the second growth state prediction information are weighted and fused to obtain the collaborative growth state prediction information.

[0035] It should be noted that traditional single-model growth state prediction information fusion methods are prone to deviations in fusion results due to unreasonable weight settings, failing to fully leverage the predictive advantages of different models. In this embodiment, weighted fusion is performed based on model prediction reliability to improve the accuracy of collaborative growth state prediction information. Specifically, prediction reliability is an indicator reflecting the accuracy of the prediction results of the prediction model. For the first prediction model, its reliability reflects the stability and accuracy of the first prediction model in predicting growth states under specific environmental conditions; for the second prediction model, its reliability reflects the reliability of the second prediction model in recognizing morphological features under specific image quality conditions. The reliability of predictions can be determined based on the prediction accuracy of the prediction model on the test set, or dynamically evaluated based on the completeness and validity of the real-time input data of the prediction model. It can also be calculated by combining the prediction deviation rate in historical applications, or by normalizing the accuracy of the model's prediction results. For example, if the accuracy of the first prediction model in the test set for predicting the leaf age of weeds in paddy fields is 90%, then the first prediction reliability of the first prediction model can be set to 0.9. If the validity of the data is reduced due to blurring of real-time images, the second prediction reliability of the second prediction model can be adjusted accordingly. For the first prediction model for accumulated temperature and the second prediction model for machine vision, the prediction reliability can also be corrected after verifying the model's prediction results through actual field observations of growth status.

[0036] Furthermore, based on the first and second prediction reliability, a first weight corresponding to the first growth state prediction information and a second weight corresponding to the second growth state prediction information are determined. The weights can be determined using a normalization method, directly using the prediction reliability as the corresponding weight; alternatively, the weights can be calculated based on the proportional relationship of the reliability; or a preset adjustment coefficient can be introduced to correct the weights, with the sum of the first and second weights being 1. Alternatively, according to the actual scenario requirements, an adjustment coefficient can be added to the visual second prediction model to increase its weight proportion. Finally, based on the first and second weights, the first and second growth state prediction information are weighted and fused to obtain the collaborative growth state prediction information. The weighted fusion is achieved by multiplying each prediction information by its corresponding weight and then summing the results: Collaborative growth state prediction information = First weight × First growth state prediction information + Second weight × Second growth state prediction information, thus making the prediction results more closely match the actual growth state.

[0037] This embodiment matches the prediction weights of different models with their prediction capabilities, effectively reducing the impact of single model prediction errors on the final results, improving the accuracy and robustness of the synergistic growth state prediction information, and providing a more reliable basis for the subsequent accurate determination of spraying strategies.

[0038] In one embodiment, based on image information, identifying the distribution information of the objects to be sprayed in the target area includes: The image information is input into the target recognition model to obtain the location and category information of the object to be sprayed, which serves as the distribution information. The target recognition model includes a lightweight feature extraction network to reduce computational cost, an attention mechanism module to enhance the expression of key features, and a multi-scale feature fusion module to fuse features at different scales.

[0039] In this embodiment, it should be noted that the location information can be represented by the coordinates of the object to be sprayed in the target area and the range of the area to which it belongs, while the category information can reflect the specific type of the object to be sprayed, such as different weed types like barnyard grass and goosegrass in rice paddies, or different types of diseases in orchards. Before inputting the target recognition model, the image information can also be preprocessed, including using super-resolution generative adversarial network SRGAN or enhanced super-resolution generative adversarial network ESRGAN to improve image resolution, while performing image enhancement, format standardization, and mesh generation processing for aerial image segmentation. The target recognition model includes a lightweight feature extraction network to reduce computational cost, an attention mechanism module to enhance the expression of key features, and a multi-scale feature fusion module to fuse features at different scales. The lightweight feature extraction network can be GhostNet, the MobileNet series of networks, or the Xception deep separable convolutional network. In this embodiment, GhostNet is preferred because it generates phantom feature maps through inexpensive linear operations, significantly reducing parameters and computational cost while maintaining performance. Such networks reduce computational cost by simplifying convolution operations and reducing the number of model parameters, making them suitable for resource-constrained applications such as UAV onboard computers and edge computing devices. The attention mechanism module can be a block convolutional attention module (CBAM), a squeezed convolutional attention module, or a similar module. The model incorporates various attention modules, including a pressure-excited attention module (SE) and a coordinate attention module (CA). In this embodiment, a CBAM (Content-Based Attention Module) is preferred, as it includes channel attention and spatial attention. This module weights the features fed into the feature fusion network, allowing the model to automatically focus on the key feature regions of the target object, reducing background interference and improving the effectiveness of feature representation. For multi-scale feature fusion modules, Feature Pyramid Network (FPN), Path Aggregation Network (PAN), and Spatial Pyramid Pooling Module (SPP) can be used. This embodiment employs a combination of FPN and PAN, integrating SPP or an improved version of SPPF. By using multiple max-pooling kernels of different sizes in parallel, the receptive field of the model is increased. These modules can fuse feature information at different scales in the image, enabling effective identification of targets of different sizes and growth stages. In this embodiment, the target recognition model is based on the YOLOv5 framework, integrating and replacing the aforementioned modules. Model reconstruction is achieved by defining new modules, registering existing modules, and modifying network configuration files. Training is also required using a paddy field weed dataset, employing data augmentation techniques such as random cropping and brightness adjustment to simulate complex field imaging conditions.

[0040] This embodiment improves the model's operating efficiency while ensuring recognition accuracy through the synergistic effect of multiple modules. It can quickly and accurately obtain the location and category information of the object to be sprayed, providing accurate spatial and category basis for subsequent differentiated spraying and avoiding spraying errors caused by recognition deviations.

[0041] In one embodiment, controlling the spraying equipment to perform spraying operations based on synergistic growth state prediction information and distribution information includes: Determine the timing of spraying based on the information on synergistic growth status prediction; Based on the distribution information, the target area is divided into multiple units; The baseline spraying amount for each unit is determined based on the distribution information of the objects to be sprayed within each unit. A spraying plan is generated based on the spraying timing and the baseline spraying amount for each unit; Control the spraying equipment to perform spraying operations based on the spraying plan.

[0042] It should be noted that the spraying timing is the period when the target plant is most sensitive to the pesticide and the spraying effect is optimal. For example, the best time to spray is when weeds such as barnyard grass and goosegrass in paddy fields are at the 2-3 leaf stage. Different types of target plants correspond to different optimal spraying times, which can be determined through agricultural production patterns and field trials. Unit division can be carried out according to fixed-size grids, or dynamically based on the distribution density of the target plant, or irregularly based on the topography of the target area. In this embodiment, farmland is preferably divided into 10m × 10m square grid units, and the grid units inherit the GPS data of the image for easy subsequent positioning and zoning management. For example, farmland can be divided into 10m × 10m square grid units, and orchards can be divided into strip units according to the planting row spacing of fruit trees. The baseline spraying rate can be determined based on information such as the number, density, and type of the target organisms within a unit, combined with field efficacy trial data. For example, when the number of weeds is 0-10, the application rate is 20% of the recommended dose; for 11-50 weeds, it is 50%; for 51-100 weeds, it is 80%; and for more than 100 weeds, it is 100%. The more numerous and denser the target organisms, the higher the corresponding baseline spraying rate can be. Different types of target organisms can have different baseline spraying rates set according to their sensitivity to the pesticide. The spraying plan can include the specific spraying time, the spraying rate for each unit, and the operating parameters of the spraying equipment. Finally, the spraying equipment is controlled to execute the spraying operation based on the spraying plan. The spraying equipment can dynamically adjust the spraying pressure and flight speed according to the spraying plan to achieve variable-rate spraying operations.

[0043] This embodiment achieves precise planning of both spraying timing and spraying amount, ensuring that the spraying operation is highly matched with the growth status and spatial distribution of the target organism, effectively improving the spraying effect, reducing the ineffective use of pesticides, and lowering agricultural production costs and environmental impact.

[0044] In one embodiment, controlling the spraying equipment to perform spraying operations based on a spraying plan includes: Obtain the real-time image information of each unit collected in real time during the spraying operation of the spraying device; For each unit: Determine the first correction coefficient corresponding to the unit according to the difference between the first quantity and the second quantity, where the first quantity is the number of objects to be sprayed in the unit based on the real-time image information, and the second quantity is the number of objects to be sprayed in the unit in the distribution information; Determine the corrected spraying amount of the unit according to the reference spraying amount of the unit and the first correction coefficient; When the spraying device flies to the unit, adjust the output parameters of the spraying device according to the corrected spraying amount corresponding to the unit, so that the spraying device performs the spraying operation based on the output parameters.

[0045] In this embodiment, it should be noted that the real-time image information can be collected by devices such as visible light cameras and multispectral cameras carried by the spraying device. The collection frequency can be set according to the flight speed and unit area of the spraying device to ensure that the actual situation of the objects to be sprayed in the unit can be completely captured. The collected real-time image information needs to be processed by the target recognition model deployed on the airborne processor of the spraying device to obtain a more accurate actual observation value of the objects to be sprayed. The first quantity is the number of objects to be sprayed in the unit based on the real-time image information, and the second quantity is the number of objects to be sprayed in the unit in the distribution information. The determination of the first correction coefficient can be calculated according to the ratio of the difference in the number of objects to be sprayed, or different correction coefficient values can be corresponding to different preset quantity difference intervals. In this embodiment, the first correction coefficient is calculated by the following formula:

[0046] Among them, α represents the first correction coefficient; k represents a preset empirical coefficient, and its value is 0.98; D represents the first quantity; D0 represents the second quantity.

[0047] In one embodiment, it is assumed that when the first quantity is 20% more than the second quantity, the first correction coefficient can be set to 1.2, and when the first quantity is 20% less than the second quantity, the first correction coefficient can be set to 0.8. And when D > D0, the spraying amount increases by 20%, and when D < D0, the spraying amount decreases by 20%. If the number of objects to be sprayed in the grid is single and the recognition confidence is high, D can be directly used to replace D0 to achieve dynamic update of the distribution information.

[0048] It should be noted that the corrected spraying amount is the product of the reference spraying amount and the first correction coefficient, which can make the spraying amount fit the actual number of objects to be sprayed in the unit. When the spraying device flies to the unit, adjust the output parameters of the spraying device according to the corrected spraying amount corresponding to the unit, so that the spraying device performs the spraying operation based on the output parameters. The output parameters can include the spraying pressure, nozzle flow rate, atomization particle size, etc. of the spraying device. The accurate control of the spraying amount is achieved by adjusting these parameters.

[0049] This embodiment realizes real-time dynamic correction of the spraying amount, effectively making up for the lag and deviation of the previous distribution information, so that the spraying amount is highly matched with the actual situation of the target in the field, further improving the accuracy of the spraying operation and avoiding the problems of missed spraying and double spraying.

[0050] Further, based on the unit's baseline spraying rate and the first correction factor, the corrected spraying rate for the unit is determined, including: Acquire the environmental parameters of the unit in real time during the spraying operation of the spraying equipment; The second correction coefficient corresponding to the unit is determined based on the implementation environment parameters; The corrected application rate of the unit is determined based on the unit's baseline application rate, the first correction factor, and the second correction factor.

[0051] It should be noted that real-time environmental parameters include at least temperature parameters, and may also include humidity parameters, wind speed parameters, and light intensity parameters. These parameters can be collected by miniature weather sensors and environmental monitoring modules mounted on the spraying equipment, accurately reflecting the real-time environmental conditions within the unit. These environmental parameters are also the basic data for calculating accumulated temperature and temperature-humidity coefficient, and can be synchronously stored in the system database. The second correction coefficient can be determined based on the influence of environmental parameters on the efficacy of the pesticide. In this embodiment, the second correction coefficient is calculated using the following formula:

[0052] Where β represents the second correction coefficient; pi represents the influence weight coefficient of each meteorological factor; and ri represents the real-time monitoring data of each meteorological factor.

[0053] For example, when the wind speed is high, the pesticide is prone to drift. The second correction factor can be appropriately increased to ensure an effective spraying amount. When the humidity is high, the pesticide adheres better to the surface of the object to be sprayed. The second correction factor can be appropriately decreased to reduce pesticide usage. In this embodiment, the corrected spraying amount is calculated using the following formula:

[0054] Among them, (D) i ) indicates the final corrected application rate; f(DL) i ) represents the baseline spraying amount per unit; α i Indicates the first correction factor; β i This represents the second correction factor.

[0055] The corrected spraying rate is the product of the baseline spraying rate and the first and second correction factors, taking into account both the number of objects to be sprayed and the environmental factors.

[0056] This embodiment realizes multi-factor dynamic adjustment of the spraying amount, so that the spraying amount not only matches the actual quantity of the target object, but also adapts to the real-time field environmental conditions, effectively ensuring the spraying effect of the pesticide, while avoiding pesticide waste and reduced efficacy caused by environmental factors.

[0057] In one embodiment, controlling the spraying equipment to perform spraying operations based on a spraying plan includes: Based on the location information of multiple units, an initial flight path covering multiple units is generated through a path planning algorithm; The initial flight path is optimized based on the trajectory optimization algorithm to generate a smooth flight trajectory that conforms to the motion characteristics of the spraying equipment; The spraying equipment is controlled to fly along a smooth flight trajectory, and when it flies to each unit, it performs spraying operations according to the spraying plan corresponding to that unit.

[0058] It should be noted that traditional flight path planning for spraying equipment does not take into account the motion characteristics of the equipment, which can easily lead to problems such as uneven paths, unstable speed during flight, and uneven spraying. In this embodiment, the initial flight path is optimized to generate a smooth trajectory that is adapted to the motion characteristics of the spraying equipment.

[0059] Specifically, the path planning algorithm can be selected from scan line method, spiral path method, ant colony algorithm, genetic algorithm, etc. In this embodiment, the scan line method based on geometric decomposition is preferred. It follows the principle of long route to reduce the number of turns. For complex concave polygon fields, the concave polygon decomposition method is first used to divide them into multiple convex polygon sub-regions. Then, the scan line method is applied in each sub-region. The scan line method is suitable for target areas with regular shapes and achieves full coverage through parallel routes. Ant colony algorithm and genetic algorithm are suitable for target areas with irregular shapes or obstacles and can generate better full coverage paths. The initial flight path is optimized using trajectory optimization algorithms to generate a smooth flight trajectory that conforms to the motion characteristics of the spraying equipment. Trajectory optimization algorithms can include minimizing SNAP trajectory optimization, Dubins path algorithm, and Bezier curve fitting algorithm. In this embodiment, a combination of minimizing SNAP trajectory optimization and Dubins path algorithm is preferred. The minimizing SNAP trajectory optimization algorithm makes the flight smoother by constructing a time allocation function and adding a position offset gradient penalty factor. The Dubins path algorithm generates the shortest path composed of arcs and straight line segments that conforms to the minimum turning radius constraint of the UAV. This type of algorithm can smooth sharp turns and polygonal lines in the initial path, ensuring that the flight trajectory conforms to the minimum turning radius, maximum acceleration, and other motion characteristics of the spraying equipment. The spraying equipment is controlled to fly along the smooth flight trajectory, and spraying operations are performed according to the corresponding spraying plan when it reaches each unit. For scenarios involving multiple scattered farmlands and multi-machine collaborative operations, a deep reinforcement learning algorithm combined with Simultaneous Localization and Mapping (SLAM) can also be used for global path optimization.

[0060] This embodiment makes the flight trajectory of the spraying equipment more closely match its motion characteristics, ensuring the stability of the flight process and effectively avoiding problems such as uneven spraying, overspraying, and missed spraying caused by uneven path. At the same time, it improves the operating efficiency of the spraying equipment and reduces flight energy consumption.

[0061] In one embodiment, controlling the spraying equipment to fly along a smooth flight trajectory includes: Obstacle information is acquired when obstacles are detected during the flight of the spraying equipment; Based on obstacle information and preset safety rules, the flight path of the spraying equipment is replanned.

[0062] It should be noted that traditional spraying equipment flight path planning is static and cannot cope with sudden obstacles that occur during flight, which can easily lead to equipment collisions, work interruptions and other problems. Therefore, this embodiment performs real-time obstacle avoidance and path replanning during flight to ensure the safe and smooth progress of spraying operations.

[0063] Specifically, the spraying equipment is controlled to fly along a smooth flight trajectory, including obtaining obstacle information when obstacles are detected during the flight of the spraying equipment. The obstacle identification can be achieved through equipment such as lidar, visual obstacle avoidance camera, and ultrasonic sensor mounted on the spraying equipment. The obstacle information includes the location, size, shape, and type of the obstacle, which can accurately reflect the actual situation of the obstacle. Based on obstacle information and preset safety rules, the flight path of the spraying equipment is replanned. These preset safety rules may include the minimum safe distance between the equipment and obstacles, the full coverage requirement after path replanning, and the shortest possible flight path. In this embodiment, the preset safety rules follow a shrinkage principle, setting shrinkage distances around field boundaries and obstacles. When there are obstacles at the boundaries, the shrinkage is one spray width; when there are no obstacles, the shrinkage is appropriately increased by about 1 meter to reduce wind-induced missed spraying. Path replanning can utilize algorithms such as the Rapid Exploratory Random Tree Algorithm (RRT), Dijkstra's algorithm, and A* algorithm. This embodiment prefers heuristic methods such as genetic algorithms, which can quickly search for the optimal solution among multiple alternative obstacle avoidance paths. It can also combine Durbins paths to generate obstacle avoidance trajectories that conform to flight characteristics, quickly generating the optimal flight path covering the remaining work units while ensuring safety. All data from the entire spraying operation, including actual spray volume, flight trajectory, and real-time images, are recorded and transmitted back to the system database for evaluating the operation effect and optimizing subsequent algorithm models.

[0064] This embodiment realizes real-time obstacle avoidance and dynamic path replanning of the spraying equipment, effectively avoiding the risk of obstacles during flight, ensuring the operational safety of the spraying equipment, while ensuring full coverage of the spraying operation, avoiding the problem of missed spraying caused by obstacles, and improving the stability and integrity of the spraying operation.

[0065] Figure 2 A schematic block diagram of an electronic device according to an embodiment of this application is shown. Figure 2 As shown, this application provides an electronic device that may include: Memory 1000 is configured to store instructions; The processor 2000 is configured to retrieve instructions from the memory 1000 and to implement the above-mentioned spraying control method when executing the instructions.

[0066] This application also provides a machine-readable storage medium storing instructions that cause a machine to perform the above-described spraying control method.

[0067] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0068] This application is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0069] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0070] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0071] In a typical configuration, a computing device includes one or more processors (CPU), input / output interfaces, network interfaces, and memory.

[0072] Memory may include non-persistent memory in computer-readable media, such as random access memory (RAM) and / or non-volatile memory, such as read-only memory (ROM) or flash RAM. Memory is an example of computer-readable media.

[0073] Computer-readable media includes both permanent and non-permanent, removable and non-removable media that can store information using any method or technology. Information can be computer-readable instructions, data structures, modules of programs, or other data. Examples of computer storage media include, but are not limited to, phase-change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technologies, CD-ROM, digital versatile optical disc (DVD) or other optical storage, magnetic tape, disk storage or other magnetic storage devices, or any other non-transferable medium that can be used to store information accessible by a computing device. As defined herein, computer-readable media does not include transient computer-readable media, such as modulated data signals and carrier waves.

[0074] It should also be noted that the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus. Unless otherwise specified, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or apparatus that includes that element.

[0075] The above are merely embodiments of this application and are not intended to limit the scope of this application. Various modifications and variations can be made to this application by those skilled in the art. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the scope of the claims of this application.

Claims

1. A spray control method characterized by, include: Acquire environmental parameters and image information of the target area, wherein the environmental parameters include at least temperature parameters; Based on the environmental parameters, the first growth state prediction information of the object to be sprayed is obtained through the first prediction model. Based on the image information, the second growth state prediction information of the object to be sprayed is obtained through the second prediction model, wherein the second prediction model is different from the first prediction model; The first growth state prediction information and the second growth state prediction information are fused to obtain the collaborative growth state prediction information of the object to be sprayed. Based on the image information, the distribution information of the object to be sprayed in the target area is identified; Based on the predicted information of the synergistic growth status and the distribution information, the spraying equipment is controlled to perform spraying operations.

2. The spray control method of claim 1, wherein The step of fusing the first growth state prediction information and the second growth state prediction information to obtain the synergistic growth state prediction information of the object to be sprayed includes: Obtain the first prediction reliability of the first prediction model and the second prediction reliability of the second prediction model; Based on the first prediction reliability and the second prediction reliability, determine the first weight corresponding to the first growth state prediction information and the second weight corresponding to the second growth state prediction information; Based on the first weight and the second weight, the first growth state prediction information and the second growth state prediction information are weighted and fused to obtain the collaborative growth state prediction information.

3. The spray control method of claim 1, wherein The step of identifying the distribution information of the object to be sprayed in the target area based on the image information includes: The image information is input into the target recognition model to obtain the location and category information of the object to be sprayed, which is used as distribution information. The target recognition model includes a lightweight feature extraction network to reduce computational load, an attention mechanism module to enhance the expression of key features, and a multi-scale feature fusion module to fuse features at different scales.

4. The spray control method of claim 1, wherein The step of controlling the spraying equipment to perform spraying operations based on the synergistic growth state prediction information and the distribution information includes: The timing of spraying is determined based on the predicted information of the synergistic growth status; Based on the distribution information, the target area is divided into multiple units; The baseline spraying amount for each unit is determined based on the distribution information of the objects to be sprayed within each unit. A spraying plan is generated based on the spraying timing and the baseline spraying amount for each unit; The spraying equipment is controlled to perform spraying operations based on the spraying plan.

5. The spraying control method according to claim 4, characterized in that, The controlled spraying equipment performs spraying operations based on the spraying plan, including: Acquire real-time image information of each unit collected in real time during the spraying operation by the spraying equipment; For each of the aforementioned units: The first correction coefficient corresponding to the unit is determined based on the difference between the first quantity and the second quantity, wherein the first quantity is the number of objects to be sprayed in the unit based on the real-time image information, and the second quantity is the number of objects to be sprayed in the unit based on the distribution information. The corrected spraying amount of the unit is determined based on the baseline spraying amount of the unit and the first correction coefficient. When the spraying equipment flies to the unit, the output parameters of the spraying equipment are adjusted according to the corrected spraying amount corresponding to the unit, so that the spraying equipment performs spraying operations based on the output parameters.

6. The spray control method of claim 5, wherein The step of determining the corrected spraying amount of the unit based on the baseline spraying amount of the unit and the first correction coefficient includes: The environmental parameters of the unit are collected in real time during the spraying operation by the spraying equipment; The second correction coefficient corresponding to the unit is determined based on the implementation environment parameters; The corrected spraying amount of the unit is determined based on the baseline spraying amount of the unit, the first correction factor, and the second correction factor.

7. The spray control method of claim 4, wherein The controlled spraying equipment performs spraying operations based on the spraying plan, including: Based on the location information corresponding to the multiple units, an initial flight path covering the multiple units is generated through a path planning algorithm; The initial flight path is optimized based on the trajectory optimization algorithm to generate a smooth flight trajectory that conforms to the motion characteristics of the spraying equipment; The spraying equipment is controlled to fly along the smooth flight trajectory, and when it flies to each of the units, it performs spraying operations according to the spraying plan corresponding to the unit.

8. The spray control method of claim 7, wherein, The controlled spraying device flies according to the smooth flight trajectory, including: If an obstacle is detected during the flight of the spraying equipment, obstacle information is acquired; Based on the obstacle information and preset safety rules, the flight path of the spraying equipment is replanned.

9. An electronic device, comprising: include: The memory is configured to store instructions; The processor is configured to retrieve the instructions from the memory and, when executing the instructions, to implement the spraying control method according to any one of claims 1 to 8.

10. A machine-readable storage medium, characterized in that, The machine-readable storage medium stores instructions for causing the machine to perform the spray control method according to any one of claims 1 to 8.

Images