JP2021114271A - Smart agriculture support system and smart agriculture support method - Google Patents

Abstract

To provide a smart agriculture support system and a smart agriculture support method which easily enable spraying by a drone for agriculture.SOLUTION: A smart agriculture support system 1000 includes a drone 300 for spraying, and a drone management terminal 200. The drone management terminal 200 includes: a machine learning part 23c having a learning model in which correlation between an NDVI value in a farm field and a spray amount of a spray material to the farm field is learned; an NDVI value calculation part 23a for calculating a current NDVI value in the farm field to be sprayed; and an optimal solution generation part 23d for inputting the current NDVI value, estimating an optimal value of the spray amount of the spray material according to the learning model, and outputting spray information defining the spray amount of the spray material to the farm field to be sprayed. The drone 200 for spray sprays the spray amount of the spray material to the farm field to be sprayed according to the spray information.SELECTED DRAWING: Figure 1

Description

The present invention relates to a smart agricultural support system and a smart agricultural support method, and more particularly to a technique for easily carrying out spraying by an agricultural drone.

Modern agriculture is becoming more dependent on technological progress to address the urgent issues of population decline and aging. For example, agricultural work that utilizes science and technology such as robots and ICT (Information and Communication Technology) is becoming lighter and more efficient. In particular, expectations are high for agricultural drones for pesticide spraying, fertilizer spraying, and seed dispersal.

For example, a drone for spraying pesticides is loaded with a pesticide (mainly a liquid agent) in an amount corresponding to the field area, and flies over the crop while injecting a fixed amount from a nozzle for spraying the liquid agent. In addition, the fertilizer spraying drone is loaded with an amount of fertilizer (mainly granules) according to the field area, and flies over the crop while discharging a fixed amount from the granule spraying skid.

The drone is piloted and flies over the entire field. At this time, the pilot may make adjustments such as increasing the amount of spraying agent applied in a specific area of the field according to the growing condition of the crop.

As one method of grasping the growing situation of crops, a method of flying over a crop with a drone equipped with a multispectral camera, calculating the normalized difference vegetation index (NDVI) of the crop, and creating a vegetation map is known. .. The vegetation map is a display (coloring, etc.) of NDVI for each field area.

Further, Patent Document 1 discloses a system in which an agricultural drone takes a picture of a crop with a multispectral camera or the like while flying over a field, collects the crop / terrain information, and transmits it to an agricultural vehicle on the ground. There is. Agricultural workers use the collected information to spray pesticides.

Patent Document 2 discloses a system that senses a field using a drone equipped with a multispectral camera and calculates a vegetation index of the field. The system provides farmers with information that can be useful for pre-cultivation spraying guidance and the need for top dressing.

JP-A-2018-198609 JP-A-2018-046787

Agricultural workers can grasp the growing situation of crops in the entire field by looking at the vegetation map and the information provided by Patent Documents 1 and 2. However, skilled farmers plan what kind of measures to take based on the grasped growing situation, for example, how much pesticides should be sprayed on which plots, based on experience, wisdom and intuition. There must be. In addition, experience and know-how are required to actually fly the drone based on the plan and spray an appropriate amount of pesticides.

Therefore, for example, it is difficult for inexperienced agricultural workers to utilize the smart agricultural system using a drone as described above. That is, the conventional smart agricultural system cannot determine the specific treatment according to the growing situation of the crop. In addition, it is not possible to fly the drone and spray it based on that judgment. It is thought that these are the factors that prevent the spread of smart agriculture using drones.

Therefore, there is a need for a method of proposing specific measures according to the growing situation of the crops collected by the drone. In addition, there is a need for a method of flying a drone according to the proposal and spraying pesticides.

The present invention is for solving such a problem, and an object of the present invention is to provide a smart agricultural support system and a smart agricultural support method capable of easily carrying out spraying by an agricultural drone.

In the smart agriculture support system according to one embodiment, a spraying drone and a drone management terminal are included, and the drone management terminal correlates the NDVI value in the field with the amount of spraying material sprayed on the field. A machine learning unit having a learning model that has learned the above, an NDVI value calculation unit that calculates the current NDVI value in the field to be sprayed, and the current NDVI value are input, and the spraying agent is sprayed according to the learning model. The spraying drone includes an optimum solution generator that estimates an optimum amount of the amount and outputs spraying information that defines the spraying amount of the spraying agent to the field to be sprayed, and the spraying drone sprays according to the spraying information. The spraying agent in the spraying amount is sprayed on the target field.
In the smart farming support system of one embodiment, the NDVI value calculation unit further includes a drone equipped with a multispectral camera that senses the field to be sprayed with a multispectral camera and transmits the sensing data to the drone management terminal. The NDVI value is calculated based on the sensing data.
In the smart agricultural support system of one embodiment, the drone management terminal further has a flight program generation unit that generates and outputs flight information that defines a flight route in the field to be sprayed, and the spraying drone is a spraying drone. Fly based on the flight information and perform the spraying.
In the smart agricultural support system of one embodiment, the spraying drone adjusts the flight speed or the amount of spraying per unit time based on the spraying information.
In the smart agriculture support system of one embodiment, the field to be sprayed has a plurality of plots, and the optimum solution generation unit generates the spraying information for each plot.
In the smart agriculture support method of one embodiment, a step of calculating the current NDVI value in the field to be sprayed, a step of inputting the current NDVI value, an NDVI value in the field, and a spraying material to the field. A drone for spraying that defines the step of estimating the optimum value of the spraying amount of the spraying agent according to the learning model that learned the correlation with the spraying amount and the spraying amount of the spraying agent to the field to be sprayed. Includes steps to output to.
In the smart agricultural support method of one embodiment, a step of adjusting the flight speed of the spraying drone or the spraying amount per unit time based on the spraying information that defines the spraying amount of the spraying agent to the field to be sprayed. including.

INDUSTRIAL APPLICABILITY According to the present invention, it is possible to provide a smart agricultural support system and a smart agricultural support method capable of easily carrying out spraying by an agricultural drone.

It is an overall view which shows the structure of smart agriculture support system 1000. It is a block diagram which shows the configuration example of the drone management terminal 200. It is a flowchart which shows the operation example of smart agriculture support system 1000. It is a comparison chart of the smart agriculture support system 1000 and the conventional system. It is a flowchart which shows the operation example of smart agriculture support system 1000. It is a figure which shows the spraying pattern of the pesticide or fertilizer by the spraying drone 300.

Specific embodiments of the present invention will be described with reference to the drawings.
<Embodiment 1>
FIG. 1 is an overall view showing the configuration of the smart agriculture support system 1000 according to the first embodiment of the present invention. The smart agriculture support system 1000 includes a drone 100 equipped with a multispectral camera, a drone management terminal 200, and a drone 300 for spraying.

The drone 100 equipped with a multispectral camera has a multispectral camera, a positioning unit, a storage unit, and a communication unit.

Multispectral cameras are typically mounted at the bottom of the fuselage with the lens facing down. A multispectral camera produces an image that contains multi-layered dispersed wavelength (spectral) information called a data cube. That is, it is a digital camera capable of simultaneously capturing and synthesizing a visible light image and an image that visualizes invisible light such as ultraviolet rays and infrared rays.

The positioning unit measures the position information of the drone 100 equipped with the multispectral camera at predetermined time intervals. For the measurement of position information, for example, a satellite positioning system (GNSS: Global Navigation Satellite System) such as GPS (Global Positioning System) or a self-position estimation technology such as SLAM (Simultaneus Localization and Mapping) that can be realized by self-position estimation is adopted.

The storage unit stores the image captured by the multispectral camera and the position information of the drone 100 mounted on the multispectral camera at the time of photographing in association with each other.

The communication unit connects to the drone management terminal 200 via a wireless (or wired) communication line 150, and transmits image information and the like stored in the storage unit. The image information and the like include an image taken by the multispectral camera and position information of the drone 100 equipped with the multispectral camera at the time of shooting.

The spraying drone 300 includes a spraying unit, a communication unit, and a storage unit.

The spraying unit has a tank for loading a liquid such as a pesticide and a nozzle for injecting a predetermined amount of liquid into the field. Alternatively, it is provided with a tank for loading particles such as fertilizer and seeds, and has a skid for discharging a predetermined amount of particles to the field. Typically, the nozzle or skid is attached to the bottom of the spray drone 300 with the nozzle or skid facing down. The spraying unit determines the spraying or discharging amount of pesticides or fertilizers according to the spraying information described later.

The communication unit connects to the drone management terminal 200 via a wireless (or wired) communication line 250, and transmits spraying information, flight information, and the like, which will be described later.

The storage unit stores spraying information, flight information, and the like. The spraying information is information that specifies the spraying amount of pesticides and the like. The spraying information defines the amount of injection or discharge per unit section of, for example, pesticides or fertilizers. The flight information is information that specifies the flight route and the like of the spraying drone 300. The flight information defines, for example, a set of coordinate points that the spraying drone 300 should pass through.

The drone management terminal 200 is a terminal that controls the drone 100 equipped with a multispectral camera and the drone 300 for spraying. Although the specific configuration will be described later, the drone management terminal 200 is communicably connected to the drone 100 equipped with a multispectral camera, the drone 300 for spraying, the server installed on the network 400, the cloud computing environment, and the like. ..

The specific configuration of the drone management terminal 200 will be described.
FIG. 2 is a block diagram showing a configuration example of the drone management terminal 200. The drone management terminal 200 has a communication unit 21, a storage unit 22, and a control unit 23. A part or all of the functions provided by the storage unit 22 and the control unit 23 may be implemented in a server installed on the network 400, or in a cloud computing, fog computing, edge computing environment, etc. (not shown). It may be implemented.

The communication unit 21 connects the drone management terminal 200 to the drone 100 equipped with the multispectral camera via the communication line 150, and receives image information or the like taken by the drone 100 equipped with the multispectral camera. Further, the drone management terminal 200 connects to the spraying drone 300 via the communication line 250, and transmits spraying information and flight information. Further, the drone management terminal 200 is connected to the network 400 and communicates with a server (not shown) or the like installed on the network 400 as needed to receive vegetation big data or the like described later.

The storage unit 22 is a NDVI (Normalized Difference Vegetation Index) value calculated based on image information such as image information transmitted from a drone 100 equipped with a multispectral camera, an NDVI map, and vegetation big data received via a network 400. , A storage device for storing a program or the like for execution by the control unit 23.

The control unit 23 includes an NDVI value calculation unit 23a, an NDVI map generation unit 23b, a machine learning unit 23c, an optimum solution generation unit 23d, and a flight program generation unit 23e.

The NDVI value calculation unit 23a synthesizes the image data of each wavelength band of the crop photographed by the drone 100 equipped with the multispectral camera from the sky above the field, and calculates the NDVI value. The NDVI value is a typical vegetation index called a normalized difference vegetation index (normalized vegetation index), and the amount and vitality of plants can be expressed by this numerical value. The NDVI value calculation unit 23a can, for example, divide the field into a plurality of sections and calculate the NDVI value for each section. Since the method for calculating the NDVI value is known, detailed description thereof will be omitted here.

The NDVI map generation unit 23b generates an NDVI map that visualizes the growing status of crops based on the NDVI value. Agricultural workers can use the NDVI map to determine at a glance variations in the growing status of crops and abnormalities in crops that are difficult to visually confirm. The form of the NDVI map is not particularly limited, but for example, a field can be divided into a plurality of sections, and a map in which the growing status of crops in each section is expressed by coloring or the like can be adopted.

The machine learning unit 23c generates a learning model that expresses the relationship and regularity between various data related to vegetation based on the vegetation big data accumulated in the storage unit 22.

Vegetation big data includes NDVI value, leaf color value, nitrogen absorption amount, number of stems, weather conditions, soil conditions, crop varieties, spray conditions (including types of pesticides to be sprayed, spray amount), vegetation coverage, yield, etc. It is a set of data related to vegetation that is associated with. These data can be collected from, for example, sensors installed in the field, providers of meteorological information, records by agricultural workers, researchers, and the like. In the present embodiment, it is assumed that the vegetation big data stored in advance in the server installed on the network 400 is downloaded to the storage unit 22.

The machine learning unit 23c can construct a learning model by using various known machine learning methods. For example, the spraying conditions and various data observed when the spraying conditions were adopted (NDVI value, leaf color value, nitrogen absorption amount, weather conditions, soil conditions, crop varieties, vegetation coverage, number of stems, yield, etc.) A large amount of data sets associated with, can be used as teacher data for supervised learning. In this way, a learning model showing the correlation between the spraying conditions and various other data is constructed. Alternatively, unsupervised learning can be performed using a large amount of learning data sets (NDVI value, leaf color value, nitrogen absorption, meteorological conditions, soil conditions, crop varieties, spraying conditions, vegetation coverage, yield, etc.). .. In this way, for example, learning including clusters showing data highly correlated with a certain spraying condition (NDVI value, leaf color value, nitrogen absorption amount, meteorological condition, soil condition, crop variety, spraying condition, vegetation coverage rate, number of stems, etc.) A model can be formed. The machine learning method, the learning data creation method, the learning step, and the like shown here are merely examples, and any other method or the like can be used. For example, reinforcement learning, deep learning, or the like may be adopted as the machine learning method. Further, the vegetation big data may be subjected to some kind of preprocessing to create learning data. Further, the learning steps may be performed in batch, or a method of learning at any time by online learning or the like may be adopted, or re-learning may be performed at an arbitrary timing. Further, as the learning data, the NDVI value or the like acquired by the present invention may be used.

As the learning is repeated, the machine learning unit 23c acquires a learning model showing the correlation between the spraying conditions and various parameters such as the NDVI value and the yield. The formed learning model is optimally sprayed for input of unknown parameter sets (NDVI value, leaf color value, nitrogen absorption, meteorological conditions, soil conditions, crop varieties, vegetation coverage, number of stems, yield, etc.). The condition can be output.

The optimum solution generation unit 23d inputs the NDVI value calculated by the NDVI value calculation unit 23a, a desired yield, meteorological conditions, soil conditions, crop varieties, vegetation coverage, etc. into the learning model constructed by the machine learning unit 23c. The training model outputs the dispersal conditions corresponding to these parameter sets. That is, the optimum solution generation unit 23d estimates the optimum type and spray amount of pesticides and the like for the current field conditions. Preferably, the optimal solution generation unit 23d estimates the spraying conditions for each section of the field. As a result, the optimum solution generation unit 23d can estimate the spraying range of the pesticide and the like, and the optimum type and spraying amount of the pesticide and the like for each section. The optimum solution generation unit 23d stores the estimated spraying conditions in the storage unit 22 as spraying information. Further, the communication unit 21 transmits the spraying information to the spraying drone 300. As a result, the drone 100 equipped with the multispectral camera can perform autonomous spraying based on the spraying information.

The NDVI value is one of the parameters of the vegetation big data. For example, the NVDI value and other vegetation big data (for example, leaf color value, nitrogen absorption amount, number of stems, coverage rate, etc.) are primary. Correlation data may be obtained (known correlation data may be obtained, or may be obtained by machine learning as described above), and this correlation data and the optimum application amount of chemicals or fertilizer for each vegetation variety are obtained. The NVDI value obtained by a known means from the sensing data of the field in which the next correlation data is acquired (known correlation data may be acquired or may be acquired by machine learning as described above). And the above quadratic correlation data may be used to calculate the optimum amount of chemicals or fertilizers applied to the vegetation varieties in the field.
As the correlation data between the NVDI value and other big vegetation data (for example, leaf color value, nitrogen absorption amount, number of stems, coverage rate, etc.), known data can be used. For example, "Hiroshi Fujii," Grant-in-Aid for Scientific Research Research Results Report "Development of Crop Diagnosis System Using Newly Developed High-performance Digital Camera", [online], June 7, 2017, Kakenhi, [Reiwa Searched on January 20, 2], Internet <URL: https://kaken.nii.ac.jp/ja/file/KAKENHI-PROJECT-26450016/26450016seika.pd>"," Shinichi Fukumoto, "V Information Processing, etc." Development of high production system by utilizing advanced technology Evaluation of warm region paddy rice growth prediction diagnostic technology by utilizing NDVI measurement Kagoshima Prefectural Agricultural Development Center ”, [online], Public Interest Incorporated Association Agriculture, Forestry and Fisheries / Food Industry Technology Promotion Association, [Reiwa 2 Search on January 20, 2014], Internet <https://www.jataff.jp/project/inasaku/pdf/30_5_12-1.pdf> ”.

The flight program generation unit 23e generates flight information for the spraying drone 300 to perform spraying. The flight information typically defines in order the coordinate points that the spraying drone 300 should pass through. The flight program generation unit 23e stores the generated flight information in the storage unit 22. In addition, the communication unit 21 transmits flight information to the spraying drone 300. As a result, the drone 100 equipped with the multispectral camera can fly autonomously based on the flight information.

An example of generating flight information will be described with reference to FIG. FIG. 6 is a diagram showing a flight route of the spraying drone 300 in the present embodiment. In this example, the entire field is divided into 4 m × 4 m square meshes (A, B, C, D, E ...).

The size of this mesh is determined depending on the specifications of the spraying drone 300. In this embodiment, the spraying drone 300 includes a pair of left and right nozzles or skids. The spray width with one nozzle or skid is 2 m. Therefore, the spraying drone 300 can be sprayed over a width of 2 m × 2 = 4 m. Therefore, the entire field is divided into 4m × 4m square meshes, and the flight route is planned so as to cover these meshes.

In this case, the flight program generation unit 23e can generate flight information in which the coordinates of the center points of each section are specified in order, for example. In the example of FIG. 6, first, the center points of the leftmost columns A, B, C, D, and E, and then the center points of the second column from the left, E, D, C, B, and A.・ ・ Specify in order.

Further, the flight program generation unit 23e may generate flight information for the drone 100 equipped with the multispectral camera to acquire the NVDI value of the field by the same method. In this case, the communication unit 21 transmits the flight information to the drone 100 equipped with the multispectral camera. As a result, the drone 100 equipped with the multispectral camera can autonomously fly and perform sensing based on the flight information.

As the drone management terminal 200, a laptop computer or tablet terminal having an input unit such as a keyboard or a touch panel and an output unit such as a display can be used.

Preferably, the drone management terminal 200 can display the NDVI map generated by the NDVI map generation unit 23b, the spray information generated by the optimum solution generation unit 23d, the flight information generated by the flight program generation unit 23e, and the like on the screen. For example, by superimposing the spraying information and the flight information on the NDVI map and displaying them, the agricultural worker can easily compare and examine the current state of the field and the spraying work to be carried out.

Further, the drone management terminal 200 can provide a means for correcting the displayed spraying information and flight information. For example, in the case of spraying information, an interface for modifying the type and amount of pesticides in each section, and in the case of flight information, the flight order of each section is provided. As a result, in addition to estimation by machine learning, the knowledge and preferences unique to agricultural workers can be reflected in the spraying. The drone management terminal 200 stores the modified spraying information and flight information in the storage unit 22.

FIG. 3 is a flowchart showing the operation of the smart agriculture support system 1000 according to the first embodiment of the present invention.

S101: The drone management terminal 200 creates flight information of the drone 100 equipped with a multispectral camera (automatic navigation). Alternatively, the pilot manually operates to fly over the field (manual navigation).
The drone 100 equipped with a multi-spectral camera captures a field while flying and performs sensing. The drone 100 equipped with a multispectral camera generates a data cube image and transmits it to the drone management terminal 200 via the communication line 150.

S102: The NDVI value calculation unit 23a analyzes the received data cube image and calculates the NDVI value.
The NDVI map generation unit 23b calculates and processes the NDVI value, and generates an NDVI map in which the growth status of the crop is visualized by coloring or the like. The drone management terminal 200 displays the NDVI map.

S103: The optimum solution generation unit 23d estimates the optimum spraying conditions according to the NDVI value of the field and the like based on the learning model generated by the machine learning unit 23c using the vegetation big data. The optimum solution generation unit 23d displays the estimated spraying conditions on the drone management terminal 200.

S104: The agricultural worker confirms the spraying conditions displayed on the drone management terminal 200, and manually corrects them if necessary. The drone management terminal 200 accepts modifications of the spraying conditions. The optimum solution generation unit 23d outputs the approved spraying conditions as spraying information.

S105: The communication unit 21 transmits flight information and spraying information to the spraying drone 300. The spraying drone 300 flies over the field according to the received flight information, and sprays pesticides and the like according to the spraying information.

According to this embodiment, the smart agriculture support system 1000 automatically sets the optimum application conditions for pesticides and the like by machine learning using vegetation big data, based on the field conditions collected by the drone 100 equipped with a multispectral camera. Estimate to. This makes it possible to easily determine the spraying conditions by the drone, which conventionally required wisdom and intuition based on the experience of skilled farmers.

<Embodiment 2>
In the second embodiment, a specific control method of the spraying drone 300 when the spraying conditions are different for each section will be described.
FIG. 5 is a flowchart showing the operation of the smart agriculture support system 1000 according to the second embodiment of the present invention. Since S201 to S205 are the same as S101 to S105 in the first embodiment, the description thereof will be omitted.

S206: The drone management terminal 200 transmits flight information and spraying information to the spraying drone 300 via the communication line 250. Here, the flight information includes, for example, a set of coordinate points to be passed through by the spraying drone 300. The spraying information includes the injection or discharge amount per unit section of, for example, pesticides or fertilizers.

S207: The spraying drone 300 navigates according to the received flight information, that is, sequentially passing through the coordinate points specified in the flight information, and sprays pesticides, fertilizers, and the like.

During navigation, the spraying drone 300 is always aware of which section it is currently sailing over. This is done by comparing the position information of the spraying drone 300 with the predefined section information (range of coordinate values of each section).

The spraying drone 300 acquires spraying information in the currently navigating section, that is, the amount of spraying per unit section. Then, in order to spray the specified amount in the section, either the flight speed or the injection or discharge amount of pesticides or fertilizers per unit time from the nozzle or skid is adjusted. That is, the specified spraying amount is realized by changing the flight speed while keeping the injection or discharge amount of pesticides or fertilizers per unit time from the nozzle or skid constant. Alternatively, the specified spraying amount is realized by changing the injection or discharge amount of pesticides or fertilizers per unit time from the nozzle or skid while keeping the flight speed constant. Agriculturalists may be allowed to choose which method to adopt.

According to the present embodiment, when the amount of pesticide or the like to be sprayed is specified for each section, the smart agricultural support system 1000 is capable of injecting pesticide or fertilizer or the like per unit time from the flight speed or nozzle or skid. Either the discharge rate is automatically adjusted. This makes it possible to autonomously and easily spray a predetermined amount of pesticides and the like.

Finally, using the chart of FIG. 4, the differences between the smart agriculture support system 1000 and the conventional system according to the first embodiment and the second embodiment will be summarized.
In STEP 1, the system senses the field using a drone equipped with a multispectral camera. In STEP2, the system analyzes the field sensing results with software, calculates the NDVI value, and creates a vegetation map (NDVI map). These are processes common to both systems.

On the other hand, the smart agriculture support system 1000 estimates the optimum spraying conditions (type, amount, plot of pesticides to be sprayed) by machine learning in STEP2. Further, in STEP3, the smart agricultural support system 1000 accepts manual correction of the spraying conditions estimated by the system. On the other hand, the conventional system does not have such a configuration, and the spraying conditions are determined by a person.

In step 4, the smart agricultural support system 1000 controls the spray drone 300 based on the spray conditions and flight conditions determined by the system. That is, the spraying drone 300 autonomously flies and sprays based on the spraying conditions and the flight conditions. In addition, when a gust of wind blows during autonomous flight or when an obstacle is found, for example, the flight control is suddenly set to manual with the drug spraying set to automatic control, or the flight speed is set to automatic control. It is also possible to switch the flight route to manual control with. On the other hand, in the conventional system, all manual operations or autonomous flight and spraying with a constant flight speed and chemical spray amount are performed.

In other words, the smart agriculture support system 1000 goes beyond the conventional level of simply providing the field crop information collected by the drone, and makes optimal and concrete proposals regarding the spraying of pesticides, etc. based on the field crop information. be able to. In addition, it can be sprayed autonomously based on the generated proposal.

As described above, the embodiment of the present invention has been described, but the specific configuration of the smart agriculture support system 1000 is not limited to the above-described embodiment, and can be appropriately changed without departing from the spirit. Is. Within the scope of the present invention, any component of the embodiment may be replaced or omitted with an equivalent. Further, any process can be replaced, omitted, or the order can be changed as long as the gist of the invention is not impaired.

For example, in the above-described embodiment, the NDVI value acquired by the drone is used as the learning data, but any data that can be used for machine learning, such as the field crop information confirmed on the ground by the farmer, can be used. Can be used.

100 Drone with multi-spectral camera 150 Wireless communication line 300 Spraying drone 250 Wireless communication line 200 Drone management terminal 21 Communication unit 22 Storage unit 23 Control unit 23a NDVI value calculation unit 23b NDVI map generation unit 23c Machine learning unit 23d Optimal solution generation unit 23e Flight Program Generator 400 Network 1000 Smart Agricultural Support System

Claims (7)

With a drone for spraying
Including drone management terminal
The drone management terminal is
A machine learning unit having a learning model that learned the correlation between the NDVI value in the field and the amount of the spraying material sprayed on the field.
An NDVI value calculation unit that calculates the current NDVI value in the field to be sprayed, and
Optimal solution generation that inputs the current NDVI value, estimates the optimum value of the spraying amount of the spraying agent according to the learning model, and outputs the spraying information that defines the spraying amount of the spraying agent to the field to be sprayed. Including the part
The spraying drone is
A smart agriculture support system that sprays the spraying agent in the spraying amount to the field to be sprayed according to the spraying information. Further including a drone equipped with a multispectral camera that senses the field to be sprayed with a multispectral camera and transmits the sensing data to the drone management terminal.
The smart agricultural support system according to claim 1, wherein the NDVI value calculation unit calculates the NDVI value based on the sensing data. The drone management terminal further has a flight program generation unit that generates and outputs flight information that defines a flight route in the field to be sprayed.
The smart agricultural support system according to claim 1, wherein the spraying drone flies based on the flight information and performs the spraying. The smart agricultural support system according to claim 1, wherein the spraying drone adjusts the flight speed or the spraying amount per unit time based on the spraying information. The field to be sprayed has a plurality of plots and has a plurality of plots.
The smart agricultural support system according to claim 1, wherein the optimum solution generation unit generates the spraying information for each section. Steps to calculate the current NDVI value in the field to be sprayed,
The step of inputting the current NDVI value and
A step of estimating the optimum value of the spraying amount of the spraying agent according to a learning model that learned the correlation between the NDVI value in the field and the spraying amount of the spraying material on the field.
A smart agricultural support method including a step of outputting spraying information defining a spraying amount of the spraying agent to the field to be sprayed to a spraying drone. Spraying information that defines the spraying amount of the spraying agent to the field to be sprayed A smart agricultural support method including a step of adjusting the flight speed of the spraying drone or the spraying amount per unit time based on the spraying information.

Images