Guiding system, estimation system, guiding method, estimation method and program

Abstract

To provide a guiding system for driving away wildlife from areas desired to be protected from wildlife damage, and guiding the wildlife to areas where they can live.SOLUTION: A guiding system includes a plurality of threatening / attracting operation parts, a plurality of detection parts, an acquisition part, a determination part, and an actuation / attraction control part. The plurality of threatening / attracting operation parts are respectively disposed in a plurality of areas and threaten / attract wildlife in each area. The plurality of detection parts are respectively disposed in the plurality of areas and detect wildlife in each area. When wildlife is detected by one of the plurality of detection parts, the acquisition part acquires location information representing a location of the wildlife. The determination part determines one guiding pattern among guiding patterns for actuating each of the plurality of threatening / attracting operation parts, based on the location information. The actuation control part actuates each of the plurality of threatening / attracting operation parts in a threatening / attracting mode based on the one guiding pattern.SELECTED DRAWING: Figure 6

Description

[Technical field]

[0001] The present invention relates to a guidance system, an estimation system, a guidance method, an estimation method, and a program. [Background technology]

[0002] Birds and animals have been causing damage to agriculture, forestry, and fisheries for a long time. The damage caused by birds and animals does not only affect agriculture, forestry, and fisheries, but also has a negative impact on the natural ecosystem. The negative impact on the natural ecosystem includes, for example, the invasion of national parks where rare plants live, and the rare plants being eaten by the birds and animals. Furthermore, there are concerns that the expected population decline and the decline of satoyama (traditional Japanese woodland forests) will exacerbate the conflict between wild animals and humans. For example, there is concern that the invasion of wild animals into human living areas could lead to the risk of zoonotic diseases.

[0003] As a measure against birds and animals, a system has been disclosed which, when a surveillance camera captures an image of a bird or animal, scares it off based on repelling pattern information corresponding to the type of bird or animal and the start and end dates and times of image capture (see, for example, Patent Document 1). [Prior art documents] [Patent documents]

[0004] [Patent Document 1] JP 2017-205033 A Summary of the Invention [Problem to be solved by the invention]

[0005] However, the conventional technology had a problem in that it was not possible to drive away birds and animals from the places where it was desired to prevent bird and animal damage and to guide them to places where they could live. Also, there was a problem in that a sustained effect of driving away birds and animals could not be expected if the birds and animals were simply chased away.

[0006] The present invention has been made in consideration of the above circumstances, and its purpose is to provide a technology that can drive away birds and animals from areas where bird and animal damage needs to be prevented, guide them to areas where they can live, and achieve a sustained repelling effect. [Means for solving the problem]

[0007] In order to solve the above-mentioned problems, the guidance system of the present invention is a guidance system comprising a plurality of intimidation / attraction operation units arranged in a plurality of areas respectively for intimidating / attracting birds and animals in each area, a plurality of detection units arranged in the plurality of areas respectively for detecting birds and animals in each area, an acquisition unit for acquiring position information indicating the position of a bird or animal when a bird or animal is detected by one of the plurality of detection units, a decision unit for determining one of the guidance patterns for operating each of the plurality of intimidation / attraction operation units based on the position information, and an operation control unit for operating each of the plurality of intimidation / attraction operation units in a intimidation / attraction mode based on the one guidance pattern.

[0008] In order to solve the above-mentioned problems, the guidance method of the present invention is a guidance method in which a management device connected to a plurality of intimidation / attraction operation units arranged in a plurality of areas respectively and intimidating / attracting birds and animals in each area, and a plurality of detection units arranged in the plurality of areas respectively and detecting birds and animals in each area, executes processing including: an acquisition step of acquiring position information indicating the position of a bird or animal when a bird or animal is detected by one of the plurality of detection units; a determination step of determining one of the guidance patterns for operating each of the plurality of intimidation / attraction operation units based on the position information; and an operation control step of operating each of the plurality of intimidation / attraction operation units in a intimidation / attraction mode based on the one guidance pattern.

[0009] In order to solve the above-mentioned problems, the program of the present invention is a program that causes a computer of a management device connected to a plurality of intimidation / attraction operation units that are placed in a plurality of areas and intimidate / attract birds and animals in each area, and a plurality of detection units that are placed in a plurality of areas and detect birds and animals in each area, to function as an acquisition unit that acquires location information indicating the location of a bird or animal when a bird or animal is detected by one of the plurality of detection units, a decision unit that determines one of the guidance patterns that operates each of the plurality of intimidation / attraction operation units based on the location information, and an operation control unit that operates each of the plurality of intimidation / attraction operation units in a intimidation / attraction mode based on the one guidance pattern.

[0010] In order to solve the above-mentioned problems, the estimation system of the present invention is an estimation system that includes a plurality of detection units that are placed in a plurality of areas inhabited by various animals and detect a target species in each area, an acquisition unit that, when the target species is detected by one of the plurality of detection units, acquires location information indicating the location of the target species, and an estimation unit that estimates a movement route of the target species based on the location information acquired by the acquisition unit. The above animals include creatures that cause bird and animal pests and endangered species, as well as amphibians such as frogs, reptiles such as snakes, insects such as butterflies, and fish such as black bass.

[0011] In order to solve the above-mentioned problems, the estimation method of the present invention is an estimation method in which a management device is placed in multiple areas inhabited by various animals and connected to multiple detection units that detect a target species in each area, and when the target species is detected by one of the multiple detection units, executes processing including: an acquisition process of acquiring location information indicating the location of the target species; and an estimation process of estimating the movement route of the target species based on the location information acquired in the acquisition process.

[0012] In order to solve the above-mentioned problems, the program of the present invention causes a computer of a management device, which is placed in multiple areas inhabited by various animals and connected to multiple detection units that detect target species in each area, to function as an acquisition unit that acquires location information indicating the location of the target species when the target species is detected by one of the multiple detection units, and an estimation unit that estimates the movement route of the target species based on the location information acquired by the acquisition unit. Effect of the Invention

[0013] According to the present invention, birds and animals can be driven out from areas where bird and animal damage is to be prevented, and guided to areas where the birds and animals can live, while a sustained repelling effect can be obtained. [Brief description of the drawings]

[0014] [Figure 1] 1 is an explanatory diagram showing a network configuration of a guide system according to a first embodiment of the present invention; [Diagram 2] FIG. 2 is a diagram illustrating an example of a hardware configuration of a guidance device. [Diagram 3] FIG. 2 illustrates an example of a hardware configuration of a management apparatus. [Figure 4] FIG. 2 is a block diagram showing an example of a functional configuration of a management device. [Diagram 5] FIG. 4 is a sequence diagram showing an operation of the guidance system. [Figure 6] This figure shows the action taken when a bird or animal enters a threat zone and is then guided to an attraction zone. [Figure 7] FIG. 1 is an explanatory diagram showing an example of a first learned model. [Figure 8] FIG. 11 is an explanatory diagram showing an example of a second learned model. [Figure 9] FIG. 2 is an explanatory diagram illustrating an example of a behavior model. [Figure 10] 13 is a flowchart showing the operation of the management device using a behavior model. [Figure 11] 13 is a flowchart showing a learning process of a value function performed by the management device. [Figure 12] FIG. 2 is an explanatory diagram showing a network configuration of a guide system according to a first modified example. [Figure 13] FIG. 11 is a block diagram showing an example of a functional configuration of an estimation system according to a second embodiment. [Figure 14] FIG. 11 is a diagram illustrating an example of a hardware configuration of a detection device according to a second embodiment. [Figure 15] FIG. 11 is a sequence diagram showing the operation of the animal estimation system of the second embodiment. [Figure 16] FIG. 11 is an explanatory diagram showing an example of a third trained model related to target species estimation. [Figure 17] FIG. 11 is an explanatory diagram showing an example of a fourth trained model related to target species estimation. [Figure 18] FIG. 11 is an explanatory diagram showing an example of a fifth learned model related to the estimation of a travel route performed by the management device 20. [Figure 19] FIG. 11 is a diagram showing an example of an estimation result of a usage area. [Figure 20] FIG. 13 is an explanatory diagram showing an example of a sixth trained model related to target species estimation. [Figure 21] FIG. 13 is an explanatory diagram showing an example of a seventh learned model related to estimation of a travel route. [Figure 22] FIG. 13 is an explanatory diagram showing an example of an eighth learned model related to estimation of a travel route. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0015] (Embodiment) (Network configuration of Guidance System 1) FIG. 1 is an explanatory diagram showing a network configuration of a guidance system 1 according to a first embodiment of the present invention. The guidance system 1 is introduced in places where damage caused by birds and animals is a concern, such as fields, national parks, and satoyama. The places where the guidance system 1 is introduced include a threatening zone 30a and an attracting zone 30b. The threatening zone 30a is an area where birds and animals are prevented from existing, such as an area where humans live or a specific national park. The birds and animals are, for example, creatures that cause bird and animal damage, and include animals such as deer, monkeys, wild boars, and bears, and birds such as crows, sparrows, and ducks. Note that the birds and animals are not limited to these. For example, when the guidance system 1 is used for the purpose of guiding protected birds and animals into a protected area other than the purpose of preventing bird and animal damage, the birds and animals include endangered species such as the Japanese crested ibis and the white stork. The threatening zone 30a includes a plurality of threatening areas 40a to 40i.

[0016] The attraction zone 30b is an area that attracts birds and animals from the intimidation zone 30a. The attraction zone 30b is an area where the presence of birds and animals is permitted and where birds and animals are likely to live. For example, the attraction zone 30b is a forest with little human activity and abundant food resources. The attraction zone 30b includes multiple attraction areas 40j to 40o.

[0017] In the drawing, the intimidation zone 30a and the attraction zone 30b are shown as areas targeted at specific birds and animals (e.g., deer) for the sake of convenience. The intimidation zone 30a and the attraction zone 30b can be different (variable) areas depending on the type of bird or animal. That is, a place that is an intimidation zone 30a for deer may not be an intimidation zone 30a for other birds and animals. Also, a place that is an attraction zone 30b for deer may not be an attraction zone 30b for other birds and animals.

[0018] The guidance system 1 includes a plurality of guidance devices 10 and a management device 20. Each device is connected by wire or wirelessly via a network 50 such as a local area network (LAN) or a wide area network (WAN). Each device is a computer device including a central processing unit (CPU), a read only memory (ROM), a random access memory (RAM), a communication unit, etc.

[0019] A plurality of guiding devices 10 are arranged in each area 40 of the intimidation zone 30a and the attraction zone 30b. The guiding device 10 arranged in the intimidation zone 30a scares off birds and animals, driving them out of the area 40 in which the device is arranged to another area 40. On the other hand, the guiding device 10 arranged in the attraction zone 30b attracts birds and animals, causing them to remain in the area 40 in which the device is arranged.

[0020] The guiding device 10 includes a detection unit 11 and a threatening / attracting operation unit 12. The detection unit 11 includes a camera that captures an image of an area 40 in which the device is placed, and detects birds and animals that have invaded the area 40 based on the image data captured by the camera. Instead of or in addition to detecting birds and animals using a camera, the detection unit 11 may detect sounds of birds and animals using a microphone or a microphone array, or may detect birds and animals using LiDAR (Light Detection And Ranging), a millimeter wave radar, an infrared sensor, or a thermal camera. When the detection unit 11 detects a bird or animal, the guiding device 10 transmits a notification (bird and animal intrusion notification) to the management device 20 that the bird or animal has invaded.

[0021] The intimidating / attracting operation unit 12 arranged in the intimidating zone 30a scares off birds and animals from the area 40 in which the device is arranged to another area 40. The intimidation is performed in a different intimidation manner for each bird and animal. The intimidation is performed, for example, by a combination of sound and light. Specifically, the intimidation manner includes the sound frequency, sound output interval, light color, light output interval, etc., as well as the sound output direction using a directional speaker and the irradiation direction using laser light. In other words, the intimidation manner is a combination of a predetermined sound manner and a predetermined light emission manner. The intimidation manner also includes a manner in which no sound or light is output.

[0022] The intimidation is not limited to sound or light, but may be vibration (e.g., vibration of wood or water), display of an image, throwing an object, spraying water, or a combination of these. In other words, the intimidation may be, for example, only sound or only light.

[0023] The intimidation / attraction operation unit 12 arranged in the attraction zone 30b outputs sound and light in an attracting manner (a manner that draws birds and animals) to attract birds and animals. For example, the attracting manner is the cry of the animal's own species to notify the presence of food. The attracting manner also includes a manner in which sound and light are not output. Note that the attraction is not limited to sound and light, but may be vibration (e.g., vibration of wood or water), display of an image, throwing an object (e.g., food), spraying water, or a combination of these, or one of these. That is, the attraction may be, for example, only sound or only light.

[0024] The management device 20 is a computer device such as a personal computer or a server. When a bird or animal is detected in the area 40 by the guiding device 10 in the intimidation zone 30a, the management device 20 determines a guiding pattern based on the location and type of the bird or animal, and transmits the determined guiding pattern to the multiple guiding devices 10. When the multiple guiding devices 10 receive the guiding pattern, they each cooperate to intimidate or attract based on the guiding pattern. This guides the bird or animal in the intimidation zone 30a to the attraction zone 30b.

[0025] (Configuration example of guidance device 10) 2 is a diagram showing an example of a hardware configuration of the guidance device 10. The guidance device 10 includes a CPU 101, a memory 102, a communication unit 103, a sound control unit 104, a light emission control unit 105, and a detection unit 11. Each unit is connected to each other via a bus 110 so as to be able to communicate with each other.

[0026] CPU 101 is a central processing unit, and controls the operation of guiding device 10 by reading and executing various programs stored in memory 102. The various programs include a guiding operation program according to the first embodiment. The guiding operation program includes a program for identifying the type of bird or animal, and a program for causing intimidation / attraction operation unit 12 to operate in a predetermined guiding mode (intimidation / attraction mode: intimidation mode or attraction mode). The memory 102 is a storage unit such as a ROM, a RAM, a hard disk, etc. The memory 102 stores various programs and other various information executed by the CPU 101. The memory 102 may also store (record) the imaging results (e.g., video and audio) captured by the detection unit 11. The communication unit 103 is an interface for transmitting and receiving information to and from other devices (eg, the management device 20).

[0027] The sound control unit 104 is included in the intimidation / attraction operation unit 12. The sound control unit 104 outputs sounds from the sound output unit. The sound control unit 104 can output a sound that repels birds and animals as a intimidation mode, and can output a sound that attracts birds and animals as an attraction mode. The sound control unit 104 can also control not only the output of sounds but also the direction in which the sounds are output.

[0028] The light emission control unit 105 is included in the intimidation / attraction operation unit 12. The light emission control unit 105 can cause the light emission unit to emit light in an emission mode that repels birds and animals as an intimidation mode, for example, by irradiating a laser light, flashing, or turning on an LED (Light Emitting Diode), and can cause the light emission unit to emit light in an emission mode that birds and animals prefer as an attraction mode. The light emission control unit 105 can also control not only the emission mode of the light emission unit, but also the direction in which the light is emitted. Note that a plurality of sound output units and light emission units may be arranged in the area 40 in which the device itself is arranged.

[0029] The detection unit 11 includes a charge coupled device (CCD) camera or a complementary metal oxide semiconductor (CMOS) camera. The detection unit 11 is not limited to being provided in the guidance device 10, and may be provided separately from the guidance device 10. A plurality of detection units 11 may be arranged in the area 40 in which the device is placed. In addition, when a device other than a camera, such as a microphone or a microphone array, is used as the detection unit 11, a plurality of these devices may also be arranged in the area 40 in which the device is placed.

[0030] (Example of configuration of management device 20) 3 is a diagram showing an example of a hardware configuration of the management device 20. The management device 20 includes a CPU 201, a memory 202, a communication unit 203, an input device 204, and an output device 205. Each unit is connected to each other via a bus 210 so as to be able to communicate with each other.

[0031] The CPU 201 is a central processing unit, and controls the operation of the management device 20 by reading and executing various programs stored in the memory 202. The various programs include a bird and animal guiding program according to the first embodiment. The bird and animal guiding program includes a program for determining an optimal guiding pattern for guiding birds and animals in the intimidation zone 30a to the attraction zone 30b using various models described later. The memory 202 is a ROM, a RAM, a hard disk, etc. The memory 202 stores various information executed by the CPU 201, including various programs.

[0032] The communication unit 203 is an interface for transmitting and receiving information to and from other devices (eg, the guide device 10). The input device 204 includes, for example, a keyboard, a mouse, a microphone, and the like, and is used to input various types of information. The output device 205 includes a display, a speaker, and the like, and outputs various types of information. When the management device 20 is realized by a server device, the management device 20 does not need to include the input device 204 and the output device 205.

[0033] (Functional configuration of management device 20) Fig. 4 is a block diagram showing an example of a functional configuration of management device 20. As shown in Fig. 4, management device 20 includes functional units, such as an acquisition unit 401, a direction estimation unit 402, a determination unit 403, an operation control unit 404, an intimidation / attraction mode estimation unit 405, and an intimidation / attraction mode generation unit 406. Each functional unit is realized by CPU 201. That is, each functional unit is realized by CPU 201 executing a bird and animal guiding program.

[0034] In the following, each functional unit shown in FIG. 4 will be described with reference to FIG. 5 and FIG. Fig. 5 is a sequence diagram showing the operation of the guidance system 1. Each process of each device shown in Fig. 5 is performed in the control cycle of each device. 6 is a diagram showing the operation when a bird or animal enters the intimidation zone 30a and is then guided to the attraction zone 30b. In the following, for the sake of simplicity, an example will be described in which a specific type of bird or animal (e.g., a deer) is guided to the attraction zone 30b. Also, the guiding devices 10 in the intimidation zone 30a are each assumed to operate in the same intimidating manner, and the guiding devices 10 in the attraction zone 30b are each assumed to operate in the same attracting manner (e.g., sounds that attract deer).

[0035] Step S501: The guidance device 10 in the intimidation zone 30a judges whether or not it has detected an intrusion of a bird or animal into the intimidation zone 30a based on the imaging data of the camera. If it has not detected an intrusion of a bird or animal (step S501: NO), the guidance device 10 proceeds to step S506.

[0036] Step S502: When the intrusion of a bird or animal is detected (Step S501: YES), the guidance device 10 transmits a bird or animal intrusion notification indicating the intrusion of the bird or animal to the management device 20. For example, as shown in FIG. 6, it is assumed that the guidance device 10 arranged in the intimidation area 40f detects a bird or animal 600. The guidance device 10 in the intimidation area 40f transmits a bird or animal intrusion notification to the management device 20. The bird or animal intrusion notification includes information indicating the type of the bird or animal 600 (for example, information indicating a deer), the position information of the bird or animal 600, and the direction from which the bird or animal 600 has intruded (intrusion direction). The position information of the bird or animal 600 is, for example, the position information of the guidance device 10 that detected the bird or animal 600. It is assumed that the position information of the guidance device 10 is stored in advance in the guidance device 10. In this way, the guidance device 10 has a function of detecting the intrusion of the bird or animal 600 and a function of determining the type of the bird or animal.

[0037] The function of detecting the bird or animal 600 and discriminating the type in step S501 may be provided in the management device 20. Specifically, the management device 20 may receive imaging data and position information of the guidance device 10 from the guidance device 10, and may identify the intrusion of the bird or animal 600 and the type of the bird or animal by performing image analysis on the received imaging data.

[0038] Step S503: The management device 20 judges whether or not a bird / animal intrusion notification has been received from the guide device 10. If the management device 20 has not received a bird / animal intrusion notification from the guide device 10 (step S503: NO), the management device 20 proceeds to step S510.

[0039] Step S504a: The management device 20 determines each zone 30 (threat zone 30a and attraction zone 30b) according to the type of bird or animal indicated by the bird or animal intrusion notification. That is, for a certain area 40, a certain bird or animal 600 may be included in the threatening zone 30a, while another bird or animal 600 may be included in the attraction zone 30b. For example, if the area 40 is an urban area, the area 40 is included in the attraction zone 30b because crows may be present, while it is included in the threat zone 30a because wild boars should not be present. The method of determining each zone 30 includes, for example, a method of determining using area boundaries for each type of bird or animal 600 set in the management device 20 by the user, and a method of the management device 20 automatically determining and determining the harvest status of agricultural crops using the camera of the detection unit 11.

[0040] Step S504b: The management device 20 predicts the direction in which the bird or animal 600 will move according to the intimidation / attraction manner and the guidance pattern for each of the intimidation areas 40a-40i and the attraction areas 40j-40o, based on the history of how the bird or animal 600 that has invaded the intimidation zone 30a in the past has progressed and on the geographic information (topography and land cover) of each area. Specifically, when the management device 20 receives a bird or animal intrusion notification from any of the guiding devices 10 (step S503: YES), the acquisition unit 401 acquires information indicating the type of the bird or animal 600, information indicating the position of the bird or animal 600, and information indicating the intrusion direction of the bird or animal 600. Then, the determination unit 403 determines one of the guidance patterns for operating each of the multiple intimidation / attraction operation units 12 for each bird or animal 600, based on the position information of the bird or animal 600 and the moving direction estimated by the direction estimation unit 402. Specifically, the determination unit 403 determines one guidance pattern using a value function described later. The guidance pattern includes the presence or absence of each operation and each intimidation / attraction mode for each of the multiple intimidation / attraction operation units 12. The guidance pattern shown in Fig. 6 shows a pattern in which, for example, the intimidation / attraction operation units 12 arranged in areas 40e and 40d are not made to perform intimidation, and the other intimidation / attraction operation units 12 are made to perform intimidation / attraction.

[0041] If the bird or animal 600 moves to the area 40e, the determination unit 403 determines, for example, a guidance pattern for making the guiding device 10 placed in the area 40e operate in a threatening manner. This allows the guiding device 10 to drive the bird or animal 600 that has invaded from the area 40f from the area 40e to the area 40d. Furthermore, if the bird or animal 600 moves to the area 40d, the determination unit 403 determines, for example, a guidance pattern for making the guiding device 10 placed in the area 40d operate in a threatening manner. This allows the guiding device 10 to drive the bird or animal 600 that has invaded from the area 40e from the area 40d to the area 40m (attraction zone 30b).

[0042] In learning the value function described later, the direction estimation unit 402 identifies the area 40 from the position information acquired by the acquisition unit 401. Then, the direction estimation unit 402 estimates the movement direction of the bird or animal 600 according to the intrusion direction of the bird or animal 600 and the intimidation / attraction mode (sound and light mode) for the identified area 40. Specifically, the direction estimation unit 402 estimates the movement direction of the bird or animal 600 using a second trained model 800 (FIG. 8) described later. For example, as shown in FIG. 6, the direction estimation unit 402 estimates that the bird or animal 600 moves from area 40f to area 40e. The intimidation / attraction mode used to estimate the movement direction is a mode with a high intimidation / attraction effect (guidance effect) estimated by the intimidation / attraction mode estimation unit 405. In addition, the intimidation / attraction mode estimation unit 405 estimates the intimidation / attraction mode with a high guiding effect using a first trained model 700 (FIG. 7) described later.

[0043] Step S505: The operation control unit 404 transmits a guidance instruction to each guiding device 10 based on the determined guidance pattern. When the bird or animal 600 invades the area 40e from the area 40f, the guiding device 10 arranged in the area 40e detects the bird or animal 600 and transmits a bird or animal intrusion notification to the management device 20. When the management device 20 receives the bird or animal intrusion notification, it performs the processes of steps S504a, S504b, and S505. That is, the management device 20 performs the processes of steps S504a, S504b, and S505 every time the bird or animal 600 is detected in each area 40 (every time the management device 20 receives a bird or animal intrusion notification). In this case, if the zones 30 have already been determined in step S504a for the type of the guidance target, the process of step S504a may be skipped.

[0044] The process of step S504b is not limited to being performed at the timing when the bird / animal intrusion notification is received. For example, the process may be performed at a predetermined cycle (e.g., 10 seconds) during the period when the bird / animal 600 is detected by any of the guiding devices 10 in the intimidation zone 30a.

[0045] Step S506: The guide device 10 judges whether or not a guide instruction has been received from the management device 20. If a guide instruction has not been received (step S506: NO), the guide device 10 proceeds to step S508.

[0046] Step S507: When a guidance instruction is received from the management device 20 (step S506: YES), the guidance device 10 performs a guidance operation (threatening operation and attracting operation) in an operation mode (threatening mode or attracting mode) according to the guidance instruction. As a result, each of the guidance devices 10 performs a guidance operation in cooperation with each other.

[0047] Step S508: The guidance device 10 judges whether it is the timing to transmit the guidance result. The guidance result indicates whether the guidance action (threatening action or attracting action) has been effective in guiding (whether the bird or animal 600 was guided or not), and also indicates the direction in which the bird or animal 600 has moved as a result of the guidance. For this purpose, the guidance device 10 estimates the moving direction of the bird or animal 600 using the detection unit 11. The timing to transmit the guidance result is a preset timing, and may be, for example, the timing when a predetermined time has elapsed since the start of the guidance action, or the timing when the bird or animal 600 has disappeared (has been driven out) from the area in which the device is placed. In addition, if the bird or animal 600 is still detected (if it has not been driven out) when the predetermined time has elapsed, the guidance device 10 transmits a guidance result indicating that it has not been guided. The guidance device 10 waits until it is the timing to transmit the guidance result (step S508: NO).

[0048] Step S509: If it is time to transmit the guidance result (step S508: YES / NO), the guidance device 10 transmits the guidance result to the management device 20.

[0049] Step S510: The management device 20 judges whether or not a guidance result has been received from the guidance device 10. If a guidance result has not been received (step S510: NO), the guidance device 10 ends the process and waits until the start of the next process.

[0050] Step S511: When the guidance result is received (step S510: YES), the management device 20 judges whether the guidance result indicates effectiveness. When the guidance result indicates effectiveness (step S511: YES), the management device 20 proceeds to step S514 and manages various data.

[0051] Step S512: If the guidance result indicates no effect (step S511: YES), that is, if the bird or animal 600 becomes accustomed to the current intimidation / attraction mode and is unable to guide the bird or animal 600, the management device 20 changes the guidance pattern. The guidance pattern to be changed is another intimidation / attraction mode with a high guidance effect estimated by the intimidation / attraction mode estimation unit 405. In addition, in step S512, the guidance pattern may be changed according to the route the bird or animal 600 has moved as a result of the change in the intimidation mode or the change in the attraction mode. Specifically, for example, the management device 20 may cause the direction estimation unit 402 to estimate the movement direction of the bird or animal 600 according to the changed intimidation / attraction mode, and may cause the determination unit 403 to determine one guidance pattern based on the movement direction. In addition, the management device 20 may suggest to the user a change in the geographical location of the guiding device 10.

[0052] Step S513: The operation control unit 404 transmits a change instruction to each guide device 10. Step S514: The management device 20 manages the various data, ends the process, and waits until the start of the next process. The various data includes information indicating the intimidation / attraction mode in which the intimidation / attraction was performed, information indicating the intimidation / attraction mode of each guiding device 10 as well as the guidance pattern, position information of the bird or animal 600, information indicating the intrusion direction, information indicating the actual movement (withdrawal) direction, guidance results, and other various information. The management of the various data includes, in addition to data accumulation, the provision of rewards in the behavior model 900, and the updating of parameters of the first trained model 700 and the second trained model 800.

[0053] Step S515: The guiding device 10 judges whether or not a change instruction has been received from the management device 20. If a change instruction has not been received (step S515: NO), the guiding device 10 ends the process and waits until the start of the next process.

[0054] Step S516: If a change instruction has been received (step S515: YES), the guide device 10 changes the guiding operation to an operation mode in accordance with the change instruction, ends the process, and waits until the start of the next process.

[0055] In this manner, the guidance system 1 can efficiently guide the bird or animal 600 that has entered the area 40f shown in FIG. 6, for example, through the areas 40e, 40d, and 40m in that order.

[0056] When the guidance of the bird or animal 600 to the attraction zone 30b is completed, the guidance device 10 ends the guidance operation. Specifically, when the detection unit 11 of the guidance device 10 in the attraction zone 30b detects the intrusion of the bird or animal 600 into the attraction zone 30b, the guidance device 10 transmits a completion notification indicating that the guidance of the bird or animal 600 is completed to the management device 20. When the management device 20 receives the completion notification, it transmits an end instruction to each guidance device 10. When each guidance device 10 receives the end instruction, it ends the guidance operation.

[0057] In the above description, the bird or animal 600 is considered to be a single individual, but in reality, the detection unit 11 may detect a plurality of individuals at the same time. For this reason, it is necessary to identify and match the individuals. The identification and matching of the individuals may be performed by the guidance device 10 through image recognition or the like, or may be estimated by the management device 20 based on the spatiotemporal distribution of the detected bird or animal.

[0058] (An example of the first trained model 700) FIG. 7 is an explanatory diagram showing an example of a first trained model 700. The first trained model 700 is generated for each bird and animal species. As shown in FIG. 7, the first trained model 700 is, for example, a model expressed using a neural network. However, the first trained model 700 and a second trained model 800 described later may be, for example, a deep neural network (DNN), a support vector machine, a random forest, or the like. The trained parameters include, for example, the number of layers of the neural network, the number of neurons in each layer, the connection relationships between the neurons, the weights (connection loads) of the connections between each neuron, and the thresholds of each neuron.

[0059] The first trained model 700 includes an input layer 701, one or more intermediate layers 702 (hidden layers), and an output layer 703. Each of the layers 701 to 703 includes one or more neurons. The number of layers in the intermediate layers 702, the number of neurons in each layer, and the like are set by trained parameters.

[0060] Information indicating a threatening / attractive mode is input to the input layer 701 as an input sample. The information indicating the threatening / attractive mode is a combination of sound information and light information. The sound information includes, for example, information indicating each of the frequency, output interval, duration, volume, and direction of the sound. The light information includes information indicating each of the color, output interval, duration, light intensity, and light direction.

[0061] The output layer 703 outputs an output value indicating "effective" and an output value indicating "ineffective" (the degree of certainty of each) as the effect of the induction (effect of intimidation / attraction). The effect of induction is, for example, "effective" when the intimidation / attraction action is performed in a target intimidating manner for a predetermined time and the bird or animal is successfully guided, and "ineffective" when the bird or animal cannot be guided. The effect of induction is not limited to "effective" or "ineffective", and can be, for example, a plurality of ranks according to the time required for induction. Specifically, the plurality of ranks can be, for example, rank 1 if the bird or animal can be guided within a predetermined time, rank 2 if the bird or animal can be guided after a predetermined time or more, and rank 3 if the bird or animal cannot be guided. The model output can be regressed as the number of seconds required for driving out, and the shortest number of seconds can be considered to be the most effective.

[0062] The first trained model 700 is a model generated by inputting an input sample to an input layer 701 and updating parameters so that the difference between each output value obtained from the output layer 703 and each output sample is reduced. The input sample is information (sound information and light information) indicating the threatening / attractive behavior input to the input layer 701 when training the learning model. The output sample is information (correct answer label) that serves as a correct answer for comparison with the output value from the output layer 703 when training the learning model, and specifically indicates "effective" or "ineffective." Note that the correct answer label may be the number of seconds of the action required until the expulsion, instead of or in addition to "effective" or "ineffective."

[0063] The input sample and the output sample are verified data obtained in advance by experiments or the like. The information indicating the intimidation / attraction mode may be generated by a user or may be randomly generated by the management device 20 or the like. For example, in generating the first trained model 700, the management device 20 causes each of the guidance devices 10 in which a bird or animal has been detected to perform an intimidation / attraction operation in a predetermined intimidation / attraction mode, and obtains the effect of guidance from each of the guidance devices 10. The management device 20 trains the learning model by using the input sample as the intimidation / attraction mode and the output sample as the effect of guidance, thereby generating the first trained model 700. The guidance system 1 according to the first embodiment includes a plurality of intimidation / attraction operation units 12, which can increase the opportunities to detect birds and animals, thereby enabling rapid learning.

[0064] Then, after the first trained model 700 is generated, the intimidation / attraction mode estimation unit 405, for example, inputs each of the multiple types of intimidation / attraction modes generated into the first trained model 700, and estimates the effect of guidance (effective, ineffective) for each intimidation / attraction mode. Then, the management device 20 selects the intimidation / attraction mode with the highest probability from among the intimidation modes estimated to be "effective," and causes the guidance device 10 to perform the intimidation / attraction operation. In this way, it is possible to obtain an intimidation / attraction mode with a high induction effect, even if the intimidation / attraction operation is not actually performed in an intimidation / attraction mode that has not been used before.

[0065] Here, when the bird or animal becomes accustomed to the intimidation / attraction mode, if the intimidation / attraction mode is similar to the previous one, the intimidation / attraction mode may not be effective in attracting the bird or animal. Therefore, the intimidation / attraction mode generating unit 406 (FIG. 4) generates a new intimidation / attraction mode having the characteristics of the multiple intimidation / attraction modes based on the multiple intimidation / attraction modes that have actually been recognized as having an effect of attracting the bird or animal. For example, when generating the intimidation / attraction mode, the intimidation / attraction mode generating unit 406 prepares a data set of variables that are recognized as having an effect of attracting the bird or animal in advance. The data set of variables is obtained, for example, by sampling those that have an effect of attracting the bird or animal based on past trial data. Specifically, the data set of variables is data based on a part of information (for example, information indicating frequency, output interval, duration, and volume) that is recognized as having an effect of attracting the bird or animal out of the information contained in the sound or light. The intimidation / attraction mode generating unit 406 generates the intimidation / attraction mode based on the data set of variables. For example, by randomly extracting the frequency, output interval, duration, and volume from multiple samples that have been effective in the past, and combining these parameters to create a sound, new data for a threatening / attractive pattern can be generated. This has the effect of inducing and can generate a new threatening pattern.

[0066] It should be noted that the generation of new intimidating / attracting patterns is not limited to this method. It is also possible to use a method in which a generative model (such as a variational autoencoder or an adversarial generative model) machine-learned from a separately prepared dataset of intimidating / attracting patterns is used to generate new intimidating / attracting patterns that are estimated to have an enticing effect using random numbers.

[0067] Furthermore, the intimidating / attracting mode estimation unit 405 may input the new intimidating / attracting mode generated by the intimidating / attracting mode generation unit 406 to the first trained model 700 and estimate the effect of the guidance (effective or ineffective). When the effect of the guidance decreases, the operation control unit 404 changes to a new intimidating / attracting mode and operates each of the multiple intimidating / attracting operation units 12. This allows the intimidating / attracting operation unit 12 to guide birds and animals in an effective intimidating / attracting mode in response to their habituation.

[0068] (An example of the second trained model 800) 8 is an explanatory diagram showing an example of a second trained model 800. The second trained model 800 is generated for each type of bird or animal. The second trained model 800 includes an input layer 801, one or more intermediate layers 802, and an output layer 803. Position information of a bird or animal, directional information of the bird or animal, and a threatening / attracting behavior are input to the input layer 801 as input samples.

[0069] The position information is, for example, position information of the guiding device 10 that detected the bird or animal. Note that the position information is not limited to this, and may be the coordinates on map data where the bird or animal is located. The intrusion direction information is, for example, information indicating the direction (for example, east, west, north, south, etc.) from which the bird or animal invaded the area 40. Note that the intrusion direction information is not limited to this, and may be information indicating the direction (head direction) in which the bird or animal is facing. The information indicating the intimidating / attracting manner is information (sound information and light information) indicating the intimidating / attracting manner that is highly effective for guidance, estimated by the intimidating / attracting manner estimation unit 405. In addition to the above information, the input sample may also include information indicating the terrain, such as the presence or absence and angle of slope, the presence or absence and size of vegetation, the presence or absence and width of roads, etc. The output layer 803 outputs output values ​​(accuracies of each) indicating the direction (for example, east, west, north, south, etc.) in which the bird or animal moves from the current area 40 to another area 40 and the moving speed.

[0070] The second trained model 800 is a model generated by inputting an input sample to an input layer 801 and updating parameters so as to reduce the difference between each output value obtained from the output layer 803 and each output sample. The input sample and the output sample are data based on past observation results observed by the guidance device 10. The input sample is data (location information, intrusion direction information, information indicating the mode of intimidation / attraction) input to the input layer 701 when training the learning model. The output sample is data (correct answer label) that is to be compared with the output value from the output layer 803 when training the learning model, and indicates, for example, any one of "moving direction 1", "moving direction 2", ... and "moving direction N".

[0071] The direction estimation unit 402 estimates the movement direction of the bird or animal within the area 40 by inputting the bird or animal's position information, the bird or animal's intrusion direction information, and information indicating the intimidation / attraction behavior into the second trained model 800.

[0072] (About the behavioral model) FIG. 9 is an explanatory diagram showing an example of a behavior model 900. The behavior model 900 is an abstract model for predicting the behavior of birds and animals, which is generated for each type of bird or animal. As shown in FIG. 9, the behavior model 900 includes an agent 901 and an environment 902. The agent 901 learns by using reinforcement learning. In reinforcement learning, an agent 901 in an environment 902 observes a current state 903 and determines an action 904 to be taken. Reinforcement learning is a method of learning a strategy that maximizes the amount of a final reward 905 obtained through a series of actions 904.

[0073] Specifically, the agent 901 executes an action 904 based on a state 903 of an environment 902. The environment 902 updates the state 903 according to the action 904 of the agent 901, and returns a reward 905 to the agent 901 as a result of the action 904. The agent 901 seeks a way to select an action 904 that maximizes the total reward. The agent 901 calculates an expected value of the reward 905 based on, for example, a state value function and an action value function, and determines the action 904.

[0074] In the first embodiment, the agent 901 is the guiding device 10. The environment 902 is represented by the position of the bird or animal (the intimidation area 40 in which the bird or animal exists) and the intrusion direction of the bird or animal. The state 903 is the moving direction of the bird or animal moving from the current area 40 to another area 40. The behavior 904 is the guiding pattern. For example, the reward 905 is given a higher value the shorter the distance between the position of the bird or animal in the intimidation zone 30a and the attraction zone 30b.

[0075] Fig. 10 is a flowchart showing the operation of the management device 20 using the behavior model 900. After executing the process related to the determination of the guiding pattern in step S504b of Fig. 5, the management device 20 executes the following process. First, the management device 20 acquires the current environment and stores the acquired environment in the memory 202 (step S1001). Next, the management device 20 calculates a reward for the previous behavior based on the environment acquired in step S1001 and the behavior (guiding pattern) at the previous control timing stored in the memory 202 (step S1002).

[0076] Next, the management device 20 generates a plurality of behavior candidates (guidance patterns) for the next control timing (step S1003). The plurality of behavior candidates are candidates for a guidance pattern in which a plurality of guidance devices 10 cooperate (link). The management device 20 selects the plurality of behavior candidates generated in step S1003 one by one, and executes the process of step S1005 for each candidate (step S1004). The management device 20 inputs the prepared simulation data and the behavior candidate selected in step S1004 into a value function stored in the memory 202, thereby calculating a value for the candidate (step S1005).

[0077] Then, the management device 20 determines, from among the multiple candidate behaviors, for example, the one with the highest value as the behavior (guidance pattern) to be applied to the control target (step S1006). The management device 20 outputs the behavior determined in step S1006 to the guiding device 10 (step S1007).

[0078] (Learning value functions) 10, the management device 20 learns the parameters of the value function in the behavior model 900 in advance. In this first embodiment, the value function is learned by simulating the intimidation and movement of birds and animals. In the simulation, the movement of birds and animals is simulated over a predetermined number of steps (P steps). 11 is a flowchart showing the learning process of the value function by the management device 20. The management device 20 randomly places birds and animals in a virtual space (step S1101). Then, the management device 20 generates candidate behaviors for the next control timing (step S1102). Furthermore, the management device 20 (direction estimation unit 402) substitutes the generated candidate behaviors into the second trained model 800 to estimate the movement direction of the birds and animals, thereby predicting the position and intrusion direction of the birds and animals at the next control timing (step S1103). Then, the management device 20 calculates a reward based on the position of the birds and animals after the movement (step S1104). For example, the reward is higher as the distance from the position of the birds and animals to the attraction zone 30b is shorter.

[0079] The management device 20 judges whether the candidate behavior generated in step S1102 is a candidate behavior related to a control timing after the P step (step S1105). If the candidate behavior generated is not a candidate behavior related to a control timing before the P step (step S1105: NO), the management device 20 returns the process to step S1102 and further predicts a state and a reward for the next control timing.

[0080] If the behavior candidate generated in step S1102 is a behavior candidate related to the control timing after the P step (step S1105: YES), the management device 20 calculates the sum of rewards in the P step (step S1106). Next, the management device 20 judges whether the number of attempts to generate behavior candidates for P steps is Q (Q is a natural number) or more (step S1107). If the number of attempts to generate behavior candidates for P steps is less than Q (step S1107: NO), the process returns to step S1101, and behavior candidates for P steps are generated again and rewards are predicted. If the number of attempts to generate behavior candidates for P steps is Q or more (step S1107: YES), the maximum of the sums of rewards calculated Q times in step S1105 is identified (step S1108). It can be said that the pattern of behavior candidates with the maximum sum of rewards is the pattern that can most efficiently guide birds and animals to the attraction zone 30b among the attempts.

[0081] The management device 20 learns the parameters of the value function using the bird and animal arrangement generated in step S1101 and the action candidates generated in step S1102 as input samples and the sum of the rewards identified in step S1108 as output samples (step S1109). The management device 20 determines whether or not a termination condition for the learning of the value function is satisfied (step S1110). The termination condition for the learning may be, for example, that the rate of change of the parameter is less than a threshold value or that the number of trials exceeds a predetermined number. If the termination condition for the learning of the value function is not satisfied (step S1110: NO), the process returns to step S1101 and the parameter update is repeatedly executed. On the other hand, if the termination condition for the learning of the value function is satisfied (step S1110: YES), the management device 20 stores the learned value function in the memory 202 and ends the process.

[0082] In the above-mentioned process, an example has been described in which a value function is learned using the results of a movement simulation performed by the direction estimation unit 402 in response to the intimidation of a bird or animal, and an action is determined using this value function, but the present invention is not limited to this. For example, it is also possible to execute a movement simulation performed by the direction estimation unit 402 in response to the intimidation of a bird or animal each time, without using the learned value function, and to find a pattern that maximizes the total reward.

[0083] In the above description, in order to simplify the description, an example of intimidation / attraction of one bird or animal of a specific type (e.g., a deer) has been described. However, the present guidance system 1 can determine a guidance pattern that provides an optimal intimidation / attraction mode for each zone 30 or area 40 and guide the bird or animal even when multiple birds or animals of different types invade multiple intimidation areas 40. For example, when a deer and a monkey invade a certain area 40, it is possible to intimidate / attract the deer for a predetermined period of time in an intimidation / attraction mode that is effective for the deer, and then intimidate / attract the monkey in an intimidation / attraction mode that is effective for the monkey.

[0084] In the above description, an example has been described in which the multiple guidance devices 10 have the same intimidation / attraction modes for each zone 30, but the present invention is not limited to this. The multiple guidance devices 10 may each have a different intimidation / attraction mode. Specifically, for example, the management device 20 may learn the parameters of the value function in the behavior model 900 using three different intimidation / attraction modes that are highly effective for guidance. This allows the management device 20 to determine a guidance pattern in which the multiple guidance devices 10 have different intimidation / attraction modes.

[0085] As described above, when a bird or animal is detected by one of the multiple guide devices 10, the guidance system 1 according to the first embodiment determines one guidance pattern based on the position information of the bird or animal, and operates the multiple intimidation / attraction operation units 12 in the guidance pattern. This makes it possible to effectively guide the bird or animal from the intimidation zone 30a to the attraction zone 30b. Therefore, it is possible to prevent the bird or animal from returning to the intimidation zone 30a, and therefore it is possible to preferably drive the bird or animal out of the intimidation zone 30a. This makes it possible to prevent bird or animal damage, and also to preferably keep protected birds and animals such as Japanese crested ibises and storks within the protected area, so it can be used for protecting the bird or animal. Therefore, by using the present guidance system 1, it becomes possible to perform zoning to guide and keep the bird or animal in an area that is not harmful to human activities. Therefore, it is possible to reduce bird or animal damage while allowing the bird or animal to live, and therefore it is possible to support the coexistence of the bird or animal and humans.

[0086] In the first embodiment, the guiding system 1 estimates the movement direction of the bird or animal according to the threatening / attracting behavior for the area specified from the location information of the bird or animal, and determines one guiding pattern based on the estimated movement direction. This makes it possible to more effectively guide the bird or animal, and therefore to more suitably drive the bird or animal out of each area 40 of the threatening zone 30a.

[0087] In the first embodiment, the guidance system 1 estimates the movement direction of the bird or animal based on the position information of the bird or animal, the information indicating the intrusion direction of the bird or animal, and the information indicating the threatening behavior. This improves the accuracy of estimating the movement direction of the bird or animal, and therefore allows the bird or animal to be more appropriately driven out of each area 40 of the intimidation zone 30a.

[0088] In the first embodiment, the guidance system 1 estimates the intimidating mode with the highest intimidating effect among a plurality of intimidating modes with different intimidating effects, and operates the intimidating / attracting operation unit 12 in the intimidating mode with the highest intimidating effect. This allows the intimidating operation to be performed in a different intimidating mode with the highest intimidating effect. This makes it possible to effectively drive away birds and animals from each area 40 of the intimidating zone 30a.

[0089] Furthermore, in the first embodiment, the guidance system 1 generates a new intimidating mode having characteristics of the multiple intimidating modes based on the multiple intimidating modes that have actually been found to have an effect of intimidation, changes the intimidating mode, and operates each of the multiple intimidating / attracting operation units 12. This makes it possible to perform a threatening operation in a different intimidating mode that has a higher intimidating effect, even if the bird or animal becomes accustomed to the threatening mode. This makes it possible to more effectively drive the bird or animal out of each area 40 of the intimidation zone 30a.

[0090] Furthermore, in the first embodiment, the guidance system 1 includes the intimidation / attraction operation unit 12 that attracts birds and animals, and causes the intimidation / attraction operation unit 12 to perform an attracting operation based on one determined guiding pattern. This makes it possible to more appropriately guide the birds and animals from the intimidation zone 30a to the attracting zone 30b, and to make the birds and animals stay in the attracting zone 30b. This makes it possible to prevent the birds and animals from returning to the intimidation zone 30a. This makes it possible to more appropriately guide the birds and animals from the intimidation zone 30a to the attracting zone 30b.

[0091] In the first embodiment, the guidance system 1 estimates the movement direction of the bird or animal based on the position information of the bird or animal, the information indicating the intrusion direction of the bird or animal, and the information indicating the threatening behavior of the bird or animal. This improves the accuracy of estimating the movement direction of the bird or animal, and therefore the bird or animal can be more suitably guided to the attraction zone 30b and can be made to stay in the attraction zone 30b.

[0092] In the first embodiment, the guiding system 1 estimates the attracting mode with the highest attracting effect among a plurality of attracting modes with different attracting effects, and operates the intimidation / attracting operation unit 12 in the estimated attracting mode with the highest attracting effect. This allows the attracting operation to be performed in another attracting mode with the highest attracting effect. This allows the bird or animal to be appropriately guided from the intimidation zone 30a to the attracting zone 30b, and allows the bird or animal to remain in the attracting zone 30b.

[0093] In the first embodiment, the guiding system 1 generates a new attracting mode having characteristics of the multiple attracting modes based on the multiple attracting modes that have been actually recognized to have an attracting effect, changes the attracting mode, and operates the multiple intimidation / attracting operation units 12. As a result, even if the bird or animal becomes accustomed to the attracting mode, the attracting operation can be performed in a different attracting mode that has a higher attracting effect. This makes it possible to more suitably keep the bird or animal in the attracting zone 30b.

[0094] (Modification of the first embodiment) Next, modified examples of the first embodiment will be described. In each of the following modified examples, the contents described in the above-mentioned first embodiment will be omitted as appropriate. In addition, each of the following modified examples and the above-mentioned first embodiment can be combined with each other.

[0095] (Variation 1) First, a first modified example of the first embodiment will be described. In the first modified example, an example in which the guidance system 1 is applied to a lotus field will be described. There are a plurality of areas within the lotus field 1200, and lotus roots are harvested in a mosaic pattern in each area. In other words, post-harvest areas and pre-harvest areas are mixed in a mosaic pattern. The pre-harvest areas are areas that may be subject to damage by birds and animals 600 such as ducks. On the other hand, the post-harvest areas are areas where small lotus roots remain that have not reached the point of being sold as products, and are areas in which the remaining lotus roots can be used as food for the birds and animals 600 without any problems.

[0096] 12 is an explanatory diagram showing a network configuration of the guidance system 1 according to the first modification. In the first modification, the lotus field 1200 includes an intimidation zone 30a and an attraction zone 30b. The intimidation zone 30a is an area where lotus roots are planted before harvest. The attraction zone 30b is an area where the lotus roots are harvested and where no damage occurs.

[0097] Guiding devices 10 are placed in the intimidation zone 30a and the attraction zone 30b, respectively. The guiding device 10 placed in the intimidation zone 30a scares off the birds and animals 600, driving them into the attraction zone 30b. On the other hand, the guiding device 10 placed in the attraction zone 30b emits sound and light to attract the birds and animals 600, causing them to stay there.

[0098] In the first modification, the intimidating / attracting mode of the intimidating / attracting operation unit 12 arranged in the intimidating zone 30a is, for example, irradiation of a laser light. Specifically, the intimidating / attracting operation unit 12 arranged in the intimidating zone 30a irradiates a laser light in a predetermined color and blinking mode toward the bird or animal 600. On the other hand, the intimidating / attracting operation unit 12 arranged in the attracting zone 30b does not irradiate a laser light, but instead outputs, for example, a cry of a bird or animal to notify the surrounding bird or animal 600 of the presence of food. The intimidating zone 30a and the attracting zone 30b may each be set by a user, or may be set automatically based on the detection result of the harvest status of the agricultural crops.

[0099] When the guiding device 10 in the intimidation zone 30a detects a bird or animal 600, it transmits a notification of the intrusion of the bird or animal 600 to the management device 20. The management device 20 determines a guidance pattern based on the position information of the guiding device 10 and transmits it to each guiding device 10. As a result, the multiple guiding devices 10 cooperate with each other to intimidate and attract based on the guidance pattern. Specifically, the guiding devices 10 in the intimidation zone 30a each intimidate in a predetermined intimidating manner. Also, the guiding devices 10 in the attraction zone 30b each attract in a predetermined attracting manner.

[0100] According to the first modification, the birds and animals 600 can be guided from the intimidation zone 30a to the attraction zone 30b where food for the birds and animals 600 is also present. This makes it possible to prevent the birds and animals 600 from returning to the intimidation zone 30a. This makes it possible to suitably drive the birds and animals 600 out of the intimidation zone 30a. Furthermore, unlike measures against bird and animal damage by extermination or the like, the birds and animals 600 can be appropriately protected.

[0101] (Variation 2) Next, a second modified example of the first embodiment will be described. In the second modified example, the management device 20 prepares a threatening / attracting mode that has been proven to have a high guiding effect, and performs a threatening / attracting operation in the threatening / attracting mode until the birds and animals become accustomed to the management device 20. In the second modified example, the proven threatening / attracting mode is a threatening / attracting mode that has actually been proven to have an guiding effect through experiments, etc. The management device 20 causes the guiding device 10 to threatening / attract in the threatening / attracting mode until the birds and animals become accustomed to the management device 20. When the birds and animals become accustomed to the management device 20, the management device 20 causes the guiding device 10 to threatening / attract in the threatening / attracting mode with high guiding effect estimated by the first trained model 700. Even in this way, optimal threatening / attracting can be performed, and therefore optimal guiding can be performed.

[0102] (Variation 3) Next, a third modified example of the first embodiment will be described. In the above-mentioned first embodiment, an example using various models (FIGS. 7 to 9) has been described. In the third modified example, an example in which a guidance pattern is determined without using various models will be described. In the third modified example, the guidance system 1 stores guidance data in which position information of birds and animals is associated with a guidance pattern. The guidance data may be stored in the management device 20 or in an external storage device. The guidance data is data prepared in advance and obtained from experimental results, past history, etc.

[0103] When the management device 20 receives a bird / animal intrusion notification from the guiding device 10, it determines a guiding pattern based on the location information of the bird / animal contained in the notification, and causes each guiding device 10 to perform a threatening / attracting operation according to the determined guiding pattern. This also makes it possible to appropriately guide the bird / animal from the threatening zone 30a to the attracting zone 30b.

[0104] The guidance data may be data in which the location information of the bird or animal, the intrusion direction information, and the guidance pattern are associated with each other. In this case, when the management device 20 receives a bird or animal intrusion notification from the guidance device 10, the management device 20 determines a guidance pattern based on the location information of the bird or animal and the intrusion direction information contained in the notification, and causes each guidance device 10 to perform a threatening / attracting operation according to the determined guidance pattern. In this way, the bird or animal can be more suitably guided from the intimidation zone 30a to the attraction zone 30b.

[0105] Second embodiment Next, a second embodiment will be described. Note that in the following second embodiment, the contents described in the above-mentioned first embodiment and the modified example of the first embodiment will not be described as appropriate.

[0106] Many animals do not stay in the same place, but use various places (hereinafter referred to as "usage areas") according to purposes such as foraging, resting, and movement. However, with conventional technology, it was sometimes impossible to efficiently grasp the actual usage of the usage area for each type of animal. As a result, it was sometimes impossible to effectively repel or attract each type of animal in an optimal manner according to the usage area. Therefore, in the second embodiment, it is possible to efficiently grasp the actual usage of the usage area for each type of animal.

[0107] In the above-mentioned first embodiment, the "birds and beasts" are described as creatures that cause bird and beast damage and endangered species, but they also include other animals. Examples of other animals include amphibians such as frogs, reptiles such as snakes, insects such as butterflies, and fish such as black bass. Target animals in the second embodiment described below include the above-mentioned birds and beasts and other animals. In the following, animals will be described as "birds and beasts".

[0108] (Functional configuration of the estimation system 1300 according to the second embodiment) In the second embodiment, an example will be described in which, after a bird or animal in the area 40 is detected by sound, the camera is started to detect the behavior of the bird or animal. Fig. 13 is a block diagram showing an example of a functional configuration of an estimation system 1300 according to the second embodiment. As shown in Fig. 13, the estimation system 1300 includes a management device 20 and a detection device 1310. The management device 20 includes functional units, an acquisition unit 401 and an estimation unit 1301. Each functional unit is realized by the CPU 201 (Fig. 3). That is, each functional unit is realized by the CPU 201 executing an estimation program.

[0109] In the second embodiment, the estimation system 1 includes a detection device 1310 instead of the guidance device 10 of the first embodiment. The detection device 1310 differs from the guidance device 10 in that it does not include the intimidation / attraction operation unit 12. The detection device 1310 is disposed in each of a plurality of areas 40 inhabited by various birds and beasts (animals). The detection device 1310 includes a detection unit 11. The detection unit 11 detects birds and beasts in each area. The detection unit 11 detects birds and beasts based on the detection results of a detector (for example, the microphone array 1404 and the camera 1405 shown in FIG. 14).

[0110] (Configuration example of detection device 1310) 14 is a diagram showing an example of a hardware configuration of a detection device 1310 according to the second embodiment. The detection device 1310 includes a CPU 1401, a memory 1402, a communication unit 1403, a microphone array 1404, and a camera 1405. Each unit is connected to each other via a bus 1410 so as to be able to communicate with each other.

[0111] The CPU 1401 is a central processing unit, and controls the operation of the detection device 1310 by reading and executing various programs stored in the memory 1402. The various programs include a detection program according to the second embodiment. The detection program includes a program for detecting the appearance of a bird or animal in the area 40, a program for identifying the type of the bird or animal, and a program for detecting the behavior of the bird or animal.

[0112] The memory 1402 is a storage unit such as a ROM, a RAM, a hard disk, etc. The memory 1402 stores various programs and various information executed by the CPU 1401. The memory 1402 may also store detection results (e.g., sound information and image information) of the microphone array 1404 and the camera 1405. The communication unit 1403 is an interface for transmitting and receiving information to and from other devices (eg, the management device 20).

[0113] The microphone array 1404 is a detector that detects the presence of birds and animals in the area 40. The microphone array 1404 is an example of a first detector. The microphone array 1404 detects sound. The microphone array 1404 also detects the direction of a sound source. Specifically, the microphone array 1404 uses the time difference between signals arriving at each microphone to emphasize or suppress a signal arriving from a specific direction, thereby estimating the direction from which the sound source arrives and separating the sound sources for each arrival direction. The first detector may be a microphone.

[0114] The detector for detecting the presence of birds and animals is not limited to the microphone array 1404. Details will be described later in a modified example, but for example, a microphone, a camera 1405, a LiDAR, a thermal sensor (thermal camera), an infrared sensor (infrared camera), a motion sensor (motion camera), or a millimeter wave radar can be used as a detector for detecting the presence of birds and animals instead of or in addition to these.

[0115] The camera 1405 is an example of a detector (second detector) that detects the behavior of a target species. The camera 1405 is, for example, a CCD or CMOS camera, and captures moving or still images. The imaging range of the camera 1405 is, for example, the range of the area 40 in which the device itself is placed.

[0116] The behavior of the target species to be detected may include any or all of the following: -Location of birds and animals within Area 40. -Distance to birds and animals. The direction from which the animal or bird came in (intrusion direction). The direction in which the animal retreated (retreat direction). -The movement speed of birds and animals.

[0117] Moreover, in the second embodiment, the second detector (detector that detects the behavior of the target species) includes a microphone array 1404. That is, the microphone array 1404 is capable of detecting the behavior of the target species in addition to detecting the presence of birds and animals in the area 40. With regard to the detection accuracy of behavior, the detection unit 11 can obtain a higher detection accuracy with the microphone array 1404 than with the microphone 140.

[0118] The detector for detecting the behavior of the target species is not limited to the microphone array 1404 and the camera 1405. Details will be described later in a modified example, but for example, a microphone, LiDAR, a thermal sensor (thermal camera), an infrared sensor (infrared camera), a motion sensor (motion camera), or a millimeter wave radar can be used as a detector for detecting the behavior of the target species instead of or in addition to these.

[0119] The detection device 1310 is equipped with, for example, a plurality of (for example, four) cameras 1405. The cameras 1405 are arranged facing four different directions (for example, north, south, east, and west). The detection device 1310 may be equipped with a single camera 1405. The detection device 1310 may also be equipped with a mechanism that enables the camera 1405 to rotate. The camera 1405 is equipped with a light source (flash). The light source is activated at night.

[0120] In the following, each functional unit shown in FIG. 13 will be described with reference to FIG. Fig. 15 is a sequence diagram showing the operation of the animal estimation system 1 of the second embodiment. Each process of each device shown in Fig. 15 is performed in the control cycle of each device.

[0121] (Processing of the detector 1310) Step S1501: Any of the detection devices 1310 (detection units 11) arranged in each of the multiple areas 40 determines whether or not a sound (a bird or animal cry) has been detected by the microphone array 1404. If the detection device 1310 does not detect a sound (step S1501: NO), it ends the process and waits until the start of the next process.

[0122] Step S1502: When the detection device 1310 detects a sound (step S1501: YES), the detection device 1310 uses the microphone array 1404 to estimate and separate sound sources in a plurality of directions, and estimates the direction of the sound source for each separated sound.

[0123] Step S1503: Based on the sound from the identified direction, the type of the bird or animal that has emitted the sound is estimated. Specifically, the detection device 1310 estimates the type of the bird or animal based on the sound information (call) and a third trained model 1600 (FIG. 16) described later.

[0124] Step S1504: Then, the detection device 1310 judges whether the type of bird or animal is a target species. The target species is a bird or animal to be tracked that is set in advance. The target species may be one type (for example, only deer) or multiple types (for example, deer and monkey). If the estimated bird or animal is not a target species (step S1504: NO), the detection device 1310 ends the process as it is and waits until the start of the next process.

[0125] Step S1505: On the other hand, if the estimated bird or animal is the target species (step S1504: YES), the detection device 1310 issues a start-up instruction to the camera 1405 facing the direction estimated in step S1504, and starts the camera 1405. If the detection device 1310 is provided with a mechanism for enabling the camera 1405 to rotate, the detection device 1310 may control the mechanism to point the camera 1405 in the direction estimated in step S1504. If the detection device 1310 is provided with a single camera 1405 and does not have a mechanism for enabling the camera to rotate, the detection device 1310 may issue a start-up instruction to the single camera 1405 without specifying the direction in which the target species exists (the process of step S1502).

[0126] Step S1506: The detection device 1310 judges whether it is nighttime or not. The detection device 1310 may judge whether it is a specific time period (nighttime period) set in advance based on time information obtained from a clock, as the judgment of whether it is nighttime or not. The clock may be, for example, an internal clock provided in the detection device 1310, or a GPS radio clock. In addition, if the detection device 1310 is provided with a light sensor (a sensor that detects the darkness (brightness) of the surroundings), the detection device 1310 may judge whether it is nighttime or not based on the detection result of the light sensor. If it is not nighttime in step S1506 (step S1506: NO), the detection device 1310 proceeds to step S1508.

[0127] Step S1507: On the other hand, if it is nighttime (step S1506: YES), the detection device 1310 turns on the light source included in the camera 1405. In this way, the light source is kept turned off during the day. This makes it possible to reduce power consumption related to the use of the light source. Note that if the camera 1405 is activated only during the day, that is, if the camera 1405 is not used at night, the light source may be turned off at all times. In this case, the detection device 1310 may not perform the processes of steps S1506 and S1507.

[0128] Step S1508: When the detection device 1310 activates the camera 1405, it obtains a detection result (hereinafter, sometimes referred to as a "behavior detection result") including sound information from the microphone array 1404 and image information from the camera 1405. Then, the detection device 1310 estimates the type of bird or animal based on the behavior detection result and a fourth trained model 1700 (FIG. 17) described later.

[0129] In addition, the estimation of the type in step S1508 may be more accurate than the estimation of the type in step S1503 because image information is used in addition to sound information. In addition, depending on the target species, the estimation based on sound information (estimation in step S1503) may be more accurate than the estimation of the type in step S1508. For this reason, it is assumed that the estimation results of the type estimation in step S1508 and the type estimation in step S1503 are different. In this case, either the type estimated in step S1503 or the type estimated in step S1508 may be prioritized as the estimation result depending on the target species. In addition, either the type estimated in step S1503 or the type estimated in step S1508 may be switched as the estimation result depending on the target species. In addition, if the estimation result in step S1508 is not the target species, the detection device 1310 may end the process as it is.

[0130] Step S1509: Next, the detection device 1310 judges whether or not it is time to transmit the location information of the target species and the behavior detection result. The location information of the target species is the location information of the detection device 1310 that detected the target species. It is assumed that the detection device 1310 stores the location information in advance.

[0131] The transmission timing is, for example, the timing when the target species disappears from the imaging range of the camera 1405. However, the transmission timing is not limited to this, and may be the lapse of a predetermined time from the start of the activation of the camera 1405. The detection device 1310 waits until the transmission timing arrives (step S1507: NO).

[0132] Step S1510: When the transmission timing arrives (step S1507: YES), the detection device 1310 transmits the position information of the target species and the behavior detection result to the management device 20. Furthermore, when the transmission timing arrives, the detection device 1310 turns off the power of the camera 1405.

[0133] Note that the function shown in step S1503 (the function of determining whether or not it is a target species) is not limited to being provided in the detection device 1310, and may be provided in the management device 20. Specifically, the management device 20 may receive sound information from the detection device 1310 and detect the type of bird or animal based on the received sound information. When the management device 20 detects a target species, it issues an activation instruction to the camera 1405 that is directed in a predetermined direction in the area 40. That is, in this case, the camera 1405 will be activated based on the activation instruction from the management device 20.

[0134] (Processing of management device 20) Step S1511: The management device 20 (acquisition unit 401) determines whether or not it has acquired the location information and the behavior detection result transmitted from the detection device 1310. If it has not acquired these (step S1511: NO), the management device 20 ends the process and waits until the start of the next process.

[0135] Step S1512: When the acquiring unit 401 acquires the location information and the behavior detection result, the estimating unit 1301 estimates a movement route of the target species based on the location information, the behavior detection result, and a fifth trained model 1800 (FIG. 18) described later. The movement route is a route including an area 40 where the target species is detected and an area 40 where the target species finally arrives.

[0136] Step S1513: The estimation unit 1301 estimates the use area where each target species is most likely to live, based on the estimated migration route. For example, various estimation methods can be used to estimate the use area. As an example, a Bayesian statistical method is used for the estimation. In Bayesian statistics, the use area can be estimated based on the posterior probability of the habitat, which is calculated by combining the prior probability obtained from prior data and the likelihood updated by new data (information on the migration route).

[0137] In this way, by estimating the utilization area, the management device 20 can quantify in which areas 40 the target species is distributed and to what extent within the areas 40 inhabited by various animals. Specifically, the management device 20 can estimate the population density of the target species and map the areas 40 in which the target species is likely to exist depending on the time of day, season, etc.

[0138] (Display on user terminal) The estimated results of the travel route and the utilization area can be displayed on a predetermined terminal (hereinafter referred to as "user terminal"). Specifically, in response to a transmission request from a user terminal that can access the management device 20, the management device 20 transmits the estimated results of the travel route and the utilization area to the requesting user terminal. When the user terminal receives the estimated results from the management device 20, it displays the estimated results. This allows the user to understand the estimated results of the travel route and the utilization area.

[0139] (An example of a trained model) Next, the trained model will be described with reference to Figs. 16 to 18. Each trained model described below indicates a model represented by using a neural network. In each trained model, the trained parameters include, for example, the number of layers of the neural network, the number of neurons in each layer, the connection relationship between the neurons, the weights (connection loads) of the connections between the neurons, and the thresholds of each neuron. Each trained model includes an input layer, one or more intermediate layers (hidden layers), and an output layer. Each layer includes one or more neurons. The number of intermediate layers, the number of neurons in each layer, and the like are set by the trained parameters. The trained model may be a DNN (Deep Neural Network), a support vector machine, a random forest, or the like.

[0140] (Third trained model: Estimation of species based on sound information (for camera activation)) 16 is an explanatory diagram showing an example of a third trained model 1600 related to target species estimation in the second embodiment. As shown in FIG. 16, the third trained model 1600 includes an input layer 1601, one or more intermediate layers 1602 (hidden layers), and an output layer 1603.

[0141] Sound information is input as an input sample to the input layer 1601. The sound information shown in Fig. 16 is information for activating the camera 1405. The sound information includes, for example, information indicating each of the frequency, output interval, duration, and volume. Note that the sound information as an input sample includes information when multiple sounds overlap, and information when there is only background sound and no sound of the detection target species.

[0142] The output layer 1603 outputs an output value (respective accuracy) indicating the type of bird or animal. The third trained model 1600 is a model generated by inputting an input sample to the input layer 1601 and updating parameters so that the difference between each output value obtained from the output layer 1603 and each output sample is reduced. The input sample is sound information input to the input layer 1601 when training the learning model. The output sample is information (correct answer label) that is the correct answer to be compared with the output value from the output layer 1603 when training the learning model. Specifically, the output sample indicates, for example, any one or more types from "Type 1 (deer)", "Type 2 (monkey)", ..., "Type N (crow)", or indicates that none of the types are included. Note that the parentheses are merely examples.

[0143] When a sound is detected by the microphone array 1404, the detection device 1310 inputs information about the detected sound to the third trained model 1600. In this way, the detection device 1310 estimates the type of bird or animal present in the area 40 in which the detection device 1310 is placed.

[0144] (4th trained model: type estimation based on behavior detection results) 17 is a diagram illustrating an example of a fourth trained model 1700 related to target species estimation. As shown in FIG. 17, the fourth trained model 1700 includes an input layer 1701, one or more intermediate layers 1702 (hidden layers), and an output layer 1703.

[0145] Sound information and image information as behavior detection results are input to the input layer 1701 as input samples. The sound information is information indicating the cries of various birds and animals. The sound information includes, for example, information indicating frequency, output interval, duration, and volume. The image information is, for example, pixel data. Note that the image information as an input sample includes information when multiple species are captured on camera at the same time.

[0146] The output layer 1703 outputs an output value (respective accuracy) indicating the type of the target species. The fourth trained model 1700 is a model generated by inputting an input sample to the input layer 1701 and updating parameters so that the difference between each output value obtained from the output layer 1703 and each output sample is reduced. The input sample is sound information and image information input to the input layer 1701 when training the learning model. The output sample is information (correct answer label) that is the correct answer to be compared with the output value from the output layer 1703 when training the learning model. Specifically, the output sample indicates, for example, any one or multiple types of "Type 1 (deer)", "Type 2 (monkey)", ..., "Type N (crow)", or indicates that none of the types are included. Note that the parentheses are merely examples.

[0147] When the detection device 1310 activates the camera 1405, the detection device 1310 inputs sound information detected by the microphone array 1404 and image information detected by the camera 1405 to the fourth trained model 1700. This enables the detection device 1310 to more accurately estimate the type of bird or animal present in the area 40 in which the detection device 1310 is placed.

[0148] (Fifth trained model: Estimation of travel route based on location information, sound information, and image information) 18 is an explanatory diagram showing an example of a fifth trained model 1800 related to the estimation of a movement route performed by the management device 20. The fifth trained model 1800 is generated for each bird and animal type. As shown in FIG. 18, the fifth trained model 1800 includes an input layer 1801, one or more intermediate layers 1802 (hidden layers), and an output layer 1803.

[0149] Position information, sound information as a behavior detection result, and image information as a behavior detection result are input as input samples to the input layer 1801. The position information is the position information of the detection device 1310 that detected the target species.

[0150] The output layer 1803 outputs an output value (respective accuracy) indicating the movement route of the target species. The fifth trained model 1800 is a model generated by inputting an input sample to the input layer 1801 and updating parameters so that the difference between each output value obtained from the output layer 1803 and each output sample is reduced. The input sample is position information, sound information, and image information input to the input layer 1801 when training the learning model. The output sample is information (correct answer label) that becomes a correct answer to be compared with the output value from the output layer 1803 when training the learning model. Specifically, the output sample indicates, for example, any one of "movement route 1 (eastward)", "movement route 2 (westward)", ... and "movement route N (northward)". Note that the parentheses are examples.

[0151] When the acquisition unit 401 acquires the position information, sound information, and image information, the management device 20 inputs these to the fifth trained model 1800. In this way, the management device 20 estimates the movement route of the target species.

[0152] (An example of an estimated usage area) Fig. 19 is a diagram showing an example of an estimation result of a usage area. In Fig. 19, an estimation result 1900 includes "type", "usage area", "period", and "individual density".

[0153] "Type" indicates the type of bird or animal. The "usage area" refers to the area 40 where various birds and animals ultimately arrive. The "period" indicates, for example, an hourly time slot, but the "period" may also be a seasonal or monthly time period. "Individual density" indicates the density (number of individuals per given area) for each utilization area.

[0154] In this way, the management device 20 can quantify the use areas for each type and the activities for each type for each period. The estimation result 1900 is not limited to being represented as a table. For example, it is also possible to graphically display the population density at a position corresponding to each area 40 on the map based on the estimation result 1900. The graphical display may be, for example, a display in a color corresponding to the population density or a display of a graph showing the population density. The illustrated estimation result may also be displayed as a video that transitions according to a time series. By graphically displaying the estimation result 1900 in this way, the use areas for each type and the activities for each type can be visualized. Therefore, the use areas for each type and the activities for each type can be more easily understood.

[0155] As described above, when a target species is detected by one of the multiple detection units 11, the estimation system 1300 according to the second embodiment estimates the movement path of the target species based on the location information indicating the location of the target species. This allows the route from the area 40 in which the target species was detected to the area 40 to which the target species will ultimately be reached to be estimated with high accuracy. Therefore, the use area in which the target species is most likely to live can be estimated with high accuracy. Therefore, the user can efficiently grasp the actual usage of the movement path of the target species and the actual usage of the use area. This allows the use area of ​​each target species to be set as the attraction zone 30b when repelling or attracting. Therefore, more optimal repelling or attraction can be performed for each target species.

[0156] Furthermore, the estimation system 1300 according to the second embodiment detects the behavior of the target species, and estimates the movement route of the target species based on the behavior detection result in addition to the position information. This makes it possible to more accurately estimate the route from the area 40 where the target species is detected to the area 40 where the target species ultimately arrives.

[0157] Furthermore, the estimation system 1300 according to the second embodiment detects a target species based on the detection result of the microphone array 1404 (first detector), and detects the behavior of the target species based on the detection result of the camera 1405 (second detector). This makes it possible to use the first detector suitable for detecting animals and the second detector suitable for detecting behavior.

[0158] For example, a device with low power consumption can be used as the first detector, and a device with high power consumption can be used as the second detector. More specifically, for detecting birds and animals, a device with low power consumption even when always turned on, such as the microphone array 1404, can be used. Then, when the detection device 1310 detects a target species based on the detection result of the microphone array 1404, it activates the camera 1405. This makes it unnecessary to keep the camera 1405 on all the time, and therefore makes it possible to reduce the power consumption of the camera 1405 (battery). Therefore, in the estimation system 1300, an optimal device can be applied to each detector.

[0159] (Modification of the second embodiment) Next, modified examples of the second embodiment will be described. The numbers of the following modified examples will be explained as numbers (modified examples 4 onwards) following modified examples 1 to 3 of the first embodiment described above. Each of the following modified examples can be combined with any of the above-mentioned first embodiment, each of the modified examples of the first embodiment, and the above-mentioned second embodiment.

[0160] (Variation 4) In the second embodiment, an example has been described in which the camera 1405 is started, and when the target species is no longer present, the power supply of the camera 1405 is turned off. In the fourth modified example, an example in which the power supply of the camera 1405 is always on will be described. For example, in a case where the power supply environment is good, such as when the power supply can be obtained directly or when the battery capacity is large, the power supply of the camera 1405 may be always on without being switched on / off.

[0161] 15. More specifically, when the detection device 1310 detects a sound by the microphone array 1404 (step S1501: YES) or detects an animal by the camera 1405, the detection device 1310 may estimate the type of bird or animal based on the detection results of the microphone array 1404 and the camera 1405 (step S1508). If the estimated type is the target species, when it is time to transmit (step S1509: YES), the detection device 1310 may transmit the position information and the behavior detection result to the management device 20.

[0162] According to the fourth modification, the camera 1405 is always ON, which eliminates the need for detection and control related to the start-up of the camera 1405. Therefore, according to the fourth modification, the behavior of the target species can be detected with simple control.

[0163] (Variation 5) Next, Modification 5 will be described. In the second embodiment, an example was described in which the microphone array 1404 and the camera 1405 are used as detectors for detecting the presence of birds and animals in the area 40 and the behavior of the target species. In Modification 5, an example is described in which another detector is used instead of or in addition to these as a detector for detecting the presence of birds and animals and the behavior of the target species. In the following, the detector in Modification 5 is referred to as a "behavior detector."

[0164] The behavior detector according to the fifth modification is one or more (L types) of a plurality of types, such as the microphone array 1404, the microphone, the camera 1405, LiDAR, an infrared sensor, an infrared camera, a motion sensor (motion camera), and a millimeter wave radar. From the viewpoint of improving the estimation accuracy of the target species and the estimation accuracy of the movement route, it is desirable to use a large number of types of behavior detectors. In the fifth modification, L is a plurality.

[0165] Each of the multiple types of behavior detectors is capable of detecting the behavior of a target species. Also, each of the multiple types of detectors is capable of detecting the presence of birds and animals within the area 40.

[0166] Also, in the fifth modification, one of the detectors (always-on detector) may be configured to detect the presence of a bird or animal, and in response to the detection of the presence, another detector (a detector capable of being switched on / off) may be activated. The always-on detector and the detector capable of being switched on / off may be set according to their respective power consumption. Specifically, the always-on detector may be a detector with low power consumption (for example, the microphone array 1404 or an infrared sensor). Moreover, the detector capable of being switched on / off may be a detector with high power consumption (for example, the camera 1405 or a LiDAR).

[0167] (6th trained model: Classification estimation based on location information and multiple behavior detection results) 20 is an explanatory diagram showing an example of a sixth trained model 2000 related to target species estimation (variation 5). As shown in FIG. 20, the sixth trained model 2000 includes an input layer 2001, one or more intermediate layers 2002 (hidden layers), and an output layer 2003.

[0168] Position information and behavior detection results 1 to L are input to the input layer 2001 as input samples. The position information is the position information of the detection device 1310 that detected the target species. L is the number of types of behavior detectors. The behavior detection results 1 to L are detected by each of multiple types (L types) of behavior detectors.

[0169] The output layer 2003 outputs an output value (respective accuracy) indicating the type of the target species. The sixth trained model 2000 is a model generated by inputting an input sample to the input layer 2001 and updating parameters so as to reduce the difference between each output value obtained from the output layer 2003 and each output sample. The input sample is position information and behavior detection results input to the input layer 2001 when training the learning model. The output sample is information (correct answer label) that is the correct answer to be compared with the output value from the output layer 2003 when training the learning model. Specifically, the output sample indicates, for example, one or more types from "Type 1 (deer)", "Type 2 (monkey)", ..., "Type N (crow)", or indicates that none of the types are included.

[0170] When the detection device 1310 detects a target species, it inputs the location information and the behavior detection results detected by the multiple types of behavior detectors to the sixth trained model 2000. In this way, the detection device 1310 estimates the type of bird or animal present in the area 40 in which the detection device 1310 is placed.

[0171] (7th trained model: Estimation of movement path based on location information and multiple behavior detection results) 21 is an explanatory diagram showing an example of a seventh trained model 2100 related to estimation of a movement path. The seventh trained model 2100 is generated for each bird and animal type. As shown in FIG. 21, the seventh trained model 2100 includes an input layer 2101, one or more intermediate layers 2102 (hidden layers), and an output layer 2103.

[0172] Position information and behavior detection results 1 to L are input to the input layer 2101 as input samples. The position information is the position information of the detection device 1310 that detected the target species. L is the number of types of behavior detectors. The behavior detection results 1 to L are detected by each of multiple types (L types) of behavior detectors.

[0173] The output layer 2103 outputs an output value (respective accuracy) indicating the movement route of the target species. The seventh trained model 2100 is a model generated by inputting an input sample to the input layer 2101 and updating parameters so as to reduce the difference between each output value obtained from the output layer 2103 and each output sample. The input sample is the position information and behavior detection result input to the input layer 2101 when training the learning model. The output sample is information (correct answer label) that becomes the correct answer to be compared with the output value from the output layer 2103 when training the learning model. Specifically, the output sample indicates, for example, any one of "movement route 1 (eastward)", "movement route 2 (westward)", ... and "movement route N (northward)".

[0174] When the acquisition unit 401 acquires the position information and the behavior detection result, the management device 20 inputs the position information and the behavior detection result to the seventh trained model 2100. In this way, the management device 20 estimates the movement route of the target species.

[0175] According to the fifth modification, the presence of birds and animals, the discrimination of the species, and the behavior of the target species are detected based on the detection results of the multiple types of behavior detectors, so that the detection accuracy can be improved. Therefore, the route from the area 40 where the target species is detected to the final destination area 40 can be estimated with higher accuracy.

[0176] (Variation 6) Next, a sixth modification will be described. In the sixth modification, an example of changing the specifications (addition, deletion, version upgrade, etc.) of the behavior detector expected in actual operation will be described. In the sixth modification, when there is a change in the specifications of the behavior detector, the estimation system 1300 performs various estimations using another trained model (a trained model corresponding to the sixth trained model 2000 or a trained model corresponding to the seventh trained model 2100) corresponding to the change in the specifications. The other trained model is a trained model generated separately. For example, when the camera 1405 is deleted and a LiDAR is added, the other trained model is a trained model generated by inputting the behavior detection result of the LiDAR as an input sample, except for the behavior detection result of the camera 1405. When there is a change in the specifications of the behavior detector, the other trained model is installed in the detection device 1310.

[0177] Specifically, when the specifications of the behavior detector are changed, the detection device 1310 acquires (installs) a trained model corresponding to the changed behavior detector. More specifically, when the specifications of the behavior detector are changed, the detection device 1310 acquires information indicating a combination of types of behavior detectors (e.g., camera 1405, LiDAR) whose specifications have been changed, and makes a request to the management device 20 to transmit a trained model corresponding to the combination.

[0178] The transmission may be performed via a line such as a wireless LAN connected via a USB (Universal Serial Bus). The detection device 1310 can install an appropriate trained model according to the changed combination of behavior detectors by receiving the trained model from the management device 20. Note that the installation is not limited to being performed by communication, and may be performed by reading from a storage medium such as a USB flash memory, for example.

[0179] In addition, the resolution and range of the behavior detector with the modified specifications may differ from the resolution and range of the behavior detector used in generating the trained model. Specifically, for example, if the behavior detector is a camera 1405, the number of pixels of the image data detected by the modified camera 1405 may differ from the number of pixels of the image data used in the input sample.

[0180] Therefore, the detection device 1310 corrects data (e.g., image data) detected by the behavior detector used so that the data becomes an index similar to that of the input sample. Specifically, the detection device 1310 may correct the captured image to image data similar to that of the input sample by, for example, enlarging or reducing the captured image to a predetermined number of pixels or by cropping the image, and input the image data to the trained model. This allows the type of bird or animal to be appropriately estimated and the movement route to be estimated even if the resolution and range of the behavior detector with the modified specifications are different from those of the behavior detector used in generating the trained model.

[0181] In addition, a trained model using all types of behavior detectors expected to be used may be generated in advance. For example, when a specific behavior detector (e.g., LiDAR) is no longer used due to a change in specifications, various estimations may be performed without inputting the behavior detection results detected by the LiDAR in the trained model.

[0182] According to the sixth modification, even if a change occurs in the specifications of the behavior detector of the estimation system 1300, various estimations (type and movement route) can be performed in response to the specification change.

[0183] (Variation 7) Next, Modification 7 will be described. In Modification 7, an example in which the intimidation / invitation operation unit 12 is retrofitted to the detection device 1310 will be described. In Modification 7, the intimidation / invitation operation unit 12 is retrofitted to each of the multiple detection devices 1310. When the intimidation / invitation operation unit 12 is retrofitted, the detection device 1310 installs the guiding operation program shown in the first embodiment. The installation may be performed by communication or by reading from a storage medium such as a USB flash memory.

[0184] When attracting a target species, if the utilization area obtained from the estimation result of the movement route shown in the second embodiment is an area in which the presence of the target species is permitted, the utilization area can be set as the attraction zone 30b. In this way, when a target species is detected in the area 40, the management device 20 can determine an optimal guidance pattern to the utilization area (attraction zone 30b) based on the position information of the target species. Therefore, it is possible to effectively attract and repel the target species to the utilization area.

[0185] (Variation 8) Next, Modification 8 will be described. In Modification 8, an example will be described in which the function of the intimidating / attracting operation unit 12 included in the guiding device 10 shown in the first embodiment can be switched ON / OFF. In Modification 8, when the function of the intimidating / attracting operation unit 12 is turned OFF, the guiding device 10 has the function of only the detection unit 11, that is, the same function as that of the detection device 1310. In other words, in Modification 8, the guiding device 10 can be used as the detection device 1310 by turning OFF the function of the intimidating / attracting operation unit 12. On the other hand, by turning ON the function of the intimidating / attracting operation unit 12, it is possible to intimidate / attract the target species.

[0186] According to such a configuration, for example, when the area of ​​use of a bird or animal is not clear, the guidance device 10 can estimate the area of ​​use by turning off the function of the intimidation / attraction operation unit 12. Then, when the estimation of the area of ​​use of the target species is completed, the guidance device 10 can effectively attract and chase away the bird or animal to the area of ​​use by setting the area of ​​use as the attraction zone 30b.

[0187] In addition, the function of the intimidation / attraction operation unit 12 may be periodically turned off. This allows the areas used by various animals according to the season and time of day to be periodically grasped, and the effectiveness of the intimidation / attraction can be confirmed.

[0188] (Variation 9) Next, a description will be given of Modification 9. In the above-described second embodiment, an example has been described in which the management device 20 estimates the movement path of the target species based on position information and a behavior detection result. In Modification 9, an example will be described in which the management device 20 estimates the movement path of the target species based on position information without using a behavior detection result.

[0189] In the ninth modification, the detection unit 11 detects a target species in the area 40 in which the device is placed, based on the detection result of the detector. The detector may be any device capable of obtaining detection results used to distinguish the species of birds and animals. The detector may be, for example, one or more of the microphone array 1404, a microphone, a camera 1405, a LiDAR, a thermal sensor, an infrared sensor, a motion sensor, a millimeter wave radar, and the like.

[0190] In the ninth modification, when a bird or animal is detected in the area 40 by the detector, the detection device 1310 estimates the type of the bird or animal based on the detection result of the detector (step S1508 in FIG. 15). If the estimated type is the target species, when the transmission timing arrives (step S1509: YES), the detection device 1310 transmits the position information of the area 40 in which the detection device is located to the management device 20 (step S1510).

[0191] In other words, the detection device 1310 according to the 9th modification may not perform the processes of steps S1502 to S1507. When the management device 20 acquires the location information (step S1511: YES), the management device 20 estimates the movement route of the target species based on the location information and an eighth trained model 2200 (FIG. 22) described later (step S1512).

[0192] (8th trained model: Estimation of movement path based on location information) 22 is an explanatory diagram showing an example of an eighth trained model 2200 related to estimation of a movement route. The eighth trained model 2200 is generated for each bird and animal type. As shown in FIG. 22, the eighth trained model 2200 includes an input layer 2201, one or more intermediate layers 2202 (hidden layers), and an output layer 2203.

[0193] Position information is input as an input sample to the input layer 2201. The position information is position information of the detection device 1310 that detected the target species.

[0194] The output layer 2203 outputs an output value (respective accuracy) indicating the movement route of the target species. The eighth trained model 2200 is a model generated by inputting an input sample to the input layer 2201 and updating parameters so as to reduce the difference between each output value obtained from the output layer 2203 and each output sample. The input sample is position information input to the input layer 2201 when training the learning model. The output sample is information (correct answer label) that becomes the correct answer to be compared with the output value from the output layer 2203 when training the learning model. Specifically, the output sample indicates, for example, any one of "movement route 1 (eastward)", "movement route 2 (westward)", ... and "movement route N (northward)".

[0195] When the acquisition unit 401 acquires the location information, the management device 20 inputs the location information to the eighth trained model 2200. As a result, the management device 20 estimates the movement route of the target species. Note that in the ninth modification, the input information to the input layer 2201 may include, in addition to the location information, the number of times the target species appears and the duration of stay in the area 40. In this way, the movement route can be estimated with higher accuracy.

[0196] According to the 9th modification, the movement path of the target species to the area 40 where the target species finally arrives is estimated based on the position information of the area 40 where the target species was detected, not based on the behavior detection result. This makes it possible to estimate the movement path with a simple configuration and simple control.

[0197] The programs for implementing the above-described guidance system 1, guidance device 10, and management device 20 may be recorded in a computer-readable recording medium, and the programs may be read into a computer system and executed. The term "computer system" as used herein includes hardware such as an OS and peripheral devices. The term "computer-readable recording medium" refers to a portable medium such as a USB (Universal Serial Bus) flash memory, an SSD (Solid State Drive), a flexible disk, an optical magnetic disk, a ROM, a CD-ROM, and a storage device such as a hard disk built into a computer system. The term "computer-readable recording medium" also refers to a storage device that holds a program for a certain period of time, such as a volatile memory (RAM) inside a computer system that becomes a server or a client when a program is transmitted via a network such as the Internet or a communication line such as a telephone line. The above-mentioned program may be transmitted from a computer system that stores the program in a storage device or the like to another computer system via a transmission medium or by a transmission wave in the transmission medium. Here, the "transmission medium" that transmits the program refers to a medium that has a function of transmitting information, such as a network (communication network) such as the Internet or a communication line (communication line) such as a telephone line. The program may be for implementing some of the functions described above, or may be a so-called differential file (differential program) that can implement the functions described above in combination with a program already recorded in the computer system. [Explanation of symbols]

[0198] 1...guidance system, 10...guidance device, 11...detection unit, 12...attraction operation unit, 20...management device, 30a...threatening zone, 30b...attraction zone, 40...area, 401...acquisition unit, 402...direction estimation unit, 403...determination unit, 404...operation control unit, 405...threatening / attracting mode estimation unit, 406...threatening / attracting mode generation unit, 600...birds and animals, 1300...estimation system, 1301...estimation unit, 1310...detection device

Claims

1. A plurality of intimidation / attraction operation units arranged in a plurality of areas, each of which intimidates or attracts birds and animals in each area; A plurality of detection units are arranged in the plurality of areas, respectively, and detect birds and animals in each area; an acquisition unit that acquires, when a bird or animal is detected by one of the plurality of detection units, location information indicating the location of the bird or animal; A determination unit that determines one of the guidance patterns for operating each of the plurality of intimidation / attraction operation units based on the position information; and an operation control unit that operates each of the plurality of intimidation / attraction operation units in a intimidation / attraction manner based on the one guidance pattern; A guidance system comprising:

2. A direction estimation unit is provided for estimating a movement direction of a bird or animal in accordance with a threatening / attracting manner for an area specified from the position information, The determination unit determines the one guidance pattern based on the moving direction estimated by the direction estimation unit. The guidance system of claim 1 .

3. The direction estimation unit estimates the movement direction based on the location information, information indicating an intrusion direction of the bird or animal into the specified area, and information indicating a threatening / attracting behavior. The guidance system of claim 2 .

4. a threat / attraction mode estimation unit for estimating a threat / attraction mode having a high threat / attraction effect from among a plurality of types of threat / attraction modes having different threat / attraction effects; the operation control unit operates each of the plurality of intimidation / attraction operation units in a intimidation / attraction mode that has a high intimidation / attraction effect estimated by the intimidation / attraction mode estimation unit, A guidance system according to any one of claims 1 to 3.

5. a threat / attraction pattern generating unit for generating a new threat / attraction pattern having characteristics of the plurality of threat / attraction patterns based on the plurality of threat / attraction patterns that have been actually found to have a threat / attraction effect, the operation control unit changes the mode to the new threatening / attracting mode generated by the threatening / attracting mode generation unit and operates each of the plurality of threatening / attracting operation units. A guidance system according to any one of claims 1 to 3.

6. A plurality of intimidation / attraction operation units arranged in a plurality of areas, each of which intimidates or attracts birds and animals in each area; A plurality of detection units are arranged in the plurality of areas, respectively, and detect birds and animals in each area; A management device that connects to an acquisition step of acquiring location information indicating a location of a bird or animal when the bird or animal is detected by one of the plurality of detection units; a determining step of determining one of the guidance patterns for operating each of the plurality of intimidation / attraction operation units based on the position information; an operation control step of operating each of the plurality of intimidation / attraction operation units in a intimidation / attraction manner based on the one guidance pattern; A method for performing a process including the steps of:

7. A plurality of intimidation / attraction operation units arranged in a plurality of areas, each of which intimidates or attracts birds and animals in each area; A plurality of detection units are arranged in the plurality of areas, respectively, and detect birds and animals in each area; A management device computer connected to an acquisition unit that acquires, when a bird or animal is detected by one of the plurality of detection units, location information indicating the location of the bird or animal; A determination unit that determines one of the guidance patterns for operating each of the plurality of intimidation / attraction operation units based on the position information; an operation control unit that operates each of the plurality of intimidation / attraction operation units in a intimidation / attraction manner based on the one guidance pattern; A program that functions as a

8. A plurality of detection units are arranged in a plurality of areas inhabited by various animals, respectively, and detect target species in each area; an acquisition unit that acquires, when the target species is detected by one of the plurality of detection units, position information indicating a position of the target species; an estimation unit that estimates a movement path of the target species based on the position information acquired by the acquisition unit; An estimation system comprising:

9. The plurality of detection units each detect a behavior of the target species, The acquisition unit acquires a detection result including a behavior of the target species, The estimation unit estimates a movement path of the target species based on the position information and the detection result. The estimation system according to claim 8 .

10. Each of the plurality of detection units includes a first detector that detects the animal and a second detector that detects behavior of the target species, activating the second detector if the animal detected by the first detector is the target species. The estimation system of claim 9 , comprising:

11. The plurality of detection units each include a plurality of types of detectors, and detect the target species and the behavior of the target species based on a plurality of detection results detected by the plurality of types of detectors; The acquisition unit acquires the plurality of detection results, The estimation unit estimates a movement path of the target species based on the position information and the plurality of detection results. The estimation system according to claim 9 .

12. A management device is arranged in a plurality of areas inhabited by various animals, and is connected to a plurality of detection units for detecting target species in each area. an acquisition step of acquiring position information indicating a position of the target species when the target species is detected by one of the plurality of detection units; an estimation step of estimating a movement path of the target species based on the position information acquired in the acquisition step; An estimation method that performs a process including:

13. A computer of a management device is arranged in a plurality of areas inhabited by various animals, and is connected to a plurality of detection units for detecting target species in each area. an acquisition unit that acquires, when the target species is detected by one of the plurality of detection units, position information indicating the position of the target species; an estimation unit that estimates a movement path of the target species based on the position information acquired by the acquisition unit; A program that functions as a

Images