TWM620697U - Precise weed processing device based on neural network analysis and target recognition - Google Patents

Abstract

This case provides a weed processing device, including a carrier, an image capturing module, a spraying module, and a processing module. The image capture module is arranged on the carrier and is structured to capture images. The spraying module is arranged on the carrier, and is structured to implement the spraying operation of the medicament. The processing module is arranged on the carrier and includes a control unit, an identification unit and a data storage unit. The control unit is communicatively connected with the image capturing module and the spraying module. The data storage unit stores the target plant identification model through neural network-like deep learning. The image capture module sends the captured image to the processing module. The identification unit of the processing module performs the target plant identification operation on the captured image according to the target plant identification mode to generate the identification result, and the control unit selects according to the identification result Control the spraying module to implement the spraying operation.

Description

Precise weed processing device based on neural network analysis and target recognition

This case is about a weed processing device, especially a precise weed processing device based on neural network analysis and target recognition.

In order to avoid the growth of weeds in farmland, it is a common practice at present to spray herbicides before the germination of crops in farmland. However, limited by the current shortage of agricultural manpower, in order to save time and manpower, farmers can only spray herbicides on a large area at one time, instead of spraying herbicides for each weed location. However, if herbicides are sprayed on a large area of farmland, the herbicide damage may cause crop loss, resulting in the weeds in the entire farmland being resistant to pesticides, and the floating of the liquid medicine will also cause environmental pollution.

In addition, in the process of crop cultivation, farmland also needs to remove weeds. In order to avoid crop damage caused by herbicides and accurately apply weeds, the current practice is to manually search for every weed in the farmland and spray it with herbicides. However, in the process of artificial herbicide spraying, the herbicide will cause health hazards to the applicator. In addition, the manual spraying of herbicides may cause a waste of human resources, and it is easy to misjudge the crop seedlings as weeds, or omit some weeds, etc., resulting in ineffective weed removal.

On the other hand, because it is usually difficult for the applicator to distinguish the types of weeds directly, and it is also difficult to directly judge whether the weeds are resistant, the applicator cannot target the types of weeds or whether the weeds are resistant. The medicament is sprayed.

In view of this, how to develop a precise weed treatment device based on neural network analysis and target identification to improve the efficiency of weed control, reduce the amount of herbicides, and reduce the environmental pollution caused by the floating of the liquid medicine is actually a need to solve at present The problem.

The purpose of this case is to provide a precise weed treatment device based on neural network analysis and target identification to improve the efficiency of weed control, reduce the amount of herbicides, and reduce environmental pollution caused by liquid drug drift.

In order to achieve the above objective, a broader implementation of this case is to provide a weed processing device, which includes a carrier, an image capturing module, a spraying module, and a processing module. The image capture module is arranged on the carrier, and is structured to provide a capture area and capture an image of the capture area. The spraying module is arranged on the carrier, and is structured to implement the spraying operation of the medicament. The processing module is arranged on the carrier and includes a control unit, an identification unit and a data storage unit. The control unit is communicatively connected with the image capturing module and the spraying module. The data storage unit stores the target plant identification model through neural network-like deep learning. The image capture module sends the image of the captured area to the processing module. The identification unit of the processing module performs the target plant identification operation on the image of the captured area according to the target plant identification mode to generate the identification result, and the control unit based on the identification As a result, the spraying module is selectively controlled to implement the spraying operation.

Some typical embodiments embodying the features and advantages of this case will be described in detail in the following description. It should be understood that this case can have various changes in different aspects, which do not depart from the scope of this case, and the descriptions and illustrations therein are essentially for illustrative purposes, rather than being constructed to limit the case.

Figure 1 is a schematic diagram of the structure of the weed treatment device according to an embodiment of the present invention, Figure 2 is a schematic diagram of the structure of the weed treatment device shown in Figure 1, and Figure 3 is a schematic diagram of the weed treatment device shown in Figure 1. A schematic cross-sectional structure diagram. Figure 4A is a schematic diagram of the first area, the second area, the third area, and the fourth area of the agricultural land according to an embodiment of the present invention. As shown in Figures 1 to 4A, the weed treatment device 100 of this embodiment is configured to identify plants growing on at least one farmland and selectively spray weeds. This embodiment takes farmland C as an example. Explain, but not limited to this. The weed processing device 100 of this embodiment includes a carrier 1, an image capturing module 2, a spraying module 3, and a processing module 4. The image capturing module 2 is disposed on the carrier 1, and is configured to provide a capturing area 20 and capturing images of the capturing area 20 (as shown in FIGS. 4B to 4E). The spraying module 3 is arranged on the carrier 1 and is configured to implement the spraying operation of the medicine. The processing module 4 is disposed on the carrier 1 and includes a control unit 41, an identification unit 42 and a data storage unit 43. The control unit 41 is communicatively connected to the image capturing module 2 and the spraying module 3. The data storage unit 43 stores a target plant identification mode, which is a type of neural network-like deep learning. The image capturing module 2 transmits the image of the capturing area 20 to the processing module 4. The identification unit 42 of the processing module 4 performs the target plant identification operation on the image of the capturing area 20 according to the target plant identification mode to generate the identification result , And the control unit 41 controls the spraying module 3 to implement the spraying operation according to the identification result. The identification result includes at least one or a combination of plant quantity information, plant type information, and plant location information, but is not limited to this.

Fig. 4B is a schematic diagram showing an image of the first area of farmland captured by the capturing area of the image capturing module of the weed processing device in an embodiment of the present invention. For example, as shown in Figures 1 to 3, 4A, and 4B, when the capture area 20 of the image capture module 2 corresponds to the first area W of the farmland C, the image capture module 2 captures the first area The image of W is sent to the processing module 4, and the identification unit 42 of the processing module 4 performs the target plant identification operation on the image of the first area W to generate an identification result, and the identification result is the image of the first area W The number of plants is 0, and the control unit 41 controls the identification unit 42 to perform the target plant identification operation on the next image according to the identification result.

Figure 4C is a schematic diagram of the second region of farmland captured by the capturing area of the image capturing module of the weed processing device according to an embodiment of the present invention. For example, as shown in Figures 1 to 3, 4A, and 4C, when the capture area 20 of the image capture module 2 corresponds to the second area X of the farmland C, the image capture module 2 captures the second area The image of X is sent to the processing module 4. The identification unit 42 of the processing module 4 performs the target plant identification operation on the image of the second area X to generate an identification result, and the identification result is the image of the second area X The image contains plant A1, and plant A1 is a weed. The control unit 41 controls the spraying module 3 to spray the plant A1 according to the identification result. After the spraying is completed, the control unit 42 controls the next image Implement target plant identification tasks. In this embodiment, the weed species is, for example, but not limited to, goosegrass.

Figure 4D is a schematic diagram of the third region of farmland captured by the capturing area of the image capturing module of the weed processing device according to an embodiment of the present invention. For example, as shown in Figures 1 to 3, 4A, and 4D, when the capture area 20 of the image capture module 2 corresponds to the third area Y of the farmland C, the image capture module 2 captures the third area The image of Y is sent to the processing module 4, and the identification unit 42 of the processing module 4 performs the target plant identification operation on the image of the third area Y to generate an identification result, and the identification result is the image of the third area Y The image contains the plant B1, and the plant B1 is a non-weed crop, the control unit 41 controls the identification unit 42 to perform the target plant identification operation on the next image according to the identification result.

Fig. 4E is a schematic diagram showing an image of the fourth area of the farmland captured by the capture area of the image capture module of the weed processing device according to an embodiment of the present invention. For example, as shown in Figures 1 to 3, 4A, and 4E, when the capturing area 20 of the image capturing module 2 corresponds to the fourth area Z of the farmland C, the image capturing module 2 captures the fourth area The image of Z is sent to the processing module 4. The identification unit 42 of the processing module 4 performs the target plant identification operation on the image of the fourth zone Z to generate an identification result, and the identification result is the image of the fourth zone Z The image contains plant A2 and plant B2. Plant A2 is a weed and plant B2 is a non-weed crop. The control unit 41 implements the chemical spraying operation on plant A2 according to the identification result, and then controls the identification after the chemical spraying operation is completed The unit 42 performs the target plant identification operation on the next image. The weed treatment device 100 is used to identify the plants of the farmland C and selectively spray the medicine, so as to realize the high-precision, unmanned spraying of the weed medicine.

In another embodiment, the target plant identification task only identifies at least one area (not shown) in the image that meets the characteristics of weeds, generates at least one mark position (not shown) in at least one area of the image, and the identification result is An image containing at least one marked position. The control unit 41 performs the medicine spraying operation on at least one marked position of the image according to the identification result, but it is not limited to this. In this way, there is no need to identify all the plants in the image, and high-precision, unmanned spraying of weed medicine can be realized.

The carrier 1 of the weed treatment device 100 of this embodiment is a movable unmanned carrier, but it is not limited to this. As shown in FIGS. 1 to 3, the carrier 1 includes a power module 11 and a supporting bracket 12. The power module 11 is configured to move the carrier 1 and includes a moving unit 111 and a power driving unit 112. The power driving unit 112 is power connected to the mobile unit 111 and is communicatively connected to the control unit 41 of the processing module 4. The control unit 41 controls the power driving unit 112 to drive the moving unit 111 to act, so that the carrier 1 is displaced through the moving unit 111. The moving unit 111 may be but not limited to a crawler or a roller, and the power driving unit 112 may be but not limited to a motor. The supporting bracket 12 is arranged on the power module 11, and the image capturing module 2, the spraying module 3 and the processing module 4 are respectively directly or indirectly connected to the supporting bracket 12.

As shown in FIGS. 1 to 4A, when the state of the carrier 1 of the weed processing device 100 is stationary, and the capturing area 20 of the image capturing module 2 corresponds to the first area W of the farmland C, the image capturing model Group 2 captures the image of the first region W, and transmits the image of the first region W to the processing module 4, so that the identification unit 42 of the processing module 4 determines whether the image of the first region W contains plants and identifies the plant type And according to the identification result, the spraying module 3 is controlled to implement the spraying operation. Thereafter, when the identification unit 42 of the processing module 4 determines that no plants are included in the image, or the identification result of the plant included in the image is non-weed, or the identification result of the plant included in the image is weed and the spraying of the medicine has been completed , The control unit 41 of the processing module 4 controls the power drive unit 112 to drive the carrier 1, so that the carrier 1 moves to the second area X of the farmland C and is stationary, and makes the capturing area 20 of the image capturing module 2 correspond to the second Area X, the image capturing module 2 captures the image of the second area X, and transmits the image of the second area X to the processing module 4, so that the identification unit 42 of the processing module 4 determines the image of the second area X Whether it contains plants, identifying plant types, and the control unit 41 of the processing module 4 selectively controls the spraying module 3 to implement the spraying operation according to the identification results, wherein the first area W of the farmland C is adjacent to the second area X . Thereafter, the weed treatment device 100 can further perform the above procedure on the adjacent third zone Y and fourth zone Z, or other adjacent areas, and the implementation is similar to the above content, so it will not be repeated here. . Through the weed treatment device 100, target plant identification operations and selective drug spraying operations are performed on each adjacent area of the farmland C, so as to achieve high-precision, unmanned spraying of weeds.

In one embodiment, the weed treatment device 100 performs the target plant identification operation and the selective spraying operation of the medicine when the carrier 1 is continuously moving. In this embodiment, the identification result generated by the target plant identification operation includes plant quantity information, plant type information, and plant location information. In this embodiment, the weed processing device 100 further includes a rate sensor (not shown). The rate sensor can be, but is not limited to, provided on the carrier 1, and the communication link control unit 41 is built on the measurement carrier Rate information of 1. The control unit 41 of the processing module 4 of the weed processing device 100 calculates the position of the plant based on the plant position information of the identification result and the velocity information of the carrier 1, and selectively controls the spraying module 3 to spray the medicine in the moving state Operation. Through the weed treatment device 100, each area of the farmland C is implemented to identify the target plants and selectively implement the drug spraying operation, so as to achieve high-precision, unmanned spraying of weeds.

As shown in FIGS. 1 to 3, the supporting frame 12 of the carrier 1 of the weed treatment device 100 includes a platform 121 and a supporting structure 122. The platform 121 includes an opening 121 a that penetrates the center of the first platform 121. The supporting structure 122 is disposed on the platform 121 and corresponding to the opening 121 a, so that the movable space S is formed between the supporting structure 122, the opening 121 a of the platform 121 and the moving unit 111. The weed treatment device 100 of this embodiment further includes a robot arm 5 disposed on the supporting structure 122 of the supporting frame 12. The robot arm 5 includes a displacement driving unit 51 and a displacement component 52. The displacement component 52 is dynamically connected to the displacement driving unit 51. The displacement driving unit 51 is connected to the control unit 41 of the processing module 4 and is structured to drive the displacement component 52 according to the control of the control unit 41 so that the end 521 of the displacement component 52 is at Move in the activity space S.

In this embodiment, the supporting structure 122 further includes a supporting base 122a and a plurality of supporting pillars 122b. One end of the plurality of support columns 122b is arranged on the platform 121 and is respectively arranged adjacent to the opening 121a. The other end of the plurality of support columns 122b is connected to the support base 122a so that the height of the support base 122a is higher than the platform 121 to provide a robot arm 5 accommodation space and scope of activity. The support base 122a further includes a hollow portion 122c. The hollow portion 122c penetrates the support base 122a and is detachably disposed in the hollow portion 122c for the robot arm 5. In this embodiment, the height of the bottom surface of the platform 121 supporting the support 12 is preferably between 70 cm and 90 cm, but it is not limited to this. The height of the platform 121 may be based on the height of at least one plant And adjust accordingly. In this embodiment, the ground clearance of the support base 122a of the support structure 122 of the support bracket 12 is preferably between 150 cm and 170 cm, but it is not limited to this. The ground clearance of the support base 122a can be based on The height of at least one plant and the moving range of the robot arm 5 are adjusted accordingly.

As shown in Figures 1 to 3, the spray module 3 of the weed treatment device 100 further includes a nozzle 31, a liquid medicine container 32, and a liquid medicine driving unit 33, wherein the liquid medicine container 32 is disposed on the platform 121 of the supporting frame 12. The structure accommodates at least one liquid medicine. The nozzle 31 is in fluid communication with the liquid medicine container 32, is disposed at the end 521 of the displacement component 52 of the robot arm 5, and moves in the movable space S with the end 521. The liquid medicine driving unit 33 is connected between the nozzle 31 and the liquid medicine container 32, and is connected to the control unit 41 of the processing module 4, and is configured to draw at least one liquid medicine from the liquid medicine container 32 in response to the control of the control unit 41. And at least one chemical liquid is sprayed from the nozzle 31. The liquid medicine driving unit 33 can be, but is not limited to, a liquid pump. The nozzle 31 is fluidly connected to the liquid medicine container 32 through a flexible hose (not shown), and the liquid medicine driving unit 33 is disposed in the flexible hose, but it is not limited to this.

When the control unit 41 of the processing module 4 controls the spraying module 3 to spray at least one plant, the control unit 41 controls the displacement driving unit 51 of the robot arm 5 so that the displacement driving unit 51 drives the end of the displacement assembly 52 521, and the nozzle 31 of the spray module 3 provided at the end 521 move to a position corresponding to at least one plant, and control the liquid medicine drive unit 33 to draw at least one liquid medicine contained in the liquid medicine container 32, and make At least one liquid medicine is sprayed from the nozzle 31 to the at least one plant. When the drug spraying operation is completed, the control unit 41 controls the displacement driving unit 51 of the robot arm 5 so that the displacement driving unit 51 drives the end 521 of the displacement assembly 52 and the nozzle 31 of the spray module 3 provided at the end 521 Reset to the initial position. In this way, at least one liquid medicine is sprayed accurately to the position of the at least one plant, so as to reduce the amount of herbicide, improve the efficiency of weed control, and reduce environmental pollution caused by liquid medicine drift.

In one embodiment, the medicinal solution container 32 includes a plurality of accommodating grooves (not shown), which are configured to accommodate a plurality of different medicinal liquids, and each of the medicinal liquids corresponds to one or more weeds. For example, when the identification unit 42 of the processing module 4 determines that at least one plant in the image is the first weed, the control unit 41 controls the liquid driving unit 33 of the spraying module 3 according to the identification result to drive the liquid The unit 33 picks up the liquid medicine corresponding to the first weed from the liquid medicine container 32, and then sprays it from the nozzle 31 to the position of the at least one plant. For example, when the identification unit 42 of the processing module 4 determines that at least one plant in the image is a second weed, the control unit 41 controls the liquid driving unit 33 of the spraying module 3 according to the identification result to drive the liquid The unit 33 draws the liquid medicine corresponding to the second weed from the liquid medicine container 32, and then sprays it from the nozzle 31 to the position of the at least one plant. In this way, targeted spraying of weed medicine is realized, and the efficiency of weed control is improved.

In this embodiment, the image capturing module 2 is disposed at the end 521 of the displacement element 52 of the robot arm 5, and moves in the movable space S with the end 521 of the displacement element 52, so that it can be based on at least one plant The image capturing height of the image capturing module 2 is adjusted correspondingly to the height, and the image capturing module 2 can be moved to a position close to at least one plant to capture a higher resolution image, which can be tailored to the number of leaves at the young stage Spraying with less weeds (such as 3-8 leaves) or smaller weeds not only improves the flexibility of image capture, but also improves the accuracy of weed identification.

In some embodiments, the image capturing module 2 can also be fixed to the platform 121 or the supporting structure 122 of the supporting bracket 12. For example, the image capturing module 2 can be installed on the bottom surface of the platform 121 of the supporting frame 12 and performing image capturing in the direction of the farmland C, but it is not limited to this. The image capturing module 2 can be installed at Make any changes based on actual implementation needs. The image capturing module 2 is fixed on the platform 121 or the supporting structure 122 of the supporting bracket 12, which is farther from the agricultural land C, and the capturing area 20 covers the agricultural land C to a larger extent, which reduces the capturing of the capturing area 20 The image does not contain plants to improve the efficiency of weed spraying.

As shown in FIG. 1, the weed treatment device 100 of this embodiment further includes a power supply module 6. The power supply module 6 of this embodiment is disposed on the bottom surface of the platform 121 of the supporting bracket 12, but it is not limited to this. The power supply module 6 is electrically connected to the power module 11, the image capturing module 2, the spraying module 3, and the processing module 4, and is configured to provide power.

In one embodiment, the neural network-like deep learning target plant identification model uses a deep learning model to train the plant target identification model, which can solve the relatively difficult problem of traditional image processing binarization in complex background discrimination. For example, you can first collect multiple images of upland goosegrass in different growth states and growth environments, and use most of the images as the training set, and a small part of the images as the test set, using a neural network-like deep learning model. Training can improve the accuracy of plant target identification. In one embodiment, the neural network-like deep learning model uses YOLO-Tiny to train the target detection model, but it is not limited to this, and any existing or about to be invented target detection deep learning model can be incorporated for reference.

To sum up, this case captures images of farmland through the image capture module, the identification unit recognizes the plants in the image, and selectively sprays drugs through the spray module to achieve high accuracy and unmanned weeds The spraying of the medicament improves the efficiency of weed control, reduces the amount of herbicides and reduces the environmental pollution caused by the accumulation of liquid medicine.

This case can be modified in many ways by those who are familiar with this technology, but none of them deviates from the protection of the scope of the patent application.

100: Weed treatment device 1: carrier 11: Power module 111: mobile unit 112: Power drive unit 12: Carrying bracket 121: platform 121a: opening 122: support structure 122a: Support seat 122b: Support column 122c: hollow part 2: Image capture module 20: Capture area 3: Spray module 31: Nozzle 32: liquid medicine container 33: Liquid medicine drive unit 4: Processing module 41: control unit 42: Identification Unit 43: data storage unit 5: Robotic arm 51: Displacement drive unit 52: Displacement component 521: End 6: Power supply module A1, A2, B1, B2: Plant C: Farmland S: Event space W: first zone X: second area Y: third area Z: The fourth zone

Figure 1 is a schematic diagram of the structure of a weed treatment device according to an embodiment of the present invention. Figure 2 is a schematic diagram of the structure of the weed treatment device shown in Figure 1. Figure 3 is a schematic cross-sectional structure diagram of the weed treatment device shown in Figure 1. Figure 4A is a schematic diagram of the first area, the second area, the third area, and the fourth area of the agricultural land according to an embodiment of the project. Fig. 4B is a schematic diagram showing an image of the first area of farmland captured by the capturing area of the image capturing module of the weed processing device in an embodiment of the present invention. . Figure 4C is a schematic diagram of the second region of farmland captured by the capturing area of the image capturing module of the weed processing device according to an embodiment of the present invention. Figure 4D is a schematic diagram of the third region of farmland captured by the capturing area of the image capturing module of the weed processing device according to an embodiment of the present invention. Fig. 4E is a schematic diagram showing an image of the fourth area of the farmland captured by the capture area of the image capture module of the weed processing device according to an embodiment of the present invention.

100: Weed treatment device

1: carrier

11: Power module

111: mobile unit

112: Power drive unit

12: Carrying bracket

121: platform

121a: opening

122: support structure

122a: Support seat

122b: Support column

122c: hollow part

2: Image capture module

3: Spray module

31: Nozzle

32: liquid medicine container

4: Processing module

5: Robotic arm

6: Power supply module

Claims (10)

A weed processing device, comprising: a carrier; an image capturing module arranged on the carrier, and configured to provide a capturing area and capturing an image of the capturing area; and a spraying module arranged in The carrier is configured to implement a drug spraying operation; and a processing module is arranged on the carrier and includes a control unit, an identification unit and a data storage unit, wherein the control unit is communicatively connected to the image capturing module and The spraying module, the data storage unit storing a target plant identification mode through neural network-like deep learning, wherein the image capturing module transmits the image of the captured area to the processing module, and the processing module The identification unit performs a target plant identification operation on the image of the captured area according to the target plant identification mode to generate an identification result, and the control unit selectively controls the spraying module to implement according to the identification result The agent spraying operation. The weed treatment device according to claim 1, wherein the weed treatment device performs the target plant identification operation and/or the drug spraying operation under the continuous movement of the carrier, wherein the identification result includes plant quantity information and plant type At least one of the message and the plant position message or a combination thereof. The weed processing device according to claim 1, wherein when the identification result is that the number of plants in the image is 0, the control unit controls the identification unit to perform the target plant identification operation on the next image according to the identification result, wherein when the The identification result is that the image contains at least one plant, and the at least one plant is a weed, and the control unit controls the spraying module to spray the medicine on the at least one plant according to the identification result, and then controls the identification unit to The next image implements the target plant identification task, and when the The recognition result is that the image contains the at least one plant and the at least one plant is a non-weed crop. The control unit controls the recognition unit to perform the target plant recognition operation on the next image according to the recognition result, wherein when the recognition result is the The image contains a plurality of plants, and the plurality of plants are a combination of weeds and non-weed crops, the control unit performs the spraying operation on the weeds according to the identification result, and then controls the identification unit to implement the next image Target plant identification task. The weed treatment device according to claim 1, wherein the carrier further includes a power module and a supporting bracket, wherein the power module is configured to move the carrier, and includes a moving unit and a power driving unit, wherein the The power drive unit is communicatively connected to the control unit, and the control unit controls the power drive unit to drive the moving unit so that the carrier is displaced through the moving unit, wherein the bearing bracket is disposed on the power module, and the image capturing module The group, the spraying module and the processing module are respectively connected to the supporting frame, wherein the moving unit is a crawler, and the power driving unit is a motor. The weed processing device according to claim 4, wherein when the carrier is stationary and the capturing area of the image capturing module is located in a first area, the image capturing module captures the first area An image, wherein after the processing module implements the target plant identification operation on the image in the first region, and selectively implements the medicine spraying operation according to the identification result, the control unit controls the power driving unit to drive the moving unit , The carrier is moved, and the capturing area of the image capturing module is located in a second area, and the image capturing module captures an image of the second area, wherein the first area and the second area The regions are adjacent, wherein the identification result includes at least one or a combination of plant quantity information, plant type information, and plant position information. The weed treatment device according to claim 4, wherein the supporting frame includes a platform and a supporting structure, wherein the platform includes an opening that penetrates the platform, and The supporting structure is arranged on the platform and arranged corresponding to the opening. The weed treatment device according to claim 6, wherein the weed treatment device further comprises a mechanical arm arranged on the supporting structure of the supporting frame and arranged toward the direction of the opening, wherein the mechanical arm comprises a displacement drive Unit and a displacement component, the displacement component is dynamically connected to the displacement driving unit, the displacement driving unit is connected to the control unit of the processing module, and is constructed to drive the displacement component under the control of the control unit to make the displacement One end of the component moves. The weed treatment device according to claim 7, wherein the spraying module further comprises a liquid medicine container, a nozzle, and a liquid medicine driving unit, wherein the liquid medicine container is arranged on the platform of the supporting frame and is constructed on At least one liquid medicine is accommodated, wherein the nozzle fluidly communicates with the liquid medicine container, is arranged at the end of the displacement component of the robot arm, and moves with the end, wherein the liquid medicine driving unit is connected to the nozzle and The liquid medicine container is connected with the control unit of the processing module, so as to draw the at least one liquid medicine from the liquid medicine container under the control of the control unit, and make the at least one liquid medicine flow from the nozzle Squirting. The weed treatment device according to claim 8, wherein when the identification result is that the image includes at least one plant and the at least one plant includes weeds, the control unit controls the spraying module to implement the at least one plant For spraying medicine, the control unit controls the displacement drive unit of the robotic arm so that the displacement drive unit drives the end of the displacement assembly and the nozzle of the spray module to move to a position corresponding to the at least one plant, And the control unit controls the liquid medicine driving unit to draw the at least one liquid medicine contained in the liquid medicine container, and causes the at least one liquid medicine to be sprayed from the nozzle to the at least one plant, wherein when the spraying operation of the medicine is completed At this time, the control unit controls the displacement driving unit of the robot arm so that the displacement driving unit drives the end of the displacement assembly and the nozzle of the spray module to an initial position. The weed processing device according to claim 7, wherein the image capturing module is arranged at the end of the displacement component of the robot arm and moves with the end of the displacement component, or the image capturing The module is arranged on the supporting bracket.

Images