Field double-layer pesticide spraying robot applied to potatoes and control method

Abstract

The invention relates to the technical field of potato crop field plant protection, in particular to a field double-layer pesticide spraying robot applied to potatoes and a control method, and aims at solving the problems that in the prior art, spraying precision is insufficient, pesticide liquid is wasted, and autonomous navigation cannot be achieved. The robot is provided with the double-layer pesticide spraying device, and accurate layered spraying is conducted on plant canopies, bottoms and potato ridges through the rear pesticide spraying device and the lower pesticide spraying device. The robot adopts a visual system to monitor plant height and disease distribution information in real time, a control device automatically adjusts the liquid medicine proportion of a water tank, a pesticide box and a sterilization pesticide box according to disease types and severity, and precise mixing and saving of pesticides are achieved. The robot can guarantee continuous operation by monitoring the stock of liquid medicine in real time and automatically adjusting the spraying concentration. After the operation is finished, the robot can automatically generate a pesticide spraying diagram. The device is suitable for disease and pest control of tuber crops such as tubers, spraying efficiency can be effectively improved, pesticide dosage is reduced, and labor cost is reduced.

Description

Technical Field

[0001] The invention relates to the technical field of field plant protection of potato crops, and in particular to a double-layer field spraying robot applied to potatoes and a control method thereof. Background Art

[0002] At present, agricultural automation equipment is gradually being used in field management. However, existing automatic spraying equipment still has deficiencies in intelligent spraying control in complex field environments. Most automatic spraying systems lack real-time disease identification and spraying ratio functions, and the efficiency of liquid medicine use is low, making it difficult to adapt to complex pest and disease control needs. The bottom and ridge areas of potato crops are more prone to diseases and pests. Traditional spraying methods cannot achieve precise spraying of the canopy and bottom diseases of potato plants, and it is difficult to effectively cover target areas at different levels.

[0003] In addition, most existing spraying equipment uses mixed agents, while insecticides and fungicides have different chemical compositions. Some of the components may react chemically after mixing, causing the agents to become ineffective or produce side effects. In actual work, only pests or diseases may appear on the same plant, and the use of mixed agents will lead to a waste of agents.

[0004] Therefore, the present invention designs a double-layer field spraying robot and control method for potatoes, which can realize precise zoning spraying and dosage control of the spraying robot, thereby effectively reducing the amount of pesticides used and reducing the impact on the environment while improving work efficiency. Summary of the invention

[0005] In view of the shortcomings of the prior art, the present invention provides a double-layer spraying robot and control method for potato fields, which realizes precise layered spraying of the plant canopy, bottom and potato ridges through a double-layer spraying device. At the same time, the integrated visual system can detect disease information in real time and automatically adjust the ratio of insecticides and fungicides to achieve efficient and economical disease and pest control.

[0006] The purpose of the present invention can be achieved through the following technical solutions:

[0007] A control method for a double-layer field spraying robot for potatoes comprises the following steps:

[0008] S1. After the spraying robot is started, it obtains the spatial information of the plot, automatically generates the optimal operation path through the path planning algorithm, and performs autonomous navigation operations;

[0009] S2. When the spraying robot moves to the operation area, it collects images of the potato field in real time, detects the disease status and height information of the plants, uses the built-in disease recognition model and disease classification model, determines the spraying strategy according to the disease type, severity and plant height, and calculates the appropriate pesticide ratio and spraying plan;

[0010] S3. Monitor the flow data of each medicine box in real time, adjust the output of the extraction pump corresponding to each medicine box according to the information fed back by the flow sensor, control the flow of each liquid medicine, and ensure that the ratio meets the preset ratio;

[0011] At the same time, the extraction pump is adjusted to control the flow rate according to the travel speed of the spraying robot;

[0012] In addition, the liquid level of each medicine box is monitored in real time. When a medicine box is running low on liquid, a replenishment reminder is issued, and the liquid ratio is automatically adjusted when liquid is insufficient.

[0013] S4. During the spraying process, the spraying robot controls the spraying angle and height of the fungicide nozzle and the insecticide nozzle in real time according to the plant height and disease distribution, accurately covers the plant canopy, plant bottom and potato ridge, and realizes the zoning treatment of the upper and lower layer diseases;

[0014] S5. During the spraying process of the spraying robot, the control device collects the spraying amount data in real time through the flow sensor, and generates a spraying amount map after the spraying of the entire field is completed.

[0015] As a further technical solution of the present invention, the following steps are also included:

[0016] The disease recognition model and disease classification model in S2 are specifically as follows: the disease recognition model adopts the YOLOv10 neural network model;

[0017] YOLOv10 is used as the basic network of the disease recognition model, and the original backbone feature extraction network is replaced with HGNet. Different convolution types are dynamically selected by introducing the num parameter to replace the ordinary convolution of the HGBlock module in HGNet.

[0018] When num=1 or num=2, ordinary convolution is replaced by ghost convolution GhostConv; when num=3 or num=4, ordinary convolution is replaced by reparameterized convolution RepConv, and the SPPF module in the backbone network is improved. Before the pooling pyramid operation, a 1×1 convolution layer is added to adjust the number of channels of the input features, and the kernel size of the pooling layer is uniformly set to 5×5, and the pooling operation is performed in a cascade manner. Finally, the channel compression convolution after the output of the pooling pyramid is replaced by grouped convolution, and the generated output features will be passed to the subsequent network layers. The original PIoU loss function in YOLOv10 is replaced by Inner-WloU;

[0019] The disease grading model uses the image segmentation algorithm U-Net to segment the diseased area, calculate the disease coverage percentage, and divide the disease into grades.

[0020] As a further technical solution of the present invention: the specific method for generating the spraying strategy is:

[0021] Determine the type of liquid medicine according to the disease type T, and dynamically adjust the liquid medicine concentration D according to the severity of the disease L:

[0022] D=D base (1+kL)

[0023] Where D base is the basic solution concentration, k is the concentration adjustment coefficient, and L is the severity level of the disease.

[0024] As a further technical solution of the present invention: the specific method of generating the dosage spraying diagram is:

[0025] During the spraying process, the pesticide flow rate Q is p and bactericide flow Q f Match with GPS location data (x, y) to form a data pair (x, y, Q p ) and (x, y, Q f ), record the flow and location data as a set of time series data;

[0026] The field is divided into multiple field grids, each grid unit is represented by the coordinates (i, j), and all sampling points (x, y, Q p ) and (x, y, Q f ) is accumulated to the corresponding grid unit (i, j), and the total spraying amount Q of each grid unit is obtained. p总 and Q f总 ;

[0027] For grids with incomplete spraying data such as uncovered or insufficient spraying amount, the IDW algorithm is used to estimate the spraying amount of insecticides and fungicides. For areas at the edge of the field or at the bend of the path, weighted smoothing is used to adjust the spraying amount in the edge area.

[0028] According to the dosage of pesticides and fungicides sprayed, the dosage spraying maps are generated respectively through the color mapping formula.

[0029] A double-layer field spraying robot for potatoes, the double-layer field spraying robot comprising:

[0030] Control device: installed at the front end of the walking device;

[0031] Vision system: installed on both sides of the front end of the walking device;

[0032] Traveling device: A medicine box group is installed on the upper middle part, two groups of lower spraying devices are installed on both sides of the lower part, and a lifting device is installed at the tail. The lifting device is fixedly installed with the rear spraying device;

[0033] The walking device includes: a chassis, a battery pack, and a side fixing frame. The front end of the chassis is equipped with a battery pack, and the sides are respectively equipped with side fixing frames;

[0034] The rear spraying device comprises: a spray rod frame, a spray head assembly, and a spray assembly is sequentially installed at the rear end of the spray rod frame;

[0035] The nozzle assembly comprises: a rotating mechanism, a bactericidal nozzle, an insecticide nozzle, and a medicine tube, wherein the bactericidal nozzle and the insecticide nozzle are sequentially installed at the lower end of the rotating mechanism, and medicine tubes are installed at the upper ends of the bactericidal nozzle and the insecticide nozzle;

[0036] The lifting device comprises: a cover shell, a flow sensor, a diverter, an extraction pump, a fixed base, a medicine tube, a medicine liquid mixer, and an electric push rod. An extraction pump is sequentially installed on the front side of the cover shell, a flow sensor is installed above each extraction pump, and two medicine liquid mixers are sequentially installed inside. The flow sensors on both sides are connected to the medicine liquid mixer through medicine tubes and are respectively connected to the two medicine liquid mixers inside the cover shell. The flow sensor in the middle is connected to the diverter through the medicine tube. The two medicine tubes separated by the diverter are respectively connected to the two medicine liquid mixers inside the cover shell. A fixed base is installed on the upper part of the cover shell, and medicine tubes are respectively installed below the two medicine liquid mixers and connected to the fixed base. Two electric push rods are sequentially installed at the lower end of the fixed base.

[0037] The medicine box group includes: a sterilization medicine box, a water tank, and an insecticide medicine box. A water tank is installed in the middle of the medicine box group, and a sterilization medicine box and an insecticide medicine box are installed on both sides respectively. The sterilization medicine box is connected to an extraction pump installed on the same side of the cover shell through a medicine pipe, and the insecticide medicine box is connected to an extraction pump installed on the same side of the cover shell through a medicine pipe. The water tank is connected to an extraction pump installed in the middle of the cover shell through a medicine pipe.

[0038] The lower spraying device comprises: a medicine tube, a bactericidal nozzle, an insecticide nozzle, and an electric slide. The two electric slide rails are installed outwards, and the bactericidal nozzle and the insecticide nozzle are installed on the slide in sequence. The ends of the two nozzles are respectively installed with medicine tubes.

[0039] Beneficial effects of the present invention:

[0040] The present invention designs a double-layer field spraying robot for potatoes. By separately controlling the spraying angle and height of the rear spraying device and the lower spraying device, the canopy and bottom of the plant can be accurately covered, thereby achieving zoning prevention and control of different disease sites and improving the effectiveness of spraying.

[0041] The present invention designs a control method for a double-layer field spraying robot for potatoes. Through real-time detection by a visual system, the ratio of insecticides and fungicides is automatically adjusted according to the type and severity of the disease, thereby reducing unnecessary use of liquid medicine, reducing pesticide costs and reducing environmental pollution. The system monitors the liquid medicine inventory in real time, automatically adjusts the spraying concentration or reminds to replenish the liquid medicine, ensuring the continuity of the spraying work and avoiding the decline in the prevention and control effect due to insufficient liquid medicine. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for use in the description of the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0043] Figure 1 Schematic diagram of the overall structure of the spraying robot in an embodiment of the present invention.

[0044] Figure 2 Schematic diagram of the walking device structure of the spraying robot in an embodiment of the present invention.

[0045] Figure 3 Schematic diagram of the structure of the rear spraying device of the spraying robot in an embodiment of the present invention.

[0046] Figure 4 Schematic diagram of the spray head assembly structure of the spray robot in an embodiment of the present invention.

[0047] Figure 5 Schematic diagram of the lifting device structure of the spraying robot in an embodiment of the present invention.

[0048] Figure 6 This is an oblique axonometric view of the lifting device of the spraying robot in an embodiment of the present invention.

[0049] Figure 7 Schematic diagram of the medicine box group structure of the spraying robot in an embodiment of the present invention.

[0050] Figure 8 Schematic diagram of the structure of the lower spraying device of the spraying robot in an embodiment of the present invention.

[0051] Fig. 9 This is a network structure diagram after the HGBlock module is improved in an embodiment of the present invention.

[0052] Fig.10 This is a network structure diagram after the SPPF module is improved in an embodiment of the present invention.

[0053] Fig.11 Flowchart of the control method of the spraying robot in an embodiment of the present invention. DETAILED DESCRIPTION

[0054] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0055] Embodiment 1

[0056] like Figure 1-8 As shown, a double-layer field spraying robot for potatoes provided in the first embodiment of the present invention includes a walking device 1, a control device 2, a visual system 3, a rear spraying device 4, a lifting device 5, a medicine box group 6, and a lower spraying device 7;

[0057] The control device 2 is installed at the front end of the walking device 1, the visual system 3 is installed on both sides of the front end of the walking device 1, a medicine box group 6 is installed on the upper middle part of the walking device 1, two groups of lower spraying devices 4 are installed on both sides below, and a lifting device 5 is installed at the tail, the lifting device 5 is fixedly installed with the rear spraying device 4, and the medicine box group 6 is respectively connected with the lower spraying device 4 and the rear spraying device 4, and is used to supply liquid to the lower spraying device 4 and the rear spraying device 4;

[0058] The walking device 1 includes: a chassis 11, a battery pack 12, and a side fixing frame 13. The front end of the chassis 11 is equipped with a battery pack 12, and the sides are respectively equipped with side fixing frames 13, and also includes walking wheels for driving the double-layer spraying robot in the field;

[0059] The rear spraying device 4 comprises: a spraying rod frame 41, a spray head assembly 42, 9 groups of spraying assemblies 42 are sequentially installed at the rear end of the spraying rod frame 41, and the front end of the spraying rod frame 41 is fixed on the fixed base 55 of the lifting device 5;

[0060] The spray head assembly 42 includes: a rotating mechanism 421, a bactericide nozzle 422, an insecticide nozzle 423, and a medicine tube 424. The bactericide nozzle 422 and the insecticide nozzle 423 are sequentially installed at the lower end of the rotating mechanism 421, and the upper ends of the bactericide nozzle 422 and the insecticide nozzle 423 are both installed with medicine tubes 424; the rotating mechanism 421 includes a fixed body and a rotating body, the rotating body is rotatably installed on the fixed body, the rotating body can be driven by a motor, and the bactericide nozzle 422 and the insecticide nozzle 423 are installed on the rotating body, and the fixed body is installed on the spray rod frame 41;

[0061] The lifting device 5 comprises: a housing 51, a flow sensor 52, a diverter 53, an extraction pump 54, a fixed base 55, a liquid medicine mixer 56, and an electric push rod 57. Three extraction pumps 54 are sequentially installed on the front side of the housing 51, a flow sensor 52 is installed above each extraction pump 54, and two liquid medicine mixers 56 are sequentially installed inside. The flow sensors 52 on both sides are connected to the liquid medicine mixer 56 through medicine tubes 424, and are respectively connected to the two liquid medicine mixers 56 inside the housing 51. The flow sensor 52 in the middle is connected to the diverter 53 through the medicine tube 424. The two medicine tubes 424 separated by the diverter 53 are respectively connected to the two liquid medicine mixers 56 inside the housing 51. A fixed base 55 is installed on the upper part of the housing 51. Medicine tubes 424 are respectively installed below the two liquid medicine mixers 56 and connected to the fixed base 55. Two electric push rods 57 are sequentially installed at the lower end of the fixed base 55.

[0062] The medicine box group 6 includes: a sterilization medicine box 61, a water tank 62, and an insecticide medicine box 63. The water tank 62 is installed in the middle of the medicine box group 6, and the sterilization medicine box 61 and the insecticide medicine box 63 are installed on both sides respectively. The sterilization medicine box 61 is connected to the extraction pump 54 installed on the same side of the cover 51 through the medicine pipe 424. Similarly, the insecticide medicine box 63 is connected to the extraction pump 54 installed on the same side of the cover 51 through the medicine pipe 424. The water tank is connected to the extraction pump 54 installed in the middle of the cover 51 through the medicine pipe 424.

[0063] The lower spraying device 7 includes: a medicine tube 424, a bactericidal nozzle 422, an insecticide nozzle 423, and an electric slide 71. The slide rails of the two electric slides 71 are installed outward, and the bactericidal nozzle 422 and the insecticide nozzle 423 are installed on the electric slide 71 in sequence, and the ends of the two nozzles are respectively installed with medicine tubes 424.

[0064] Embodiment 2

[0065] like Figure 9-11 As shown, a control method for a double-layer field spraying robot applied to potatoes provided in the second embodiment of the present invention includes the following steps:

[0066] S1: After the spraying robot is started, the spatial information of the plot acquired by the visual system 3 is transmitted to the control device 2, and the optimal operation path is automatically generated through the path planning algorithm. Then the control device 2 drives the walking device 1 to perform autonomous navigation operation;

[0067] S2: When the spraying robot moves to the operation area, the visual system 3 starts to collect potato field images in real time, detects the disease status and height information of the plants, and the control device 2 uses the built-in disease recognition model and disease classification model to determine the spraying strategy according to the disease type, severity and plant height, and calculates the appropriate pesticide ratio and spraying plan;

[0068] In some implementation schemes: the disease recognition model and the disease grading model in step S2 are specifically: the disease recognition model adopts the YOLOv10 neural network model, and the following improvements are made on the basis of the YOLOv10 neural network model;

[0069] HGNet is used as the backbone feature extraction network. Different convolution types are dynamically selected by introducing the num parameter to replace the ordinary convolution of the HGBlock module in HGNet. When num=1 or num=2, the ordinary convolution is replaced by GhostConv; when num=3 or num=4, the ordinary convolution is replaced by RepConv. The SPPF module in the backbone network is improved. Before the pooling pyramid operation, a 1×1 convolution layer is added to adjust the number of channels of the input feature. The kernel size of the pooling layer is uniformly set to 5×5, and the pooling operation is performed in a cascade manner. Finally, the channel compression convolution after the output of the pooling pyramid is replaced by grouped convolution. The generated output features will be passed to the subsequent network layers, and the original PIoU loss function in YOLOv10 is replaced by Inner-WloU.

[0070] Use the trained disease recognition model to classify diseases and output the preset disease type T, for example, T 1 For early blight, T 2 For late blight, T 3 For powdery mildew, T 4 For leaf roll disease, T 5 For leaf spot, T 6 For leaf blight, T 7 For viral diseases, etc.

[0071] The disease classification model uses the image segmentation algorithm U-Net to segment the disease area and calculate the disease coverage percentage S. The disease coverage percentage calculation formula is:

[0072]

[0073] The disease level L is divided according to the disease coverage percentage S.

[0074] No disease (L 0 ): S = 0%;

[0075] Mild (L 1 ): 0%<S≤10%;

[0076] Moderate (L 2 ): 10%<S≤30%;

[0077] Severe (L 3 ): 30%<S;

[0078] The control device 2 outputs the diagnosis result according to the disease type T and severity L. For example, the diagnosis result of late blight is (T 2 , L 1 );

[0079] The method for obtaining the height information in step S2 is to scan the point cloud data of the plants through a visual system and calculate the average height of the plants;

[0080]

[0081] where h i is the height of the i-th point, and N is the total number of point cloud data;

[0082] The specific method of generating the spraying strategy in step S2 is:

[0083] Determine the type of liquid medicine according to the disease type T, and dynamically adjust the liquid medicine concentration D according to the severity of the disease L;

[0084] D=D base (1+kL)

[0085] Where D base is the basic liquid concentration, k is the concentration adjustment coefficient, and L is the severity level of the disease;

[0086] S3: The three flow sensors 52 respectively monitor the flow data of each medicine box in the medicine box group 6 in real time. The control device 2 adjusts the output of the extraction pump 54 corresponding to each medicine box according to the information fed back by the flow sensor 52, accurately controls the flow of each liquid medicine, and ensures that the ratio meets the preset ratio. At the same time, the control device 2 will adjust the extraction pump 54 to control the flow according to the travel speed of the spraying robot to ensure uniform coverage of each area. The liquid medicine output by each medicine box is transported to the liquid medicine mixer 56 through the extraction pump 54. The liquid medicine mixer 56 dynamically mixes the liquid medicine to ensure that the concentration of the generated liquid medicine is uniform. Liquid level sensors are installed in the medicine box group 6. The control device 2 monitors the liquid level height of each medicine box in real time through the liquid level sensor. When a medicine box is short of liquid medicine, the control device 2 issues a replenishment reminder. At the same time, the control device 2 automatically adjusts the liquid medicine ratio when the liquid medicine is insufficient to ensure continuous operation without interruption.

[0087] S4: During the spraying process, the control device 2 adjusts the rotating mechanism 421, the electric push rod 57 and the electric slide 71 in real time according to the plant height and disease distribution, controls the spraying angle and height of the fungicide nozzle 422 and the insecticide nozzle 423, accurately covers the plant canopy, the plant bottom and the potato ridge, and realizes the zoning treatment of the upper and lower layer diseases;

[0088] S5: During the spraying process of the spraying robot, the control device 2 collects spraying amount data in real time through the flow sensor 52, and generates a spraying amount map after the spraying of the entire field is completed;

[0089] In some embodiments: the specific method of generating the drug dosage spraying diagram in step S5 is:

[0090] During the spraying process, the control device 2 regularly samples the amount of medicine extracted from the flow sensor 52 connected to the insecticide box 63 and the bactericidal box 61, with a sampling interval of 1 second. The RTK GPS module carried by the spraying robot control device 2 records the position data of the spraying robot. The control device 2 calculates the insecticide flow Q at each sampling moment. p and bactericide flow Q f Match with GPS location data (x, y) to form a data pair (x, y, Q p ) and (x, y, Q f ), record the traffic and location data as a set of time series data:

[0091] Q pData ={(t 1 ,x 1 ,y 1 ,Q p1 ),(t 2 ,x 2 ,y 2 ,Q p2 ),...,(t n ,x n ,y n ,Q pn )}

[0092] Q fData ={(t 1 ,x 1 ,y 1 ,Q f1 ),(t 2 ,x 2 ,y 2 ,Q f2 ),...,(t n ,x n ,y n ,Q fn )}

[0093] The field is divided into multiple 1m×1m field grids, each grid unit is represented by the coordinates (i, j), and all sampling points (x, y, Q p ) and (x, y, Q f ) is accumulated to the corresponding grid unit (i, j), and the total spraying amount Q of each grid unit is obtained. p Sum Q f total:

[0094]

[0095] Among them, (x k , y k , Q k ) represents the information of a spraying event. x k , y k represent the location coordinates of the spraying event, and Q k represents the amount of medicine released during the spraying event. cell(i, j) represents the range of the grid cell (i, j), including all spraying times within the grid cell.

[0096] For grid cells with incomplete spraying data such as uncovered or insufficient spraying amount, the IDW algorithm is used to estimate the spraying amounts of insecticides and fungicides. The IDW interpolation formula is as follows:

[0097]

[0098] Among them, N represents the number of neighboring grid cells used for interpolation, and Q l is the amount value of the neighboring grid cell, d l is the distance from the current grid cell to the neighboring grid cell, and p is the distance weight, which takes the value of 2 here.

[0099] For areas at the edge of the field or at the turning of the path, the spraying amount may be affected by the movement path and speed changes. Weighted smoothing is used to adjust the spraying amount in the edge area to make the transition smoother. The edge processing formula is as follows:

[0100] Q (i,j) ′ = αQ (i,j) + (1 - α)Q s

[0101] Among them, Q (i,j) ′ is the adjusted spraying amount at the edge, Q s is the average spraying amount of the adjacent grid cell, and α is the weighting factor, which takes the value of 0.7.

[0102] According to the amounts of insecticides and fungicides sprayed, dosage spraying maps are generated respectively through the color mapping formula. The color mapping formula is as follows:

[0103]

[0104] Among them, f is the color mapping function, and different gradient colors are used for insecticides and fungicides to represent the dosage intensity.

[0105] Working principle of the present invention: the visual system starts to collect potato field images in real time, detects the disease status and height information of the plants, and the control device uses the built-in disease recognition model and disease classification model to determine the spraying strategy according to the disease type, severity and plant height, and calculates the appropriate drug ratio and spraying plan; three flow sensors respectively monitor the flow data of each medicine box in the medicine box group in real time, and the control device adjusts the output of the extraction pump corresponding to each medicine box according to the information fed back by the flow sensor, accurately controls the flow of each liquid medicine, and ensures that the ratio meets the preset ratio. At the same time, the control device adjusts the extraction pump control flow according to the travel speed of the spraying robot. To ensure uniform coverage of each area, the liquid medicine output from each medicine box is transported to the liquid medicine mixer through the extraction pump, and the liquid medicine mixer dynamically mixes the liquid medicine to ensure that the generated liquid medicine concentration is uniform; during the spraying process of the spraying robot, the control device adjusts the rotating mechanism, electric push rod and electric slide in real time according to the plant height and disease distribution, controls the spray angle and height of the fungicide nozzle and the insecticide nozzle, and accurately covers the plant canopy, plant bottom and potato ridge, so as to achieve zoning treatment of upper and lower layer diseases; during the spraying process of the spraying robot, the control device collects the spraying amount data in real time through the flow sensor, and generates a spray amount map after the spraying of the entire field is completed.

[0106] The above is a detailed description of an embodiment of the present invention, but the content is only a preferred embodiment of the present invention and cannot be considered to limit the scope of implementation of the present invention. All equivalent changes and improvements made within the scope of the present invention should still fall within the scope of the patent coverage of the present invention.

Claims

1. A control method for a double-layer field spraying robot for potatoes, characterized in that: The following steps are involved: S1. After the spraying robot is started, it obtains the spatial information of the plot, automatically generates the optimal operation path through the path planning algorithm, and performs autonomous navigation operations; S2. When the spraying robot moves to the operation area, it collects images of the potato field in real time, detects the disease status and height information of the plants, uses the built-in disease recognition model and disease classification model, determines the spraying strategy according to the disease type, severity and plant height, and calculates the appropriate pesticide ratio and spraying plan; S3. Monitor the flow data of each medicine box in real time, adjust the output of the extraction pump corresponding to each medicine box according to the information fed back by the flow sensor, control the flow of each liquid medicine, and ensure that the ratio meets the preset ratio; At the same time, the extraction pump is adjusted to control the flow rate according to the travel speed of the spraying robot; In addition, the liquid level of each medicine box is monitored in real time. When a medicine box is running low on liquid, a replenishment reminder is issued, and the liquid ratio is automatically adjusted when liquid is insufficient. S4. During the spraying process, the spraying robot controls the spraying angle and height of the fungicide nozzle and the insecticide nozzle in real time according to the plant height and disease distribution, accurately covers the plant canopy, plant bottom and potato ridge, and realizes the zoning treatment of the upper and lower layer diseases; S5. During the spraying process of the spraying robot, the control device collects the spraying amount data in real time through the flow sensor, and generates a spraying amount map after the spraying of the entire field is completed.

2. The control method of a double-layer field spraying robot for potatoes according to claim 1 is characterized in that: The disease recognition model and disease classification model in S2 are specifically as follows: the disease recognition model adopts the YOLOv10 neural network model; YOLOv10 is used as the basic network of the disease recognition model, and the original backbone feature extraction network is replaced with HGNet. Different convolution types are dynamically selected by introducing the num parameter to replace the ordinary convolution of the HGB lock module in HGNet. When num=1 or num=2, ordinary convolution is replaced by ghost convolution GhostConv; when num=3 or num=4, ordinary convolution is replaced by reparameterized convolution RepConv, and the SPPF module in the backbone network is improved. Before the pooling pyramid operation, a 1×1 convolution layer is added to adjust the number of channels of the input features, and the kernel size of the pooling layer is uniformly set to 5×5, and the pooling operation is performed in a cascade manner. Finally, the channel compression convolution after the output of the pooling pyramid is replaced by grouped convolution, and the generated output features will be passed to the subsequent network layers. The original PI oU loss function in YOLOv10 is replaced by Inner-WloU; The disease grading model uses the image segmentation algorithm U-Net to segment the diseased area, calculate the disease coverage percentage, and divide the disease into grades.

3. The control method of a double-layer field spraying robot for potatoes according to claim 1 is characterized in that: The specific method for generating a spraying strategy is: Determine the type of liquid medicine according to the disease type T, and dynamically adjust the liquid medicine concentration D according to the severity of the disease L: D=D base ·(1+kL) Where D base is the basic solution concentration, k is the concentration adjustment coefficient, and L is the severity level of the disease.

4. The control method of a double-layer field spraying robot for potatoes according to claim 1 is characterized in that: The specific method of generating a spraying map is as follows: During the spraying process, the pesticide flow rate Q is p and bactericide flow Q f Match with GPS location data (x, y) to form a data pair (x, y, Q p ) and (x, y, Q f ), record the flow and location data as a set of time series data; The field is divided into multiple field grids, each grid unit is represented by the coordinates (i, j), and all sampling points (x, y, Q p ) and (x, y, Q f ) is accumulated to the corresponding grid unit (i, j), and the total spraying amount Q of each grid unit is obtained. p总 and Q f总 ; For grids with incomplete spraying data such as uncovered or insufficient spraying amount, the IDW algorithm is used to estimate the spraying amount of insecticides and fungicides. For areas at the edge of the field or at the bend of the path, weighted smoothing is used to adjust the spraying amount in the edge area. According to the dosage of pesticides and fungicides sprayed, the dosage spraying maps are generated respectively through the color mapping formula.

5. A double-layer field spraying robot for potatoes, characterized in that: The double-layer field spraying robot is used to perform the method described in any one of claims 1 to 4 above, and the double-layer field spraying robot comprises: Control device (2): installed at the front end of the walking device (1); Vision system (3): installed on both sides of the front end of the walking device (1); Walking device (1): a medicine box group (6) is installed on the upper middle part, two groups of lower spraying devices (4) are installed on both sides of the lower part, and a lifting device (5) is installed on the tail, and the lifting device (5) is fixedly installed with the rear spraying device (4); The walking device (1) comprises: a chassis (11), a battery pack (12), and a side fixing frame (13); the battery pack (12) is installed at the front end of the chassis (11), and the side fixing frames (13) are respectively installed at the sides; The rear spraying device (4) comprises: a spraying rod frame (41), a spray head assembly (42), and nine groups of spraying assemblies (42) are sequentially mounted on the rear end of the spraying rod frame (41); The nozzle assembly (42) comprises: a rotating mechanism (421), a bactericidal nozzle (422), an insecticide nozzle (423), and a medicine tube (424); the bactericidal nozzle (422) and the insecticide nozzle (423) are sequentially mounted on the lower end of the rotating mechanism; and the upper ends of the bactericidal nozzle (422) and the insecticide nozzle (423) are both mounted with medicine tubes (424); The lifting device (5) comprises: a housing (51), a flow sensor (52), a flow divider (53), an extraction pump (54), a fixed base (55), a medicine tube (424), a medicine liquid mixer (56), and an electric push rod (57). Three extraction pumps (54) are sequentially installed on the front side of the housing (51), a flow sensor (52) is installed above each extraction pump (54), and two medicine liquid mixers (56) are sequentially installed inside the housing. The flow sensors (52) on both sides are connected to the medicine liquid mixer (56) through the medicine tube (424), respectively. Two liquid medicine mixers (56) are connected to the inner side of the housing (51); a flow sensor (52) in the middle is connected to the diverter (53) through a medicine tube (424); two medicine tubes (424) separated from the diverter (53) are respectively connected to the two liquid medicine mixers (56) inside the housing (51); a fixed base (55) is installed on the upper part of the housing (51); medicine tubes (424) are respectively installed below the two liquid medicine mixers (56) and are connected to the fixed base (55); two electric push rods (57) are sequentially installed at the lower end of the fixed base (55); The medicine box group (6) comprises: a sterilization medicine box (61), a water tank (62), and an insecticide medicine box (63). The water tank (62) is installed in the middle of the medicine box group (6), and the sterilization medicine box (61) and the insecticide medicine box (63) are installed on both sides respectively. The sterilization medicine box (61) is connected to the extraction pump (54) installed on the same side of the cover (51) through the medicine pipe (424). Similarly, the insecticide medicine box (63) is connected to the extraction pump (54) installed on the same side of the cover (51) through the medicine pipe (424). The water tank is connected to the extraction pump (54) installed in the middle of the cover (51) through the medicine pipe (424). The lower spraying device (7) comprises: a medicine tube (424), a bactericidal nozzle (422), an insecticide nozzle (423), and an electric slide (71). The two electric slides (71) are installed with their slide rails facing outwards, and the bactericidal nozzle (422) and the insecticide nozzle (423) are installed on the slide in sequence. The ends of the two nozzles are respectively installed with medicine tubes (424).

Images