Poultry house cleaning system and cleaning control method

Abstract

The invention provides a poultry house cleaning system and a cleaning control method, and the method comprises the steps: constructing global path data according to the three-dimensional information of a chicken house region; acquiring chicken coop image data in the advancing direction of the chicken coop cleaning process; inputting the chicken house image data into a smudginess identification model to obtain a target smudginess identification result; on the basis of a preset mapping relation between the smudge classification result and a smudge cleaning mode, according to the target smudge identification result, determining cleaning control data of a region corresponding to the chicken house image data; according to the target smudginess identification result, smudginess area information in the chicken house image data is acquired; performing path planning according to the global path data and the dirty area information so as to drive the walking module to a target cleaning position of the chicken house area; and after it is determined that the walking module is driven to the target cleaning position, corresponding cleaning operation is executed on the target cleaning position according to the cleaning control data. While the henhouse cleaning effect is ensured, the henhouse cleaning efficiency is also improved.

Description

Technical Field

[0001] The present invention relates to the field of intelligent breeding technology, and in particular to a poultry house cleaning system and a cleaning control method. Background Art

[0002] In the chicken farming process, environmental factors play a vital role and directly affect the health status and production performance of chickens.

[0003] In the current chicken house farming practice, cleaning work mainly relies on manual operation. Although this method can maintain the sanitation of the chicken house to a certain extent, it has significant limitations. Manual cleaning is not only time-consuming and labor-intensive, but also inefficient. In the face of the complex distribution of pollutants inside the chicken house, it is often difficult to achieve thorough and immediate cleaning. Although there are some automatic cleaning devices available on the market, most of these devices only have the function of timed and quantitative cleaning, lack flexibility and pertinence, resulting in poor cleaning effect.

[0004] Therefore, there is an urgent need for a poultry house cleaning system and a cleaning control method to solve the above problems. Summary of the invention

[0005] In view of the problems existing in the prior art, the present invention provides a poultry house cleaning system and a cleaning control method.

[0006] The present invention provides a poultry house cleaning system, comprising a laser radar module, a first image acquisition module, a controller module, a walking module and a cleaning module, wherein: The laser radar module is used to collect three-dimensional information about the surrounding environment of the chicken house area and construct global path data based on the obtained three-dimensional information; The first image acquisition module is used to acquire images of the chicken house area in the direction of travel of the walking module during the chicken house cleaning process to obtain chicken house image data; The controller module is used to perform dirt recognition on the chicken house image data, and generate corresponding dirt area information and cleaning control data according to the dirt recognition result; The walking module is used to plan a path according to the global path data and the dirty area information, so as to drive to the target cleaning position of the chicken house area; The cleaning module is used to perform a corresponding cleaning operation on the target cleaning position according to the cleaning control data after the walking module is driven to the target cleaning position.

[0007] According to a poultry house cleaning system provided by the present invention, the cleaning module at least includes a floor cleaning brush device, a cleaning nozzle and a dust suction device; the cleaning control data includes brush control information, nozzle control information and dust suction motor control information, wherein: The floor cleaning brush device is used to clean the target cleaning position according to the cleaning angle corresponding to the brush control information; The cleaning nozzle is used to spray cleaning water on the target cleaning position according to the spraying angle and spraying water pressure corresponding to the nozzle control information; The dust suction device is used to perform dust suction cleaning on the target cleaning position according to the dust suction angle and dust suction wind force corresponding to the dust suction motor control information.

[0008] According to a poultry house cleaning system provided by the present invention, the poultry house cleaning system further comprises a second image acquisition module, the second image acquisition module is used to acquire an image of the target cleaning position after cleaning to obtain cleaning state image data; The controller module is also used to determine whether the cleaning condition of the target cleaning position meets the preset cleaning conditions based on the cleaning status image data, and if not, generate a secondary cleaning instruction for the target cleaning position, wherein the secondary cleaning instruction is used to drive the cleaning module to perform a cleaning operation on the target cleaning position again.

[0009] The present invention also provides a cleaning control method based on the above poultry house cleaning system, comprising: Construct global path data based on the three-dimensional information corresponding to the surrounding environment of the chicken house area; Acquire chicken house image data corresponding to the chicken house area in the direction of travel during the chicken house cleaning process; Inputting the chicken house image data into a dirt recognition model to obtain a target dirt recognition result of the chicken house area output by the dirt recognition model, wherein the dirt recognition model is trained based on sample chicken house image data marked with dirt category labels; Based on a preset mapping relationship between the dirt classification result and the dirt cleaning mode, the cleaning control data of the area corresponding to the chicken house image data is determined according to the target dirt identification result; and the dirt area information in the chicken house image data is obtained according to the target dirt identification result; Path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area; and after determining that the walking module is driven to the target cleaning position, a corresponding cleaning operation is performed on the target cleaning position according to the cleaning control data.

[0010] According to a cleaning control method provided by the present invention, the dirt recognition model is trained by the following steps: Obtain sample chicken house image data for each area of ​​the chicken house; According to the chicken house dirt category in the sample chicken house image, a corresponding dirt category label is constructed to obtain sample chicken house image data marked with the dirt category label, wherein the chicken house dirt category includes at least feed dirt, chicken feather dirt and excrement dirt; Based on the sample chicken house image data marked with the dirt category label, the deep learning model is trained to obtain the dirt recognition model.

[0011] According to a cleaning control method provided by the present invention, after constructing a corresponding dirt category label according to the dirt category of the chicken house in the sample chicken house image and obtaining the sample chicken house image data marked with the dirt category label, the method further includes: Obtaining a training sample set by obtaining obstacle labels corresponding to obstacle marks in the sample chicken house image data of the dirty category label; Based on the sample chicken house image data marked with the dirt category label, the deep learning model is trained to obtain the dirt recognition model, including: Based on the training sample set, the deep learning model is trained to obtain the dirt recognition model; The path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area, including: When it is determined that there is an obstacle between the current position of the walking module and the dirty area according to the dirty area information, generating a corresponding obstacle avoidance path; If the size of the obstacle between the current position of the walking module and the dirty area is larger than the preset avoidance size, a cleaning pause instruction and an obstacle warning prompt message for stopping the cleaning process of the current chicken house area are generated.

[0012] According to a cleaning control method provided by the present invention, after performing a corresponding cleaning operation on the target cleaning position according to the cleaning control data, the method further includes: Acquire the image data of the chicken house after the cleaning operation is performed on the target cleaning position; Inputting the image data of the cleaned chicken house into the dirt recognition model, and if there is residual dirt in the dirt recognition result output by the dirt recognition model, generating corresponding residual dirt control data according to the dirt type of the residual dirt; The target cleaning position is cleaned again according to the residual dirt control data.

[0013] According to a cleaning control method provided by the present invention, the path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area, and further includes: Convert the dirty image coordinate data in the dirty area information into a world coordinate system to obtain the real position coordinate data of the dirty area in the world coordinate system; Path planning is performed according to the global path data and the real position coordinate data, and a driving instruction for driving the walking module to the target cleaning position is generated.

[0014] The present invention also provides an electronic device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein when the processor executes the program, any one of the cleaning control methods described above is implemented.

[0015] The present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the cleaning control method described in any one of the above is implemented.

[0016] The present invention provides a poultry house cleaning system and a cleaning control method, which perform cleaning operations in a three-dimensional layered environment of a chicken house. A navigation and moving map is constructed through a laser radar, and dirt is identified based on the collected chicken house images. The poultry house cleaning system is navigated to the target area to complete the cleaning of each area, thereby ensuring the cleaning effect of the chicken house while also improving the cleaning efficiency of the chicken house. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0018] Figure 1 A schematic diagram of the structure of a poultry house cleaning system provided by the present invention; Figure 2 This is one of the device and layout diagrams of a cleaning robot provided by the present invention; Figure 3 The second schematic diagram of the components and layout of a cleaning robot provided by the present invention; Figure 4 The third schematic diagram of the components and layout of a cleaning robot provided by the present invention; Figure 5 A fourth schematic diagram of the components and layout of a cleaning robot provided by the present invention; Figure 6 A schematic diagram of the cleaning control method provided by the present invention; Figure 7 A schematic diagram of the overall flow of the cleaning control method based on the cleaning robot provided by the present invention; Figure 8 A schematic diagram of world coordinate transformation of the cleaning robot provided by the present invention; Fig. 9 This is a schematic structural diagram of an electronic device provided by the present invention. DETAILED DESCRIPTION

[0019] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described below in conjunction with the drawings of the present invention. Obviously, the described embodiments are 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 are within the scope of protection of the present invention.

[0020] With the vigorous development of the livestock and poultry industry, the scale of poultry farming continues to expand, large-scale poultry farms continue to emerge, and farming methods are becoming increasingly standardized and intensive. Although this trend has brought significant economic benefits, it has also put forward higher requirements for the standardization and healthy farming of chickens. In the process of chicken farming, environmental factors play a vital role in their health and production performance. However, with the continuous improvement of the degree of intensification and industrialization, the internal environment of the chicken house faces many challenges.

[0021] Specifically, problems such as high humidity, complex and diverse sources of dust, slow air flow, and lack of ultraviolet radiation in the chicken house provide favorable conditions for the growth of dust and microorganisms. This not only easily causes respiratory diseases in chickens, but also poses a threat to the health of farm workers. If the dust on the floor and in the environment of the chicken house is not cleaned in time, it will also lead to reduced chicken production and may even cause lung diseases in workers. In extreme cases, if the chickens are in a high-concentration dust environment for a long time, a large number of deaths may occur in a short period of time.

[0022] Therefore, it is particularly important to regularly clean and dust the inside of the chicken house. In particular, the use of unmanned cleaning can not only more effectively ensure the breeding conditions of the chickens, but also improve the breeding welfare of the poultry, which in turn has a positive impact on economic benefits.

[0023] In the current chicken farming process, the cleaning of the chicken house is mainly done manually. However, this traditional manual cleaning method is not only time-consuming and labor-intensive, but also difficult to ensure that pollutants in the chicken house are removed in a timely and thorough manner. Over time, this will lead to a decline in air quality, which in turn affects production efficiency. In order to overcome the limitations of manual cleaning, the present invention provides a cleaning system specifically suitable for the complex environment inside a chicken house, and based on the cleaning system, constructs an intelligent robot for cleaning chicken houses. This robot can replace manual labor to complete cleaning tasks, effectively save time, and significantly improve the problem of low efficiency of manual cleaning. It can better prevent the occurrence of respiratory diseases in chickens, reduce the probability of illness in chickens, and thus increase the economic benefits of chicken farms.

[0024] The present invention performs efficient cleaning operations in a three-dimensional layer-raising environment of a chicken house. It uses lidar technology to construct a navigation mobile map, and then uses a camera to identify dirty areas in the chicken house aisles in turn. Then, through the precise transformation of image coordinates and world coordinates, it can accurately navigate to the target area and complete the cleaning work of each aisle.

[0025] Figure 1 The schematic diagram of the structure of the poultry house cleaning system provided by the present invention is as follows: Figure 1 As shown, the present invention provides a poultry house cleaning system, including a laser radar module 101, a first image acquisition module 102, a controller module 103, a walking module 104 and a cleaning module 105, wherein: The laser radar module 101 is used to collect three-dimensional information of the surrounding environment of the chicken house area and construct global path data based on the obtained three-dimensional information; The first image acquisition module 102 is used to acquire images of the chicken house area in the direction of travel of the walking module 104 during the chicken house cleaning process to obtain chicken house image data; The controller module 103 is used to perform dirt recognition on the chicken house image data, and generate corresponding dirt area information and cleaning control data according to the dirt recognition result; The walking module 104 is used to plan a path according to the global path data and the dirty area information to drive to the target cleaning position of the chicken house area; The cleaning module 105 is used to perform a corresponding cleaning operation on the target cleaning position according to the cleaning control data after the walking module 104 is driven to the target cleaning position.

[0026] In the present invention, a robot for cleaning a chicken house (hereinafter referred to as a cleaning robot) constructed based on a poultry house cleaning system is described. Figure 2 This is one of the device and layout diagrams of a cleaning robot provided by the present invention. Figure 3 The second schematic diagram of the components and layout of a cleaning robot provided by the present invention is as follows: Figure 4 The third schematic diagram of the components and layout of a cleaning robot provided by the present invention is as follows: Figure 5 The fourth schematic diagram of the device and layout of a cleaning robot provided by the present invention can be referred to Figures 2 to 5 As shown, the present invention provides a cleaning robot based on a poultry house cleaning system, the cleaning robot comprises a front camera 1, a 3D laser radar 2, a robot shell 3, a driving wheel 4, a universal wheel 5, a brush opening and closing driver 6, a vacuum cleaner 7, a brush angle adjustment motor 8, a brush 9, a nozzle 10, a charger 11, a water tank and a dust box 12, a dust suction motor 13, a controller module 14, a branching terminal block 15, a dust suction slide 16, an inertial measurement unit (IMU) 17, a battery 18, a driving wheel motor 19, a driving wheel reduction A first image acquisition module 102 is constructed based on the front camera 1; a laser radar module 101 is constructed by a 3D laser radar 2; a cleaning module 105 is constructed by a brush opening and closing driver 6, a vacuum cleaner 7, a brush angle adjustment motor 8, a brush 9, a nozzle 10, a water tank and a dust box 12, a vacuum motor 13 and a vacuum slide 16; a walking module 104 is constructed by a driving wheel 4, a universal wheel 5, an inertial measurement unit 17, a driving wheel motor 19 and a driving wheel reducer 20.

[0027] Specifically, the front camera 1 can collect dirt images for the cleaning robot, and then the controller module 14 can identify dirt and evaluate the road cleanliness of the collected images, and provide visual distance information of dirt; the 3D laser radar 2 is used for map drawing and robot positioning to ensure the navigation and movement of the cleaning robot itself; the robot shell 3 provides support, dust and pollution prevention, waterproof and other functions, protects internal devices and circuits, and avoids collisions; the driving wheel 4 provides support and steering, performs differential movement, and cooperates with the universal wheel 5 to achieve forward and backward movement requirements as needed; the brush opening and closing driver 6 can open and close the brush 9, and control the opening and closing size according to the dirt location; the vacuum cleaner 7 is used to absorb dirt of small mass such as feathers; the nozzle 10 can be used for different types of dirt on the ground The brush 9 is used to clean different dirt on the ground; the nozzle 10 can further clean the chicken house environment; the charger 11 can charge the 48V power supply that provides power for the entire system; the water tank and dust box 12 provide water source and dust collection box; the vacuum motor 13 provides vacuum air pressure; the controller module 14 and the branch terminal block 15 provide power and logic control for the system; the vacuum slide 16 can control the lifting and lowering of the vacuum cleaner 7, reducing it when the cleaning robot is working and raising it when it is not working; the inertial measurement unit 17 can supplement the moving state information of the cleaning robot to the navigation system; the driving wheel motor 19 and the driving wheel reducer 20 provide power and deceleration torque for the cleaning robot.

[0028] In the present invention, the laser radar module 101 can accurately measure the position, shape and distance of each corner, wall, ground and obstacle inside the chicken house by emitting a laser beam and receiving the reflected signal, thereby sensing the three-dimensional information of the chicken house and its surrounding environment. Based on this three-dimensional information, the laser radar module 101 can construct the global path data of the entire chicken house, providing a basis for the autonomous navigation of the cleaning robot. This data not only helps the robot understand the layout of the environment, but also updates the position information in real time during the robot's walking process to ensure that it will not get lost or collide with obstacles.

[0029] The first image acquisition module 102 may be composed of a high-definition camera, which is responsible for capturing image data in the direction of travel of the walking module 104 during the chicken house cleaning process. These image data provide the controller module 103 with intuitive visual information of dirt. The camera can capture various dirt on the ground, such as feces, feathers, feed residues, etc., and transmit these image data to the controller module 103 in real time for further processing. Through image analysis, the system can identify the type, scope and severity of dirt, and provide accurate guidance for subsequent cleaning operations.

[0030] The controller module 103 is responsible for processing the data from each module and making decisions. For the chicken house image data received from the first image acquisition module 102, the controller module 103 will use the image recognition algorithm to identify dirt. Once the dirty area is identified, the controller module 103 will generate corresponding dirty area information and cleaning control data according to the type and degree of dirt. These data include parameters such as the time, strength and method required for cleaning, as well as the specific coordinates of the target cleaning position. The controller module 103 will also plan the optimal walking path for the walking module 104 based on this information and the global path data provided by the laser radar module 101.

[0031] The walking module 104 is the mobile part of the cleaning robot, which performs path planning based on the global path data and dirty area information provided by the controller module 103. The walking module 104 is usually composed of a driving wheel, a universal wheel, a driving wheel motor and a driving wheel reducer. The driving wheel provides support and driving power, and realizes functions such as steering and forward and backward movement through differential motion. The universal wheel increases the flexibility of the robot, allowing it to move freely in a narrow or complex environment. After receiving the command from the controller module, the walking module 104 will drive the cleaning robot to the target cleaning position according to the planned path.

[0032] The cleaning module 105 is responsible for performing specific cleaning operations after the walking module 104 is driven to the target cleaning position. The cleaning module 105 is usually composed of components such as a brush, a vacuum cleaner, a nozzle, a water tank and a dust box, and a vacuum motor. The brush is used to clean the dirt on the ground, and the vacuum cleaner is responsible for absorbing the brushed dirt and dust in the air. The nozzle can spray cleaning liquid or water mist at different angles and pressures according to the type and degree of dirt to further clean the ground. The water tank and the dust box provide cleaning liquid and store the absorbed dirt respectively. The vacuum motor provides the necessary suction for the vacuum cleaner. Under the instruction of the controller module 103, the cleaning module 105 will perform accurate and efficient cleaning operations on the target cleaning position according to the cleaning control data.

[0033] The present invention provides a poultry house cleaning system, which performs cleaning operations in a three-dimensional layered environment of a chicken house. A navigation and moving map is constructed through a laser radar, and dirt is identified based on the collected chicken house images. The poultry house cleaning system is navigated to the target area to complete the cleaning of each area, thereby ensuring the cleaning effect of the chicken house while also improving the cleaning efficiency of the chicken house.

[0034] On the basis of the above embodiment, the cleaning module at least includes a floor cleaning brush device, a cleaning nozzle and a dust suction device; the cleaning control data includes brush control information, nozzle control information and dust suction motor control information, wherein: The floor cleaning brush device is used to clean the target cleaning position according to the cleaning angle corresponding to the brush control information; The cleaning nozzle is used to spray cleaning water on the target cleaning position according to the spraying angle and spraying water pressure corresponding to the nozzle control information; The dust suction device is used to perform dust suction cleaning on the target cleaning position according to the dust suction angle and dust suction wind force corresponding to the dust suction motor control information.

[0035] In the present invention, the cleaning module is responsible for performing specific cleaning tasks after the cleaning robot reaches the target cleaning position. The cleaning module integrates a variety of cleaning tools and technologies to ensure comprehensive and efficient cleaning of the chicken house floor. Specifically, the floor cleaning brush device is mainly used to clean the dirt on the ground. The device is usually composed of a rotating brush or a vibrating brush, and can adjust the cleaning angle according to different cleaning needs. In the present invention, the controller module generates brush control information based on the dirt identification result, which includes parameters such as the cleaning angle and rotation speed that the brush should adopt. Among them, the cleaning angle refers to the angle between the brush and the ground, which determines the cleaning area that the brush can cover and the cleaning effect. By adjusting the cleaning angle, the brush can more effectively remove dirt such as feces, feathers, and feed residues on the ground.

[0036] The cleaning nozzle is mainly used to spray clean water on the ground to soften or dissolve stubborn dirt, which is convenient for subsequent vacuuming or scrubbing. In the present invention, the nozzle control information is an instruction generated by the controller module, which includes parameters such as spraying angle and spraying water pressure, wherein the spraying angle determines the ground area that the cleaning water can cover. By adjusting the spraying angle, it can be ensured that the cleaning water can be sprayed evenly on the target cleaning position to improve the cleaning effect. The spraying water pressure refers to the pressure of the cleaning water when it is sprayed from the nozzle. Different types of dirt require different water pressures for cleaning. For example, for stubborn feces or feed residues, higher water pressure may be required to soften or dissolve them; for lighter feathers or dust, lower water pressure may be required to avoid blowing the dirt away.

[0037] The dust suction device is mainly used to absorb the dirt that is brushed or dissolved and the dust in the air. In the present invention, the dust suction motor control information generated by the controller module includes parameters such as the dust suction angle and the dust suction wind force. The dust suction angle refers to the angle between the dust suction port and the ground. By adjusting the dust suction angle, it can be ensured that the dust suction device can more effectively absorb the dirt on the ground. The dust suction wind force determines how much dirt the dust suction device can absorb. A larger wind force can more effectively absorb heavier dirt or dust from a distance; while a smaller wind force is suitable for light dirt or close-range cleaning.

[0038] The cleaning module in the present invention integrates multiple cleaning tools and technologies such as a floor cleaning brush device, a cleaning nozzle and a dust suction device. It can perform accurate and efficient cleaning operations on the target cleaning position according to the cleaning control data provided by the controller module, fully considering the complexity of the chicken house environment and the diversity of dirt, ensuring the cleaning effect while improving cleaning efficiency and safety.

[0039] On the basis of the above embodiment, the poultry house cleaning system further comprises a second image acquisition module, and the second image acquisition module is used to acquire an image of the target cleaning position after cleaning to obtain cleaning state image data; The controller module is also used to determine whether the cleaning condition of the target cleaning position meets the preset cleaning conditions based on the cleaning status image data, and if not, generate a secondary cleaning instruction for the target cleaning position, wherein the secondary cleaning instruction is used to drive the cleaning module to perform a cleaning operation on the target cleaning position again.

[0040] In the present invention, reference may be made to Figure 2 As shown, the cleaning robot also includes a rear camera 21, through which a second image acquisition module of the poultry house cleaning system is constructed. The second image acquisition module is used to perform another image acquisition on the target cleaning position after the cleaning module completes the first cleaning operation.

[0041] In the present invention, the second image acquisition module is usually composed of a high-definition camera or other image sensors, which can automatically start and capture images of the target cleaning position after the cleaning robot completes a cleaning task. These image data, i.e., cleaning state image data, will serve as an important basis for evaluating the cleaning effect. Specifically, the working time of the second image acquisition module is when the cleaning module completes the cleaning operation and returns or pauses, so as to ensure that the collected image data truly reflects the state of the ground after cleaning.

[0042] After receiving the cleaning status image data provided by the second image acquisition module, the controller module uses an image recognition algorithm or a machine learning model to analyze the cleaning status image data to determine whether the cleanliness of the target cleaning position meets the preset cleaning conditions, which may include no residual feces, feathers and other dirt on the ground, no obvious water stains or cleaning liquid residues, etc. If the cleaning status of the target cleaning position does not meet the preset conditions, the controller module will automatically generate a secondary cleaning instruction, which includes detailed parameters such as the specific coordinates of the target cleaning position, the required cleaning force, the brush angle, and the suction wind force, to ensure that the cleaning module can re-clean the position that does not meet the cleaning conditions in a targeted manner.

[0043] Furthermore, after receiving the secondary cleaning instruction, the cleaning module will immediately enter the standby state and prepare to perform the secondary cleaning task. According to the parameters in the secondary cleaning instruction, the cleaning module will adjust the cleaning angle of the brush, the wind force of the dust suction device, and the spraying angle and water pressure of the cleaning nozzle to ensure that the cleaning effect meets the preset standards. In the present invention, after adjusting the cleaning parameters, the cleaning module will be driven to the target cleaning position again and perform the secondary cleaning operation according to the requirements in the instruction. This process may include steps such as brushing the floor again, spraying cleaning fluid, and vacuuming until the cleaning condition of the target cleaning position meets the preset conditions.

[0044] The present invention enables the poultry house cleaning system to more accurately evaluate the cleaning effect and perform a secondary cleaning operation when necessary by adding the second image acquisition module and related functions, thereby ensuring the cleanliness and hygiene of the chicken house environment.

[0045] Figure 6 A schematic diagram of the cleaning control method provided by the present invention is shown in FIG. Figure 6 As shown, the present invention provides a cleaning control method of the poultry house cleaning system based on the above embodiments, comprising: Step 601, constructing global path data according to the three-dimensional information corresponding to the surrounding environment of the chicken house area; Step 602, obtaining chicken house image data corresponding to the chicken house area in the direction of travel during the chicken house cleaning process; Step 603, inputting the chicken house image data into a dirt recognition model to obtain a target dirt recognition result of the chicken house area output by the dirt recognition model, wherein the dirt recognition model is trained based on sample chicken house image data marked with dirt category labels; Step 604, based on a preset mapping relationship between the dirt classification result and the dirt cleaning mode, and according to the target dirt identification result, determining the cleaning control data of the area corresponding to the chicken house image data; and obtaining the dirt area information in the chicken house image data according to the target dirt identification result; Step 605, perform path planning based on the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area; and after determining that the walking module is driven to the target cleaning position, perform corresponding cleaning operations on the target cleaning position according to the cleaning control data.

[0046] In the present invention, the laser radar module obtains the surrounding distance information for the poultry house cleaning system, and the first image acquisition module transmits image information for the poultry house cleaning system, senses the road condition information ahead, establishes a three-dimensional global map, and identifies dirt and other information when the poultry house cleaning system is moving. In the present invention, radar information, image information, posture information and other data will be periodically transmitted to the controller module in real time for analysis and processing, and navigation control will be performed according to the target posture results of image recognition, and the speed and direction of the vehicle body will be adjusted to navigate to the cleaning destination. When the poultry house cleaning system arrives at the destination, the cleaning brush and nozzle will be turned on for cleaning purposes.

[0047] Specifically, before the chicken house cleaning task begins, the poultry house cleaning system first needs to construct global path data based on the three-dimensional information corresponding to the surrounding environment of the chicken house area. This step usually involves the following operations: Three-dimensional information acquisition: Through laser radar, depth camera and other sensor equipment, the interior and surrounding environment of the chicken house are scanned to obtain high-precision three-dimensional point cloud data.

[0048] Map construction: Use SLAM (Simultaneous Localization and Mapping) algorithm or similar technology to build a three-dimensional map of the chicken house based on the acquired three-dimensional point cloud data.

[0049] Global path planning: Based on the constructed three-dimensional map, the cleaning robot plans a global path from the starting point to each target cleaning location according to information such as the layout of the chicken house and the location of obstacles.

[0050] Furthermore, while the cleaning robot is moving along the global path, the poultry house cleaning system needs to obtain image data corresponding to the chicken house area in real time. This step usually involves the following operations: Using a high-definition camera or other image sensor installed on the cleaning robot, images of the chicken house area in the direction of travel are collected.

[0051] The collected image data is preprocessed, such as denoising, contrast enhancement, etc., to improve the accuracy of subsequent dirt recognition.

[0052] Furthermore, the preprocessed chicken house image data is input into the dirt recognition model to obtain the target dirt recognition result. In the present invention, the dirt recognition model is trained based on a large amount of sample chicken house image data marked with dirt category labels, and the dirt recognition model can recognize and classify the types of dirt in the image, such as feces, feathers, feed residues, etc. In the present invention, the preprocessed image data is input into the model, and the model outputs the target dirt recognition result to obtain information such as the type, location, area, etc. of the dirt in the image.

[0053] Furthermore, based on the preset mapping relationship between the dirt classification result and the dirt cleaning mode, the cleaning control data is determined according to the target dirt recognition result, and the dirt area information is obtained. In the present invention, the corresponding cleaning control data is determined according to the dirt type, area and other information, combined with the preset cleaning strategy (such as different cleaning forces, brush angles, dust suction wind force, etc. for different types of dirt). At the same time, the specific location, shape, size and other information of the dirty area are extracted from the target dirt recognition result to provide a basis for subsequent path planning and cleaning operations.

[0054] Furthermore, the optimal path from the current position to the target cleaning position is planned by combining the global path data and the dirty area information. In the present invention, the walking module of the cleaning robot is controlled to move to the target cleaning position according to the planned path. After reaching the target cleaning position, the cleaning module is controlled to perform corresponding cleaning operations, such as scrubbing the floor, vacuuming, spraying cleaning fluid, etc., according to the previously determined cleaning control data.

[0055] Figure 7 The overall flow diagram of the cleaning control method based on the cleaning robot provided by the present invention can be referred to Figure 7 As shown, in the present invention, after the cleaning robot body is started, the poultry house cleaning system first checks whether all devices are connected correctly, otherwise it cannot be started. When it is determined that there is no problem, each sensor device will be started in turn to collect various data.

[0056] In the present invention, the poultry house cleaning system waits for the staff to input the target point or target path. At this time, although the sensors of the vehicle body have started working, the whole body has not moved. When the poultry house cleaning system reads the input, it will first generate a global path according to the previously built global map planning. It should be noted here that if it is used for the first time, the global map can be first established through the laser radar module and saved locally for easy loading and use later.

[0057] After the path is generated, the industrial computer will publish the corresponding vehicle speed information and transmit it to the motor driver through the serial port for wheel drive control, and realize acceleration, deceleration and direction change through the speed difference. During the movement, each sensor will transmit the corresponding information to the industrial computer. The industrial computer will correct the previously planned global path in real time according to the road cleaning situation, and transmit the corrected vehicle speed to the microcontroller to realize the road cleaning work of the cleaning robot. While cleaning the road, the rear camera will remain on to record and measure the cleaning situation of this road. If it does not meet the requirements, it will stop and reverse to the dirty area for secondary cleaning until the indicators are met, and then enter the next aisle for cleaning until the cleaning is completed.

[0058] The present invention provides a cleaning control method for performing cleaning operations in a three-dimensional layered environment of a chicken house. A navigation and moving map is constructed through a laser radar, and dirt is then identified based on the collected chicken house images. The poultry house cleaning system is navigated to the target area to complete the cleaning of each area, thereby ensuring the cleaning effect of the chicken house while also improving the cleaning efficiency of the chicken house.

[0059] Based on the above embodiment, the dirt recognition model is trained by the following steps: Obtain sample chicken house image data for each area of ​​the chicken house; According to the chicken house dirt category in the sample chicken house image, a corresponding dirt category label is constructed to obtain sample chicken house image data marked with the dirt category label, wherein the chicken house dirt category includes at least feed dirt, chicken feather dirt and excrement dirt; Based on the sample chicken house image data marked with the dirt category label, the deep learning model is trained to obtain the dirt recognition model.

[0060] In the present invention, a large amount of sample chicken house image data is first collected. These data can cover various areas inside the chicken house, including but not limited to walkways, feed areas, drinking areas, rest areas, etc., to ensure the generalization ability of the model. In the present invention, the sample chicken house image data can be obtained from multiple channels, such as field photography, existing surveillance videos, and other public data sets of similar chicken houses.

[0061] In order to ensure the accuracy of the model, the collected sample images should be diverse, including different lighting conditions, different angles, different levels of dirtiness, etc. After collecting the original images, some preprocessing work is required, such as cropping, scaling, denoising, etc., to facilitate subsequent processing and analysis.

[0062] Furthermore, after the sample images are collected, it is necessary to construct a corresponding dirt category label for each image according to the dirt category of the chicken house in the image. According to the actual needs of chicken house cleaning, the present invention divides the dirt categories into at least feed dirt, chicken feather dirt and excrement dirt. These categories can be further subdivided according to actual conditions. For example, the dirt category also includes broken eggs, etc.

[0063] After obtaining the sample chicken house image data marked with the dirty category label, select the appropriate deep learning model according to the specific needs of dirt recognition and the characteristics of the sample data, such as convolutional neural network (CNN), residual network (ResNet), target detection network (such as YOLO, Faster R-CNN), etc. Then, input the labeled sample image data into the model, calculate the loss value through forward propagation, and then adjust the model parameters through back propagation until the performance of the model on the validation set is stable or optimal. During the training process, some hyperparameters such as learning rate, batch size, number of iterations, etc. may need to be adjusted to optimize the performance of the model. After training, the model needs to be evaluated on the test set to check its generalization ability and accuracy. The evaluation indicators can include accuracy, recall rate, F1 score, etc.

[0064] After the training and evaluation of the above steps, a dirt recognition model is obtained. This model can receive new chicken house images as input and output the category and location information of dirt in the image. Finally, the trained model can be deployed on chicken house cleaning robots or other equipment that requires dirt recognition to achieve real-time dirt detection and classification. In practical applications, new sample data can be continuously collected, and the model can be continuously optimized and updated to adapt to the ever-changing chicken house environment and dirt types.

[0065] On the basis of the above embodiment, after constructing a corresponding dirt category label according to the dirt category of the chicken house in the sample chicken house image and obtaining the sample chicken house image data marked with the dirt category label, the method further includes: Obtaining a training sample set by obtaining obstacle labels corresponding to obstacle marks in the sample chicken house image data of the dirty category label; Based on the sample chicken house image data marked with the dirt category label, the deep learning model is trained to obtain the dirt recognition model, including: Based on the training sample set, the deep learning model is trained to obtain the dirt recognition model; The path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area, including: When it is determined that there is an obstacle between the current position of the walking module and the dirty area according to the dirty area information, generating a corresponding obstacle avoidance path; If the size of the obstacle between the current position of the walking module and the dirty area is larger than the preset avoidance size, a cleaning pause instruction and an obstacle warning prompt message for stopping the cleaning process of the current chicken house area are generated.

[0066] In the present invention, in the process of building the dirt recognition model, in addition to marking the dirt categories in the sample chicken house image data, it is also necessary to mark the obstacles in the image and assign corresponding obstacle labels to them. Obstacles may include fixed obstacles such as equipment, fences, chicken cages, etc. in the chicken house, as well as temporary mobile obstacles such as chickens, staff, etc.

[0067] The present invention combines sample chicken house image data marked with dirt category labels and obstacle labels to form a complete training sample set, which will be used to train the deep learning model to ensure that the model can simultaneously identify dirt and obstacles. After training, a deep learning model that can identify dirt and obstacles is obtained, namely, a dirt recognition model. This model will be used in subsequent chicken house cleaning tasks to detect dirt and obstacles in real time.

[0068] In the chicken house cleaning task, path planning is required based on the global path data and dirty area information to drive the walking module to the target cleaning position. However, obstacles may be encountered in the actual process, so obstacle avoidance and corresponding processing are required. During the path planning process, the dirt recognition model will detect in real time whether there is an obstacle between the current position of the walking module and the dirty area. If there is an obstacle, the model will generate a corresponding obstacle avoidance path to ensure that the walking module can bypass the obstacle and reach the target cleaning position.

[0069] For detected obstacles, the model also needs to determine whether its size is larger than the preset avoidance size. If the obstacle is too large to be safely avoided, the model will generate a cleaning pause instruction and issue an obstacle warning prompt message, which will notify the operator or system administrator to deal with the obstacle in time to ensure the safe progress of the chicken house cleaning task.

[0070] After avoiding or dealing with obstacles, the model will re-plan the path and drive the walking module to the target cleaning position along the new path. After reaching the target position, the system will perform the corresponding cleaning operation according to the dirt category and location information output by the dirt recognition model to ensure the safe and efficient cleaning of the chicken house.

[0071] In the present invention, the floor inside the chicken house is made of black-gray cement. The main sources of dirt are yellow powdered feed and chicken feathers, followed by a small amount of chicken manure and broken eggs from broken chicken cages. Therefore, according to the above dirt recognition model, during the cleaning process, the cleaning robot can dynamically adjust the cleaning strategy through an intelligent algorithm according to the identified dirt category, automatically increase the cleaning time, adjust the nozzle angle and water pressure, and ensure comprehensive cleaning coverage. At the same time, if the operator leaves tools or other obstacles in the chicken house that are not suitable for cleaning, the cleaning robot will send an alarm to the operating platform, stop cleaning, and wait for the tools to be taken away before cleaning again.

[0072] Table 1 Treatment methods for different types of dirt

[0073] For example, when there is powdered feed on the ground, the controller module will judge that the feed is dirty by the pixel grayscale difference, and judge the distance information and the depth of the dirt by combining the depth data. The robot will be controlled to go to the area for cleaning by combining the distance information with the world coordinate transformation, and the cleaning angle of the brush, the wind force of the vacuum cleaner, the water outlet angle, pressure and dwell time of the nozzle will be controlled by the depth of the dirt to complete the cleaning. If obstacles such as hammers and screwdrivers appear, the cleaning will be stopped and an alarm will be sent to remind the operator. Table 1 shows the cleaning methods for various dirt. For specific cleaning methods, please refer to Table 1.

[0074] After confirming that each cleaning step is completed, the cleaning robot uses the rear camera to conduct real-time inspection and evaluation, and evaluates the cleaning effect through image comparison. For areas that do not meet the standards, the system will control the robot to reverse and perform a second cleaning.

[0075] On the basis of the above embodiment, after performing a corresponding cleaning operation on the target cleaning position according to the cleaning control data, the method further includes: Acquire the image data of the chicken house after the cleaning operation is performed on the target cleaning position; Inputting the image data of the cleaned chicken house into the dirt recognition model, and if there is residual dirt in the dirt recognition result output by the dirt recognition model, generating corresponding residual dirt control data according to the dirt type of the residual dirt; The target cleaning position is cleaned again according to the residual dirt control data.

[0076] In the present invention, in the actual chicken house scene, there are multiple types of dirt mixed. At this time, the identified dirt type is superimposed on part of the image. At this time, the robot will turn on all cleaning devices and adjust all intensities to medium for cleaning. After the first cleaning is completed, the system will re-evaluate through the rear camera. If it is unqualified, the dirt type will be identified again. If it was cleaned to a single type last time, it will be treated as a single type. If there are still multiple types of dirt, the cleaning intensity of various cleaning equipment will be increased until the cleaning is completed. This operation is repeated many times. If the cleaning is not completed after the threshold is met, the location will be marked as abnormal and sent to the operating platform to remind the operator. The specific adjustment formula is as follows: The state vector containing all control variables, where is the moving speed of the cleaning robot, is the brush speed, It's wind pressure. It's the water pressure. It is the water outlet angle.

[0077] ; A vector containing the basis values ​​of all variables: ; , are the proportional, integral, and derivative gain coefficient matrices for each control variable, for example: ; Accordingly, The corresponding gain coefficient matrix can be found in .

[0078] The error vector, defined as the difference between the target and actual values ​​of each controlled variable: ; The control matrix equation of the poultry house cleaning system is: .

[0079] In the present invention, after each cleaning is completed, the system will produce a clean cloud map and a dirty map based on the cleaning situation, and record the location of each cleaning to determine whether there is a damaged chicken coop where chicken manure and broken eggs appear, and whether there is a fight between chickens where there are too many chicken feathers. The predicted situation will then be recorded in the local system and uploaded to the cloud system to prompt the producer whether it is necessary to maintain the production structure.

[0080] On the basis of the above embodiment, the path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area, and further includes: Convert the dirty image coordinate data in the dirty area information into a world coordinate system to obtain the real position coordinate data of the dirty area in the world coordinate system; Path planning is performed according to the global path data and the real position coordinate data, and a driving instruction for driving the walking module to the target cleaning position is generated.

[0081] Figure 8 The world coordinate transformation diagram of the cleaning robot provided by the present invention can be referred to Figure 8 As shown, the present invention abstracts and simplifies the cleaning robot, removes the complex structure, abstracts each sensor into a coordinate system to simplify the robot coordinate system, and calculates the movement through the transformation relationship between each coordinate system, including the transformation of position and posture.

[0082] Specifically, the industrial computer obtains coordinates through the model imported by SolidWorks software, obtains the position data of each sensor, takes the forward direction of the cleaning robot as the X-axis, and establishes a three-dimensional space coordinate system according to the right-hand spiral rule. After the coordinate transformation of the sensor data, the relative information of each position of the cleaning robot can be obtained. Figure 8 As shown in the figure, the front camera is built at the front of the robot. Through visual recognition, it is found that there is dirt at (0.4, -0.3, -0.3) in front of the front camera as the origin. According to the coordinate transformation of the robot model, the coordinate relationship between the robot cleaning center base_link and the camera is (0.4, 0, 0.1). Through the coordinate transformation, the distance from the cleaning center to the origin can be obtained as (0.8, -0.3, -0.2). At this time, after obtaining the dirty coordinate point, the robot can be controlled to go to the dirty place for cleaning.

[0083] In the present invention, the front camera will continuously collect images when working, and the collected RGB images and depth images will be processed separately. The RGB image taken by the front camera will be used for dirt recognition through the target detection model (i.e., dirt recognition model). After the model recognizes the dirt, it inputs the position of the detection frame into the depth image, builds a bridge of the pixel coordinates of the image within the range through the physical size, performs perspective projection transformation to the image physical coordinates, and then constructs a rigid body transformation of the projection relationship through the internal parameters to form the camera coordinates. Finally, the rotation and translation relationship is constructed according to the external parameters of the camera to obtain the world coordinate system of the camera, which corresponds to the above-mentioned (0.4, -0.3, -0.3).

[0084] Fig. 9A schematic diagram of the structure of an electronic device provided by the present invention, such as Fig. 9 As shown, the electronic device may include: a processor (Processor) 901, a communication interface (Communications Interface) 902, a memory (Memory) 903 and a communication bus 904, wherein the processor 901, the communication interface 902, and the memory 903 communicate with each other through the communication bus 904. The processor 901 can call the logic instructions in the memory 903 to execute the cleaning control method, which includes: constructing global path data according to the three-dimensional information corresponding to the surrounding environment of the chicken house area; obtaining chicken house image data corresponding to the chicken house area in the direction of travel during the chicken house cleaning process; inputting the chicken house image data into the dirt recognition model to obtain the target dirt recognition result of the chicken house area output by the dirt recognition model, wherein the dirt recognition model is trained based on sample chicken house image data marked with dirt category labels; based on the preset mapping relationship between the dirt classification result and the dirt cleaning mode, according to the target dirt recognition result, determining the cleaning control data of the area corresponding to the chicken house image data; and according to the target dirt recognition result, obtaining the dirty area information in the chicken house image data; performing path planning according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area; and after determining that the walking module is driven to the target cleaning position, performing the corresponding cleaning operation on the target cleaning position according to the cleaning control data.

[0085] In addition, the logic instructions in the above-mentioned memory 903 can be implemented in the form of a software functional unit and can be stored in a computer-readable storage medium when it is sold or used as an independent product. Based on such an understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product, and the computer software product is stored in a storage medium, including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0086] On the other hand, the present invention also provides a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium, and the computer program includes program instructions. When the program instructions are executed by a computer, the computer can execute the cleaning control method provided by the above methods, and the method includes: constructing global path data according to the three-dimensional information corresponding to the surrounding environment of the chicken house area; obtaining chicken house image data corresponding to the chicken house area in the direction of travel of the chicken house cleaning process; inputting the chicken house image data into a dirt recognition model to obtain a target dirt recognition result of the chicken house area output by the dirt recognition model, wherein the The dirt recognition model is trained based on sample chicken house image data marked with dirt category labels; based on the preset mapping relationship between the dirt classification result and the dirt cleaning mode, the cleaning control data of the area corresponding to the chicken house image data is determined according to the target dirt recognition result; and according to the target dirt recognition result, the dirty area information in the chicken house image data is obtained; path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area; and after determining that the walking module is driven to the target cleaning position, the corresponding cleaning operation is performed on the target cleaning position according to the cleaning control data.

[0087] On the other hand, the present invention also provides a non-transitory computer-readable storage medium having a computer program stored thereon, which is implemented when the processor executes the cleaning control method provided in the above-mentioned embodiments, the method comprising: constructing global path data according to the three-dimensional information corresponding to the surrounding environment of the chicken house area; obtaining chicken house image data corresponding to the chicken house area in the direction of travel during the chicken house cleaning process; inputting the chicken house image data into a dirt recognition model to obtain a target dirt recognition result of the chicken house area output by the dirt recognition model, wherein the dirt recognition model is based on a target dirt recognition result marked with a dirt category label. The method is based on the sample chicken house image data trained; based on the preset mapping relationship between the dirt classification result and the dirt cleaning mode, the cleaning control data of the area corresponding to the chicken house image data is determined according to the target dirt recognition result; and according to the target dirt recognition result, the dirty area information in the chicken house image data is obtained; according to the global path data and the dirty area information, the path planning is carried out to drive the walking module to the target cleaning position of the chicken house area; and after determining that the walking module is driven to the target cleaning position, according to the cleaning control data, the corresponding cleaning operation is performed on the target cleaning position.

[0088] The device embodiments described above are merely illustrative, wherein the units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the scheme of this embodiment. Ordinary technicians in this field can understand and implement it without paying creative labor.

[0089] Through the description of the above implementation methods, those skilled in the art can clearly understand that each implementation method can be implemented by means of software plus a necessary general hardware platform, and of course, can also be implemented by hardware. Based on this understanding, the above technical solution is essentially or the part that contributes to the prior art can be embodied in the form of a software product, and the computer software product can be stored in a computer-readable storage medium, such as ROM / RAM, a disk, an optical disk, etc., including a number of instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to execute the methods described in each embodiment or some parts of the embodiments.

[0090] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or make equivalent replacements for some of the technical features therein. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A poultry house cleaning system, characterized in that: It includes a laser radar module, a first image acquisition module, a controller module, a walking module and a cleaning module, wherein: The laser radar module is used to collect three-dimensional information about the surrounding environment of the chicken house area and construct global path data based on the obtained three-dimensional information; The first image acquisition module is used to acquire images of the chicken house area in the direction of travel of the walking module during the chicken house cleaning process to obtain chicken house image data; The controller module is used to perform dirt recognition on the chicken house image data, and generate corresponding dirt area information and cleaning control data according to the dirt recognition result; The walking module is used to plan a path according to the global path data and the dirty area information, so as to drive to the target cleaning position of the chicken house area; The cleaning module is used to perform a corresponding cleaning operation on the target cleaning position according to the cleaning control data after the walking module is driven to the target cleaning position.

2. The poultry house cleaning system according to claim 1, characterized in that: The cleaning module at least includes a floor cleaning brush device, a cleaning nozzle and a dust suction device; the cleaning control data includes brush control information, nozzle control information and dust suction motor control information, wherein: The floor cleaning brush device is used to clean the target cleaning position according to the cleaning angle corresponding to the brush control information; The cleaning nozzle is used to spray cleaning water on the target cleaning position according to the spraying angle and spraying water pressure corresponding to the nozzle control information; The dust suction device is used to perform dust suction cleaning on the target cleaning position according to the dust suction angle and dust suction wind force corresponding to the dust suction motor control information.

3. The poultry house cleaning system according to claim 2, characterized in that: The poultry house cleaning system further comprises a second image acquisition module, which is used to acquire images of the target cleaning position after cleaning to obtain cleaning status image data; The controller module is also used to determine whether the cleaning condition of the target cleaning position meets the preset cleaning conditions based on the cleaning status image data, and if not, generate a secondary cleaning instruction for the target cleaning position, wherein the secondary cleaning instruction is used to drive the cleaning module to perform a cleaning operation on the target cleaning position again.

4. A cleaning control method based on the poultry house cleaning system according to any one of claims 1 to 3, characterized in that: include: Construct global path data based on the three-dimensional information corresponding to the surrounding environment of the chicken house area; Acquire chicken house image data corresponding to the chicken house area in the direction of travel during the chicken house cleaning process; Inputting the chicken house image data into a dirt recognition model to obtain a target dirt recognition result of the chicken house area output by the dirt recognition model, wherein the dirt recognition model is trained based on sample chicken house image data marked with dirt category labels; Based on a preset mapping relationship between the dirt classification result and the dirt cleaning mode, the cleaning control data of the area corresponding to the chicken house image data is determined according to the target dirt identification result; and the dirt area information in the chicken house image data is obtained according to the target dirt identification result; Path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area; and after determining that the walking module is driven to the target cleaning position, a corresponding cleaning operation is performed on the target cleaning position according to the cleaning control data.

5. The cleaning control method according to claim 4, characterized in that: The dirt recognition model is trained by the following steps: Obtain sample chicken house image data for each area of ​​the chicken house; According to the chicken house dirt category in the sample chicken house image, a corresponding dirt category label is constructed to obtain sample chicken house image data marked with the dirt category label, wherein the chicken house dirt category includes at least feed dirt, chicken feather dirt and excrement dirt; Based on the sample chicken house image data marked with the dirt category label, the deep learning model is trained to obtain the dirt recognition model.

6. The cleaning control method according to claim 5, characterized in that: After constructing a corresponding dirt category label according to the dirt category of the chicken house in the sample chicken house image and obtaining the sample chicken house image data marked with the dirt category label, the method further includes: Obtaining a training sample set by obtaining obstacle labels corresponding to obstacle marks in the sample chicken house image data of the dirty category label; Based on the sample chicken house image data marked with the dirt category label, the deep learning model is trained to obtain the dirt recognition model, including: Based on the training sample set, the deep learning model is trained to obtain the dirt recognition model; The path planning is performed according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area, including: When it is determined that there is an obstacle between the current position of the walking module and the dirty area according to the dirty area information, generating a corresponding obstacle avoidance path; If the size of the obstacle between the current position of the walking module and the dirty area is larger than the preset avoidance size, a cleaning pause instruction and an obstacle warning prompt message for stopping the cleaning process of the current chicken house area are generated.

7. The cleaning control method according to claim 5, characterized in that: After performing a corresponding cleaning operation on the target cleaning position according to the cleaning control data, the method further includes: Acquire the image data of the chicken house after the cleaning operation is performed on the target cleaning position; Inputting the image data of the cleaned chicken house into the dirt recognition model, and if there is residual dirt in the dirt recognition result output by the dirt recognition model, generating corresponding residual dirt control data according to the dirt type of the residual dirt; The target cleaning position is cleaned again according to the residual dirt control data.

8. The cleaning control method according to claim 4, characterized in that: The method of performing path planning according to the global path data and the dirty area information to drive the walking module to the target cleaning position of the chicken house area also includes: Convert the dirty image coordinate data in the dirty area information into a world coordinate system to obtain the real position coordinate data of the dirty area in the world coordinate system; Path planning is performed according to the global path data and the real position coordinate data, and a driving instruction for driving the walking module to the target cleaning position is generated.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the cleaning control method according to any one of claims 4 to 8 is implemented.

10. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the cleaning control method according to any one of claims 4 to 8 is implemented.

Images