Spray control method and system for a spray robot

Abstract

The present application discloses a spray control method and system for a spray robot, which relates to the technical field of spray robot control. The method determines a target planning path within a spray area, controls the spray robot to move along the target planning path, and collects real-time images on the target planning path in real time during the movement. Then, a target recognition model pre-installed on the spray robot is used to recognize the real-time image on the target planning path to determine the real-time target on the target planning path. Finally, based on the real-time target, the spray robot is controlled to spray during the movement, thereby realizing on-demand spraying, improving the utilization rate of liquid medicine or cleaning agent, reducing waste, and ensuring that the target area is fully and evenly covered. In addition, a machine learning algorithm is used to enable the spray robot to adapt to complex and changeable working environments, independently decide on the optimal spray strategy, and improve the intelligence level of the robot.

Description

Technical Field

[0001] The present application relates to the technical field of spray robot control, and in particular to a spray control method and system for a spray robot. Background Art

[0002] Spray robots are automated devices primarily used in agriculture, sanitation, and environmental management. They precisely spray liquids, such as pesticides and disinfectants, onto targeted areas based on preset programs or real-time sensor data. Their high efficiency and low labor intensity significantly reduce manual operations and improve work quality. Their built-in intelligent control systems optimize spray paths and dosage to ensure even coverage, avoiding waste and environmental pollution. Furthermore, some spray robots feature remote monitoring and automatic obstacle avoidance, enhancing safety and convenience. Spray robots are becoming indispensable tools in modern agriculture and public health. With the advancement of automation technology, spray robots are widely used in a variety of fields, including indoor air purification, space disinfection, precision agricultural pesticide application, and gardening maintenance. Traditional spray robots typically rely on preset programs or simple sensor feedback for spraying, which has several drawbacks: low spray efficiency and an inability to adjust spray volume based on specific conditions in the target area (such as pollution levels, pest and disease density, and plant growth status). This can lead to wasted pesticides or detergents, or inadequate dosage that fails to achieve the desired effect. Uneven coverage: Fixed patterns are difficult to adapt to complex and changing working environments, which may result in repeated spraying in some areas and insufficient spraying in others. Low intelligence: The inability to achieve true on-demand and precise spraying limits the robot's potential for application in delicate work scenarios. Summary of the Invention

[0003] The present application provides a spray control method and system for a spray robot, aiming to solve the problems of low spray efficiency, uneven coverage and / or low intelligence in the prior art.

[0004] A first aspect of the present application provides a spray control method for a spray robot, comprising:

[0005] Obtaining the spray area range input by the staff, and performing path planning within the spray area range to determine the target planning path within the spray area range;

[0006] Controlling the spray robot to move along the target planned path within the spray area, and collecting real-time images on the target planned path in real time during the movement;

[0007] Using a target recognition model pre-installed on the spray robot, the real-time image on the target planning path is recognized to determine the real-time target on the target planning path;

[0008] Based on the real-time target on the target planning path, the spray robot is controlled to spray during the movement, thereby completing the spray control of the spray robot.

[0009] In a possible implementation, obtaining a spray area range input by a worker, performing path planning within the spray area range, and determining a target planning path within the spray area range include:

[0010] Obtaining the spray area range input by the staff; wherein the spray area range is the spray area input on the electronic map, and the spray area range is a polygon;

[0011] Determine the longest side of the spray area to obtain a target side;

[0012] Rotating the spray area so that the target side is vertical to obtain a rotated spray area;

[0013] Based on the scanning width of the spray robot, the spray area after rotation is divided into multiple sub-areas using vertical lines; wherein, during the process of dividing the sub-areas, if the width of the remaining area is smaller than the scanning width of the spray robot, the remaining area is directly used as the sub-area;

[0014] The vertical center lines of each sub-area are determined, and the vertical center lines are connected in an S-shape to obtain the target planning path within the spray area.

[0015] In one possible embodiment, controlling the spray robot to move along a target planned path within the spray area and collecting real-time images on the target planned path in real time during the movement includes:

[0016] Determine the two endpoints of the target planning path within the spray area, and determine the endpoint closest to the spray robot to obtain the spray starting point;

[0017] Controlling the spray robot to reach the spray starting point and move along the target planned path within the spray area;

[0018] During the movement of the spray robot, a real-time image on the target planning path is collected by a camera carried by the spray robot.

[0019] In a possible implementation, the target recognition model pre-installed on the spray robot is set to a YOLO v3 model.

[0020] In a possible embodiment, the method for loading the target recognition model pre-installed on the spray robot is:

[0021] Initializing model parameters of the target recognition model, encoding the initialized model parameters into vectors to obtain vector codes, and obtaining multiple different vector codes;

[0022] Obtaining loss function values ​​for all vector encodings, and determining an optimal encoding based on the loss function values;

[0023] According to the optimal code, adopting an adaptive variable speed search strategy to perform a local search on the vector code, and determining the vector code after the local search;

[0024] Performing a local interpolation search on the vector code after the local search using a multi-code interpolation search strategy to determine the vector code after the local interpolation search;

[0025] Performing a global search on the vector code after the local interpolation search using a global information mutation strategy to determine the vector code after the global search;

[0026] Repeating the adaptive speed change search strategy, the multi-code interpolation search strategy, and the global information mutation strategy until the number of training times reaches a preset maximum number of training times, and re-obtaining the optimal code to obtain the target vector code;

[0027] The model parameters in the target vector encoding are used as the final model parameters of the target recognition model, and the target recognition model is deployed on the spray robot.

[0028] In a possible implementation, performing a local search on the vector code using an adaptive variable speed search strategy according to the optimal code, and determining the vector code after the local search includes:

[0029] Based on the current number of training times, an adaptive speed change factor is obtained, and a speed change search speed is determined according to the adaptive speed change factor and the optimal encoding;

[0030] According to the variable speed search speed and the optimal code, a local search is performed on the vector code to determine the vector code after the local search.

[0031] In a possible implementation, a multi-code interpolation search strategy is used to perform a local interpolation search on the vector code after the local search, and determining the vector code after the local interpolation search includes:

[0032] Obtaining a first neighborhood code and a second neighborhood code of the vector code after the local search;

[0033] The vector code after the local search is interpolated according to the optimal code, the first neighborhood code, and the second neighborhood code to obtain the vector code after the local interpolation search.

[0034] In a possible implementation, a global information mutation strategy is used to perform a global search on the vector code after the local interpolation search, and determining the vector code after the global search includes:

[0035] Obtaining the amount of information integration, and determining comprehensive integrated information based on the amount of information integration;

[0036] Performing a global search on the vector code after the local interpolation search based on the comprehensive integration information and a historical optimal value of the vector code after the local interpolation search to obtain a global search code;

[0037] Based on the global search code, a probabilistic inferior solution acceptance method is adopted to obtain the vector code after the global search.

[0038] In a possible implementation, based on the real-time target on the target planning path, controlling the spray robot to spray during movement includes:

[0039] Based on the real-time target on the target planning path, query the association relationship table between the target and the spray amount to determine the spray amount corresponding to the real-time target;

[0040] According to the spray volume corresponding to the real-time target, the spray robot is controlled to spray during the movement.

[0041] A second aspect of the present application provides a spray control system for a spray robot, comprising: a path planning module, a travel control module, a target recognition module, and a spray control module;

[0042] The path planning module is used to obtain the spray area range input by the staff, perform path planning within the spray area range, and determine the target planning path within the spray area range;

[0043] The travel control module is used to control the spray robot to travel along the target planning path within the spray area, and to collect real-time images on the target planning path in real time during the travel process;

[0044] The target recognition module is used to recognize the real-time image on the target planning path using the target recognition model pre-installed on the spray robot, and determine the real-time target on the target planning path;

[0045] The spray control module is used to control the spray robot to spray during the movement based on the real-time target on the target planning path, thereby completing the spray control of the spray robot.

[0046] Beneficial effects:

[0047] The present application provides a spray control method and system for a spray robot, which performs path planning within the spray area, determines the target planning path within the spray area, and controls the spray robot to move along the target planning path within the spray area, and collects real-time images on the target planning path in real time during the movement; then uses the target recognition model pre-installed on the spray robot to recognize the real-time image on the target planning path and determine the real-time target on the target planning path; finally, based on the real-time target on the target planning path, controls the spray robot to spray during the movement, thereby realizing on-demand spraying, improving the utilization rate of liquid medicine or detergent, reducing waste, and ensuring that the target area is fully and evenly covered. In addition, the machine learning algorithm is used to enable the spray robot to adapt to complex and changeable working environments, autonomously decide on the optimal spray strategy, and improve the intelligence level of the robot. The present application utilizes intelligent algorithms such as machine learning to enable the robot to adapt to complex and changeable working environments, autonomously decide on the optimal spray strategy, and improve the intelligence level of the robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following briefly introduces the drawings required for use in the description of the embodiments of the present application. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0049] Figure 1 This is a flow chart of a spray control method of a spray robot proposed in one embodiment of the present application.

[0050] Figure 2 This is a schematic diagram of obtaining the target planning path proposed in an embodiment of the present application.

[0051] Figure 3 This is a structural diagram of a spray control system of a spray robot proposed in one embodiment of the present application.

[0052] Description of reference numerals: 301 - path planning module, 302 - travel control module, 303 - target recognition module, 304 - spray control module. DETAILED DESCRIPTION

[0053] The following will be combined with the drawings in the embodiments of this application to clearly and completely describe the technical solutions in the embodiments of this application. Obviously, the embodiments described are part of the embodiments of this application, not all of them. Based on the embodiments in this application, all other embodiments obtained by ordinary technicians in this field without making creative efforts are within the scope of protection of this application.

[0054] like Figure 1 As shown, the embodiment of the present application provides a spray control method of a spray robot, comprising:

[0055] S101, obtaining a spray area range input by a staff member, performing path planning within the spray area range, and determining a target planning path within the spray area range;

[0056] An electronic map can be provided to the staff, and then the spray area range input by the staff through human-computer interaction can be obtained. The spray area range is a polygonal area (such as a rectangular area or a triangular area) selected by the staff on the electronic map. Then, path planning is performed within the spray area to determine the target planning path.

[0057] The spray robot can be a land-walking robot or a spray drone robot. In the embodiment of the present invention, a spray drone robot is preferably used as the spray robot described in the embodiment.

[0058] S102, controlling the spray robot to move along the target planned path within the spray area, and collecting real-time images on the target planned path in real time during the movement;

[0059] In order to control the spray more precisely, a real-time image of the target planning path can be collected in real time during the movement along the target planning path. The real-time image reflects the target situation on the target planning path and can be analyzed to determine the spray amount.

[0060] S103, using a target recognition model pre-installed on the spray robot to recognize the real-time image on the target planning path, and determining the real-time target on the target planning path;

[0061] The target recognition model can be a model pre-deployed through a deep learning model, through which the real-time target on the target planning path can be accurately identified.

[0062] S104 : Based on the real-time target on the target planning path, control the spray robot to spray during the moving process, thereby completing the spray control of the spray robot.

[0063] This application achieves on-demand spraying, improves the utilization rate of liquid medicine or cleaning agent, reduces waste, and ensures that the target area is fully and evenly covered. Furthermore, the application uses machine learning algorithms to enable the spray robot to adapt to complex and changing working environments, independently decide on the optimal spray strategy, and improve the robot's intelligence level. This application uses machine learning and other intelligent algorithms to enable the robot to adapt to complex and changing working environments, independently decide on the optimal spray strategy, and improve the robot's intelligence level.

[0064] In an embodiment of the present application, the spray control method of the spray robot can be used in scenarios such as pesticide spraying, fertilization, and environmental disinfection. However, it is worth noting that for scenarios such as environmental disinfection that do not require target identification, after determining the target planning path, the robot can directly move along the target planning path and perform disinfection.

[0065] like Figure 2 As shown, the spray area range input by the staff is obtained, and path planning is performed within the spray area range to determine the target planning path within the spray area range, including:

[0066] Obtaining the spray area range input by the staff; wherein the spray area range is the spray area input on the electronic map, and the spray area range is a polygon; considering the actual situation, the spray area range can be limited to a triangle, a quadrilateral, and a pentagon;

[0067] Determine the longest side of the spray area to obtain a target side;

[0068] Rotating the spray area so that the target side is vertical to obtain a rotated spray area;

[0069] Based on the spray robot's scanning width (i.e., the real-world width corresponding to the spray robot's image acquisition width at a fixed flight altitude), the rotated spray area is divided into multiple sub-areas using vertical lines. During the sub-area division process, if the width of the remaining area is smaller than the spray robot's scanning width, the remaining area is directly used as the sub-area.

[0070] It is worth noting that the remaining area may be irregular. In this case, when acquiring the vertical centerline, the vertical centerline can be extended to both ends to ensure that the spray robot can fully capture the image within the remaining area. In addition to the remaining area, if other sub-areas have the same situation, the vertical centerline can also be extended to ensure complete image acquisition.

[0071] The vertical center lines of each sub-area are determined, and the vertical center lines are connected in an S-shape to obtain the target planning path within the spray area.

[0072] In one possible embodiment, controlling the spray robot to move along a target planned path within the spray area and collecting real-time images on the target planned path in real time during the movement includes:

[0073] Determine the two endpoints of the target planning path within the spray area, and determine the endpoint closest to the spray robot to obtain the spray starting point;

[0074] Controlling the spray robot to reach the spray starting point and move along the target planned path within the spray area;

[0075] During the movement of the spray robot, a real-time image on the target planning path is collected by a camera carried by the spray robot.

[0076] In one possible embodiment, the target recognition model pre-installed on the spray robot is set to a YOLO v3 model. It is worth noting that the YOLO v3 model is only a preferred example of the embodiment of the present application, and other target recognition models can also be used.

[0077] In a possible embodiment, the method for loading the target recognition model pre-installed on the spray robot is:

[0078] Initializing model parameters of the target recognition model, encoding the initialized model parameters into vectors to obtain vector codes, and obtaining multiple different vector codes;

[0079] For example, the connection weights between network layers of a target recognition model generally have an upper limit and a lower limit. The connection weights can be initialized between the upper limit and the lower limit, and the initialized connection weights can be encoded into a vector to obtain a vector encoding.

[0080] Obtaining loss function values ​​for all vector encodings, and determining an optimal encoding based on the loss function values;

[0081] The loss function of the YOLO v3 model is a relatively common setting and will not be further described in detail in this embodiment. However, it should be noted that when constructing training samples, images containing the actual target objects should be labeled to ensure that the target recognition model can accurately identify the target. For example, if the crops to be identified are generally pear trees, orange trees, apple trees, etc., then the images in the training samples should contain manual labels for these three types of fruit trees.

[0082] According to the optimal code, adopting an adaptive variable speed search strategy to perform a local search on the vector code, and determining the vector code after the local search;

[0083] Performing a local interpolation search on the vector code after the local search using a multi-code interpolation search strategy to determine the vector code after the local interpolation search;

[0084] Performing a global search on the vector code after the local interpolation search using a global information mutation strategy to determine the vector code after the global search;

[0085] Repeating the adaptive speed change search strategy, the multi-code interpolation search strategy, and the global information mutation strategy until the number of training times reaches a preset maximum number of training times, and re-obtaining the optimal code to obtain the target vector code;

[0086] The model parameters in the target vector encoding are used as the final model parameters of the target recognition model, and the target recognition model is deployed on the spray robot.

[0087] By improving the training algorithm, the embodiment of the present application can enable the target recognition model to identify the target more accurately, thereby achieving precise control of the spray volume.

[0088] Optionally, after deploying the target recognition model on the spray robot, it can be used to identify targets. After each target recognition, the real-time image and the corresponding recognition result can be associated and stored in a database, allowing users to revise the recognition results. With each learning cycle, the self-learning mechanism learns from the data in the database, gradually improving the recognition accuracy of the target recognition model and ensuring accurate target recognition.

[0089] In a possible implementation, performing a local search on the vector code using an adaptive variable speed search strategy according to the optimal code, and determining the vector code after the local search includes:

[0090] Based on the current number of training times, the adaptive speed change factor is obtained as:

[0091]

[0092] in, represents the adaptive speed change factor, represents a natural constant, cos represents the cosine function, represents pi, t represents the current number of training times, T represents the maximum number of training times, Represents the first random number between (0,1);

[0093] According to the adaptive speed change factor and the optimal encoding, the speed change search speed is determined as:

[0094]

[0095] in, Indicates the t During the training k vector encoding, k =1,2,…,NP, NP represents the total number of vector codes, represents the optimal encoding, represents the second random number between (0,1), Indicates variable search speed;

[0096] According to the variable speed search speed and the optimal code, a local search is performed on the vector code to determine the vector code after the local search as follows:

[0097]

[0098] in, Represents the vector encoding after local search .

[0099] An adaptive variable-speed search strategy is used to perform a local search on the vector code, allowing all vector codes to be searched around a location within an optimal region (i.e., the location after perturbing the optimal code). This strategy provides stronger search capabilities in the early stages and more refined searches in the later stages of the algorithm. The search position is continuously shifted, enabling searches in more areas and providing the algorithm with stronger comprehensive search capabilities.

[0100] In a possible implementation, a multi-code interpolation search strategy is used to perform a local interpolation search on the vector code after the local search, and determining the vector code after the local interpolation search includes:

[0101] Obtaining a first neighborhood code and a second neighborhood code of the vector code after the local search;

[0102] For example, a longhorn beetle whisker search algorithm may be used to search the vector code after the local search to determine a first neighborhood code corresponding to the longhorn beetle left whisker position and a second neighborhood code corresponding to the longhorn beetle right whisker position;

[0103] The vector code after the local search is interpolated according to the optimal code, the first neighborhood code, and the second neighborhood code, and the vector code after the local interpolation search is obtained as follows:

[0104]

[0105] in, The optimal coding d dimensional model parameters, d =1,2,…,D, where D represents the total number of model parameters, Indicates the s The first neighborhood code corresponding to the vector code after the local search is d dimensional model parameters, Indicates the s The vector code after the local search corresponds to the second neighborhood code d dimensional model parameters, Indicates the s The vector encoding after the local interpolation search d dimensional model parameters, s =1,2,…,NP, Indicates the s The fitness of the second neighborhood code corresponding to the vector code after the local search, Indicates the s The fitness of the first neighborhood code corresponding to the vector code after the local search, Indicates the fitness corresponding to the optimal encoding, fitness = 1 / (loss function value + 0.001).

[0106] A multi-coding interpolation search strategy is adopted to perform local interpolation search on the vector coding after the local search, so that the algorithm can fuse the optimal position and the position in its neighborhood for fusion search, which is more conducive to searching for a better solution. Moreover, as the algorithm proceeds, the mutation probability gradually decreases, and the search shifts from global optimal to local optimal search, effectively improving the search ability of the algorithm.

[0107] In a possible implementation, a global information mutation strategy is used to perform a global search on the vector code after the local interpolation search, and determining the vector code after the global search includes:

[0108] The number of information integrations obtained is:

[0109]

[0110] in, Indicates the amount of information integration, Indicates rounding up. Represents a random number between (0,1) generated by Levy flight;

[0111] Determining comprehensive integrated information based on the amount of information integration;

[0112]

[0113] in, Indicates the t During the training n The vector encoding after the local interpolation search, n =1,2,…,NP, Represents the randomly selected m The vector encoding after the local interpolation search, represents the third random number between (0,1), Indicates the n Vector encoding after local interpolation search Corresponding comprehensive integrated information;

[0114] According to the comprehensive integration information and the historical optimal value of the vector code after the local interpolation search, a global search is performed on the vector code after the local interpolation search, and the global search code is obtained as follows:

[0115]

[0116] in, Indicates the t During the training n The historical optimal value of the vector encoding after the local interpolation search, that is, in all previous training processes The state where the corresponding loss function value is minimum; Indicates the n A global search code;

[0117] Based on the global search code, a probabilistic inferior solution acceptance method is adopted to obtain the vector code after the global search.

[0118] For example, when the loss function value of the global search code is less than the loss function value of the corresponding vector code after the local interpolation search, the global search code can be used as the vector code after the global search. Otherwise, the simulated annealing algorithm can be used to probabilistically accept the inferior solution of the global search code.

[0119] The global information mutation strategy is used to perform a global search on the vector code after the local interpolation search. This can effectively use the population information to realize the reverse search, so that the vector code is searched in an area far away from the population, thereby effectively improving the global search capability of the algorithm and avoiding the possibility of the algorithm falling into the local optimum. At the same time, the probabilistic inferior solution acceptance method is used for control, which will not affect the convergence of the algorithm and can also improve the global search capability of the algorithm.

[0120] The training algorithm provided in the embodiment of the present application improves the global search capability of the algorithm while ensuring the training speed and training accuracy of the algorithm through the mutual cooperation of several strategies, thereby improving the target recognition accuracy and ultimately improving the spray accuracy.

[0121] Optionally, after each update of the vector code, the vector code may be processed for out-of-bounds to ensure the validity of the model parameters.

[0122] In a possible implementation, based on the real-time target on the target planning path, controlling the spray robot to spray during movement includes:

[0123] Based on the real-time target on the target planning path, query the association relationship table between the target and the spray amount to determine the spray amount corresponding to the real-time target;

[0124] According to the spray volume corresponding to the real-time target, the spray robot is controlled to spray during the movement.

[0125] For example, when identifying real-time targets, the number should also be identified, and the spray volume should be controlled based on the number of real-time targets in the unit area. The target-spray volume association table should store the relationship between the number of targets and the spray volume per unit area. Here, the unit area refers to the area divided by unit area. However, in the actual recognition process, there may be a situation where a real-time target occupies multiple unit areas. In this case, the real-time target will be divided into the unit area with the largest proportion in the area where the predicted box of the real-time target occupies the largest proportion.

[0126] When a spray robot is working, it usually sprays a single target. Therefore, the embodiment of the present application does not need to consider the situation where multiple targets exist within the same unit area. For example, for fruit trees, the fruit tree planting area can be divided into unit areas of unit area while the spray robot is moving, and the number of fruit trees in the unit area can be identified. Then, the association table between target and spray amount is queried to determine the spray amount. The spray robot is then controlled to spray according to the spray amount corresponding to the real-time target during the movement, thereby achieving fine spraying.

[0127] However, it is difficult to determine the unit area more accurately during actual operation. Therefore, during the flight of the spray robot, it flies at a fixed altitude. At this time, the size ratio of the real-time image collected by the spray robot to the size of the real world is fixed. Therefore, the real-time image can be divided into equal parts using a grid. The area of ​​the real world corresponding to each sub-image after equal division is regarded as a unit area. The actual positioning of each unit area can be determined based on the positioning information of the spray robot and the relative relationship between the camera coordinate system and the world coordinate system, thereby realizing regional control of the spray volume and improving the fineness of the spray.

[0128] The present application provides a spray control method for a spray robot, which performs path planning within the spray area, determines the target planning path within the spray area, and controls the spray robot to move along the target planning path within the spray area, and collects real-time images on the target planning path in real time during the movement; then uses the target recognition model pre-installed on the spray robot to recognize the real-time image on the target planning path and determine the real-time target on the target planning path; finally, based on the real-time target on the target planning path, controls the spray robot to spray during the movement, thereby realizing on-demand spraying, improving the utilization rate of liquid medicine or detergent, reducing waste, and ensuring that the target area is fully and evenly covered. The machine learning algorithm is used to enable the spray robot to adapt to complex and changeable working environments, autonomously decide on the optimal spray strategy, and improve the intelligence level of the robot. The present application utilizes intelligent algorithms such as machine learning to enable the robot to adapt to complex and changeable working environments, autonomously decide on the optimal spray strategy, and improve the intelligence level of the robot.

[0129] like Figure 3 As shown, the embodiment of the present application provides a spray control system of a spray robot, including: a path planning module 301, a travel control module 302, a target recognition module 303 and a spray control module 304;

[0130] The path planning module 301 is used to obtain the spray area range input by the staff, perform path planning within the spray area range, and determine the target planning path within the spray area range;

[0131] The travel control module 302 is used to control the spray robot to travel along the target planned path within the spray area, and to collect real-time images on the target planned path in real time during the travel process;

[0132] The target recognition module 303 is used to recognize the real-time image on the target planning path using the target recognition model pre-installed on the spray robot, and determine the real-time target on the target planning path;

[0133] The spray control module 304 is used to control the spray robot to spray during the movement based on the real-time target on the target planning path, thereby completing the spray control of the spray robot.

[0134] The spray control system of a spray robot provided in an embodiment of the present application is capable of executing the above-mentioned method and technical solution. Its principles and beneficial effects are similar and will not be repeated here.

[0135] The various embodiments in this specification are described in a progressive manner, and each embodiment focuses on the differences from other embodiments. The same or similar parts between the various embodiments can be referenced to each other.

[0136] The embodiments of the present application are described with reference to the flowcharts and / or block diagrams of the methods, apparatuses, electronic devices, and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing terminal device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing terminal device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 a process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0137] These computer program instructions may also be stored in a computer readable memory that can direct a computer or other programmable data processing terminal device to operate in a specific manner, so that the instructions stored in the computer readable memory produce a manufactured product including an instruction device, which implements the process Figure 1 a process or multiple processes and / or boxes Figure 1 The function specified in one or more boxes.

[0138] These computer program instructions can also be loaded onto a computer or other programmable data processing terminal device so that a series of operating steps are executed on the computer or other programmable terminal device to produce a computer-implemented process, thereby providing instructions for executing on the computer or other programmable terminal device to implement the process. Figure 1 a process or multiple processes and / or boxes Figure 1 A step that specifies a function in one or more boxes.

[0139] Although preferred embodiments of the present invention have been described, those skilled in the art may make additional changes and modifications to these embodiments once they become aware of the basic inventive concepts. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the embodiments of the present invention.

[0140] Finally, it should be noted that, in this document, relational terms such as first and second, etc., are used only to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the terms "comprises," "includes," or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article, or terminal device that includes a series of elements includes not only those elements, but also other elements not explicitly listed, or elements that are inherent to such process, method, article, or terminal device. In the absence of further restrictions, an element defined by the phrase "comprises a ..." does not exclude the presence of additional identical elements in the process, method, article, or terminal device that includes the element.

[0141] The above is a detailed introduction to the hierarchical relationship analysis method for influencing factors provided by this application. Specific examples are used in this article to illustrate the principles and implementation methods of this application. The description of the above embodiments is only used to help understand the method of this application and its core idea; at the same time, for general technical personnel in this field, based on the ideas of this application, there will be changes in the specific implementation methods and application scope. In summary, the content of this specification should not be understood as a limitation on this application.

Claims

1. A spray control method for a spray robot, characterized in that: include: Obtaining the spray area range input by the staff, and performing path planning within the spray area range to determine the target planning path within the spray area range; Controlling the spray robot to move along the target planned path within the spray area, and collecting real-time images on the target planned path in real time during the movement; Using a target recognition model pre-installed on the spray robot, the real-time image on the target planning path is recognized to determine the real-time target on the target planning path; Based on the real-time target on the target planning path, the spray robot is controlled to spray during the movement, thereby completing the spray control of the spray robot; The method for loading the target recognition model pre-installed on the spray robot is as follows: Initializing model parameters of the target recognition model, encoding the initialized model parameters into vectors to obtain vector codes, and obtaining multiple different vector codes; Obtaining loss function values ​​for all vector encodings, and determining an optimal encoding based on the loss function values; According to the optimal code, adopting an adaptive variable speed search strategy to perform a local search on the vector code, and determining the vector code after the local search; Performing a local interpolation search on the vector code after the local search using a multi-code interpolation search strategy to determine the vector code after the local interpolation search; Performing a global search on the vector code after the local interpolation search using a global information mutation strategy to determine the vector code after the global search; Repeating the adaptive speed change search strategy, the multi-code interpolation search strategy, and the global information mutation strategy until the number of training times reaches a preset maximum number of training times, and re-obtaining the optimal code to obtain the target vector code; Using the model parameters in the target vector encoding as the final model parameters of the target recognition model, and deploying the target recognition model on the spray robot; According to the optimal code, a local search is performed on the vector code using an adaptive variable speed search strategy to determine the vector code after the local search, including: Based on the current number of training times, the adaptive speed change factor is obtained as: in, represents the adaptive speed change factor, represents a natural constant, cos represents the cosine function, represents pi, t represents the current number of training times, T represents the maximum number of training times, Represents the first random number between (0,1); According to the adaptive speed change factor and the optimal encoding, the speed change search speed is determined as: in, Indicates the t During the training k vector encoding, k =1,2,…,NP, NP represents the total number of vector codes, represents the optimal encoding, represents the second random number between (0,1), Indicates variable search speed; According to the variable speed search speed and the optimal code, a local search is performed on the vector code to determine the vector code after the local search as follows: in, Represents the vector encoding after local search .

2. The spray control method of the spray robot according to claim 1, characterized in that: Obtain the spray area range input by the staff, perform path planning within the spray area range, and determine the target planning path within the spray area range, including: Obtaining the spray area range input by the staff; wherein the spray area range is the spray area input on the electronic map, and the spray area range is a polygon; Determine the longest side of the spray area to obtain a target side; Rotating the spray area so that the target side is vertical to obtain a rotated spray area; Based on the scanning width of the spray robot, the spray area after rotation is divided into multiple sub-areas using vertical lines; wherein, during the process of dividing the sub-areas, if the width of the remaining area is smaller than the scanning width of the spray robot, the remaining area is directly used as the sub-area; The vertical center lines of each sub-area are determined, and the vertical center lines are connected in an S-shape to obtain the target planning path within the spray area.

3. The spray control method of the spray robot according to claim 1, characterized in that: Controlling the spray robot to move along the target planning path within the spray area, and collecting real-time images on the target planning path in real time during the movement, including: Determine the two endpoints of the target planning path within the spray area, and determine the endpoint closest to the spray robot to obtain the spray starting point; Controlling the spray robot to reach the spray starting point and move along the target planned path within the spray area; During the movement of the spray robot, a real-time image on the target planning path is collected by a camera carried by the spray robot.

4. The spray control method of the spray robot according to claim 1, characterized in that: The target recognition model pre-installed on the spray robot is set to the YOLO v3 model.

5. The spray control method of the spray robot according to claim 1, characterized in that: Performing a local interpolation search on the vector code after the local search using a multi-code interpolation search strategy to determine the vector code after the local interpolation search includes: Obtaining a first neighborhood code and a second neighborhood code of the vector code after the local search; The vector code after the local search is interpolated according to the optimal code, the first neighborhood code, and the second neighborhood code to obtain the vector code after the local interpolation search.

6. The spray control method of the spray robot according to claim 5, characterized in that: Performing a global search on the vector code after the local interpolation search using a global information mutation strategy to determine the vector code after the global search includes: Obtaining the amount of information integration, and determining comprehensive integrated information based on the amount of information integration; Performing a global search on the vector code after the local interpolation search based on the comprehensive integration information and a historical optimal value of the vector code after the local interpolation search to obtain a global search code; Based on the global search code, a probabilistic inferior solution acceptance method is adopted to obtain the vector code after the global search.

7. The spray control method of the spray robot according to claim 1, characterized in that: Based on the real-time target on the target planning path, controlling the spray robot to spray during the movement includes: Based on the real-time target on the target planning path, query the association relationship table between the target and the spray amount to determine the spray amount corresponding to the real-time target; According to the spray volume corresponding to the real-time target, the spray robot is controlled to spray during the movement.

8. A spray control system for a spray robot, capable of executing the spray control method for a spray robot according to any one of claims 1 to 7, characterized in that: include: Path planning module, travel control module, target recognition module and spray control module; The path planning module is used to obtain the spray area range input by the staff, perform path planning within the spray area range, and determine the target planning path within the spray area range; The travel control module is used to control the spray robot to travel along the target planning path within the spray area, and to collect real-time images on the target planning path in real time during the travel process; The target recognition module is used to recognize the real-time image on the target planning path using the target recognition model pre-installed on the spray robot, and determine the real-time target on the target planning path; The spray control module is used to control the spray robot to spray during the movement based on the real-time target on the target planning path, thereby completing the spray control of the spray robot.

Images