Pose recognition method and device of photovoltaic module, and cleaning equipment

Abstract

The invention discloses a pose recognition method and device of a photovoltaic module and cleaning equipment. The method is used for a laser radar and a camera to carry out pose recognition on a photovoltaic module, and comprises the steps: obtaining a first image; obtaining first point cloud data; determining a first edge parameter of the photovoltaic module based on the first image; determining a plane point cloud of the photovoltaic module based on the first point cloud data; second edge parameters are determined based on the first pose relation, the plane point cloud and the first edge parameters, and the first pose relation is determined based on the installation relation of the laser radar and the camera; determining a first angular point parameter of the plane point cloud based on the second edge parameter and a preset size parameter of the photovoltaic module; determining a second angular point parameter based on a preset size parameter; and determining a second pose relationship based on the first corner parameter and the second corner parameter. According to the invention, the laser radar and the camera in a small view field range can be utilized, so that the accuracy of short-distance pose recognition in the photovoltaic cleaning process can be improved, and the production cost is saved.

Description

Technical Field

[0001] The present application relates to the technical field of photovoltaic cleaning, and particularly relates to a method and device for identifying the pose of a photovoltaic module, and a cleaning device. Background Art

[0002] With the continuous development of new energy technologies, the newly installed capacity of photovoltaic power generation has been continuously expanding, and photovoltaic energy has become an important part of the new energy structure. Since photovoltaic devices are long-term exposed to the outdoor environment, especially photovoltaic modules are prone to dust accumulation, resulting in a decrease in the photoelectric conversion efficiency, it is necessary to clean the photovoltaic modules in a timely manner to ensure the stable operation of the devices. Therefore, the pose identification of photovoltaic modules deserves attention. Summary of the Invention

[0003] In view of this, embodiments of the present application provide a method and device for identifying the pose of a photovoltaic module, and a cleaning device, in order to improve the pose identification range of the photovoltaic module. In a first aspect, a method for identifying the pose of a photovoltaic module is provided, which is used for identifying the pose of the photovoltaic module by combining a lidar and a camera, and includes: obtaining a first image, the first image being determined based on the camera; obtaining first point cloud data, the first point cloud data being determined based on the lidar; determining first edge parameters of the photovoltaic module based on the first image; determining a planar point cloud of the photovoltaic module based on the first point cloud data; determining second edge parameters based on a first pose relationship, the planar point cloud, and the first edge parameters, wherein the first pose relationship is determined based on the installation relationship between the lidar and the camera; determining first corner point parameters of the planar point cloud based on the second edge parameters and preset size parameters of the photovoltaic module; determining second corner point parameters of the photovoltaic module based on the preset size parameters; and determining a second pose relationship based on the first corner point parameters and the second corner point parameters.

[0004] In the above method for identifying the pose of a photovoltaic module, the heterogeneous sensors combined by the camera and the lidar can simultaneously meet the different requirements of the two sensors in terms of the focus of the field of view. Especially when the photovoltaic cleaning robot is close to the photovoltaic module and cannot obtain a complete image of the photovoltaic module, it can accurately identify the pose of the photovoltaic module by only obtaining partial structures of the photovoltaic module (such as characteristic structures such as the lower edge and the upper edge), and assist the photovoltaic cleaning robot to complete the cleaning work. By using a lidar and a camera with a relatively small field of view, the accuracy of pose identification at a relatively short distance during the photovoltaic cleaning process can be improved, and the production cost can be saved.

[0005] In some embodiments, the first edge parameters include: a first coordinate set of one or more feature points of the first edge in the camera coordinate system; the second edge parameters include: a second coordinate set of one or more feature points of the first edge in the lidar coordinate system.

[0006] In some embodiments, determining the first edge parameter of a photovoltaic module based on a first image includes: determining the structure of the photovoltaic module in the first image based on the first image and an image segmentation algorithm; extracting and fitting the linear pixels of the first edge based on the structure of the photovoltaic module; and determining the first edge parameter based on the linear pixels.

[0007] In some embodiments, the preset size parameters include: the length size and the width size of the photovoltaic module; the second coordinate set includes: the first corner point coordinate and the second corner point coordinate; determining the first corner point parameter of the planar point cloud based on the second edge parameter and the preset size parameters of the photovoltaic module includes: determining the third corner point coordinate based on the first corner point coordinate and the width size; determining the fourth corner point coordinate based on the second corner point coordinate and the width size; and determining the first corner point parameter based on the first corner point coordinate, the second corner point coordinate, the third corner point coordinate, and the fourth corner point coordinate.

[0008] In some embodiments, determining the second corner point parameter based on the preset size parameters includes: setting the first preset point of the photovoltaic module as the origin; and determining the second corner point parameter based on the origin, the length size, and the width size.

[0009] In some embodiments, determining the planar point cloud of the photovoltaic module based on the first point cloud data includes: determining the planar point cloud based on the first point cloud data and a point cloud depth threshold.

[0010] In a second aspect, there is provided a pose recognition device for a photovoltaic module, including: an acquisition unit configured to acquire a first image, the first image being determined based on a camera, and acquire first point cloud data, the first point cloud data being determined based on a lidar; an edge determination unit configured to determine a first edge parameter of the photovoltaic module based on the first image, determine the planar point cloud of the photovoltaic module based on the first point cloud data, and determine a second edge parameter based on a first pose relationship, the first point cloud data, and the first edge parameter, wherein the first pose relationship is determined based on the installation relationship between the lidar and the camera; a corner determination unit configured to determine a first corner point parameter of the planar point cloud based on the second edge parameter and the preset size parameters of the photovoltaic module, and determine a second corner point parameter based on the preset size parameters; and a pose determination unit configured to determine a second pose relationship based on the first corner point parameter and the second corner point parameter.

[0011] The preset size parameters include: the length size and the width size of the photovoltaic module; the second coordinate set includes: the first corner point coordinate and the second corner point coordinate;

[0012] In some embodiments, the corner determination unit is further configured to determine a third corner point coordinate based on the first corner point coordinate and the width size; determine a fourth corner point coordinate based on the second corner point coordinate and the width size; and determine the first corner point parameter based on the first corner point coordinate, the second corner point coordinate, the third corner point coordinate, and the fourth corner point coordinate.

[0013] In a third aspect, a pose recognition device for a photovoltaic module is provided, including: a camera for acquiring a first image of the photovoltaic module; a lidar for acquiring first point cloud data of the photovoltaic module; and a processor configured to be coupled to the camera and the lidar and configured to perform pose recognition on the photovoltaic module according to the pose recognition method provided in the first aspect.

[0014] In a fourth aspect, a photovoltaic cleaning device is provided, including the pose recognition device provided in the third aspect. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] The following provides a brief introduction to the drawings used in the description of the embodiments of the present application:

[0016] Figure 1 FIG. shows a schematic structural diagram of a photovoltaic cleaning device provided in some embodiments of the present application;

[0017] Figure 2 FIG. shows a schematic flowchart of a method for recognizing the pose of a photovoltaic module provided in some embodiments of the present application;

[0018] Figure 3 FIG. shows a schematic flowchart of a method for determining first edge parameters provided in some embodiments of the present application;

[0019] Figure 4 FIG. shows a schematic structural diagram of a pose recognition device for a photovoltaic module provided in some embodiments of the present application;

[0020] Figure 5 FIG. shows a schematic structural diagram of another pose recognition device for a photovoltaic module provided in some embodiments of the present application. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0021] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will describe the specific embodiments of the present application with reference to the accompanying drawings. The drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings or embodiments can be obtained based on these drawings or embodiments. Adjustments and improvements made without departing from the concept of the present application all fall within the protection scope of the present application.

[0022] To make the drawings concise, each drawing only schematically shows the parts related to the embodiments, and they do not represent the actual structure of the product. Additionally, to make the drawings concise and easy to understand, in some drawings, parts with the same structure or function are only schematically shown partially, and there may actually be more or fewer parts with the same structure or function.

[0023] Photovoltaic power stations are often located in large photovoltaic sites with sufficient sunlight and open environments. Photovoltaic modules are usually installed at a relatively high position and are arranged in a scattered manner. Due to being exposed to the outdoor environment for a long time, a large amount of dust is likely to accumulate on the photovoltaic modules, and it is difficult to support the huge workload and cleaning frequency by manual cleaning. Therefore, the daily cleaning of photovoltaic equipment can be achieved through photovoltaic module cleaning equipment, such as using a photovoltaic cleaning robot to complete the lifting and placing actions of the cleaning machine through the robotic arm on it. Among them, the placing action is to place the cleaning machine on the photovoltaic module, and the pose of the module needs to be accurately identified. However, in order to maximize the utilization of sunlight and improve the photoelectric conversion efficiency, the size of photovoltaic modules is often designed to be relatively large. When performing the action of placing the cleaning machine in a short-distance space, the field of view of the sensor on the robotic arm is difficult to cover the entire picture of a photovoltaic module with a length of more than 2 meters and a width of more than 1 meter, and usually can only cover a part of it. A photovoltaic module can be composed of a metal outer frame and a combined part of photovoltaic cells connected in the middle. When identifying the pose of a photovoltaic module, the entire or most of the module can be irradiated by a 3D camera sensor to identify 3 - 4 edges or 4 corners of the photovoltaic module, and then the pose of the module can be calculated. This method is applicable to a robotic arm with a large range of motion space. For a robotic arm with a small range of motion space for close-range operations, it often cannot meet the requirement that the 3D camera sensor irradiates most of the module. Due to the limited field of view at close range, the field of view range of the 3D sensor on the robotic arm can only cover a part of the module. Therefore, this application provides a method, device, and cleaning equipment for identifying the pose of a photovoltaic module, which uses a lidar and a camera with non-parallel optical axes to form a larger field of view within a short distance range, identify the pose of the photovoltaic module, and determine the pose of the photovoltaic module in the camera coordinate system by identifying the image obtained by the camera and the point cloud data obtained by the lidar.

[0024] The following is a description with reference to the accompanying drawings:

[0025] Please refer to Figure 1, which shows a schematic structural diagram of a photovoltaic cleaning device provided in some embodiments of the present application. The photovoltaic cleaning robot cleans the photovoltaic module 20 through the robotic arm 11, and a camera 12 and a lidar 13 are arranged on the robotic arm 11. The camera 12 can cover the first field of view F1, and the lidar 13 can cover the second field of view F2. Among them, the optical axes of the camera 12 and the lidar 13 are not parallel. Compared with the setting where the optical axes are parallel, the combination of the first field of view F1 and the second field of view F2 can jointly cover a larger field of view range, and it is easier to identify the photovoltaic module 20 when the photovoltaic cleaning robot operates at a short distance. For example, the first field of view F1 of the camera 12 covers the lower edge of the photovoltaic module 20, and the second field of view F2 of the lidar 13 covers most of the plane of the photovoltaic module. By using the method for identifying the pose of the photovoltaic module of the present application, the characteristic structure of part of the photovoltaic module 20 can be covered by the camera 12 without fully covering the photovoltaic module 20. By superimposing the other characteristic structures of the photovoltaic module covered by the lidar 13, and combining the two, all the plane characteristic structures of the photovoltaic module are covered, and the pose identification can be achieved. The present application can use a lidar and a camera with a small field of view range to take into account the identification at different distances, especially improving the accuracy of pose identification at a relatively short distance during the photovoltaic cleaning process and saving production costs.

[0026] Please refer to Figure 2 , which shows a schematic flow diagram of a method for identifying the pose of a photovoltaic module provided in some embodiments of the present application. The method for identifying the pose of the photovoltaic module is used for the combination of a lidar and a camera to identify the pose of the photovoltaic module, and at least includes the following steps:

[0027] S210: Obtain a first image, where the first image is determined based on the camera;

[0028] S220: Obtain first point cloud data, where the first point cloud data is determined based on the lidar;

[0029] S230: Determine the first edge parameter of the photovoltaic module based on the first image;

[0030] S240: Determine the planar point cloud of the photovoltaic module based on the first point cloud data;

[0031] S250: Determine the second edge parameter based on the first pose relationship, the planar point cloud, and the first edge parameter, where the first pose relationship is determined based on the installation relationship between the lidar and the camera;

[0032] S260: Determine the first corner point parameter of the planar point cloud based on the second edge parameter and the preset size parameter of the photovoltaic module;

[0033] S270: Determine the second corner point parameter of the photovoltaic module based on the preset size parameter;

[0034] S280: Determine the second pose relationship based on the first corner point parameter and the second corner point parameter.

[0035] In the above pose recognition method, when the photovoltaic cleaning robot is close to the photovoltaic module for operation, it is difficult to capture the entire structure of the photovoltaic module. Therefore, the first image determined by the camera can be a partial image of the photovoltaic module. For example, only the lower edge or the upper edge of the photovoltaic module is presented in the figure. The presentation of this first image does not affect the final pose recognition result. The point cloud image can be obtained through the scanning of the lidar to form the first point cloud data. Based on the first image, the image segmentation algorithm in deep learning can be used to segment and recognize the part of the photovoltaic module in the first image to determine the first edge parameter of the photovoltaic module. For example, the first edge parameter can be the center point and the two side points in the recognized lower edge of the photovoltaic module. When processing the first point cloud data, the point cloud representing the plane of the photovoltaic module can be extracted through the plane extraction method to determine the plane point cloud. Since the first pose relationship can be determined by debugging and other means when the lidar and the camera are installed, based on this first pose relationship, the points corresponding to the first edge parameter in the plane point cloud can be determined to form the second edge parameter. The length and width dimensions of the photovoltaic module are known parameters, and the photovoltaic cleaning robot is allowed to translate in the Y-axis direction when placing the cleaning machine on the photovoltaic module. Therefore, the first corner point parameter can be determined based on the preset dimension parameter and the second edge parameter. For example, the center point of the lower edge of the photovoltaic module represented by the second edge parameter obtained in the above steps can be considered as the center point at the bottom of the photovoltaic module. Starting from this center point, extending to the left and right ends, the straight line of the lower edge of the photovoltaic module is determined according to the preset dimension parameter, and then the two ends of the upper edge straight line are determined by continuing to extend upward from the two ends of this lower edge straight line. So far, the 4 corner points of the photovoltaic module in the laser point cloud plane can be obtained, that is, the spatial coordinates of the 4 corner points of the photovoltaic module in the camera coordinate system, forming the first corner point parameter. On the photovoltaic module body, since the size of the photovoltaic module can be known in advance, a feature point can be anchored. For example, the center point of the photovoltaic module is used as the origin, and the length and width dimensions of the photovoltaic module are used as coordinates to determine the second corner point parameter corresponding to the first corner point parameter. For example, the first corner point parameter involves the upper left, upper right, lower left, and lower right four corner points, and the second corner point parameter can also determine the above four points. After the corresponding points in different coordinate systems can be determined, the second pose relationship can be determined through the coordinate point conversion relationship. In the above determination process, the heterogeneous sensors combined with the camera and the lidar can simultaneously meet the different requirements of the visual field focus of the two sensors. Especially when the photovoltaic cleaning robot is close to the photovoltaic module and cannot obtain a complete image of the photovoltaic module, only by obtaining partial structures of the photovoltaic module (such as the lower edge, upper edge and other characteristic structures) can the pose of the photovoltaic module be accurately recognized to assist the photovoltaic cleaning robot to complete the cleaning work.

[0036] In one embodiment, the first edge parameter includes: a first coordinate set of one or more feature points of the first edge in the camera coordinate system; the second edge parameter includes: a second coordinate set of one or more feature points of the first edge in the lidar coordinate system.

[0037] The coordinate points in the first coordinate set and the second coordinate set have a corresponding relationship, and this corresponding relationship can be determined through the first pose relationship. When determining the first pose relationship, first, the camera internal parameter matrix can be obtained from the factory settings of the camera or camera calibration, as shown in Formula 1:

[0038]

[0039] The installation bracket structure size and calibration can obtain the installation pose relationship from the lidar to the camera, as shown in Formula 2:

[0040] Shown in the formula.

[0041]

[0042] Assume that the point cloud coordinates in a certain lidar coordinate system are (x Li , y Li , z Li ). Then, when converted to the camera coordinate system, the coordinates of the point cloud are represented by (x Ci , y Ci , z Ci ), and there is Formula 3:

[0043]

[0044] Thereby, the second edge parameter is determined.

[0045] Please refer to Figure 3 , which shows a schematic flowchart of a method for determining the first edge parameter provided in some embodiments of the present application. Step S230 determines the first edge parameter of the photovoltaic module based on the first image, including:

[0046] S310: Based on the first image and the image segmentation algorithm, determine the structure of the photovoltaic module in the first image;

[0047] S320: Based on the structure of the photovoltaic module, extract and fit the straight-line pixels of the first edge;

[0048] S330: Based on the straight-line pixels, determine the first edge parameter.

[0049] To determine the structure of the photovoltaic module in the first image, a deep learning image segmentation method can be used to segment the photovoltaic panel of the photovoltaic module in the camera's field of view. For example, a deep learning method such as the YOLO series can be used. Take the largest piece in the first image, and this part can be considered as the photovoltaic panel. In the segmented image, detect the part with characteristic structure in the segmented image. For example, the longest edge line at the bottom. Since most of the camera's field of view covers the lower part of the photovoltaic panel, the longest edge line at the bottom of the segmented image can be considered as the lower edge of the photovoltaic module. Based on this lower edge, the straight-line pixels of the first edge can be extracted and fitted, and the coordinates of the feature points can be determined according to the position of the pixels. For example, determine the center point of the first edge, the endpoint coordinates on the left and right sides of the first edge, and determine the first edge parameters.

[0050] In one embodiment, the preset size parameters include: the length size and width size of the photovoltaic module; the second coordinate set includes: the first corner point coordinates and the second corner point coordinates; based on the second edge parameters and the preset size parameters of the photovoltaic module, determining the first corner point parameters of the planar point cloud includes:

[0051] Determining the third corner point coordinates based on the first corner point coordinates and the width size;

[0052] Determining the fourth corner point coordinates based on the second corner point coordinates and the width size;

[0053] Determining the first corner point parameters based on the first corner point coordinates, the second corner point coordinates, the third corner point coordinates, and the fourth corner point coordinates.

[0054] The first corner point coordinates can be the left endpoint of the lower edge of the photovoltaic module. Based on the known width size of the photovoltaic module, extending upward from this first corner point coordinates by the width size, the third corner point coordinates can be determined. The second corner point coordinates can be the right endpoint of the lower edge of the photovoltaic module. Based on the known width size of the photovoltaic module, extending upward from this second corner point coordinates, the fourth corner point coordinates can be determined. The first corner point parameters can include the set of the above four corner point coordinates. After determining the first corner point parameters, the second pose relationship can be jointly determined through the coordinates of the feature points corresponding to the first corner point parameters in the actual photovoltaic module.

[0055] In one embodiment, determining the second corner point parameters based on the preset size parameters includes: setting the first preset point of the photovoltaic module as the origin; determining the second corner point parameters based on the origin, the length size, and the width size.

[0056] For example, assuming that the coordinates of the geometric center point of the photovoltaic module are (0.0,0.0,0.0), the length of the photovoltaic module is a, and the width is b, then the coordinates of the upper left, upper right, lower left, and lower right corner points of the module are (0.0,0.0,0.0) respectively. Combined with the actual size of the four corner points of the component (-0.5*b,0.5*a,0.0), (0.5*b,0.5*a,0.0), (-0.5*b,-0.5*a,0.0), (0.5*b,-0.5*a,0.0) and the above second corner point parameters, combined with the first corner point parameters, you can call the function in the pcl library to perform rigid body pose transformation, such as calling pcl::registration::TransformationEstimationSVD<PointType,PointType> The estimateRigidTransformation function can obtain the component's pose transformation matrix to determine the pose of the PV component in the camera coordinate system.

[0057] In one embodiment, determining a plane point cloud of the photovoltaic assembly based on the first point cloud data includes: determining the plane point cloud based on the first point cloud data and a point cloud depth threshold.

[0058] When the laser radar is scanning, the point clouds of different depths in the point cloud can reflect the distance of the object represented by the point from the laser radar. Since the photovoltaic panel of a photovoltaic module is often a complete plane, the depth of the point determined by the echo reflected from the plane is often similar. By presetting the point cloud depth threshold, the point cloud that does not belong to the photovoltaic module can be screened out and removed, leaving the point cloud that represents the photovoltaic panel of the photovoltaic module, thereby improving the accuracy of photovoltaic panel identification.

[0059] Please refer to Figure 4 , which shows a schematic structural diagram of a posture recognition device for a photovoltaic component provided in some embodiments of the present application. The posture recognition device 400 for a photovoltaic component includes: an acquisition unit 410, which is used to acquire a first image, the first image is determined based on a camera, and acquire first point cloud data, the first point cloud data is determined based on a laser radar; an edge determination unit 420, which is used to determine a first edge parameter of the photovoltaic component based on the first image, determine a plane point cloud of the photovoltaic component based on the first point cloud data, and determine a second edge parameter based on a first posture relationship, the first point cloud data and the first edge parameter, wherein the first posture relationship is determined based on the installation relationship between the laser radar and the camera; a corner point determination unit 430, which is used to determine a first corner point parameter of the plane point cloud based on the second edge parameter and a preset size parameter of the photovoltaic component, and determine a second corner point parameter based on the preset size parameter; a posture determination unit 440, which is used to determine a second posture relationship based on the first corner point parameter and the second corner point parameter.

[0060] In some embodiments, the corner determination unit 430 is further configured to determine the coordinates of a third corner based on the coordinates of the first corner and the width dimension; determine the coordinates of a fourth corner based on the coordinates of the second corner and the width dimension; and determine a first corner parameter based on the coordinates of the first corner, the coordinates of the second corner, the coordinates of the third corner, and the coordinates of the fourth corner.

[0061] For the specific implementation manners and beneficial effects of the above pose recognition device of the photovoltaic module, reference may be made to the specific descriptions of the embodiments of the above pose recognition method, which will not be elaborated herein. The above division of each unit is only a division of logical functions. In actual implementation, all or part of them may be integrated into a physical entity, or physically separated. In addition, the above units may be implemented in the form of a processor calling software. Alternatively, the above units may be implemented in the form of a hardware circuit. The functions of some or all of the units may be implemented by designing the hardware circuit, and the hardware circuit may be understood as one or more processors. For example, in some embodiments, the hardware circuit is an application specific integrated circuit (ASIC), and the functions of some or all of the above units are implemented by designing the logical relationship between the components in the circuit. Again, in another implementation, the hardware circuit may be implemented by a programmable logic device (PLD), which may include a large number of logic gate circuits, and the logical relationship between the logic gate circuits is configured through a configuration file to implement the functions of some or all of the above units. The units of the above device may all be implemented in the form of a processor calling a program, or all be implemented in the form of a hardware circuit, or part be implemented in the form of a processor calling a program, and the remaining part be implemented in the form of a hardware circuit.

[0062] Please refer to Figure 5 , which shows a schematic structural diagram of another pose recognition device of a photovoltaic module provided in some embodiments of the present application. The pose recognition device 500 of the photovoltaic module includes: a camera 510, configured to acquire a first image of the photovoltaic module 20; a lidar 520, configured to acquire first point cloud data of the photovoltaic module 20; and a processor 530, configured to be coupled to the camera 510 and the lidar 520, and configured to perform pose recognition on the photovoltaic module 20 according to the pose recognition method provided in any one of the above embodiments.

[0063] In this application, unless otherwise clearly specified and defined, ordinal numbers such as "first", "second", etc. are only used to distinguish and describe related objects, and cannot be understood as indicating or implying the relative importance or order between related objects; in addition, they do not represent the quantity of related objects. "A plurality of" includes two or more, and other quantifiers are similar. " / " is used to describe the relationship between related objects, which represents the "or" relationship between related objects. "And / or" is used to describe the relationship between related objects, which includes any combination relationship between related objects. For example, "a and / or b" includes: "a alone", "b alone", or "a and b". "One or more" or "at least one" among a plurality of objects refers to any object or any combination of a plurality of objects. For example, "one or more of a1, a2, a3" or "at least one of a1, a2, a3" includes: "a1 alone", "a2 alone", "a3 alone", "a1 and a2", "a1 and a3", "a2 and a3", or "a1, a2 and a3".

Claims

1. A method for identifying the pose of a photovoltaic module, characterized in that Pose recognition of a photovoltaic module using a lidar and a camera with non-parallel optical axes, comprising: Obtaining a first image, which is determined based on the camera; Obtaining first point cloud data, which is determined based on the lidar; Determining first edge parameters of the photovoltaic module based on the first image; Determining a planar point cloud of the photovoltaic module based on the first point cloud data; Determining second edge parameters based on a first pose relationship, the planar point cloud, and the first edge parameters, wherein the first pose relationship is determined based on the installation relationship between the lidar and the camera; Determining first corner point parameters of the planar point cloud based on the second edge parameters and preset dimension parameters of the photovoltaic module; Determining second corner point parameters of the photovoltaic module based on the preset dimension parameters; Determining a second pose relationship based on the first corner point parameters and the second corner point parameters.

2. The pose recognition method according to claim 1, wherein The first edge parameters include: a first coordinate set of one or more feature points of a first edge in a camera coordinate system; The second edge parameters include: a second coordinate set of one or more of the feature points of the first edge in a lidar coordinate system.

3. The pose recognition method according to claim 2, characterized in that, The determining the first edge parameters of the photovoltaic module based on the first image includes: Determining the structure of the photovoltaic module in the first image based on the first image and an image segmentation algorithm; Extracting and fitting the straight line pixels of the first edge based on the structure of the photovoltaic module; Determining the first edge parameters based on the straight line pixels.

4. The pose recognition method according to claim 3, wherein The preset dimension parameters include: the length dimension and the width dimension of the photovoltaic module; the second coordinate set includes: a first corner point coordinate and a second corner point coordinate; The determining the first corner point parameters of the planar point cloud based on the second edge parameters and the preset dimension parameters of the photovoltaic module includes: Determining a third corner point coordinate based on the first corner point coordinate and the width dimension; Determining a fourth corner point coordinate based on the second corner point coordinate and the width dimension; Determining the first corner point parameters based on the first corner point coordinate, the second corner point coordinate, the third corner point coordinate, and the fourth corner point coordinate.

5. The pose recognition method according to claim 4, characterized in that, The determining the second corner point parameters based on the preset dimension parameters includes: Setting a first preset point of the photovoltaic module as the origin; Determining the second corner point parameters based on the origin, the length dimension, and the width dimension.

6. The pose recognition method according to any one of claims 1-5, characterized in that, The determining the planar point cloud of the photovoltaic module based on the first point cloud data includes: Determining the planar point cloud based on the first point cloud data and a point cloud depth threshold.

7. A pose recognition device for a photovoltaic module, characterized in that, Comprising: An acquisition unit for acquiring a first image, which is determined based on a camera, and acquiring first point cloud data, which is determined based on a lidar; An edge determination unit, configured to determine first edge parameters of the photovoltaic module based on the first image, determine a planar point cloud of the photovoltaic module based on the first point cloud data, and determine second edge parameters based on a first pose relationship, the first point cloud data, and the first edge parameters, where the first pose relationship is determined based on the installation relationship between the lidar and the camera; A corner point determination unit, configured to determine first corner point parameters of the planar point cloud based on the second edge parameters and preset dimension parameters of the photovoltaic module, and determine second corner point parameters of the photovoltaic module based on the preset dimension parameters; A pose determination unit, configured to determine a second pose relationship based on the first corner point parameters and the second corner point parameters.

8. The pose recognition device according to claim 7, characterized in that The preset dimension parameters include: the length dimension and the width dimension of the photovoltaic module; the second edge parameters include: a second coordinate set of one or more feature points of the first edge in the lidar coordinate system; the second coordinate set includes: a first corner point coordinate and a second corner point coordinate; The corner point determination unit is further configured to determine a third corner point coordinate based on the first corner point coordinate and the width dimension; determine a fourth corner point coordinate based on the second corner point coordinate and the width dimension; and determine the first corner point parameters based on the first corner point coordinate, the second corner point coordinate, the third corner point coordinate, and the fourth corner point coordinate.

9. A pose recognition device for a photovoltaic module, characterized in that, Comprising: A camera, configured to acquire a first image of the photovoltaic module; A lidar, configured to acquire first point cloud data of the photovoltaic module, where the optical axes of the camera and the lidar are not parallel; A processor, configured to be coupled to the camera and the lidar, and configured to perform pose recognition on the photovoltaic module according to the pose recognition method according to any one of claims 1-6.

10. A photovoltaic cleaning device, characterized in that, Comprising the pose recognition device according to claim 9.

Images