Photovoltaic cleaning robot based on visual SLAM

Abstract

The utility model relates to a photovoltaic cleaning robot based on visual SLAM (Simultaneous Localization and Mapping), which comprises a robot body, cleaning components are respectively arranged at the front end and the rear end of the robot body, track driving components are respectively arranged on two sides of the robot body, and the photovoltaic cleaning robot walks under the driving of the track driving components. Photovoltaic modules in a photovoltaic array in the photovoltaic power station are cleaned through the cleaning module; the track driving assembly is provided with a wheel type encoder, the wheel type encoder is used for obtaining the rotating speed of the track driving assembly and obtaining a wheel type odometer, the track driving assembly is further provided with an inertia measurement unit, and the inertia measurement unit is used for measuring attitude information of the fuselage. And then the wheel type odometer and the attitude information are coupled to serve as basic positioning of the photovoltaic cleaning robot. The automation degree of the photovoltaic cleaning robot can be improved, and therefore the purposes of saving manpower and reducing the maintenance cost of a power station are achieved.

Description

Technical Field

[0001] The utility model relates to a photovoltaic cleaning robot based on visual SLAM. Background Art

[0002] As a renewable and clean energy, solar photovoltaic has become an important force in today's global energy transformation. The surface of solar panels is prone to accumulate dust, sand and other dirt. If there is no timely, scientific and professional cleaning and monitoring and other maintenance operations, it can cause the power generation of the components to decay by up to 40%-60%, and the power generation to drop by 20%-30%. Therefore, the concept of improving the power generation and efficiency of power stations by reasonably and scientifically cleaning and maintaining solar panels and carefully maintaining components has been recognized by the industry.

[0003] Currently, the market mainly relies on adding specific sensing devices or fixtures to the photovoltaic array for positioning to maintain the machine's walking, cleaning or monitoring operations on the photovoltaic array. This induction method of adding specific sensing devices or fixtures not only increases the cost of the equipment, but is also easily limited by the size and shape of the photovoltaic array, making it impossible to install additional devices. After installing specific sensing devices such as rails or signal base stations, the equipment can only be used in the current area and cannot be used in other unmodified power stations, which seriously affects operating efficiency and economic benefits.

[0004] Currently, photovoltaic cleaning robots that rely on visual SLAM (simultaneous localization and mapping) are mainly used in distributed power stations. After cleaning a certain array or area, each time a new array or area is replaced, the initial positioning and mapping will be re-initialized, which is inefficient. Moreover, it is impossible to repeatedly clean a specific area in a power station for a long time. Utility Model Content

[0005] In view of the above problems existing in the prior art, the main purpose of the utility model is to provide a photovoltaic cleaning robot based on visual SLAM.

[0006] The technical solution of the utility model is as follows:

[0007] A photovoltaic cleaning robot based on visual SLAM, the photovoltaic cleaning robot comprising a fuselage, a front end and a rear end of the fuselage are respectively provided with cleaning components, and two sides of the fuselage are respectively provided with crawler drive components, the photovoltaic cleaning robot walks under the drive of the crawler drive components, and cleans the photovoltaic components in the photovoltaic array in the photovoltaic power station through the cleaning components;

[0008] The crawler drive assembly is provided with a wheel encoder, and the wheel encoder is used to obtain the rotation speed of the crawler drive assembly and obtain the wheel odometer. The crawler drive assembly is also provided with an inertial measurement unit, and the inertial measurement unit is used to measure the posture information of the fuselage, and then the wheel odometer and the posture information are coupled as the basic positioning of the photovoltaic cleaning robot;

[0009] A first visual sensor is fixedly arranged at the front end of the bottom of the fuselage, the first visual sensor faces the ground, and the first visual sensor is used to detect the photovoltaic components at the bottom of the fuselage and obtain a first line feature; a second visual sensor is fixedly arranged at the top of the front of the fuselage, the second visual sensor faces the front of the fuselage, and the second visual sensor is used to detect the photovoltaic components at the front of the fuselage and obtain a second line feature;

[0010] A satellite navigation system positioning module is fixedly arranged on the rear side of the top of the fuselage, and the satellite navigation system positioning module is used to provide the latitude and longitude information of the photovoltaic cleaning robot in real time;

[0011] By using the first line feature obtained by the first visual sensor or the second line feature obtained by the second visual sensor, a map is constructed and positioned in real time, and navigation points and paths are planned in real time according to default specific rules and manual settings to complete the SLAM operation of autonomous exploration on the photovoltaic array;

[0012] When exploring a new area for the first time and pre-loading a new map, the longitude and latitude information provided by the satellite navigation system positioning module is recorded or provided, and the longitude and latitude range of the cleaning operation area is recorded and stored accordingly; when starting a new round of cleaning operation tasks, the longitude and latitude provided by the satellite navigation system positioning module are first compared with the longitude and latitude range of the preset historical map data. If there is a corresponding range, it is recommended to load the corresponding map, and provide a rough initialization position and area map for the cleaning operation based on the corresponding coordinates.

[0013] The first visual sensor is a first camera component, and the parameters of the first camera component are: 720P and 80-degree lens.

[0014] The first visual sensor uses an OV7720 camera capable of night vision.

[0015] The first visual sensor is used to detect the grid lines of the solar cells in the photovoltaic array at the bottom of the fuselage and the interval between two adjacent solar cells.

[0016] The first visual sensor is also used to detect the frame of the photovoltaic components in the photovoltaic array at the bottom of the body of the photovoltaic cleaning robot, the connecting components between two adjacent photovoltaic components, and the distance between two connected photovoltaic components.

[0017] The second visual sensor is a second camera component, and the parameters of the second camera component are: 1080P and 140-degree lens.

[0018] The second visual sensor adopts the H7650 camera with night vision.

[0019] The second visual sensor is used to detect the interval between two adjacent solar cells in the photovoltaic array in front of the body of the photovoltaic cleaning robot.

[0020] The second visual sensor is also used to detect the frame of the photovoltaic components in the photovoltaic array in front of the fuselage, the connecting components between two adjacent photovoltaic components, and the distance between two connected photovoltaic components.

[0021] The model of the satellite navigation system positioning module is WTGPS-300P, and the positioning accuracy is 1.5m.

[0022] The utility model has the following advantages and beneficial effects: the photovoltaic robot based on visual SLAM provided by the embodiment of the utility model senses the surface image of the photovoltaic component through the first visual sensor and the second visual sensor (i.e., the photosensitive device) during the cleaning operation, and performs semantic segmentation processing and extracts line features on the image data, uses the boundary and line features to build maps and locate in real time, and integrates the longitude and latitude information provided by the positioning module of the satellite navigation system as information record and auxiliary positioning of the global map, thereby realizing efficient operation in multiple scenarios, across regions and on duty. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] Figure 1 A schematic diagram of the three-dimensional structure of a photovoltaic cleaning robot provided in an embodiment of the utility model.

[0024] Figure 2 A schematic diagram of a photovoltaic cleaning robot provided in an embodiment of the utility model performing cleaning operations on a photovoltaic array. DETAILED DESCRIPTION

[0025] In order to make the purpose, technical scheme and advantages of the embodiment of the utility model clearer, the technical scheme in the embodiment of the utility model will be clearly and completely described below in conjunction with the drawings in the embodiment of the utility model. Obviously, the described embodiment is a part of the embodiment of the utility model, rather than all the embodiments. The components of the embodiment of the utility model generally described and shown in the drawings here can be arranged and designed in various different configurations. Therefore, the following detailed description of the embodiment of the utility model provided in the drawings is not intended to limit the scope of the utility model for protection, but merely represents the selected embodiment of the utility model. Based on the embodiments in the utility model, all other embodiments obtained by ordinary technicians in this field without making creative work belong to the scope of protection of the utility model.

[0026] In the description of the present invention, it should be noted that the terms "center", "upper", "lower", "left", "right", "vertical", "horizontal", "inner", "outer", etc., indicating the orientation or positional relationship, are based on the orientation or positional relationship shown in the drawings, and are only for the convenience of describing the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. In addition, the terms "first", "second", and "third" are used for descriptive purposes only, and cannot be understood as indicating or implying relative importance.

[0027] In the description of the present invention, it is also necessary to explain that, unless otherwise clearly specified and limited, the terms "set", "install", "connect", and "connect" should be understood in a broad sense, for example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, or it can be the internal communication of two components. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances.

[0028] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0029] like Figure 1 to Figure 2 As shown: A photovoltaic cleaning robot based on visual SLAM is provided in an embodiment of the utility model, the photovoltaic cleaning robot comprises a body 100, a cleaning assembly 101 is respectively provided at the front end and the rear end of the body 100, and a crawler drive assembly 102 is respectively provided at both sides of the body 101, the photovoltaic cleaning robot walks under the drive of the crawler drive assembly 102, and cleans the photovoltaic assembly 201 in the photovoltaic array 200 in the photovoltaic power station through the cleaning assembly 101;

[0030] The crawler drive assembly 102 is provided with a wheel encoder (not shown in the figure), and the wheel encoder is used to obtain the rotation speed of the crawler drive assembly 102 and obtain the wheel odometer. The crawler drive assembly 102 is also provided with an inertial measurement unit (not shown in the figure), and the inertial measurement unit is used to measure the posture information of the fuselage, and then the wheel odometer and the posture information are coupled as the basic positioning of the photovoltaic cleaning robot;

[0031] A first visual sensor 111 is fixedly arranged at the front end of the bottom of the fuselage 100, the first visual sensor 111 faces the ground, and is used to detect the photovoltaic component at the bottom of the fuselage 100 and obtain a first line feature; specifically, the first line feature includes the grid lines of the battery cells in the photovoltaic component 201 at the bottom of the fuselage 100 and the interval between two adjacent battery cells; a second visual sensor 112 is fixedly arranged at the top end of the front part of the fuselage 100, the second visual sensor 112 faces the front of the fuselage 100, and is used to detect the photovoltaic component at the front of the fuselage 100 and obtain a second line feature; specifically, the second line feature includes the interval between two adjacent battery cells in the photovoltaic component 201 at the front of the fuselage 100, the frame of the photovoltaic component 201, and the connecting component between two adjacent photovoltaic components 201.

[0032] A satellite navigation system positioning module 300 is fixedly arranged on the rear side of the top of the fuselage 100, wherein the model of the satellite navigation system positioning module 300 is WTGPS-300P, and the positioning accuracy is 1.5m; the satellite navigation system positioning module 300 is used to provide the latitude and longitude information of the photovoltaic cleaning robot in real time;

[0033] By using the first line features obtained by the first visual sensor 111 or the second line features obtained by the second visual sensor 112, a map and positioning are constructed in real time, and navigation points and paths are planned in real time according to default specific rules and manual settings to complete the SLAM operation of autonomous exploration on the photovoltaic array.

[0034] Specifically: first, through the second line feature, the second line feature is projected on a two-dimensional plane from a bird's-eye view through perspective transformation, the corresponding lines are pieced together and supplemented with objects such as photovoltaic components or connecting bridges, and then the positions of components or connecting bridges are matched on the map, and at the same time, the current posture of the photovoltaic cleaning robot and the distribution of photovoltaic components in the cleaning map are combined to deduce the global rough position of the photovoltaic cleaning robot relative to the current component plane in the longitudinal and lateral directions. Then, the image captured by the first visual sensor 111 is first classified by deep learning to determine whether the photovoltaic cleaning robot is currently on the plane of objects such as photovoltaic components, or on the connection of objects such as photovoltaic components or connecting bridges, and then paired again.

[0035] If the photovoltaic cleaning robot is on the plane of an object such as a photovoltaic module, a local straight reference line path of the observation plane can be obtained based on the first line feature. The local straight reference line path may include the grid lines of the battery cells in the photovoltaic module and the interval between two adjacent battery cells, and the artificially processed guide lines on the connecting bridge. The coordinates of the first line feature straight line are obtained through straight line detection, and the lateral distance and angle deviation are calculated with the feature straight line of the map in the forward direction. The local relative lateral precise position of the current first visual sensor 111 and the battery cells in the photovoltaic module is then calculated through coordinate conversion.

[0036] If the photovoltaic cleaning robot is at the connection of objects such as photovoltaic components or connecting bridges, the coordinates of the second line feature straight line obtained by straight line detection and the starting position of the longitudinal interval between other objects such as photovoltaic components and connecting bridges in the map of the forward direction can be calculated based on the first line feature, the horizontal line of the connection interval between the current plane and the plane of the object in the forward direction, and the deviation can be calculated to calculate the local relative longitudinal precise position of the current first visual sensor 111 and the connection. Finally, based on the global rough position of the second line feature and the local relative horizontal or vertical precise position of the first line feature, the global relative precise position of the photovoltaic cleaning robot in the map can be calculated, so as to correct the posture of the photovoltaic cleaning robot in time.

[0037] When exploring a new area for the first time and pre-loading a new map, the longitude and latitude information provided by the satellite navigation system positioning module is recorded or provided, and the longitude and latitude range of the cleaning operation area is recorded and stored accordingly; when starting a new round of cleaning operation tasks, the longitude and latitude provided by the satellite navigation system positioning module are first compared with the longitude and latitude range of the preset historical map data. If there is a corresponding range, it is recommended to load the corresponding map, and provide a rough initialization position and area map for the cleaning operation based on the corresponding coordinates.

[0038] The first visual sensor 111 is a first camera component, and the parameters of the first camera component are: 720P and 80-degree lens; specifically, the first visual sensor adopts an OV7720 camera with night vision; the first visual sensor 111 is used to detect the grid lines of the battery cells in the photovoltaic array 200 at the bottom of the fuselage 100 and the interval between two adjacent battery cells.

[0039] The first visual sensor 111 is also used to detect the frame of the photovoltaic component 201 in the photovoltaic array 200 at the bottom of the body 100 of the photovoltaic cleaning robot, the connecting component between two adjacent photovoltaic components 201, and the distance between two connected photovoltaic components 201.

[0040] The second visual sensor 112 is a second camera component, and the parameters of the second camera component are: 1080P and 140-degree lens. Specifically, the second visual sensor adopts a H7650 camera with night vision; the second visual sensor 112 is used to detect the interval between two adjacent cells in the photovoltaic array 200 in front of the body of the photovoltaic cleaning robot.

[0041] The second visual sensor 112 is also used to detect the frame of the photovoltaic components 201 in the photovoltaic array 200 in front of the fuselage, the connecting components between two adjacent photovoltaic components 201, and the distance between two connected photovoltaic components 201.

[0042] Cleaning robots for fixed track and fixed model base stations cannot change the cleaning area, and photovoltaic cleaning robots relying on visual SLAM (simultaneous localization and mapping) cannot effectively fix the cleaning area, which leads to low equipment utilization and low efficiency. A photovoltaic cleaning robot based on visual SLAM is adopted and integrated with a satellite navigation system to improve the automation level of the photovoltaic cleaning robot and improve the applicability of the photovoltaic cleaning robot, so that the photovoltaic cleaning robot can not only move autonomously in the distributed photovoltaic power station array scene, but also be suitable for the customized duty operation of the fixed unattended power station photovoltaic array, ultimately saving manpower and reducing the maintenance cost of the power station.

[0043] The photovoltaic cleaning robot based on visual SLAM in the embodiment of the utility model senses the surface image of the photovoltaic component through the first visual sensor and the second visual sensor (i.e., photosensitive device) during the cleaning operation, and performs semantic segmentation processing and extracts line features on the image data, uses the boundary and line features to build maps and locate in real time, and integrates the longitude and latitude information provided by the positioning module of the satellite navigation system as information record and auxiliary positioning of the global map, thereby realizing efficient operation in multiple scenarios, across regions and on duty.

[0044] The photovoltaic operation robot starts with one button and autonomously builds maps and locates in the photovoltaic array in real time to achieve the cleaning operation effect. It also uses the satellite navigation system positioning module to record and save the longitude and latitude coordinates and range of the current area, and sets a guard point. Alternatively, the regional map can be manually entered in advance according to the on-site scene or construction drawings, and its longitude and latitude coordinates and range can be included. If it is operated again within the area, the previously stored map can be directly loaded and the route can be planned according to the cleaning needs, providing the photovoltaic cleaning robot with global map information, reducing the time of global exploration of the environment again, and improving cleaning efficiency.

[0045] It is common to obtain the speed of the drive device through a wheel encoder, and convert the wheel mileage according to the mechanical structure, and then tightly couple it with the posture information measured by the inertial measurement unit (IMU) as the basic positioning. The camera on the top of the front of the robot body (hereinafter referred to as the front camera) faces forward, and the projected image is used to extract the boundary data of the component cell area, such as the frame of the component, the frame bracket installed between the components and the gap interval, and the width of the middle cell of some half-cell components is 1.5-4cm, which is relatively large, and the boundary processing is also performed. The camera at the bottom of the front of the robot body (hereinafter referred to as the bottom camera) faces the component, and the projected image is used to extract the boundary information of the component cell area, such as the frame of the component, the frame bracket installed between the components and the gap interval, and the width of the middle cell of some half-cell components is 1.5-4cm, which is relatively large, and the boundary processing is also performed. Then the camera installed at the bottom of the front of the robot body (hereinafter referred to as the bottom camera) faces the component, and the line feature extraction of the image is performed to extract the grid lines of the component cells, the intervals between the component cells and the frame of the component, the line features of the objects installed between the components and the gap interval, and whether they are in the photovoltaic component panel.

[0046] The boundary information and line features are then matched with the PV module cell area boundary data and module specifications in the map coordinate system that is constructed in real time or loaded in advance, and the reprojection error is calculated. The visual mileage that minimizes the reprojection error is then obtained through nonlinear optimization. Finally, the visual mileage and wheel mileage are fused through the extended Kalman filter to obtain the final pose. Visual pairing correction and auxiliary positioning are used to obtain a globally accurate positioning system.

[0047] The signal antenna of the satellite navigation system positioning module is installed near the non-metallic surface on the rear side of the top of the fuselage of the photovoltaic cleaning robot. When the photovoltaic cleaning robot explores the new area array for the first time, the latitude and longitude information provided by the satellite navigation system positioning module in the map corresponding process established by SLAM is saved accordingly; when the scene layout and component specifications are manually understood, the map is manually drawn according to the scene in advance and sent to the photovoltaic cleaning robot, and the longitude and latitude information is provided. When the photovoltaic cleaning robot starts a new operation task, the longitude and latitude provided by the current satellite navigation system positioning module is first compared with the longitude and latitude range of the stored historical map data. If there is a corresponding range, it is recommended to load the corresponding map first, and provide the regional map for the operation according to the corresponding coordinates. Although the horizontal positioning accuracy of the satellite navigation system positioning module is easily affected by many factors, it can also provide a rough initialization position. Alternatively, the stored regional map and current position can be manually selected to provide the photovoltaic cleaning robot with global map information. The photovoltaic cleaning robot can directly plan the route according to the cleaning needs, reduce the time of re-exploring the environment globally and repositioning the preliminary screening process, and improve the cleaning efficiency.

[0048] Finally, it should be noted that the above-described embodiments are only used to illustrate the technical solution of the utility model, rather than to limit it. Although the utility model 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 replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the various embodiments of the utility model.

Claims

1. A photovoltaic cleaning robot based on visual SLAM, characterized in that: The photovoltaic cleaning robot comprises a fuselage, a front end and a rear end of the fuselage are respectively provided with cleaning components, and both sides of the fuselage are respectively provided with crawler drive components, the photovoltaic cleaning robot walks under the drive of the crawler drive components, and cleans the photovoltaic components in the photovoltaic array in the photovoltaic power station through the cleaning components; The crawler drive assembly is provided with a wheel encoder, and the wheel encoder is used to obtain the rotation speed of the crawler drive assembly and obtain the wheel odometer. The crawler drive assembly is also provided with an inertial measurement unit, and the inertial measurement unit is used to measure the posture information of the fuselage, and then the wheel odometer and the posture information are coupled as the basic positioning of the photovoltaic cleaning robot; A first visual sensor is fixedly arranged at the front end of the bottom of the fuselage, the first visual sensor faces the ground, and the first visual sensor is used to detect the photovoltaic components at the bottom of the fuselage and obtain a first line feature; A second visual sensor is fixedly arranged at the top of the front part of the fuselage, the second visual sensor faces the front of the fuselage, and the second visual sensor is used to detect the photovoltaic assembly in front of the fuselage and obtain a second line feature; A satellite navigation system positioning module is fixedly arranged on the rear side of the top of the fuselage, and the satellite navigation system positioning module is used to provide the latitude and longitude information of the photovoltaic cleaning robot in real time.

2. The photovoltaic cleaning robot based on visual SLAM according to claim 1, characterized in that: The first visual sensor is a first camera component, and the parameters of the first camera component are: 720P and 80-degree lens.

3. The photovoltaic cleaning robot based on visual SLAM according to claim 2, characterized in that: The first visual sensor uses an OV7720 camera capable of night vision.

4. The photovoltaic cleaning robot based on visual SLAM according to claim 2, characterized in that: The first visual sensor is used to detect the grid lines of the solar cells in the photovoltaic array at the bottom of the fuselage and the interval between two adjacent solar cells.

5. The photovoltaic cleaning robot based on visual SLAM according to claim 4, characterized in that: The first visual sensor is also used to detect the frame of the photovoltaic components in the photovoltaic array at the bottom of the body of the photovoltaic cleaning robot, the connecting components between two adjacent photovoltaic components, and the distance between two connected photovoltaic components.

6. The photovoltaic cleaning robot based on visual SLAM according to claim 1, characterized in that: The second visual sensor is a second camera component, and the parameters of the second camera component are: 1080P and 140-degree lens.

7. The photovoltaic cleaning robot based on visual SLAM according to claim 6, characterized in that: The second visual sensor adopts the H7650 camera with night vision.

8. The photovoltaic cleaning robot based on visual SLAM according to claim 5, characterized in that: The second visual sensor is used to detect the interval between two adjacent solar cells in the photovoltaic array in front of the body of the photovoltaic cleaning robot.

9. The photovoltaic cleaning robot based on visual SLAM according to claim 5, characterized in that: The second visual sensor is also used to detect the frame of the photovoltaic components in the photovoltaic array in front of the fuselage, the connecting components between two adjacent photovoltaic components, and the distance between two connected photovoltaic components.

10. The photovoltaic cleaning robot based on visual SLAM according to claim 5, characterized in that: The model of the satellite navigation system positioning module is WTGPS-300P, and the positioning accuracy is 1.5m.

Images