Vision-based local control method and system for photovoltaic cleaning robot

Abstract

The invention discloses a local control method and system of a photovoltaic cleaning robot based on vision, belongs to the technical field of photovoltaic cleaning robot control, and solves the problem that an existing photovoltaic cleaning robot is low in operation efficiency. The photovoltaic panel surface state information and the position state information of the photovoltaic cleaning robot are obtained, so that the photovoltaic panel surface area needing to be cleaned by the photovoltaic cleaning robot is determined, path planning of the photovoltaic cleaning robot is conducted, and overall effective cleaning of the photovoltaic panel is achieved; local effective cleaning of a photovoltaic panel is realized by acquiring real-time positioning state information of a photovoltaic cleaning robot, planning an optimal path of a to-be-cleaned area, acquiring pollution degree information of the surface of the photovoltaic panel and determining cleaning frequency; local path planning of the photovoltaic cleaning robot is achieved, path planning is conducted autonomously according to the range where stains are located, local powerful cleaning is achieved, and the operation efficiency of the photovoltaic cleaning robot is improved.

Description

Technical Field

[0001] The invention belongs to the technical field of photovoltaic cleaning robot control, and relates to a vision-based local control method and system of a photovoltaic cleaning robot. Background Art

[0002] Photovoltaic panels are set up in outdoor environments, which are used to absorb sunlight and convert it into electrical energy. When photovoltaic panels are affected by the outdoor environment for a long time, dust and other particles will inevitably accumulate on their surfaces, causing the sunlight reaching the surface of the photovoltaic panels to be scattered and reducing the photovoltaic panels' absorption efficiency of sunlight, thereby affecting the photovoltaic panels' power generation efficiency. In order to ensure the normal and stable power generation of photovoltaic panels, the surface of the photovoltaic panels needs to be cleaned regularly. Photovoltaic cleaning robots are photovoltaic cleaning robots specifically used to clean photovoltaic panels. By using water sprays, brushes, etc. to clean photovoltaic panels, the use of photovoltaic cleaning robots can greatly reduce the cost and labor intensity of manual cleaning, while also ensuring efficient power generation of photovoltaic panels. Existing photovoltaic cleaning robots usually clean along a specified path on the surface of the photovoltaic panel according to a pre-set mode, making it impossible for the photovoltaic cleaning robot to choose a cleaning method according to the actual situation on site, which can easily cause problems such as missed cleaning or incomplete cleaning, thereby affecting the cleaning efficiency and reliability of the photovoltaic panel. Summary of the invention

[0003] The technical solution of the present invention is used to solve the problem of low operating efficiency of the existing photovoltaic cleaning robot.

[0004] The present invention solves the above technical problems through the following technical solutions:

[0005] A local control method of a photovoltaic cleaning robot based on vision comprises the following steps:

[0006] Step 1: Use a drone to cruise and shoot to obtain information about the surface status of the photovoltaic panels and the location status of the photovoltaic cleaning robot;

[0007] Step 2: According to the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot, the A-Star algorithm is used to plan the motion path of the photovoltaic cleaning robot;

[0008] Step 3: Obtain the real-time positioning status information of the photovoltaic cleaning robot and use the A-Star algorithm to plan the optimal path for the area to be cleaned;

[0009] Step 4: Use image segmentation algorithm to detect the surface color of photovoltaic panels, obtain the degree of contamination on the surface of photovoltaic panels, and plan the cleaning frequency;

[0010] Step 5: confirm whether to continue cleaning based on the cleaning image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process.

[0011] Furthermore, the method described in step 2 for using the A-Star algorithm to plan the motion path of the photovoltaic cleaning robot according to the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot is as follows: the surface of the photovoltaic panel is divided into a series of grids, each grid is set to 1m×1m, the photovoltaic panel is divided into a whole grid, the initial position of the photovoltaic cleaning robot is set as the starting point S, and the end point of the photovoltaic panel or the last area to be cleaned is set as the target point G; when searching for the path, the weight is calculated for each possible path node, and the path with the minimum total cost is selected; the weight includes the known cost g(n) from the starting point to the current node and the heuristic estimated cost h(n) from the current node to the target point; the total cost function f(n) of each node n is defined as: f(n)=g(n)+h(n); wherein g(n) is the actual cost from the starting point to the current node, and h(n) is the heuristic estimate to the target point calculated by the Manhattan distance, and the calculation formula is h(n)=|x current -x goal ∣+∣y current -y goal ∣.

[0012] Furthermore, the method of obtaining the real-time positioning status information of the photovoltaic cleaning robot and using the A-Star algorithm to plan the optimal path for the area to be cleaned as described in step three is as follows: when the position or map information of the photovoltaic cleaning robot is updated in real time, the A-Star algorithm is used to recalculate the path cost: assuming that node n is updated, g(n′) is recalculated for the neighboring node n′ and f(n′) is updated, that is, f(n′)=g(n′)+h(n′). If it is found that the path cost increases or encounters new obstacles, the A-Star algorithm backtracks forward or backward and continues searching from the known optimal node until a new optimal path is found.

[0013] Furthermore, the method of obtaining the pollution degree information of the photovoltaic panel surface and planning the cleaning frequency according to the pollution degree described in step 4 is specifically as follows: using an image segmentation algorithm to detect the color of the photovoltaic panel surface, separate the polluted area from the background, and form a binary image I of the polluted area binary , according to the center coordinates of the binary image recognition area (x c ,y c ) and the pollution area A; if the pollution area A is greater than the set threshold, a cleaning path is generated; the pollution factor p(i,j) is set to represent the pollution level on the grid (i,j), and the pollution factor p(i,j) is graded according to the sensor data. The value range of p(i,j) is set to [0,1], where 0 represents the clean state and 1 represents the heavy pollution state; the cleaning frequency f is defined clean(i,j) is the number of times or priority of the photovoltaic cleaning robot to clean the grid, f clean(i,j)The larger the value, the higher the cleaning frequency. clean(i,j) = k × p (i, j), k is a proportional constant used to adjust the overall baseline value of the cleaning frequency; the cleaning frequency and path cost are unified into one optimization goal, and the total cost function F (n) is updated to take into account both the path cost f (n) and the cleaning frequency f clean(i,j) , that is, F(n)=f(n)+α×f clean(i,j) , to ensure that the cleaning work is carried out normally.

[0014] Furthermore, the method for confirming whether to continue cleaning according to the cleaned image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process described in step 5 is specifically as follows: collecting N×M images of the cleaned area, binarizing the image, obtaining the contaminated pixel P, calculating the ratio of the contaminated residual area to the total area, and obtaining the contamination residual degree Set the cleaning threshold R of the residual contamination threshold , when R≤R threshold , the cleaning is considered completed, and the photovoltaic cleaning robot stops cleaning the area; if R ≥ R threshold , then repeat the path planning and cleaning process for the area.

[0015] A local control system of a photovoltaic cleaning robot based on vision, comprising the following steps: a position state information acquisition module, a motion path planning module, an optimal path planning module for the area to be cleaned, a pollution degree acquisition and cleaning frequency planning module, and a cleaning completion judgment module;

[0016] The position status information acquisition module is used to obtain the surface status information of the photovoltaic panel and the position status information of the photovoltaic cleaning robot by performing cruise shooting by the drone;

[0017] The motion path planning module: according to the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot, the A-Star algorithm is used to plan the motion path of the photovoltaic cleaning robot;

[0018] The optimal path planning module for the area to be cleaned: obtains the real-time positioning status information of the photovoltaic cleaning robot and uses the A-Star algorithm to plan the optimal path for the area to be cleaned;

[0019] The pollution degree acquisition and cleaning frequency planning module: uses an image segmentation algorithm to detect the surface color of the photovoltaic panel, obtains the pollution degree information of the photovoltaic panel surface, and plans the cleaning frequency;

[0020] The cleaning completion judgment module is used to determine whether to continue cleaning based on the cleaning image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process.

[0021] Furthermore, the method of using the A-Star algorithm to plan the motion path of the photovoltaic cleaning robot according to the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot described in the motion path planning module is specifically as follows: the surface of the photovoltaic panel is divided into a series of grids, each grid is set to 1m×1m, the photovoltaic panel is divided into a whole grid, the initial position of the photovoltaic cleaning robot is set as the starting point S, and the end point of the photovoltaic panel or the last area to be cleaned is set as the target point G; when searching for the path, the weight is calculated for each possible path node, and the path with the minimum total cost is selected; the weight includes the known cost g(n) from the starting point to the current node and the heuristic estimated cost h(n) from the current node to the target point; the total cost function f(n) of each node n is defined as: f(n)=g(n)+h(n); wherein g(n) is the actual cost from the starting point to the current node, and h(n) is the heuristic estimate to the target point calculated by the Manhattan distance, and the calculation formula is h(n)=|x current -x goal ∣+∣y current -y goal ∣.

[0022] Furthermore, the method of obtaining the real-time positioning status information of the photovoltaic cleaning robot and using the A-Star algorithm to plan the optimal path of the area to be cleaned described in the optimal path planning module of the area to be cleaned is specifically as follows: when the position or map information of the photovoltaic cleaning robot is updated in real time, the A-Star algorithm is used to recalculate the path cost: assuming that node n is updated, g(n′) is recalculated for the neighboring node n′ and f(n′) is updated, that is, f(n′)=g(n′)+h(n′); if it is found that the path cost increases or encounters new obstacles, the A-Star algorithm backtracks forward or backward and continues searching from the known optimal node until a new optimal path is found.

[0023] Furthermore, the method for obtaining the pollution degree information on the surface of the photovoltaic panel and planning the cleaning frequency according to the pollution degree described in the pollution degree acquisition and cleaning frequency planning module is as follows: using an image segmentation algorithm to detect the color of the surface of the photovoltaic panel, separating the polluted area from the background, and forming a binary image I of the polluted area binary , according to the center coordinates of the binary image recognition area (x c ,y c ) and the pollution area A; if the pollution area A is greater than the set threshold, a cleaning path is generated; the pollution factor p(i,j) is set to represent the pollution level on the grid (i,j), and the pollution factor p(i,j) is graded according to the sensor data. The value range of p(i,j) is set to [0,1], where 0 represents the clean state and 1 represents the heavy pollution state; the cleaning frequency f is defined clean(i,j) is the number of times or priority of the photovoltaic cleaning robot to clean the grid, fclean(i,j) The larger the value, the higher the cleaning frequency. clean(i,j) = k × p (i, j), k is a proportional constant used to adjust the overall baseline value of the cleaning frequency; the cleaning frequency and path cost are unified into one optimization goal, and the total cost function F (n) is updated to take into account both the path cost f (n) and the cleaning frequency f clean(i,j) , that is, F(n)=f(n)+α×f clean(i,j) , to ensure that the cleaning work is carried out normally.

[0024] Furthermore, the method for confirming whether to continue cleaning according to the cleaned image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process described in the cleaning completion judgment module is specifically as follows: collecting N×M images of the cleaned area, binarizing the image, obtaining the contaminated pixel P, calculating the ratio of the contaminated residual area to the total area, and obtaining the contamination residual degree Set the cleaning threshold R of the residual contamination threshold , when R≤R threshold , the cleaning is considered completed, and the photovoltaic cleaning robot stops cleaning the area; if R ≥ R threshold , then repeat the path planning and cleaning process for the area.

[0025] The advantages of the present invention are:

[0026] The vision-based local control method of the photovoltaic cleaning robot of the present invention cruises and shoots the space where the photovoltaic panels are located, obtains the surface state information of the photovoltaic panels and the position state information of the photovoltaic cleaning robot, so as to determine the surface area of ​​the photovoltaic panels to be cleaned by the photovoltaic cleaning robot, and performs path planning of the photovoltaic cleaning robot to achieve global and effective cleaning of the photovoltaic panels; by acquiring the real-time positioning state information of the photovoltaic cleaning robot, planning the optimal path of the area to be cleaned, and acquiring the pollution degree information of the surface of the photovoltaic panels, determining the cleaning frequency, and achieving local and effective cleaning of the photovoltaic panels; the present invention realizes the local path planning of the photovoltaic cleaning robot, autonomously performs path planning according to the range of the stains, achieves local and powerful cleaning, and greatly improves the operating efficiency of the photovoltaic cleaning robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] Figure 1 is a flow chart of a local control method of a photovoltaic cleaning robot based on vision according to Embodiment 1 of the present invention;

[0028] Figure 2 It is a structural diagram of a local control system of a vision-based photovoltaic cleaning robot according to the second embodiment of the present invention. DETAILED DESCRIPTION

[0029] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in combination with the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0030] The technical solution of the present invention is further described below in conjunction with the accompanying drawings and specific embodiments:

[0031] Embodiment 1

[0032] like Figure 1 As shown, this embodiment provides a local control method of a photovoltaic cleaning robot based on vision, including:

[0033] Step 1: Use a drone to cruise and shoot to obtain information about the surface status of the photovoltaic panels and the location status of the photovoltaic cleaning robot.

[0034] In order to obtain a global image of the area where the photovoltaic panels are located, the aerial photography drone is divided into regions based on the boundary area, and the global image information obtained by the aerial photography drone is summarized to obtain a global image of the area where the photovoltaic panels are located, which can improve the efficiency of aerial photography and avoid the occurrence of missed shots; the global image is subjected to picture contour recognition to obtain picture contour information of the global image; based on the picture contour information, the position information of all photovoltaic cleaning robots in the area where the photovoltaic panels are located and the distribution position information of stains on the photovoltaic surface are obtained, so as to accurately locate and identify all photovoltaic cleaning robots and stains on the surface of the photovoltaic panels.

[0035] Step 2: Plan the motion path of the photovoltaic cleaning robot based on the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot.

[0036] In order to realize the path planning of the photovoltaic cleaning robot, the A-Star algorithm is used to generate the optimal path. The surface of the photovoltaic panel is divided into a series of grids, each grid is set to 1m×1m, the photovoltaic panel is divided into a whole grid, the initial position of the photovoltaic cleaning robot is set as the starting point S, and the end point of the photovoltaic panel or the last area to be cleaned is set as the target point G. When searching for a path, the A-Star algorithm calculates weights for each possible path node and selects the path that minimizes the total cost. These weights include the known cost g(n) from the starting point to the current node and the heuristic estimated cost h(n) from the current node to the target point. The total cost function f(n) of each node n is defined as: f(n)=g(n)+h(n). Among them, g(n) is the actual cost from the starting point to the current node, for example, each step forward on the path accumulates 1 unit distance (such as the g(n) value from the starting point to (2,3) is 5); h(n) is the heuristic estimate to the target point calculated by Manhattan distance, and the calculation formula is h(n)=|x current -x goal ∣+∣y current -y goal |; For example, if the current position is (2,3) and the target is (5,6), then h(n) = |2-5|+|3-6|=6.

[0037] Step 3: Obtain the real-time positioning status information of the photovoltaic cleaning robot and plan the optimal path for the area to be cleaned.

[0038] When the location or map information of the photovoltaic cleaning robot is updated in real time, the A-Star algorithm will recalculate the path cost. If node n is updated, g(n′) is recalculated for the neighboring node n′ and f(n′) is updated, that is, f(n′)=g(n′)+h(n′). If it is found that the path cost increases or encounters new obstacles, the A-Star algorithm will backtrack forward or backward and continue searching from the known optimal node until a new optimal path is found.

[0039] Step 4: Obtain information on the degree of contamination on the surface of the photovoltaic panels and plan the cleaning frequency based on the degree of contamination.

[0040] The image segmentation algorithm is used to detect the surface color of the photovoltaic panel, separate the polluted area from the background, and form a binary image I of the polluted area. binary , according to the center coordinates of the binary image recognition area (x c ,y c) and the contaminated area A; if the contaminated area A is greater than the set threshold, a cleaning path is generated. Plan the cleaning frequency according to the degree of contamination, and set the contamination factor p(i,j) to represent the contamination level on the grid (i,j). Classify the contamination factor p(i,j) according to the sensor data, and set the value range of p(i,j) to [0,1], where 0 represents a clean state and 1 represents a heavily contaminated state. Define the cleaning frequency f clean(i,j) is the number of times or priority of the photovoltaic cleaning robot to clean the grid, f clean(i,j) The larger the value, the higher the cleaning frequency. clean(i,j) = k × p(i, j), k is a proportional constant used to adjust the overall baseline value of the cleaning frequency. In order to unify the cleaning frequency and path cost into one optimization goal, the total cost function F(n) can be updated to take into account both the path cost f(n) and the cleaning frequency f clean(i,j) , that is, F(n)=f(n)+α×f clean(i,j) , to ensure that the cleaning work is carried out normally.

[0041] Step 5: Determine whether to continue cleaning based on the cleaning image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process.

[0042] After the photovoltaic cleaning robot completes local cleaning, it collects the N×M images of the cleaned area through the visual system, performs binarization on the image, obtains the polluted pixel P, calculates the ratio of the polluted residual area to the total area, and obtains the residual degree of pollution. Set the cleaning threshold R of the residual contamination threshold =0.05, when R≤R threshold , cleaning is considered complete, and the photovoltaic cleaning robot can stop cleaning the area. threshold , the path planning and cleaning process is repeated. The cleaning intensity and path planning parameters are dynamically adjusted according to the cleaning effect to adapt to different pollution conditions and ensure that the cleaning effect reaches the expected level.

[0043] Embodiment 2

[0044] like Figure 2 As shown, this embodiment provides a local control system of a photovoltaic cleaning robot based on vision, comprising the following steps: a position state information acquisition module, a motion path planning module, an optimal path planning module for the area to be cleaned, a pollution degree acquisition and cleaning frequency planning module, and a cleaning completion judgment module;

[0045] The position status information acquisition module is used to obtain the surface status information of the photovoltaic panel and the position status information of the photovoltaic cleaning robot by performing cruise shooting by the drone;

[0046] The motion path planning module: according to the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot, the A-Star algorithm is used to plan the motion path of the photovoltaic cleaning robot. The specific method is: the surface of the photovoltaic panel is divided into a series of grids, each grid is set to 1m×1m, the photovoltaic panel is divided into a whole grid, the initial position of the photovoltaic cleaning robot is set as the starting point S, and the end point of the photovoltaic panel or the last area to be cleaned is set as the target point G; when searching for the path, the weight is calculated for each possible path node, and the path with the minimum total cost is selected; the weight includes the known cost g(n) from the starting point to the current node and the heuristic estimated cost h(n) from the current node to the target point; the total cost function f(n) of each node n is defined as: f(n)=g(n)+h(n); wherein g(n) is the actual cost from the starting point to the current node, and h(n) is the heuristic estimate to the target point calculated by Manhattan distance, and the calculation formula is h(n)=|x current -x goal ∣+∣y current -y goal ∣.

[0047] The optimal path planning module for the area to be cleaned: obtains the real-time positioning status information of the photovoltaic cleaning robot, and uses the A-Star algorithm to plan the optimal path for the area to be cleaned. The specific method is: when the position or map information of the photovoltaic cleaning robot is updated in real time, the A-Star algorithm is used to recalculate the path cost: if the node n is updated, then the neighboring node n′ is recalculated g(n′) and f(n′) is updated, that is, f(n′)=g(n′)+h(n′). If it is found that the path cost increases or encounters a new obstacle, the A-Star algorithm backtracks forward or backward and continues to search from the known optimal node until a new optimal path is found.

[0048] The pollution degree acquisition and cleaning frequency planning module described above: uses an image segmentation algorithm to detect the surface color of the photovoltaic panel, obtains the pollution degree information of the photovoltaic panel surface, and plans the cleaning frequency. The specific method is: uses an image segmentation algorithm to detect the surface color of the photovoltaic panel, separates the polluted area from the background, and forms a binary image I of the polluted area. binary , according to the center coordinates of the binary image recognition area (x c ,y c ) and the pollution area A; if the pollution area A is greater than the set threshold, a cleaning path is generated; the pollution factor p(i,j) is set to represent the pollution level on the grid (i,j), and the pollution factor p(i,j) is graded according to the sensor data. The value range of p(i,j) is set to [0,1], where 0 represents the clean state and 1 represents the heavy pollution state; the cleaning frequency f is defined clean(i,j) is the number of times or priority of the photovoltaic cleaning robot to clean the grid, fclean(i,j) The larger the value, the higher the cleaning frequency. clean(i,j) = k × p (i, j), k is a proportional constant used to adjust the overall baseline value of the cleaning frequency; the cleaning frequency and path cost are unified into one optimization goal, and the total cost function F (n) is updated to take into account both the path cost f (n) and the cleaning frequency f clean(i,j) , that is, F(n)=f(n)+α×f clean(i,j) , to ensure that the cleaning work is carried out normally.

[0049] The cleaning completion judgment module is used to determine whether to continue cleaning based on the clean image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process. The specific method is: collect N×M images of the cleaned area, and perform binarization on the image to obtain the contaminated pixel P, calculate the ratio of the contaminated residual area to the total area, and obtain the contamination residual degree. Set the cleaning threshold R of the residual contamination threshold , when R≤R threshold , the cleaning is considered completed, and the photovoltaic cleaning robot stops cleaning the area; if R ≥ R threshold , then repeat the path planning and cleaning process for the area.

[0050] Embodiment 3

[0051] An electronic device includes a memory and a processor, wherein the memory is used to store a program that supports the processor to execute the local control method of the vision-based photovoltaic cleaning robot in Example 1, and the processor is configured to execute the program stored in the memory.

[0052] Embodiment 4

[0053] A storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the local control method of a photovoltaic cleaning robot based on vision in embodiment 1 are executed.

[0054] The above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A local control method of a photovoltaic cleaning robot based on vision, characterized in that: The following steps are involved: Step 1: Use a drone to cruise and shoot to obtain information about the surface status of the photovoltaic panels and the location status of the photovoltaic cleaning robot; Step 2: According to the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot, the A-Star algorithm is used to plan the motion path of the photovoltaic cleaning robot; Step 3: Obtain the real-time positioning status information of the photovoltaic cleaning robot and use the A-Star algorithm to plan the optimal path for the area to be cleaned; Step 4: Use image segmentation algorithm to detect the surface color of photovoltaic panels, obtain information on the degree of contamination on the surface of photovoltaic panels, and plan the cleaning frequency; Step 5: confirm whether to continue cleaning based on the cleaning image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process.

2. The local control method of the vision-based photovoltaic cleaning robot according to claim 1, characterized in that: The method described in step 2 for using the A-Star algorithm to plan the motion path of the photovoltaic cleaning robot based on the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot is as follows: the surface of the photovoltaic panel is divided into a series of grids, each grid is set to 1m×1m, the photovoltaic panel is divided into a whole grid, the initial position of the photovoltaic cleaning robot is set as the starting point S, and the end point of the photovoltaic panel or the last area to be cleaned is set as the target point G; when searching for the path, the weight is calculated for each possible path node, and the path with the minimum total cost is selected; the weight includes the known cost g(n) from the starting point to the current node and the heuristic estimated cost h(n) from the current node to the target point; the total cost function f(n) of each node n is defined as: f(n)=g(n)+h(n); where g(n) is the actual cost from the starting point to the current node, and h(n) is the heuristic estimate to the target point calculated by the Manhattan distance, and the calculation formula is h(n)=|x current -x goal ∣+∣y current -y goal ∣.

3. The local control method of the vision-based photovoltaic cleaning robot according to claim 2, characterized in that: The method described in step 3 for obtaining the real-time positioning status information of the photovoltaic cleaning robot and using the A-Star algorithm to plan the optimal path for the area to be cleaned is as follows: when the position or map information of the photovoltaic cleaning robot is updated in real time, the A-Star algorithm is used to recalculate the path cost: assuming that node n is updated, g(n′) is recalculated for the neighboring node n′ and f(n′) is updated, that is, f(n′)=g(n′)+h(n′). If it is found that the path cost increases or encounters new obstacles, the A-Star algorithm backtracks forward or backward and continues to search from the known optimal node until a new optimal path is found.

4. The local control method of the vision-based photovoltaic cleaning robot according to claim 3 is characterized in that: The method for obtaining the pollution degree information of the photovoltaic panel surface and planning the cleaning frequency according to the pollution degree described in step 4 is as follows: using the image segmentation algorithm to detect the color of the photovoltaic panel surface, separate the polluted area from the background, and form a binary image I of the polluted area. binary , according to the center coordinates of the binary image recognition area (x c ,y c ) and the contaminated area A; if the contaminated area A is greater than the set threshold, a cleaning path is generated; The pollution factor p(i,j) is set to represent the pollution level on the grid (i,j). The pollution factor p(i,j) is classified according to the sensor data. The value range of p(i,j) is set to [0,1], where 0 represents a clean state and 1 represents a heavily polluted state. Define the cleaning frequency f clean(i,j) is the number of times or priority of the photovoltaic cleaning robot to clean the grid, f clean(i,j) The larger the value, the higher the cleaning frequency. clean(i,j) =k×p(i,j), where k is a proportional constant used to adjust the overall reference value of the cleaning frequency; Unify the cleaning frequency and path cost into one optimization objective, and update the total cost function F(n) to take into account both the path cost f(n) and the cleaning frequency f clean(i,j) , that is, F(n)=f(n)+α×f clean(i,j) , to ensure that the cleaning work is carried out normally.

5. The local control method of the vision-based photovoltaic cleaning robot according to claim 4, characterized in that: The method for confirming whether to continue cleaning according to the cleaned image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process described in step 5 is as follows: collect N×M images of the cleaned area, perform binarization on the image, obtain the contaminated pixel P, calculate the ratio of the contaminated residual area to the total area, and obtain the contamination residual degree. Set the cleaning threshold R of the residual contamination threshold , when R≤R threshold , the cleaning is considered completed, and the photovoltaic cleaning robot stops cleaning the area; if R ≥ R threshold , then repeat the path planning and cleaning process for the area.

6. A local control system of a photovoltaic cleaning robot based on vision, characterized in that: The following steps are involved: Position status information acquisition module, motion path planning module, optimal path planning module for the area to be cleaned, pollution degree acquisition and cleaning frequency planning module, cleaning completion judgment module; The position status information acquisition module is used to obtain the surface status information of the photovoltaic panel and the position status information of the photovoltaic cleaning robot by performing cruise shooting by the drone; The motion path planning module: according to the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot, the A-Star algorithm is used to plan the motion path of the photovoltaic cleaning robot; The optimal path planning module for the area to be cleaned: obtains the real-time positioning status information of the photovoltaic cleaning robot and uses the A-Star algorithm to plan the optimal path for the area to be cleaned; The pollution degree acquisition and cleaning frequency planning module: uses an image segmentation algorithm to detect the surface color of the photovoltaic panel, obtains the pollution degree information of the photovoltaic panel surface, and plans the cleaning frequency; The cleaning completion judgment module is used to determine whether to continue cleaning based on the cleaning image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process.

7. The local control system of the vision-based photovoltaic cleaning robot according to claim 6, characterized in that: The method described in the motion path planning module for using the A-Star algorithm to plan the motion path of the photovoltaic cleaning robot based on the surface state information of the photovoltaic panel and the position state information of the photovoltaic cleaning robot is as follows: the surface of the photovoltaic panel is divided into a series of grids, each grid is set to 1m×1m, the photovoltaic panel is divided into a whole grid, the initial position of the photovoltaic cleaning robot is set as the starting point S, and the end point of the photovoltaic panel or the last area to be cleaned is set as the target point G; when searching for the path, the weight is calculated for each possible path node, and the path with the minimum total cost is selected; the weight includes the known cost g(n) from the starting point to the current node and the heuristic estimated cost h(n) from the current node to the target point; the total cost function f(n) of each node n is defined as: f(n)=g(n)+h(n); where g(n) is the actual cost from the starting point to the current node, and h(n) is the heuristic estimate to the target point calculated by Manhattan distance, and the calculation formula is h(n)=|x current -x goal ∣+∣y current -y goal ∣.

8. The local control system of the vision-based photovoltaic cleaning robot according to claim 7, characterized in that: The method for obtaining the real-time positioning status information of the photovoltaic cleaning robot and using the A-Star algorithm to plan the optimal path of the area to be cleaned described in the optimal path planning module of the area to be cleaned is specifically as follows: when the position or map information of the photovoltaic cleaning robot is updated in real time, the A-Star algorithm is used to recalculate the path cost: assuming that node n is updated, g(n′) is recalculated for the neighboring node n′ and f(n′) is updated, that is, f(n′)=g(n′)+h(n′). If it is found that the path cost increases or encounters new obstacles, the A-Star algorithm backtracks forward or backward and continues to search from the known optimal node until a new optimal path is found.

9. The local control system of the vision-based photovoltaic cleaning robot according to claim 8, characterized in that: The method for obtaining the pollution degree information on the surface of the photovoltaic panel and planning the cleaning frequency according to the pollution degree described in the pollution degree acquisition and cleaning frequency planning module is as follows: using the image segmentation algorithm to detect the color of the photovoltaic panel surface, separate the polluted area from the background, and form a binary image I of the polluted area binary , according to the center coordinates of the binary image recognition area (x c ,y c ) and the contaminated area A; if the contaminated area A is greater than the set threshold, a cleaning path is generated; The pollution factor p(i,j) is set to represent the pollution level on the grid (i,j). The pollution factor p(i,j) is classified according to the sensor data. The value range of p(i,j) is set to [0,1], where 0 represents a clean state and 1 represents a heavily polluted state. Define the cleaning frequency f clean(i,j) is the number of times or priority of the photovoltaic cleaning robot to clean the grid, f clean(i,j) The larger the value, the higher the cleaning frequency. clean(i,j) =k×p(i,j), where k is a proportional constant used to adjust the overall reference value of the cleaning frequency; Unify the cleaning frequency and path cost into one optimization objective, and update the total cost function F(n) to take into account both the path cost f(n) and the cleaning frequency f clean(i,j) , that is, F(n)=f(n)+α×f clean(i,j) , to ensure that the cleaning work is carried out normally.

10. The local control system of the vision-based photovoltaic cleaning robot according to claim 9, characterized in that: The method for confirming whether to continue cleaning according to the cleaned image of the photovoltaic panel surface obtained by the photovoltaic cleaning robot during the cleaning process described in the cleaning completion judgment module is as follows: collecting N×M images of the cleaned area, binarizing the image, obtaining the contaminated pixel P, calculating the ratio of the contaminated residual area to the total area, and obtaining the contamination residual degree Set the cleaning threshold R of the residual contamination threshold , when R≤R threshold , the cleaning is considered completed, and the photovoltaic cleaning robot stops cleaning the area; if R ≥ R threshold , then repeat the path planning and cleaning process for the area.

Images