Cleaning robot turning method

Abstract

The invention discloses a cleaning robot turning method which comprises the following steps: shooting a photovoltaic module through a depth camera, collecting environment information, and generating an RGB image and a depth image; the main controller inputs the RGB image and the depth image into the trained model according to the RGB image and the depth image, identifies the frame of the photovoltaic module, and linearly runs along the frame; the main controller judges whether the front of the cleaning robot reaches the end of the photovoltaic module or not; if not, proceeding; and if yes, stopping advancing, and meanwhile, judging whether an obstacle exists in the rotating radius by the depth camera based on the point cloud information, if not, turning by adopting an in-situ rotating turning mode, and if yes, turning by adopting a multi-step turning mode. According to the method, the turning strategy can be flexibly adjusted according to different working environments, and the adaptive capacity and flexibility of the cleaning robot are enhanced.

Description

Technical Field

[0001] The present invention relates to the technical field of photovoltaic cleaning, and particularly to a turning method for a cleaning robot. Background Art

[0002] Distributed photovoltaic power stations are generally built on industrial and commercial rooftops. In industrial areas or suburbs, there is a lot of dust, and the components need to be cleaned and maintained regularly to avoid affecting the power generation efficiency of the power station. At present, manual cleaning has low efficiency and high intensity, and the autonomous cleaning of robots has become an irresistible trend, and the demand for distributed cleaning robots is increasing continuously. Due to obstacles such as cables, maintenance channels, and railings being relatively common around the arrays, they are distributed around the array on industrial and commercial rooftops.

[0003] When the cleaning robot completes the cleaning task of a row of photovoltaic components, it needs to turn at the end of the component array to continue the cleaning work of the next row. Currently, the tracked cleaning robots on the market mainly use two methods to turn:

[0004] 1. Rotating in place: This turning method uses the speed difference between the left and right wheels to achieve the turning of the robot, and can complete the turning action at any angle quickly and efficiently. However, due to the relatively long roller brush, when the robot rotates in place, the roller brush may extend beyond the edge of the photovoltaic component by 0.3 to 0.5 meters, which may cause collisions with surrounding obstacles, thus posing a safety hazard.

[0005] 2. Multi-step turning: Another method is to achieve turning by moving forward and backward multiple times in cooperation with small-angle turning. Although this method can ensure that the roller brush is always inside the photovoltaic component and avoid direct contact with external obstacles, the operation is complex and time-consuming, and it takes 4 to 5 actions to complete a complete turning process. In addition, this method may also leave cleaning dead corners, affecting the overall cleaning effect.

[0006] The above two common turning strategies for cleaning robots have their own advantages and disadvantages, but neither can perfectly solve the contradiction between cleaning efficiency and safety. Therefore, developing a new method that can ensure the safety of the cleaning robot during turning and maintain a high working efficiency has become a problem to be solved. Summary of the Invention

[0007] Therefore, a turning method for a cleaning robot is needed to solve the problems raised in the background art.

[0008] To achieve the above object, the present invention provides a turning method for a cleaning robot, including the following steps:

[0009] Taking pictures of the photovoltaic components through a depth camera, collecting environmental information, and generating an RGB image and a depth image;

[0010] The main controller inputs the RGB image and the depth image into the trained model to identify the frame of the photovoltaic module and drive straight along the frame.

[0011] The main controller determines whether the front of the cleaning robot has reached the end of the photovoltaic module.

[0012] If not, it moves forward.

[0013] If so, it stops moving forward. At the same time, the depth camera determines whether there are obstacles within the rotation radius based on the point cloud information. If not, it turns using the in-place rotation turning mode. If so, it turns using the multi-step turning mode.

[0014] Furthermore, the multi-step turning method includes the following steps:

[0015] Control the cleaning robot to retreat a first preset distance L1.

[0016] Control the cleaning robot to rotate 180° towards the inside of the photovoltaic module, making the tail of the cleaning robot face the end of the array.

[0017] Control the cleaning robot to retreat a second preset distance L2.

[0018] Control the cleaning robot to rotate 90° in the reverse direction.

[0019] Furthermore, the first preset distance L1 is obtained through the following steps:

[0020] Obtain the maximum rotation radius r of the robot and the distance m2 between the sweeping robot and the obstacle, where the maximum rotation radius r of the robot is the distance from the rotation center of the robot to the farthest point of the roller brush.

[0021] The first preset distance L1 is obtained through the following formula:

[0022] L1 = r - m2.

[0023] Furthermore, the second preset distance L2 is obtained through the following steps:

[0024] Obtain the distance m3 between the robot and the bottom edge of the photovoltaic module and the width w of the photovoltaic panel.

[0025] The second preset distance L2 is obtained through the following formula:

[0026] L2 = 0.5 * w - m3.

[0027] Furthermore, the first preset distance L1 is equal to the second preset distance L2.

[0028] Furthermore, in the process of photographing the photovoltaic module by the depth camera, the following steps are also included:

[0029] The depth camera is arranged to incline downwards for shooting the photovoltaic module obliquely downwards.

[0030] Further, when the main controller inputs the RGB image and the depth image into the trained model to identify the border of the photovoltaic module and travels straight along the border, the following steps are further included:

[0031] The motion control module of the cleaning robot calculates the speed information based on the coordinate position of the border and the position of the robot to ensure that the robot always travels straight along the border.

[0032] Further, a roller brush is arranged at the front end of the cleaning robot, with a length ranging from 0.8 m to 1.4 m, and a squeegee is arranged at the rear end.

[0033] The above technical solution has the following beneficial effects:

[0034] Through the monitoring of the surrounding environment, the method of the present invention can select a faster in-situ rotation turning mode in the case of no obstacles, effectively avoiding the collision risk with other objects. When there are obstacles around, the robot will select a more cautious multi-step turning method, so as to shorten the turning time of the robot as much as possible on the premise of ensuring safety. This turning strategy not only ensures the safety of turning, but also reduces the deviation correction amount of the next edge patrol and reduces the cleaning blind area. This method can flexibly adjust the turning strategy according to different working environments, enhancing the adaptability and flexibility of the cleaning robot. Description of the Drawings

[0035] Figure 1 It is one of the flowcharts of the turning method of the cleaning robot in this embodiment;

[0036] Figure 2 It is the second flowchart of the turning method of the cleaning robot in this embodiment;

[0037] Figure 3 It is the structural diagram of the cleaning robot in this embodiment;

[0038] Figure 4 It is the cleaning path diagram of the cleaning robot in this embodiment.

[0039] Description of the Reference Numerals:

[0040] 1. Cleaning robot; 11. Roller brush; 12. Depth camera;

[0041] 2. Rotation center;

[0042] 3. Photovoltaic module;

[0043] 4. Cleaning path;

[0044] 5. Next cleaning path. Detailed implementation mode

[0045] To describe in detail the technical content, structural features, achieved objectives and effects of the technical solution, the following will be described in detail with reference to specific embodiments and in conjunction with the accompanying drawings.

[0046] Please refer to Figure 1 , this embodiment provides a turning method for a cleaning robot, including the following steps:

[0047] The photovoltaic module is photographed by a depth camera to collect environmental information, and an RGB image and a depth image are generated;

[0048] Step S102, the main controller inputs the RGB image and the depth image into the trained model, identifies the border of the photovoltaic module, and travels straight along the border;

[0049] Step S103, the main controller determines whether the end of the photovoltaic module has been reached ahead;

[0050] If not, proceed forward;

[0051] If so, stop moving forward. As shown in step S104, at the same time, the depth camera determines whether there are obstacles within the rotation radius based on the point cloud information. As shown in step S105, if not, a turning mode of rotating in place is adopted for turning, as shown in step S107. If so, a multi-step turning mode is adopted for turning, as shown in step S106, until the turning is completed, as shown in step S108.

[0052] First, the photovoltaic module is scanned by a depth camera installed on the cleaning robot to obtain an RGB image (color image) and a depth image (distance information) in the current environment. The RGB image is a common color image format, where "RGB" represents three color channels: red, green, and blue. In digital image processing, each color channel is encoded as a series of numerical values, which represent the color intensity or brightness of each pixel. The depth image (DepthImage), sometimes also called the distance image or depth map, is an image that records the distance information from each pixel point in the scene to the camera. The main purpose of the depth image is to provide information about the three-dimensional space. Through the depth image, the relative position, distance, and shape of objects can be determined. This is particularly important for robots because they need to understand the three-dimensional structure of the surrounding environment to plan paths, avoid obstacles, or perform other complex operations. In the application scenario of the cleaning robot, the depth image can be used to detect the exact position of the photovoltaic module, identify obstacles, and determine the positioning of the robot itself.

[0053] Subsequently, the main controller inputs this image data into a pre-trained algorithm model to identify the specific position and its border of the photovoltaic module. After the identification is completed, the robot will drive straight along the border of the photovoltaic module.

[0054] When the robot approaches the end of the photovoltaic module, the main controller will detect whether there is enough space ahead to continue moving forward. If it has not reached the end, the robot will continue to move forward; if it has reached the end, it will stop moving forward and further check the environment around the robot. At this time, the depth camera uses the acquired point cloud data to determine whether there are obstacles within the rotation radius of the robot. If no obstacles are detected, the robot will complete the turning action by rotating in place; on the contrary, if there are obstacles, the robot will adopt a more efficient multi-step turning method.

[0055] The above technical solution has the following beneficial effects:

[0056] Through real-time monitoring of the surrounding environment, the method of the present invention can select a faster in-place rotation turning mode in the absence of obstacles, effectively avoiding the risk of collision with other objects. When there are obstacles around, the robot will choose a more cautious multi-step turning method, so as to shorten the turning time of the robot as much as possible on the premise of ensuring safety. This turning strategy not only ensures the safety of turning, but also reduces the amount of deviation correction for the next edge patrol, reducing the cleaning blind area. This method can flexibly adjust the turning strategy according to different working environments, enhancing the adaptability and flexibility of the cleaning robot.

[0057] Please refer to Figure 2 , in this embodiment, the multi-step turning method includes the following steps:

[0058] Control the cleaning robot to retreat a first preset distance L1;

[0059] Control the cleaning robot to rotate 180° towards the inside of the photovoltaic module, so that the tail of the cleaning robot faces the end of the array;

[0060] Control the cleaning robot to retreat a second preset distance L2;

[0061] Control the cleaning robot to rotate 90° in the reverse direction (i.e., the cleaning robot rotates 90° towards the outside of the photovoltaic module).

[0062] When the robot detects that it has reached the end of the photovoltaic module and there are no obstacles within the rotation radius, it will first move backward a first preset distance L1. The selection of this distance is to ensure that the robot has enough space for subsequent rotation actions. Next, the robot will rotate 180° towards the inside of the photovoltaic module, making the tail of the robot point to the end of the array. This action enables the robot to keep the cleaning component away from the photovoltaic module while preparing for the next backward movement. Then, let the robot move backward a second preset distance L2 to ensure that the rotation center of the robot remains unchanged. Finally, the robot rotates 90° in the opposite direction, thus avoiding obstacles and ensuring the convenience of the next edge patrol step and reducing the cleaning blind area.

[0063] In this embodiment, the first preset distance L1 is obtained through the following steps:

[0064] Obtain the length m1 of the roller brush, the maximum rotation radius r of the robot, and the distance m2 between the sweeping robot and the obstacle, where the maximum rotation radius r of the robot is the distance from the rotation center of the robot to the farthest point of the roller brush;

[0065] The first preset distance L is obtained through the following formula 1：

[0066] L1 = r - m2.

[0067] Calculate the backward distance according to the distance of the obstacle and the length of the roller brush to ensure that the roller brush does not touch the obstacle when the cleaning robot rotates.

[0068] In this embodiment, the second preset distance L2 is obtained through the following steps:

[0069] Obtain the distance m3 between the robot and the bottom edge of the photovoltaic module and the width w of the photovoltaic panel;

[0070] The second preset distance L2 is obtained through the following formula:

[0071] L2 = 0.5 * w - m3.

[0072] Calculate the backward distance according to the distance of the obstacle and the length of the roller brush to ensure that the roller brush does not touch the obstacle when the cleaning robot rotates.

[0073] In this embodiment, the first preset distance L1 is equal to the second preset distance L2. The consistent backward distance helps to improve the reliability of the robot and reduces the risk of operation failure or equipment damage caused by improper setting of variables. In some embodiments, the first preset distance L1 is not equal to the second preset distance L2.

[0074] Please refer to Figure 3 , in this embodiment, when shooting the photovoltaic module with a depth camera, the following steps are further included:

[0075] The depth camera 12 is set to incline downward to photograph the photovoltaic module obliquely downward.

[0076] The depth camera 12 can be installed at an appropriate position on the top of the cleaning robot 1 through a bracket, and its installation angle can be adjusted so that it can photograph the photovoltaic module obliquely downward. The selection of the inclination angle should ensure that the depth camera can cover the effective area of the photovoltaic module and provide clear RGB images and depth images.

[0077] Through the depth camera set to incline, the surface of the photovoltaic module and the ground conditions around it can be better covered, information on the photovoltaic module and its surrounding environment can be photographed, and RGB images and depth images can be generated. Based on the data obtained from the optimized perspective, the main controller can more accurately identify the position, boundary, and other related features of the photovoltaic module, thereby better planning the driving path and turning mode of the robot.

[0078] In this embodiment, when the main controller inputs the RGB image and the depth image into the trained model to identify the border of the photovoltaic module and drives straight along the border, the following steps are further included:

[0079] The motion control module of the cleaning robot calculates the speed information based on the coordinate position of the border and the position of the robot to ensure that the robot always drives straight along the border.

[0080] The main controller accurately obtains the coordinate position of the border of the photovoltaic module through visual recognition technology. Based on the calculated speed information, the motion control module dynamically adjusts the driving parameters of the robot so that it can always stay on the correct driving path in real time, avoiding the situation of deviating from the track and improving the driving accuracy.

[0081] Please refer to Figure 1 , in this embodiment, before photographing the photovoltaic module through the depth camera, the following steps are further included:

[0082] Step S101, start the cleaning robot.

[0083] Please refer to Figure 3 , in this embodiment, a roller brush 11 with a length of 0.8 meters to 1.4 meters is provided at the front end of the cleaning robot 1, and a squeegee is provided at the rear end. A roller brush with a length of 0.8 meters to 1.4 meters is installed at the front end of the cleaning robot. The selection of this length range is to balance the cleaning efficiency and the mobility of the robot, ensure that the roller brush can cover a sufficient cleaning area, and at the same time not affect the flexibility of the robot. A squeegee is installed at the rear end of the robot, and its function is to further clean the residual dirt brought up by the roller brush during the cleaning process to ensure that the surface of the photovoltaic module is clean and tidy.

[0084] Please refer to Figure 4, the cleaning robot 1 walks on the photovoltaic module 3. The cleaning robot turning method can not only ensure the safety of the cleaning robot 1, but also improve the cleaning efficiency. There is no need to perform multi-step turning actions at each turning point. The turning method is intelligently selected, and the multi-step turning strategy can ensure the invariance of the rotation center 2. During the process from the cleaning path 4 to the next cleaning path 5, the complexity of the next edge-following trajectory is reduced, and the cleaning blind area is decreased.

[0085] It should be noted that in this article, relational terms such as first and second are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any actual relationship or order between these entities or operations. Moreover, the term "comprising", "including" or any other variant thereof is intended to cover non-exclusive inclusion, so that a process, method, article or terminal device comprising a series of elements not only includes those elements, but also includes other elements not expressly listed, or also includes elements inherent to such process, method, article or terminal device. Without further limitation, the elements defined by the statement "comprising..." or "including..." do not exclude the existence of additional elements in the process, method, article or terminal device comprising the elements. In addition, in this article, "greater than", "less than", "exceeding", etc. are understood not to include the present number; "above", "below", "within", etc. are understood to include the present number.

[0086] Although the above embodiments have been described, those skilled in the art can make additional changes and modifications once they know the basic creative concept. Therefore, the above are only embodiments of the present invention, and do not limit the patent protection scope of the present invention. Any equivalent structure or equivalent process transformation made by using the specification and drawings of the present invention, or directly or indirectly applied in other related technical fields, shall be equally included in the patent protection scope of the present invention.

Claims

1. A turning method for a cleaning robot, characterized in that: The steps include: Use a depth camera to photograph photovoltaic modules, collect environmental information, and generate RGB images and depth maps; The main controller inputs the RGB image and the depth image into the trained model, identifies the border of the PV panel, and drives in a straight line along the border; The main controller determines whether the front of the cleaning robot has reached the end of the photovoltaic panel; If not, proceed forward; If so, the vehicle stops moving forward. At the same time, the depth camera determines whether there is an obstacle within the rotation radius based on the point cloud information. If not, the vehicle turns in place using a turning mode. If so, the vehicle turns in a multi-step turning mode.

2. The turning method according to claim 1, characterized in that: The multi-step turning method includes the following steps: Control the cleaning robot to retreat a first preset distance L1; Control the cleaning robot to rotate 180° toward the inside of the photovoltaic module, so that the tail of the cleaning robot faces the end of the array; Control the cleaning robot to retreat a second preset distance L2; Control the cleaning robot to rotate 90° in the opposite direction.

3. The turning method according to claim 2, characterized in that: The first preset distance L1 is obtained by the following steps: Obtain the maximum rotation radius r of the robot and the distance m2 between the sweeping robot and the obstacle, where the maximum rotation radius r of the robot is the distance from the rotation center of the robot to the farthest point of the roller brush; The first preset distance L is obtained by the following formula 1： L1=r-m2.

4. The turning method according to claim 2, characterized in that: The second preset distance L2 is obtained by the following steps: Get the current distance m3 between the robot and the bottom edge of the photovoltaic module and the width w of the photovoltaic panel; The second preset distance L is obtained by the following formula 2： L2 = 0.5*w-m3.

5. The turning method according to claim 2, characterized in that: The first preset distance L1 is equal to the second preset distance L2.

6. The turning method according to claim 1, characterized in that: In the method of photographing the photovoltaic assembly by using the depth camera, the following steps are also included: The depth camera is tilted downward to photograph the photovoltaic components obliquely downward.

7. The turning method according to claim 1, characterized in that: The main controller inputs the RGB image and the depth image into the trained model, identifies the border of the photovoltaic module, and drives in a straight line along the border, and also includes the following steps: The motion control module of the cleaning robot calculates the speed information based on the coordinate position of the frame and the position of the robot to ensure that the robot always moves in a straight line along the frame.

8. The turning method according to claim 1, characterized in that: The front end of the cleaning robot is provided with a roller brush with a length of 0.8 meters to 1.4 meters, and the rear end is provided with a scraper bar.

Images