Control method of underwater cleaning robot

Abstract

The invention discloses a control method for an underwater cleaning robot, and the scheme can comprise the steps: S1, collecting an environment image of a to-be-cleaned region through a global camera, and carrying out the distortion correction and perspective transformation preprocessing, and obtaining a preprocessed image; s2, generating a static environment map according to the preprocessed image, wherein the static environment map comprises obstacles and dirt sub-regions; s3, adopting a multi-scale path planning strategy: firstly, dividing a map by using a first-scale grid, determining a cleaning end point, and planning an access sequence based on an A * algorithm; setting the first dirt sub-region as a target, and planning a path to a boundary turning point of the first dirt sub-region; dividing a map by using a second scale grid, and planning a path from a turning point to a cleaning end point; and updating the operation starting point and removing the processed area, and circulating until the operation path sequence is generated. And S4, the underwater cleaning robot completes cleaning operation on all the dirt sub-areas according to the planned path.

Description

Technical Field

[0001] The present invention relates to the technical field of underwater robot control, and more particularly, to a control method for an underwater cleaning robot. Background Art

[0002] With the country's promotion of the "Healthy China" strategy and the construction of "Smart Cities", the water quality safety and cleaning problems of pools have attracted increasing attention. Especially in the context of national fitness and public health, how to clean swimming pools efficiently and accurately has become an urgent problem to be solved. Traditional manual cleaning and automatic sewage suction equipment have problems such as low efficiency and incomplete cleaning in large areas and complex pool shapes, and cannot perform precise cleaning according to the actual distribution of dirt (dirt in the water environment is prone to change its position with the flow of water), easily resulting in "cleaning blind spots".

[0003] As a classic path planning algorithm, the A* algorithm is widely used in cleaning robots. However, in complex environments, the traditional A* algorithm has limitations such as too coarse grid division resulting in incomplete cleaning and too fine grid division resulting in large computational amounts. In addition, existing methods usually ignore the matching problem between the actual coverage area of the robot and the cleaning area, resulting in insufficient cleaning accuracy.

[0004] Therefore, there is an urgent need for a new path planning method that can solve the "cleaning blind spot" problem while improving computational efficiency, ensuring the comprehensiveness and accuracy of pool cleaning. Summary of the Invention

[0005] The present invention provides a control method for an underwater cleaning robot to overcome at least one technical problem existing in the prior art.

[0006] The present invention provides a control method for an underwater cleaning robot, including:

[0007] S1. Collect the environmental image of the area to be cleaned through a global camera, and perform preprocessing operations on the environmental image to obtain a preprocessed environmental image; wherein, the preprocessing operations include distortion correction and perspective transformation;

[0008] S2. Generate a static environmental map of the area to be cleaned based on the preprocessed environmental image, and the map elements of the static environmental map include obstacle sub-regions and dirt sub-regions;

[0009] S3. Adopt a multi-scale path planning strategy to hierarchically generate an operation path sequence from the operation starting point of the underwater cleaning robot to the cleaning end point in each dirt sub-region, specifically including:

[0010] S31. Divide the static environment map using grids of the first scale to generate a grid map of the first scale; determine the cleaning endpoints within each dirt sub-region to obtain multiple cleaning endpoints; based on the positions of these multiple cleaning endpoints and the position of the operation starting point of the underwater cleaning robot, plan the access sequence of the underwater cleaning robot to each dirt sub-region in the grid map of the first scale based on the A* algorithm.

[0011] S32. Take the first dirt sub-region in the access sequence as the target dirt sub-region, and then plan the travel path of the underwater cleaning robot from the operation starting point to the turning point on the boundary of the target dirt sub-region in the grid map of the first scale.

[0012] S33. Divide the static environment map using grids of the second scale to generate a grid map of the second scale, where the grid size of the grids of the second scale is smaller than the grid size of the grids of the first scale; plan the cleaning path from the turning point to the cleaning endpoint within the target dirt sub-region in the grid map of the second scale.

[0013] S34. Take the cleaning endpoint within the target dirt sub-region as the updated operation starting point of the cleaning robot, and then remove the target dirt sub-region from the access sequence to obtain the updated access sequence.

[0014] S35. Repeat steps S32 to S34 until the operation path sequence is obtained.

[0015] S4. Complete the cleaning operation of all dirt sub-regions in the area to be cleaned by the underwater cleaning robot according to the paths planned in the operation path sequence.

[0016] In an alternative embodiment, the global camera is set at the central position above the area to be cleaned, and the acquisition range of the global camera covers the entire area to be cleaned; wherein, the area to be cleaned is a pool or fishpond to be subjected to a cleaning operation.

[0017] In an alternative embodiment, the preprocessing operation on the environmental image to obtain the preprocessed environmental image specifically includes:

[0018] Perform distortion correction on the environmental image using the Zhang-Zhengyou calibration method to eliminate the distortion generated by the lens of the global camera when acquiring the environmental image, and obtain the environmental image after distortion correction processing.

[0019] Perform perspective transformation on the environment image after distortion correction according to the coordinates of the four corner points of the area to be cleaned pre-calibrated, so as to convert the tilted image from the perspective of the global camera into a top-down plan view from a bird's-eye view.

[0020] In an alternative embodiment, the positioning of the dirt sub-region in step S2 adopts a target detection algorithm to identify the dirt distribution by analyzing color, texture or shape features in the image.

[0021] In an alternative embodiment, the side length of the grid of the first scale is 3 to 4 times the side length of the grid of the second scale, and the area of the grid of the second scale is smaller than the coverage area of the underwater cleaning robot body.

[0022] In an alternative embodiment, an AprilTag code is installed on the top of the underwater cleaning robot; during the movement of the underwater cleaning robot, the global camera continuously identifies the AprilTag code to obtain the current position of the underwater cleaning robot, and determines the actual position of the underwater cleaning robot in the area to be cleaned by combining perspective transformation and coordinate mapping, and then adjusts the movement of the underwater cleaning robot by combining the PID control algorithm to make it move along the planned path to the target area.

[0023] In an alternative embodiment, the PID control algorithm dynamically adjusts parameters according to the distance between the current position of the underwater cleaning robot and the target point; wherein, when the distance is large, the proportional coefficient Kp is increased to increase the moving speed of the underwater cleaning robot, and when the distance is small, the differential coefficient Kd is increased to reduce the position deviation of the underwater cleaning robot.

[0024] In an alternative embodiment, the method further includes re-planning the cleaning path and performing subsequent cleaning operations if it is monitored that there is a dirt sub-region in the area to be cleaned that has not been cleaned or there is a cleaning blind spot.

[0025] At least one embodiment of this specification can achieve the following beneficial effects:

[0026] The technical solution of this application first uses a first-scale grid to divide the static environment map, and combines the A* algorithm to plan the access sequence of the underwater cleaning robot to each dirt sub-region. The A* algorithm can improve the search efficiency while ensuring to find the optimal path, enabling the robot to macroscopically determine the reasonable order to each dirt sub-region, reducing unnecessary movements, and improving the overall cleaning efficiency. At the same time, the second-scale grid size is smaller than the first scale, and a path from the turning point to the cleaning end point is planned in the grid map generated by it. The smaller grid size can more precisely consider the environmental details within the target dirt sub-region, such as tiny obstacles, etc., enabling the robot to avoid obstacles and accurately reach the cleaning end point, improving the accuracy of the cleaning path and the quality of the cleaning operation. In this way, both the macroscopic route is quickly planned through the first-scale grid, and the local fine planning is carried out using the second-scale grid, taking into account both the planning efficiency and accuracy, and effectively improving the cleaning efficiency of the robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0028] Figure 1 FIG. is a schematic diagram of the path generated when planning the cleaning path based on a large grid in the prior art;

[0029] Figure 2 FIG. is a schematic diagram of the path generated when planning the cleaning path based on a small grid in the prior art;

[0030] Figure 3 FIG. is a schematic diagram of the path generated when planning the cleaning path in the technical solution of this application based on a multi-scale (combination of large and small grids) planning method;

[0031] Figure 4 FIG. is a schematic flow chart of a control method for an underwater cleaning robot provided by an embodiment of this specification. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0032] To make the objectives, technical solutions, and advantages of one or more embodiments of this specification clearer, the following will clearly and completely describe the technical solutions of one or more embodiments of this specification in conjunction with the specific embodiments of this specification and the corresponding drawings. Obviously, the described embodiments are only some, rather than all, of the embodiments of this specification. Based on the embodiments in this specification, all other embodiments obtained by those of ordinary skill in the art without creative efforts fall within the scope protected by one or more embodiments of this specification.

[0033] It should be understood that although terms such as first, second, and third may be used in this application document to describe various information, such information should not be limited to these terms. These terms are only used to distinguish information of the same type from each other.

[0034] As Figure 1 shown, Figure 1 Figure (a) is a schematic diagram of a path generated when planning a cleaning path based on a large grid in the prior art; Figure 2 Figure (b) is a schematic diagram of a path generated when planning a cleaning path based on a small grid in the prior art. Comparing the grid division methods in Figure 1 and Figure 2 it can be known that Figure 1 the large grid in Figure (a) has a larger size. When using the large grid division method in Figure 1 to plan the cleaning path of an underwater cleaning robot in the area to be cleaned, it is difficult for this grid division method to accurately represent details such as small obstacles and narrow channels in the environment. In practical applications such as robot navigation, it may cause the robot to be unable to perceive and avoid small obstacles in advance, affecting the safety and accuracy of travel. And Figure 2 the small grids in Figure (b) are dense and numerous, which will greatly increase the computational complexity and data storage requirements during path planning. Moreover, small grids focus on local details and it is difficult to quickly grasp the overall environmental layout when dealing with a larger-scale environment, which is not conducive to global path planning and decision-making. From a global perspective, it may lose the optimal travel direction.

[0035] To solve the defects in the prior art, the following embodiments are given in this solution:

[0036] As Figure 3 shown, Figure 3 Figure (c) is a schematic diagram of a path generated when planning a cleaning path based on a multi-scale (combination of large and small grids) planning method in the technical solution of this application. The grid division in this figure adopts a multi-scale (combination of large and small grids) method, and this division method can take into account the efficiency of global planning and the accuracy of local planning. For this, the specific explanation is as follows: Figure 3 The larger squares outlined by the thick red lines in Figure (c) are large grids. The side length of the large grid is usually 3 to 4 times that of the small grid, which is used for macroscopic division of the overall environment, can quickly process large-scale data, determine the general travel direction and path framework, and reduce the computational complexity of global path planning. For example, in the application scenario of the underwater cleaning robot in the embodiment of this application, it can quickly plan the general route for the robot to go to each dirt sub-area, improving the planning efficiency. Figure 3Among them, the smaller squares formed by the black thin lines are small grids. When the robot approaches a specific target area (such as a dirt sub-area), the small grids are enabled to divide the area more finely. The small grids can present more environmental details, such as tiny obstacles, etc., providing detailed information for the robot to plan a precise path within the target area and ensuring that the robot can reach the cleaning position more safely and accurately. When planning the path, first perform global path planning based on the large grid to determine the approximate path and access order of the robot to each target area; when approaching a specific target area, switch to the small grid for fine path planning to adapt to the complexity of the local environment. This way of combining large and small grids balances the efficiency and accuracy of the planning and enables the robot to complete tasks more efficiently in a complex environment.

[0037] Based on the grid division method of the embodiments of the present application described above, the technical solution of the present application will be introduced below. As Figure 4 shown, Figure 4 Figure 4 FIG. is a schematic flowchart of a control method for an underwater cleaning robot provided by an embodiment of the present specification. From a program perspective, the execution subject of the process can be an application program installed in the corresponding hardware device. The method includes the following steps:

[0038] Step S1: Collect an environmental image of the area to be cleaned through a global camera, and perform preprocessing operations on the environmental image to obtain a preprocessed environmental image; wherein, the preprocessing operations include distortion correction and perspective transformation;

[0039] Step S2: Generate a static environmental map of the area to be cleaned based on the preprocessed environmental image. The map elements of the static environmental map include obstacle sub-areas and dirt sub-areas;

[0040] Step S3: Adopt a multi-scale path planning strategy to hierarchically generate an operation path sequence from the operation starting point of the underwater cleaning robot to the cleaning end point within each dirt sub-area, specifically including:

[0041] Step S31: Use grids of the first scale to divide the static environmental map to generate a grid map of the first scale; determine the cleaning end points within each dirt sub-area to obtain a plurality of cleaning end points; based on the positions of the plurality of cleaning end points and the position of the operation starting point of the underwater cleaning robot, plan the access sequence of the underwater cleaning robot to each dirt sub-area in the grid map of the first scale;

[0042] In robot path planning, it is usually necessary to discretize the actual environmental map. In this step, according to the selected first scale, the static environmental map is divided into grid cells of the same size. Each grid cell can be identified by coordinates and corresponding attributes are assigned to it, such as whether it is occupied by an obstacle, whether it is a passable area, etc. In this way, a grid map at the first scale is generated, which is the basis for subsequent path planning. For each dirt sub-region, a specific point within the region can be selected as the cleaning end point. For example, the center point, the farthest point, etc. of the dirt sub-region can be selected. The principle of selection is to ensure that the robot can cover the entire dirt sub-region as much as possible after reaching this point. The starting point of the underwater cleaning robot's operation is used as the starting point, and the cleaning end points within each dirt sub-region are used as the target points. Using the A* algorithm, calculate the path cost from the starting point to each cleaning end point in the grid map at the first scale. By comparing these path costs and comprehensively considering factors such as path length and whether to avoid obstacles, determine the access order of the robot to each dirt sub-region, that is, the access sequence.

[0043] Step S32: Take the first dirt sub-region in the access sequence as the target dirt sub-region, and then plan the travel path of the underwater cleaning robot from the operation starting point to the turning point of the target dirt sub-region in the grid map at the first scale, where the turning point is located at the boundary of the target dirt sub-region.

[0044] In this step, among the access sequences of the underwater cleaning robot to each dirt sub-region planned by the A* algorithm, select the first dirt sub-region as the target dirt sub-region. This access sequence is obtained by applying the A* algorithm based on the positions of the cleaning end points in multiple dirt sub-regions and the position of the operation starting point, and it is an optimized access order. The grid map at the first scale is obtained by initially dividing the static environmental map into grids. Its grid size is relatively large. This larger-scale division helps to plan the travel path of the robot from a macroscopic level and improve the efficiency of path planning. Planning the path on this map can quickly plan the general travel route and avoid falling into local optimal solutions in a complex environment. For example, when the area to be cleaned is large and the dirt is distributed dispersedly, planning in the grid map at the first scale can quickly determine the general direction for the robot to go to each dirt sub-region.

[0045] Meanwhile, in this step, taking the operation starting point of the underwater cleaning robot as the starting point, plan a travel path to the turning points of the boundary of the target dirt sub-region in the first-scale grid map. Here, the turning points are located at the boundary of the target dirt sub-region and are the nodes for transitioning from global path planning (first-scale grid map planning) to local fine path planning (second-scale grid map planning). By planning the path to the turning points, the robot can quickly approach the target dirt sub-region first, and then perform subsequent more refined path planning to enter the sub-region for cleaning operations. When planning the travel path in this step, it is necessary to comprehensively consider the obstacle sub-regions in the map to avoid the robot colliding with obstacles during travel and ensure the safety and feasibility of the path.

[0046] Step S33: Use the grids of the second scale to divide the static environment map to generate a second-scale grid map. The grid size of the grids of the second scale is smaller than the grid size of the grids of the first scale; plan a cleaning path from the turning point to the cleaning end point within the target dirt sub-region in the second-scale grid map.

[0047] In this step, the grid size of the second scale is smaller than that of the first scale. Usually, the side length of the grids of the first scale can be 3 to 4 times the side length of the grids of the second scale. Such a size design can provide map information with different precisions for path planning at different stages. The larger-scale first grid is used for macroscopic planning to determine the general route for the robot to go to each dirt sub-region; the smaller-scale second grid is used for microscopic planning to perform more refined path planning when the robot approaches the target dirt sub-region.

[0048] Use the grids of the second scale to re-divide the static environment map to generate a second-scale grid map. In this process, the entire area to be cleaned is divided into denser and smaller grid cells. Each grid cell represents a smaller actual area, which makes the map present environmental details more precisely, including the positions and shapes of obstacles and the specific conditions within the target dirt sub-region, etc. For example, in the first-scale grid map, one grid may cover a large area, and some small obstacles or dirt distribution details are ignored; while in the second-scale grid map, these details can be shown more clearly.

[0049] Meanwhile, in this step, the turning points planned in the first-scale grid map are used as the starting points, and the cleaning end points within the target dirt sub-region are used as the end points. The turning points can be located at the boundaries of the target dirt sub-region and are key positions for the cleaning robot to transition from the macroscopic path to the microscopic cleaning path. The cleaning end points are the target positions where the robot finally completes the cleaning task. By performing path planning in the second-scale grid map, more detailed environmental information within the target dirt sub-region can be fully considered. Since the grid size is small, the robot can avoid obstacles more precisely and select a better route to reach the cleaning end point. For example, the path planned in the first-scale grid map may only roughly point to the target dirt sub-region, but in the second-scale grid map, an accurate path around the obstacles can be planned according to the specific distribution of the obstacles within the sub-region, ensuring that the robot can reach the cleaning end point safely and efficiently to complete the cleaning operation.

[0050] Step S34: Use the cleaning end point within the target dirt sub-region as the updated operation starting point of the cleaning robot, and then remove the target dirt sub-region from the access sequence to obtain the updated access sequence.

[0051] In this step, after completing the cleaning path planning for a target dirt sub-region and reaching its cleaning end point, the cleaning end point is set as the updated operation starting point of the robot. This is because after the robot completes the cleaning of the current dirt sub-region, starting from this position to the next target area is the most reasonable choice, which can reduce unnecessary movement paths and improve cleaning efficiency. For example, assume the robot is cleaning in a large pool, and the current dirt sub-region is at a corner of the pool. After completing the cleaning of this region, using this cleaning end point as the new starting point, it can directly move to the next target area.

[0052] Meanwhile, in this solution, the processed target dirt sub-regions need to be removed from the access sequence. The access sequence records the order of the dirt sub-regions that the robot needs to visit in sequence. When a sub-region is cleaned, it no longer needs to be visited. Removing this sub-region can ensure that the access sequence always accurately reflects the remaining areas that need to be cleaned. For example, initially, there are 5 dirt sub-regions in the access sequence. After completing the cleaning of the first sub-region and removing it, the access sequence becomes 4, and the robot can clearly know which area to go to for cleaning next.

[0053] Through the above operations, an updated access sequence is obtained. This new sequence provides accurate information for the next path planning. The robot will continue to execute the subsequent path planning and cleaning operations based on this updated sequence. In the subsequent steps, the robot will start from the updated operation starting point, plan the travel path to the turning point of the boundary of the next target dirt sub-region in the first-scale grid map, and then plan the path to the cleaning end point within this sub-region in the second-scale grid map. This process repeats until the access sequence is empty, completing the cleaning of all dirt sub-regions.

[0054] Step S35: Repeat Step S32 to Step S34 until the operation path sequence is obtained.

[0055] In Step S34 and S35, after completing the cleaning path planning of a target dirt sub-region and reaching its cleaning end point, use this cleaning end point as the updated operation starting point of the robot, remove the processed target dirt sub-region from the access sequence, and obtain the updated access sequence. Then repeat the above process of planning the path to the turning point of the next target dirt sub-region in the first-scale grid map and planning the path to the cleaning end point in the second-scale grid map until the access sequence is empty, generating a complete operation path sequence to ensure that the robot completes the cleaning operation of all dirt sub-regions.

[0056] Step S4: Complete the cleaning operation of all dirt sub-regions in the area to be cleaned by the underwater cleaning robot according to the path planned in the operation path sequence.

[0057] The technical solution of this application first uses the first-scale grid to divide the static environment map and combines the A* algorithm to plan the access sequence of the underwater cleaning robot for each dirt sub-region. The A* algorithm can improve the search efficiency while ensuring finding the optimal path, enabling the robot to macroscopically determine the reasonable order to each dirt sub-region, reducing unnecessary movements, and improving the overall cleaning efficiency. At the same time, the second-scale grid size is smaller than the first scale, and the path from the turning point to the cleaning end point is planned in the grid map generated by it. The smaller grid size can more precisely consider the environmental details within the target dirt sub-region, such as tiny obstacles, etc., allowing the robot to avoid obstacles and accurately reach the cleaning end point, improving the accuracy of the cleaning path and the quality of the cleaning operation. In this way, both the macroscopic route is quickly planned through the first-scale grid and the local fine planning is carried out using the second-scale grid, taking into account both the planning efficiency and accuracy, and effectively improving the cleaning efficiency of the robot.

[0058] Based on Figure 1 the method, the embodiments of this specification also provide some specific implementation schemes of this method, which will be described below.

[0059] In an alternative embodiment of the technical solution, the global camera is disposed at the central position above the area to be cleaned, and the acquisition range of the global camera covers the entire area of the area to be cleaned; wherein, the area to be cleaned is a pool or a fishpond to be subjected to a cleaning operation.

[0060] In the technical solution of this embodiment, considering that the acquisition range of the global camera needs to cover the entire area of the area to be cleaned, and setting it at the central position above can achieve this goal. From the perspective of the spatial position relationship, the central position enables the camera to capture information about each corner of the pool or fishpond with a relatively uniform view in the horizontal direction, avoiding acquisition dead angles. When cleaning the pool, whether it is the obstacles or the dirt distribution in the pool edge or the central area, they can be clearly captured by the camera. In the vertical direction, this position can reduce image distortion and occlusion problems caused by the tilt of the viewing angle. If the camera position deviates from the center, it may cause image distortion or partial information loss in some areas due to the occlusion of the pool wall or excessive tilt of the viewing angle, affecting the subsequent accurate judgment of the environmental information. Therefore, in the technical solution of this embodiment, the global camera is disposed at the central position above the area to be cleaned.

[0061] In an alternative embodiment of the technical solution, the preprocessing operation on the environmental image to obtain the preprocessed environmental image specifically includes:

[0062] Performing distortion correction on the environmental image using the Zhang Zhengyou calibration method to eliminate the distortion generated by the lens of the global camera when acquiring the environmental image, and obtaining the environmentally distorted image after distortion correction;

[0063] Performing perspective transformation on the environmentally distorted image after distortion correction according to the coordinates of the four corner points of the area to be cleaned pre-calibrated, so as to convert the tilted image from the perspective of the global camera into a top-down plan view from the bird's-eye view.

[0064] In an alternative embodiment of the technical solution, the positioning of the dirt sub-region in step S2 adopts a target detection algorithm to identify the dirt distribution by analyzing the color, texture or shape features in the image.

[0065] In this step, a target detection algorithm is used to locate the dirt sub-region from the pre-processed environmental images collected by the global camera. This algorithm analyzes and judges each pixel or pixel region in the image, and identifies the dirt distribution by analyzing the color, texture, or shape features in the image. In terms of color, dirt usually has a different color from the surrounding environment. For example, the dirt in a pool may appear dark, with an obvious color difference from the pool wall and the water body. This target detection algorithm can filter out the areas that may be dirt by analyzing the color information. In terms of texture, the surface texture of dirt is different from that of normal areas. For example, moss-like dirt has a unique texture structure, and this target detection algorithm can identify these texture features to locate the dirt. Shape features can also be used as a judgment basis. For example, some dirt may appear in a clump or strip shape, and this target algorithm can further determine the location and distribution range of the dirt based on these shape differences.

[0066] In an optional embodiment technical solution, the side length of the grid of the first scale is 3 to 4 times the side length of the grid of the second scale, and the area of the grid of the second scale is smaller than the coverage area of the underwater cleaning robot body.

[0067] In the technical solution of this embodiment, the grid of the first scale is suitable for global path planning. It can quickly process large-scale data. When using a large-scale grid for path search in the global environment, it can quickly determine the skeleton of the cleaning path, avoid major obstacles at the same time, reduce the search space and calculation time. The efficient processing ability of the large grid provides a basic framework for the overall path planning. The grid of the second size is used for fine planning. When the robot approaches the dirt center area, the system automatically switches to a small-scale grid for more refined path planning, making the path more conform to the dirt distribution and improving the cleaning coverage rate. This multiple relationship ensures both the efficiency of the large grid in global planning and the accuracy of the small grid in local fine planning.

[0068] Since the small grid is used for fine planning in the dirt-intensive area or complex area, the smaller grid area allows the robot to more accurately cover the dirt area during detailed planning, avoiding omission. If the small grid area is too large, it may cause the robot to miss some dirt points during the cleaning process, affecting the cleaning quality. The setting of this relationship ensures the comprehensiveness and efficiency of the robot in dealing with dirt, and helps to improve the cleaning coverage rate.

[0069] In an alternative embodiment of the technical solution, an AprilTag code is installed on the top of the underwater cleaning robot; during the movement of the underwater cleaning robot, the global camera continuously identifies the AprilTag code to obtain the current position of the underwater cleaning robot, and determines the actual position of the underwater cleaning robot in the area to be cleaned by combining perspective transformation and coordinate mapping. Then, the movement of the underwater cleaning robot is adjusted by combining the PID control algorithm to make it move along the planned path to the target area. The PID control algorithm dynamically adjusts the parameters according to the distance between the current position of the underwater cleaning robot and the target point; among them, when the distance is large, the proportional coefficient Kp is increased to improve the moving speed of the underwater cleaning robot, and when the distance is small, the differential coefficient Kd is increased to reduce the position deviation of the underwater cleaning robot.

[0070] In the technical solution of this embodiment, the AprilTag code is installed on the top of the robot and serves as a unique identifier of the robot in the global environment. During the movement of the robot, the global camera can continuously identify the AprilTag code. According to the characteristics of the AprilTag code, by identifying the position of its center point, the current position (X, Y coordinates) of the underwater cleaning robot can be obtained. This provides a reliable position reference point for the robot in the complex underwater environment. Since there is a certain perspective distortion in the image obtained by the global camera, in order to obtain the accurate actual position of the robot in the area to be cleaned (such as a pool or a fish pond), perspective transformation is required. According to the pre-calibrated relevant parameters of the area to be cleaned, the image obtained by identifying the AprilTag code is subjected to perspective transformation to convert the image into a more intuitive perspective, and then combined with coordinate mapping to correspond the coordinates in the image with the coordinates in the actual physical space, so as to accurately determine the actual position of the robot in the area to be cleaned.

[0071] After determining the actual position of the robot, in order to enable the robot to move smoothly along the planned path to the target area, the technical solution of this application adopts the PID control algorithm. During the cleaning process, the center point of the dirt area is set as the target point (desired point) of the robot, and the center point of the AprilTag code represents the current position of the robot. By comparing the deviation between the current position and the target point, when the robot is far from the dirt area, the proportional coefficient Kp is appropriately increased, which can improve the moving speed of the robot and make the robot approach the target area faster; when the robot approaches the dirt area, Kp is appropriately reduced, and at the same time the differential coefficient Kd is increased to reduce the position deviation of the robot and avoid overshoot when the robot approaches the target, ensuring that the robot can accurately reach the target area for cleaning operations.

[0072] In an alternative embodiment of the technical solution, the method further includes, if it is detected that there is a dirt sub-region in the area to be cleaned where no cleaning operation has been performed or there is a cleaning blind spot, re-planning the cleaning path and performing subsequent cleaning operations.

[0073] In the technical solution of this application, if it is detected that there is a dirt sub-region in the area to be cleaned where no cleaning operation has been performed or there is a cleaning blind spot, the operation path sequence from the operation starting point of the underwater cleaning robot to the cleaning end point in each dirt sub-region can be re-generated hierarchically according to the multi-scale path planning strategy described above, and then subsequent specific cleaning operations can be performed. In this way, by promptly discovering and dealing with the uncleaned area, it can ensure that every corner of the pool or fishpond is effectively cleaned, reduce the pollution of the water quality by dirt, and ensure the hygiene of the water body. At the same time, this feedback-based cleaning path adjustment mechanism forms a closed-loop control system for the cleaning process of the underwater cleaning robot. It can continuously optimize the cleaning path, improve the cleaning efficiency, and adapt to different cleaning environments and dirt distributions.

[0074] In the technical solution of this application, there are grids of the first scale and the second scale. The side length of the grid of the first scale is 3 to 4 times the side length of the grid of the second scale. The grid of the first scale is used for global path planning, initially dividing the static environment map to generate a grid map of the first scale. Its grid size is relatively large, which can quickly process large-scale data, search for paths in the global environment, quickly determine the general framework of the cleaning path, and avoid major obstacles at the same time, reducing the search space and calculation time. For example, in a large pool, the large-scale grid can quickly plan the general direction for the robot to reach each dirt sub-region. The grid of the second scale is used for fine planning. When the robot approaches the target dirt sub-region, the static environment map is re-divided using the grid of the second scale to generate a grid map of the second scale. Its grid size is small, which can more precisely present the environmental details, including the positions, shapes of small obstacles, and the specific conditions within the target dirt sub-region, etc. For example, small obstacles that may be ignored in the grid map of the first scale can be clearly shown in the grid map of the second scale, providing a basis for the robot to plan a more accurate path.

[0075] In the path planning stage, an access sequence and a preliminary path are planned based on the first-scale grid. Specifically, first, the cleaning end points within each dirt sub-region are determined. Based on the positions of these cleaning end points and the operation start point, the A* algorithm is used to plan the access sequence of the underwater cleaning robot to each dirt sub-region in the grid map of the first scale. The A* algorithm can improve the search efficiency while ensuring finding the optimal path, enabling the robot to macroscopically determine the reasonable order to reach each dirt sub-region and reducing unnecessary movement. After determining the access sequence, the first dirt sub-region is taken as the target dirt sub-region, and the travel path from the operation start point to the turning point on the boundary of the target dirt sub-region is planned in the grid map of the first scale. The turning point is located on the boundary of the target dirt sub-region and is a key node for transitioning from global path planning to local fine path planning. By planning the path to the turning point, the robot can quickly approach the target dirt sub-region first.

[0076] Then, a fine cleaning path is planned based on the second-scale grid. Specifically, taking the turning point planned in the grid map of the first scale as the starting point and the cleaning end point within the target dirt sub-region as the end point, the cleaning path is planned in the grid map of the second scale. Since the second-scale grid has a small size and can fully consider the more detailed environmental information within the target dirt sub-region, the robot can more precisely avoid obstacles and select a better route to reach the cleaning end point. For example, the path planned in the grid map of the first scale may only generally point to the target dirt sub-region, but in the grid map of the second scale, an accurate path bypassing the obstacles can be planned according to the specific distribution of the obstacles within the sub-region, ensuring that the robot safely and efficiently reaches the cleaning end point and completes the cleaning operation.

[0077] After completing the cleaning path planning of a target dirt sub-region and reaching its cleaning end point, the cleaning end point is taken as the updated operation start point of the robot, and the processed target dirt sub-region is removed from the access sequence to obtain the updated access sequence. Then, repeat the above process of planning the path to the turning point of the next target dirt sub-region in the grid map of the first scale and planning the path to the cleaning end point in the grid map of the second scale until the access sequence is empty, generating a complete operation path sequence to ensure that the robot completes the cleaning operation of all dirt sub-regions.

[0078] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit them; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions described in the foregoing embodiments, or perform equivalent replacements for some of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A control method for an underwater cleaning robot, characterized in that: The following steps are involved: S1. Collecting an environmental image of the area to be cleaned through a global camera, and performing a preprocessing operation on the environmental image to obtain a preprocessed environmental image; wherein the preprocessing operation includes distortion correction and perspective transformation; S2. Generate a static environment map of the area to be cleaned based on the preprocessed environment image, wherein map elements of the static environment map include an obstacle sub-area and a dirt sub-area; S3, adopting a multi-scale path planning strategy to hierarchically generate an operation path sequence from the operation starting point of the underwater cleaning robot to the cleaning end point in each dirt sub-area, specifically including: S31, using a grid of a first scale to grid the static environment map to generate a grid map of the first scale; determining the cleaning end points in each dirt sub-area to obtain multiple cleaning end points; based on the positions of the multiple cleaning end points and the position of the operation starting point of the underwater cleaning robot, planning the visit sequence of the underwater cleaning robot to each dirt sub-area in the grid map of the first scale based on the A* algorithm; S32, taking the first dirt sub-region in the access sequence as the target dirt sub-region, and then planning a travel path of the underwater cleaning robot from the operation starting point to a turning point of the target dirt sub-region in the grid map of the first scale, wherein the turning point is located at a boundary of the target dirt sub-region; S33, using a second-scale grid to divide the static environment map into grids to generate a second-scale grid map, wherein the grid size of the second-scale grid is smaller than the grid size of the first-scale grid; planning a cleaning path from the turning point to the cleaning end point in the target dirt sub-area in the second-scale grid map; S34, taking the cleaning end point in the target dirt sub-area as the updated operation starting point of the cleaning robot, and then removing the target dirt sub-area from the access sequence to obtain an updated access sequence; S35, repeating step S32 to step S34 until the operation path sequence is obtained; S4. Complete the cleaning operation of all the dirty sub-areas in the area to be cleaned by the underwater cleaning robot according to the path planned in the operation path sequence.

2. The control method of the underwater cleaning robot according to claim 1, characterized in that: The global camera is arranged at a central position above the area to be cleaned, and the acquisition range of the global camera covers the entire area of ​​the area to be cleaned; wherein the area to be cleaned is a pool or a fish pond to be cleaned.

3. The control method of the underwater cleaning robot according to claim 1, characterized in that: The performing a preprocessing operation on the environment image to obtain the preprocessed environment image specifically includes: Using the Zhang Zhengyou calibration method to perform distortion correction on the environmental image, so as to eliminate the distortion generated by the lens of the global camera when collecting the environmental image, and obtain the environmental image after distortion correction; According to the pre-calibrated coordinates of the four corner points of the area to be cleaned, a perspective transformation is performed on the environment image after the distortion correction process, so as to convert the oblique image of the global camera perspective into a top-down plan view of a bird's-eye view.

4. The control method of the underwater cleaning robot according to claim 1, characterized in that: The dirt sub-region in step S2 is located using a target detection algorithm to identify the dirt distribution by analyzing the color, texture or shape features in the image.

5. The control method of the underwater cleaning robot according to claim 1, characterized in that: The side length of the grid of the first scale is 3 to 4 times the side length of the grid of the second scale, and the area of ​​the grid of the second scale is smaller than the coverage area of ​​the underwater cleaning robot body.

6. The control method of the underwater cleaning robot according to claim 1, characterized in that: An AprilTag code is installed on the top of the underwater cleaning robot; during the movement of the underwater cleaning robot, the global camera recognizes the AprilTag code in real time to obtain the current position of the underwater cleaning robot, and determines the actual position of the underwater cleaning robot in the area to be cleaned in combination with perspective transformation and coordinate mapping, and then adjusts the movement of the underwater cleaning robot in combination with the PID control algorithm, so that it moves to the target area along the planned path.

7. The control method of the underwater cleaning robot according to claim 6, characterized in that: The PID control algorithm dynamically adjusts parameters according to the distance between the current position of the underwater cleaning robot and the target point; when the distance is large, the proportional coefficient Kp is increased to increase the moving speed of the underwater cleaning robot, and when the distance is small, the differential coefficient Kd is increased to reduce the position deviation of the underwater cleaning robot.

8. The control method of the underwater cleaning robot according to claim 1, characterized in that: The method further includes replanning a cleaning path and performing subsequent cleaning operations if it is detected that there is a dirt sub-area that has not been cleaned or a cleaning blind area in the area to be cleaned.

Images