Control method, device and storage medium of wall identification system

Abstract

The application discloses a wall surface recognition system control method, equipment and storage medium, relates to the technical field of sensor positioning, and the method comprises the following steps: determining target point cloud data corresponding to a wall surface according to a wall surface boundary line in point cloud data; determining a boundary point set of the target point cloud data on a horizontal projection plane; traversing all subsets of the boundary point set, determining the circumscribed circle corresponding to each subset, and obtaining a circumscribed circle sequence, wherein the subset contains a preset number of boundary points; in the circumscribed circle sequence, determining the minimum circumscribed circle that surrounds the boundary point set; comparing the radius of the minimum circumscribed circle with a preset radius threshold value, and generating obstacle detection information according to the comparison result. The application determines the wall surface by performing transverse straight line detection through the top laser radar and recognizing the vertical direction of the wall surface through multiple frames of line lasers, and then performs wall surface obstacle recognition based on the algorithm of the method, thereby improving the wall surface obstacle recognition capability of the sweeping robot.

Description

Technical Field

[0001] This application relates to the field of sensor positioning technology, and in particular to a control method, device and storage medium for a wall recognition system. Background Technology

[0002] The core functions of a robotic vacuum cleaner are autonomous navigation and obstacle avoidance, and wall recognition is a key technology for achieving this goal. Furthermore, walls not only define the cleaning path but also contain various obstacles. In related technologies, robotic vacuum cleaners typically place wall recognition sensors near the drive wheel axis, resulting in a small point cloud acquisition area and consequently poor obstacle recognition accuracy.

[0003] The above content is only used to help understand the technical solution of this application and does not represent an admission that the above content is prior art. Summary of the Invention

[0004] The main purpose of this application is to provide a control method, device and storage medium for a wall recognition system, which aims to solve the technical problem in the related art where the physical installation position and detection field of view of the sensor of the sweeping robot are limited, so that the recognition sensor cannot provide complete wall information, which leads to the sweeping robot colliding with obstacles.

[0005] To achieve the above objectives, this application proposes a method for a multi-sensor time-series wall recognition system, the method comprising:

[0006] Based on the wall boundary lines in the point cloud data, determine the target point cloud data corresponding to the wall;

[0007] Determine the set of boundary points of the target point cloud data on the horizontal projection plane;

[0008] Traverse all subsets of the boundary point set, determine the circumcircle corresponding to each subset, and obtain the circumcircle sequence, wherein the subset contains a preset number of boundary points;

[0009] In the sequence of circumcircles, determine the smallest circumcircle that encloses the set of boundary points;

[0010] By comparing the radius of the minimum circumcircle with a preset radius threshold, obstacle detection information is generated based on the comparison result.

[0011] In one embodiment, the collected point cloud data is fitted with a straight line using a random sampling consensus algorithm, and multiple random straight lines are generated. The straight line with the highest proportion of interior points is selected as the initial baseline.

[0012] The portion of the initial baseline whose angle change exceeds a preset angle change threshold is divided into multiple sub-segments;

[0013] Based on the initial baseline and the multiple sub-segments, the initial baseline is locally modified to generate the wall boundary line.

[0014] In one embodiment, the point cloud data is divided into first point cloud data and second point cloud data based on the wall boundary line;

[0015] By combining the lateral projection results of the top lidar, the portion of the first point cloud data and the second point cloud data that is above the ground is determined as the target point cloud data.

[0016] In one embodiment, based on the radius corresponding to each of the minimum circumcircles, a comparison is made with the radius threshold to determine each target circumcircle with a radius smaller than the radius threshold;

[0017] Based on the circumcircle of each target and in conjunction with the horizontal projection plane, determine the center coordinates of each target circumcircle.

[0018] Based on the center coordinates of the target's circumcircle, obstacle detection information including the center coordinates is generated.

[0019] In one embodiment, based on the center coordinates of each circle in the obstacle detection information, the distance between any two adjacent center coordinates is calculated to obtain several center distances;

[0020] The center distances of several circles are compared and filtered, and the center distance with the largest corresponding value is taken as the target distance;

[0021] Based on the target distance, it is compared with a preset distance threshold, and the wall recognition result is determined according to the comparison result.

[0022] In one embodiment, if the comparison result shows that the target distance is less than the distance threshold, the wall recognition result, including a wall detection success notification and the target point cloud data, is released to indicate that the wall detection conforms to the preset protection rules.

[0023] If the comparison result shows that the target distance is greater than the distance threshold, the wall recognition result, including a wall detection failure notification, is issued to indicate that the wall does not meet the preset protection rules.

[0024] In one embodiment, the projection result corresponding to the line lidar and the horizontal projection result corresponding to the top lidar are generated based on the laser illumination perpendicular to the ground from the line lidar and the horizontal laser illumination from the top lidar.

[0025] The projection result and the horizontal projection result are identified and judged. If both the projection result and the horizontal projection result are straight lines, then the sampling position in the projection result contains a wall.

[0026] In one embodiment, a projection surface is formed based on the linear laser line projected by the line lidar and combined with the lateral projection result of the top lidar;

[0027] Based on the projection surface, the robot is controlled to move in a direction parallel to the wall to obtain multiple time-series projection planes;

[0028] The horizontal projection plane is generated by projecting onto the horizontal plane based on the multi-time projection plane.

[0029] In addition, to achieve the above objectives, this application also proposes a wall recognition device, the device comprising: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the control method of the wall recognition system as described above.

[0030] In addition, to achieve the above objectives, this application also proposes a storage medium, which is a computer-readable storage medium, on which a computer program is stored, and when the computer program is executed by a processor, it implements the steps of the control method of the wall recognition system as described above.

[0031] This application provides a control method for a wall recognition system, including: determining target point cloud data corresponding to a wall based on wall boundary lines in point cloud data; determining a set of boundary points of the target point cloud data on a horizontal projection plane; traversing all subsets of the boundary point set to determine the circumcircle corresponding to each subset, obtaining a sequence of circumcircles, wherein each subset contains a preset number of boundary points; determining the smallest circumcircle surrounding the boundary point set in the sequence of circumcircles; comparing the radius of the smallest circumcircle with a preset radius threshold, and generating obstacle detection information based on the comparison result. This method utilizes LiDAR to illuminate the wall, identifies the wall position based on the generated point cloud data, and then uses algorithms to identify obstacles on the wall, improving the robot vacuum cleaner's ability to recognize wall obstacles.

[0032] In summary, this application uses a top-mounted lidar to detect horizontal straight lines and uses multi-frame line laser recognition to identify the vertical direction of the wall to determine the wall's length and position. Then, based on the algorithm of this method, it identifies obstacles on the wall, thereby improving the robot vacuum cleaner's ability to identify obstacles on the wall. Attached Figure Description

[0033] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0034] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, for those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0035] Figure 1 This is a flowchart illustrating the first embodiment of the control method for the wall recognition system of this application;

[0036] Figure 2 This is a top-down view of the point cloud when there are obstacles.

[0037] Figure 3 A top-down point cloud map showing the unobstructed view;

[0038] Figure 4 This is a flowchart illustrating the second embodiment of the control method for the wall recognition system of this application;

[0039] Figure 5 This is a diagram showing the arrangement of the sensors in this application;

[0040] Figure 6 This is a cross-sectional view of the sensor arrangement method in this application;

[0041] Figure 7 This is a flowchart illustrating the fourth embodiment of the control method for the wall recognition system of this application;

[0042] Figure 8 This is a flowchart illustrating the fifth embodiment of the control method for the wall recognition system of this application;

[0043] Figure 9 An image showing the effect of sampling a straight wall after the robot moves;

[0044] Figure 10 This is a flowchart of the control method for the wall recognition system of this application;

[0045] Figure 11 This is a structural schematic diagram of the wall recognition device of this application.

[0046] The purpose, features, and advantages of this application will be further explained in conjunction with the embodiments and with reference to the accompanying drawings. Detailed Implementation

[0047] It should be understood that the specific embodiments described herein are merely illustrative of the technical solutions of this application and are not intended to limit this application.

[0048] In related technologies, robotic vacuum cleaners generally place wall recognition sensors near the drive wheel axis, resulting in a small point cloud acquisition area, which in turn leads to poor obstacle recognition accuracy.

[0049] This application provides a solution: First, based on the wall boundary line in the point cloud data, the target point cloud data corresponding to the wall is determined. Then, the boundary point set of the target point cloud data on the horizontal projection plane is determined. Next, all subsets of the boundary point set are traversed to determine the circumcircle corresponding to each subset, resulting in a sequence of circumcircles. The subset contains a preset number of boundary points. Then, in the sequence of circumcircles, the smallest circumcircle surrounding the boundary point set is determined. Finally, the radius of the smallest circumcircle is compared with a preset radius threshold, and obstacle detection information is generated based on the comparison result.

[0050] It should be noted that the executing entity in this embodiment can be a computing service device with data processing, network communication, and program execution functions, such as a tablet computer, personal computer, or mobile phone, or an electronic device or wall recognition device capable of performing the above functions. The following description uses a wall recognition device as an example to illustrate this embodiment and the subsequent embodiments.

[0051] To better understand the technical solution of this application, a detailed description will be provided below in conjunction with the accompanying drawings and specific implementation methods.

[0052] This application provides a control method for a wall recognition system, referring to... Figure 1 , Figure 1 This is a flowchart illustrating the first embodiment of the control method for the wall recognition system of this application.

[0053] In this embodiment, the control method of the wall recognition system includes steps S10 to S50:

[0054] Step S10: Determine the target point cloud data corresponding to the wall based on the wall boundary line in the point cloud data.

[0055] In this embodiment, point cloud data refers to a collection of a large number of discrete three-dimensional coordinate points, representing the spatial distribution characteristics of the surface of the scanned object. The wall surface refers to a spatial region above the ground and whose vertical coordinate values ​​exceed the ground baseline, with the ground reference height as a reference. The boundary line refers to a valid line segment that has been geometrically verified and represents the wall surface baseline. The target point cloud data refers to a subset of the point cloud representing the wall surface, which must satisfy constraints on height continuity, flatness, and relative position with the robot.

[0056] As an optional implementation, in a scenario where the robot vacuum cleaner cruises parallel to the wall, a coordinate system is established based on the straight line segments corresponding to the wall's dividing line, using these straight line segments as the dividing lines. Then, within this coordinate system, point cloud data above the ground is defined as the target point cloud data corresponding to the wall, while point cloud data parallel to the dividing lines is defined as ground point cloud data.

[0057] Step S20: Determine the set of boundary points of the target point cloud data on the horizontal projection plane.

[0058] In this embodiment, the horizontal projection plane refers to the two-dimensional coordinate space after projecting the three-dimensional point cloud along the vertical axis onto the two-dimensional plane, ignoring height information. The boundary point set refers to the discrete set of points on the projection plane that constitute the boundary of the point cloud outline, usually located at the outer edge of the point cloud or at abrupt changes in shape.

[0059] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, a two-dimensional scatter plot is generated by projecting the target point cloud data along the height direction of the wall onto a horizontal projection plane. Discrete points of the point cloud outline boundary are then selected from the two-dimensional scatter plot, and a set of boundary points is generated based on the discrete points.

[0060] Step S30: Traverse all subsets of the boundary point set, determine the circumcircle corresponding to each subset, and obtain the circumcircle sequence, wherein the subset contains a preset number of boundary points.

[0061] In this embodiment, a subset refers to a sequence of continuous or discontinuous point segments extracted from the boundary point set. The preset quantity refers to the number of points contained in the subset, defined by the user, used for constructing the circumcircle. The circumcircle is the smallest closed circle that completely encloses all points within the subset; its center and radius are determined by the extreme points. The circumcircle sequence refers to a set of multiple circumcircles generated in the order of subset traversal, used to characterize the local geometric features of the boundary contour.

[0062] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, a circumscribed circle is constructed based on a subset containing a preset number of boundary points. A circumscribed circle is determined by the boundary points in the subset. The construction process of the circumscribed circle is repeated for each subset to obtain several circumscribed circles. The several circumscribed circles are then combined to generate a sequence of circumscribed circles.

[0063] Step S40: In the sequence of circumcircles, determine the smallest circumcircle that surrounds the set of boundary points.

[0064] In this embodiment, the minimum circumcircle is the circle with the smallest radius that completely encloses all points of the boundary point set, and it is dynamically determined by the extreme points. The Welzl algorithm is a geometric algorithm that recursively solves for the minimum enclosing circle of a point set.

[0065] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, all circumcircles in the circumcircle sequence are traversed, and after being sorted in ascending order by radius, the first tenth of the smallest radius is selected as candidate circles. The candidate circles are then verified to check whether all points in the boundary point set are located on or inside the candidate circle. Those that meet the requirements are selected as valid candidate circles, and the smallest radius among the valid candidate circles is selected as the smallest circumcircle.

[0066] As an optional implementation of screening circumcircles, in the scenario where the robot vacuum cleaner cruises parallel to the wall, if none of the candidate circles contain a circumcircle with a set of boundary points, then the circumcircle sequence arranged in ascending order of radius is used for verification. The verification is performed according to the radius from smallest to largest to see if all points of the boundary point set are located on or inside the circle, until a circumcircle that meets the requirements is found. The circumcircle that meets the requirements is then taken as the smallest circumcircle.

[0067] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, based on the target point cloud data, the projection data of the target point cloud data on the horizontal projection plane is obtained, and the Welz minimum enclosing circle algorithm is applied to the projection data corresponding to the target point cloud data. By calculating and determining the unique minimum enclosing circle through any three non-collinear points on the projection data.

[0068] Step S50: Compare the radius of the minimum circumcircle with a preset radius threshold, and generate obstacle detection information based on the comparison result.

[0069] In this embodiment, the preset radius threshold refers to a critical radius value that is custom-set according to the application scenario. The comparison judgment refers to comparing the actually calculated minimum circumscribed circle radius with the threshold to determine whether it meets the preset conditions. The detection result refers to the classification result output based on the threshold comparison.

[0070] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, based on the smallest circumcircle calculated through projection data, a comparison is made with a preset radius threshold. The radius of each smallest circumcircle is compared with the radius threshold, and obstacle detection information is generated according to the comparison results.

[0071] As an optional implementation method for generating obstacle detection information, in a scenario where the robot vacuum cleaner cruises parallel to the wall, if the radius of each smallest circumscribed circle is greater than a radius threshold, then based on this comparison result, an operation to clear the wall detection result is generated, thus generating obstacle detection information indicating the presence of obvious obstacles.

[0072] As an optional implementation method for generating obstacle detection information, in a scenario where the robot vacuum cleaner cruises parallel to the wall, the radius of each smallest circumscribed circle is smaller than the radius threshold. Based on the comparison results, point cloud data of the corresponding wall and detection information for no obvious obstacles are generated.

[0073] The system generates point cloud data for the wall surface and provides obstacle detection information for areas without obvious obstacles.

[0074] For example, refer to Figure 2 and Figure 3 , Figure 2 This is a top-down view of the point cloud with obstacles present. Figure 3 This is a top-down point cloud map when there are no obstacles. In a scenario where the robot vacuum is cruising parallel to the wall, a T-shaped projection surface is formed by LiDAR and line lasers. When both the horizontal projection of the LiDAR and the vertical projection of the line lasers are straight lines, the sampling location of the line laser point can be considered part of the wall surface. As the robot moves, multiple sampled line laser lines can form a plane. The higher the sampling frequency of the line lasers and the smaller the spacing between the lines, the lower the probability of missing small obstacles.

[0075] Based on the point cloud data generated by the line LiDAR illumination, a random sampling consensus algorithm is used to fit a straight line. Any two points in the point cloud are selected to generate a candidate straight line equation y=kx+b. Then, the number of interior points whose distance to this straight line is less than a set threshold is counted. If the number of interior points is 100, after 50 attempts, the line with the highest proportion of interior points is selected as the initial baseline with the equation y=kx+b. The initial baseline is then segmented using the random sampling consensus algorithm. For cases where the actual wall surface has bends or unevenness, local corrections are made to the fitted initial baseline. The change in orientation angle between adjacent point sets is calculated using a sliding window (e.g., calculating the angle difference between the first and last points in groups of 5). When the angle change exceeds 5 degrees, a segmentation point is inserted, breaking the long straight line into multiple sub-segments. Invalid segments are filtered out by combining physical constraints (e.g., the sub-segment length must exceed 1.2 times the robot's width). The initial baseline is then locally corrected using several sub-segments to obtain and identify at least one straight line segment that traverses the side of the robot closest to the wall. Point cloud data is segmented by straight line segments to identify and acquire target point cloud data above the horizontal plane. Initial projection data of the target point cloud data onto the horizontal projection plane is obtained. Based on the distribution of the target point cloud data, the projection data undergoes secondary filtering, removing isolated discrete points and retaining the continuously distributed core point set to obtain the projection data. The Welz minimum bounding circle algorithm is applied to the projection data, calculating and determining a unique minimum bounding circle using any three non-collinear points on the projection data. This process of selecting points and calculating the minimum bounding circle is repeated to obtain several minimum bounding circles. These minimum bounding circles are compared with a preset radius threshold (e.g., 3mm). A comparison result is obtained where the radius of each minimum bounding circle is less than the radius threshold (3mm). Based on this comparison result, obstacle detection information is generated for the corresponding wall point cloud data where no obvious obstacles exist.

[0076] By using top-mounted lidar for horizontal line detection and multi-frame line laser recognition to determine the vertical direction of the wall, as well as the length and position of the wall, and then using the algorithm based on this method to identify obstacles on the wall, the robot vacuum cleaner's ability to identify obstacles on the wall is improved.

[0077] Based on any of the above embodiments, in Embodiment 2 of this application, referring to Figure 4 , Figure 4 This is a flowchart illustrating the second embodiment of the control method for the wall recognition system of this application. Before step S10, steps A11-A13 are also included:

[0078] Step A11: The collected point cloud data is fitted with a straight line using a random sampling consensus algorithm. Multiple random straight lines are generated, and the straight line with the highest proportion of interior points is selected as the initial baseline.

[0079] In this embodiment, the Random Sample Consensus Algorithm (RANSAC) is an iterative optimization algorithm that establishes candidate models by repeatedly randomly sampling the smallest dataset and evaluates the reliability of the model based on the number of inliers. Line fitting refers to the mathematical modeling process of solving the optimal straight line equation from a discrete set of points based on the least squares method or geometric optimization criteria. Inliers are data points that meet a preset error threshold and are considered support points for a valid fit. The initial baseline is the preliminary fitting result determined by the candidate line with the highest proportion of inliers, serving as the basis for subsequent optimization.

[0080] As an optional implementation, in scenarios where the robotic vacuum cleaner cruises parallel to the wall, the collected point cloud data is preprocessed, including height-based calibration and outlier filtering, retaining the valid point set. Random sampling algorithm parameters are initialized, setting the maximum number of iterations and a dynamically adjusted inlier distance threshold. In each iteration, two non-collinear points are randomly selected to generate candidate straight line models. The geometric distances from all points to this line are calculated, and the number of support points meeting the threshold condition is counted. Simultaneously, the current optimal model and its corresponding number of inliers are recorded, and the candidate straight line with the highest inlier percentage is selected as the initial baseline.

[0081] Furthermore, in a scenario where the robot vacuum cleaner cruises parallel to the wall, based on the point cloud data generated by the line LiDAR illuminating the wall, a straight line is fitted using the random sampling consensus algorithm. Two points are randomly selected to generate candidate straight line equations, and the number of interior points whose distance to the straight line is less than a set threshold is counted. After several random attempts, the straight line with the highest percentage of interior points is selected as the initial baseline. Then, line segmentation is performed using the random sampling consensus algorithm. For cases where the wall surface has bends or unevenness, local corrections are made to the fitted initial baseline. The change in the directional angle of adjacent point sets is calculated using a sliding window. When the angle change exceeds a preset threshold, a segmentation point is inserted, decomposing the long straight line into several sub-segments. The initial baseline is then locally corrected using these sub-segments to obtain and identify at least one straight line segment that traverses the side of the robot closest to the wall, and this straight line segment is used as the wall boundary line.

[0082] Step A12: Divide the portion of the initial baseline whose angle change exceeds a preset angle change threshold into multiple sub-segments.

[0083] In this embodiment, line segment splitting refers to the operation of detecting curvature abrupt changes on a fitted straight line and dividing the continuous straight line into multiple sub-segments that satisfy local linear constraints. The angle change refers to the directional offset of adjacent points relative to the baseline, and is calculated using vector angles to characterize the degree of structural abrupt changes. The preset angle change threshold is the critical angle deviation for determining whether splitting is necessary. The straight line with the highest proportion of interior points refers to the candidate model that has the largest proportion of supporting points and meets the error range in multiple random samplings.

[0084] As an optional implementation, in scenarios where the robot vacuum cleaner cruises parallel to the wall, for cases where the actual wall surface has bends or unevenness, local corrections are made on the fitted initial baseline. The change in orientation angle of adjacent point sets is calculated using a sliding window. When the angle change exceeds a preset threshold, the system inserts a dynamic segmentation point along the change location and decomposes the original long line segment into several sub-segments.

[0085] Step A13: Based on the initial baseline and the multiple sub-segments, locally modify the initial baseline to generate the wall boundary line.

[0086] In this embodiment, local correction refers to segmenting and adjusting the initial baseline based on the spatial distribution of sub-segments to improve the continuity of the baseline at corners.

[0087] As an optional implementation, in a scenario where the robotic vacuum cleaner cruises parallel to the wall, the baseline coverage area is analyzed along the robot's movement direction based on the initial baseline and the set of disassembled sub-segments. Density clustering and orientation consistency checks are performed on the point cloud subset corresponding to each sub-segment to remove discrete noise points, retain continuously distributed core point groups, and check the geometric coherence between adjacent sub-segments. If the angle difference at the connection point is less than a preset merging threshold, each sub-segment is merged with the initial baseline into a single straight line segment. The robot identifies the straight line segment crossing one side of itself and uses this straight line segment as the wall boundary line.

[0088] For example, in a scenario where a robotic vacuum cleaner cruises parallel to a wall, the point cloud data collected by sensors is denoised and height-filtered before being projected onto a horizontal two-dimensional plane. A random sampling consensus algorithm is used to randomly select two points in the point cloud multiple times to generate candidate lines, and the interior point support rate of each model is calculated. The line with the densest interior point distribution is selected as the initial baseline. The change in the normal vector angle of the point cloud is analyzed segment by segment along the initial baseline using a sliding window. When the angle offset of a local area exceeds a preset threshold, a dynamic segmentation point is inserted to divide the long baseline into multiple geometrically coherent sub-segments. Secondary clustering verification is performed based on the interior point density and directional consistency of the sub-segments. Based on the initial baseline and the set of sub-segments, a priority analysis region is defined near the robot's real-time pose. Each sub-segment is sorted by lateral distance and its motion direction fit is evaluated, and the candidate line segment with the closest distance and the smallest angle with the robot's trajectory is selected. After verifying its physical existence through multi-sensor feedback, endpoint extension and curvature smoothing are performed, and finally an optimized baseline that fits the real shape of the wall and meets the obstacle avoidance distance constraint is output. This baseline is then updated to the environmental map as a reliable basis for navigation control.

[0089] By using a line fitting and line segment splitting process based on the principle of random sampling consensus algorithm, the wall baseline is accurately identified, thus improving the accuracy of the robot vacuum cleaner in recognizing wall obstacles.

[0090] Based on any of the above embodiments, in Embodiment 3 of this application, step S10 includes steps B11 to B12:

[0091] Step B11: Based on the wall boundary line, the point cloud data is divided into first point cloud data and second point cloud data.

[0092] In this embodiment, segmentation refers to the data segmentation operation that divides the mixed point cloud into different semantic categories, including ground or wall, based on geometric features. The first point cloud data and the second point cloud data refer to the point cloud data generated by the line LiDAR illumination being segmented into two parts by the wall boundary line.

[0093] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, the point cloud data is divided into two parts based on the wall boundary line and the straight line segment corresponding to the wall boundary line. One side of the straight line segment is the ground point cloud data, and the other side of the straight line segment is the target point cloud data.

[0094] Step B12: Combining the lateral projection results of the top lidar, determine the portion of the first point cloud data and the second point cloud data that is above the ground as the target point cloud data.

[0095] In this embodiment, the horizontal projection result of the top lidar refers to the result of the lidar installed on the top of the robot illuminating horizontally, projecting the laser vertically onto the wall to supplement the environmental structure information in the vertical direction.

[0096] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, based on the horizontal laser beam that is vertically irradiated onto the wall by the top lidar, the projection result of the horizontal laser beam is used to determine whether the projection result of the horizontal laser beam intersects with part of the point cloud data on one side of the straight line segment. This part of the point cloud data is identified as the target point cloud data, and the part of the point cloud data on the other side of the straight line segment is identified as the ground point cloud data.

[0097] For example, in a scenario where the robot vacuum cleaner cruises parallel to the wall, refer to Figure 5 and Figure 6 , Figure 5 This is a diagram showing the arrangement of the sensors in this application. Figure 6 This is a cross-sectional view of the sensor arrangement method in this application. Line lasers are a type of 2D solid-state laser sensor with a limited field of view (FOV), typically around 100°. When arranging the sensors, the line laser is emitted diagonally to the right front and tilted at a certain angle from top to bottom, allowing the line laser to cover the entire wall surface. During edge-fitting operations, the highest point on the wall hit by the line laser must not be lower than the height of the top lidar's emitting surface. This allows the line laser and the top lidar to form an intersection point, enabling the robot's movement to determine whether the results from multiple frames can form a surface. By identifying straight line segments, the point cloud data is divided into two parts. Combining the intersection of the top laser projection result and the point cloud data, the target point cloud data and the ground point cloud data are identified.

[0098] By combining the projection results of top lidar and line lidar, ground point cloud data and target point cloud data can be effectively distinguished, thus improving the accuracy of the robot vacuum cleaner in identifying target point cloud data.

[0099] Based on any of the above embodiments, in Embodiment 4 of this application, referring to Figure 7 , Figure 7 This is a flowchart illustrating the fourth embodiment of the control method for the wall recognition system of this application. Step S50 includes steps C11 to C13:

[0100] Step C11: Based on the radius corresponding to each of the minimum circumcircles, compare and determine with the radius threshold to identify each target circumcircle with a radius smaller than the radius threshold.

[0101] In this embodiment, the target circumcircle refers to the circumcircle that is determined to be abnormal or require attention after threshold screening.

[0102] As an optional implementation, in a scenario where the robotic vacuum cleaner cruises parallel to the wall, a radius threshold is dynamically calculated based on the minimum circumscribed circle radius of each point set, combined with the robot's real-time motion state. By comparing the relationship between each minimum circumscribed circle radius and the preset radius threshold, all target circumscribed circles whose radii exceed the limit are identified and marked.

[0103] Step C12: Determine the center coordinates of each target circumcircle based on the horizontal projection plane.

[0104] In this embodiment, the center coordinates refer to the position parameters of the geometric center of the circumscribed circle, which are used to describe the absolute or relative spatial positioning of the target area.

[0105] As an optional implementation, in a scenario where the robotic vacuum cleaner cruises parallel to the wall, coordinates are established on a horizontal projection plane based on the point cloud covered by the circumcircle of each target. The coordinates of the center of the circumcircle of the target on the horizontal projection plane are then determined using the coordinate system constructed on the horizontal projection plane.

[0106] Step C13: Based on the center coordinates of the target's circumcircle, generate obstacle detection information including the center coordinates.

[0107] In this embodiment, the detection result including the center coordinates means writing the center coordinates into the corresponding detection result to determine the coordinates of the obstacle on the wall.

[0108] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, the center coordinates of each target's circumcircle are used to correlate the center coordinates with the obstacle's position, generating obstacle detection information including the center coordinates.

[0109] As an implementation method with optional preset obstacle conditions, in a scenario where the robot vacuum cleaner cruises parallel to the wall, dynamic center coordinates are generated based on the identified dynamic point cloud data. Combined with the temporary obstacle judgment condition in the preset obstacle judgment conditions, when the center coordinates are detected to be dynamically passing by, the center coordinates are filtered out as temporary obstacles.

[0110] For example, in a scenario where a robotic vacuum cleaner cruises parallel to a wall, the radius of each point cloud cluster is calculated based on its minimum circumcircle parameter and compared with a dynamic radius threshold. After filtering out valid target circumcircles with radii below the threshold, the spatial distribution and normal vector features of the target point cloud data are combined with multi-frame data alignment and coordinate transformation to map the target circumcircles to the global coordinate system, eliminating abnormal circles caused by projection distortion or momentary occlusion. For valid target circumcircles, the height gradient distribution of the target point cloud data and the lateral scanning results of the top LiDAR are fused to perform reverse verification and dynamic compensation on the center coordinates, determining the updated center coordinates. By associating obstacles with the center coordinates, obstacle detection information including the center coordinates is generated.

[0111] By combining the Welz minimum enclosing circle algorithm with a preset radius threshold, the robot effectively identifies obstacles on the wall, improving the accuracy of obstacle recognition for the robot vacuum cleaner.

[0112] Based on any of the above embodiments, in Embodiment 5 of this application, referring to Figure 8 , Figure 8 This is a flowchart illustrating the fifth embodiment of the control method for the wall recognition system of this application. Following step C13, steps D11-D13 are also included:

[0113] Step D11: Based on the center coordinates of each circle in the obstacle detection information, calculate the distance between any two adjacent center coordinates to obtain several center distances.

[0114] In this embodiment, adjacent center coordinates refer to two consecutive center locations in spatial distribution or temporal correlation. Distance calculation refers to the numerical quantification process of obtaining the linear interval between two points through geometric operations. Center distance refers to the actual spatial interval between adjacent center points, used to analyze the continuity of environmental structure or the density of obstacle distribution.

[0115] As an optional implementation, in a scenario where the robot vacuum cleaner cruises parallel to the wall, based on the coordinates of each circle's center in the detection results, the distance between any two adjacent circle center coordinates is calculated to determine the distance between those adjacent circle center coordinates. This distance calculation step is repeated to generate several circle center distances.

[0116] Step D12: Compare and filter the center distances of several circles, and take the center distance with the largest corresponding value as the target distance.

[0117] In this embodiment, the distance between several center points refers to multiple spatial interval values ​​obtained by calculating the coordinates of adjacent center points. The target distance refers to the maximum center point distance determined after screening, which is a core parameter used for decision-making or path planning.

[0118] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, the numerical values ​​of several center distances are compared, and the center distance with the largest numerical value is selected as the target distance.

[0119] As an alternative implementation method where the center distance is too large, in the scenario where the robot vacuum cleaner cruises parallel to the wall, if the target distance reaches a first preset threshold, local point cloud resampling and center coordinate interpolation correction are triggered to generate the corrected target distance.

[0120] Step D13: Based on the target distance, compare it with a preset distance threshold, and determine the wall recognition result according to the comparison result.

[0121] In this embodiment, the preset distance threshold refers to the distance threshold between adjacent circle centers that is custom-set according to scene requirements. The comparison result refers to the decision conclusion generated based on the threshold comparison, used to determine whether the wall obstacle detection was successful. The wall recognition result refers to the corresponding recognition result generated based on the comparison result, indicating whether the obstacle exists.

[0122] As an optional implementation, in scenarios where the robot vacuum cleaner cruises parallel to the wall, the distance between the centers of any two adjacent circles is compared with a preset distance threshold, and a judgment is made according to a preset center-to-center distance rule. A corresponding judgment result is generated based on whether the center-to-center distance is greater than or less than the distance threshold.

[0123] For example, in a scenario where the robot vacuum cleaner cruises parallel to the wall, based on the center coordinates of each circle in the detection results, two adjacent center coordinates are arbitrarily selected. The center distance between the two adjacent center coordinates is calculated using the coordinate distance formula, resulting in several center distances. These center distances are then compared numerically, and the center distance with the largest value is selected as the target distance, which is 1cm. This target distance is then compared with a preset distance threshold of 1.5cm. Since the target distance is less than the preset distance threshold, a judgment result corresponding to "target distance is less than distance threshold" is generated.

[0124] By using a preset distance threshold to determine the distance between the center coordinates of adjacent circles, the success or failure of the target point cloud data detection can be identified. This distance threshold-based judgment improves the accuracy of the robot vacuum cleaner in recognizing obstacles on walls.

[0125] Based on any of the above embodiments, in Embodiment Six of this application, step D13 includes steps E11 to E12:

[0126] Step E11: If the comparison result shows that the target distance is less than the distance threshold, then the wall recognition result, including a wall detection success notification and the target point cloud data, is released to indicate that the wall detection conforms to the preset protection rules.

[0127] In this embodiment, successful wall detection means that the target point cloud data is determined to meet the preset judgment rules when the target distance is less than a threshold. The preset protection rules refer to the set protection measures. If the detection distance between two frames is large, obstacles that may appear in between will be skipped, and the detection will be considered a failure.

[0128] As an optional implementation, in scenarios where the robot vacuum cleaner cruises parallel to the wall, a preset center-distance judgment rule is used. If the judgment result shows that the target distance is less than a distance threshold, a notification of successful wall detection is issued, along with the target point cloud data generated by the wall detection.

[0129] As one implementation method for sending notifications, in a scenario where the robot vacuum cleaner cruises parallel to the wall, after the judgment result is generated, a corresponding notification is issued based on the judgment result, and the robot broadcasts the corresponding notification, or the notification is sent to a mobile device.

[0130] Step E12: If the comparison result shows that the target distance is greater than the distance threshold, then the wall recognition result, including a wall detection failure notification, is issued to indicate that the wall does not meet the preset protection rules.

[0131] In this embodiment, wall detection failure means that when the target distance exceeds a threshold, the target point cloud data is determined to not conform to the preset judgment rules.

[0132] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, according to the preset center distance judgment rule, if the judgment result shows that the target distance is greater than the distance threshold, a notification of wall detection failure is issued, and the target point cloud data generated by the wall detection is cleared.

[0133] For example, in a scenario where a robotic vacuum cleaner cruises parallel to a wall, based on a real-time comparison of the target distance and a dynamic distance threshold, if the target distance is determined to be less than the threshold, a successful wall detection process is triggered. The spatiotemporal consistency of the center coordinates of the most recent three frames is traced back via a sliding window to verify the continuity of the target point cloud data (e.g., azimuth angle fluctuation less than 5°, center-to-center distance standard deviation less than 0.1W). Once confirmed as valid, a notification containing the normalized wall azimuth angle, fitted boundary points, and confidence labels is issued. Simultaneously, this is injected into the navigation system to generate a wall-following tracking path. At the same time, the height gradient distribution of the top LiDAR and the inertial navigation attitude angle data are fused to compensate for motion distortion in the wall coordinates. If the deviation between the infrared ranging feedback and the target distance exceeds 15%, local point cloud resampling is initiated. A random sampling consensus algorithm is used to fit and correct a straight line on the wall within a preset range around the center. Conversely, if the target distance exceeds the threshold, the wall detection is deemed a failure.

[0134] By setting a distance threshold and combining it with the center distance judgment rule, the validity and accuracy of the target point cloud data can be detected, and the target point cloud data can be judged and corrected in a timely manner, thus improving the accuracy of the robot vacuum cleaner in recognizing obstacles on the wall.

[0135] Based on any of the above embodiments, in Embodiment 7 of this application, before step S10, steps F11 to F12 are further included:

[0136] Step F11: Based on the laser illumination perpendicular to the ground from the line lidar and the horizontal laser illumination from the top lidar, generate the projection result corresponding to the line lidar and the horizontal projection result corresponding to the top lidar.

[0137] In this embodiment, a line lidar refers to a sensor device that generates three-dimensional coordinate data of an object's surface by emitting a fan-shaped laser beam to scan the environment and receiving reflected signals. Vertical laser illumination refers to a scanning method where the laser beam is pointed at a 90° angle towards the ground, used to capture the height and distance information of vertical structures such as walls and columns. A top lidar refers to a horizontal scanning lidar installed on the top of the robot. Horizontal laser illumination refers to a detection mode where the laser beam scans horizontally, used to measure the distance to obstacles or the width of passages in a horizontal plane. The projection result refers to the projection result formed on the wall surface after the lidar scan. The horizontal projection result is a set of horizontal spatial point clouds output by the top lidar's horizontal scan.

[0138] As an optional implementation, in a scenario where the robot vacuum cleaner cruises parallel to the wall, the projection result corresponding to the line lidar is generated based on the laser emitted by the line lidar perpendicular to the ground, which illuminates the wall. Similarly, the horizontal projection result corresponding to the top lidar is generated based on the horizontal laser illumination of the top lidar onto the wall.

[0139] Step F12: Identify and judge the projection result and the horizontal projection result. If both the projection result and the horizontal projection result are straight lines, then the sampling position in the projection result contains a wall.

[0140] In this embodiment, identification and judgment refers to the classification process of performing geometric feature analysis on point cloud data.

[0141] As an optional implementation, in a scenario where the robotic vacuum cleaner cruises parallel to the wall, the projection results corresponding to the line lidar and the horizontal projection results corresponding to the top lidar are identified to determine whether the projection results and the horizontal projection results are straight lines. If the projection results and the horizontal projection results are not straight lines, dynamic distortion compensation is performed based on the robot's motion state. If the compensated projection results and the horizontal projection results are straight lines, then the wall is present at the position used in the projection result.

[0142] For example, in a scenario where a robotic vacuum cleaner cruises parallel to a wall, coordinate transformation and height gradient filtering are performed on the vertical scan data from a line LiDAR to extract the vertical point cloud projection result. After removing ground reflection noise, a random sampling consensus algorithm is used to fit candidate straight lines on the wall (with a residual threshold set to 0.03m). Simultaneously, the horizontal scan data from the top LiDAR is processed, and horizontal projection results are generated through motion distortion compensation and dynamic clustering. The least squares method is then used to fit a straight line in the horizontal direction. Spatial logic verification is performed on the straight line fitting results from the two LiDARs: if the angle between the normal vectors of the vertical and horizontal straight lines is less than 5° and the extension direction is consistent with the direction of the robot's motion trajectory, then a valid wall surface is determined to exist. Multi-level verification is implemented for the fitting results. By backtracking the straight line parameters of the most recent 3 frames through a time window (direction angle fluctuation tolerance ±3°, length change rate <15%), the continuity of the wall surface is confirmed.

[0143] The combination of linear and top-mounted lidar generates projection and horizontal projection results, effectively identifying the presence of walls and improving the accuracy of the robot vacuum cleaner in recognizing wall obstacles.

[0144] Based on any of the above embodiments, in Embodiment 8 of this application, before step S20, steps G11~G13 are further included:

[0145] Step G11: Based on the linear laser line projected by the line lidar, and combined with the horizontal projection result of the top lidar, a projection surface is formed.

[0146] In this embodiment, the projection surface refers to a three-dimensional spatial plane model constructed by fusing vertical and horizontal laser scanning data, which is used to describe the geometric continuity of the environmental structure.

[0147] As an optional implementation, in a scenario where the robotic vacuum cleaner cruises parallel to the wall, a straight line is extracted from the line laser data of the line LiDAR by performing height gradient filtering and a random sampling consistency algorithm for line fitting. After removing dynamic obstacle noise, the straight line is mapped to the global coordinate system, and the lateral scan data from the top LiDAR is processed simultaneously. A lateral projection result is generated based on motion distortion compensation and cluster analysis. The fitted line in the vertical direction is spatially aligned with the boundary line of the horizontal channel, and the alignment is checked by verifying the angle between the normal vectors and the consistency of the extension direction. If both satisfy the planar constraint conditions in three-dimensional space, it is determined to be a valid projection surface.

[0148] Step G12: Based on the projection surface, control the robot to move in a direction parallel to the wall to obtain multiple time-series projection planes.

[0149] In this embodiment, the direction parallel to the wall refers to the motion trajectory direction that is perpendicular to the wall's normal vector direction and maintains a constant distance. The multi-time projection plane refers to the plane generated by the projection surface based on displacement and time during the robot's movement parallel to the wall.

[0150] As an optional implementation, in scenarios where the robotic vacuum cleaner cruises parallel to the wall, the heading angle of the robotic vacuum cleaner is adjusted by controlling the operating parameters based on the wall plane parameters fitted in the projection surface, so that its movement direction remains parallel to the wall. During the movement, the projection surface formed by the line LiDAR and top LiDAR is acquired in real time, and a horizontal projection plane is generated through continuous mapping of the projection surface.

[0151] Step G13: Project the multi-time-series projection plane onto the horizontal plane to generate the horizontal projection plane.

[0152] In this embodiment, the horizontal plane refers to a two-dimensional reference plane defined by the XY axes of the robot or global coordinate system, used for projection calculations. Projection is a dimensionality reduction operation that orthogonally maps a 3D point cloud to the horizontal plane along the vertical axis, preserving XY coordinate information. The horizontal projection plane refers to the two-dimensional point set generated after projection, representing the geometric distribution of the 3D spatial data in the horizontal direction.

[0153] As an optional implementation, in the scenario where the robot vacuum cleaner cruises parallel to the wall, the multi-time sequence projection plane is projected in the top-down direction and projected onto the horizontal plane to form a horizontal projection plane.

[0154] For example, refer to Figure 9 , Figure 9This is an image showing the effect of sampling along a straight wall after the robot's movement. In a scenario where the robot vacuum cruises parallel to the wall, based on the linear laser lines projected onto the wall by the linear LiDAR, several linear laser lines are projected onto the wall as the robot moves parallel to the wall. An initial surface is generated by calibrating these linear laser lines and combining them with the horizontal projection results from the top LiDAR. Point cloud data is interpolated onto the initial surface to generate projection surfaces. Based on these projection surfaces, continuous projection surfaces are obtained by the robot moving parallel to the wall. These continuous projection surfaces are then synthesized to generate a multi-temporal projection plane. This multi-temporal projection plane is then projected onto a horizontal plane to form a horizontal projection plane.

[0155] By combining the line lidar and the top lidar, the projection forms a horizontal projection plane, which facilitates the calculation of the minimum circumscribed circle and the identification of obstacles on the wall, thus improving the accuracy of the robot vacuum cleaner in identifying obstacles on the wall.

[0156] Based on any of the above embodiments, in Embodiment Nine of this application, referring to Figure 10 , Figure 10 This is a flowchart of the control method for the wall recognition system of this application.

[0157] For example, the process begins by extracting the point cloud from the LiDAR on the right front of the robot vacuum. Polar coordinate filtering conditions are set (azimuth angle 60°-120°, distance 0.1-2.0m) to extract valid point cloud clusters on the right side. The filtered point cloud is then fitted with the RANSAC algorithm (Random Sample Consensus Algorithm) for line fitting (preset iteration count 500 times, interior point distance threshold 0.02m). After fitting, the point cloud is segmented into independent line segments according to line segment splitting rules (abrupt distance between adjacent points exceeding 0.1m or azimuth angle shift exceeding 3°). It is then determined whether there are valid line segments with a length ≥ 0.8m that extend across the robot's right geometric boundary (with the robot's center as the origin, right boundary X = +0.25m). If no valid line segment is detected, the current frame's wall detection result cache is cleared. If a valid line segment is detected, data points with a Z-axis height > 0.05m are extracted from the LiDAR point cloud in that area (filtering ground reflection interference). The XY plane projection points are then input into the Welzl algorithm (Welzl minimum enclosing circle algorithm) to calculate the minimum circumcircle. When the circle radius r ≤ 0.15m, the circle center coordinates (cx, cy) are added to the wall detection result queue; otherwise, the queue is cleared. The Euclidean distance between adjacent circle center coordinates in the queue is calculated over time. If the maximum distance between two consecutive points exceeds the threshold of 0.4m, a "Wall Detection Failed" command is issued and the queue data is cleared. If the threshold is not exceeded, a "Wall Detection Successful" command is issued, and all circle center coordinates and the corresponding wall point cloud set are output. At the end of the process, the current frame processing status (success / failure flag, circle center coordinate list) is written to a log file for offline verification of algorithm effectiveness.

[0158] This application provides a wall recognition device, which includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, which are executed by the at least one processor to enable the at least one processor to perform the control method of the wall recognition system in the first embodiment described above.

[0159] The following is for reference. Figure 11 This document illustrates a structural schematic diagram of a wall recognition device suitable for implementing embodiments of this application. The wall recognition device in these embodiments may include, but is not limited to, mobile terminals such as mobile phones, laptops, robotic vacuum cleaners, personal digital assistants (PDAs), tablet computers (PADs), portable media players (PMPs), linear LiDAR, etc., as well as fixed terminals such as intelligent robotic vacuum cleaners, desktop computers, etc. Figure 11 The wall recognition device shown is merely an example and should not be construed as limiting the functionality and scope of use of the embodiments of this application.

[0160] like Figure 11As shown, the wall recognition device may include a processing unit 1001 (e.g., a central processing unit, a graphics processing unit, etc.), which can perform various appropriate actions and processes according to a program stored in a read-only memory (ROM) 1002 or a program loaded from a storage device 1003 into a random access memory (RAM) 1004. The RAM 1004 also stores various programs and data required for the operation of the wall recognition device. The processing unit 1001, the read-only memory 1002, and the RAM 1004 are interconnected via a bus 1005. An input / output (I / O) interface 1006 is also connected to the bus. Typically, the following systems can be connected to I / O interface 1006: input devices 1007 including, for example, touchscreens, touchpads, keyboards, mice, image sensors, microphones, accelerometers, gyroscopes, etc.; output devices 1008 including, for example, liquid crystal displays (LCDs), speakers, vibrators, etc.; storage devices 1003 including, for example, magnetic tapes, hard disks, etc.; and communication devices 1009. Communication device 1009 allows the wall recognition device to communicate wirelessly or wiredly with other devices to exchange data. Although wall recognition devices with various systems are shown in the figures, it should be understood that it is not required to implement or possess all the systems shown. More or fewer systems can be implemented alternatively.

[0161] Specifically, according to the embodiments disclosed in this application, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via a communication device, or installed from storage device 1003, or installed from read-only memory 1002. When the computer program is executed by processing device 1001, it performs the functions defined in the methods of the embodiments disclosed in this application.

[0162] The wall recognition device provided in this application, employing the control method of the wall recognition system in the above embodiments, can solve the technical problem in related technologies where sweeping robots generally place wall recognition sensors near the drive wheel axis, resulting in a small point cloud acquisition area and consequently poor obstacle recognition accuracy. Compared with the prior art, the beneficial effects of the wall recognition device provided in this application are the same as those of the control method of the wall recognition system provided in the above embodiments, and other technical features of this wall recognition device are the same as those disclosed in the previous embodiment method, and will not be repeated here.

[0163] It should be understood that the various parts disclosed in this application can be implemented using hardware, software, firmware, or a combination thereof. In the description of the above embodiments, specific features, structures, materials, or characteristics can be combined in any suitable manner in one or more embodiments or examples.

[0164] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

[0165] This application provides a computer-readable storage medium having computer-readable program instructions (i.e., a computer program) stored thereon, the computer-readable program instructions being used to execute the control method of the wall recognition system in the above embodiments.

[0166] The computer-readable storage medium provided in this application may be, for example, a USB flash drive, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this embodiment, the computer-readable storage medium may be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, system, or device. The program code contained on the computer-readable storage medium may be transmitted using any suitable medium, including but not limited to: wires, optical cables, radio frequency (RF), etc., or any suitable combination thereof.

[0167] The aforementioned computer-readable storage medium may be included in the wall recognition device; or it may exist independently and not assembled into the wall recognition device.

[0168] The aforementioned computer-readable storage medium carries one or more programs. When the one or more programs are executed by the wall recognition device, the wall recognition device causes the wall recognition device to: determine the target point cloud data corresponding to the wall based on the wall boundary line in the point cloud data; determine the boundary point set of the target point cloud data on the horizontal projection plane; traverse all subsets of the boundary point set, determine the circumcircle corresponding to each subset, and obtain a sequence of circumcircles, wherein the subset contains a preset number of boundary points; determine the smallest circumcircle surrounding the boundary point set in the sequence of circumcircles; compare the radius of the smallest circumcircle with a preset radius threshold, and generate obstacle detection information based on the comparison result.

[0169] Computer program code for performing the operations of this application can be written in one or more programming languages ​​or a combination thereof, including object-oriented programming languages ​​such as Java, Smalltalk, and C++, as well as conventional procedural programming languages ​​such as the "C" language or similar programming languages. The program code can be executed entirely on the user's computer, partially on the user's computer, as a standalone software package, partially on the user's computer and partially on a remote computer, or entirely on a remote computer or server. In cases involving remote computers, the remote computer can be connected to the user's computer via any type of network—including a local area network (LAN) or a wide area network (WAN)—or can be connected to an external computer (e.g., via the Internet using an Internet service provider).

[0170] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of this application. In this regard, each block in a flowchart or block diagram may represent a module, segment, or portion of code containing one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those indicated in the drawings. For example, two consecutively indicated blocks may actually be executed substantially in parallel, and they may sometimes be executed in reverse order, depending on the functions involved. It should also be noted that each block in the block diagrams and / or flowcharts, and combinations of blocks in the block diagrams and / or flowcharts, can be implemented using a dedicated hardware-based system that performs the specified function or operation, or using a combination of dedicated hardware and computer instructions.

[0171] The modules described in the embodiments of this application can be implemented in software or hardware. The names of the modules do not necessarily limit the functionality of the unit itself.

[0172] The readable storage medium provided in this application is a computer-readable storage medium that stores computer-readable program instructions (i.e., a computer program) for executing the control method of the aforementioned wall recognition system. This addresses the technical problem in related technologies where robotic vacuum cleaners typically place wall recognition sensors near the drive wheel axis, resulting in a small point cloud acquisition area and consequently poor obstacle recognition accuracy. Compared to existing technologies, the beneficial effects of the computer-readable storage medium provided in this application are the same as those of the control method of the wall recognition system provided in the above embodiments, and will not be elaborated upon here.

[0173] The above description is only a part of the embodiments of this application and does not limit the patent scope of this application. All equivalent structural transformations made under the technical concept of this application and using the contents of the specification and drawings of this application, or direct / indirect applications in other related technical fields, are included in the patent protection scope of this application.

Claims

1. A control method for a wall recognition system, characterized in that, The method includes: The collected point cloud data is fitted with a straight line using a random sampling consensus algorithm. Multiple random straight lines are generated, and the line with the highest percentage of interior points is selected as the initial baseline. The portion of the initial baseline whose angle change exceeds a preset angle change threshold is divided into multiple sub-segments; Based on the initial baseline and the multiple sub-segments, the initial baseline is locally modified to generate the wall boundary line; Based on the wall boundary line, the point cloud data is divided into first point cloud data and second point cloud data; Based on the horizontal laser beam that is vertically irradiated onto the wall by the top lidar, the projection result of the horizontal laser beam is used to determine whether the projection result of the horizontal laser beam intersects with part of the point cloud data on one side of the straight line segment. The part of the point cloud data is identified as the target point cloud data, and the part of the point cloud data on the other side of the straight line segment is identified as the ground point cloud data. Determine the set of boundary points of the target point cloud data on the horizontal projection plane; Traverse all subsets of the boundary point set, determine the circumcircle corresponding to each subset, and obtain the circumcircle sequence, wherein the subset contains a preset number of boundary points; In the sequence of circumcircles, determine the smallest circumcircle that encloses the set of boundary points; Based on the radius corresponding to each of the minimum circumcircles, the radius is compared with the radius threshold to determine each target circumcircle with a radius smaller than the radius threshold; Based on the point cloud covered by the circumcircle of each target, coordinates are established on the horizontal projection plane. The coordinates of the center of the target's circumcircle on the horizontal projection plane are determined using a coordinate system constructed on the horizontal projection plane. Based on the generated center coordinates, and combined with the temporary obstacle judgment condition in the preset obstacle judgment condition, when the center coordinates are detected to be dynamically passing by, the center coordinates are filtered out as temporary obstacles.

2. The control method for the wall recognition system as described in claim 1, characterized in that, Following the step of filtering out the center coordinates as temporary obstacles when they are detected to be dynamically passing through the circle, based on the generated center coordinates and combined with the temporary obstacle judgment condition in the preset obstacle judgment condition, the method further includes: Based on the center coordinates of each circle in the obstacle detection information, the distance between any two adjacent center coordinates is calculated to obtain several center distances; The center distances of several circles are compared and filtered, and the center distance with the largest corresponding value is taken as the target distance; Based on the target distance, it is compared with a preset distance threshold, and the wall recognition result is determined according to the comparison result.

3. The control method for the wall recognition system as described in claim 2, characterized in that, The step of determining the wall recognition result based on the comparison result includes: If the comparison result shows that the target distance is less than the distance threshold, then the wall recognition result, including a wall detection success notification and the target point cloud data, is released to indicate that the wall detection conforms to the preset protection rules. If the comparison result shows that the target distance is greater than the distance threshold, the wall recognition result, including a wall detection failure notification, is issued to indicate that the wall does not meet the preset protection rules.

4. The control method for the wall recognition system as described in claim 1, characterized in that, Before the step of locally correcting the initial baseline based on the initial baseline and the multiple sub-line segments to generate the wall boundary line, the method further includes: Based on the vertical laser illumination from the line lidar and the horizontal laser illumination from the top lidar, the projection result corresponding to the line lidar and the horizontal projection result corresponding to the top lidar are generated. The projection result and the horizontal projection result are identified and judged. If both the projection result and the horizontal projection result are straight lines, then the sampling position in the projection result contains a wall.

5. The control method for the wall recognition system as described in claim 1, characterized in that, Before the step of determining the boundary point set of the target point cloud data on the horizontal projection plane, the method further includes: The projection surface is formed by combining the linear laser lines projected by the linear lidar with the horizontal projection results of the top lidar. Based on the projection surface, the robot is controlled to move in a direction parallel to the wall to obtain multiple time-series projection planes; The horizontal projection plane is generated by projecting onto the horizontal plane based on the multi-time projection plane.

6. A wall recognition device, characterized in that, The wall recognition device includes: a memory, a processor, and a computer program stored in the memory and executable on the processor, the computer program being configured to implement the steps of the control method for the wall recognition system as described in any one of claims 1 to 5.

7. A storage medium, characterized in that, The storage medium is a computer-readable storage medium, and a computer program is stored on the storage medium. When the computer program is executed by a processor, it implements the steps of the control method of the wall recognition system as described in any one of claims 1 to 5.

Images