Ceiling boundary acquisition system and method

Abstract

The invention relates to a ceiling boundary acquisition system and method.The ceiling is provided with a plurality of cross beams, the base plane of the ceiling is divided into a plurality of ceiling areas through the cross beams, and the acquisition method comprises the following steps that three-dimensional point cloud data of each ceiling area are collected; performing plane extraction on the acquired three-dimensional point cloud data to obtain a corresponding ceiling base plane; screening the ceiling base surface to obtain the required ceiling base surface; carrying out boundary extraction on a required ceiling basal plane to obtain a first intersection point set; and carrying out mathematical description on boundary points of the first intersection point set so as to complete acquisition of the ceiling boundary. According to the ceiling boundary acquisition system and method, the three-dimensional point cloud data of the ceiling area scanned by the laser radar scanner is extracted through the set calculation process, the boundary points of the ceiling area with the precision meeting the requirement can be obtained, and mathematical description of the boundary points of the ceiling area is completed; therefore, a reference can be provided for the construction of the mobile robot.

Description

Technical Field

[0001] The present invention relates to the technical field of information data processing, and in particular to a ceiling boundary acquisition system and method. Background Art

[0002] Mobile robots are widely used in existing construction projects. They can achieve automated construction results and save a lot of labor costs. However, when mobile robots are performing construction, they need to model the three-dimensional space with a certain degree of accuracy to form a navigation map and plan the construction tasks based on it.

[0003] Existing navigation maps are usually completed by manual modeling, which has certain limitations on height. For example, the dimension measurement of structures below 2 meters is easy to complete and can be used as the benchmark for laser SLAM navigation maps. However, it is difficult to measure the position and size of objects above 2 meters, especially those that lack ground reference benchmarks, such as I-beams, air ducts, etc.

[0004] The benchmark for laser SLAM navigation and positioning is the facade features in the environment, and the benchmark for ceiling construction planning is the position relationship between the ceiling area boundary and the global coordinate system. The inconsistency of the two benchmarks and the modeling error will affect the accuracy of ceiling construction, which will cause the mobile robot to collide with obstacles or fail to fully cover the construction and repeat the construction. Therefore, it is urgent to provide a new technical solution for ceiling boundary extraction. Summary of the invention

[0005] The purpose of the present invention is to overcome the defects of the prior art and provide a ceiling boundary acquisition system and method to solve the problem that the existing laser SLAM navigation map has large errors when acquiring ceiling boundaries due to inconsistent benchmarks.

[0006] The technical solution to achieve the above purpose is:

[0007] The present invention provides a ceiling boundary acquisition method, wherein the ceiling is provided with a plurality of beams, and the base surface of the ceiling is divided into a plurality of ceiling areas by the beams. The acquisition method comprises the following steps:

[0008] Collect 3D point cloud data of each sky area;

[0009] Perform plane extraction on the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface;

[0010] Screening the ceiling base surface to obtain the desired ceiling base surface;

[0011] Extract the boundary of the required ceiling base surface to obtain the first intersection point set;

[0012] The boundary points of the first intersection point set are mathematically described, thereby completing the acquisition of the ceiling boundary.

[0013] A further improvement of the ceiling boundary acquisition method of the present invention is that plane extraction is performed on the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface, which includes the following steps:

[0014] Performing point cloud segmentation on the collected three-dimensional point cloud data to obtain a number of planar point cloud clusters;

[0015] The plane whose normal vector of the planar point cloud cluster is parallel to the gravity line is taken as the first round of candidate ceiling regions;

[0016] The plane point cloud clusters in the first round of candidate ceiling regions are sorted in descending order according to the number of points, and the top three plane point cloud clusters with the highest number of points are selected as the second round of candidate ceiling regions;

[0017] The second round of candidate ceiling areas are arranged in ascending order according to the distance between the centroid of the plane point cloud cluster and the set position, and the plane point cloud cluster closest to the set position is selected as the ceiling base surface.

[0018] A further improvement of the ceiling boundary acquisition method of the present invention is that screening the ceiling base surface to obtain the required ceiling base surface comprises the following steps:

[0019] Segmenting the ceiling base surface by clustering to form a plurality of candidate point cloud clusters as the first round of candidate ceiling regions;

[0020] The first round of candidate ceiling regions are arranged in ascending order according to the distances of the centroids of the candidate point cloud clusters from the set positions, and three groups of candidate point cloud clusters closest to the set positions are selected as the second round of candidate ceiling regions;

[0021] The second round of candidate ceiling areas are arranged in descending order according to the number of points, and the candidate point cloud cluster with the largest number of points is selected as the required ceiling base surface.

[0022] A further improvement of the ceiling boundary acquisition method of the present invention is that the boundary extraction of the required ceiling base surface to obtain the first intersection point set includes the following steps:

[0023] Extract the boundary of the required ceiling base surface to obtain the ceiling boundary points;

[0024] Perform straight line fitting on the obtained ceiling boundary points and complete boundary fitting recursively;

[0025] Push the linear coefficients and the corresponding number of internal points obtained by fitting into the coefficient and point number structure;

[0026] Arrange the coefficient and point number structures in descending order according to the number of internal points;

[0027] The coefficient and point number structures of the previously set number of groups are retained, and any straight line in the retained coefficient and point number structures and other non-parallel straight lines are traversed to find the intersection points to obtain the first intersection point set.

[0028] A further improvement of the ceiling boundary acquisition method of the present invention is that mathematically describing the boundary points of the first intersection point set comprises the following steps:

[0029] Move any boundary point in the first intersection point set to the second intersection point set;

[0030] Selecting a boundary point closest to the boundary point moved to the second intersection point set from the remaining boundary points in the first intersection point set, and moving the selected boundary point to the second intersection point set, and repeating this process until all boundary points in the first intersection point set are moved to the second intersection point set;

[0031] Connecting the boundary points in the second intersection point set in sequence in a moving order to obtain a vector set including a plurality of vectors;

[0032] The second intersection point set and the connected set are used to complete the mathematical description of the ceiling area boundary.

[0033] The present invention also provides a ceiling boundary acquisition system, wherein the ceiling is provided with a plurality of beams, and the base surface of the ceiling is divided into a plurality of ceiling areas by the beams, and the acquisition system comprises:

[0034] A collection unit, used to collect three-dimensional point cloud data of each skylight area;

[0035] A plane extraction unit connected to the acquisition unit is used to perform plane extraction on the acquired three-dimensional point cloud data to obtain a corresponding ceiling base surface;

[0036] A screening unit connected to the plane extraction unit, used for screening the ceiling base surface to obtain a desired ceiling base surface;

[0037] A boundary extraction unit connected to the screening unit is used to extract the boundary of the required ceiling base surface to obtain a first intersection point set;

[0038] The mathematical description unit connected to the boundary extraction unit is used to mathematically describe the boundary points of the first intersection point set, thereby completing the acquisition of the ceiling boundary.

[0039] A further improvement of the ceiling boundary acquisition system of the present invention is that the plane extraction unit includes a point cloud segmentation module connected to the acquisition unit, a region division module connected to the point cloud segmentation module, and a base surface selection module connected to the region division module;

[0040] The point cloud segmentation module is used to perform point cloud segmentation on the collected three-dimensional point cloud data to obtain a plurality of planar point cloud clusters;

[0041] The region division module is used to use the plane whose normal vector of the plane point cloud cluster is parallel to the gravity line as the first round of candidate ceiling regions; and is also used to sort the plane point cloud clusters in the first round of candidate ceiling regions in descending order according to the number of points, and select the top three plane point cloud clusters in terms of the number of points as the second round of candidate ceiling regions;

[0042] The base surface selection module is used to arrange the second round of candidate ceiling areas in ascending order according to the distance between the center of mass of the plane point cloud cluster and the set position, and select the plane point cloud cluster closest to the set position as the ceiling base surface.

[0043] A further improvement of the ceiling boundary acquisition system of the present invention is that the screening unit includes a cluster segmentation module connected to the plane extraction unit and a screening selection module connected to the cluster segmentation module;

[0044] The cluster segmentation module is used to segment the ceiling base surface by clustering to form a plurality of candidate point cloud clusters as the first round of candidate ceiling regions;

[0045] The screening and selection module is used to arrange the first round of candidate ceiling areas in ascending order according to the distance of the centroid of the candidate point cloud cluster from the set position, and select the three groups of candidate point cloud clusters closest to the set position as the second round of candidate ceiling areas; it is also used to arrange the second round of candidate ceiling areas in descending order according to the number of points, and select the candidate point cloud cluster with the largest number of points as the required ceiling base surface.

[0046] A further improvement of the ceiling boundary acquisition system of the present invention is that the boundary extraction unit includes a boundary point extraction module, a straight line fitting module connected to the boundary point extraction module, and an intersection calculation module connected to the straight line fitting module;

[0047] The boundary point extraction module is used to extract the boundary of the required ceiling base surface to obtain the ceiling boundary points;

[0048] The straight line fitting module is used to perform straight line fitting on the obtained ceiling boundary points and recursively complete the boundary fitting;

[0049] The intersection calculation module is used to push the straight line coefficients and the corresponding number of internal points obtained by fitting into the coefficient and point number structure; it is also used to arrange the coefficient and point number structure in descending order according to the number of internal points; it is also used to retain a previously set number of groups of coefficient and point number structures, traverse any straight line in the retained coefficient and point number structure and other non-parallel straight lines and find the intersection points to obtain a first intersection point set.

[0050] A further improvement of the ceiling boundary acquisition system of the present invention is that the mathematical description unit includes a first description module connected to the boundary extraction unit and a second description module connected to the first description module;

[0051] The first description module is used to move any boundary point in the first intersection point set to the second intersection point set;

[0052] Also used for selecting a boundary point closest to the boundary point moved to the second intersection point set from the remaining boundary points in the first intersection point set, and moving the selected boundary point to the second intersection point set, and repeating this process until all boundary points in the first intersection point set are moved to the second intersection point set;

[0053] The second description unit is used to sequentially connect the boundary points in the second intersection point set in a moving order to obtain a vector set including a plurality of vectors.

[0054] The beneficial effects of the ceiling boundary acquisition system and method of the present invention are:

[0055] The ceiling boundary acquisition system and method of the present invention extracts the three-dimensional point cloud data of the ceiling area scanned by the laser radar scanner through a set calculation process, obtains the boundary points of the ceiling area with accuracy that meets the requirements, and completes the mathematical description of the boundary points of the ceiling area, so that it can provide a benchmark for the construction of the mobile robot.

[0056] The ceiling boundary acquisition system and method of the present invention reduce the requirements for manual modeling accuracy, and in particular can reduce the requirements for building accuracy when scenes need to be built.

[0057] The ceiling boundary acquisition system and method of the present invention greatly reduce the measurement accuracy requirements for the position and size of objects above 2 meters, especially those lacking a ground reference benchmark.

[0058] The ceiling boundary acquisition system and method of the present invention provide feasible technical support for fence-type scene construction, complete construction, and achieve one-time modeling and planning for multiple constructions, thereby spreading the cost of modeling and planning, which is a necessary non-value-added process.

[0059] The ceiling boundary acquisition system and method of the present invention reduce the communication cost between modelers and planners.

[0060] The ceiling boundary acquisition system and method of the present invention improve the construction coverage rate and avoid the problem of repeated construction. BRIEF DESCRIPTION OF THE DRAWINGS

[0061] Figure 1 This is a system diagram of the ceiling boundary acquisition system of the present invention.

[0062] Figure 2 The figure is a flow chart of the ceiling boundary acquisition method of the present invention.

[0063] Figure 3 This is an original image of a three-dimensional point cloud of an embodiment of the system and method for acquiring ceiling boundaries of the present invention.

[0064] Figure 4 for Figure 3 The point cloud image of the three-dimensional point cloud after filtering.

[0065] Figure 5 For Figure 4 Point cloud image of the base surface extracted from the 3D point cloud.

[0066] Figure 6 For Figure 5 The point cloud image of the ceiling area obtained by filtering the three-dimensional point cloud in .

[0067] Figure 7 For Figure 6 The ceiling boundary point cloud map is obtained by describing the boundary of the three-dimensional point cloud in . DETAILED DESCRIPTION

[0068] The present invention will be further described below in conjunction with the accompanying drawings and specific embodiments.

[0069] See also Figure 1 The present invention provides a ceiling boundary acquisition system and method, which aims to improve the automation and stability of mobile robots in semi-structured environments, reduce the operation requirements for operators, reduce the communication costs between planners and construction personnel, reduce the requirements for the on-site environment, and reduce non-value-added costs. The ceiling boundary acquisition system and method of the present invention are described below in conjunction with the accompanying drawings.

[0070] See also Figure 1 , showing the system diagram of the ceiling boundary acquisition system of the present invention. Figure 1 , the ceiling boundary acquisition system of the present invention is described.

[0071] like Figure 1 As shown, the ceiling boundary acquisition system of the present invention is used to acquire the boundary of the ceiling. The ceiling is provided with a plurality of beams, and the base surface of the ceiling is divided into a plurality of ceiling areas by the beams. The acquisition system of the present invention is used to acquire the boundary of each ceiling area.

[0072] The acquisition system of the present invention comprises an acquisition unit 21, a plane extraction unit 22, a screening unit 23, a boundary extraction unit 24 and a mathematical description unit 25, wherein the acquisition unit 21 is connected to the plane extraction unit 22, the plane extraction unit 22 is connected to the screening unit 23, the screening unit 23 is connected to the boundary extraction unit 24, and the boundary extraction unit 24 is connected to the mathematical description unit 25; wherein the acquisition unit 21 is used to acquire the three-dimensional point cloud data of each skylight area, and the acquired three-dimensional point cloud data is as follows Figure 3 The plane extraction unit 22 is used to perform plane extraction on the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface. The extracted ceiling base surface is as shown in FIG. Figure 5 The screening unit 23 is used to screen the ceiling base surface to obtain the required ceiling base surface, such as Figure 6 The boundary extraction unit 24 is used to extract the boundary of the required ceiling base surface to obtain the first intersection point set. The mathematical description unit 25 is used to mathematically describe the boundary points of the first intersection point set, thereby completing the acquisition of the ceiling boundary, such as Figure 7 shown.

[0073] In a specific embodiment of the present invention, the acquisition unit 21 is a laser radar scanner, which includes a protective cover, a rotating pan-tilt platform, a servo motor, a servo motor driver and a single-line laser radar. The single-line laser radar is installed on the rotating pan-tilt platform, and the protective cover is an openable cover arranged on the single-line laser radar. In the closed state, the single-line laser radar can be completely covered. The single-line laser radar can perform environmental measurements when it is opened.

[0074] When scanning the ceiling area of ​​the ceiling, the laser radar scanner is installed on the top of the construction robot, and the construction robot does not block the scanning area above the laser radar scanner. The construction robot is preferably a ceiling construction robot, and can also be other construction robots, such as a paint robot.

[0075] During the construction process of the construction robot, let the construction robot first temporarily set the corresponding construction action and move to the set position. The set position can be a position corresponding to the center position of each ceiling area. Then the laser radar scanner performs a position scan on each ceiling area. During the scanning, the protective cover is in the open state, the pan-tilt head is controlled to rotate and feedback the angle information, and the pan-tilt head angle and the point cloud data of the single-line laser radar scan are synchronously spliced ​​to obtain three-dimensional point cloud data.

[0076] The rated speed of the rotating gimbal is greater than 10° / s; the position control and feedback accuracy of the rotating gimbal is better than 0.1°; the rotating gimbal has a position control mode; the encoder of the rotating gimbal is an absolute encoder; the ranging range of the single-line laser radar is greater than 20m; the sampling frequency of the single-line laser radar is not less than 50hz; the ranging accuracy of the single-line laser radar is within the range of ±1cm; the visual angle range of the single-line laser radar is greater than 180°; the angular resolution of the single-line laser radar is better than 0.09°. The single-line laser radar is installed on the gimbal in a lying position, with the middle position of the radar visual angle pointing upward along the direction of gravity, and the laser ranging emission window of the radar coincides with the rotation center of the gimbal;

[0077] The construction phase of the construction robot is often interspersed with other construction processes, which makes it difficult to clear the site. This will cause two effects: one is that the positioning reference of the navigation is blocked, and the other is that the difficulty of obstacle avoidance is increased. In large scenes, there are also problems such as new wall surfaces that cause the model to be inconsistent with the physical environment. In the glass curtain wall scene of a public office, a metal anti-fall net will be placed in front of the glass curtain wall during the construction process. The laser radar point cloud generates a layer of point cloud at the metal anti-fall net and the glass curtain wall, and the data errors of these two layers of point clouds are large, resulting in positioning difficulties. The present invention installs a laser radar scanner on the top of the construction robot, and uses the construction robot to complete the scanning of the ceiling area during the construction phase, without the need to clear the site. For the problems caused by the metal anti-fall net and the glass curtain wall, a low-precision fence area can be artificially built as a navigation reference, and obstacles in the fence area can be cleared.

[0078] In another specific embodiment of the present invention, the acquisition unit 21 is communicatively connected to the laser radar scanner, and can obtain the three-dimensional point cloud data of each ceiling area scanned by the laser radar scanner. The specific structure and scanning method of the laser radar scanner are exactly the same as those of the above-mentioned laser radar scanner.

[0079] Preferably, the acquisition unit 21 includes a communication module, which is communicatively connected to the signal module in the laser radar scanner, and the communication module can receive the three-dimensional point cloud data sent by the signal module in the laser radar scanner. The connection between the communication module and the signal module can be a wireless connection, such as a 4G, 5G or other mobile communication connection.

[0080] In a specific embodiment of the present invention, the acquisition system of the present invention further includes a filtering unit, which is connected to the acquisition unit 21 and the plane extraction unit 22. The filtering unit is used to filter the three-dimensional point cloud data acquired by the acquisition unit 21. The point cloud data after filtering is as follows: Figure 4 As shown, the three-dimensional point cloud data after filtering is then subjected to plane extraction by the plane extraction unit 22 .

[0081] The filtering unit includes a NAN filtering module, a tail filtering module and a straight-through filtering module, wherein the NAN filtering module is used to filter out abnormal values ​​of points in the three-dimensional point cloud data collected by the acquisition unit 21, and the abnormal values ​​are non-numeric values; the tail filtering module is used to filter out abnormal light spots in the three-dimensional point cloud data. The laser point can be regarded as a light spot. When this light spot hits two adjacent surfaces, it will produce two echoes. These two echoes cannot reflect the actual distance, so they are called abnormal light spots, which need to be filtered out; the straight-through filtering module is used to retain data within the set limit according to the set limit, so as to reduce the amount of processed data. The set limit includes the upper and lower limits in the three directions of XYZ, and only retains data within a certain range in the three directions of XYZ, so as to reduce the amount of data for subsequent processing of point cloud data. For example, for the three-dimensional point cloud data of the ceiling area, the point cloud data of the area greater than 2m in the Z direction can be set.

[0082] The NAN filter module, the tail filter module and the straight-through filter module of the filter unit of the present invention are connected in sequence to filter the three-dimensional point cloud data in sequence, thereby obtaining the filtered three-dimensional point cloud data.

[0083] In a specific embodiment of the present invention, the plane extraction unit 22 of the present invention includes a point cloud segmentation module connected to the acquisition unit, a region division module connected to the point cloud segmentation module, and a base plane selection module connected to the region division module;

[0084] The point cloud segmentation module is used to perform point cloud segmentation on the collected three-dimensional point cloud data to obtain a number of planar point cloud clusters;

[0085] The region division module is used to take the plane whose normal vector of the plane point cloud cluster is parallel to the gravity line as the first round of candidate ceiling regions; it is also used to sort the plane point cloud clusters in the first round of candidate ceiling regions in descending order according to the number of points, and select the top three plane point cloud clusters in terms of the number of points as the second round of candidate ceiling regions;

[0086] The base surface selection module is used to sort the second round of candidate ceiling areas in ascending order according to the distance between the center of mass of the plane point cloud cluster and the set position, and select the plane point cloud cluster closest to the set position as the ceiling base surface.

[0087] Furthermore, after the filtering unit is set, the point cloud segmentation module is connected to the filtering unit and is used to perform point cloud segmentation on the filtered three-dimensional point cloud data to obtain a plurality of planar point cloud clusters.

[0088] Furthermore, the point cloud segmentation module of the plane extraction unit 22 of the present invention is used to extract the plane, remove the inner points (the inner points are points belonging to the plane), and then recursively extract the plane from the remaining points. The plane extraction method can use the ransac algorithm to extract the plane. The plane extracted by the point cloud segmentation module includes point cloud data that satisfies the mathematical description of the plane, and a set of points that satisfies a certain plane equation is a plane point cloud cluster.

[0089] The calculation principle of the area division module of the present invention is: based on the point cloud data in a structured environment, the plane of the ceiling area is a plane perpendicular to the Z direction, and the area of ​​the ceiling area is the largest. After plane fitting, the number of point cloud clusters is first arranged from large to small and the top three are taken, because the ceiling area we need is the area closest to the scanning position point, and the top three are taken according to the safety redundancy designed according to the distance sorting.

[0090] In a specific embodiment of the present invention, the screening unit 23 of the present invention includes a cluster segmentation module connected to the plane extraction unit 22 and a screening selection module connected to the cluster segmentation module;

[0091] The cluster segmentation module is used to segment the ceiling base surface by clustering to form multiple candidate point cloud clusters as the first round of candidate ceiling areas;

[0092] The screening and selection module is used to arrange the first round of candidate ceiling areas in ascending order according to the distance of the centroid of the candidate point cloud cluster from the set position, and select the three groups of candidate point cloud clusters closest to the set position as the second round of candidate ceiling areas; it is also used to arrange the second round of candidate ceiling areas in descending order according to the number of points, and select the candidate point cloud cluster with the largest number of points as the required ceiling base surface.

[0093] Furthermore, the clustering segmentation module segments the ceiling base surface through Euclidean clustering to form multiple candidate point cloud clusters as the first round of candidate ceiling areas.

[0094] The purpose of the Euclidean clustering in the present invention is to classify the point cloud data of the beam segmentation according to the distance. The principle is that points with a distance less than a certain threshold are considered to be a point cloud cluster (that is, what we call the ceiling area). According to the prior information, the required ceiling area is closest to the scanned position point, so it is sorted from small to large according to the centroid distance and the top three are taken. For safety redundancy (consider a situation where a circle of beams is also extracted, and its centroid is actually the same as the centroid of the ceiling area), the above point cloud clusters are sorted again from large to small according to the number and the point cloud cluster with the largest number of points is taken as the required ceiling area, and the point cloud data of the extracted beams are removed.

[0095] In a specific embodiment of the present invention, the boundary extraction unit 24 of the present invention includes a boundary point extraction module connected to the screening unit 23, a straight line fitting module connected to the boundary point extraction module, and an intersection calculation module connected to the straight line fitting module;

[0096] The boundary point extraction module is used to extract the boundary of the required ceiling base surface to obtain the ceiling boundary points;

[0097] The straight line fitting module is used to perform straight line fitting on the obtained ceiling boundary points and recursively complete the boundary fitting;

[0098] The intersection calculation module is used to push the straight line coefficients and the corresponding number of internal points obtained by fitting into the coefficient and point number structure; it is also used to arrange the coefficient and point number structure in descending order according to the number of internal points; it is also used to retain the previously set number of coefficient and point number structures, traverse any straight line in the retained coefficient and point number structure and other non-parallel straight lines and find the intersection points to obtain the first intersection point set.

[0099] Furthermore, the number of groups is set to n, and the intersection calculation module retains the first n groups of straight line coefficients in the structure according to the a priori number of ceiling area boundaries n, traverses the straight lines in the structure and finds intersections with other non-parallel straight lines in the structure, and presses n+1 intersections into the first intersection set to complete the extraction of ceiling boundary points.

[0100] The required ceiling area obtained by the screening unit 23 of the present invention is not its mathematical description, so it is necessary to perform boundary extraction through the boundary extraction unit 24. After the boundary points are extracted, RANSAC straight line fitting is performed to obtain the mathematical description of the straight line. According to the mathematical equation of the straight line, their intersection can be obtained and the mathematical description of the ceiling area is completed (taking a rectangle as an example, to describe a rectangle in space, it is necessary to know an intersection and two vectors starting from the intersection).

[0101] In a specific embodiment of the present invention, the mathematical description unit 25 of the present invention includes a first description module connected to the boundary extraction unit 24 and a second description module connected to the first description module;

[0102] The first description module is used to move any boundary point in the first intersection point set to the second intersection point set;

[0103] Also used for selecting a boundary point closest to the boundary point moved to the second intersection point set from the remaining boundary points in the first intersection point set, and moving the selected boundary point to the second intersection point set, and repeating this process until all boundary points in the first intersection point set are moved to the second intersection point set;

[0104] The second description unit is used to sequentially connect the boundary points in the second intersection point set in a moving order to obtain a vector set including a plurality of vectors.

[0105] Specifically, the mathematical description unit 25 of the present invention takes any boundary point as a starting point, moves the point from the first intersection point set A to the second intersection point set B, calculates and selects the remaining point in the first intersection point set A that is closest to the starting point as a new starting point and moves from the first intersection point set A to the second intersection point set B, repeats the above steps until all points in the first intersection point set A are pressed into the second intersection point set B, and all points in the second intersection point set B are sequentially connected according to the index to obtain a vector set C of n vectors;

[0106] The second intersection point set B and vector set C are used to complete the mathematical description of the boundary of the ceiling area.

[0107] The present invention also provides a ceiling boundary acquisition method, wherein a plurality of beams are provided on the ceiling, and the base surface of the ceiling is divided into a plurality of ceiling areas by the beams.

[0108] like Figure 2 As shown, the acquisition method of the present invention comprises the following steps:

[0109] Execute step S11 to collect three-dimensional point cloud data of each ceiling area at a set position; then execute step S12;

[0110] Execute step S12, perform plane extraction on the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface; then execute step S13;

[0111] Execute step S13 to screen the ceiling base surface to obtain the required ceiling base surface; then execute step S14;

[0112] Execute step S14, extract the boundary of the required ceiling base surface to obtain a first intersection point set; then execute step S15;

[0113] Execute step S15 to mathematically describe the boundary points of the first intersection point set, thereby completing the acquisition of the ceiling boundary.

[0114] In a specific implementation of the present invention, before performing plane extraction on the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface, the three-dimensional point cloud data is further subjected to filtering processing.

[0115] Furthermore, the three-dimensional point cloud data is processed by NAN filtering, tailing filtering and straight-through filtering. NAN filtering refers to filtering out the outliers of Not A Number in the point cloud data; tailing means that the laser point is a light spot, and this light spot will produce two echoes when it hits two adjacent surfaces, which does not reflect the actual distance. These abnormal points must be eliminated through algorithms; straight-through filtering takes the upper and lower limits according to the xyz directions, and only retains the data between the upper and lower limits, in order to reduce the amount of data for subsequent processing.

[0116] Furthermore, performing plane extraction on the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface includes the following steps:

[0117] Performing point cloud segmentation on the collected three-dimensional point cloud data to obtain a number of planar point cloud clusters;

[0118] The plane whose normal vector of the planar point cloud cluster is parallel to the gravity line is taken as the first round of candidate ceiling regions;

[0119] The plane point cloud clusters in the first round of candidate ceiling regions are sorted in descending order according to the number of points, and the top three plane point cloud clusters with the highest number of points are selected as the second round of candidate ceiling regions;

[0120] The second round of candidate ceiling areas are arranged in ascending order according to the distance between the centroid of the plane point cloud cluster and the set position, and the plane point cloud cluster closest to the set position is selected as the ceiling base surface.

[0121] Specifically, base surface extraction is divided into three parts: point cloud segmentation, point cloud cluster point cloud quantity sorting and point cloud cluster centroid distance sorting. The point cloud segmentation part extracts the plane and removes the inner points (the inner points are points belonging to the plane), and then recursively extracts the plane from the remaining points (ransac algorithm is used for plane extraction). The plane whose extracted normal vector is parallel to the gravity line is used as a round of candidate ceiling areas. For a round of candidate ceiling areas, the point cloud clusters (the plane extracted above refers to the point cloud data that satisfies the mathematical description of the plane, and a set of points that satisfies a certain plane equation is a point cloud cluster) are first sorted in descending order according to the number of points, and the point cloud clusters with the top three points are taken as the second round of candidate ceiling areas. For the second round of candidate ceiling areas, the point cloud clusters are sorted in ascending order according to the distance from the centroid of the laser radar scanner, and the point cloud cluster closest to the centroid of the laser radar scanner is the ceiling base surface we need.

[0122] In a specific embodiment of the present invention, screening the ceiling base surface to obtain the desired ceiling base surface comprises the following steps:

[0123] Segmenting the ceiling base surface by clustering to form a plurality of candidate point cloud clusters as the first round of candidate ceiling regions;

[0124] The first round of candidate ceiling regions are arranged in ascending order according to the distances of the centroids of the candidate point cloud clusters from the set positions, and three groups of candidate point cloud clusters closest to the set positions are selected as the second round of candidate ceiling regions;

[0125] The second round of candidate ceiling areas are arranged in descending order according to the number of points, and the candidate point cloud cluster with the largest number of points is selected as the required ceiling base surface.

[0126] Furthermore, the ceiling base surface is segmented by Euclidean clustering to form a plurality of candidate point cloud clusters as the first round of candidate ceiling regions.

[0127] In a specific implementation of the present invention, performing boundary extraction on the required ceiling base surface to obtain a first intersection point set includes the following steps:

[0128] Extract the boundary of the required ceiling base surface to obtain the ceiling boundary points;

[0129] Perform straight line fitting on the obtained ceiling boundary points and complete boundary fitting recursively;

[0130] Push the linear coefficients and the corresponding number of internal points obtained by fitting into the coefficient and point number structure;

[0131] Arrange the coefficient and point number structures in descending order according to the number of internal points;

[0132] The coefficient and point number structures of the previously set number of groups are retained, and any straight line in the retained coefficient and point number structures and other non-parallel straight lines are traversed to find the intersection points to obtain the first intersection point set.

[0133] Specifically, the boundary of the ceiling area point cloud obtained in the above steps is extracted, and then the obtained ceiling boundary point cloud is fitted with a straight line, and the boundary fitting is completed recursively. The fitted straight line coefficients and the corresponding number of internal points are pressed into the [coefficient & number of points] structure, and the structure is sorted in descending order according to the number of internal points. The first n groups of straight line coefficients in the structure are retained according to the a priori number of ceiling area boundaries n, and the straight lines in the structure are traversed and the intersection points with other non-parallel straight lines in the structure are found, and the n+1 intersection points are pressed into the intersection set A to complete the ceiling boundary point extraction.

[0134] In a specific implementation of the present invention, mathematically describing the boundary points of the first intersection point set includes the following steps:

[0135] Move any boundary point in the first intersection point set to the second intersection point set;

[0136] Selecting a boundary point closest to the boundary point moved to the second intersection point set from the remaining boundary points in the first intersection point set, and moving the selected boundary point to the second intersection point set, and repeating this process until all boundary points in the first intersection point set are moved to the second intersection point set;

[0137] Connecting the boundary points in the second intersection point set in sequence in a moving order to obtain a vector set including a plurality of vectors;

[0138] The second intersection point set and the connected set are used to complete the mathematical description of the ceiling area boundary.

[0139] Specifically, take any boundary point as the starting point, move the point from set A to set B, calculate and select the remaining point in A that is closest to the starting point as the new starting point and move from set A to set B, repeat the above steps until all points in set A are pressed into set B, connect all points in set B in sequence according to the index to obtain a set C of n vectors; use the intersection set B and the vector set C to complete the mathematical description of the boundary of the ceiling area.

[0140] The present invention is described in detail above in conjunction with the embodiments of the accompanying drawings. A person skilled in the art can make various variations of the present invention according to the above description. Therefore, certain details in the embodiments should not constitute a limitation of the present invention, and the scope of protection of the present invention shall be defined by the scope of the attached claims.

Claims

1. A method for obtaining a ceiling boundary, wherein a plurality of beams are provided on the ceiling, and a base surface of the ceiling is divided into a plurality of ceiling areas by the beams, characterized in that: The acquisition method comprises the following steps: Collect 3D point cloud data of each sky area; Perform plane extraction on the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface; Screening the ceiling base surface to obtain the desired ceiling base surface; Extract the boundary of the required ceiling base surface to obtain the first intersection point set; The boundary points of the first intersection point set are mathematically described, thereby completing the acquisition of the ceiling boundary.

2. The ceiling boundary acquisition method according to claim 1, characterized in that: The plane extraction of the collected three-dimensional point cloud data to obtain the corresponding ceiling base surface includes the following steps: Performing point cloud segmentation on the collected three-dimensional point cloud data to obtain a number of planar point cloud clusters; The plane whose normal vector of the planar point cloud cluster is parallel to the gravity line is taken as the first round of candidate ceiling regions; The plane point cloud clusters in the first round of candidate ceiling regions are sorted in descending order according to the number of points, and the top three plane point cloud clusters with the highest number of points are selected as the second round of candidate ceiling regions; The second round of candidate ceiling areas are arranged in ascending order according to the distance between the centroid of the plane point cloud cluster and the set position, and the plane point cloud cluster closest to the set position is selected as the ceiling base surface.

3. The ceiling boundary acquisition method according to claim 1, characterized in that: Screening the ceiling base surface to obtain the desired ceiling base surface includes the following steps: Segmenting the ceiling base surface by clustering to form a plurality of candidate point cloud clusters as the first round of candidate ceiling regions; The first round of candidate ceiling regions are arranged in ascending order according to the distances of the centroids of the candidate point cloud clusters from the set positions, and three groups of candidate point cloud clusters closest to the set positions are selected as the second round of candidate ceiling regions; The second round of candidate ceiling areas are arranged in descending order according to the number of points, and the candidate point cloud cluster with the largest number of points is selected as the required ceiling base surface.

4. The ceiling boundary acquisition method according to claim 1, characterized in that: Extracting the boundary of the required ceiling base surface to obtain the first intersection point set includes the following steps: Extract the boundary of the required ceiling base surface to obtain the ceiling boundary points; Perform straight line fitting on the obtained ceiling boundary points and complete boundary fitting recursively; Push the linear coefficients and the corresponding number of internal points obtained by fitting into the coefficient and point number structure; Arrange the coefficient and point number structures in descending order according to the number of internal points; The coefficient and point number structures of the previously set number of groups are retained, and any straight line in the retained coefficient and point number structures and other non-parallel straight lines are traversed to find the intersection points to obtain the first intersection point set.

5. The ceiling boundary acquisition method according to claim 1, characterized in that: Mathematically describing the boundary points of the first intersection point set includes the following steps: Move any boundary point in the first intersection point set to the second intersection point set; Selecting a boundary point closest to the boundary point moved to the second intersection point set from the remaining boundary points in the first intersection point set, and moving the selected boundary point to the second intersection point set, and repeating this process until all boundary points in the first intersection point set are moved to the second intersection point set; Connecting the boundary points in the second intersection point set in sequence in a moving order to obtain a vector set including a plurality of vectors; The second intersection point set and the connected set are used to complete the mathematical description of the ceiling area boundary.

6. A ceiling boundary acquisition system, wherein the ceiling is provided with a plurality of beams, and the base surface of the ceiling is divided into a plurality of ceiling areas by the beams, characterized in that: The acquisition system comprises: A collection unit, used to collect three-dimensional point cloud data of each skylight area; A plane extraction unit connected to the acquisition unit is used to perform plane extraction on the acquired three-dimensional point cloud data to obtain a corresponding ceiling base surface; A screening unit connected to the plane extraction unit, used for screening the ceiling base surface to obtain a desired ceiling base surface; A boundary extraction unit connected to the screening unit is used to extract the boundary of the required ceiling base surface to obtain a first intersection point set; The mathematical description unit connected to the boundary extraction unit is used to mathematically describe the boundary points of the first intersection point set, thereby completing the acquisition of the ceiling boundary.

7. The ceiling boundary acquisition system according to claim 6, characterized in that: The plane extraction unit includes a point cloud segmentation module connected to the acquisition unit, a region division module connected to the point cloud segmentation module, and a base plane selection module connected to the region division module; The point cloud segmentation module is used to perform point cloud segmentation on the collected three-dimensional point cloud data to obtain a plurality of planar point cloud clusters; The region division module is used to use the plane whose normal vector of the plane point cloud cluster is parallel to the gravity line as the first round of candidate ceiling regions; and is also used to sort the plane point cloud clusters in the first round of candidate ceiling regions in descending order according to the number of points, and select the top three plane point cloud clusters in terms of the number of points as the second round of candidate ceiling regions; The base surface selection module is used to arrange the second round of candidate ceiling areas in ascending order according to the distance between the center of mass of the plane point cloud cluster and the set position, and select the plane point cloud cluster closest to the set position as the ceiling base surface.

8. The ceiling boundary acquisition system according to claim 6, characterized in that: The screening unit includes a cluster segmentation module connected to the plane extraction unit and a screening selection module connected to the cluster segmentation module; The cluster segmentation module is used to segment the ceiling base surface by clustering to form a plurality of candidate point cloud clusters as the first round of candidate ceiling regions; The screening and selection module is used to arrange the first round of candidate ceiling areas in ascending order according to the distance of the centroid of the candidate point cloud cluster from the set position, and select the three groups of candidate point cloud clusters closest to the set position as the second round of candidate ceiling areas; it is also used to arrange the second round of candidate ceiling areas in descending order according to the number of points, and select the candidate point cloud cluster with the largest number of points as the required ceiling base surface.

9. The ceiling boundary acquisition system according to claim 6, characterized in that: The boundary extraction unit includes a boundary point extraction module, a straight line fitting module connected to the boundary point extraction module, and an intersection calculation module connected to the straight line fitting module; The boundary point extraction module is used to extract the boundary of the required ceiling base surface to obtain the ceiling boundary points; The straight line fitting module is used to perform straight line fitting on the obtained ceiling boundary points and recursively complete the boundary fitting; The intersection calculation module is used to push the straight line coefficients and the corresponding number of internal points obtained by fitting into the coefficient and point number structure; it is also used to arrange the coefficient and point number structure in descending order according to the number of internal points; it is also used to retain a previously set number of groups of coefficient and point number structures, traverse any straight line in the retained coefficient and point number structure and other non-parallel straight lines and find the intersection points to obtain a first intersection point set.

10. The ceiling boundary acquisition system according to claim 6, characterized in that: The mathematical description unit includes a first description module connected to the boundary extraction unit and a second description module connected to the first description module; The first description module is used to move any boundary point in the first intersection point set to the second intersection point set; Also used for selecting a boundary point closest to the boundary point moved to the second intersection point set from the remaining boundary points in the first intersection point set, and moving the selected boundary point to the second intersection point set, and repeating this process until all boundary points in the first intersection point set are moved to the second intersection point set; The second description unit is used to sequentially connect the boundary points in the second intersection point set in a moving order to obtain a vector set including a plurality of vectors.

Images