A method for processing a garage point cloud

Abstract

This application relates to the field of point cloud data processing, and in particular to a method for processing point clouds in a parking garage, comprising the following steps: S100, acquiring the point cloud of a target area in the parking garage; S200, segmenting the point cloud of the target area in the parking garage to obtain a set of point cloud clusters A; S300, traversing A and extracting A n The straight line segment in the middle, we get A. n The corresponding set of line segments L n S400, obtain A n The first parameter s1; S500, if |b m,1 -b m,2 If |≤b′, then obtain l max,1 and l max,2 S600, obtain the first target line; if A max,1 If each point is located to the right of the mobile charging robot's direction of movement, then the first target straight line is to bring l max,1 The straight line obtained by shifting d / 2 to the left. This invention solves the problem that existing mobile charging robots, relying on vision, cannot recognize the dividing line between two lanes in a garage.

Description

Technical Field

[0001] This invention relates to the field of point cloud data processing, and in particular to a method for processing point clouds of a parking garage. Background Technology

[0002] In a parking garage with a two-lane main driveway, there is typically a dividing line in the middle to separate the opposing lanes. A mobile charging robot can use this dividing line to distinguish the opposing lanes, providing a reference for mapping and path planning. Currently, mobile charging robots generally identify the dividing line in the middle of the two lanes visually. However, when this dividing line is covered by dirt or scratched, the mobile charging robot will be unable to visually identify it. Summary of the Invention

[0003] The purpose of this invention is to provide a method for processing point clouds in a parking garage, which can acquire the dividing line between two lanes in a parking garage when it is covered by dirt or scratched.

[0004] According to the present invention, a method for processing garage point clouds is provided, comprising the following steps:

[0005] S100, acquire the point cloud of the target area in the garage.

[0006] S200, the point cloud of the target area in the garage is segmented to obtain a point cloud cluster set A = {A1, A2, ..., A...} N}, A n The nth point cloud cluster is obtained by segmenting the point cloud of the target area in the garage. The value of n ranges from 1 to N, and N is the number of point cloud clusters obtained by segmenting the point cloud of the target area in the garage.

[0007] S300, iterate through A, extract A n The straight line segment in the middle, we get A. n The corresponding set of line segments L n = (l1, l2, ..., l M ), l m To extract A n The m-th line segment in the equation, where m ranges from 1 to M, and M is the extracted value of A. n The number of straight line segments in the equation.

[0008] S400, iterate through A and get A n The first parameter s1 = w1*s′2 + w2*(1-s′3), where w1 and w2 are the first and second weights, respectively, and s′2 is the weight of A. n The normalized value of the second parameter s2. c j For l jThe length of l j To extract A n The j-th line segment in the diagram, s′3 is the line segment corresponding to A. n The normalized value of the third parameter s3. b j For l j The angle of inclination of the line in question. To extract A n The mean of the inclination angles of the lines containing all line segments.

[0009] S500, if Then A max,1 The midpoint is fitted to a straight line segment l max,1 , will A max,2 The midpoint is fitted to a straight line segment l max,2 ;in, To extract A max,1 The mean angle of inclination of the lines containing all line segments in the equation, A max,1 For the point cloud cluster with the largest s1 in A, To extract A max,2 The mean angle of inclination of the lines containing all line segments in the equation, A max,2 Let b' be the point cloud cluster with the largest value of s1 in A, and b' be the preset angle threshold.

[0010] S600, obtain the first target line; if A max,1 If each point is located to the right of the mobile charging robot's direction of movement, then the first target straight line is to bring l max,1 The straight line obtained by shifting it to the left by d / 2, where d is l max,1 and l max,2 The distance between them; if A max,2 If each point is located to the right of the mobile charging robot's direction of movement, then the first target straight line is to bring l max,2 The straight line obtained by shifting it to the left by d / 2.

[0011] Compared with the prior art, the present invention has significant advantages. Through the above technical solution, the garage point cloud processing method provided by the present invention can achieve considerable technical progress and practicality, and has broad industrial application value. It has at least the following advantages:

[0012] This invention acquires a point cloud of a target area in a garage and segments it to obtain multiple point cloud clusters. The two point cloud clusters with the largest corresponding first parameter are determined by the length of the line segment corresponding to each point cloud cluster and the inclination angle of that line segment. If these two point cloud clusters meet preset conditions, they are determined to be the point clouds of the two walls acquired by the mobile charging robot while it is moving in the target area of ​​the garage. Furthermore, based on these two point cloud clusters, the first target straight line, i.e., the dividing line between two lanes, is obtained. This invention solves the problem that existing mobile charging robots cannot recognize the dividing line between two lanes in a garage using vision, providing a reference for the mapping and path planning of mobile charging robots. Attached Figure Description

[0013] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0014] Figure 1 A flowchart illustrating a method for processing garage point clouds provided in an embodiment of the present invention. Detailed Implementation

[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0016] According to the present invention, such as Figure 1 As shown, the methods for processing garage point clouds include:

[0017] S100, acquire the point cloud of the target area in the garage.

[0018] Optionally, a point cloud of the target area in the garage can be acquired using a multi-line lidar system. Those skilled in the art will understand that any existing method for acquiring point clouds falls within the protection scope of this invention.

[0019] This invention is applicable to scenarios where the walls on both sides of the main driveway in a garage are symmetrical about the dividing line in the middle of the main driveway, and the main driveway is a two-lane roadway; the target area is the area scanned by LiDAR when the mobile charging robot travels on the main driveway. The point cloud of the target area acquired in S100 of this invention is a two-dimensional point cloud, excluding height information; it should be understood that in the two-dimensional point cloud, the walls are straight line segments, the rectangular support columns in the garage are rectangles, and the irregularly shaped obstacles on the driveway are irregular polygons.

[0020] S200, the point cloud of the target area in the garage is segmented to obtain a point cloud cluster set A = {A1, A2, ..., A...} N}, A n The nth point cloud cluster is obtained by segmenting the point cloud of the target area in the garage. The value of n ranges from 1 to N, and N is the number of point cloud clusters obtained by segmenting the point cloud of the target area in the garage.

[0021] Optionally, a distance-based point cloud segmentation algorithm, such as an Euclidean distance-based segmentation algorithm, can be used to segment the point cloud of the target area in the garage. According to the present invention, after segmenting the point cloud of the target area, N point cloud clusters are obtained. It should be understood that different point cloud clusters correspond to relatively independent objects in a two-dimensional plane (i.e., a two-dimensional plane constructed by the x-axis and y-axis, excluding the z-axis). As an example, N=4, and the four point cloud clusters obtained after segmenting the point cloud of the target area correspond to the point cloud on the left wall in the direction of travel of the mobile charging robot, the point cloud on the right wall in the direction of travel of the mobile charging robot, the point cloud corresponding to a rectangular support column in the target area, and the point cloud of irregular obstacles on the road where the mobile charging robot is located.

[0022] S300, iterate through A, extract A n The straight line segment in the middle, we get A. n The corresponding set of line segments L n = (l1, l2, ..., l M ), l m To extract A n The m-th line segment in the equation, where m ranges from 1 to M, and M is the extracted value of A. n The number of straight line segments in the equation.

[0023] According to the present invention, A is extracted using a multi-line fitting extraction method. n The straight line segment in the middle can be used to obtain L. n Any existing method for extracting multiple lines by fitting them together falls within the scope of protection of this invention. The process of extracting multiple lines from a point cloud is existing technology and will not be described in detail here.

[0024] S400, iterate through A and get A n The first parameter s1 = w1*s′2 + w2*(1-s′3), where w1 and w2 are the first and second weights, respectively, and s′2 is the weight of A. n The normalized value of the second parameter s2. c j For l j The length of l j To extract A n The j-th line segment in the diagram, s′3 is the line segment corresponding to A. n The normalized value of the third parameter s3. b j For l j The angle of inclination of the line in question. To extract A n The mean of the inclination angles of the lines containing all line segments.

[0025] According to the present invention, compared with the point cloud clusters corresponding to rectangular support columns and obstacles in the garage, the line segments in the set of line segments corresponding to the point cloud clusters of the wall are longer and the inclination angles of the lines on which the line segments are located are more consistent. Therefore, the present invention obtains a first parameter based on the above-mentioned second and third parameters to distinguish the point cloud clusters corresponding to the wall from other types of point cloud clusters (e.g., the point cloud clusters corresponding to rectangular support columns and the point cloud clusters corresponding to irregular obstacles).

[0026] Those skilled in the art will understand that any normalization method in the prior art falls within the protection scope of this invention.

[0027] According to the present invention, w1 + w2 = 1, where w1 and w2 are both constants greater than 0 and less than 1. Optionally, w1 = w2 = 0.5. It should be understood that if the length of the reference line segment is a significant factor when distinguishing point cloud cluster types, then w1 > w2; if the tilt angle of the line containing the reference line segment is a significant factor when distinguishing point cloud cluster types, then w1 > w2. <w2。

[0028] S500, if Then A max,1 The midpoint is fitted to a straight line segment l max,1 , will A max,2 The midpoint is fitted to a straight line segment l max,2 ;in, To extract A max,1 The mean angle of inclination of the lines containing all line segments in the equation, A max,1 For the point cloud cluster with the largest s1 in A, To extract A max,2 The mean angle of inclination of the lines containing all line segments in the equation, A max,2Let b' be the point cloud cluster with the largest value of s1 in A, and b' be the preset angle threshold.

[0029] According to the present invention, A max,1 and A max,2 The two point cloud clusters in A with the largest corresponding first parameter are considered. According to the present invention, when the point cloud cluster corresponds to the wall surface, the value of the corresponding first parameter is the largest; the present invention further... and The magnitude of the difference between them is used as the basis for the above judgment A. max,1 and A max,2 The verification condition of whether it is a point cloud cluster corresponding to two walls in the target area improves the accuracy of obtaining the first target line in this invention. According to this invention, if and If the difference between them is greater than b′, then repeat S100-S400, where b′ is a certain angle value slightly greater than 0.

[0030] Optionally, A can be solved using the least squares method. max,1 The midpoint is fitted to a straight line segment l max,1 And A max,2 The midpoint is fitted to a straight line segment l max,2 The fitting process is an existing technique and will not be described in detail here.

[0031] S600, obtain the first target line; if A max,1 If each point is located to the right of the mobile charging robot's direction of movement, then the first target straight line is to bring l max,1 The straight line obtained by shifting it to the left by d / 2, where d is l max,1 and l max,2 The distance between them; if A max,2 If each point is located to the right of the mobile charging robot's direction of movement, then the first target straight line is to bring l max,2 The straight line obtained by shifting it to the left by d / 2.

[0032] According to the present invention, the distance d in the present invention is l in the point cloud coordinate system. max,1 and l max,2 The distance between them is determined by the first target straight line, which is the line corresponding to the dividing line in the middle of the two lanes in the point cloud. Based on the position of the first straight line in the point cloud and the transformation relationship between the point cloud coordinate system and the world coordinate system, the coordinates of the dividing line in the middle of the two lanes in the world coordinate system can be obtained. The methods for constructing the point cloud coordinate system and the world coordinate system, as well as the methods for obtaining the transformation relationship between the two coordinate systems, are existing technologies and will not be elaborated here.

[0033] Thus, this invention solves the problem that existing mobile charging robots cannot recognize the dividing line between two lanes in a garage by relying on vision, and provides a reference for the mapping and path planning of mobile charging robots.

[0034] As a preferred embodiment, the garage point cloud processing method of the present invention further includes the following steps:

[0035] S700, for A excluding A max,1 and A max,2 For any point cloud cluster A' other than A': obtain the set B of the inclination angles of the lines containing each line segment in the set of line segments corresponding to A'.

[0036] S800, cluster B according to the difference between any two tilt angles in B to obtain the first tilt angle cluster B1 and the second tilt angle cluster B2.

[0037] Optionally, the k-means clustering algorithm is used to cluster B, where k=2. This invention uses the k-means clustering algorithm to cluster B, including:

[0038] S810, randomly select two tilt angles from B as the centroids.

[0039] It should be noted that each centroid in this invention is a tilt angle.

[0040] S820, for each tilt angle in B, obtain its distance to each centroid and assign it to the set of the nearest centroid; where the p-th tilt angle Bb in B p With the qth tilt angle Bb q distance d p,q =|b p -b q | p and q both range from 1 to Q B Q B This represents the number of tilt angles in B.

[0041] S830, after dividing B into each tilt angle, re-obtain the centroids of the two sets.

[0042] According to the present invention, the centroid of the two re-obtained sets is the average tilt angle of all tilt angles in each of the two sets.

[0043] S840, repeat steps S820-S830 until the obtained centroids no longer change or the number of repetitions reaches the set number, then the clustering ends.

[0044] It should be understood that after clustering, two sets will be obtained, namely B1 and B2.

[0045] S900, obtain the mean eB1 of all tilt angles in B1 and the mean eB2 of all tilt angles in B2; if Then A' is determined to be the point cloud cluster corresponding to the rectangular support column.

[0046] According to the present invention, when A' is the point cloud cluster corresponding to the rectangular support column, the tilt angles in B are mainly concentrated near two tilt angles, and the angle difference between these two tilt angles is π / 2. Therefore, the present invention can identify the rectangular support column in the garage based on the relationship between the mean eB1 of all tilt angles in B1 and the mean eB2 of all tilt angles in B2.

[0047] It should be understood that the positions of the support columns in the garage are fixed and their number is relatively small. Based on the positions of the support columns in the garage, the mobile charging robot can quickly locate its own position. Therefore, after identifying the point cloud clusters corresponding to the rectangular support columns, this invention also uses the rectangular support column corresponding to A' as a landmark for subsequent relocalization of the mobile charging robot.

[0048] Generally, the shapes of objects in a garage's architectural design are regular, such as the supporting columns, which are mostly rectangular and a few are circular. However, the shapes of non-architectural objects in a garage are irregular, such as people and vehicles. Given the typical architectural design characteristics of garages, in this invention, regular shapes can be understood as straight lines, rectangles, circles, ellipses, and regular polygons, while irregular shapes can be understood as shapes other than the aforementioned regular shapes.

[0049] Furthermore, the shapes of the objects in the architectural design of the garage can be obtained in advance; that is, the shapes of the objects in the current architectural design of the garage are known, and the number of shapes is finite. Therefore, in this invention, if A' does not match any of the aforementioned known regular shapes, then A' is determined to be a point cloud cluster corresponding to an irregular obstacle. In the first embodiment, the shapes of the objects in the architectural design of the garage only include straight lines and rectangles. Therefore, if A' is not a point cloud cluster corresponding to straight lines and rectangles, then A' can be determined to be a point cloud cluster corresponding to an irregular obstacle. In the second embodiment, the shapes of the objects in the architectural design of the garage only include straight lines, rectangles, and circles. Therefore, if A' is not a point cloud cluster corresponding to straight lines, rectangles, and circles, then A' can be determined to be a point cloud cluster corresponding to an irregular obstacle.

[0050] According to the present invention, if A' is determined to be a point cloud cluster corresponding to an irregular obstacle, the following steps are performed:

[0051] S910, obtain A h Point P1 is the closest point to the mobile charging robot, and point P2 is the farthest point from the mobile charging robot.

[0052] S920: Use the line connecting P1 and P2 as the diameter of the circumcircle of the irregular obstacle to obtain the circumcircle of the irregular obstacle.

[0053] S930, the circumcircle of the irregular obstacle is used as the size of the irregular obstacle for obstacle avoidance.

[0054] It should be noted that once the size of the obstacle is determined, the process of obstacle avoidance is existing technology and will not be elaborated here.

[0055] This invention, based on S700-S900, identifies rectangular support columns in a garage and uses these columns as landmarks, providing important reference for the subsequent positioning of the mobile charging robot. Based on S910-S930, this invention effectively avoids irregular obstacles, ensuring the safety of the mobile charging robot during its operation.

[0056] While specific embodiments of the invention have been described in detail by way of example, those skilled in the art should understand that the examples are for illustrative purposes only and not intended to limit the scope of the invention. It should also be understood that various modifications can be made to the embodiments without departing from the scope and spirit of the invention. The scope of the invention is defined by the appended claims.

Claims

1. A method for processing garage point clouds, characterized in that, Includes the following steps: S100, acquire the point cloud of the target area in the garage; S200, the point cloud of the target area in the garage is segmented to obtain a point cloud cluster set A = {A1, A2, ..., A...} N }, A n The nth point cloud cluster is obtained by segmenting the point cloud of the target area in the garage. The value of n ranges from 1 to N, and N is the number of point cloud clusters obtained by segmenting the point cloud of the target area in the garage. S300, iterate through A, extract A n The straight line segment in the middle, we get A. n The corresponding set of line segments L n = (l1, l2, ..., l M ), l m To extract A n The m-th line segment in the equation, where m ranges from 1 to M, and M is the extracted value of A. n The number of straight line segments in the text; S400, iterate through A and get A n The first parameter s1 = w1*s′2 + w2*(1-s′3), where w1 and w2 are the first and second weights, respectively, and s′2 is the weight of A. n The normalized value of the second parameter s2. c j For l j The length of l j To extract A n The j-th line segment in the diagram, s′3 is the line segment corresponding to A. n The normalized value of the third parameter s3. b j For l j The angle of inclination of the line in question. To extract A n The mean of the angles of inclination of the lines containing all line segments in the equation; S500, if Then A max,1 The midpoint is fitted to a straight line segment l max,1 , will A max,2 The midpoint is fitted to a straight line segment l max,2 ;in, To extract A max,1 The mean angle of inclination of the lines containing all line segments in the equation, A max,1 For the point cloud cluster with the largest s1 in A, To extract A max,2 The mean angle of inclination of the lines containing all line segments in the equation, A max,2 Let b′ be the point cloud cluster with the largest s1 order in A, and b′ be the preset angle threshold. S600, obtain the first target line; if A max,1 If each point is located to the right of the mobile charging robot's direction of movement, then the first target straight line is to bring l max,1 The straight line obtained by shifting it to the left by d / 2, where d is l max,1 and l max,2 The distance between them; if A max,2 If each point is located to the right of the mobile charging robot's direction of movement, then the first target straight line is to bring l max,2 The straight line obtained by shifting it to the left by d / 2.

2. The method according to claim 1, characterized in that, It also includes the following steps: S700, for A excluding A max,1 and A max,2 For any cloud cluster A' other than A': obtain the set B of inclination angles of the lines containing each line segment in the set of line segments corresponding to A'; S800, cluster B according to the difference between any two tilt angles in B to obtain the first tilt angle cluster B1 and the second tilt angle cluster B2; S900, obtain the mean eB1 of all tilt angles in B1 and the mean eB2 of all tilt angles in B2; if Then A' is determined to be the point cloud cluster corresponding to the rectangular support column.

3. The method according to claim 2, characterized in that, In S900, if A' is determined to be the point cloud cluster corresponding to the rectangular support column, then the rectangular support column corresponding to A' is used as a landmark for the subsequent relocalization of the mobile charging robot.

4. The method according to claim 2, characterized in that, If A' is determined to be a point cloud cluster corresponding to an irregular obstacle, then the following steps are performed: S910, obtain A h Point P1 is the closest point to the mobile charging robot and point P2 is the farthest point from the mobile charging robot; S920, take the line connecting P1 and P2 as the diameter of the circumcircle of the irregular obstacle, and obtain the circumcircle of the irregular obstacle; S930, the circumcircle of the irregular obstacle is used as the size of the irregular obstacle for obstacle avoidance.

5. The method according to claim 2, characterized in that, In S800, the k-means clustering algorithm is used to cluster B, where k=2.

6. The method according to claim 5, characterized in that, Clustering B using the k-means clustering algorithm includes: S810, randomly select two tilt angles from B as the centroid; S820, for each tilt angle in B, obtain its distance to each centroid and assign it to the set of the nearest centroid; where the p-th tilt angle Bb in B p With the qth tilt angle Bb q distance d p,q =|b p -b q | p and q both range from 1 to Q B Q B The number of tilt angles in B; S830, after dividing B into each tilt angle, the centroids of the two sets are obtained again; S840, repeat steps S820-S830 until the obtained centroids no longer change or the number of repetitions reaches the set number, then the clustering ends.

7. The method according to claim 2, characterized in that, In S500, if Then repeat S100-S400.

8. The method according to claim 1, characterized in that, In S100, a multi-line lidar is used to acquire point clouds of the target area in the garage.

9. The method according to claim 1, characterized in that, The point cloud of the target area in the garage is a two-dimensional point cloud, which does not include height information.

Images