Granted Patent B2

US 12,339,363 · App. 17/244,665 · Granted Jun 24, 2025

Inventors: Nam Gyun Kim (Seongnam-si, KR); Mu Gwan Jeong (Seoul, KR)

Assignees: Hyundai Motor Company; Kia Corporation

G01S17/89G06V10/751G06V20/56

Abstract

Claims (99)

1. An apparatus for recognizing a high-elevation structure using a light detection and ranging (LiDAR) sensor, the apparatus comprising:

an object-sensing unit configured to output a point cloud, acquired by sensing an object, as LiDAR data;

a point-extracting unit configured to extract points associated with the high-elevation structure from among points included in the LiDAR data, the high-elevation structure being an object having a height greater than or equal to a predetermined level; and

an information-generating unit configured to:

generate a contour of the high-elevation structure;

estimate a location of the high-elevation structure using the extracted points; and

output the generated contour and the estimated location as information representing the high-elevation structure,

wherein the point-extracting unit further comprises:

a verticality inspection unit configured to select points satisfying verticality from among the points included in the LiDAR data;

a horizontality inspection unit configured to select points satisfying horizontality from among the points satisfying the verticality; and

a point-filtering unit configured to remove points having noise from among points satisfying both the verticality and the horizontality, through a density inspection.

2. The apparatus according to claim 1 , wherein the verticality inspection unit is configured to:

when points of 0 th to M−1 th channels are sensed simultaneously at every predetermined unit angle (x) where M is a positive integer of 2 or greater, and a point of each channel vertically overlaps a point of another channel, determine that the point of each channel satisfies the verticality.

3. The apparatus according to claim 1 , wherein the point-filtering unit is configured to:

construct a tree based on points satisfying both the verticality and the horizontality;

detect the number of points that are within a predetermined range of respective points of the constructed tree;

when the detected number is not greater than a threshold number, determine that the points satisfying both the verticality and the horizontality are points having noise; and

remove the points having noise.

4. The apparatus according to claim 1 , wherein the information-generating unit further comprises:

a map-generating unit configured to generate a grid map based on the extracted points associated with the high-elevation structure;

a point-labeling unit configured to label points on the grid map; and

a contour-generating and location-estimating unit configured to:

generate a contour of the high-elevation structure by connecting contours of groups labeled by the point-labeling unit; and

estimate a location where the generated contour is present as a location of the high-elevation structure.

5. The apparatus according to claim 4 , wherein the information-generating unit further comprises:

a post-processing unit configured to remove a group having a size smaller than a group threshold size from among the groups labeled by the point-labeling unit.

6. An apparatus for recognizing a high-elevation structure using a light detection and ranging (LiDAR) sensor, the apparatus comprising:

an object-sensing unit configured to output a point cloud, acquired by sensing an object, as LiDAR data;

a point-extracting unit configured to extract points associated with the high-elevation structure from among points included in the LiDAR data, the high-elevation structure being an object having a height greater than or equal to a predetermined level; and

an information-generating unit configured to:

generate a contour of the high-elevation structure;

estimate a location of the high-elevation structure using the extracted points; and

output the generated contour and the estimated location as information representing the high-elevation structure,

wherein the point-extracting unit further comprises:

a verticality inspection unit configured to select points satisfying verticality from among the points included in the LiDAR data;

a horizontality inspection unit configured to select points satisfying horizontality from among the points satisfying the verticality; and

a point-filtering unit configured to remove points having noise from among points satisfying both the verticality and the horizontality, through a density inspection;

wherein the verticality inspection unit is configured to:

when points of 0 th to M−1 th channels are sensed simultaneously at every predetermined unit angle (x) where M is a positive integer of 2 or greater, and a point of each channel vertically overlaps a point of another channel, determine that the point of the each channel satisfies the verticality:

in order to check whether a point of an m th (1≤m≤M) channel satisfies the verticality, obtain a verticality vector between a point (PNT(f,m)) (1≤f≤F, F=360/x) of the m th channel and a point (PNT(f,m−k)) of an m−k th (1≤k≤M) channel;

obtain a magnitude and an angle of the verticality vector; and

determine that the point of the m th channel is a point satisfying the verticality when the obtained magnitude is less than a vertical threshold magnitude and the obtained angle is less than a vertical threshold angle.

7. The apparatus according to claim 6 , wherein the horizontality inspection unit is configured to:

select a size and a moving direction of a moving window;

detect the number of points that satisfy the verticality within the moving window to which points satisfying the verticality among the points of the 0 th to M−1 th channels belong; and

determine that the points satisfying the verticality also satisfy the horizontality when the detected number is greater than a predetermined number.

8. The apparatus according to claim 7 , wherein the horizontality inspection unit is configured to:

in order to check whether a point (PNT(f,n)) of an n th channel (0≤n≤M−1) sensed in an f th sequential position satisfies the horizontality, obtain a horizontality vector between a point (PNT(f+1,n)) of the n th channel sensed in an f+1 th sequential position and the point (PNT(f,n)) of the n th channel sensed in the f th sequential position when the point (PNT(f,n)) of the n th channel is a point satisfying the verticality;

obtain a magnitude and an angle of the horizontality vector; and

determine that the point (PNT(f,n)) of the n th channel is a point satisfying the horizontality when the obtained magnitude is less than a horizontal threshold magnitude and the obtained angle is less than a horizontal threshold angle.

9. The apparatus according to claim 6 , wherein the point-filtering unit is configured to:

construct a tree based on points satisfying both the verticality and the horizontality;

detect the number of points that are within a predetermined range of respective points of the constructed tree;

when the detected number is not greater than a threshold number, determine that the points satisfying both the verticality and the horizontality are points having noise; and

remove the points having noise.

10. The apparatus according to claim 6 , wherein the information-generating unit further comprises:

a map-generating unit configured to generate a grid map based on the extracted points associated with the high-elevation structure;

a point-labeling unit configured to label points on the grid map; and

a contour-generating and location-estimating unit configured to:

generate a contour of the high-elevation structure by connecting contours of groups labeled by the point-labeling unit; and

estimate a location where the generated contour is present as a location of the high-elevation structure.

11. The apparatus according to claim 10 , wherein the information-generating unit further comprises:

a post-processing unit configured to remove a group having a size smaller than a group threshold size from among the groups labeled by the point-labeling unit.

12. A method of recognizing a high-elevation structure using a light detection and ranging (LiDAR) sensor, the method comprising:

acquiring a point cloud as LiDAR data by sensing an object;

extracting points associated with a high-elevation structure from among points included in the LiDAR data, the high-elevation structure being an object having a height greater than or equal to a predetermined level;

generating a contour of the high-elevation structure; estimating a location of the high-elevation structure using the extracted points; and

determining that the generated contour and the estimated location is information representing the high-elevation structure, wherein extracting the points comprises:

selecting points satisfying verticality from among the points included in the LiDAR data;

selecting points satisfying horizontality from among the points satisfying the verticality; and

removing points having noise from among points satisfying both the verticality and the horizontality, through a density inspection.

13. The method according to claim 12 , wherein selecting the points satisfying the verticality comprises:

when points of 0 th to M−1 th channels are sensed simultaneously at every predetermined unit angle (x) where M is a positive integer of 2 or greater, and a point of each channel vertically overlaps a point of another channel, determining that the point of each channel satisfies the verticality.

14. The method according to claim 13 , wherein determining that the point of each channel satisfies the verticality comprises:

checking whether a point of an m th (1≤m≤M) channel satisfies the verticality;

obtaining a verticality vector between a point (PNT(f,m)) (1≤f≤F, F=360/x) of the m th channel and a point (PNT(f,m−k)) of an m−k th (1≤k≤M) channel;

obtaining a magnitude and an angle of the verticality vector; and

when the obtained magnitude is less than a vertical threshold magnitude and the obtained angle is less than a vertical threshold angle, determining that the point of the m th channel is a point satisfying the verticality.

15. The method according to claim 14 , wherein selecting points satisfying the horizontality comprises:

selecting a size and a moving direction of a moving window;

detecting the number of points that satisfy the verticality within the moving window to which points satisfying the verticality among the points of the 0 th to M−1 th channels belong; and

determining that the points satisfying the verticality also satisfy the horizontality when the detected number is greater than a predetermined number.

16. The method according to claim 15 , wherein determining that the points satisfying the verticality satisfy the horizontality comprises:

checking whether a point (PNT(f,n)) of an n th channel (0≤n≤M−1) sensed in an f th sequential position satisfies the horizontality;

obtaining a horizontality vector between a point (PNT(f+1,n)) of the n th channel sensed in an f+1 th sequential position and the point (PNT(f,n)) of the n th channel sensed in the f th sequential position when the point (PNT(f,n)) of the n th channel is a point satisfying the verticality;

obtaining a magnitude and an angle of the horizontality vector; and

when the obtained magnitude is less than a horizontal threshold magnitude and the obtained angle is less than a horizontal threshold angle, determining that the point (PNT(f,n)) of the n th channel is a point satisfying the horizontality.

17. The method according to claim 12 , wherein removing points having noise through the density inspection comprises:

constructing a tree based on points satisfying both the verticality and the horizontality;

detecting the number of points that are within a predetermined range of respective points of the constructed tree;

when the detected number is not greater than a threshold number, determining that the points satisfying both the verticality and the horizontality are points having noise; and

removing the points having noise.

18. The method according to claim 12 , wherein generating the contour of the high-elevation structure and estimating the location of the high-elevation structure comprises:

generating a grid map based on the extracted points associated with the high-elevation structure;

labeling points on the grid map;

generating the contour of the high-elevation structure by connecting contours of labeled groups; and

estimating a location where the generated contour is present as the location of the high-elevation structure.

19. The method according to claim 18 , wherein generating the contour of the high-elevation structure and estimating the location of the high-elevation structure further comprises:

removing a group having a size smaller than a group threshold size from among the labeled groups.

Assignments (1)

Priority Claims (1)

Continuity (1)

References Cited (11)