KR102114558B1 - Ground and non ground detection apparatus and method utilizing lidar - Google Patents

Abstract

The present invention provides a ground and non-ground detection apparatus using a LiDAR and a method thereof. The apparatus comprises: a LiDAR sensor performing a scanning operation while rotating at 360 degrees in a predetermined direction to obtain a three dimensional point group composed of multiple points each of which has a coordinate value of the three dimensional coordinate system (x, y, z); a distortion correcting unit correcting a distortion generated due to movement in the three dimensional point group while the LiDAR sensor rotates and performs the scanning operation; an ID line identification unit discriminating the multiple points of the distortion-corrected three dimensional point group according to a field of view (FOV) of the LiDAR sensor to identify N ID lines; a feature point extraction unit detecting the nearest point of the same ID line and the nearest point of other ID line on the xy plane with respect to each of the multiple points of the N ID lines and determining points of which distances from the detected points on the xy plane are within a predetermined first reference distance and of which z value differences are equal to or greater than a predetermined second reference distance to extract the points as feature points; and a non-ground detection unit detecting the determined feature points as the non-ground and detecting the points except the feature points as the ground.

Description

Ground and non-ground detection device and method using riders {GROUND AND NON GROUND DETECTION APPARATUS AND METHOD UTILIZING LIDAR}

The present invention relates to a ground and non-ground detection device and method using a lidar, and to a ground and non-ground detection device and method utilizing the geometric features of the 3D point group obtained through the lidar.

The LiDAR sensor is a 3D point cloud based on a relative distance centered on itself by scanning the surroundings by emitting light such as a laser pulse while rotating 360 degrees and detecting the light that is reflected back. By obtaining a, unlike other depth measurement sensors, it has a great advantage in that it can acquire a wide range of depth information outdoors. Accordingly, many studies have been conducted using riders, and in particular, studies on object detection, road detection, and the like have been conducted on point groups obtained from riders. In addition, many researches on SLAM (Simultaneous Localization and Mapping) using riders have been conducted.

Currently, researches using Lida acquire characteristic points used in Odometry / Mapping using separate RGB images or extract them from the 3D point group itself. However, the feature points extracted from the 3D point group cannot be considered to be constant every frame, and there is a disadvantage that the geometric features of the entire point group are not utilized. In addition, as many researches on lidar use are carried out by mounting a lidar sensor on a vehicle such as a vehicle, the 3D point cloud extracted from the lidar contains distortion due to the movement of the lidar sensor. Accordingly, a post-processing operation must be performed to compensate for distortion.

1 shows an example of distortion generated in a point cloud by the movement of a lidar.

In FIG. 1 (a), area A and area B, and area (b) area A represent a part of the area where point group distortion is generated. Looking at area A and area B enlarged in (a), the point group obtained from lidar It can be seen that the distortion-generated region has a relatively non-uniform spacing compared to other regions.

As described above, the distortion is due to the movement of the position of the lidar sensor, and it is unlikely that distortion is present in the point group acquired while the lidar is fixed at a specific position, but in the point group obtained by moving the lidar, While the Ida is rotated 360 degrees to acquire a point cloud, the position of the start and end points of the scan is different due to the positional movement of the Lida, which causes distortion.

Fig. 2 shows an example of a misleading metric path for distortion correction due to the movement of a lidar.

Conventionally, in order to correct the motion distortion of the 3D point group acquired by the lidar, a method of correcting motion using a sensor (eg, IMU / GPS) other than the lidar is mainly used. And the evaluation in SLAM-related studies is performed as a method of presenting the prediction results of the odometrics as shown in Fig. 2 than the actual 3D map, and expressing the error of the rotation matrix and the motion vector in the proposed odometric predictions.

However, the motion distortion correction method using IMU / GPS has a limitation that it cannot be used in urban areas with high-rise buildings, unless a very expensive IMU sensor is used. Therefore, a technique capable of correcting motion distortion using only a lidar sensor is required.

In addition, in the SLAM study, feature points of a 3D point group were extracted based on an edge / planar, or only 100 points were extracted by comparing reference values.

3 shows an example of feature points extracted from the existing SLAM.

As shown in FIG. 3, in 3D point clusters, low / sparse feature points obtained through comparison with edge / planar-based or reference values may cause a large error in subsequent mapping / matching. there is a problem.

Korean Registered Patent No. 10-1899549 (Registration on September 11, 2018)

An object of the present invention is to provide a ground and non-ground detection apparatus and method that can detect ground and non-ground, that is, objects by using geometric features of a 3D point cloud obtained through a lidar.

Another object of the present invention is to provide a ground and non-ground detection apparatus and method using a lidar capable of compensating for motion distortion caused by movement of a lidar.

A ground and non-ground detection device using a lidar according to an embodiment of the present invention for achieving the above object is scanned while rotating 360 degrees in a predetermined direction, and coordinate values of a three-dimensional coordinate system (x, y, z), respectively A lidar sensor acquiring a 3D point group consisting of a plurality of points having a; A distortion correction unit for correcting distortion caused by the rotation of the rider sensor during the scan in the 3D point group; An ID line discriminating unit for dividing a plurality of points of the distortion-corrected 3D point group into N ID lines by dividing them according to a field of view (FOV) of the lidar sensor; For each of the plurality of points of the N ID lines, a first reference distance in which a distance on the xy plane is determined by searching for the closest point of the same ID line and the closest point of another ID line on the xy plane A feature point extracting unit for determining points within a difference between the z values and the second reference distance or more as a feature point; And a non-ground detection unit that detects the determined feature points as non-ground and detects points other than the feature points as ground. It includes.

The distortion correction unit searches for a start point and an end point in the 3D point group, and calculates and obtains a distance difference in the x-axis direction and a distance in the y-axis direction between the searched start and end points as offset values, and scans the direction of the lidar sensor In consideration of, it is possible to compensate using offset values obtained for points of a predetermined section that is first scanned in a 3D point group.

The distortion correction unit may compensate for distortions of points in a section obtained during a section in which the lidar sensor is initially rotated 1/2.

The ground and non-ground detection method using a lidar according to another embodiment of the present invention for achieving the above object is scanned while rotating 360 degrees in a predetermined direction using a lidar sensor, and each of the three-dimensional coordinate systems (x, y , z) obtaining a 3D point group consisting of a plurality of points having coordinate values; Correcting distortion caused by the rider sensor rotating and moving during the scan in the 3D point group; Dividing a plurality of points of the distortion-corrected 3D point group into N ID lines by dividing them according to the field of view (FOV) of the lidar sensor; Searching for a plurality of points of the N ID lines on the xy plane, the closest point of the same ID line and the closest point of another ID line; Determining a point in which the distance on the xy plane with respect to the searched points is within a predetermined first reference distance, and a difference in the z value is greater than or equal to a predetermined second reference distance and extracting the feature point; And detecting the determined feature point as non-ground, and detecting points other than the feature point as ground. It includes.

Therefore, the ground and non-ground detection apparatus and method using the lidar according to the embodiment of the present invention search for the position of the scan start point and the position of the end point, obtain an offset value from the difference between the start point and the end point to move the lidar. Distortion can be corrected, and feature points are extracted by using the distance difference in the xy plane and the distance in the z-axis direction between each point in the distortion-corrected point group, thereby accurately detecting the ground and the non-ground. Therefore, it is possible to provide depth information with excellent performance in mapping / matching, and can be used in various ways, such as SLAM, object detection, and autonomous driving algorithms.

1 shows an example of distortion generated in a point cloud by the movement of a lidar.
Fig. 2 shows an example of a misleading metric path for distortion correction due to the movement of a lidar.
3 shows an example of feature points extracted from the existing SLAM.
4 is a view for explaining the concept of the 3D point cloud obtained from the lidar sensor.
5 shows an example of feature points extracted based on geometric features of an object in a 3D point group.
6 shows a schematic structure of a ground and non-ground detection device using a lidar according to an embodiment of the present invention.
7 and 8 show concepts and algorithms in which the distortion correction unit of FIG. 6 corrects motion distortion of the lidar sensor, respectively.
9 is a view for explaining the concept of the feature point extraction unit of FIG. 6 to distinguish non-ground points and ground points from a 3D point group.
10 shows a method of detecting ground and non-ground using a lidar according to an embodiment of the present invention.
11 and 12 show results obtained by dividing the points of the 3D point group obtained using the ground and non-ground detection apparatus and method using the lidar according to the present embodiment into a ground and a non-ground.

In order to fully understand the present invention, the operational advantages of the present invention, and the objects achieved by the practice of the present invention, reference should be made to the accompanying drawings and the contents described in the accompanying drawings, which illustrate preferred embodiments of the present invention.

Hereinafter, the present invention will be described in detail by explaining preferred embodiments of the present invention with reference to the accompanying drawings. However, the present invention can be implemented in various different forms, and is not limited to the described embodiments. And, in order to clearly describe the present invention, parts irrelevant to the description are omitted, and the same reference numerals in the drawings indicate the same members.

Throughout the specification, when a part “includes” a certain component, it means that the component may further include other components, not to exclude other components, unless otherwise stated. In addition, terms such as "... part", "... group", "module", and "block" described in the specification mean a unit that processes at least one function or operation, which is hardware or software or hardware. And software.

4 is a view for explaining the concept of the 3D point cloud obtained from the lidar sensor.

As described above, the lidar sensor rotates 360 degrees around the z-axis, and emits light such as a laser pulse, and scans the surroundings by detecting the light that has returned to the ground or an object and reflected back. At this time, the lidar sensor acquires the reflection position of the light in the form of a 3D point group consisting of a plurality of points represented by coordinate values (x, y, z) of the 3D coordinate system.

As shown in Fig. 4 (a), in the three-dimensional coordinate system, the x-axis represents the moving direction in the forward direction of the lidar sensor, the y-axis represents the left direction of the lidar sensor, and the z-axis represents the upper direction of the lidar sensor. Shows.

The 3D point group obtained by rotating the lidar sensor 360 degrees includes N lines (where N is a natural number) each composed of a plurality of points, and each of the N lines is called an ID line (ID).

In FIG. 4 (a), for example, among the N IDs of the 3D point group, the k-1 ID line (ID _k-1 ), the k ID line (ID _k ), and the k + 1 ID line (ID _{k + 1} ). Only 3 ID lines are shown. In addition, in FIG. 4, p _j ^{k + 1} means a j-th point in the k + 1 ID line (ID _{k + 1} ).

Therefore, the coordinate value (x, y, z) of the j-th point (p _j ^{k + 1} ) of the k + 1 ID line (ID _{k + 1} ) is not a point on the xy plane, but a point having a z-axis size (z ≠ 0), the jth point (p _j ^{k + 1} ) of the k + 1 ID line (ID _{k + 1} ) is a point projected on the xy plane with respect to the x axis, as shown in (b) ( It has a vertical angle θ in the z-axis direction with respect to the xy plane along with a horizontal angle φ for the position of p _j ^{k + 1} ′).

5 shows an example of a feature point extracted based on a geometric feature of an object in a 3D point group.

In FIG. 5, the feature points obtained by distinguishing the ground from the non-ground using geometric features in the 3D point group obtained from the lidar are shown. In FIG. 5, the gray dots represent points on the ground, and the purple dots represent points on the non-ground, that is, the feature points.

In general, it can be determined that a geometrical characteristic of an object other than the ground having a planar property, that is, non-ground, is in a relative height difference with respect to the ground.

Accordingly, in the present invention, in the 3D point group, the difference between the coordinate values in the x and y axes is not large compared to the surrounding points, but the concept that the point in which the difference in coordinate values in the z-axis protrudes can be selected as a feature point indicating non-ground surface. Therefore, the ground and the non-ground should be detected separately.

As can be seen in FIG. 5, if the ground and the non-ground are distinguished using geometric features in the 3D point group, and the points on the divided ground can be extracted as the feature points, then the object can be easily detected, and the object When mapping / matching, errors can be greatly reduced.

6 shows a schematic structure of a ground and non-ground detection device using a lidar according to an embodiment of the present invention, and FIGS. 7 and 8 are concepts in which the distortion correction unit of FIG. 6 corrects motion distortion of the lidar sensor, respectively And an algorithm, and FIG. 9 is a diagram for explaining a concept in which the feature point extraction unit of FIG. 6 classifies non-ground points and ground points in a 3D point group.

Referring to FIG. 6, the ground and non-ground detection apparatus according to the present embodiment is a lidar sensor 100, a distortion correction unit 200, an ID line discrimination unit 300, a feature point extraction unit 400, and a non-ground detection unit 500.

The lidar sensor 100 rotates 360 degrees, emits light such as a laser pulse, and senses the returned light when the emitted light reaches the ground or an object and returns, thereby generating a group of 3D points each containing a plurality of points. To acquire.

The distortion correcting unit 200 compensates for motion distortion generated by the movement of the lidar sensor in the 3D point group obtained from the lidar sensor 100. As shown in FIG. 7, the motion distortion is a movement change in the x-axis and y-axis directions while the lidar sensor 100 scans, so that the positions of the start point (A) and the end point (B) of the scan are different. Occurs by Therefore, distortion compensation is performed based on the x and y values of the scan start point (A) and the end point (B). In FIG. 7, the origin (O) means the position of the lidar sensor 100 at the time when the scan is started from the designated direction and rotated once to end the scan.

The distortion correction unit 200 first searches the positions of the start point (A) and the end point (B) of the scan. The distortion correction unit 200 determines the start point A and the end point B from the 3D point group P, and determines the positions of the determined start point A and end point B.

At this time, the distortion correction unit 200 does not distinguish N ID lines constituting the 3D point group. Therefore, the distortion correction unit 200 cannot directly check the start and end points of each of the N ID lines (ID ₁ to ID _n ), and thus the x and y coordinate values of the coordinate values (x, y, z) of each point Based on the N ID lines, the start and end points of the outermost ID line (ID _n ) or the innermost ID line (ID ₁ ) are searched.

Hereinafter, as an example, the lidar sensor 100 moves in the x-axis direction, and the lidar sensor 100 scans clockwise (-180 degrees to 180 degrees) from the ?? x direction, and the starting point of the outermost ID line And it is assumed that the end point is searched.

As shown in FIG. 8, the distortion correction unit 200 searches for the starting point A in a first angle range designated as (-180 <φ <-180 + th ₁ ) based on the horizontal angle φ. Here, th ₁ means a predetermined angle threshold. For example, the angle threshold value th ₁ may be set to an angle at which the lidar sensor 100 that periodically emits light rotates for one period that emits light.

The distortion correction unit 200 analyzes the coordinate values (x, y, z) of at least one point (p _i ) obtained within the first angle range specified by the angle threshold value (th ₁ ), the smallest x The value (min x) is set to the starting point (A) x value (starting x), and the smallest y value (min y) is set to the starting point (A) y value (starting y). That is, the position of the starting point A is set to a coordinate value (starting x, starting y) on the xy coordinate plane.

Then, the distortion correction unit 200 searches for the end point B in the range of (180-th ₁ <φ <180) based on the horizontal angle φ, similar to the starting point A. The distortion correction unit 200 analyzes coordinate values (x, y, z) of at least one point obtained within a specified angular range, and the smallest x value (min x) is the x value of the ending point B (ending) x), and the largest y value (max y) is set to the y value of the ending point B (ending y). That is, the position of the end point B is set to a coordinate value (ending x, ending y) on the xy coordinate plane.

When the position of the starting point (A) (starting x, starting y) and the position of the ending point (B) (ending x, ending y) are set, the distortion correcting unit 200 starts the position (A) of the starting point (starting x, starting y) ) And the end point (B) of the position (ending x, ending y) to obtain the x-axis offset value (d ₁ ) (offset _x ) and y-axis offset value (d ₂ ) (offset _y ). Here, the x-axis offset value (offset _x ) and the y-axis offset value (offset _y ) are shown as d ₁ and d ₂ in FIG. 7, respectively. And, as shown in FIG. 8, the offset values (offset _x , offset _y ) in the x-axis direction and the y-axis direction are each of the starting point (A) at the x and y coordinate values (ending x, ending y) of the end point (B), respectively. It can be obtained by subtracting the x and y coordinate values (starting x, starting y).

When the offset values (offset _x , offset _y ) are obtained, the distortion correction unit 200 y-axis for points (p _i ) (p _i .y> 0) whose y value is greater than 0 in the 3D point group (P). the y value of the offset level (offset _y) and the horizontal angle (φ) of the y value is greater point (p _i) greater than 0 using a (p _i .y> 0) the direction is compensated as shown in the equation (1).

In Equation 1, since the (ID / N) item is a point included in one of the N ID lines (ID ₁ to ID _n ) constituting a 3D point group, an offset of a size corresponding to the included ID line (ID) This is to ensure that it is reflected.

In addition, the value of y is greater than 0 points (p _i .y> 0) the offset value of the x-axis direction with respect to the _(x offset) the group is greater than a specified threshold offset value (th _off), the value of y is greater than 0 points (p The y values of _i ) (p _i .y> 0) are compensated using Equation 2 using the offset value ( _x ) in the x-axis direction and the horizontal angle (φ).

Here, the distortion correction unit 200 compensates for the offset values of only half from the scan start point of the lidar sensor 100, where the y value is greater than 0 (p _i .y> 0), that is, the lidar sensor at the end of the scan This is because the location of the (100) is set as the origin (O), so that distortion occurs in the initial scan section.

If the scan start position or direction of the lidar sensor 100 is different, or search for the start point (A) and end point (B) of the innermost ID line (ID ₁ ) among the N ID lines (ID ₁ to ID _n ) In the case, the algorithm of Fig. 8 of the start point (A) and the end point (B) may also be partially modified.

)과 수직각(

)을 획득하고, 라이다 센서(100)의 FOV(Field of View)에 기반하여, 수학식 3에 따라 다수의 구간으로 구분함으로써, N개의 ID 라인(ID)을 획득한다.When the distortion correction unit 200 compensates for the distortion of the 3D point group, the ID line distinguishing unit 300 vectorizes the distortion-compensated 3D point group and distinguishes them into N ID lines. The ID line discrimination unit 300 is a horizontal angle for each of a plurality of points in the 3D point group (

) And vertical angle (

), And based on the field of view (FOV) of the lidar sensor 100, divided into a plurality of sections according to Equation 3 to obtain N ID lines ID.

이고, 채널 수(# of channels), 상한 및 하한은 라이다 센서(100)의 고정 변수들이다.here,

And the number of channels (# of channels), the upper limit and the lower limit are fixed variables of the lidar sensor 100.

The ID line distinguishing unit 300 may arrange a plurality of points p _i of each of the obtained N ID lines ID in ascending order based on the horizontal angle φ. That is, points p _{i of} each of the N ID lines ID ₁ to ID _n may be sequentially arranged from the scan start position.

On the other hand, the feature point extracting unit 400 analyzes the geometric features of the points p _i of the N ID lines ID ₁ to ID _n to determine points other than the ground, and extracts the feature points.

As mentioned above, the geometrical features of the non-ground can be considered to be in a difference in height relative to the ground.

Accordingly, in the present invention, the difference in coordinate values in the x and y axes is not large compared to the surrounding points in the 3D point group, but a point in which the difference in coordinate values in the z axis is large is extracted as a feature point.

The feature point extracting unit 400 is adjacent to the closest point of the same ID line and the closest point of another ID line on the xy plane for each point p _i of the N ID lines ID ₁ to ID _n according to the Nearest Neighbors Search technique. Search for points. As an example, when the j-th point (p _j ⁴ ) of the fourth ID line (ID ₄ ) in FIG. 9 is A, the feature point extracting unit 400 is the j-1-th point in the same fourth ID line (ID ₄ ). Search for (p _j-1 ⁴ ) and j + 1th point (p _{j + 1} ⁴ ) and designate the nearest point as C. Then, the point closest to the point A in the point group p ³ and p ^{5 of} the third ID line ID ₃ and the fifth ID line ID ₅ is searched and designated as B.

The feature point extracting unit 400 calculates a distance on the xy plane between points A and B and C, and if the distance to point B or C on the xy plane is within a predetermined first reference distance, point A And z-value difference between point B and point C. If at least one of the difference of the z value from the B point or the C point is greater than or equal to a second predetermined reference distance, a point greater than or equal to the second reference distance is extracted as a feature point.

The non-ground detection unit 500 detects the feature points determined by the feature point extraction unit 400 as non-ground, that is, objects, and detects points other than the feature points as the ground. The non-ground detection unit 500 divides a plurality of points of a 3D point group into points on the ground and points on the non-ground, and expresses them in different colors, so that the user can easily identify objects and the ground.

10 shows a method of detecting ground and non-ground using a lidar according to an embodiment of the present invention.

Referring to Figures 5 to 9, when explaining the ground and non-ground detection method using the lidar of FIG. 10, first, the lidar sensor 100 rotates around 360 degrees and scans to obtain a 3D point group (P). (S10).

The distortion correction unit 200 compensates for distortion caused by the movement of the rider sensor 100 during scanning in the 3D point group P acquired by the rider sensor 100 (S20). The distortion correction unit 200 first searches for a start point and an end point in the 3D point group P. At this time, the distortion correction unit 200 searches for the start and end points of the outermost ID line ID _n or the innermost ID line ID ₁ of the N ID lines ID ₁ to ID _n of the 3D point group P. . And the distance difference (d1) in the x-axis direction and the distance difference (d2) in the y-axis direction between the searched start and end points are calculated and obtained as offset values (S22). When the offset value is calculated, the distortion correcting unit 200 distorts the points in the section that is first scanned in the 3D point group (in this example, the section where the rider sensor 100 initially rotates 1/2) in consideration of the scan direction. Compensation is performed (S23).

)과 수직각(

)으로 표현되도록 벡터화하고, 벡터화된 3D 점군의 다수의 점들 각각을 라이다 센서(100)의 FOV에 기반하여 다수의 구간으로 구분함으로써, N개의 ID 라인(ID)을 획득한다(S30). ID 라인 구별부(300)는 획득된 N개의 ID 라인(ID) 각각의 다수의 점(p_i) 을 수평각(φ)을 기준으로 오름차순 정렬할 수 있다.Thereafter, each of the plurality of points of the 3D point group in which the ID line discrimination unit 300 is distortion-compensated is horizontal (

) And vertical angle (

Vectorized to be represented by), and each of the plurality of points of the vectorized 3D point group is divided into a plurality of sections based on the FOV of the lidar sensor 100, thereby obtaining N ID lines (ID) (S30). The ID line distinguishing unit 300 may arrange a plurality of points p _i of each of the obtained N ID lines ID in ascending order based on the horizontal angle φ.

The feature point extracting unit S40 searches for the closest point of the same ID line and the closest point of another ID line on the xy plane for each of the points p _i of the N ID lines ID ₁ to ID _n , and searches The points on the xy plane with respect to the determined points are within a predetermined first reference distance, and points with a difference in z value equal to or greater than the predetermined second reference distance are determined and extracted as a feature point (S40).

The non-ground detection unit 500 detects the determined feature point as an empty surface, that is, an object, and detects points other than the feature point as the ground (S50).

11 and 12 show results obtained by dividing the points of the 3D point group obtained using the ground and non-ground detection apparatus and method using the lidar according to the present embodiment into a ground and a non-ground.

In FIG. 11, the ground is shown in gray, and the non-ground indicates a ground / non-ground detection image in pink. As shown in Fig. 11, the apparatus and method for detecting the ground and the ground using the lidar according to the present embodiment can accurately detect the ground and the ground.

In FIG. 12, the dots for the ground are not separately displayed, and the color of the dots for the non-ground is expressed differently according to the z value. As shown in FIG. 12, when colors are differently expressed according to the z value, it can be seen that the user can more accurately recognize the shape of the ground surface.

11 and 12, it can be seen that distortion due to movement of the lidar sensor 100 is compensated.

The method according to the present invention can be implemented as a computer program stored in a medium for execution on a computer. The computer readable medium herein can be any available medium that can be accessed by a computer, and can also include any computer storage medium. Computer storage media includes both volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data, and ROM (readable) Dedicated memory), RAM (random access memory), CD (compact disk) -ROM, DVD (digital video disk) -ROM, magnetic tape, floppy disk, optical data storage, and the like.

The present invention has been described with reference to the embodiments shown in the drawings, but these are merely exemplary, and those skilled in the art will understand that various modifications and other equivalent embodiments are possible therefrom.

Therefore, the true technical protection scope of the present invention should be defined by the technical spirit of the appended claims.

100: lidar sensor 200: distortion correction unit
300: ID line distinguishing unit 400: feature point extraction unit
500: non-ground detection unit

Claims (10)

A lidar sensor that scans while rotating 360 degrees in a predetermined direction to obtain a 3D point group consisting of a plurality of points each having coordinate values of a three-dimensional coordinate system (x, y, z);
A distortion correction unit for correcting distortion caused by the rotation of the rider sensor during the scan in the 3D point group;
An ID line discriminating unit for dividing a plurality of points of the distortion-corrected 3D point group into N ID lines by dividing them according to a field of view (FOV) of the lidar sensor;
For each of the plurality of points of the N ID lines, a first reference distance in which a distance on the xy plane is determined by searching for the closest point of the same ID line and the closest point of another ID line on the xy plane A feature point extracting unit for determining points within a difference between the z values and the second reference distance or more as a feature point; And
A non-ground detection unit that detects the determined feature points as non-ground and detects points other than the feature points as ground; Including,
The distortion correction unit
Search for a start point and an end point in the 3D point group,
It is obtained by calculating the distance difference in the x-axis direction and the distance in the y-axis direction between the searched start and end points as offset values.
Compensation using offset values obtained for points in a predetermined section that is first scanned in a 3D point group in consideration of the scan direction of the lidar sensor,
The distortion correction unit
When the lidar sensor moves in the x-axis direction and rotates from the -x direction (-180 degrees) to start scanning,
-180 to the start point by comparing the groups x and y values of the at least one point obtained from a range of an angle threshold value (th ₁₎ within (-180 <horizontal angle (φ) <-180 th + ₁₎ at the Set it up,
Ground and non-ground using a rider that sets the end point by comparing the x value and the y value of at least one point obtained within a range of 180 degrees within a predetermined angle threshold (180-th ₁ <φ <180). Detection device.

delete The method of claim 1, wherein the distortion correction unit
A ground and non-ground detection device using a lidar that compensates for distortion of points obtained during a section in which the lidar sensor is initially rotated for 1/2 turn.
delete The method of claim 1, wherein the distortion correction unit
Of the offset value and the y value of the compensated offset value points along the y axis _(y offset) Equation using the horizontal angle (φ)

Reward according to,
The y value of the points to be compensated is expressed by using the offset value in the x-axis direction (offset _x ) and the horizontal angle (φ) among the offset values.

Ground and non-ground detection device using lidar to compensate according to. The method of claim 1, wherein the ID line discrimination unit
Equation

(Where θ is the vertical angle,

And # of channels, upper and lower limits are fixed variables of the lidar sensor.)
By dividing into a plurality of sections according to, to obtain the N ID line,
A ground and non-ground detection device using a lidar that sorts a plurality of points of each of the acquired N ID lines in ascending order based on a horizontal angle. Scanning while rotating 360 degrees in a predetermined direction using a lidar sensor, to obtain a 3D point group consisting of a plurality of points each having a coordinate value of a 3D coordinate system (x, y, z);
Correcting distortion caused by the rider sensor rotating and moving during the scan in the 3D point group;
Dividing a plurality of points of the distortion-corrected 3D point group into N ID lines by dividing them according to the field of view (FOV) of the lidar sensor;
Searching for a plurality of points of the N ID lines on the xy plane, the closest point of the same ID line and the closest point of another ID line;
Determining a point in which the distance on the xy plane with respect to the searched points is within a predetermined first reference distance, and a difference in the z value is greater than or equal to a predetermined second reference distance and extracting it as a feature point; And
Detecting the determined feature point as a non-ground surface, and detecting points other than the feature point as a ground; Including,
Correcting the distortion is
Searching for a start point and an end point in the 3D point group;
Calculating and obtaining a distance difference in the x-axis direction and a distance in the y-axis direction between the searched start point and end point as an offset value; And
Compensating by using the offset value obtained for points of a predetermined section that is first scanned in a 3D point group in consideration of the scan direction of the lidar sensor; Including,
Compensating the distortion,
When the lidar sensor moves in the x-axis direction and starts scanning by rotating from the -x direction (-180 degrees), within a predetermined angle threshold (th ₁ ) at -180 degrees (-180 <horizontal angle (φ ) <-180 + th ₁ ) Set the starting point by comparing the x and y values of at least one point obtained in the range, and within a predetermined angle threshold at 180 degrees (180-th ₁ <φ <180 Method of detecting ground and non-ground using a lidar that sets the end point by comparing x and y values of at least one point obtained in the range of).
delete The method of claim 7, wherein correcting the distortion is
A method of detecting ground and non-ground using a lidar that compensates for distortion of points in a section obtained during a section in which the lidar sensor is initially rotated 1/2. The method of claim 7, wherein the step of distinguishing by the ID line
Equation

(Where θ is the vertical angle,

And # of channels, upper and lower limits are fixed variables of the lidar sensor.)
Acquiring the N ID lines by dividing them into a plurality of sections according to; And
Arranging a plurality of points of each of the obtained N ID lines in ascending order based on a horizontal angle; Ground and non-ground detection method using a lidar comprising a.

Images