Vehicle and method for selectively removing specific points included in multi-layer data

Abstract

The present disclosure relates to a vehicle and a method of selectively removing specific points included in multi-layers of data received from a sensor. The vehicle can include a lidar sensor as the sensor; and a signal processor configured to select static objects included in the multi-layers, select target layers from the multi-layers, define shapes formed by points included in each of the selected target layers, select a reference layer based on the shapes formed by the points, and remove respective points from at least one of the remaining layers based on distances between the outline formed by the reference layer and each of the points of the remaining layers and a reference distance.

Description

[0001] Related cross-references

[0002] This application claims the benefit of Korean Patent Application No. 10-2020-0128642, filed on October 6, 2020, the entire contents of which are incorporated herein by reference. Technical Field

[0003] This disclosure relates to a method for selectively removing specific points that distinguish the static object from the object after defining a static object to be identified, which are included in a multilayer collected by a lidar (LiDAR) sensor. Background Technology

[0004] A lidar sensor, which constitutes a lidar system, is a device that radiates high-power laser pulses into the surrounding atmosphere and receives laser pulses reflected from target objects present near the lidar sensor. The received signals are used to measure the distance between the lidar system and the target object, measure the target object's velocity and direction, analyze substances in the surrounding atmosphere, and measure the concentration of substances.

[0005] Because automotive LiDAR systems primarily use laser pulses with a short wavelength of 905 nanometers (nm), they offer excellent 3D map reconstruction performance, making them potentially suitable for use in autonomous vehicles. The high linearity of short-wavelength laser pulses also makes them advantageous for achieving high resolution and accuracy, enabling object perception in three dimensions rather than two.

[0006] Unlike indoor robots that use only one layer to perceive objects, autonomous vehicles use multiple layers to accurately perceive objects, taking into account the slope and curvature of the road.

[0007] The lidar layer data (hereinafter referred to as layer data) received by the lidar system installed in the main vehicle includes information, for example, about static objects such as guardrails, as well as information about structures such as shrubs existing near the guardrails. In order to minimize errors when matching lidar layer data with high-definition maps, it is necessary to distinguish between guardrails, which are important factors to be identified, and shrubs, which are unimportant structures, and to remove data about shrubs.

[0008] In traditional techniques that distinguish between a fence as a static object and shrubs as a structure, and ultimately remove data about shrubs from the layer data, static objects and structures are identified using only points included in a single layer of information received by the lidar system. However, when the points of the structure form a straight line, there is a problem that the points of the static object and the points of the structure cannot be completely distinguished. Summary of the Invention

[0009] Therefore, this disclosure relates to a vehicle and method for selectively removing specific points included in multi-layer data received from a sensor, i.e., a lidar sensor, which substantially eliminates one or more problems caused by the limitations and disadvantages of the prior art.

[0010] The objective of this disclosure is to provide an apparatus for selectively removing specific points included in multiple layers, wherein, in cases where a static object and another structure are logically clustered into one object, the layer forming the shape most similar to the static object is selected as a reference layer, and points associated with structures included in the other layer are removed based on this reference layer.

[0011] Another objective of this disclosure is to provide a method for selectively removing specific points included in multiple layers, wherein, in cases where a static object and another structure are logically clustered into one object, the layer forming the shape most similar to the static object is selected as a reference layer, and points associated with structures included in the other layer are removed based on this reference layer.

[0012] However, the objectives to be achieved by this disclosure are not limited to those described above, and other objectives not mentioned herein will be clearly understood by those skilled in the art from the following description.

[0013] According to one aspect of this disclosure, the above and other objectives can be achieved by providing a vehicle equipped with means for selectively removing specific points included in multiple layers, the vehicle including a signal processor configured to perform the following processes: selecting static objects included in multiple layers received from a lidar sensor; selecting a target layer from the multiple layers; defining a shape formed by a plurality of points included in each of the selected target layers; setting an equation corresponding to the shape; applying regression to the equation to select the layer with the minimum error as a reference layer; and removing the corresponding points from the remaining layers when the distance between the contour in the reference layer and the points in the remaining layers is longer than the reference distance.

[0014] According to another aspect of this disclosure, a method for selectively removing specific points included in multiple layers is provided. The method includes: a target layer selection step, selecting a target layer from multiple layers received from a lidar sensor; a shape determination step, determining a shape formed by a plurality of points included in each of the target layers; a reference layer selection step, selecting one of the target layers as a reference layer based on the shape; and a specific point removal step, comparing points included in the reference layer with points included in remaining layers of the target layers that were not selected as reference layers, to remove points from the remaining layers that have low correlation with a predefined static object. The above steps of this method can be performed by a signal processor.

[0015] According to another aspect of this disclosure, a non-transitory computer-readable recording medium is provided, comprising program instructions executable by a processor, the non-transitory computer-readable recording medium comprising: program instructions for selecting a target layer from multiple layers received from a lidar sensor; program instructions for determining a shape formed by a plurality of points included in each of the target layers; program instructions for selecting one of the target layers as a reference layer based on the shape; and program instructions for comparing the points included in the reference layer with points included in the remaining layers of the target layers that were not selected as reference layers, to remove points from the remaining layers that have low correlation with a predefined static object. Attached Figure Description

[0016] The accompanying drawings, which are included, incorporated in, and constitute a part of this application, to provide a further understanding of this disclosure, illustrate embodiments of the disclosure and provide a description for explaining the principles of the disclosure. In the drawings:

[0017] Figure 1 An apparatus is shown for selectively removing specific points included in a multilayer according to an embodiment of the present disclosure.

[0018] Figure 2 A method for selectively removing specific points included in a multilayer, according to this disclosure, is shown, executed by a signal processor.

[0019] Figure 3 The diagram illustrates the straight line generation step and the point symbol determination step in the shape determination process. In the point symbol determination step, the symbol of the point located between the start and end points is determined based on the point distribution location.

[0020] Figure 4 This illustrates the shape determination steps for determining the shape formed by the points included in each of the target layers using the average distance and distance variance between the straight connecting lines and each point.

[0021] Figure 5 This shows the error generated after performing regression in the reference layer selection step; and

[0022] Figure 6 This illustrates the process of removing specific points based on the distance between points in the reference layer and the remaining layers. Detailed Implementation

[0023] It should be understood that the term "vehicle" or "of a vehicle," or other similar terms as used herein, includes motor vehicles, generally such as passenger cars including sport utility vehicles (SUVs), buses, trucks, various commercial vehicles, water transport vehicles including various small boats and ships, aircraft, etc., and includes hybrid vehicles, electric vehicles, plug-in hybrid vehicles, hydrogen-powered vehicles, and other alternative fuel vehicles (e.g., fuels derived from resources other than petroleum). As referred to herein, a hybrid vehicle is a vehicle with two or more power sources, such as a gasoline-powered and electric vehicle.

[0024] The terminology used herein is for the purpose of describing particular embodiments only and is not intended to limit this disclosure. As used herein, the singular forms “a,” “an,” and “the” are intended to include the plural forms as well, unless the context clearly indicates otherwise. It will be further understood that, when used in this specification, the terms “comprising” and / or “including” specifically describe the presence of stated features, integers, steps, operations, elements, and / or components, but do not exclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof. As used herein, the term “and / or” includes any and all combinations of one or more of the associated terms listed. Throughout this specification, unless explicitly stated otherwise, the word “comprising” and variations such as “including” or “including” will be understood to imply inclusion of stated elements but not exclusion of any other elements. Additionally, the terms “unit,” “device,” “apparatus,” and “module” described herein refer to a unit for performing at least one function or operation and may be implemented by hardware components or software components and combinations thereof.

[0025] Furthermore, the control logic of this disclosure can be implemented as a non-transitory computer-readable medium containing executable program instructions that run by a processor, controller, etc. Examples of computer-readable media include, but are not limited to, ROM, RAM, optical disc (CD)-ROMs, magnetic tape, floppy disks, flash drives, smart cards, and optical data storage devices. The computer-readable medium can also be distributed across a network-connected computer system so that the computer-readable medium is stored and operated in a distributed manner, for example, by a telematics server or a controller area network (CAN).

[0026] To fully understand this disclosure, its operational advantages, and the objectives achieved by implementing it, reference should be made to the accompanying drawings illustrating exemplary embodiments of the disclosure and the description therein.

[0027] Preferred embodiments of the present disclosure will be described in detail below with reference to the accompanying drawings. In the drawings, the same or similar elements are indicated by the same reference numerals.

[0028] Figure 1 An apparatus is shown for selectively removing specific points included in a multilayer according to an embodiment of the present disclosure.

[0029] Reference Figure 1 The apparatus 100 for selectively removing specific points included in a multilayer according to the present disclosure includes a lidar sensor 110 and a signal processor 120.

[0030] The lidar sensor 110 radiates laser pulses onto the target object 10 and receives laser pulses reflected from the target object 10, thereby collecting multi-layer data (hereinafter referred to as "multi-layer data"), which is sensor information on a transverse plane defined at predetermined intervals in the vertical direction.

[0031] The signal processor 120 uses the multi-layer data received from the lidar sensor 110 to perform the following processes: (1) selecting a static object; (2) selecting a target layer from the multi-layer data; (3) defining a shape (hereinafter referred to as a "shape") formed by a plurality of points included in each of the selected target layers; (4) setting an equation corresponding to the shape defined in each of the target layers and applying regression to the equation to select the layer with the minimum error as the reference layer; and (5) comparing the points in the reference layer with those in the remaining layers, removing the corresponding point from the corresponding layer when the distance between a segment of the reference layer and a point in the remaining layer is longer than the reference distance. Here, a segment refers to an imaginary line connecting the points included in the reference layer.

[0032] In some embodiments, the operation of the apparatus 100 for selectively removing specific points included in multiple layers according to the present disclosure can be performed by a conventional lidar system including a lidar sensor and a controller. However, the controller of a conventional lidar system does not perform the functions performed by the signal processor 120 of the present disclosure. Therefore, if additional installation programs are installed in the conventional controller to perform the functions performed by the signal processor 120 of the present disclosure, the apparatus 100 for selectively removing specific points included in multiple layers can be replaced by a conventional lidar system.

[0033] Figure 1 The device shown for selectively removing specific points included in multiple layers is preferably installed in a vehicle.

[0034] The steps performed by the signal processor 120 will be described below.

[0035] Figure 2 A method for selectively removing specific points included in a multilayer, according to this disclosure, is shown, executed by a signal processor.

[0036] Reference Figure 2The method 200 for selectively removing specific points included in multiple layers according to this disclosure includes a static object definition step 210, a target layer selection step 220, a shape determination step 230, a reference layer selection step 240, and a specific point removal step 250, all of which are executed by a signal processor 120.

[0037] In the static object definition step 210, the static object is defined based on the object's length *a* and the number of points *b* forming the object's outline. For example, when selecting a highway guardrail as the static object, the determination of the static object should not be affected by dynamic objects including vehicles. In particular, since the shape of the vehicle's outline should not be distorted, considering the length of a bus (12m) or a trailer (16.7m), the reference length of the object is preferably set to 16.7 meters (m), and the number of points forming the outline of the determined valid static object is preferably set to at least seven (b = 7). Static objects can be selected differently depending on the objective of the method 200 for selectively removing specific points included in multiple layers according to this disclosure, and the length of the object and the number of points forming the object's outline can be varied depending on the selected static object.

[0038] In target layer selection step 220, a group of layers with a higher probability of being selected as a reference layer is selected from the multiple layers collected by the lidar sensor and stacked vertically on top of each other. Essentially, layers where static objects constitute a large proportion are preferably selected as reference layers. Because the length of the static object is considered in static object definition step 210, layer selection in target layer selection step 220 can be performed based on the height of the static object as a supplement to static object definition step 210. In the case where the guardrail is a static object, it is assumed that the target layers relative to the guardrail are the two consecutive layers at the lowest position in the multiple layers.

[0039] In shape determination step 230, the shape formed by multiple points included in each of the target layers selected in target layer selection step 220 is determined. At this time, the determined shape can be classified as, for example, "line", "arc", and "unknown". Shape determination includes the following steps: generating straight connecting lines that connect the start and end points of multiple points to each other (straight connecting line generation step 230-1); determining the sign of the points based on the straight connecting lines (point sign determination step 230-2); measuring the distance between the straight connecting lines and each point to calculate the average distance and the variance of the distance (point distance average / variance calculation step 230-3); and using the average distance and the variance of the distance to determine the shape formed by the points included in the corresponding layer in the target layer (layer shape determination step 230-4).

[0040] The reference layer selection step 240 includes: an equation setting step 240-1, setting an equation corresponding to the shape formed by the points included in each of the target layers as determined in the shape determination step 230; a regression application step 240-2, applying regression to the equation; and a selection step 240-3, selecting the layer with the smallest error as the result of the applied regression as the reference layer.

[0041] In the specific point removal step 250, the distance d between the points included in the reference layer selected in the reference layer selection step 240 and the points included in the remaining layers of the target layer that were not selected as reference layers is compared with the removal reference distance Rd, and points that satisfy the condition d>Rd are removed from the corresponding layer.

[0042] The following text will describe it in detail. Figure 2 The steps are shown.

[0043] exist Figure 2 In the shape determination steps shown, the straight connection line generation step 230-1 and the dot symbol determination step 230-2 can be performed in various ways, one of which will be described below.

[0044] Figure 3 The diagram illustrates the straight line generation step and the point symbol determination step in the shape determination process. In the point symbol determination step, the symbol of the point located between the start and end points is determined based on the point distribution location.

[0045] Reference Figure 3 Straight connecting cable It is a line connecting the starting point A and the ending point B. The starting point A is the first point c1 among the points forming the contour, and the ending point B is the sixth point c6 among the points forming the contour. Figure 3 The diagram on the left shows that, excluding points c1 and c6, points c2 to c5 are located on the straight connecting line. The case where it is to the left and therefore has a positive sign (+), Figure 3 The diagram on the right shows that three of the four points, c2, c4, and c5, lie on the straight connecting line. The left side of the line and therefore has a positive sign (+), while the third point c3 lies on the straight connecting line. Therefore, the right side of the symbol has a negative sign (-).

[0046] Figure 3 The sign of the point shown is determined by the signal processor 120 that performs the following calculation process.

[0047] Connect the straight wire Rotate by 90° to obtain the vector of the rotated straight connection. The target vector that connects the starting point A and the corresponding point Ci. The inner product of , so that the sign y of the corresponding point can be represented by the following equation 1.

[0048] [Equation 1]

[0049]

[0050] Here, Ci represents the final contour point, Ni represents the contour point located between the starting point A and the ending point B, and i represents a variable.

[0051] The values ​​of the corresponding points obtained from Equation 1 are in two vectors and A positive sign is given when the angle between two vectors is less than 90°, and when the angle between two vectors is less than 90°. and An angle greater than 90° has a negative sign.

[0052] The point distribution R can be calculated based on the number of points with positive signs and the number of points with negative signs. AB And it can be based on the point distribution R AB This is to confirm that these points are concentrated on the sides of the straight connecting lines. The point distribution R used in shape determination step 230 will be described later. AB .

[0053] Figure 4 This illustrates the shape determination steps for determining the shape formed by the points included in each of the target layers using the average distance and distance variance between the straight connecting lines and each point.

[0054] The distances between the direct connecting line and each point are not difficult to calculate, so a detailed description will be omitted.

[0055] The average distance A between the straight line and each point can be obtained using Equation 2 below. AB and distance variance V AB .

[0056] [Equation 2]

[0057]

[0058]

[0059] exist Figure 4 In the middle, A Line V represents the average distance to the first reference point. Line R represents the reference variance. pnts1 Let R represent the first reference distribution. pnts1 This indicates the second reference distribution.

[0060] Reference Figure 4 Using the average distance A ABand distance variance V AB The layer shape determination step 230-4, which determines the shape formed by the points included in each of the target layers, includes a first distance reference comparison step 410, a variance reference comparison step 420, a point distribution comparison step 430, a second distance reference comparison step 440, and a determination step 450.

[0061] In the first distance reference comparison step 410, the average distance A is determined. AB Is it less than the average value A of the first reference distance? Line .

[0062] When the average distance A is determined in the first distance reference comparison step 410 AB Greater than the average value of the first reference distance A Line If not in step 410, perform variance reference comparison step 420. In variance reference comparison step 420, determine the distance variance V. AB Is it less than the reference variance V? Line .

[0063] When the distance variance V is determined in variance reference comparison step 420 AB Greater than the reference variance V Line If (no) in step 420, then perform point distribution comparison step 430. In point distribution comparison step 430, determine the point distribution R. AB Is it greater than the first reference distribution R? pnts1 Or less than the second reference distribution R pnts2 .

[0064] When the point distribution R is determined in step 430 of the point distribution comparison... AB Greater than the first reference distribution R pnts1 Or less than the second reference distribution R pnts2 (In step 430) when the second distance reference comparison step 440 is performed. In the second distance reference comparison step 440, the average distance A is determined. AB Is it greater than the average value of the second reference distance A? Unknown .

[0065] In step 450, when the average distance A is determined in the first distance reference comparison step 410... AB Less than the average value of the first reference distance A Line When (in step 410) and when the distance variance V is determined in variance reference comparison step 420 AB Less than the reference variance V Line At that time (in step 420), it is determined that the corresponding layer has a "line" shape (steps 451 and 452). Additionally, when the average distance A is determined in the second distance reference comparison step 440...AB Less than the average value of the second reference distance A Unknown If (no in step 440), it is determined that the corresponding layer has an "arc" shape (step 453). Additionally, when the point distribution R is determined in point distribution comparison step 430... AB The first reference distribution R pnts1 With the second reference distribution R pnts2 The value between (no in step 430) and when the average distance A is determined in the second distance reference comparison step 440. AB Greater than the average value of the second reference distance A Unknown At that time (in step 440), it is determined that the corresponding layer has an "unknown" shape (steps 455 and 454).

[0066] To help understand Figure 4 The process shown assumes that the average first reference distance A Line Reference variance V Line First reference distribution R pnts1 Second reference distribution R pnts2 The average distance A between the second reference and the second reference Unknown The values ​​were 0.35m, 0.01m, 85%, 15%, and 1m, respectively.

[0067] In the first distance reference comparison step 410, based on the average distance A AB This determines whether the layers to be compared have a "line" shape. At this point, the baseline is determined as the distance from the average value A. AB Less than 0.35m, i.e., the average value of the first reference distance A Line .

[0068] In the variance reference comparison step 420, based on the distance variance V AB Re-determine based on the average distance A AB It has been determined whether layers that do not have a "linear" shape possess a "linear" shape. At this point, the baseline is determined as the distance variance V. AB Less than 0.01, i.e., the reference variance V Line .

[0069] As described above, the first distance reference comparison step 410 and the variance reference comparison step 420 determine whether the layer to be compared has a "line" shape. When it is determined that the layer does not have a "line" shape, the following process is performed.

[0070] In the point distribution comparison step 430, based on the point distribution R AB To initially determine whether layers that have been determined not to have a "line" shape have an "arc" shape, specifically, based on... Figure 3 Are the points shown clustered along the straight connecting line? left side or straight connecting line To the right. At this point, based on either of the two symbols, the first point reference distribution R... pnts1 It was set to 85%, while the second point reference distribution R pnts2 It was set to 15%. In other words, the condition for determining that the corresponding layer may have an "arc" shape is the point distribution R. AB Greater than 85% or less than 15%. If this condition is not met, it is also determined that layers that have been determined not to have a "line" shape do not have an "arc" shape. Therefore, the corresponding layer is determined to have an "unknown" shape (step 455).

[0071] In the second distance reference comparison step 440, the average distance A of the layer that may have an "arc" shape, which was preliminarily determined in the point distribution comparison step 430, is determined. AB Is it greater than the average value of the second reference distance A? Unknown To ultimately determine whether the layer has an "arc" shape. Although the average distance A of the corresponding layer has been determined. AB The length is greater than 0.35m, i.e., the average value A of the first reference distance. Line However, in the first distance reference comparison step 410 before the second distance reference comparison step 440, the average distance A AB It is not infinitely long. Therefore, determine the average distance A. AB The upper limit is used to ultimately determine whether the corresponding layer has an "arc" shape. This is based on the average value of the second reference distance A. Unknown If the average distance A of the corresponding layer is set to 1m, AB If the depth is between 0.35m and 1m, the corresponding layer is determined to have an “arc” shape (step 453); otherwise, the corresponding layer is determined to have an “unknown” shape (step 454).

[0072] In passing Figure 4 After the process described above determines that the layer has an "arc", "line", or "unknown" shape, a reference layer is selected as follows. Since selecting a layer with an "unknown" shape as the reference layer is meaningless, the process of selecting an "arc" or "line" shaped layer as the reference layer will be described below.

[0073] Figure 5 This shows the error that occurs after performing regression in the reference layer selection step.

[0074] Reference Figure 5 The left-hand "arc" in the diagram is modeled using a quadratic equation, and the right-hand "line" is modeled using a linear equation, as shown in Equation 3 below.

[0075] [Equation 3]

[0076] y = a2x 2 +a1x+a0

[0077] y = a1x + a0

[0078] In principle, among multilayer structures with defined shapes and models, the model with the minimum error after performing regression is selected as the reference layer. More precisely, the model with the minimum sum of squared errors is selected as the reference layer.

[0079] Equation 4 below is used to calculate the sum of squared errors S for determining a layer with an "arc" shape, while Equation 5 below is used to calculate the sum of squared errors S for a layer with a "line" shape.

[0080] [Equation 4]

[0081]

[0082]

[0083]

[0084]

[0085]

[0086] [Equation 5]

[0087]

[0088]

[0089]

[0090]

[0091] Equations 4 and 5 above are applied to the corresponding layers, such that the layer with the smallest sum of squared errors S is selected as the reference layer. Subsequently, specific point removal step 250 is performed as follows.

[0092] Figure 6 This illustrates the process of removing specific points based on the distance between points in the reference layer and the remaining layers.

[0093] Figure 6 The image on the left shows a situation where guardrails and shrubs coexist. Figure 6 The middle diagram shows the distance value di calculated after sorting the points along the X-axis, while Figure 6 The diagram on the right shows the result obtained by removing points in the remaining layer when the distance between the line segment of the reference layer and each point in the remaining layer is greater than the removal distance Rd.

[0094] In this embodiment, we assume the static object is a guardrail, so the removal reference distance Rd is set to 0.5m. However, the removal reference distance can be set differently depending on the static object to be identified. When the distance between the reference layer and each point in the remaining layer is greater than 0.5m, i.e., when the removal reference distance Rd is reached, the corresponding point is removed from the remaining layer.

[0095] As is evident from the above description, the selective removal of vehicles and methods from specific points within multiple layers completely overcomes the limitations of traditional single-layer techniques. Furthermore, the accurate identification of the shape of static objects improves the performance of classification or tracking algorithms during LiDAR signal processing. In particular, matching LiDAR layer data with high-definition maps minimizes errors, thereby ensuring the driving safety of autonomous vehicles.

[0096] While this disclosure has been specifically shown and described with reference to exemplary embodiments thereof, these embodiments are provided for illustrative purposes only and are not intended to limit the scope of the disclosure. Furthermore, it will be apparent to those skilled in the art that various modifications, additions, and substitutions may be made without departing from the scope and spirit of the disclosure as disclosed in the appended claims.

Claims

1. A vehicle comprising: LiDAR sensor; as well as A signal processor configured to: select static objects included in multiple layers received from the lidar sensor, wherein the static objects have been clustered into a single object along with another object; select a target layer from the multiple layers received from the lidar sensor that includes at least a portion of the static objects; and define a shape formed by points included in each of the selected target layers. A reference layer is selected in the target layer based on the shape formed by the points; and at least one point associated with the static object, belonging to the other object, is removed from at least one of the remaining layers based on the distance and reference distance between the contour formed by the reference layer and each point in the remaining layers.

2. A method for selectively removing specific points included in a multilayer, said multilayer being received from a lidar sensor, the method comprising: The step of determining static objects included in the multilayer received from the lidar sensor, wherein the static objects have been clustered together with another object into a single object; The target layer selection step involves selecting a target layer that includes at least a portion of the static object from the multiple layers received from the lidar sensor by a signal processor. The shape determination step involves using the signal processor to determine the shape formed by a plurality of points included in each of the target layers; The reference layer selection step involves the signal processor selecting one of the target layers as a reference layer based on the shape; and The specific point removal step involves comparing points included in the reference layer with points included in the remaining layers of the target layer that were not selected as the reference layer, using the signal processor, to remove at least one point associated with the static object from the remaining layers, the at least one point belonging to the other object.

3. The method according to claim 2, wherein, The static object is defined with reference to its length and the number of points that form its outline.

4. The method according to claim 2, wherein, The target layer is selected based on the height of the static object.

5. The method according to claim 2, wherein, The shape determined in the shape determination step is one of a "line" shape, an "arc" shape, and an "unknown" shape.

6. The method according to claim 5, wherein, The shape determination step includes: The straight connection generation step generates a straight connection that connects the start and end points of each point included in the target layer. The point symbol determination step involves determining the symbol of the point based on the straight connecting line to calculate the point distribution; The point distance average / variance calculation steps involve measuring the distance between the straight connecting line and each point to calculate the average distance and the variance of the distance; and The layer shape determination step uses the average distance and the variance of the distance to determine the shape formed by the points included in the corresponding layer in the target layer.

7. The method according to claim 6, wherein, The steps for determining the dot symbol include: Rotate the straight connecting line by 90°; and Obtain the inner product of the vector of the rotated straight connection line and the target vector that connects the starting point and the target point. The value of the target point is determined to have a positive sign when the angle between the vector of the rotated straight connecting line and the target vector is less than 90°, and a negative sign when the angle is greater than 90°.

8. The method according to claim 6, wherein, The layer shape determination step includes: The first distance reference comparison step determines whether the average distance is less than the first reference average distance. The variance reference comparison step determines whether the distance variance is less than the reference variance. If the average distance is determined to be greater than the first reference average distance in the first distance reference comparison step, the variance reference comparison step is executed. The point distribution comparison step determines whether the point distribution is greater than the first point reference distribution or less than the second point reference distribution. The point distribution comparison step is executed when the distance variance is determined to be greater than the reference variance in the variance reference comparison step. The second distance reference comparison step determines whether the average distance is greater than the second reference average distance. When the point distribution comparison step determines that the point distribution is greater than the first point reference distribution or less than the second point reference distribution, the second distance reference comparison step is executed. The determination steps are as follows: when the average distance is determined to be less than the first reference average distance in the first distance reference comparison step and when the distance variance is determined to be less than the reference variance in the variance reference comparison step, the corresponding layer is determined to have a "line" shape; when the average distance is determined to be less than the second reference average distance in the second distance reference comparison step, the corresponding layer is determined to have an "arc" shape; and when the point distribution is determined to be a value between the first point reference distribution and the second point reference distribution in the point distribution comparison step and when the average distance is determined to be greater than the second reference average distance in the second distance reference comparison step, the corresponding layer is determined to have an "unknown" shape.

9. The method according to claim 2, wherein, The reference layer selection step includes: The equation setting step involves setting equations corresponding to the shapes of each of the target layers determined in the shape determination step. The regression application step involves applying regression to the equation; and In the selection step, the layer with the smallest error as the result of applying the regression is selected as the reference layer.

10. The method of claim 9, wherein, The equation setting steps include: When the corresponding layer in the target layer has an "arc" shape, a quadratic equation is set; and When the corresponding layer in the target layer has a "line" shape, a linear equation is set.

11. The method according to claim 9, wherein, In the selection step, the layer with the smallest error is the layer with the smallest sum of squared errors.

12. The method according to claim 2, wherein, In the specific point removal step, a low correlation is determined when the distance between a line segment in the reference layer and each of the points included in the remaining layer is greater than the removal reference distance.

13. A non-transitory computer-readable recording medium comprising program instructions executable by a processor, the non-transitory computer-readable recording medium comprising: Program instructions for determining static objects included in multiple layers received from a lidar sensor, wherein the static objects have been clustered together with another object into a single object; Program instructions to select a target layer comprising at least a portion of the static object from the multilayers received from the lidar sensor; Program instructions that determine the shape formed by multiple points included in each of the target layers; The program instruction to select one of the target layers as a reference layer based on the shape; and The program instructions for removing at least one point associated with the static object, belonging to the other object, are compared between points included in the reference layer and points included in the remaining layers of the target layer that were not selected as the reference layer.

Images