Device and method for extracting a feature point for detecting an obstacle, using a laser scanner

Abstract

Device (100) for extracting a feature point to detect an obstacle, wherein a laser scanner (120) is used, the device (100) having: a laser scanner (120) installed on the front of a moving vehicle and configured to receive laser scanner data, which has a multitude of layers, in real time; and a control element (160) which is configured to separate the laser scanner data obtained by laser scanner (120) into a multitude of layers in order to extract measurement data present in each layer and to determine feature points of the measurement data in order to classify the type of obstacle based on a multitude of stored feature points, where the control element (160) is configured, to determine a form of measurement data that is present in the layer, to save a three-dimensional coordinate value for the measurement data again, using layer information, and to calculate a distance difference up to a virtual plane in order to minimize a distance from the three-dimensional coordinate value.

Description

BACKGROUND Area of ​​the invention

[0001] The present invention relates to a device and a method for extracting a feature point to detect an obstacle, wherein a laser scanner is used, and more specifically to a device and a method for extracting a feature point to detect an obstacle, wherein a laser scanner is used which extracts the feature point present in the laser scanner data to classify a type of obstacle around a moving car body, wherein a multi-layer laser scanner is used. Description of the state of the art

[0002] Correspondingly, with the generalization of navigation systems integrated into vehicles to provide services such as road guidance, obstacle avoidance, and similar functions, ongoing research has been conducted into algorithms for detecting obstacles, such as other vehicles, pedestrians, and the like. Specifically, the vehicle incorporates sensors, such as radar, laser scanners, image sensors, ultrasonic sensors, and similar devices. Since each sensor has its advantages and disadvantages, a technology for sensor aggregation has been developed that mitigates the drawbacks of each individual sensor by utilizing two or more sensors.

[0003] Among the various types of sensors, the laser scanner has a high degree of use because it provides more accurate distance and angle information than other sensors. However, since the laser scanner data only provides angle and distance information, it can be difficult to classify obstacles, such as vehicles, pedestrians, and the like, based on this data alone.

[0004] Furthermore, from EP 2 026 096 A1 and DE 10 2011 007 133 A1, a device for extracting a feature point to detect an obstacle is already known, using a laser scanner, wherein the device comprises: a laser scanner installed on the front of a moving vehicle, configured to receive laser scanner data having a multitude of layers in real time; and a control element configured to separate the laser scanner data obtained by the laser scanner into a multitude of layers in order to extract measurement data present in each layer and to determine feature points of the measurement data in order to classify the type of obstacle based on a multitude of stored feature points. OVERVIEW

[0005] It is an object of the present invention to provide a device and a method for extracting a feature point in order to detect an obstacle, wherein a laser scanner is used which extracts the feature point in the laser scanner data in order to classify a type of obstacle which is present around a moving car body, wherein a multi-layer laser scanner is used.

[0006] The problem is solved by a device for extracting a feature point with the features of claim 1 and a method for extracting a feature point with the features of claim 7. Advantageous further developments are found in the dependent claims.

[0007] In one aspect of the present invention, a device for extracting a feature point to detect an obstacle, using a laser scanner, may comprise: a laser scanner mounted on the front of a vehicle body and configured to obtain laser scanner data, configured from a plurality of layers, in real time; and a control element configured to separate the laser scanner data obtained from the laser scanner into a plurality of layers in order to extract measurement data present in each layer and to determine the feature points of the measurement data in order to classify a type of obstacle based on a plurality of feature points previously stored.The control element is configured to determine a form of measurement data present in the layer, to restore a three-dimensional coordinate value for the measurement data using layer information, and to calculate a distance difference up to a virtual plane in order to minimize a distance from the three-dimensional coordinate value.

[0008] Furthermore, the control element can be configured to determine the number or position of measurement data present in each layer and whether the layer lacking measurement data exists. The control element can be configured to calculate a standard deviation between the moving vehicle body and the obstacle, using a mean derived from the three-dimensional coordinate value. The control element can be configured to calculate a distance difference up to a virtual plane by minimizing a distance from the coordinate values ​​and to calculate the sum of the areas of each layer for which the three-dimensional coordinate value exists.The control element can also be configured to calculate an average value for an area of ​​each layer in which the three-dimensional coordinate value is present, and to calculate a gradient of a line segment or a coefficient of a curve generated from the measurement data.

[0009] In another aspect of the present invention, a method for extracting a feature point for obstacle detection using a laser scanner may include: obtaining, by a control element, laser scanner data configured from a plurality of layers in real time from a laser scanner installed on the front of a moving car body; separating, by the control element, the laser scanner data into a plurality of layers; extracting, by the control element, measurement data present in each layer; and determining, by the control element, feature points from the measurement data to classify a type of obstacle present on the front of the car body, based on a plurality of previously stored feature points.Determining the feature points involves: determining, by the control element, a form of measurement data present in the layer; restoring, by the control element, a three-dimensional coordinate value for the measurement data, using the layer information; and calculating, by the control element, a distance difference up to a virtual plane in order to minimize a distance, from the three-dimensional coordinate value.

[0010] Determining the feature points of the measurement data can involve determining the number or position of the measurement data points present in each layer, and whether the layer in which the measurement data are missing exists. Determining the feature points of the measurement data can further involve calculating a standard deviation of a distance between the moving vehicle body and the obstacle, using a mean value derived from the three-dimensional coordinate value. Additionally, determining the feature points of the measurement data can involve calculating a distance difference up to a virtual plane by minimizing a distance from the coordinate values, and calculating the summation of an area or...of an area of ​​each layer in which the three-dimensional coordinate value is present, calculating an average value for an area of ​​each layer in which the three-dimensional coordinate value is present, and calculating a gradient of a line segment or a coefficient of a curve generated from the measurement data. BRIEF DESCRIPTION OF THE DRAWINGS

[0011] The above and other problems, features and advantages of the present invention will become more apparent from the following detailed description, which is given in conjunction with the accompanying drawings, in which: Fig. 1 is an exemplary block diagram showing a main configuration of a device for extracting a feature point to detect an obstacle, wherein a multilayer laser scanner according to an exemplary embodiment of the present invention is used; Fig. 2. An exemplary flowchart is provided to describe a method for extracting a feature point to detect an obstacle, wherein a multi-layer laser scanner according to an exemplary embodiment of the present invention is used; and Fig. Figures 3 to 10 are exemplary views which describe the method to extract the feature point in order to detect the obstacle, wherein the multilayer laser scanner is used according to the exemplary embodiment of the present invention. DETAILED DESCRIPTION

[0012] It is to be assumed that the term "vehicle" or "vehicle-like" or any other similar term as used herein is inclusive of motor vehicles in general, such as passenger cars, including sports vehicles (SUVs), buses, trucks, various commercial vehicles, watercraft, including a variety of boats and ships, aircraft and the like, and including hybrid vehicles, electric vehicles, internal combustion engine vehicles, plug-in hybrid electric vehicles, hydrogen-powered vehicles and other vehicles using alternative fuels (e.g., fuels derived from resources other than oil).

[0013] Although an exemplary embodiment is described using a plurality of units to perform the exemplary process, it should be understood that the exemplary processes can also be performed by one or a plurality of modules. Additionally, it should be understood that the term "control element" refers to a hardware device that includes memory and a processor. The memory is configured to store the modules, and the processor is specifically configured to execute these modules to perform one or more processes, which are described below.

[0014] Furthermore, the control logic of the present invention can be embedded as non-transitory, computer-readable media on a computer-readable medium containing executable program instructions that are executed by a processor, controller, or the like. Examples of computer-readable media include, but are not limited to, ROM, RAM, compact disc (CD-)ROMs, magnetic tapes, floppy disks, flash drives, smart cards, and optical data storage devices. The computer-readable recording medium can also be distributed across networked computer systems, so that the computer-readable media are stored and executed in a distributed manner, e.g., by a telematics server or a CAN network.

[0015] The terminology used herein serves only to describe individual embodiments and is not intended to limit the invention. As used herein, the singular forms "a," "an," "one," and "the" are to include the plural forms as well, unless otherwise clearly indicated in the context. Furthermore, it is to be understood that the terms "includes" and / or "including," when used in this specification, specify the presence of the listed 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 related listed terms.

[0016] Unless specifically stated otherwise or evident from the context, as used here, the term "approximately" is to be understood as within a range of normal tolerance in the field, for example, within two standard deviations from the mean. "Approximately" can be understood as within 10%, 9%, 8%, 7%, 6%, 5%, 4%, 3%, 2%, 1%, 0.5%, 0.1%, 0.05%, or 0.01% of the stated value. Unless otherwise clearly indicated by the context, all values ​​provided here are modified by "approximately".

[0017] Exemplary embodiments of the present invention are described in greater detail below with reference to the accompanying drawings. However, when describing these exemplary embodiments, technological aspects that are well known in the field, to which the present invention relates, and which are not directly related to the present invention, are omitted where possible. This serves to clarify the focus of the present invention by omitting any unnecessary description, thus avoiding obscuring the invention.

[0018] Fig. Figure 1 is an exemplary block diagram showing a main configuration of a device to extract a feature point to detect an obstacle, wherein a multilayer laser scanner is used according to an exemplary embodiment of the present invention.

[0019] With reference to Fig. 1 comprises a device 100 (hereinafter referred to as a feature point extractor 100) for extracting a feature point to detect an obstacle, wherein a multi-layer laser scanner according to an exemplary embodiment of the present invention is used, an image-capturing device 110, a laser scanner 120, an input unit 130, an output unit 140, a storage unit 150 and a control element 160. The control element 160 can be configured to operate the image-capturing device 110, the laser scanner 120, the input unit 130, the output unit 140 and the storage unit 150.

[0020] The image-capturing device 110 can be installed on the front of a moving vehicle body (i.e., a vehicle being driven) to acquire image data of the front or front side of the moving vehicle at its current location and to provide this image data to the control unit 160. Additionally, the laser scanner 120 can be positioned inside the moving vehicle to acquire laser scanner data from the front of the moving vehicle and to provide this laser scanner data to the control unit 160. The laser scanner data can contain a variety of layers, and the laser scanner can be a multi-layer scanner.Specifically, the Laserscanner 120 can include a light detection and distance measurement (LIDAR) laser radar, but is not specifically limited to this, and various types of sensors and laser scanners that correspond to it can be used.

[0021] The input unit 130 can be configured to receive numerical and textual information and to transmit a key signal input regarding the function control of the feature point extractor 100 from a plurality of functions to the control element 160. The input unit 130 can be configured as a touch pad or plate, or a keypad or plate with a general key arrangement, according to a provisioning mode of the feature point extractor 100. The input unit 130 can include a touchscreen. Specifically, the input unit 130 can be displayed on an output unit 140. According to the exemplary embodiment of the present invention, the input unit 130 can be configured by means of the touch pad or the touchscreen to improve user convenience.The output unit 140 can be configured to display screen data, for example, various menu data, external image data of the front of the moving vehicle, and similar data generated while a program is running through the operation of the control element 160. The storage unit 150 can be configured to store application programs (e.g., a program for each separation of a plurality of layers that configure the laser scanner data, a program for extracting the feature point contained in the laser scanner data, and similar programs) necessary to operate functions according to the exemplary embodiment of the present invention.

[0022] The control element 160 can be configured to separate the laser scanner data acquired by the laser scanner 120 into multiple layers, to extract measurement data present in each layer, and to determine a feature point within that measurement data to classify an obstacle type based on a multitude of previously stored feature points. More specifically, the control element 160 can be configured to determine the form of measurement data present in each layer, the number or position of measurement data in each layer, and whether a layer lacking measurement data is present.

[0023] Additionally, the control element 160 can be configured to store a three-dimensional coordinate value for the measurement data, using layer information to calculate the standard deviation of a distance between the moving vehicle and the obstacle, using an average value derived from the coordinate value. The control element 160 can be configured to calculate a distance difference up to a virtual plane to minimize a distance from the coordinate values ​​and to calculate a sum of a range of each layer in which the coordinate value is present. Furthermore, the control element 160 can be configured to calculate an average value of the range of each layer in which the coordinate value is present and to calculate a gradient of a line segment or a coefficient of a curve generated from the measurement data.The type of obstacle, which is determined from the laser scanner data, can be classified more precisely by the operations mentioned above in the control element 160.

[0024] Fig. Figure 2 is an exemplary flowchart describing a method for extracting a feature point to detect an obstacle, wherein a multilayer laser scanner according to an exemplary embodiment of the present invention is used. Fig. Figures 3 to 10 are exemplary views that describe the method for extracting the feature point in order to detect the obstacle, wherein the multilayer laser scanner is used according to the exemplary embodiment of the present invention.

[0025] With reference to Fig.At S11, control element 160 can be configured to receive the laser scanner data, which comprises the multiple layers received by laser scanner 120, installed at the front of the moving vehicle (the moving body). At S13, control element 160 can be configured to separate the laser data received from laser scanner 120 into multiple layers. These multiple layers can be configured to consist of four layers, from layer 0 to layer 3, as shown in... Fig. 5 is shown.

[0026] At S15, the control element 160 can be configured to analyze the multiple, separated layers and extract the measurement data present in each layer. Specifically, the measurement data is generated when the laser collides with the obstacle (for example, hits it, reaches it, or similar). Furthermore, at S17, the control element 160 can be configured to determine the feature point of the measurement data and compare this feature point with feature points stored in the memory unit 150. The stored feature point definitions are shown in Table 1 below, and these feature points can be mapped to a reference value (e.g., compared to it) to classify the obstacle, based on the measurement data, as either a vehicle or a pedestrian. Table 1 Key features Meaning f1 an L-shape, which best represents the distribution of the points f2 a width along the axis of a straight line, which best represents the distribution of the points f3 a height along a line axis of a straight line, which best represents the distribution of the points f4 An area or region along the axis of a straight line that best represents the distribution of the points. f5 the number of data points in layer 0 f6 the number of data points in layer 1 f7 the number of data from layer 2 f8 the number of data points in layer 3 f9 the number of layers in which the measurement data is available f10 Standard deviation of a distance from the average point of (x, y, z) values f11 Planarity of the (x, y, z) values f12 a summation of the areas of each layer f13 an average of the areas of each layer f14 Gradient of a line segment (linear expression), which best represents the number of points f15 Coefficient of a quadratic term of a curve (quadratic expression), which best represents the number of points f16 Coefficient of a linear term of a curve (linear expression) that best represents the number of points

[0027] More specifically, the control element 160 can be configured to analyze the extracted measurement data and determine a feature point corresponding to f1. Specifically, if the extracted measurement data (reference digit 10) forms a shape similar to an L-shape, which corresponds to a reference digit 12 of the Fig. 3 is, can r min The data can be extracted by using a point (reference digit 11) that is closest to an origin within the measurement data. Additionally, the control element 160 can be configured to generate a straight line that best represents the distribution of the measurement data, such as a reference digit 13. Fig.4, and to calculate a width, height, and area (corresponding to f2, f3, and f4 of Table 1) represented by the measurement data based on reference digit 13, in order to extract a feature that is slightly varied based on the direction of the obstacle. As a result, the control element 160 can be configured to determine an approximate dimension of the detected obstacle.

[0028] Additionally, the control element 160 can be configured to determine the number of measurement data points present in layer 0 to layer 3, which is the multitude of layers, as described in Fig. 5 is shown so that they correspond to the feature points of f5 to f8 of Table 1, and the control element 160 can be configured to determine the number of layers in which the measurement data are present, as shown in Fig.6 is shown, so that they correspond to the feature point of f9 in Table 1. For example, if a moving pedestrian is detected as the obstacle, the different number of measurement data for each layer can be determined, as shown in Fig. 5 is shown, and the layer in which the measurement data is not present may be based on the height or size of the pedestrian.

[0029] The control element 160 can be configured to restore the x, y, and z coordinates of the obstacle, using information layers 0 to 3, as described in Fig.Figure 7 is shown to correspond to the feature point of f10 in Table 1. Specifically, although the drawing shows that an angle formed by each layer is approximately 0.8°, the angle is not necessarily limited to this and can be modified by those skilled in the art. The control element 160 can be configured to use an average of the restored x, y, and z values ​​as a standard deviation of the distance between the moving vehicle and the obstacle. Furthermore, to restore the x, y, and z coordinates of the obstacle, the control element 160 can be configured to use the following equation 1. {x=r*cos(θ)*cos(−1.2+0.8L)y=r*sin(θ)*cos(−1.2+0.8L)z=r*sin(θ)(−1.2+0.8L)(r,θ,L)= where r represents a distance of a distance between the laser scanner 120 and a data set within the measurement data to the plane, θ represents an angle which is produced by the respective layers based on the laser scanner 120, and L represents a position of the layers. f10=1np∑n=1np‖(x,y,z)n−(x,y,z)mean‖2 where n p represents the number of measurement data points.

[0030] Additionally, the control element 160 can be configured to derive the planarity of the calculated x, y, and z coordinate values ​​to correspond to the feature point f11 of Table 11. Therefore, the control element 160 can be configured to calculate a distance difference up to a plane 14 that has a minimum distance from the points that are the reference numbers 20a to 20f, which are formed by an obstacle 20, as shown in Fig. 8 is shown, using the following equation 3. f11=1 / np∑n=1np(xn−Pln)2 where x n an area in which the measurement data is distributed, and P 1,n a plane which allows the distance between the measurement data that can be detected to be at a minimum.

[0031] The control element 160 can be configured to calculate a sum of the area or region of each layer, using the x, y, and z coordinate values, as described in Fig. 9 is shown, so that they correspond to the feature point of f12 in Table 1, and to calculate a mean value of the area or range of each layer, using the x, y, and z coordinate values, as shown in Fig. 9 is shown, so that it corresponds to the feature point of f13 in Table 1. Furthermore, the control element 160 can be configured to calculate a gradient of a line segment or a coefficient of a curve, using the number of measurement data points available in layer 0 to layer 3, such as L0 to L3, as shown in Fig. 10 is shown, so that they correspond to f14 to f16 of Table 1 according to the feature point.

[0032] More specifically, if the measurement data present in the layer is essentially similar to the L-shape, such as the reference digit 10 of the Fig. 3. Control element 160 can be configured to recognize that the measurement data is generated by the vehicle. Additionally, since the measurement data present in the layer can be present in every layer, as described in Fig.As shown in Figure 5, and since the number of measurement data points present in each layer may differ, the control element 160 can be configured to recognize that the measurement data is generated by the pedestrian 20. Therefore, in the exemplary embodiment of the present invention, the control element can be configured to determine the type of obstacle positioned in front of the moving vehicle and can determine the number of obstacles using the feature points extracted by analyzing the measurement data present in the laser scanner data. According to the exemplary embodiment of the present invention, the feature point present in the laser scanner data obtained by the multi-layer laser scanner can be extracted to more precisely classify the type of obstacle present around the moving vehicle body.

[0033] The device and method for extracting the feature point to detect the obstacle, using the laser scanner, have been described above with reference to the exemplary embodiment of the present invention. The exemplary embodiments of the present invention have been disclosed in this specification and the accompanying drawings, and special terms have been used, but only in a general sense to easily describe the technical content of the present invention and to aid understanding of the present invention; they do not limit the scope of the present invention. SYMBOL OF EACH OF THE ELEMENTS IN THE DRAWINGS 110 CAMERA 120 laser scanners 130 INPUT UNIT 140 OUTPUT UNIT 150 STORAGE UNIT 160 CONTROL ELEMENT S11 RECEIVE LASER DATA S13 SEPARATE LAYER S15 EXTRACT MEASUREMENT DATA FOR EACH LAYER S17 CHECK FEATURE

Claims

[1] Device (100) for extracting a feature point to detect an obstacle, wherein a laser scanner (120) is used, the device (100) having: a laser scanner (120) installed on the front of a moving vehicle and configured to receive laser scanner data, which has a multitude of layers, in real time; and a control element (160) which is configured to separate the laser scanner data obtained by laser scanner (120) into a multitude of layers in order to extract measurement data present in each layer and to determine feature points of the measurement data in order to classify the type of obstacle based on a multitude of stored feature points, where the control element (160) is configured, to determine a form of measurement data that is present in the layer, to save a three-dimensional coordinate value for the measurement data again, using layer information, and to calculate a distance difference up to a virtual plane in order to minimize a distance from the three-dimensional coordinate value. [2] Device according to claim 1, wherein the control element (160) is configured to determine a number or position of the measurement data present in each layer and to determine whether the layer in which the measurement data are not present exists. [3] Device according to claim 1, wherein the control element (160) is configured to calculate a standard deviation of a distance between the moving vehicle and the obstacle, using an average value derived from the three-dimensional coordinate value. [4] Device according to claim 1, wherein the control element (160) is configured to calculate a summation of an area or region of each layer in which the three-dimensional coordinate value is present. [5] Device according to claim 1, wherein the control element (160) is configured to calculate an average value for an area of ​​each layer in which the three-dimensional coordinate value is present. [6] Device according to claim 1, wherein the control element (160) is configured to calculate a gradient of a line segment or a coefficient of a curve generated from the measurement data. [7] Method for extracting a feature point to detect an obstacle, wherein a laser scanner (120) is used, the method comprising: Received, by a control element (160), from laser scanner data, which has a multitude of layers, in real time from the laser scanner (120), which is installed on the front of a moving vehicle; Separation, by the control element (160), of the laser scanner data into a multitude of layers; Extracting, through the control element (160), measurement data present in each layer; and Determine, by the control element (160), feature points of the measurement data to classify a type of obstacle present at the front of the moving vehicle, based on a multitude of stored feature points, which includes determining the characteristic points: Determine, by means of the control element (160), a form of measurement data which is present in the layer, Re-saving, by the control element (160), of a three-dimensional coordinate value for the measurement data, using the layer information; and Calculate, by means of the control element (160), a distance difference up to a virtual plane in order to minimize a distance from the three-dimensional coordinate value. [8] Method according to claim 7, wherein determining the feature points of the measurement data includes: Determine, by means of the control element (160), the number or position of the measurement data which are present in each layer, and whether the layer in which the measurement data are not present exists. [9] Method according to claim 7, wherein determining the feature points of the measurement data includes: Calculate, by the control element (160), a standard deviation of a distance between the moving vehicle and the obstacle, using an average value derived from the three-dimensional coordinate value. [10] Method according to claim 7, wherein determining the feature points of the measurement data includes: Calculate, by means of the control element (160), a summation of an area of ​​each layer in which the three-dimensional coordinate value is present. [11] Method according to claim 7, wherein determining the feature points of the measurement data includes: Calculate, by the control element (160), an average value for an area of ​​each layer in which the three-dimensional coordinate value is present. [12] Method according to claim 7, wherein determining the feature points of the measurement data includes: Calculate, by the control element (160), a gradient of a line segment or a coefficient of a curve generated from the measurement data. [13] Non-transitory, computer-readable medium containing program instructions which are executed by a control element (160), wherein the computer-readable medium has: Program instructions which receive laser scanner data which have a multitude of layers, in real time from a laser scanner (120) which is installed on the front of a moving vehicle; Program instructions that separate the laser scanner data into a multitude of layers; Program instructions that extract the measurement data present in each layer; Program instructions that determine feature points of the measurement data in order to classify a type of obstacle present on the front of the moving vehicle, based on a large number of stored feature points; Program instructions that determine a form of measurement data present in the layer; Program instructions that restore a three-dimensional coordinate value for the measurement data, using layer information; and Program instructions that calculate a distance difference up to a virtual plane, in order to minimize a distance, from the three-dimensional coordinate value. [14] Non-transitory, computer-readable medium according to claim 13, which further comprises: Program instructions that specify a number or position of the measurement data present in each layer, and determine whether the layer in which the measurement data is not present exists. [15] Non-transitory, computer-readable medium according to claim 13, which further comprises: Program instructions that calculate a standard deviation of a distance between the moving vehicle and the obstacle, using an average value derived from the three-dimensional coordinate value.

Images