Distance distortion correction method and device for improving laser radar detection accuracy

Abstract

The distance distortion correction method for improving the detection precision of a laser radar according to an embodiment of the present application comprises the following steps: obtaining measurement data obtained by detecting a target by the laser radar; calculating a range of distance offsets by comparing the measurement data and reference data; calculating a distance offset within the range of distance offsets based on a correlation coefficient between the measurement data and the reference data; and correcting the measurement data using the calculated distance offset.

Description

Technical Field

[0001] This invention relates to a method and apparatus for correcting distance distortion in lidar. Background Technology

[0002] Distortion is unavoidable in the signals received by LiDAR (Light Detection and Ranging) systems. For example, during target detection, distance offset occurs due to: delays in the LiDAR transmitter caused by the difference between the trigger time and the actual transmission time, and transmission delays after signal reception. As another example, distortion occurs due to the distance resolution during the Time-to-Digital-Converter (TDC) process, resulting in a plane appearing as multiple cut arcs.

[0003] Distance offset and the cut arc will deform a plane into a curved surface, resulting in inaccurate distance calculations. For example, in LiDAR applications such as autonomous driving, where precise target detection information is required, distance distortion can be a fatal problem.

[0004] On the other hand, even with data acquired from the same lidar, the distance offset values ​​within the lidar will vary depending on each channel. Therefore, distortion can occur not only in a single channel (single-channel) but also in planes with significant vertical deviations in the signals from multiple channels (multi-channel).

[0005] Depending on the vertical distance and horizontal angle between the flat target and the lidar, the cut arc observed in the lidar will exhibit different shapes. As mentioned above, lidar is affected by distance and angle; therefore, changes in distance due to the movement of the field of view (FOV) or the lidar will cause distortion, resulting in the same plane being deformed into other shapes. Summary of the Invention

[0006] In this invention, a method for correcting distance offset (amount) and cut arc in distance distortion is proposed, and a correction method that can correct not only the distortion of a single channel, but also the distance offset (amount) and cut arc of multiple channels is further proposed.

[0007] The problem the invention aims to solve

[0008] The purpose of this invention is to provide a distance distortion correction method and apparatus to further improve the detection performance of lidar.

[0009] The purpose of this invention is to provide a method and apparatus for correcting distance offset and distance distortion of cut arcs not only on a single channel but also on multiple channels.

[0010] means for solving problems

[0011] An embodiment of the present invention provides a range distortion correction method for improving the detection accuracy of lidar, comprising: a step of acquiring measurement data obtained by the lidar detecting a target; a step of estimating a range of distance offset by comparing the measurement data with reference data; a step of estimating the distance offset within the range of distance offset based on the correlation coefficient between the measurement data and the reference data; and a step of correcting the measurement data using the estimated distance offset.

[0012] The step of calibrating the measurement data may include: calibrating the measurement data acquired in each channel using the distance offset calculated in each channel.

[0013] The step of estimating the range of distance offsets may include: estimating the range of distance offsets using the reference data set for each distance offset.

[0014] The step of estimating the distance offset range may include: identifying the distance offset range where the difference between the distance offset of a specific pixel in the measurement data and a specific pixel in the reference data is the smallest.

[0015] The step of estimating the distance offset may include: within the range of the distance offset, estimating the distance offset in which the correlation coefficient between the measured data and the reference data is the highest.

[0016] The step of correcting the measurement data may include: correcting the measurement data by subtracting the calculated distance offset from the measurement data.

[0017] An embodiment of the present invention provides a range distortion correction device for improving the detection accuracy of lidar, comprising a processor that performs the following functions: acquiring measurement data obtained by the lidar detecting a target; calculating a range offset range by comparing the measurement data with reference data; calculating a range offset within the range offset based on a correlation coefficient between the measurement data and the reference data; and correcting the measurement data using the calculated range offset.

[0018] The processor can use the distance offset calculated on each channel to correct the measurement data acquired on each channel.

[0019] The processor can use the reference data set for each distance offset to calculate the range of the distance offset.

[0020] The processor is able to identify the minimum distance offset range between a specific pixel in the measurement data and a specific pixel in the reference data.

[0021] The processor is able to calculate the distance offset within the range where the correlation coefficient between the measured data and the reference data is the highest.

[0022] The processor can correct the measurement data by subtracting the calculated distance offset from the measurement data.

[0023] A method for correcting a cut arc for improving the detection accuracy of a lidar according to an embodiment of the present invention includes: acquiring measurement data obtained by the lidar detecting a target; calculating a horizontal angle between the lidar and the target using the measurement data; identifying errors in the measurement data using an error map related to errors caused by the range resolution of the lidar; and correcting the measurement data using the identified errors.

[0024] The correction method may further include the step of generating the error map by mapping the error to each coordinate within the detection area of ​​the lidar.

[0025] The step of generating the error map may include generating the error map for each horizontal angle between the lidar and the target.

[0026] The step of correcting the measurement data may include: correcting the measurement data for each pixel using the error identified for each pixel.

[0027] The step of estimating the horizontal angle may include: estimating the horizontal angle using a regression line, wherein the regression line is calculated from the measured data using a linear regression method.

[0028] An arc correction device for improving the detection accuracy of a lidar according to an embodiment of the present invention includes a processor that performs the following functions: acquiring measurement data obtained by the lidar detecting a target; using the measurement data to calculate the horizontal angle between the lidar and the target; identifying errors in the measurement data using an error map related to errors caused by the distance resolution of the lidar; and correcting the measurement data using the identified errors.

[0029] The processor is able to generate the error map by mapping the error to each coordinate within the detection area of ​​the lidar.

[0030] The processor is capable of generating the error map for each horizontal angle between the lidar and the target.

[0031] The processor can use the error identified for each pixel to correct the measurement data for each pixel.

[0032] The processor can calculate the horizontal angle using a regression line, which is calculated from the measured data using a linear regression method.

[0033] The processor can use the linear regression method to calculate a regression line from the measured data.

[0034] Invention Effects

[0035] According to one embodiment of the present invention, since the process is divided into two steps: a step of compressing the range of distance offset using reference data; and a step of calculating the accurate distance offset within the range of the compressed distance offset using a correlation coefficient, the distance offset can be calculated more accurately.

[0036] According to one embodiment of the present invention, since the distance offset is accurately calculated and corrected on each channel, accurate correction of the overall measurement data can be achieved.

[0037] According to one embodiment of the present invention, since the range of distance offset is compressed, the final distance offset can be calculated more quickly and with less data processing.

[0038] According to one embodiment of the present invention, the accurate shape of the target can be restored by correcting (compensating) the error of the cut arc, and more accurate distance information can be obtained.

[0039] According to one embodiment of the present invention, since the error is accurately calculated on each channel and corrected using the error, accurate correction of the overall measurement data can be achieved. Attached Figure Description

[0040] Figure 1 This is a diagram illustrating the distortion caused by distance offset in an embodiment of the present invention.

[0041] Figure 2 These are bird's-eye view and front view diagrams showing distance distortion measurement data according to an embodiment of the present invention.

[0042] Figure 3This is a block diagram illustrating the structure of a lidar and correction device according to an embodiment of the present invention.

[0043] Figure 4 This is a flowchart illustrating the operation of a correction device for correcting distance offset according to an embodiment of the present invention.

[0044] Figure 5 This is a graph illustrating reference data and measurement data of an embodiment of the present invention.

[0045] Figure 6 This is a diagram illustrating the range of calculated distance offsets according to an embodiment of the present invention.

[0046] Figure 7 This is a diagram illustrating the calculation of distance offset according to an embodiment of the present invention.

[0047] Figure 8 This is a graph showing the results of correcting the measurement data on each channel.

[0048] Figure 9 This is a diagram illustrating the distortion caused by the cut arc in an embodiment of the present invention.

[0049] Figure 10 This is a flowchart illustrating the operation of a correction device for correcting distortion caused by a cut arc, according to an embodiment of the present invention.

[0050] Figure 11 This is a diagram illustrating the calculation of the horizontal angle between the lidar and the target according to an embodiment of the present invention.

[0051] Figure 12 This is an error diagram of an embodiment of the present invention.

[0052] Figure 13 This is a graph showing the results of the measurement data corrected on each channel.

[0053] Figure 14 This is a graph showing the results of correcting the measured data corresponding to the horizontal angle between the lidar and the target. Detailed Implementation

[0054] Preferred embodiments of the present invention will now be described in detail with reference to the accompanying drawings. Hereinafter, they will be discussed in conjunction with the accompanying drawings. Figure 1 The detailed description disclosed herein is intended to illustrate exemplary embodiments of the invention, but is not intended to represent the only embodiments in which the invention can be implemented. For the purpose of clearly illustrating the invention in the drawings, parts unrelated to the description may be omitted, and throughout the specification, the same or similar constituent elements may use the same reference numerals.

[0055] Figure 1This is a diagram illustrating the distortion caused by distance offset in an embodiment of the present invention.

[0056] exist Figure 1 The image shows distortion due to distance offset in a lidar received signal according to an embodiment of the present invention. Figure 1 In this context, it is assumed that the horizontal field of view (θ) of a lidar in an embodiment of the present invention is... k The range is between -60 degrees and 60 degrees, and it is assumed that the actual target 1 is located at a position represented by a solid line with a vertical distance a from the lidar.

[0057] As mentioned above, the measurement data of lidar can be distorted due to distance offset for the following reasons: delay caused by the difference between the emission trigger time and the actual emission time in the lidar's transmitting unit, and signal transmission delay after signal reception, etc.

[0058] Therefore, as Figure 1 As shown by the dashed line, the lidar can acquire data of the measured target 2 that is distorted by an additional distance offset c compared to the actual target 1 (hereinafter referred to as the measurement data).

[0059] At this point, the distance offset c is a constant that is the same regardless of the pixel in a single channel. Therefore, the distance offset can be calculated by subtracting (removing) the distance offset from the LiDAR signal received in each individual channel. Distortion correction is then performed. Therefore, the distance offset needs to be accurately calculated for each channel.

[0060] Figure 2 These are bird's-eye view 210 and front view 220, which show distance distortion measurement data according to an embodiment of the present invention.

[0061] Bird's-eye view 210 is a diagram showing the point cloud along the direction from above to the ground. Relative to the entire passage, the planar target 211 appears as a thicker and distorted curved target 212 at a distance from its actual location.

[0062] In this case, it can also be confirmed through the main view 220 that the planar target 221, which is the actual target, appears as a thicker and distorted curved target 222 at a position farther than the actual position.

[0063] At this point, by referring in detail to surface target 212, it can be confirmed that multiple measurement data points formed along the horizontal direction are superimposed to form the entire data. That is, Figure 1 The target 2 shown is Figure 2The curved surface target 212 is a measurement data obtained from one channel measurement, which is then measured for each channel and displayed across all channels.

[0064] In this invention, not only can the distance offset on a single channel be calculated, but the entire measurement data can also be corrected (compensated) by calculating on each channel, thereby reducing distance distortion.

[0065] Below, refer to Figures 3 to 8 The method and apparatus for performing the above correction will be explained in detail.

[0066] Figure 3 This is a block diagram illustrating the structure of a lidar and correction device according to an embodiment of the present invention.

[0067] Reference Figure 3 In one embodiment of the present invention, the lidar 10 includes a transmitter 11 and a receiver 12, and the correction device 100 includes a processor 110.

[0068] According to one embodiment of the present invention, the lidar 10 is disposed in front of and / or behind the vehicle to detect targets within the vehicle's driving radius. In this case, there are no limitations on the placement or number of lidar 10s.

[0069] According to one embodiment of the present invention, the lidar 10 emits a pulsed light signal via a transmitting unit 11, and then receives the reflected light from the target via a receiving unit 12. The distance to the target can then be calculated by measuring the time elapsed from the transmission of the signal to the reception of the signal. At this time, the lidar 10 can classify the signals received by the receiving unit 12 according to different positions, thereby acquiring measurement data for each channel. Although in Figure 3 The diagram shows a lidar 10 consisting only of a transmitter 11 and a receiver 12. However, the lidar 10 may also include a light source, a reflector, a reflector drive motor, a signal amplification unit, a microcontroller unit (MCU), and a control unit such as an electronic control unit (ECU).

[0070] According to one embodiment of the present invention, the calibration device 100 receives measurement data measured by the lidar 10, thereby enabling the setting of a distance offset. In this case, when setting the distance offset while the lidar 10 emits a large amount of light, the calibration device 100 can be implemented as a device independent of the lidar 10, such as a computer or server. In this case, in addition to the processor 110, the calibration device 100 may further include: an input unit for receiving user input; a communication unit for sending data to or receiving data from an external device such as the lidar 10; a display unit with a display screen; and a memory for storing data or operating programs.

[0071] However, it is not limited to this. The lidar 10 itself can calculate the distance offset and correct distorted measurement data. In this case, the correction device 100 can be a microcontroller (MCU) or microprocessor installed inside and operating the lidar 10, and the correction device 100 can perform the overall operation for controlling the lidar 10.

[0072] According to one embodiment of the present invention, the calibration device 100 is capable of receiving measurement data measured by the lidar 10 and correcting (compensating) errors in the measurement data. The calibration device 100 can receive measurement data from the receiving unit 12 using CAN communication, LIN communication, or the like.

[0073] The processor 110 of one embodiment of the present invention can control at least one other component (e.g., hardware component or software component) of the correction device 100 by running software such as programs, and can perform various data processing or operations.

[0074] In one embodiment of the present invention, the processor 110 can acquire measurement data obtained by the lidar 10 detecting (measuring) a target, then calculate the range of distance offset by comparing the measurement data and reference data, then calculate the distance offset within the range of distance offset based on the correlation coefficient between the measurement data and reference data, and finally use the calculated distance offset to correct the measurement data.

[0075] In one embodiment of the present invention, the processor 110 can acquire measurement data obtained by the lidar 10 detecting a target, then use the measurement data to calculate the horizontal angle between the lidar and the target, then use an error map related to the error caused by the distance resolution of the lidar to identify the error in the measurement data, and finally use the identified error to correct the measurement data.

[0076] On the other hand, in the process where the processor 110 acquires measurement data obtained by the lidar 10 detecting a target, then calculates the range of distance offset by comparing the measurement data and reference data, then calculates the distance offset within the range of distance offset based on the correlation coefficient between the measurement data and reference data, and finally uses the calculated distance offset to correct the action of the measurement data; or, in the process where the processor 110 acquires measurement data obtained by the lidar 10 detecting a target, then calculates the horizontal angle between the lidar and the target using the measurement data, then identifies the error of the measurement data using an error map related to the error generated by the distance resolution of the lidar, and finally uses the identified error to correct the action of the measurement data, at least one of machine learning, neural network, or deep learning algorithms can be used as a rule-based algorithm or artificial intelligence algorithm to perform at least a part of data analysis, data processing, and result information generation. Examples of neural networks include models such as convolutional neural networks (CNNs), deep neural networks (DNNs), and recurrent neural networks (RNNs).

[0077] Figure 4 A flowchart illustrating the operation of a calibration apparatus according to an embodiment of the present invention is shown.

[0078] According to one embodiment of the present invention, the processor 110 is capable of acquiring measurement data obtained by the lidar 10 from detecting a target (step S10).

[0079] At this point, it is assumed that the acquired measurement data is distorted by a distance offset c. When the correction device 100 is configured as a device independent of the lidar 10, it can receive the measurement data from the lidar 10 via the communication unit. Alternatively, if the correction device 100 is installed inside the lidar 10, it can receive the measurement data from the receiving unit 12 using CAN communication, LIN communication, or the like.

[0080] According to one embodiment of the present invention, the processor 110 can estimate the range of distance offset by comparing the measured data and the reference data (step S20).

[0081] The reference data in one embodiment of the present invention refers to distorted data that occurs when there is a specific distance offset between the correct position of the target and the target.

[0082] According to the present invention, reference data can be set for each distance offset to calculate the range of distance offset of the measured data. For example, the reference data can be set in intervals of 0.01m from a distance offset of 0.01m to a distance offset of 10m. However, the range of distance offsets and the intervals of distance offsets for the reference data of the present invention are not limited thereto. In addition, the reference data can be provided by the processor 110, or it can be received from an external device.

[0083] The processor 110 calculates the range of distance offsets with the smallest difference by comparing the measured data and the reference data. At this time, distance offsets are calculated using distance information in polar coordinates for both the measured data and the reference data.

[0084] More specifically, as shown in Formula 1, in order to calculate the range of distance offset, the difference between the distance offset between the measured data and the reference data can be applied.

[0085] [Formula 1]

[0086] d n =|r m，n -r ref，n |

[0087] In Formula 1 above, d n The difference in distance offset between the measured data and the reference data in the nth pixel; r m，n : The distance between the lidar and the measurement data in the nth pixel; r ref，n : The distance between the lidar and the reference data in the nth pixel.

[0088] The difference d between the measured data and the reference data. n In this case, it can be done through d n The distance offset of the measured data is obtained by taking the distance offset of the smallest reference data. This d n The smaller the value, the more similar the distance offset between the measured data and the reference data, which can be compared with d. n The distance offset of the measured data is calculated by taking the distance offset corresponding to the smallest reference data value.

[0089] In this invention, in order to calculate a more accurate distance offset, the difference d between distance offsets is first compressed. n The range of smaller distance offset values.

[0090] [Formula 2]

[0091]

[0092] That is, processor 110 can use Formula 2 to identify the range of distance offsets where the difference between the distance offsets of a specific pixel in the measurement data and a specific pixel in the reference data is minimized. More details will be provided later. Figure 6 Please provide an explanation.

[0093] According to one embodiment of the present invention, the processor 110 is capable of calculating the distance offset within the range of the distance offset based on the correlation coefficient between the measured data and the reference data. (Step S30)

[0094] In this context, the correlation coefficient refers to the relationship between the data aspect of the measured data and the data aspect of the reference data. More specifically, the correlation coefficient is used to represent the relationship between the increase or decrease of the reference data as the measured data increases or decreases. Its value ranges from 0 to 1. The closer the value is to 1, the more consistent the data aspects of the measured data and the reference data are.

[0095] As shown below, the correlation coefficient can be determined using Formula 3.

[0096] [Formula 3]

[0097]

[0098] According to an embodiment of the present invention, the processor 110 can, within the range of the distance offset compressed in step S20 above, increase the correlation coefficient. The highest distance offset is used to calculate the final distance offset.

[0099] Refer to later Figure 7 Detailed explanation of the final distance offset with the highest correlation coefficient The situation under which calculations are being made.

[0100] According to one embodiment of the present invention, the processor 110 can utilize the calculated distance offset. To calibrate the measurement data (step S40).

[0101] According to an embodiment of the present invention, as shown in Formula 4, the processor 110 can calculate the distance offset by removing (subtracting) the calculated distance offset from the measured data. The measurement data is then corrected. At this point, the correction is performed on each pixel, ensuring that all pixels within the same channel have the same distance offset. Therefore, the processor 110 can remove the same distance offset from the measurement data of all pixels within the same channel. To calibrate the measurement data.

[0102] [Formula 4]

[0103]

[0104] In formula 4 above, The corrected (compensated) distance; rm: the distance between the lidar and the measurement data. The calculated distance offset.

[0105] Additionally, the processor 110 can use the distance offset calculated for each channel to correct the measurement data acquired in each channel. More details will be provided later. Figure 8 Please provide an explanation.

[0106] According to one embodiment of the present invention, since the distortion correction method for measurement data is performed in two steps: a step of compressing the range of distance offset using reference data; and a step of calculating the accurate distance offset within the compressed range of distance offset using a correlation coefficient, the distance offset can be calculated more accurately. Furthermore, since the distance offset can be accurately calculated for each channel and then used for correction, the entire measurement data can be accurately corrected.

[0107] Figure 5 This is a graph illustrating reference data and measurement data of an embodiment of the present invention.

[0108] exist Figure 5 In, it is shown that in Figure 4 The reference data 510 and measurement data 520 described in step S20. In the reference data 510 and measurement data 520, the horizontal axis represents the pixel index, and the vertical axis represents the distance between the lidar and the measurement data. That is, the distance between the lidar and the measurement data is shown for each pixel. At this time, the pixel index refers to the distance from the lidar to the measurement data. Figure 1 The corresponding pixels in the lidar measurement data shown are numbered from left to right.

[0109] Therefore, when roughly observing the data level, it can be confirmed that the closer to the center of the measured data, the smaller the distance between the lidar and the measured data.

[0110] At this point, only one reference data 510 is represented. However, in reality, there are multiple reference data 510s for each distance offset. From these multiple reference data 510s, the reference data 510 that is similar to or the same as the measured data 520 is identified. In order to calculate a more accurate final distance offset, in this step, the range of the distance offset is calculated based on the measured data 520 that is similar to the reference data 510.

[0111] Therefore, the finer the interval between distance offsets, the more accurately the range of distance offsets can be calculated.

[0112] For example, when the distance offset is between 0.001m and 10m and the distance offset is set to an interval of 0.001m, compared to the case where the distance offset is between 0.01m and 10m and the distance offset is set to an interval of 0.01m, the range of distance offset can be set in units of 0.001m, thus allowing for a more accurate calculation of the range of distance offset.

[0113] Figure 6 This is a diagram illustrating the range of calculated distance offsets according to an embodiment of the present invention.

[0114] As above Figure 4 As explained in step S20, the calculation of the range of distance offset utilizes the difference in distance offset between the reference data and the measured data.

[0115] exist Figure 6 In the curve graph 610, the horizontal axis represents the distance offset, and the vertical axis represents the difference in distance offset between the reference data and the measured data.

[0116] Therefore, the processor 110 can identify the range where the difference in distance offset between a specific pixel in the measurement data and a specific pixel in the reference data is the smallest. In this case, the specific pixel can be a pixel located at the center position within a single channel. However, the invention is not limited to this; the pixel can be applied in different ways, for example, calculating the difference in distance offset between a specific pixel in the measurement data and a specific pixel in the reference data for all pixels and using their average value, etc.

[0117] That is, the processor 110 can identify the range of distance offsets where the difference between the measured data and the reference data is the smallest.

[0118] Reference Figure 6 The difference in distance offset was calculated when the distance offset of the reference data was between 9.5m and 10.5m.

[0119] The difference between the distance offsets is identified as minimal at the first distance offset_a and the second distance offset_b, and the processor 110 can extrapolate the range between the first distance offset_a and the second distance offset_b as the range of the distance offsets.

[0120] According to one embodiment of the present invention, by compressing the range of distance offset, the final distance offset can be calculated more quickly with less data processing.

[0121] Figure 7 This is a diagram illustrating the calculation of distance offset according to an embodiment of the present invention.

[0122] like Figure 4 As explained in step S30, the calculation of the final distance offset is performed using a correlation coefficient.

[0123] exist Figure 7 In graph 710, the horizontal axis represents the distance offset, and the vertical axis represents the correlation coefficient between the reference data and the measured data. At this point, graph 710 is in relation to... Figure 6 Data under the same conditions, Figure 6 and Figure 7 In this context, the first distance offset_a and the second distance offset_b have the same value. Additionally, as shown in the reference... Figure 6 As explained, the range of distance offsets is compressed into a first distance offset_a and a second distance offset_b. Furthermore, to further reduce the amount of data processing, the correlation coefficient can be identified only within the range of the compressed distance offsets.

[0124] As mentioned above, the correlation coefficient is calculated using Formula 4. Referring to the results in Curve 710, it can be confirmed that within the range of distance offsets, the first distance offset_a has the largest correlation coefficient Coef_a.

[0125] The processor 110 can calculate the correlation coefficient and extrapolate the first distance offset offset_a corresponding to the first correlation coefficient Coef_a with the largest correlation coefficient as the final distance offset.

[0126] Figure 8 This is a graph showing the results of correcting the measurement data on each channel.

[0127] Figure 8 It displays the distorted measurement data from channel 1 to channel 8, as well as the data after correction (compensation).

[0128] According to an embodiment of the present invention, such as Figure 4 As described in step S40, processor 110 can utilize the calculated distance offset. To correct the measurement data. The processor 110 can remove (subtract) the calculated distance offset from the measurement data. To calibrate the measurement data.

[0129] Reference Figure 8The data from the typical first channel Ch1 shows the corrected data 812 obtained after correcting the measurement data 811 using the distortion correction method of the present invention.

[0130] The processor 110 can use the distance offset calculated on each channel to correct the measurement data acquired on each channel.

[0131] According to one embodiment of the present invention, the distance offset is corrected on each channel, thereby enabling accurate correction for the entire measurement data.

[0132] Figure 9 This is a diagram illustrating the distortion caused by the cut arc in an embodiment of the present invention.

[0133] exist Figure 9 The image shows distortion caused by a cut-off arc in a lidar received signal according to an embodiment of the present invention. Figure 9 In this invention, for a lidar according to one embodiment of the present invention, it is assumed that the horizontal field of view angle θ k The actual target 1 is located on a plane with a horizontal angle of α between the lidar and the target, between -60 degrees and 60 degrees, and is represented by a solid line with a vertical distance a from the lidar.

[0134] As mentioned above, the measurement data from lidar may be distorted due to the distance resolution during the time-to-digital-converter (TDC) process, resulting in the plane appearing as multiple cut arcs.

[0135] LiDAR can acquire: unlike the actual target 1, by means of... Figure 9 The data of the measurement target 2 (hereinafter referred to as measurement data) consists of multiple cut arcs represented by the thick (red) dashed line.

[0136] More specifically, regarding the received signal from the lidar, data measured from the direction and distance of the actual target 1 is received. However, in representing this data according to the lidar's range resolution, all values ​​within a specified distance are converted to the same distance due to the limitation of the range resolution. That is, there is an error in the displayed measurement data, and the data set with this error is displayed as multiple cut arcs.

[0137] For example, regarding the range resolution of a lidar, if the lidar can display a range resolution of 0.1m, it cannot display values ​​after the decimal point below 0.1. Therefore, even if it detects 0.11m, it will be displayed as 0.1m. Thus, in this case, an error equivalent to 0.01m will occur, and these measurement data will be displayed as multiple cut arcs.

[0138] At this point, the error caused by the distance resolution varies depending on each pixel and also on the horizontal angle between the target and the lidar.

[0139] Therefore, the lidar can acquire: unlike the actual target 1, by means of... Figure 9 The data of the measurement target 2 (hereinafter referred to as measurement data) consists of multiple cut arcs represented by the thick (red) dashed line.

[0140] Below, we will refer to Figures 10 to 14 Specifically, this describes a method and apparatus for correcting measurement data using information related to errors caused by distance resolution. In this case, the structure and functions associated with the lidar 10 and the correction device 100 performing the aforementioned functions will be described using previously referenced methods. Figure 4 The content described.

[0141] Figure 10 This is a flowchart illustrating the operation of an apparatus for correcting distortion caused by a cut arc, according to an embodiment of the present invention.

[0142] According to one embodiment of the present invention, the processor 110 can acquire measurement data obtained by the lidar 10 detecting the target (step S1010).

[0143] At this point, it is assumed that the acquired measurement data is distorted due to distance resolution limitations. If the correction device 100 is configured as a device independent of the lidar 10, it can receive measurement data from the lidar 10 via a communication unit. Alternatively, if the correction device 100 is installed inside the lidar 10, it can receive measurement data using CAN communication, LIN communication, or the like.

[0144] According to an embodiment of the present invention, the processor 110 is able to use measurement data to calculate the horizontal angle α (hereinafter referred to as angle α) between the lidar and the target (step S1020).

[0145] According to one embodiment of the present invention, angle α refers to the angle between the plane in which the lidar is located and the plane in which the target is located. Angle α is a necessary factor for calculating the error caused by the range resolution of the lidar, as described later. The processor 110 can use a linear regression method to calculate angle α.

[0146] If linear regression is used, a regression line can be extracted to represent any data cluster. Therefore, as... Figure 11 As shown, the processor 110 can extract the equation of a straight line (y = P1x + P2) representing the measurement data from the measurement data. As shown in Equation 5, the processor 110 can use the slope of the extracted straight line to calculate the horizontal angle between the lidar and the target. More detailed information will be provided later. Figure 11 Please provide an explanation.

[0147] [Formula 5]

[0148]

[0149] In formula 5 above, The horizontal angle between the lidar and the target; P1: the slope of the regression line of the measured data.

[0150] According to one embodiment of the present invention, the processor 110 is able to identify errors in the measurement data using an error map related to the error caused by the distance resolution of the lidar (step S1030).

[0151] According to an embodiment of the present invention, as described above, the error caused by the distance resolution of the lidar refers to the error that occurs when all values ​​within a specified distance are converted to the same distance due to the limitation of the distance resolution.

[0152] An error map, as described in one embodiment of the present invention, refers to information on the error calculated for each pixel due to distance resolution and the information set therein. In practice, the error map can consist of specific pixels and vector information regarding the error corresponding to those pixels, or it can be set in tabular form.

[0153] The error in one embodiment of the present invention can be calculated using the following formulas 6 to 8.

[0154] [Formula 6]

[0155]

[0156] In Formula 6, rk: the theoretical distance between the target and the lidar; θ k: Horizontal field of view of the lidar; α: Horizontal angle between the lidar and the target; a: Vertical distance between the target and the lidar.

[0157] [Formula 7]

[0158]

[0159] In Formula 7, The distance between the target and the lidar includes distortion; r res : The range resolution of the lidar; r k The theoretical distance between the target and the lidar.

[0160] [Formula 8]

[0161]

[0162] In Equation 8, ε: the error caused by distance resolution.

[0163] Referring to the above formula, Formula 6 shows that, theoretically, the horizontal field of view θ of the lidar can be utilized. k The distance r between the target and the lidar is calculated using the horizontal angle α between the lidar and the target, and the vertical distance a between the target and the lidar. k Distance r k It is actually included in the received signal of the lidar, and the distance r k It can be the distance measured before distortion.

[0164] Then, due to the range resolution r of the lidar res In the event of data distortion, the distance between the target and the lidar, including the distortion, can be calculated using Formula 7.

[0165] Then, the distance r calculated by Equations 6 and 7 can be used. k and distance The difference is used to calculate the error ε caused by the distance resolution. This is shown in Equation 8.

[0166] In one embodiment of the present invention, the processor 110 can generate an error map by mapping the error to each coordinate within the detection area of ​​the lidar 10. The detection area refers to the horizontal field of view θ. k The area within the radius. More specifically, the processor 110 can calculate the error ε for each pixel, thereby generating an error map. At this time, the processor 110 can generate an error map for each horizontal angle α between the lidar 10 and the target.

[0167] According to one embodiment of the present invention, the processor 110 is able to use the identified error ε to correct the measurement data (step S1040).

[0168] According to an embodiment of the present invention, as shown in Formula 9, the processor 110 is able to measure the distance, including distortion. The measured data are corrected by adding the calculated error ε to the calculated error.

[0169] [Formula 9]

[0170]

[0171] In Formula 9, The distance has been corrected; The distance between the target and the lidar includes distortion; ε: error caused by range resolution.

[0172] At this point, the calibration of the measurement data is performed for each pixel, and the same error map is used for calibration at the same angle α. Therefore, the processor 110 is able to calibrate the measurement data by removing (subtracting) the error from the measurement data of all pixels.

[0173] Furthermore, the processor 110 can use the error calculated for each channel to correct the measurement data acquired in each channel. More details will be provided later. Figure 13 illustrate.

[0174] According to one embodiment of the present invention, the accurate shape of the target can be restored by correcting (compensating) for errors in the cut arc, thereby making the distance information more accurate. Furthermore, since the error is precisely calculated for each channel and the calculated error is used for correction, the entire measurement data can be accurately corrected.

[0175] Figure 11 This is a diagram illustrating the calculation of the horizontal angle between the lidar and the target according to an embodiment of the present invention.

[0176] exist Figure 11 The diagram shows measurement data 1110, which includes distortion due to distance resolution. In one embodiment of the invention, the processor 110 can calculate the horizontal angle using a regression line calculated from the measurement data 1110 using a linear regression method. In this case, the processor 110 can calculate the regression line from the measurement data 1110 using a linear regression method.

[0177] Reference Figure 11The regression line calculated from the measured data 1110 can be derived from y = P1x + P2. At this point, the horizontal angle α refers to the angle between the plane where the lidar is located and the plane where the target is located, which can be obtained using the slope P1 of the regression line.

[0178] Processor 110 can be used in conjunction with Figure 10 The horizontal angle is calculated using Formula 5 as described in the explanation of step S1020.

[0179] As described above, the processor 110 can utilize the calculated angle To calculate the error caused by distance resolution.

[0180] Figure 12 This is a diagram illustrating an error graph of an embodiment of the present invention.

[0181] Figure 12 This is an example of an error graph created using Formula 9. Since the error graph is generated for each calculated angle... Therefore, if the angle of the generated error map is... Changes will result in different forms of error maps.

[0182] because Figure 12 The error diagram 1210 shown is based on Figure 9 It was created based on the detection area under the horizontal field of view, therefore it can be regarded as... Figure 9 The detection area corresponds to the area on the error map.

[0183] exist Figure 12 In the error diagram 1210 shown, the error value of each pixel in the detection area is represented by color. In one embodiment of the present invention, the error value 1220 has a value between 0 and 0.1, and each value has its corresponding color. However, the colors corresponding to the range of error values ​​or the individual error values ​​are not limited to this embodiment.

[0184] Therefore, in the error diagram 1210, the color corresponding to the error value 1220 is represented on the coordinate plane accordingly. For example, the color corresponding to the error value 1220 is displayed at position A with coordinates (10, 10) in the coordinate plane.

[0185] The error diagram of the present invention refers to information related to the error caused by distance resolution. The error diagram 1210 of this embodiment is a method for displaying the information, but is not limited thereto.

[0186] According to one embodiment of the present invention, the error of the error map can be represented by color, thereby making it more intuitive to confirm the degree of error and the deviation of the error of each pixel.

[0187] Figure 13 This is a graph showing the results of the measurement data corrected on each channel.

[0188] exist Figure 13 The image shows the measurement data of distortion from the first channel Ch1 to the eighth channel Ch8, as well as the data after compensation.

[0189] According to an embodiment of the present invention, as with Figure 10 As described in step S1040, the processor 110 can use the calculated error ε to correct the measurement data. The processor 110 can correct the measurement data by removing (subtracting) the calculated error ε from the measurement data.

[0190] When observing symbolically Figure 13 When examining the data on the first channel Ch1, the corrected data 1320 is shown, which is formed after the measurement data 1310 is corrected using the distortion correction method of the present invention.

[0191] The processor 110 is capable of correcting (compensating) the measurement data acquired in each channel using the error ε calculated in each channel. According to an embodiment of the invention, it can be confirmed that the linear component of the target measured after error correction based on the error map in individual channels is recovered.

[0192] According to one embodiment of the present invention, since the error is corrected on each channel, accurate correction for the entire measurement data can be achieved.

[0193] Figure 14 This is a graph showing the results of correcting the measured data corresponding to the horizontal angle between the lidar and the target.

[0194] exist Figure 14 The image shows the results of correcting (compensating) for errors in the cut arc of a target at various angles. That is, in... Figure 14 It can be confirmed that even when there is a horizontal angle between the lidar and the target, the plane will be restored regardless of the horizontal angle.

Claims

1. A method for improving the detection accuracy of lidar, wherein, include: The step of acquiring measurement data obtained from the target detected by the lidar; The step of estimating the range of distance offset by the difference between the measured data and the reference data; The step of estimating the distance offset within the estimated distance offset range based on the correlation coefficient between the measured data and the reference data; as well as The step of correcting the measurement data using the calculated distance offset. The difference in distance offset between the measured data and the reference data is calculated based on a distance offset comparison performed on a specific pixel or the average of the distance offset differences of all pixels. The reference data is distorted data that occurs when a specific distance offset exists between the target and its correct position, and is set for each distance offset. The distorted data refers to the distortion that occurs in the plane due to distance resolution during the time-to-digital conversion process, resulting in multiple cut arcs.

2. The distance distortion correction method for improving the detection accuracy of lidar according to claim 1, wherein, The steps for calibrating the measurement data include: The step of correcting the measurement data acquired in each channel using the distance offset calculated in each channel.

3. The distance distortion correction method for improving the detection accuracy of lidar according to claim 1, wherein, The steps for calculating the range of the distance offset include: The step of identifying the minimum range of distance offsets between specific pixels in the measured data and specific pixels in the reference data.

4. The distance distortion correction method for improving the detection accuracy of lidar according to claim 1, wherein, The steps for calculating the distance offset include: The step of estimating the highest correlation coefficient between the measured data and the reference data within the range of the distance offset.

5. The distance distortion correction method for improving the detection accuracy of lidar according to claim 1, wherein, The steps for calibrating the measurement data include: The step of correcting the measurement data by subtracting the calculated distance offset from the measurement data.

6. The distance distortion correction method for improving the detection accuracy of lidar according to claim 1, wherein, Also includes: The step of using the measured data to calculate the horizontal angle between the lidar and the target; The step of identifying errors in the measurement data using an error map related to the error caused by the range resolution of the lidar; and The step of correcting the measurement data using the identified error.

7. The distance distortion correction method for improving the detection accuracy of lidar according to claim 6, wherein, Also includes: The step of generating the error map by mapping the error to each coordinate within the detection area of ​​the lidar.

8. The distance distortion correction method for improving the detection accuracy of lidar according to claim 7, wherein, The steps for generating the error map include: The step of generating the error map for each horizontal angle between the lidar and the target.

9. The distance distortion correction method for improving the detection accuracy of lidar according to claim 6, wherein, The steps for calibrating the measurement data include: The step of correcting the measurement data for each pixel using the error identified for each pixel.

10. The distance distortion correction method for improving the detection accuracy of lidar according to claim 6, wherein, The steps for calculating the horizontal angle include: The step of calculating the horizontal angle using the regression line. The regression line is calculated from the measured data using a linear regression method.

11. A range distortion correction device for improving the detection accuracy of lidar, the range distortion correction device comprising a processor, wherein, The processor performs the following functions: Acquire measurement data obtained from the target detected by the lidar; The range of distance offset is estimated by the difference between the measured data and the reference data; Based on the correlation coefficient between the measured data and the reference data, the distance offset is estimated within the estimated distance offset range; The calculated distance offset is used to correct the measurement data. The difference in distance offset between the measured data and the reference data is calculated based on a distance offset comparison performed on a specific pixel or the average of the distance offset differences of all pixels. The reference data is distorted data that occurs when a specific distance offset exists between the target and its correct position, and is set for each distance offset. The distorted data refers to the distortion that occurs in the plane due to distance resolution during the time-to-digital conversion process, resulting in multiple cut arcs.

12. The distance distortion correction device for improving the detection accuracy of lidar according to claim 11, wherein, The processor uses the distance offset calculated for each channel to correct the measurement data acquired in each channel.

13. The distance distortion correction device for improving the detection accuracy of lidar according to claim 11, wherein, The processor identifies the minimum difference in distance offset between a specific pixel in the measurement data and a specific pixel in the reference data.

14. The distance distortion correction device for improving the detection accuracy of lidar according to claim 11, wherein, The processor calculates the highest correlation coefficient between the measured data and the reference data within the specified distance offset range.

15. The distance distortion correction device for improving the detection accuracy of lidar according to claim 11, wherein, The processor corrects the measurement data by subtracting the calculated distance offset from the measurement data.

16. The distance distortion correction device for improving the detection accuracy of lidar according to claim 11, wherein, The processor performs the following functions: The horizontal angle between the lidar and the target is calculated using the measured data. Errors in the measurement data are identified using an error map relating to the error caused by the range resolution of the lidar. The identified errors are used to correct the measurement data.

17. The distance distortion correction device for improving the detection accuracy of lidar according to claim 16, wherein, The processor generates the error map for each horizontal angle between the lidar and the target by mapping the error to each coordinate within the detection area of ​​the lidar.

18. The distance distortion correction device for improving the detection accuracy of lidar according to claim 16, wherein, The processor uses the error identified for each pixel to correct the measurement data for each pixel.

Images