Multi-light-source distance and angle calibration and correction method for laser radar

Abstract

The invention discloses a multi-light-source distance and angle calibration correction method for a laser radar, which optimizes the calibration process of the laser radar through an algorithm, and can obtain a calibration result only through one-time calibration wall surface measurement. According to the multi-light-source distance and angle calibration correction method for the laser radar, the calibration precision of the distance and the angle of the laser radar is effectively improved, the calibration process is optimized through the centroid algorithm, the complexity of the calibration process is greatly reduced, and the calibration efficiency is improved.

Description

Technical Field

[0001] The present invention relates to the field of laser radar technology, and in particular to a method for calibrating and correcting multi-light source distances and angles of a laser radar. Background Art

[0002] LiDAR (Light Detection And Ranging) is a device that uses laser beams for three-dimensional distance measurement and imaging. LiDAR emits short pulses of laser light to the surrounding environment, senses the echo signal reflected from the target, and measures the distance and position of the target, thereby accurately obtaining the three-dimensional spatial information of environmental objects. In recent years, the LiDAR market has grown rapidly and is widely used in many fields, such as the Earth Information System (GIS), environmental monitoring, architectural design, smart agriculture, aerospace, and autonomous driving.

[0003] The laser radars currently on the market usually use multiple sets of light sources and mechanical scanning components to achieve target coverage and three-dimensional reconstruction. For example, the multi-light source uses a laser array to cover the pixels in the vertical direction; the mechanical scanning component uses a one-dimensional rotating multi-faceted prism to cover the pixels in the horizontal direction. The outgoing light from the transmitting end is reflected by the prism and projected onto the target surface; the scattered light from the target is reflected by the prism and returned to the detector at the receiving end. Through photoelectric detection and data processing, the distance and angle information of each pixel is calculated and finally converted into the three-dimensional coordinates of the target.

[0004] Due to manufacturing tolerances and process precision, the laser patch position / angle, optical lens assembly position / angle, and multi-faceted prism assembly position / angle of each LiDAR are bound to be different. Therefore, it is necessary to calibrate the distance and angle of each pixel of the LiDAR to obtain accurate coordinate information.

[0005] The traditional calibration process currently used in this field includes: the first step is that the laser radar illuminates the calibration plate or the calibration wall to obtain the projection spot position Pmn of all pixels (m×n); the second step is to move the laser radar or the calibration plate forward and backward to generate the projection spot position P'mn of all pixels for the second time. According to the coordinate relationship between the projection spot positions of the two times, the horizontal angle, vertical angle, and distance deviation of the (m×n) group of pixels are calculated and calibrated. However, this calibration method has obvious disadvantages, mainly: a high-cost and high-precision electromechanical control module is required to ensure that the translation process of the laser radar or the calibration plate does not produce unnecessary angle or distance deviations, the operation is difficult, and the accuracy is poor. At the same time, the size of the calibration plate needs to be large enough to ensure that all the laser radar spots cover the calibration plate after translation. Finally, the whole process takes a long time, the amount of calculation is large, and the calibration efficiency is low. Summary of the invention

[0006] In order to solve the above technical problems, a technical solution adopted by the present invention is: A method for calibrating and correcting the distance and angle of multiple light sources of a laser radar is provided, the steps of which include: (1) Obtain the system parameter information of the laser radar and the coordinate information of the light source Q (Q x , Q y , Q z ), the incident light direction vector s of the light emitted by each pixel light source p (s px , s py , s pz ), calculate the coordinates of the reflection point O of each light ray on the multifaceted prism ij (O ijx , O ijy , O ijz ), where p is the actual number of light sources in the vertical direction of the laser radar, 1≤p≤n, and the system parameter information includes the number of pixels in the laser radar, which is m columns × n rows, 1≤i≤m, 1≤j≤n; (1.1) Based on the system parameter information, the coordinate information of the light source and the incident light direction vector s p (s px , s py ,s pz ), obtain the reflection surface normal vector and plane equation of each reflection surface; The coordinates of the normal vector N of the reflection surface are: (-sinθ r ·sinα i , sinθ r ·cosα i , -cosθ r ); The plane equation of the reflecting surface is: -sinθ r ·sinα i ·x + sinθ r ·cosα i y - cosθ r z – sinθ r d = 0; (1.2) According to the intersection formula, the coordinates of each reflection point O are calculated. ij (O ijx , O ijy , O ijz ): ; Among them, a three-dimensional coordinate system is established with the center of the multi-faceted prism as the origin and the z axis is perpendicular to the paper surface, α i is the rotation angle of the multifaceted prism around the z axis, 1≤i≤m, d is the rotation radius of the multifaceted prism, θr is the angle between each reflective surface M and the z-axis, and r is the number of reflective surfaces; (2) Each light beam passes through a multi-faceted prism and is irradiated onto the calibration wall to form m columns × n rows of light spots. The center coordinates P of each light spot are obtained. ij (P ijx , P ijy , P ijz ), 1≤i≤m, 1≤j≤n; (3) All reflection points in each row are taken as a group of reflection areas to obtain n groups of reflection areas. The centroid coordinates C of each group of reflection areas are calculated by the centroid algorithm. j (C jx , C jy , C jz ), 1≤j≤n; in, ; (4) According to the centroid coordinates C of the m groups of reflection areas j And (m×n) spot center coordinates P mn , to obtain the calibration distance and angle, thereby generating a distance lookup table, an azimuth lookup table, and an elevation lookup table; According to the following formula, an m-column × n-row distance parameter matrix is ​​calculated to generate a calibrated distance lookup table: , 1≤i≤m, 1≤j≤n; According to the following formula, an m-column × n-row azimuth parameter matrix is ​​calculated to generate a calibrated azimuth lookup table: , 1≤i≤m, 1≤j≤n; The m-column × n-row pitch angle parameter matrix is ​​calculated according to the following formula to generate the calibrated pitch angle lookup table: , 1≤i≤m, 1≤j≤n.

[0007] In a preferred embodiment of the present invention, in a laser radar with a one-dimensional rotating structure, a plurality of light sources are arranged in a vertical direction, collimated by an optical system, and share the same light source coordinate Q (Q x , Q y , Q z ).

[0008] In a preferred embodiment of the present invention, the number of light sources is less than or equal to the number of pixels of the laser radar.

[0009] In a preferred embodiment of the present invention, when p<n, p*r=n.

[0010] In a preferred embodiment of the present invention, the step of acquiring the center coordinates of the light spot includes: (2.1) Use at least one camera to calibrate the wall to be measured and obtain a mapping table between camera pixels and wall coordinates; (2.2) Compile an automated program, the laser radar emits lasers at preset horizontal and vertical angles, generates (m×n) laser spots on the wall, and uses a camera to synchronously capture and calibrate the wall image; (2.3) For all laser spots, use the image processing algorithm and the camera pixel-wall coordinate mapping table to extract the center coordinates P of the (m×n) spots. mn .

[0011] The beneficial effects of the present invention are: not only the calibration accuracy of the distance and angle of the laser radar is effectively improved, but also the calibration process is optimized through the centroid algorithm, which greatly reduces the complexity of the calibration process and improves the calibration efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative work, among which: Figure 1 It is a schematic diagram of calibration positions of a preferred embodiment of a method for calibrating and correcting distance and angle of multiple light sources of a laser radar according to the present invention; Figure 2 It is a top view schematic diagram of a multi-faceted prism of a preferred embodiment of a method for calibrating and correcting distance and angle of multiple light sources of a laser radar according to the present invention; Figure 3 It is a schematic diagram of the centroid position of a preferred embodiment of a multi-light source distance and angle calibration and correction method for a laser radar of the present invention; Figure 4 It is a schematic diagram of horizontal angle deviation obtained by testing in a preferred embodiment of a multi-light source distance and angle calibration and correction method for a laser radar of the present invention; Figure 5 It is a schematic diagram of vertical angle deviation obtained by testing in a preferred embodiment of a multi-light source distance and angle calibration and correction method for a laser radar of the present invention; Figure 6 It is a schematic diagram of distance deviation obtained by testing in a preferred embodiment of a multi-light source distance and angle calibration and correction method for a laser radar of the present invention; DETAILED DESCRIPTION

[0013] The technical solutions in the embodiments of the present invention are described clearly and completely below. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0014] See also Figure 1-6 , the embodiment of the present invention includes: A multi-light source distance and angle calibration and correction method for laser radar is proposed. The laser radar calibration process is optimized by an algorithm. The calibration result can be obtained by only measuring the calibration wall once. The specific steps include: Step (1), obtain the system parameter information of the laser radar and the coordinate information Q of the light source (Q x , Q y , Q z ), the incident light direction vector s of the light emitted by each pixel light source p (s px , s py , s pz ), calculate the coordinates of the reflection point O (O) of each light ray on the multifaceted prism x , O y , O z ). Where p is the actual number of light sources in the vertical direction of the laser radar.

[0015] In the embodiment of the present application, in a laser radar with a one-dimensional rotating structure, a plurality of light sources are arranged in a vertical direction, collimated by an optical system, and share the same light source coordinate Q (Q x , Q y , Q z ). If there are k groups of pixels in the vertical direction of the laser radar, then the incident light direction vector s of the light emitted by each pixel light source is p (s px , s py , s pz ), 1≤p≤n.

[0016] Optionally, the actual number of light sources does not have to be consistent with the number of pixels of the laser radar. For example, when the number of target pixels is 128 and each of the four rotating mirrors has an inclination angle, only 32 sets of light sources are needed to achieve the final effect of presenting 128 pixels. At this time, p*4=128, that is, the direction vector of the incident light has s 1、 s 2…… s 32 .

[0017] In the embodiment of the present application, the laser radar includes m (columns) × n (rows) groups of pixels, where m and n are both positive integers, and the coordinates of the reflection point can be expressed as ij (O ijx , O ijy , O ijz ), 1≤i≤m, 1≤j≤n, that is, the pixel coordinates of the first row and first column are represented by O 11 (O 11x , O 11y , O 11z ), the pixel coordinates of the 2nd column and the 1st row are represented as O 21 (O 21x , O 21y , O 21z ), ... and so on, the pixel coordinates of the i-th column and j-th row are expressed as O ij (O ijx , O ijy , O ijz ).

[0018] (1.1) Based on the system parameter information, the coordinate information of the light source and the incident light direction vector s p (s px , s py ,s pz ), obtain the normal vector of the reflection surface M and the plane equation of the reflection surface M.

[0019] according to Figure 1-2 It can be seen that Q is the position of the light source, O is the reflection point of any light on the surface of the multi-faceted prism, and a three-dimensional coordinate system is established with the center of the multi-faceted prism as the origin (the z-axis is perpendicular to the paper surface). The rotation angle of the multi-faceted prism when it rotates counterclockwise around the z-axis is α i (1≤i≤m), the rotation radius is d and the angle between each reflection surface M and the z-axis is θ r , that is, the system parameter information includes the rotation angle α i , rotation radius d, angle θ r .

[0020] Among them, α i The number of i in the equation is consistent with the number of columns in the pixel matrix, that is, the number of rotations is the same as the number of pixel columns. For example, when there are 600 columns of pixels, the rotation angle is switched 600 times, then α i There is α 1、 α 2…… α 600 There are 600 in total. Since the inclination angles of different reflective surfaces may be different, θ r To represent the inclination angle of each reflecting surface, r is the number of reflecting surfaces. For example, when there are 4 reflecting surfaces, 1≤r≤4.

[0021] The coordinates of the normal vector N of the reflection surface M are: (-sinθ r ·sinα i , sinθ r ·cosα i , -cosθ r ); The plane equation of the reflecting surface M is: -sinθ r ·sinα i ·x + sinθ r ·cosα i y - cosθ r z – sinθ r ·d=0.

[0022] (1.2) According to the intersection formula, the coordinates of each reflection point O are calculated. ij (O ijx , O ijy , O ijz ): .

[0023] Step (2), see Figure 3 Each light beam passes through a multi-faceted prism and illuminates the calibration wall to form m (columns) × n (rows) light spots, and the center coordinates P of each light spot are obtained. ij (P ijx , P ijy , P ijz ), 1≤i≤m, 1≤j≤n.

[0024] The center coordinates of the light spot can be obtained by using a variety of common technical methods in the field, as long as the center coordinates of each light spot can be obtained. The following is an example of an acquisition method: (2.1) Use at least one camera to calibrate the wall to be measured and obtain a mapping table between camera pixels and wall coordinates; (2.2) Compile an automated program, the laser radar emits lasers at specific horizontal and vertical angles, generates (m×n) laser spots on the wall, and uses a camera to synchronously capture and calibrate the wall image; (2.3) For all laser spots, use the image processing algorithm and the camera pixel-wall coordinate mapping table to extract the center coordinates P of the (m×n) spots. mn .

[0025] Step (3): take all reflection points in each row as a group of reflection areas to obtain n groups of reflection areas. Calculate the centroid coordinates C of each group of reflection areas by using the centroid algorithm. j (C jx , C jy , C jz ), 1≤j≤n.

[0026] All reflection points in the first row form the first reflection area, and the centroid coordinates of the first reflection area are C1 (C 1x , C 1y ,C 1z ); All reflection points in the jth row form a second reflection area, and the centroid coordinates of the second reflection area are C j (C jx , C jy ,C jz ); ... and so on, all the reflection points in the nth row form the nth reflection area, and the centroid coordinates of the nth reflection area are C n (C nx , C ny , C nz ).

[0027] The calculation formula of the center coordinates is as follows: ; Step (4), according to the centroid coordinates C of n groups of reflection areas j And the corresponding (m×n) spot coordinates P mn , calculate and calibrate the actual distance and angle to generate distance lookup table, azimuth lookup table and elevation lookup table to improve the accuracy and stability of lidar.

[0028] The centroid of each group of reflection areas is calculated correspondingly to the center of the light spot corresponding to the group, for example, C1 is calculated correspondingly to all the light spot centers of the first row, C2 is calculated correspondingly to all the light spot centers of the second row, and so on.

[0029] The calibrated distance lookup table is calculated according to the following formula to obtain the corresponding (m×n) matrix value: , 1≤i≤m, 1≤j≤n.

[0030] The calibrated azimuth lookup table is calculated according to the following formula to obtain the corresponding (m×n) matrix value: , 1≤i≤m, 1≤j≤n.

[0031] The calibrated pitch angle lookup table is calculated according to the following formula to obtain the corresponding (m×n) matrix value: , 1≤i≤m, 1≤j≤n.

[0032] The traditional calibration method requires calibrating the (m×n) reflection points with the (m×n) light spot centers respectively. That is, if the total number of pixels is 600×128, then it is necessary to calculate the coordinate lookup table of 600×128 reflection points, and then calibrate according to the 600×128 light spot coordinates and 600×128 reflection point coordinates. Such a calibration process is too cumbersome and complicated, which is not conducive to the lidar manufacturing process.

[0033] The present application can simplify the (m×n) reflection points into n groups of reflection areas, and then use the centroid coordinates of the n groups of reflection areas to perform calibration calculations with the (m×n) light spot centers, which can greatly simplify the calibration process and improve the calibration efficiency.

[0034] according to Figure 4-6 It can be seen that through simulation tests, the calibration wall is set to 2m*2m, the horizontal field of view angle is 120°, the vertical field of view angle is 25.4°, the horizontal angle resolution is 0.2°, the vertical angle resolution is 0.2°, and the total number of pixels is 600×128. The above-mentioned traditional calibration method and the calibration method of this application are used to compare the distance and angle calibration. The comparison deviation structure is as follows Figure 4-6 As shown, the horizontal axis is in the range of -60°~+60°, and the vertical axes are distance deviation, horizontal angle deviation and vertical angle deviation respectively. It can be seen that the present application can greatly reduce the calibration workload and improve the calibration efficiency while ensuring the radar accuracy.

[0035] The beneficial effect of the multi-light source distance and angle calibration and correction method of the laser radar of the present invention is: not only effectively improving the calibration accuracy of the distance and angle of the laser radar, but also optimizing the calibration process through the centroid algorithm, greatly reducing the complexity of the calibration process and improving the calibration efficiency.

[0036] The above descriptions are merely embodiments of the present invention and are not intended to limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made using the contents of the present invention specification, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A method for calibrating and correcting the distance and angle of multiple light sources of a laser radar, characterized in that the steps include: (1) Obtain the system parameter information of the laser radar and the coordinate information of the light source Q (Q x , Q y , Q z ), the incident light direction vector s of the light emitted by each pixel light source p (s px , s py , s pz ), calculate the coordinates of the reflection point O of each light ray on the multifaceted prism ij (O ijx , O ijy , O ijz ), where p is the actual number of light sources in the vertical direction of the laser radar, 1≤p≤n, and the system parameter information includes the number of pixels in the laser radar, the number of pixels is m columns × n rows, 1≤i≤m, 1≤j≤n; (1.1) Based on the system parameter information, the coordinate information of the light source and the incident light direction vector s p (s px , s py , s pz ), obtain the reflection surface normal vector and plane equation of each reflection surface; The coordinates of the normal vector N of the reflection surface are: (-sinθ r ·sinα i , sinθ r ·cosα i , -cosθ r ); The plane equation of the reflecting surface is: -sinθ r ·sinα i ·x + sinθ r ·cosα i y - cosθ r z – sinθ r d = 0; (1.2) According to the intersection formula, the coordinates of each reflection point O are calculated. ij (O ijx , O ijy , O ijz ): ； Among them, a three-dimensional coordinate system is established with the center of the multi-faceted prism as the origin and the z axis is perpendicular to the paper surface, α i is the rotation angle of the multifaceted prism around the z axis, 1≤i≤m, d is the rotation radius of the multifaceted prism, θ r is the angle between each reflective surface M and the z-axis, and r is the number of reflective surfaces; (2) Each light ray passes through a multi-faceted prism and is irradiated onto the calibration wall to form m columns × n rows of light spots. The center coordinates P of each light spot are obtained. ij (P ijx , P ijy , P ijz ), 1≤i≤m, 1≤j≤n; (3) Take all the reflection points in each row as a group of reflection areas to obtain n groups of reflection areas. Use the centroid algorithm to calculate the centroid coordinates C of each group of reflection areas. j (C jx , C jy , C jz )1≤j≤n; in, ； (4) According to the centroid coordinates C of the m groups of reflection areas j And (m×n) spot center coordinates P mn , to obtain the calibration distance and angle, thereby generating a distance lookup table, an azimuth lookup table, and an elevation lookup table; According to the following formula, an m-column × n-row distance parameter matrix is ​​calculated to generate a calibrated distance lookup table: , 1≤i≤m,1≤j≤n; According to the following formula, an m-column × n-row azimuth parameter matrix is ​​calculated to generate a calibrated azimuth lookup table: , 1≤i≤m,1≤j≤n; The m-column × n-row pitch angle parameter matrix is ​​calculated according to the following formula to generate the calibrated pitch angle lookup table: , 1≤i≤m,1≤j≤n。 2. The method for calibrating and correcting the multi-light source distance and angle of a laser radar according to claim 1, characterized in that: In a one-dimensional rotating laser radar, several groups of light sources are arranged vertically, collimated by an optical system, and share the same light source coordinate Q (Q x , Q y , Q z ).

3. The method for calibrating and correcting multi-light source distance and angle of a laser radar according to claim 1, characterized in that: The number of light sources is less than or equal to the number of pixels of the lidar.

4. The method for calibrating and correcting multi-light source distance and angle of a laser radar according to claim 1, characterized in that: When p<n, p*r=n.

5. The method for calibrating and correcting multi-light source distance and angle of a laser radar according to claim 1, characterized in that: The steps for obtaining the center coordinates of the light spot include: (2.1) Use at least one camera to calibrate the wall to be measured and obtain a mapping table between camera pixels and wall coordinates; (2.2) Compile an automated program, the laser radar emits lasers at preset horizontal and vertical angles, generates (m×n) laser spots on the wall, and uses a camera to synchronously capture and calibrate the wall image; (2.3) For all laser spots, use the image processing algorithm and the camera pixel-wall coordinate mapping table to extract the center coordinates P of the (m×n) spots. mn .

Images