A method for online training and calibration of MEMS scanning mirrors for laser radar

Abstract

A method for online training and calibration of a MEMS scanning mirror for a laser radar, wherein a target plate is placed in front of the laser radar, a laser radar reference mirror is placed above the laser radar, a high-speed camera is placed on the side of the laser radar, and a red light source emits red light to illuminate the target plate. The theodolite system aligns the laser radar reference mirror to establish a laser radar reference mirror coordinate system. By placing the target plate at two different positions, the relationship between the camera image plane coordinates and the position under the laser radar reference mirror system is established by using the target plate and the theodolite system, and a high-speed camera is used to measure the scanning trajectory of the image plane coordinates at two different positions, thereby obtaining the relationship between the laser radar reference mirror coordinate system and the scanning system. On this basis, the laser radar scanning trajectory at one of the positions is obtained, and the error between it and the target scanning trajectory is calculated, and the error is made to meet the specified threshold through PID control. The present invention has the characteristics of high precision and high efficiency compared with traditional methods.

Description

Technical Field

[0001] The invention belongs to the field of aerospace optical remote sensing and relates to a method for online training and calibration of a MEMS scanning mirror for a laser radar. Background Art

[0002] As a solid-state laser radar, MEMS scanning laser radar has the characteristics of high precision, high reliability, miniaturization and light weight. It is increasingly becoming the mainstream sensor in aerospace optical remote sensing, civil mapping, autonomous driving and other fields.

[0003] The current MEMS scanning mirrors, whether they are electrostatic, piezoelectric, thermoelectric, or electromagnetic, do not have an integrated angle encoder inside the device, and generally use an open-loop control method. For example, electrostatically driven MEMS scanning mirrors, when driven by high-frequency voltage in an open loop, are prone to oscillation and deviate from the expected angle scanning trajectory, causing the three-dimensional point cloud image output by the lidar to be distorted and deformed.

[0004] In order to solve the above oscillation problem, Veljko A device and method for offline training of MEMS scanning mirrors using PSD detectors is proposed (“Iterative learning control (ILC) algorithm for greatly increased bandwidth and linearity of MEMS mirrors in LiDAR and related imaging applications”). This method has three problems:

[0005] First, the PSD detector directly measures the light spot. Compared with the light spot size, the detector image surface is smaller, resulting in a shorter measurement baseline and limited angle measurement accuracy.

[0006] Second, the measurement benchmark of the PSD detector is difficult to derive, so the optical path structure parameters of the PSD detector device are difficult to accurately calibrate, and the accuracy of MEMS scanning mirror training and calibration is limited;

[0007] Third, it is offline training and can only calibrate the MEMS scanning mirror device itself. After it is assembled on the lidar product, the angle of the whole machine must be calibrated. Training and calibration are divided into two production processes, which is inefficient and time-consuming. Summary of the invention

[0008] The technical problem solved by the present invention is: to overcome the deficiencies of the prior art and provide an online training and calibration method for a MEMS scanning mirror for a laser radar, which can significantly improve the calibration accuracy and training efficiency of the MEMS scanning mirror.

[0009] The technical solution of the present invention is: an online training and calibration method for a MEMS scanning mirror for a laser radar, comprising the following steps:

[0010] (101) placing a target plate in front of a laser radar, placing a laser radar reference mirror above the laser radar, placing a high-speed camera on the side of the laser radar, and illuminating the target plate with a red light source; a plurality of return light marks are arranged on the target plate;

[0011] (102) Using the theodolite system to align the laser radar reference mirror and establish the laser radar reference mirror coordinate system;

[0012] (103) According to the factory test parameters of the laser radar, the target scanning angle trajectory (α d (t), β d (t)) is converted into the initial digital quantity (X) given by the laser radar to the laser radar MEMS scanning mirror p0 (t), Y p0 (t));

[0013] (104) Move the target plate to position A, adjust the exposure time of the high-speed camera, image the target plate, and calculate the image plane centroid position (U) of a local return light mark in the field of view using the centroid algorithm. b1 , V b1 );

[0014] (105) Using the theodolite system to measure the coordinate value (X Jb1 , Y Jbl , Z Jb1 ), and thus establish the image plane centroid position (U1, V1) and the coordinate value (X J1 , Y J1 , Z J1 )

[0015] (X J1 , Y J1 , Z J1 ) = f1(U1, V1);

[0016] (106) Turn off the red light source, control the laser radar to scan the target plate, and the control digital quantity is (X p0 (t), Y p0 (t)), at the same time, the high-speed camera continuously images the target area, and obtains the scanning trajectory of the laser beam on the image plane (u1(t), v1(t)) through the centroid algorithm calculation;

[0017] (107) According to the relationship of step (105), the scanning trajectory of the laser beam on the target plate is obtained:

[0018] (X J1 (t), Y J1 (t), Z J1 (t)) = f1(u1(t), v1(t));

[0019] (108) Move the target plate to position B, which is farther away from the laser radar than position A, use a red light source to illuminate the target plate, adjust the exposure time of the high-speed camera to image the target plate, and calculate the image plane centroid position (U) of all the return light marks in the field of view by the centroid algorithm. b2 , V b2 );

[0020] (109) Using the theodolite system to measure the coordinate values ​​(X) of all the return light marks in the laser radar reference mirror coordinate system Jb2 , Y Jb2 , Z Jb2 ), and thus establish the image plane centroid position (U2, V2) and the coordinate value (X J2 , Y J2 , Z J2 )

[0021] (X J2 , Y J2 , Z J2 ) = f2(U2, V2);

[0022] (110) Turn off the red light source, control the laser radar to scan the target plate, and the control digital quantity is (X p0 (t), Y p0 (t)), at the same time, the high-speed camera continuously images the target area, and obtains the scanning trajectory of the laser beam on the image plane (u2(t), v2(t)) through the centroid algorithm calculation;

[0023] (111) According to the relationship in step (109), the scanning trajectory of the laser beam on the target plate is obtained:

[0024] (X J2 (t), Y J2 (t), Z J2 (t)) = f2(u2(t), v2(t))

[0025] (112) Establish the optical scanning angle (α, β) of the MEMS scanning mirror of the laser radar and the light spot position (X J , Y J , Z J), the expression is

[0026]

[0027] in Representation vector The modulus of , R is the rotation matrix, T is the translation matrix;

[0028] (113) According to the relationship in step (112), the angle scanning trajectory (α(t), β(t)) is solved:

[0029]

[0030]

[0031]

[0032] To measure the coordinates in this system;

[0033] (114) Calculate the angle scanning trajectory (α(t), β(t)) and the target scanning angle trajectory (α d (t), β d (t)) to find the maximum absolute value of the two-axis angle error. If the maximum absolute values ​​of the two-axis errors are both less than the set value θ, the training and calibration are completed; otherwise, the initial digital value (X p0 (t), Y p0 (t)) and obtain the new maximum absolute value of the two-axis angular error again, until the maximum absolute value of the two-axis angular error is less than the set value θ, and then exit the iterative process.

[0034] The advantages of the present invention compared with the prior art are:

[0035] (1) The present invention adopts a reflective spot position measurement method based on a high-speed camera. Due to the long measurement baseline, it has higher angle measurement accuracy than the direct spot measurement method based on the PSD device, and can improve the angle measurement accuracy from 0.1° to 0.01°.

[0036] (2) The present invention uses a theodolite system to directly establish the relationship between the image plane coordinates of the high-speed camera and the spatial coordinates under the laser radar reference mirror system, which overcomes the shortcomings of the prior art PSD detector equipment that the measurement reference is difficult to derive and accurately calibrate, and the training accuracy of the MEMS scanning mirror is improved from 0.3° to 0.03°.

[0037] (3) The present invention adopts an online training method for the MEMS scanning mirror based on the whole machine product, and outputs the whole machine angle calibration result at the same time. Compared with the existing method of offline training and whole machine angle calibration in two stages, it simplifies the production process, improves production efficiency, and reduces time cost. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 is a structural diagram of the device of the present invention;

[0039] Figure 2 This is a composition diagram of the laser radar of the present invention;

[0040] Figure 3 is the amplitude-frequency response curve of the MEMS scanning mirror of the present invention;

[0041] Figure 4 is the phase-frequency response curve of the MEMS scanning mirror of the present invention;

[0042] Figure 5 A distribution diagram of the return light marks on the target plate of the present invention;

[0043] Figure 6 It is a flow chart of the online training and calibration method of the present invention. DETAILED DESCRIPTION

[0044] In order to make the objectives, technical solutions and advantages of the present invention more clear, the embodiments of the present invention are further described in detail below with reference to the accompanying drawings.

[0045] like Figure 1 As shown, the device involved in the online training and calibration method of the MEMS scanning mirror for laser radar of the present invention comprises a laser radar 1, a laser radar reference mirror 2, a high-speed camera 3, a target plate 4, a processing computer 5, a theodolite system 6, and a red light source 7. The target plate 4 is placed in front of the laser radar 1, the laser radar reference mirror 2 is placed above the laser radar 1, the high-speed camera 3 is placed on the side of the laser radar 1, and the red light source 7 emits uniform red light to illuminate the target plate 4. The theodolite system 6 aligns the laser radar reference mirror 2 to establish the laser radar reference mirror 2 coordinate system. By placing the target plate 4 at two different positions, the conversion relationship between the image plane coordinates of the high-speed camera 3 and the three-dimensional coordinates of the position under the laser radar reference mirror 2 system is established respectively by using the return light mark on the target plate 4 and the theodolite system 6, and the high-speed camera 3 is used to measure the image plane coordinate scanning trajectory of the laser radar 1 at two different positions, thereby obtaining the rotation and translation matrix of the laser radar reference mirror 2 coordinate system to the scanning system. On this basis, the optical angle scanning trajectory of the laser radar 1 at one of the positions is obtained, and the error between it and the target angle scanning trajectory is calculated. Through PID control, the error is made to meet the specified threshold, thereby completing the online training and calibration of the MEMS scanning mirror of the laser radar 1.

[0046] The laser radar 1 is composed of a laser 101, a laser collimator 102, a MEMS scanning mirror 103, a MEMS scanning mirror driving circuit 104, a D / A conversion circuit 105, a receiving lens 106, an APD detector driving circuit 107, and a processor 108. Figure 2 shown.

[0047] The laser 101 emits a pulsed laser with a central wavelength of 1064nm and a repetition frequency of 75kHz. After passing through the laser collimator 102, a collimated beam is formed, and the collimated beam is incident on the MEMS scanning mirror 103. The MEMS scanning mirror 103 is based on the electrostatic drive principle and performs scanning in two orthogonal directions under the drive of two pairs of differential high-voltage signals. When the MEMS scanning mirror 103 leaves the factory, the manufacturer will provide the amplitude-frequency response curves (A) of the corresponding serial number product in two axes. x (f), A y (f)) and phase-frequency response curve like Figure 3 As shown, the amplitude-frequency response curve A of the X-axis x (f) If Figure 4 As shown, it is the phase-frequency response curve of the X-axis Under the control of the processor 108, two axial digital control quantities are generated, the D / A conversion circuit 105 converts the digital control quantity into an analog control quantity, and the MEMS scanning mirror drive circuit 104 converts the analog control quantity into two pairs of differential high-voltage signals to control the MEMS scanning mirror 103 to swing in two dimensions to form a 30°×30° scanning beam. The receiving lens 106 is a wide-field energy receiving optical system that collects the 30°×30° echo beam reflected by the target onto the photosensitive surface of the APD detector. The APD detector drive circuit 107 converts the analog echo laser pulse into a termination timing pulse digital signal, and the processor 108 converts the pulse flight time into a distance according to the pulse flight time measurement principle, and generates a three-dimensional point cloud of the target in combination with the scanning angle trajectory of the MEMS scanning mirror 103.

[0048] The time it takes for the laser radar 1 to complete a 30°×30° full-field scan is 1 second, and the total number of scanning points is 75,000.

[0049] The laser radar reference mirror 2 is a cubic mirror of 15mm×15mm×15mm.

[0050] The response spectrum of the high-speed camera 3 ranges from 400 nm to 1700 nm, the resolution is 1024×1024, the imaging frame rate is 10000 frames / s, the field of view is 32°×32°, and the instantaneous receiving field of view corresponding to a single pixel is about 0.03°.

[0051] The red light source 7 emits red light with a wavelength of about 635 nm.

[0052] The target plate 4 is a standard flat target plate with a plane size of 1600mm*1600mm and a flatness better than 0.1 / 100mm. The surface material of the target plate is a diffuse reflection material, which is highly reflective to the 1064nm laser wavelength with a reflectivity greater than 90%, and lowly reflective to the 635nm red light with a reflectivity less than 10%. The target plate 4 is installed on a mobile platform.

[0053] like Figure 5 As shown, 100 light-returning marks are distributed on the target plate 4, and the light-returning marks are distributed at equal intervals, with an interval d of 160 mm. Each light-returning mark on the target plate is circular, with a diameter Φ of 5 mm. The light-returning mark is highly reflective to 635nm red light and 1064nm laser wavelength, and the reflectivity is greater than 90%. A crosshair is set at the center of the light-returning mark.

[0054] like Figure 6 As shown, the steps of online training of the laser radar MEMS scanning mirror 103 are as follows:

[0055] (1) Based on the factory test parameters of the MEMS scanning mirror 103, namely the amplitude-frequency response (A) of the two axes x (f), A y (f)) and the phase-frequency response The target scanning angle trajectory (α d (t), β d (t)) is converted into the initial driving amount (X) of the laser radar 1 to the laser radar MEMS scanning mirror 103 d0 (t), Y d0 (t)):

[0056]

[0057]

[0058] in, is the Fourier transform, is the inverse Fourier transform.

[0059] (2) According to the design parameters of the laser radar 1, the initial driving amount (X d0 (t), Y d0 (t)) is inversely converted into the initial digital value (X) given to the laser radar 1 by the processor 108 p0 (t), Y p0 (t)):

[0060] (X p0 (t), Y p0 (t)) = H -1 (X d0(t), Y d0 (t))

[0061] The H function is a linear function, expressed as:

[0062] X d0 (t) = kx*X p0 (t)+bx

[0063] Y d0 (t) = ky*Y p0 (t)+by

[0064] Among them, kx, ky, bx, and by are the design parameters of the lidar, among which kx and ky are proportional coefficients, which physically include the combined effects (products) of digital-to-analog conversion, low-voltage drive circuit amplification factor, high-voltage drive circuit amplification factor, etc. The circuit part is not expanded here, and bx and by are constant terms.

[0065] (3) Place the target plate 4 at position A (a position close to the laser radar 1), that is, 2 meters away from the light outlet of the laser radar 1. Use the red light source 7 to illuminate the target plate 4, adjust the exposure time of the high-speed camera 3, and image the target plate 4 so that the imaging grayscale values ​​of all the return light marks on the target plate 4 are between 100 and 200. The image plane centroid position (U) of all 36 return light marks in the field of view (other values ​​can also be used, such as a local return light mark) is calculated by the known centroid algorithm. b1 , V b1 ). According to the known sub-pixel subdivision technology, the positioning accuracy of the image plane centroid can at least reach 1 / 3 of the instantaneous receiving field of view, that is, 0.01°.

[0066] (4) The theodolite system 6 aligns the laser radar reference mirror 2 to establish the laser radar reference mirror coordinate system.

[0067] (5) Theodolite system 6 measures the coordinate values ​​of 36 return light marks in the laser radar reference mirror coordinate system (X Jb1 , Y Jb1 , Z Jb1 ).

[0068] (6) Using the known camera calibration method, the image plane centroid position (U1, V1) and the coordinate value (X) in the laser radar reference mirror coordinate system are established based on the measurement data of steps (3) and (5). J1 , Y J1 , Z J1 )

[0069] (X J1 , Y J1 , Z J1 )=f1(U1,V1)

[0070] (7) Turn off the red light source 7, and the processing computer 5 controls the laser radar 1 to scan the target plate 4. The control digital quantity is (X p0 (t), Y p0 (t)), at the same time, the high-speed camera 3 continuously images the target plate 4 area, and obtains the scanning trajectory (u1(t), v1(t)) of the laser beam on the image plane through the centroid algorithm calculation, where t takes discrete values, i.e. 0, T, 2T, 3T, .... Wherein T is the laser pulse emission period, and in this embodiment, T is 1 / (75kHz).

[0071] (8) According to step (6), obtain the scanning trajectory of the laser beam on the target plate 4:

[0072] (X J1 (t), Y J1 (t), Z J1 (t)) = f1(u1(t), v1(t))

[0073] (9) The target plate is placed at position B (far away from the laser radar 1), that is, 3 m away from the light outlet of the laser radar 1, and the target plate 4 is illuminated by a red light source 7. The exposure time of the high-speed camera 3 is adjusted to image the target plate 4 so that the grayscale values ​​of all the return light marks on the target plate are between 100 and 200. The image plane centroid position (U) of all 100 return light marks (i.e., all return light marks) in the field of view is calculated by the known centroid algorithm. b2 , V b2 ).

[0074] (10) Theodolite system 6 measures the coordinate values ​​of 100 return light marks in the laser radar reference mirror coordinate system (X Jb2 , Y Jb2 , Z Jb2 ).

[0075] (11) By using a known camera calibration method, the image plane centroid position (U2, V2) and the coordinate value (X) in the reference mirror coordinate system are established based on the measurement data of steps (9) and (10). J2 , Y J2 , Z J2 )

[0076] (X J2 , Y J2 , Z J2 )=f2(U2,V2)

[0077] (12) Turn off the red light illumination source 7, and the processing computer 5 controls the laser radar 1 to scan the target plate 4. At the same time, the high-speed camera 3 continuously images the target plate 4 area, and obtains the scanning trajectory (u2(t), v2(t)) of the laser beam on the image plane through the centroid algorithm calculation.

[0078] (13) According to step (11), the scanning trajectory of the laser beam on the target plate 4 is obtained:

[0079] (X J2 (t), Y J2 (t), Z J2 (t)) = f2(u2(t), v2(t))

[0080] (14) Establish the optical scanning angle (α, β) of the laser radar MEMS scanning mirror 103 and the light spot position (X J , Y J , Z J ), the expression is

[0081]

[0082] in Representation vector Model.

[0083] According to this expression, the calibration model of rotation matrix R and translation matrix T is established:

[0084]

[0085]

[0086] In the above formula, t has 75,000 values ​​(the total number of points in a full-field scan, as explained above), and is solved by the least squares method to obtain the regression values ​​of the rotation matrix and the translation matrix.

[0087] (15) According to the solved rotation matrix R and translation matrix T, the angle scanning trajectory (α(t), β(t)) is solved:

[0088]

[0089]

[0090]

[0091] Coordinate representation for the measurement system (angle scanning trajectory can only be expressed in the measurement system)

[0092] (16) Calculate the angle scanning trajectory (α(t), β(t)) and the target scanning angle trajectory (α d (t), β d The error between (t)) is:

[0093]

[0094] (17) Determine the maximum absolute value of the two-axis angular error. If the maximum value is less than the set value θ, the training calibration is completed; if the maximum value is greater than the set value θ, the next digital control amount is calculated according to the following learning law:

[0095]

[0096] Among them, P, I, and D are 2*2 constant matrices and can be set manually.

[0097] (18) The processing computer 5 inputs the digital control amount to the laser radar 1, and repeats steps (12), (13), (15), and (16) to obtain a new absolute value of the angle error e1(t). This cycle is repeated until the maximum absolute value of the two-axis angle error is less than the set θ, and the iterative process is exited.

[0098] In this embodiment, θ is 0.03°.

[0099] Compared with the offline calibration method of the traditional laser radar MEMS scanning mirror, the online calibration method of the present invention is carried out at the whole machine stage, and the angle calibration accuracy of the two axes is better than 0.03°, which has the significant advantages of high precision and high efficiency.

[0100] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A method for online training and calibration of a MEMS scanning mirror for laser radar, characterized in that The steps include: (101) placing a target plate (4) in front of a laser radar (1), placing a laser radar reference mirror (2) above the laser radar (1), placing a high-speed camera (3) on the side of the laser radar (1), and illuminating the target plate (4) with a red light source (7); a plurality of return light marks are provided on the target plate (4); (102) using the theodolite system (6) to align the laser radar reference mirror (2) and establish a laser radar reference mirror coordinate system; (103) According to the factory test parameters of the laser radar (1), the target scanning angle trajectory (α d (t), β d (t)) is converted into the initial digital quantity (X) from the laser radar (1) to the laser radar MEMS scanning mirror (103) p0 (t), Y p0 (t)); (104) Move the target plate (4) to position A, adjust the exposure time of the high-speed camera (3), image the target plate (4), and calculate the image plane centroid position (U) of a local return light mark in the field of view by using the centroid algorithm. b1 , V b1 ); (105) Using the theodolite system (6), the coordinate value (X Jb1 , Y Jb1 , Z Jb1 ), and thus establish the image plane centroid position (U1, V1) and the coordinate value (X J1 , Y J1 , Z J1 ) (X J1 ,Y J1 ,Z J1 )=f1(U1,V1); (106) Turn off the red light source (7), control the laser radar (1) to scan the target plate (4), and control the digital quantity to be (X p0 (t), Y p0 (t)), at the same time, the high-speed camera (3) continuously images the target plate (4) area, and obtains the scanning trajectory (u1(t), v1(f)) of the laser beam on the image plane through the centroid algorithm calculation; (107) According to the relationship in step (105), the scanning trajectory of the laser beam on the target plate (4) is obtained: (X J1 (t),Y J1 (t),Z J1 (t))=f1(u1(t),v1(t)); (108) moving the target plate (4) to position B, wherein the position B is further away from the laser radar (1) than the position A, using a red light source (7) to illuminate the target plate (4), adjusting the exposure time of the high-speed camera (3) to image the target plate (4), and calculating the image plane centroid position (U) of all the return light marks in the field of view by a centroid algorithm. b2 , V b2 ); (109) Using the theodolite system (6) to measure the coordinate values ​​(X Jb2 , Y Jb2 , Z Jb2 ), and thus establish the image plane centroid position (U2, V2) and the coordinate value (X J2 , Y J2 , Z J2 ) <h2 style=";text-align:left;direction:ltr">(X<h2 style=";text-align:left;direction:ltr"> J2 <h2 style=";text-align:left;direction:ltr"> ,Y<h2 style=";text-align:left;direction:ltr"> J2 <h2 style=";text-align:left;direction:ltr"> Z<h2 style=";text-align:left;direction:ltr"> J2 <h2 style=";text-align:left;direction:ltr"> )=f2(U2,V2); (110) Turn off the red light source (7), control the laser radar (1) to scan the target plate (4), and control the digital quantity to be (X p0 (t), Y p0 (t)), at the same time, the high-speed camera (3) continuously images the target plate (4) area, and obtains the scanning trajectory (u2(t), v2(t)) of the laser beam on the image plane through the centroid algorithm calculation; (111) According to the relationship in step (109), the scanning trajectory of the laser beam on the target plate (4) is obtained: (X J2 (t),Y J2 (t),Z J2 (t))=f2(u2(t),v2(t)); (112) Establish the optical scanning angle (α, β) of the MEMS scanning mirror (103) of the laser radar (1) and the light spot position (X) in the laser radar reference mirror coordinate system J , Y J , Z J ), the expression is in Representation vector The modulus of , R is the rotation matrix, T is the translation matrix; (113) According to the relationship in step (112), the angle scanning trajectory (α(t), β(t)) is solved: To measure the coordinates in this system; (114) Calculate the angle scanning trajectory (α(t), β(t)) and the target scanning angle trajectory (α d (t), β d (t)) to find the maximum absolute value of the two-axis angle error. If the maximum absolute values ​​of the two-axis errors are both less than the set value θ, the training and calibration are completed; otherwise, the initial digital value (X p0 (t), Y p0 (t)) and obtain the new maximum absolute value of the two-axis angular error again, until the maximum absolute value of the two-axis angular error is less than the set value θ, and then exit the iterative process.

2. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 1, characterized in that: The target plate (4) is a standard plane target plate, on which 10*10 light-returning marks are distributed in a square shape. The light-returning marks are distributed at equal intervals, and the interval d is 160 mm.

3. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The light-returning mark is circular, and its diameter Φ is 5 mm.

4. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The position A is 2 m away from the light outlet of the laser radar (1).

5. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The imaging of the target plate (4) should be performed so that the imaging grayscale values ​​of all the return light marks on the target plate (4) are between 100 and 200.

6. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The position B is 3 m away from the light outlet of the laser radar (1).

7. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The local light-returning marks are 36 light-returning marks.

8. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The t takes discrete values, namely 0, T, 2T, 3T, ..., where T is the emission period of the laser pulse.

9. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The rotation matrix R and the translation matrix T are solved by the least square method through different values ​​of t to obtain the regression value.

10. The method for online training and calibration of a MEMS scanning mirror for laser radar according to claim 2, characterized in that: The θ is 0.03°.

Images