Calibration system and method for point laser module pointing to center

Abstract

The invention discloses a calibration system and method for a point laser module pointing to a center, and relates to the technical field of optics. The system comprises a basic positioning unit, a sensing feedback unit, a motion control unit and an auxiliary positioning device. The basic positioning unit comprises a six-degree-of-freedom module carrying platform and a high-precision displacement platform; the sensing feedback unit comprises a multi-axis attitude sensor array and a contact type probe group; the motion control unit comprises a multi-axis motion controller and a servo driving module; the auxiliary positioning device comprises a reference calibration assembly and an environment compensation module. According to the method, the contact probe extends out to be in contact with the standard ball target, an initial coordinate system reference is established, the environment is monitored in real time, the three-dimensional attitude and environment parameters are synchronously input into the LSTM prediction model, the real-time compensation amount is generated, the measurement light path is constructed, sub-pixel-level center locking is achieved through moire fringe verification, and the calibration precision is improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of optics, and particularly relates to a calibration system and method for the pointing center of a point laser module. Background Art

[0002] In the fields of precision optical measurement, industrial robot positioning, aerospace, etc., the calibration accuracy of the pointing center of the point laser module directly determines the overall performance of the system. In the prior art, the calibration process mainly relies on manual intervention and static calibration methods, and there are the following technical bottlenecks:

[0003] (1) Limited calibration accuracy

[0004] Traditional methods use monocular vision or mechanical probe contact calibration. Limited by the image resolution (usually >1μm) or mechanical backlash error, it is difficult to achieve sub-micron calibration accuracy. Especially in dynamic working conditions, factors such as environmental temperature drift and mechanical vibration will further reduce the calibration repeatability.

[0005] (2) Lack of real-time performance and adaptability

[0006] Existing systems mostly use fixed calibration parameters and cannot be dynamically adjusted according to environmental changes (temperature, vibration, load disturbance). When the module undergoes micro-deformation due to thermal expansion or mechanical stress, the calibration process needs to be frequently restarted, seriously affecting the operation continuity.

[0007] (3) Reliability defects

[0008] Conventional sensor layouts (such as orthogonal three-axis accelerometers) have a risk of single-point failure, and are vulnerable to interference in a complex electromagnetic environment, resulting in cumulative attitude measurement errors.

[0009] (4) Lack of data management

[0010] Historical calibration parameters are mostly stored in a linear database, lacking the association with working condition characteristics and abnormal recovery mechanism. When the calibration data is abnormal, the system completely relies on manual intervention for recalibration, resulting in an extended downtime. Summary of the Invention

[0011] The purpose of the present invention is to provide a calibration system and method for the pointing center of a point laser module. By extending a contact probe to contact a standard ball target, an initial coordinate system reference is established, the environment is monitored in real time, a measurement optical path is constructed, and Moiré fringes are used to verify and achieve sub-pixel center locking, solving the problems of insufficient existing calibration accuracy, lack of real-time performance and adaptability, and lack of reliability.

[0012] To solve the above technical problems, the present invention is realized through the following technical solutions:

[0013] The present invention relates to a calibration system for a point laser module to point to the center, which includes a basic positioning unit, a sensing feedback unit, a motion control unit and an auxiliary positioning device;

[0014] The basic positioning unit includes a six-degree-of-freedom module stage and a high-precision displacement platform; the six-degree-of-freedom module stage is a combined mechanism of a slide rail and a rotary table, used to achieve translational motion in the X, Y, and Z axes and rotational motion around the axes; the high-precision displacement platform uses a nanoscale electro-ceramic driver and is internally equipped with a grating scale closed-loop feedback system;

[0015] The sensing feedback unit includes a multi-axis attitude sensor array and a contact probe group; the multi-axis attitude sensor array is used to monitor the attitude angles of the six-degree-of-freedom module stage in real time, construct a spatial position measurement network, and establish a three-dimensional coordinate system mapping; the contact probe group is used to realize physical reference point contact detection through a tungsten carbide probe array;

[0016] The motion control unit includes a multi-axis motion controller and a servo drive module; the multi-axis motion controller is a real-time control module using an FPGA and an ARM architecture, and integrates a PID parameter self-tuning algorithm to dynamically adjust the servo response characteristics;

[0017] The auxiliary positioning device includes a reference calibration component and an environmental compensation module; the reference calibration component includes a grating plate with Moiré fringes and an infrared opposed sensor; the grating plate is used to provide a sub-micron-level positioning reference; the infrared opposed sensor constitutes a zero-point calibration system; the motion compensation module includes an active vibration isolation platform and a constant-temperature liquid cooling system; the active vibration isolation platform is used to eliminate ground vibration interference; the constant-temperature liquid cooling system is used to maintain the thermal stability of the mechanical structure.

[0018] As a preferred technical solution, the bottom of the six-degree-of-freedom module stage is rigidly connected to the high-precision displacement platform through an air-floating vibration isolation interface; the multi-axis attitude sensor array is integrated on the surface of the six-degree-of-freedom module stage; the axis attitude sensor array includes a MEMS gyroscope and an accelerometer; the measurement extension arm of the six-degree-of-freedom module stage is equipped with a contact probe group; the tip of the contact probe group is in physical contact with the reference calibration component.

[0019] As a preferred technical solution, the high-precision displacement platform is a three-layer structure; the upper layer of the high-precision displacement platform is an air-floating rotary platform, used for pitch angle and yaw angle adjustment; the middle layer of the high-precision displacement platform is a cross-roller guide rail, used to achieve fine adjustment in the X-axis and Y-axis directions; the bottom layer of the high-precision displacement platform is a piezoelectric ceramic nano-displacement stage, used for fine adjustment in the Z-axis direction; the high-precision displacement platform is electrically connected to the multi-axis motion controller through an optical fiber encoder.

[0020] As a preferred technical solution, the multi-axis motion controller is electrically connected to the motion control unit through a CAN bus; the servo drive module is installed on the high-precision displacement platform; the reference calibration component further includes a laser interferometer mirror and a calibration ball target; the optical path of the laser interferometer mirror is parallel to the axis of the displacement platform; the standard ball target is coaxially aligned with the contact probe.

[0021] The present invention relates to a calibration method for the center pointing of a point laser module, comprising the following steps:

[0022] Step S1: The contact probe extends out to contact the standard ball target to establish the initial coordinate system reference.

[0023] Step S2: After the six-degree-of-freedom module stage is roughly positioned through the crossed roller guide, the piezoelectric ceramic nano-displacement stage completes the fine adjustment in the Z-axis direction, and the air-bearing rotary stage provides pitch and yaw angle compensation.

[0024] Step S3: The multi-axis attitude sensor array collects the three-dimensional attitude of the module.

[0025] Step S4: The environment compensation module monitors the environmental parameters in real time and compensates for the change in the air refractive index through gradient field modeling.

[0026] Step S5: The three-dimensional attitude and environmental parameters are synchronously input into the LSTM prediction model to generate a real-time compensation amount.

[0027] Step S6: A measurement optical path is constructed, and sub-pixel-level center locking is achieved through Moiré fringe verification.

[0028] Step S7: The calibration data is stored in the self-healing database, and the historical optimal parameters are automatically called under abnormal working conditions.

[0029] Step S8: The historical optimal parameters are input into the motion control unit for calibration.

[0030] In step S1, when the contact probe extends out to contact the standard ball target to establish the initial coordinate system reference, the specific process is as follows:

[0031] Step S11: The contact probe extends out, and the tip of the probe contacts the surface of the standard ball target with a constant pressure. The constant pressure generally ranges from 0.5 N to 2 N, and the trigger stroke error is controlled within ±5 μm.

[0032] Step S12: The micro-force sensor built in the contact probe monitors the contact force in real time. When the preset threshold is reached, a signal is triggered to lock the current position coordinates of the probe.

[0033] Step S13: Synchronously collect the three-dimensional coordinates (X1, Y1, Z1) of the probe tip in the robotic arm coordinate system, the known geometric center coordinates (X0, Y0, Z0) of the standard ball target in the global coordinate system, and the environmental temperature parameter collected by the environmental compensation module, where the environmental temperature parameter is used for thermal expansion compensation;

[0034] Step S14: Solve the coordinate transformation matrix by the least squares method. The specific formula is as follows:

[0035] (X1, Y1, Z1) T = R × (X0, Y0, Z0) T + T;

[0036] In the formula, R is the rotation matrix, which is calculated by fitting the contact point between the probe and the target ball; T is the translation vector, which is directly obtained from the displacement of the probe robotic arm;

[0037] Step S15: The probe makes multiple-point contacts (at least four non-coplanar points) along the surface of the standard ball target, and the positioning error of single-point contact is eliminated by spherical equation fitting. The specific formula is as follows:

[0038] (x - a) 2 +(y - b) 2 +(z - c) 2 = r 2 ;

[0039] In the formula, (a, b, c) are the coordinates of the ball center, and r is the radius of the standard ball;

[0040] Step S16: Modify the coordinate system reference according to the environmental temperature parameter collected by the environmental compensation module. The formula for modifying the coordinate system reference is as follows:

[0041] ΔL = α × L × ΔT + β × ΔT 2 ;

[0042] In the formula, α is the material thermal expansion coefficient, and β is the non-linear compensation coefficient;

[0043] Step S17: When the error of the coordinate transformation matrix for three consecutive calibrations is less than the threshold, where the threshold is 5 μm, it is determined that the initial coordinate system is established. The stored reference parameters include the transformation matrix R / T, the environmental compensation coefficient, the timestamp, and the temperature record.

[0044] As a preferred technical solution, in step S3, the multi-axis attitude sensor arrays (including three-axis accelerometers + gyroscopes) are respectively installed at the four vertex positions on the surface of the six-degree-of-freedom module stage to form a spatial redundant measurement network. By uniformly converting the calibration matrices of the four multi-axis attitude sensor arrays to the stage coordinate system, the attitude calculation reference is ensured to be consistent. The acceleration and angular velocity data of the four sensors are fused using the Kalman filter to eliminate the single-point measurement noise, and the three-dimensional attitude angles of the stage are calculated by the least squares method. The three-dimensional attitude angles include pitch, yaw, and roll; satisfying the formula:

[0045]

[0046] In the formula, S i is the measured value of the sensor, and T i (θ) is the theoretical model value;

[0047] The data deviation of each sensor is monitored in real time, and abnormal nodes are removed, such as failed sensors caused by temperature drift; then, the measurement error caused by mechanical deformation is compensated by associating the sensor spacing with the kinematic model.

[0048] The geometric body formed by the 4 sensor nodes is parameterized, and an ideal regular tetrahedron mathematical model (the side length L and vertex coordinates can be analytically expressed) is established as the reference form of the mechanical structure. When mechanical deformation occurs, by measuring the node spacings {L ij} in real time, the deviation value between the actual geometric body and the ideal regular tetrahedron is calculated to quantify the degree of deformation.

[0049] Derive the influence transfer equation of mechanical deformation on the sensor measurement value:

[0050]

[0051] In the formula, K is the stiffness matrix (obtained through finite element analysis or experimental calibration), is the process noise, realizing the joint suppression of deformation error and sensor noise;

[0052] Through the annularly arranged temperature sensors, a three-dimensional temperature field ΔL thermal = α × f V (T - T n ) is established; α is the material thermal expansion data, and T0 is the reference temperature.

[0053] As a preferred technical solution, in step S5, the training process of the LSTM prediction model is as follows:

[0054] Step S51: Collect the sensor data and environmental data and align them according to the time stamp to construct a multi-dimensional time series input matrix;

[0055] Step S52: Divide the input sequence using a time series generator, and the output is the compensation amount Δθ for the next moment;

[0056] Step S53: Perform normalization processing on the multi-source data;

[0057] Step S54: Build an LSTM model architecture; among them, the loss function uses weighted MSE, and the pose parameter weight is set to 3 times that of the environmental parameter; regularization is implemented by setting weight_decay = 0.01 in the Adam optimizer to achieve L2 regularization;

[0058] Step S55: Convert the trained model into ONNX format and deploy it to the embedded AI acceleration module to achieve a real-time inference latency of <5ms.

[0059] When deploying the LSTM prediction model, convert the trained model into ONNX format and deploy it to the embedded AI acceleration module (such as NVIDIA Jetson) to achieve a real-time inference latency of <5ms. Send the predicted compensation amount to the six-degree-of-freedom platform controller through the EtherCAT bus.

[0060] By measuring the phase difference of the laser beam between the target mirror and the reference mirror and combining the periodic characteristics of the Moiré fringes, sub-pixel displacement detection is achieved. When two coherent light beams are superimposed, a small displacement will cause a phase shift of the interference fringes. By analyzing the deformation and position change of the Moiré fringes, the sub-micron center offset can be calculated.

[0061] As a preferred technical solution, in step S6, the specific process of constructing the measurement optical path and verifying through Moiré fringes is as follows:

[0062] Step S61: Configure the interferometer parameters, including configuring the parameters of the reference mirror and the target mirror;

[0063] Step S62: Insert a high-density diffraction grating (3000 lines / mm) into the laser path to generate a periodic Moiré fringe substrate;

[0064] Step S63: Real-time monitor the optical path offset through a quadrant detector, and drive the piezoelectric ceramic fine-tuning mirror to complete beam collimation (accuracy ≤0.1μrad);

[0065] Step S64: The piezoelectric ceramic modulator finely tunes the phase of the reference light at a frequency of 10kHz to generate dynamic Moiré fringes;

[0066] Step S65: The global shutter camera captures fringe images at a rate of 2000fps and transmits them to the image processing unit through an optical fiber;

[0067] Step S66: The image processing unit performs a two-dimensional Fourier transform on the image and extracts the phase information of the fundamental frequency component. The specific formula is as follows:

[0068]

[0069] In the formula, F(u, v) represents the frequency-domain signal, and φ represents the phase distribution FF1B

[0070] Step S67: The cubic spline interpolation method is used to perform sub-pixel fitting on the fringe center, and the positioning accuracy reaches 1 / 50 pixel (equivalent to 0.02 μm).

[0071] Step S68: The center offset Δd is input into the PID controller to generate a high-precision displacement platform drive signal, which drives the servo drive module to complete position correction. The calculation formula for generating the high-precision displacement platform drive signal is as follows:

[0072]

[0073] Step S69: When the standard deviation of the offset for 5 consecutive iterations < 0.1 μm, it is determined that the calibration is completed.

[0074] As a preferred technical solution, in step S7, the self-healing database architecture includes a real-time layer, an optimization layer, and a disaster recovery layer; the real-time layer uses a time-series database to store the calibration raw data updated at the second level; the optimization layer is based on a time-series clustering algorithm, classifies and stores the historical optimal parameters according to the working conditions in a relational database, and establishes an index of the working condition feature vectors; the disaster recovery layer includes realizing distributed redundant storage of the parameters through blockchain technology;

[0075] The self-healing database uses an LSTM network deployed to monitor the calibration data stream in real time. When it detects that the sensor noise > 3σ or the environmental parameters change suddenly, it triggers the self-healing process, matches the current working condition with the historical optimal parameter library based on the weighted Euclidean distance, and preferentially calls the calibration parameter group with a matching degree > 90%.

[0076] The present invention has the following beneficial effects:

[0077] (1) The present invention establishes an initial coordinate system reference by extending the contact probe to contact the standard ball target, monitors the environment in real time, synchronously inputs the three-dimensional attitude and environmental parameters into the LSTM prediction model to generate a real-time compensation amount, constructs a measurement optical path, and uses Moiré fringes to verify and achieve sub-pixel level center locking, improving the calibration accuracy.

[0078] (2) The present invention establishes an initial coordinate system by contacting a standard ball target with a contact probe, eliminates single-point positioning errors by using the four-point contact method, improves the accuracy to 0.3 μm, and adjusts the contact pressure by using the PID algorithm to avoid damage to the surface of the target ball. At the same time, it maintains the coordinate system of the robotic arm and the global coordinate system, supports real-time coordinate conversion, and improves the positioning accuracy.

[0079] (3) The present invention locks the deformation propagation path through the rigid geometric relationship of the regular tetrahedron structure, reduces the dimension of the compensation model from 12 degrees of freedom in the traditional cubic layout to 6 degrees of freedom. The real-time feedback of the deformation amount ΔL controls the compensation delay within 1 ms, which is better than the 10 ms level of the traditional temperature compensation scheme, and improves the environmental adaptability of the system.

[0080] (4) The present invention stores the calibration data in a self-healing database, automatically calls the historical optimal parameters under abnormal working conditions, and inputs the historical optimal parameters into the motion control unit for calibration, which is convenient for the user. In the next abnormal situation, the motion control unit can directly call the optimal parameters to improve the calibration efficiency.

[0081] Of course, it is not necessary for any product implementing the present invention to achieve all the above-mentioned advantages simultaneously. BRIEF DESCRIPTION OF THE DRAWINGS

[0082] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0083] Figure 1 It is a flowchart of a calibration method for a point laser module pointing to the center of the present invention; Figure 2 It is a schematic structural diagram of a calibration system for a point laser module pointing to the center of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0084] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0085] In addition, the technical features involved in the various embodiments of the present invention described below can be combined with each other as long as they do not conflict with each other.

[0086] To make the objectives, technical solutions, and advantages of this application clearer, the following will further describe the embodiments of this application in detail in conjunction with the appended Figure 1 drawings.

[0087] Before introducing the embodiments of this application, the functions of the calibration technology are as follows:

[0088] (1) Calibrate the pointing center accuracy

[0089] Ensure that the actual emission direction of the laser beam coincides strictly with the theoretical optical axis, and eliminate the pointing deviation caused by assembly errors; through the precise matching of the geometric profile center and the reference calibration component (such as the center of a round hole), sub-pixel level positioning is achieved.

[0090] (2) Ensure the stability of system performance

[0091] Compensate for the influence of environmental disturbances such as temperature, humidity, and vibration on the laser beam quality, and maintain long-term working stability; verify the fluctuation range of the laser wavelength and energy output to meet the industrial application standards.

[0092] (3) Support the collaborative work of multiple devices

[0093] In the laser display system, ensure the spatial synchronization of the images of multiple projection modules and avoid image misalignment; provide a unified spatial coordinate system reference for robot navigation and automated production lines.

[0094] To make the objectives, technical solutions, and advantages of this application clearer and more understandable, the following will further elaborate on this application in conjunction with the Figure 1 drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain this application and are not used to limit this application.

[0095] Embodiment 1

[0096] Please refer to Figure 1 shown in the figure. The present invention is a calibration method for the pointing center of a point laser module, including the following steps:

[0097] Step S1: The contact probe extends and contacts the standard spherical target to establish the initial coordinate system reference;

[0098] Step S2: After the six-degree-of-freedom module stage is roughly positioned through the crossed roller guide, the piezoelectric ceramic nano-displacement stage completes the fine adjustment in the Z-axis direction, and the air-bearing rotary stage provides pitch and yaw angle compensation;

[0099] Step S3: The multi-axis attitude sensor array collects the three-dimensional attitude of the module;

[0100] Step S4: The environmental compensation module monitors the environmental parameters in real time and compensates for the change in the air refractive index through gradient field modeling;

[0101] Step S5: The three-dimensional posture and environmental parameters are synchronously input into the LSTM prediction model to generate real-time compensation;

[0102] Step S6: construct a measurement optical path and achieve sub-pixel center locking through moiré fringe verification;

[0103] Step S7: The calibration data is stored in the self-healing database, and the historical optimal parameters are automatically called under abnormal conditions;

[0104] Step S8: Input the historical optimal parameters into the motion control unit for calibration.

[0105] In step S1, the contact probe is extended to contact the standard ball target, and the specific process of establishing the initial coordinate system reference is as follows:

[0106] Step S11: the contact probe is extended, and the tip of the probe contacts the surface of the standard ball target with a constant pressure, the constant pressure is usually in the range of 0.5N-2N, and the trigger stroke error is controlled within ±5μm;

[0107] Step S12: The contact probe has a built-in micro-force sensor that monitors the contact force in real time. When the preset threshold is reached, a signal is triggered to lock the current position coordinates of the probe;

[0108] Step S13: synchronously collect the three-dimensional coordinates (X1, Y1, Z1) of the probe tip in the robot arm coordinate system, the known geometric center coordinates (X0, Y0, Z0) of the standard ball target in the global coordinate system, and the environmental compensation module collects the ambient temperature parameters, which are used for thermal expansion compensation;

[0109] Step S14: Solve the coordinate change matrix by the least square method. The specific formula is as follows:

[0110] (X1,Y1,Z1) T =R×(X0,Y0,Z0) T +T;

[0111] Where R is the rotation matrix, which is calculated by fitting the contact point between the probe and the target ball; T is the translation vector, which is directly obtained from the displacement of the probe mechanical arm;

[0112] Step S15: The probe makes multi-point contact (at least four non-coplanar points) along the surface of the standard spherical target, and eliminates the positioning error of single-point contact by fitting the spherical equation; the specific formula is as follows:

[0113] (x) 2 +(yb) 2 +(zc) 2 =r 2 ;

[0114] In the formula, (a, b, c) are the coordinates of the center of the sphere, and r is the radius of the standard sphere;

[0115] Step S16: Modify the coordinate system reference according to the environmental temperature parameters collected by the environmental compensation module; the formula for modifying the coordinate system reference is as follows:

[0116] ΔL = α × L × ΔT + β × ΔT 2 ;

[0117] In the formula, α is the material thermal expansion coefficient, and β is the non-linear compensation coefficient;

[0118] Step S17: When the error of the coordinate transformation matrix for three consecutive calibrations is less than the threshold value, the threshold value is 5μm, it is determined that the initial coordinate system is established; the stored reference parameters include the transformation matrix R / T, the environmental compensation coefficient, the time stamp, and the temperature record.

[0119] In step S3, the multi-axis attitude sensor array (including a three-axis accelerometer + gyroscope) is respectively installed at the four vertex positions on the surface of the six-degree-of-freedom module stage to form a spatial redundant measurement network. By uniformly converting the calibration matrices of the four multi-axis attitude sensor arrays to the stage coordinate system, the attitude calculation reference is ensured to be consistent. The acceleration and angular velocity data of the four sensors are fused using the Kalman filter to eliminate the single-point measurement noise, and the three-dimensional attitude angles of the stage are calculated by the least squares method. The three-dimensional attitude angles include pitch, yaw, and roll; satisfying the formula:

[0120]

[0121] In the formula, S i is the measured value of the sensor, and T i (θ) is the theoretical model value;

[0122] Monitor the data deviation of each sensor in real time, and eliminate abnormal nodes, such as failed sensors caused by temperature drift; then, compensate for the measurement error caused by mechanical deformation by associating the sensor spacing with the kinematic model.

[0123] Parameterize the geometric body formed by the 4 sensor nodes, establish an ideal regular tetrahedron mathematical model (the side length L and vertex coordinates can be analytically expressed), and use it as the reference form of the mechanical structure. When mechanical deformation occurs, by measuring the node spacing {L ij} in real time, calculate the deviation value between the actual geometric body and the ideal regular tetrahedron to quantify the degree of deformation.

[0124] Derive the influence transfer equation of mechanical deformation on the sensor measurement value:

[0125]

[0126] In the formula, K is the stiffness matrix (obtained through finite element analysis or experimental calibration), is the process noise, which realizes the joint suppression of deformation error and sensor noise;

[0127] Establish a three-dimensional temperature field ΔL through annularly arranged temperature sensors thermal = α × f V (T - T n ); α is the material thermal expansion data, and T0 is the reference temperature.

[0128] In step S5, the training process of the LSTM prediction model is as follows:

[0129] Step S51: Align the sensor data and environmental data according to the time stamp, and construct a multi-dimensional time series input matrix;

[0130] Step S52: Use the time series generator to divide the input sequence, and the output is the compensation amount Δθ at the next moment;

[0131] Step S53: Normalize the multi-source data;

[0132] Step S54: Construct the LSTM model architecture; among them, the loss function uses weighted MSE, and the pose parameter weight is set to 3 times that of the environmental parameter; regularization is achieved by setting weight_decay = 0.01 in the Adam optimizer to implement L2 regularization;

[0133] Step S55: Convert the trained model into the ONNX format and deploy it to the embedded AI acceleration module to achieve a real-time inference delay of <5ms.

[0134] When deploying the LSTM prediction model, convert the trained model into the ONNX format and deploy it to the embedded AI acceleration module (such as NVIDIA Jetson) to achieve a real-time inference delay of <5ms. Send the predicted compensation amount to the six-degree-of-freedom platform controller through the EtherCAT bus.

[0135] By measuring the phase difference between the laser beam on the target mirror and the reference mirror and combining the periodic characteristics of the Moiré fringes, sub-pixel displacement detection is realized. When two coherent light beams are superimposed, a small displacement will cause a phase shift of the interference fringes. By analyzing the deformation and position change of the Moiré fringes, the sub-micron-level center offset can be calculated.

[0136] In step S6, the specific process of constructing the measurement optical path and verifying through the Moiré fringes is as follows:

[0137] Step S61: Configure the interferometer parameters, including configuring the parameters of the reference mirror and the target mirror;

[0138] Step S62: Insert a high-density diffraction grating (3000 lines / mm) into the laser path to generate a periodic Moiré fringe substrate;

[0139] Step S63: Monitor the optical path offset in real time through a quadrant detector, and drive the piezoelectric ceramic micromirror to complete beam collimation (accuracy ≤ 0.1 μrad);

[0140] Step S64: The piezoelectric ceramic modulator finely tunes the phase of the reference light at a frequency of 10 kHz to generate dynamic Moiré fringes;

[0141] Step S65: The global shutter camera captures the fringe image at a rate of 2000 fps and transmits it to the image processing unit through an optical fiber;

[0142] Step S66: The image processing unit performs a two-dimensional Fourier transform on the image to extract the phase information of the fundamental frequency component; the specific formula is as follows:

[0143]

[0144] In the formula, F(u, v) represents the frequency domain signal, and φ represents the phase distribution FF1B

[0145] Step S67: Use the cubic spline interpolation method to perform sub-pixel fitting on the fringe center, and the positioning accuracy reaches 1 / 50 pixel (equivalent to 0.02 μm);

[0146] Step S68: Input the center offset Δd into the PID controller to generate a high-precision displacement platform drive signal, and drive the servo drive module to complete position correction; the calculation formula for generating the high-precision displacement platform drive signal is as follows:

[0147]

[0148] Step S69: When the standard deviation of the offset for 5 consecutive iterations < 0.1 μm, it is determined that the calibration is completed.

[0149] In Step S7, the self-healing database architecture includes a real-time layer, an optimization layer, and a disaster recovery layer; the real-time layer uses a time series database to store the calibration raw data updated at the second level; the optimization layer is based on a time series clustering algorithm, stores the historical optimal parameters classified by working conditions in a relational database, and establishes a working condition feature vector index; the disaster recovery layer includes implementing distributed redundant storage of the parameters through blockchain technology;

[0150] The self-healing database uses an LSTM network deployed to monitor the calibration data stream in real time. When it detects that the sensor noise > 3σ or the environmental parameters mutate, it triggers the self-healing process, matches the current working condition with the historical optimal parameter library based on the weighted Euclidean distance, and preferentially calls the calibration parameter group with a matching degree > 90%.

[0151] Embodiment 2

[0152] Refer to Figure 2As shown in the figure, the present invention is a calibration system for a point laser module to point to the center, which can be used to execute the method content of Embodiment 1 of the present invention, including: a basic positioning unit, a sensing feedback unit, a motion control unit, and an auxiliary positioning device;

[0153] The basic positioning unit includes a six-degree-of-freedom module stage and a high-precision displacement platform; the six-degree-of-freedom module stage is a combined mechanism of a slide rail and a rotary table, used to achieve translational motion in the X, Y, and Z axes and rotational motion around the axes; the high-precision displacement platform uses a nanoscale electro-ceramic driver and is built-in with a grating scale closed-loop feedback system;

[0154] The sensing feedback unit includes a multi-axis attitude sensor array and a contact probe group; the multi-axis attitude sensor array is used to monitor the attitude angle of the six-degree-of-freedom module stage in real time, construct a spatial position measurement network, and establish a three-dimensional coordinate system mapping; the contact probe group is used to achieve physical reference point contact detection through a tungsten carbide probe array;

[0155] The motion control unit includes a multi-axis motion controller and a servo drive module; the multi-axis motion controller is a real-time control module using the FPGA and ARM architectures, and integrates a PID parameter self-tuning algorithm to dynamically adjust the servo response characteristics;

[0156] The auxiliary positioning device includes a reference calibration component and an environment compensation module; the reference calibration component includes a grating plate with Moiré fringes and an infrared opposed sensor; the grating plate is used to provide a sub-micron positioning reference; the infrared opposed sensor constitutes a zero calibration system; the motion compensation module includes an active vibration isolation platform and a constant temperature liquid cooling system; the active vibration isolation platform is used to eliminate ground vibration interference; the constant temperature liquid cooling system is used to maintain the thermal stability of the mechanical structure.

[0157] The bottom of the six-degree-of-freedom module stage is rigidly connected to the high-precision displacement platform through an air-floating vibration isolation interface; the surface of the six-degree-of-freedom module stage is integrated with a multi-axis attitude sensor array; the multi-axis attitude sensor array includes a MEMS gyroscope and an accelerometer; the measurement extension arm of the six-degree-of-freedom module stage is equipped with a contact probe group; the tip of the contact probe group is in physical contact with the reference calibration component.

[0158] The high-precision displacement platform is a three-layer structure; the upper layer of the high-precision displacement platform is an air-floating rotary platform, used for pitch angle and yaw angle adjustment; the middle layer of the high-precision displacement platform is a crossed roller guide rail, used to achieve fine adjustment in the X-axis and Y-axis directions; the bottom layer of the high-precision displacement platform is a piezoelectric ceramic nano-displacement table, used for fine adjustment in the Z-axis direction; the high-precision displacement platform is electrically connected to the multi-axis motion controller through an optical fiber encoder.

[0159] The multi-axis motion controller is electrically connected to the motion control unit through the CAN bus; the servo drive module is installed on the high-precision displacement platform; the reference calibration component further includes a laser interferometer mirror and a calibration ball target; the optical path of the laser interferometer mirror is parallel to the axis of the displacement platform; the standard ball target is coaxially aligned with the contact probe.

[0160] It should be noted that in the above system embodiments, the included units are only divided according to functional logic, but are not limited to the above division as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present invention.

[0161] In addition, those of ordinary skill in the art can understand that all or part of the steps in implementing the above embodiments of the method can be completed by instructing relevant hardware through a program, and the corresponding program can be stored in a computer-readable storage medium.

[0162] The preferred embodiments of the present invention disclosed above are only used to help explain the present invention. The preferred embodiments do not describe all the details in detail, nor do they limit the invention to the specific embodiments described. Obviously, many modifications and variations can be made according to the content of this specification. These embodiments are selected and specifically described in this specification to better explain the principles and practical applications of the present invention, so that those skilled in the art can well understand and utilize the present invention. The present invention is only limited by the claims and their full scope and equivalents.

Claims

1. A point laser module pointing center calibration system, comprising a basic positioning unit, a sensor feedback unit, a motion control unit and an auxiliary positioning device, characterized in that: The basic positioning unit includes a six-degree-of-freedom module stage and a high-precision displacement platform; the six-degree-of-freedom module stage is a combination of a slide rail and a rotating table, which is used to achieve X, Y, Z three-axis translation and rotation around the axis; the high-precision displacement platform adopts a nano-level electric ceramic driver and has a built-in grating ruler closed-loop feedback system; The sensor feedback unit includes a multi-axis attitude sensor array and a contact probe group; the multi-axis attitude sensor array is used to monitor the attitude angle of the six-degree-of-freedom module platform in real time, and to build a spatial position measurement network and establish a three-dimensional coordinate system mapping; the contact probe group is used to realize physical reference point contact detection through a tungsten carbide probe array; The motion control unit includes a multi-axis motion controller and a servo drive module; the multi-axis motion controller is a real-time control module using FPGA and ARM architecture, and integrates a PID parameter self-tuning algorithm to dynamically adjust the servo response characteristics; The auxiliary positioning device includes a reference calibration component and an environmental compensation module; the reference calibration component includes a grating plate with moiré fringes and an infrared counter-radiation sensor; The grating plate is used to provide a sub-micron positioning reference; the infrared radiation sensor constitutes a zero-point calibration system; the motion compensation module includes an active vibration isolation platform and a constant temperature liquid cooling system; the active vibration isolation platform is used to eliminate ground vibration interference; the constant temperature liquid cooling system is used to maintain the thermal stability of the mechanical structure.

2. The calibration system for the pointing center of a dot laser module according to claim 1, wherein, The bottom of the six-degree-of-freedom module carrier is rigidly connected to the high-precision displacement platform through an air-floating vibration isolation interface; the surface of the six-degree-of-freedom module carrier is integrated with a multi-axis attitude sensor array; the axis attitude sensor array includes a MEMS gyroscope and an accelerometer; the measurement extension arm of the six-degree-of-freedom module carrier is equipped with a contact probe group; the tip of the contact probe group is in physical contact with the reference calibration component.

3. A calibration system for pointing the center of a dot laser module according to claim 1, characterized in that, The high-precision displacement platform has a three-layer structure; the upper layer of the high-precision displacement platform is an air-floating rotary platform for adjusting the pitch angle and yaw angle; the middle layer of the high-precision displacement platform is a cross roller guide for fine-tuning in the X-axis and Y-axis directions; the bottom layer of the high-precision displacement platform is a piezoelectric ceramic nano-displacement table for fine-tuning in the Z-axis direction; the high-precision displacement platform is electrically connected to the multi-axis motion controller through an optical fiber encoder.

4. A calibration system for pointing the center of a dot laser module according to claim 1, characterized in that, The multi-axis motion controller is electrically connected to the motion control unit via a CAN bus; the servo drive module is installed on a high-precision displacement platform; the benchmark calibration component also includes a laser interferometer reflector and a calibration ball target; the optical path of the laser interferometer reflector is parallel to the axis of the displacement platform; the standard ball target is coaxially aligned with the contact probe.

5. A calibration method for a point laser module to point to the center, characterized in that, The steps include: Step S1: The contact probe is extended to contact the standard ball target to establish an initial coordinate system reference; Step S2: After the six-degree-of-freedom module stage is coarsely positioned by the cross roller guide, the piezoelectric ceramic nano-displacement stage completes the Z-axis fine adjustment, and the air-floating rotary platform provides pitch and yaw angle compensation; Step S3: The multi-axis attitude sensor array collects the three-dimensional attitude; Step S4: The environmental compensation module monitors the environmental parameters in real time and compensates for the change in air refractive index through gradient field modeling; Step S5: The three-dimensional attitude and environmental parameters are synchronously input into the LSTM prediction model to generate a real-time compensation amount; Step S6: Construct a measurement optical path and achieve sub-pixel level center locking through Moiré fringe verification; Step S7: Calibration data is stored in the self-healing database, and historical optimal parameters are automatically called under abnormal working conditions; Step S8: Input the historical optimal parameters into the motion control unit for calibration.

6. A calibration method for the center pointing of a dot laser module according to claim 5, characterized in that, In the said Step S1, when the contact probe extends out and contacts the standard ball target to establish the initial coordinate system benchmark, the specific process is as follows: Step S11: The contact probe extends out, and the probe tip contacts the surface of the standard ball target with a constant pressure; Step S12: The micro-force sensor built in the contact probe monitors the contact force in real time. When the preset threshold is reached, a signal is triggered to lock the current position coordinates of the probe; Step S13: Synchronously collect the three-dimensional coordinates (X1, Y1, Z1) of the probe tip in the robotic arm coordinate system, the known geometric center coordinates (X0, Y0, Z0) of the standard ball target in the global coordinate system, and the environmental temperature parameters collected by the environmental compensation module; Step S14: Solve the coordinate transformation matrix through the least squares method. The specific formula is as follows: (X1, Y1, Z1) T = R × (X0, Y0, Z0) T + T; In the formula, R is the rotation matrix and T is the translation vector; Step S15: The probe makes multi-point contacts along the surface of the standard ball target, and the positioning error of single-point contact is eliminated through spherical equation fitting; Step S16: Correct the coordinate system benchmark according to the environmental temperature parameters collected by the environmental compensation module; Step S17: When the error of the coordinate transformation matrix in three consecutive calibrations is less than the threshold, it is determined that the initial coordinate system is established.

7. A calibration method for a point laser module to point to the center, according to claim 5, characterized in that In the said Step S3, the multi-axis attitude sensor arrays are respectively installed at the four vertex positions on the surface of the six-degree-of-freedom module stage. The calibration matrices of the four multi-axis attitude sensor arrays are uniformly converted to the stage coordinate system. The acceleration and angular velocity data of the four sensors are fused by using the Kalman filter, and the three-dimensional attitude angles of the stage are calculated by the least squares method. The data deviation of each sensor is monitored in real time, and abnormal nodes are removed; then, the measurement error caused by mechanical deformation is compensated by associating the sensor spacing with the kinematic model.

8. A calibration method for the center pointing of a dot laser module according to claim 5, characterized in that, In the said Step S5, the training process of the LSTM prediction model is as follows: Step S51: Collect sensor data and environmental data and align them according to the time stamp to construct a multi-dimensional time series input matrix; Step S52: Use the time series generator to divide the input sequence, and the output is the compensation amount Δθ at the next moment; Step S53: Perform normalization processing on multi-source data; Step S54: Construct the LSTM model architecture; among them, the loss function uses weighted MSE, and the weight of the attitude parameter is set to 3 times that of the environmental parameter; regularization is realized by setting weight_decay = 0.01 in the Adam optimizer to achieve L2 regularization; Step S55: Convert the trained model into the ONNX format and deploy it to the embedded AI acceleration module to achieve a real-time inference delay of <5ms.

9. A calibration method for pointing the center of a dot laser module according to claim 5, characterized in that, In step S6, the specific process of constructing the measurement optical path and verifying it through Moiré fringes is as follows: Step S61: Configure the interferometer parameters; Step S62: Insert a high-density diffraction grating into the laser path to generate a periodic Moiré fringe substrate; Step S63: Real-time monitor the optical path offset through a quadrant detector, and drive the piezoelectric ceramic micromirror to complete beam collimation; Step S64: The piezoelectric ceramic modulator finely adjusts the phase of the reference light at a frequency of 10 kHz to generate dynamic Moiré fringes; Step S65: The global shutter camera captures the fringe image at a rate of 2000 fps and transmits it to the image processing unit through an optical fiber; Step S66: The image processing unit performs a two-dimensional Fourier transform on the image to extract the phase information of the fundamental frequency component; Step S67: Use the cubic spline interpolation method to perform sub-pixel fitting on the fringe center; Step S68: Input the center offset Δd into the PID controller to generate a high-precision displacement platform drive signal, and drive the servo drive module to complete position correction; Step S69: When the standard deviation of the offset for 5 consecutive iterations is less than the threshold, it is determined that the calibration is completed.

10. A calibration method for the pointing center of a dot laser module according to claim 5, characterized in that, In step S7, the self-healing database architecture includes a real-time layer, an optimization layer, and a disaster recovery layer; the real-time layer uses a time series database to store the calibrated raw data updated at the second level; the optimization layer is based on a time series clustering algorithm, classifies and stores the historical optimal parameters according to the working conditions in a relational database, and establishes an index of the working condition feature vectors; the disaster recovery layer includes implementing distributed redundant storage of the parameters through blockchain technology; The self-healing database uses an LSTM network deployed to monitor the calibration data stream in real time. When it detects that the sensor noise > 3σ or the environmental parameters mutate, it triggers the self-healing process, matches the current working condition with the historical optimal parameter library based on the weighted Euclidean distance, and preferentially calls the calibration parameter group with a matching degree > 90%.

Images