A method for calibrating a three-dimensional line laser sensor with a cylindrical rotating shaft

Abstract

The present application relates to the technical field of industrial precision vision measurement, and particularly relates to a three-dimensional line laser sensor and a calibration method of a cylindrical rotating shaft, which aims at the problems that the existing cylindrical shaft calibration method is difficult to consider precision, efficiency and robustness, relies on a precision motion mechanism, needs to know the cylindrical parameters in advance, and has poor working condition adaptability. The present application is configured with two parameter-accurate and differentiated standard constraint rings, which are sleeved on the to-be-calibrated cylinder, relies on a rotating mechanism to cooperate with a PLC and an encoder to complete global synchronous data acquisition, reconstructs multiple groups of dense three-dimensional point clouds with angle information, builds a global optimization objective function that fuses multi-dimensional geometric constraints, and obtains a high-precision cylindrical space axis and body radius through iterative optimization solution, without the need for a precision linear motion mechanism and prior parameters of a workpiece, and has strong anti-interference capability, can quickly complete online precision calibration of industrial cylindrical workpieces, and is suitable for various intelligent manufacturing measurement scenes.

Description

Technical Field

[0001] This invention relates to the field of industrial precision vision measurement technology, and in particular to a calibration method for a three-dimensional line laser sensor and a cylindrical rotating axis, which is applicable to high-precision online calibration of the spatial axis of cylindrical workpieces such as shaft parts, rollers, pipes, and robotic arm joint shafts. Background Technology

[0002] In the fields of industrial precision measurement, automated assembly, and intelligent manufacturing, high-precision and rapid calibration of the spatial axes of cylindrical workpieces (such as shafts, rollers, pipes, and robotic arm joints) is a key technology. The accuracy of axis calibration directly determines the accuracy and reliability of subsequent positioning, docking, assembly, and geometric tolerance inspection.

[0003] Currently, vision-based non-contact axis calibration technology is widely used due to its high efficiency and flexibility. Among them, line structured light 3D vision technology projects a laser line onto the surface of an object, captures its deformed image with a camera, and reconstructs the 3D contour information of the cross-section where the laser line is located by combining the principles of triangulation. It is one of the mainstream techniques for this type of calibration.

[0004] Existing methods for calibrating the axis of a cylinder based on line structured light can be mainly divided into the following two categories: 1. Calibration method based on multi-section fitting This method is the most common calibration approach. The steps typically involve fixing a line structured light sensor (consisting of a camera and a line laser) so that the laser plane forms a certain angle (non-parallel) with the theoretical axis of the cylinder to be calibrated. Then, the sensor moves along the cylinder's axis, or the cylinder rotates around its axis, thereby acquiring laser stripe images of a series of cross-sections on the cylinder's surface. The point cloud of each cross-section is then fitted with a circle or ellipse to obtain the three-dimensional coordinates of a series of circle centers (or ellipse centers). Finally, these center points are fitted with a spatial straight line, and the resulting straight line represents the cylinder's spatial axis. The problems and drawbacks of this method are as follows:

[0005] (1) Complex operation and low efficiency: It requires precise relative motion of sensors or workpieces to collect data from multiple cross sections. The calibration process is long and not suitable for scenarios that require rapid online calibration.

[0006] (2) High dependence on the accuracy of motion mechanism: The guiding accuracy of sensor movement or workpiece rotation will directly introduce errors, affecting the positioning accuracy of multiple cross-section centers, thereby reducing the accuracy of final axis fitting.

[0007] (3) High equipment cost: In order to achieve precise motion, a high-precision translation stage or turntable is often required, which increases the system cost and volume.

[0008] 2. Calibration method based on single-frame images To improve calibration speed, some studies employ single-shot calibration. Typically, a laser plane is projected at a large angle onto the surface of a cylinder, forming an elliptical (or hyperbolic) light stripe on the cylinder. By fitting this light stripe curve and utilizing the known geometric constraints of the cylinder's radius, the direction of the axis within the laser plane is derived in reverse, and the spatial axis is estimated by combining this with pose information from other sensors. The problems and drawbacks of this method are as follows: (1) Limited accuracy and poor robustness: This method relies heavily on the fitting accuracy of a single light stripe and is easily affected by the reflection characteristics of the workpiece surface, ambient light noise and image processing errors. The stability and repeatability of the calibration results are poor.

[0009] (2) The precise radius of the cylinder must be known: The actual radius of the cylinder must be known in advance as a priori condition, which is often difficult to guarantee in practical applications (such as manufacturing tolerances or wear), leading to calibration errors.

[0010] (3) The complete spatial axis cannot be obtained directly: a single view and a single image can usually only determine the directional component of the axis in the laser plane. To obtain the complete spatial straight line equation, it is still necessary to combine complex sensor calibration data or multi-view information, which is a complicated process.

[0011] In summary, existing methods for calibrating the axis of a cylindrical object using line structured light struggle to achieve a good balance between accuracy, efficiency, robustness, and ease of operation. They either sacrifice efficiency and convenience for high accuracy by employing complex motion mechanisms and lengthy data acquisition processes, or they prioritize speed in a single shot at the expense of accuracy and robustness, but this approach is subject to stringent application conditions and lacks versatility.

[0012] Therefore, there is a need in this field for a new method for calibrating the spatial axis of a cylinder that can achieve high precision and robustness without relying on precision motion mechanisms, in order to meet the urgent needs of modern intelligent manufacturing for online, real-time, and flexible measurement. Summary of the Invention

[0013] The purpose of this invention is to overcome the problems of the prior art and provide a calibration method for a three-dimensional line laser sensor and a cylindrical rotating shaft. This method addresses the industry pain points of existing cylindrical rotating shaft laser calibration technology. Traditional calibration methods either rely on high-precision linear motion mechanisms, resulting in cumbersome and time-consuming calibration processes with significant error accumulation, or use single-frame image calibration, which has poor calibration accuracy and robustness to operating conditions. Furthermore, the cylinder radius must be known in advance to complete the calibration, limiting its applicability and making it difficult to balance calibration accuracy, efficiency, and versatility. This fails to meet the practical application needs of industrial online precision calibration.

[0014] The above objectives are achieved through the following technical solutions: A calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis includes the following steps: Step 1, System Layout and Assembly: Fix the three-dimensional line laser sensor so that the laser plane of the three-dimensional line laser sensor is approximately parallel to the theoretical rotation axis of the cylinder to be calibrated; assemble the cylinder into the rotation mechanism; take at least two standard constraint rings with precisely known inner radii and unequal radii; coaxially fit each of the standard constraint rings onto the cylinder, with each standard constraint ring located within the field of view of the three-dimensional line laser sensor camera; Step 2, Synchronous Rotation Scanning Acquisition: Control the rotation mechanism to drive the cylinder to rotate uniformly around its theoretical axis for a complete revolution; the rotation angle is acquired in real time by an angle encoder coaxially connected to the rotation mechanism; the PLC is used to synchronously trigger the three-dimensional line laser sensor to capture laser stripe images and lock the corresponding rotation angles at preset equal angle intervals, thereby acquiring a one-to-one corresponding image sequence and synchronous angle sequence. Step 3: Spatiotemporal Fusion 3D Point Cloud Reconstruction: Process the image sequence frame by frame, extract the pixel coordinates of the laser stripe center line in a single frame image, and distinguish and separate the laser line segments corresponding to the cylinder body, the first standard constraint ring, and the second standard constraint ring; based on the pre-calibrated 3D line laser sensor parameters, use triangulation to reconstruct the pixel coordinates of each laser line segment into a 3D spatial point set; bind the rotation angle corresponding to the acquisition to each 3D spatial point, and generate three sets of dense 3D point clouds carrying rotation angle information, namely the cylinder body point cloud, the first constraint ring point cloud, and the second constraint ring point cloud; Step 4: Construct a global nonlinear optimization objective function: Parametrically construct the spatial axis L of the cylinder to be solved. The spatial axis L is represented by a reference point and a unit direction vector. The variables to be optimized include the reference point, the unit direction vector, and the unknown radius of the cylinder. Construct an objective function, which includes three sums of squared errors. The first term is the sum of the squares of the differences between the Euclidean distances from all points on the cylinder to the spatial axis L and the unknown radius of the cylinder. The second term is the sum of the squares of the differences between the Euclidean distances from all points on the first constraint loop to the spatial axis L and the known radius of the first constraint loop. The third term is the sum of the squares of the differences between the Euclidean distances from all points on the second constraint loop to the spatial axis L and the known radius of the second constraint loop. Step 5: Optimization and Result Output: The objective function is solved iteratively using a nonlinear least squares algorithm until the objective function converges. The high-precision spatial axis equation composed of the optimal reference point and the optimal unit direction vector is output, and the optimal cylinder body radius is output simultaneously.

[0015] As a further optimization of this method, the rotating mechanism is a servo motor or a stepper motor, and the angle encoder is rigidly connected coaxially to the rotating mechanism for real-time, high-precision output of the real-time rotation angle of the cylinder. .

[0016] As a further optimization of this method, the preset equal angle interval of the PLC is selected from 0.1° or 0.5°, and the acquisition time for the cylinder to complete one revolution is controlled within 1 second.

[0017] As a further optimization of this method, the standard constraint ring is a detachable rigid ring. The inner cylindrical surface of each standard constraint ring is calibrated with high precision by a metrology device. The known radius of the first constraint ring is not equal to the known radius of the second constraint ring. After each standard constraint ring is fitted onto the cylinder, its inner cylindrical surface is forced to be collinear with the axis to be calibrated of the cylinder.

[0018] As a further optimization of this method, the laser line segments are distinguished in step 3 by dividing them according to the spatial contour shape of the laser stripes. When the cylinder is stationary, the three-dimensional line laser sensor projects straight laser stripes. During the rotation scanning process of the cylinder, the laser stripes completely sweep across the entire outer surface of the cylinder and the inner surfaces of the two standard constraint rings. In the collected point cloud of the cylinder body, the point cloud of the first constraint ring, and the point cloud of the second constraint ring, each three-dimensional point in each group of point clouds is bound to the rotation angle corresponding to the acquisition time.

[0019] As a further optimization of this method, the equation for the spatial axis L is: Any point in space To the spatial axis Euclidean distance It is obtained by calculating the cross product of vectors.

[0020] As a further optimization of this method, the initialization method of the optimized variables in step 5 is as follows: the point cloud of the cylindrical body is coarsely fitted using principal component analysis (PCA) to obtain the initial value of the unit direction vector; the initial estimated value of the unknown radius of the cylindrical body is set according to the nominal size of the workpiece.

[0021] As a further optimization of this method, the complete mathematical expression of the objective function is as follows: , In the formula: For any three-dimensional point within the point cloud of the cylindrical body, Let be any three-dimensional point within the first constraint loop point cloud. For any three-dimensional point within the second constraint loop point cloud, Representative point To the spatial axis Euclidean distance, The radius of the first constraint loop is known. The radius of the second constraint loop is known. The radius of the cylinder is unknown.

[0022] As a further optimization of this method, the three sets of dense 3D point cloud storage formats are as follows: Point cloud of cylindrical body: , First constraint loop point cloud: , Second constraint loop point cloud: , in, These represent the rotation angles of the cylinder during the acquisition of the corresponding 3D points.

[0023] As a further optimization of this method, the optimal cylindrical body radius output after calibration is simultaneously used for the detection of the cylindrical workpiece's form and position tolerances and outer diameter, realizing the integrated output of axis calibration and workpiece size measurement.

[0024] The present invention provides a calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis. Through a structural arrangement of coaxial assembly of dual standard constraint rings, it establishes calibration constraints for the cylindrical axis without prior radius conditions. Through a global rotation synchronous scanning acquisition method, it achieves complete acquisition of high-precision and high-density working condition data. Through the construction of a global optimization model by fusing multiple sets of point clouds, it achieves accurate solution of spatial axis and cylindrical body parameters. This effectively avoids the error accumulation problem of traditional calibration, simplifies the calibration hardware structure and process, and significantly improves calibration accuracy, efficiency and industrial working condition adaptability. Attached Figure Description

[0025] Figure 1 This is a structural diagram of the calibration system for a three-dimensional line laser sensor and a cylindrical rotation axis in the calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis described in this invention. Figure 2 This is a flowchart of a calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis as described in this invention. Detailed Implementation

[0026] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments. The described embodiments are merely some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0027] This solution provides a calibration method for a 3D line laser sensor and a cylindrical rotation axis. The core of this method lies in controlling the cylinder to rotate around its own axis, simultaneously triggering a line structured light sensor for dense sampling. The absolute spatial dimensional constraints provided by two standard constraint loops with known radii are combined with the full-surface 3D point cloud obtained from the rotational scan to construct a unified objective function containing axis parameters. Through nonlinear optimization, the spatial axis parameters of the cylinder are solved with high precision in a single operation. The innovation of this method lies in: The core innovation at the hardware level is to set up at least two rigid standard constraint rings with different inner diameters after metrological calibration. During the calibration process, the rings are coaxially fitted onto the cylinder being measured. The inner cylindrical surfaces of the two constraint rings are naturally collinear with the axis to be calibrated, generating two sets of spatial curve features with known radii in the laser plane. This provides an absolute dimension anchor point that cannot be shifted for global optimization, completely eliminating the dependence on the prior conditions of the workpiece's own radius. The core innovation of the software algorithm is to construct a unified objective function and simultaneously integrate the geometric constraints of three types of point clouds: statistical fitting constraints of massive point clouds of the cylinder body, and absolute hard constraints of fixed radii of two standard constraint loops; it optimizes and solves all spatial parameters of the axis in one go, which is different from the existing segmented and single-frame local fitting approach. It eliminates the accumulation of segmented errors from the bottom layer of the algorithm and improves the stability and accuracy of calibration. Unlike traditional methods of acquiring discrete cross-sections using an inclined laser plane, this invention uses a laser plane parallel to the theoretical axis of the cylinder, which, in conjunction with an encoder, synchronously triggers a full-domain scan to completely acquire a 360° full-circumference surface point cloud. This eliminates the need for axial linear motion and removes the systematic errors introduced by linear mechanisms.

[0028] (1) Overall composition of the calibration system like Figure 1 The diagram shown is a structural diagram of the three-dimensional line laser sensor and cylindrical rotation axis calibration system of the present invention. The system consists of five main components: Line structured light projection unit and image acquisition unit: together they constitute the fixed three-dimensional line laser sensor; the key arrangement requirement is that the laser plane is approximately parallel to the theoretical rotation axis of the cylinder to be calibrated; when the cylinder is stationary, the laser projects straight stripes, and during rotation, the laser stripes completely scan the entire surface of the cylinder.

[0029] Standard constraint ring set: contains at least two inner radii that are precisely known ( , The rigid circular rings are the first standard constraint ring and the second standard constraint ring. The standard constraint rings are detachable and are coaxially sleeved on the cylinder during the calibration process. The inner cylindrical surfaces of the two constraint rings are forced to be collinear with the axis to be calibrated, providing two sets of independent and accurate absolute cylindrical dimension geometric references.

[0030] Rotation drive and synchronization control unit: includes a precision rotation mechanism, an angle encoder, and a PLC programmable logic controller; the precision rotation mechanism uses a servo motor or a stepper motor to drive the cylinder to rotate at a uniform speed; the angle encoder is rigidly connected to the rotation shaft on the same axis and outputs a high-precision rotation angle θ in real time; the PLC is the core of the system control, receives the encoder angle signal, and synchronously triggers the camera to capture and lock the current angle value at preset small angle intervals for each rotation, so as to achieve a strict one-to-one correspondence between the image and the angle.

[0031] The cylindrical object to be tested: a shaft, roller, or pipe workpiece to be calibrated, assembled on the precision rotating mechanism.

[0032] Data processing and control unit: Receives camera image sequences and PLC synchronized angle sequences, runs a global nonlinear optimization algorithm, and outputs the spatial axis equation of the cylinder and the workpiece radius measurement results.

[0033] (2) Define the complete implementation process like Figure 2 The diagram shown is a flowchart of the calibration system implementation of this invention. The complete calibration method consists of five core steps, as detailed below: Step 1, System Layout and Assembly: Fix the three-dimensional line laser sensor and adjust the laser output plane to make it approximately parallel to the theoretical rotation axis of the cylinder to be calibrated; assemble the cylinder into the precision rotation mechanism; prepare two rigid standard constraint rings with precisely known inner radii and unequal values, and coaxially fit the two standard constraint rings at different positions along the axial direction of the cylinder to ensure that both standard constraint rings fall within the field of view of the three-dimensional line laser sensor camera.

[0034] Step 2, Synchronous Rotation Scanning and Acquisition: The precision rotation mechanism is activated, driving the cylinder to rotate uniformly around its theoretical axis for a complete 360° revolution. The PLC receives the rotation angle signal output by the coaxial angle encoder in real time. When the cylinder rotates to a preset equal angular interval position, two operations are executed simultaneously: First, the camera triggered by the three-dimensional line laser sensor captures a single-frame image containing laser stripes; Second, the current precise rotation angle output by the latch encoder. After the cylinder completes one full rotation, an image sequence is obtained. Angle sequences with strict synchronization The total acquisition time for a single loop should be controlled within 1 second; the preset equal angle interval is preferably 0.1° or 0.5°.

[0035] Step 3: Spatiotemporal fusion and 3D point cloud reconstruction: 1) Frame-by-frame image processing: For each frame of image... Extract all pixel coordinates of the center line of the laser stripe; based on the spatial contour shape of the laser stripe, distinguish and separate three types of laser segments: laser segments on the surface of the cylinder body, laser segments on the inner surface of the first standard constraint ring, and laser segments on the inner surface of the second standard constraint ring.

[0036] 2) 3D point cloud reconstruction: Retrieve the pre-calibrated intrinsic and extrinsic parameters of the 3D line laser sensor and the laser plane pose parameters. Based on the principle of optical triangulation, convert the two-dimensional pixel coordinates of each separated laser line segment into three-dimensional spatial coordinates to obtain the corresponding rotation angle. The three-dimensional point set below .

[0037] 3) Spatiotemporal information binding: for Each point in the three-dimensional space is assigned attributes, including the surface type to which the point belongs (cylinder body / first constraint ring / second constraint ring) and the rotation angle corresponding to the acquisition time. ; Set of points from all angles Merge the points to generate three sets of dense 3D point clouds carrying angle labels: Point cloud of cylindrical body: , First constraint loop point cloud: , Second constraint loop point cloud: , in, These represent the rotation angles of the cylinder during the acquisition of the corresponding 3D points; Step 4: Construct the global nonlinear optimization objective function: 1) Parametric spatial axis model: Let the parametric equation of the spatial axis L of the cylinder to be solved be... ; among which the benchmark point Unit direction vector The two parameters together define the spatial axis L; additional variables to be optimized are also added. , representing the unknown actual radius of the cylinder.

[0038] 2) Definition of point-axis distance geometric constraint: Represents any point in space To a straight line in space The Euclidean distance is calculated using the cross product of vectors; for the rotation angle... Collected cylindrical body points The point is on the axis The distance is theoretically equal to the radius of the cylinder. For the first constraint loop point The point is on the axis Distance theory equals known constants For the second constraint loop point The point is on the axis Distance theory equals known constants .

[0039] 3) The constraints are transformed into a nonlinear least squares objective function. The mathematical expression for the sum of squares of the distance errors between all observation points and the theoretical cylindrical surfaces is as follows: , The first term in the formula is the sum of squared distance errors of all points on the cylinder body; the second term is the sum of squared distance errors of all points on the first constraint loop; and the third term is the sum of squared distance errors of all points on the second constraint loop.

[0040] Step 5: Nonlinear optimization solution and result output: 1) Initialize optimization variables: Construct the overall optimization variable vector Point cloud of the cylindrical body Principal component analysis (PCA) is performed for coarse fitting to obtain an initial estimate of the unit direction vector v; this estimate is then combined with the workpiece's factory nominal dimensions. Initial estimate; benchmark point The center of the point cloud space is taken as the initial value.

[0041] 2) Iterative optimization solution: A nonlinear least squares algorithm is used to iteratively minimize the objective function F, automatically adjusting all parameters in the vector X to continuously reduce the value of the objective function until the function converges, thus obtaining the optimal solution. .

[0042] Output: Optimal Solution within , The combination yields the high-precision equations for the spatial axes of the cylinder; the optimal solution contains... It provides the actual precise radius of the cylinder body and can be simultaneously output for workpiece outer diameter and geometric tolerance detection.

[0043] As a specific embodiment of this method, this embodiment constructs a three-dimensional line laser sensor and cylindrical rotation axis calibration system, using a shaft-like workpiece as the cylinder to be calibrated. The system hardware composition is as follows: Figure 1 As shown; the overall calibration process strictly follows the following... Figure 2 The process nodes shown are executed step by step, and the implementation steps are as follows: Step 1: System Initialization and Deployment Hardware assembly: Fix the three-dimensional line laser sensor and adjust the light output plane of the line laser so that the laser plane is approximately parallel to the theoretical rotation axis of the cylinder to be calibrated; clamp the shaft-type cylindrical workpiece to the servo motor precision rotation mechanism to ensure that the workpiece can rotate smoothly and uniformly around its own theoretical axis; prepare two rigid standard constraint rings, the inner circles of the two constraint rings are both calibrated with high precision by metrology equipment, and the inner radius values ​​of the two are not equal.

[0044] Constraint ring assembly: The first standard constraint ring and the second standard constraint ring are sequentially and coaxially fitted onto different positions along the axis of the cylinder. The axial position of the constraint rings is adjusted to ensure that the two constraint rings fall completely within the field of view of the three-dimensional line laser sensor camera. The inner cylindrical surface of the constraint ring is in close contact with the outer surface of the cylindrical workpiece, and their axes are forced to be collinear.

[0045] Step 2: Install constraint ring and acquire rotational scanning data 1. Parameter configuration: The PLC controller presets the angle acquisition interval to 0.1° or 0.5°; the coaxial angle encoder is rigidly coaxially connected to the output shaft of the servo motor; the servo motor speed is set to ensure that the total time for the cylinder to rotate 360° is controlled within 1 second.

[0046] 2. Synchronous data acquisition process: Start the servo motor to drive the cylinder to rotate at a constant speed; the PLC continuously reads the real-time angle from the encoder. Whenever the cumulative rotation angle reaches an integer multiple of the preset equal angle interval, the PLC synchronously outputs two signals: the first signal triggers the 3D line laser sensor camera to capture an image containing a complete laser stripe. The second latch stores the precise angle output by the current encoder. And store; after the cylinder completes one full rotation, the rotation mechanism stops, ultimately obtaining a one-to-one corresponding image sequence. With synchronization angle sequence ,in The corresponding rotation angle is The laser stripe image at that time.

[0047] Step 3: Data Processing and Spatial Point Cloud Reconstruction 1. Frame-by-frame image processing: Read all images sequentially. Extract all pixel coordinates of the center line of the laser stripes in the image; based on the segmentation features of the stripe contour, divide the single-frame laser stripes into three independent line segments: the laser line segment of the cylinder body, the laser line segment of the first standard constraint ring, and the laser line segment of the second standard constraint ring.

[0048] 2. 3D Point Cloud Reconstruction: Retrieve the parameters of the pre-calibrated 3D line laser sensor system offline, and based on the mathematical model of line structured light triangulation, convert the two-dimensional pixel coordinates of the three laser line segments into three-dimensional spatial coordinates to obtain the current angle. The corresponding three-dimensional point set .

[0049] 3. Spatiotemporal information fusion and binding: for Add a dual attribute label to each 3D point: the first attribute is the surface type (body / constraint loop A / constraint loop B), and the second attribute is the acquisition rotation angle. ; Traverse the set of corresponding points in the entire image Merge all 3D points to generate three sets of complete spatiotemporally fused dense point clouds: Point cloud of cylindrical body: , First constraint loop point cloud: , Second constraint loop point cloud: , Each 3D point in the three sets of point clouds simultaneously carries 3D coordinates. With the acquisition rotation angle .

[0050] Step 4: Construct the overall optimization model 1. Given fixed parameters: precise radius of the first standard constraint ring. The precise radius of the second standard constraint ring ,and .

[0051] 2. Observation input data: Three sets of dense 3D point clouds with angle labels output from step 3. .

[0052] 3. Definition of variables to be optimized: Spatial axis L parameter: reference point Unit direction vector Spatial axis parametric equations ; Additional optimization variable: Unknown radius of the cylindrical workpiece body .

[0053] 4. Complete construction of the objective function: , In the formula, For spatial points To a straight line in space The Euclidean distance is obtained by calculating the cross product of three-dimensional vectors; the first term is the sum of the squared distance errors of all points on the cylinder body, the second term is the sum of the squared distance errors of all points A in the constraint loop, and the third term is the sum of the squared distance errors of all points B in the constraint loop.

[0054] Step 5: Nonlinear optimization solution Optimization variable initialization: Constructing the optimization variable vector ; On the point cloud Perform PCA principal component analysis to extract the principal directions in the point cloud space as the initial values ​​of the unit direction vector v; take The average coordinates of all three-dimensional points in space are used as the reference point. Initial value; set according to the nominal dimensions of the workpiece. Initial estimate.

[0055] Iterative convergence solution: A nonlinear least squares algorithm is used to iteratively minimize the objective function F. All parameters in the vector X are adjusted iteratively until the difference between two adjacent iterations of the objective function is less than a preset convergence threshold. At this point, convergence is determined, and the optimal parameter solution is output. .

[0056] Step 6: Output calibration results Axis calibration result: Optimal solution Inside , The combined equations form a high-precision spatial axis equation for a cylinder, which serves as the core calibration output of this method. Workpiece dimension measurement results: Optimal solution Inside The actual and accurate outer diameter of the cylindrical workpiece is synchronously output to the host computer for workpiece form and position tolerance and dimensional accuracy detection.

[0057] Once the calibration process is complete, the entire data acquisition phase lasts no more than 1 second, and the calibration cycle for the next workpiece can begin directly.

[0058] The above description is merely illustrative of the embodiments of the present invention and is not intended to limit the present invention. For those skilled in the art, any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis, characterized in that, Includes the following steps: Step 1, System Layout and Assembly: Fix the three-dimensional line laser sensor so that the laser plane of the three-dimensional line laser sensor is parallel to the theoretical rotation axis of the cylinder to be calibrated; assemble the cylinder into the rotation mechanism, take at least two standard constraint rings with precisely known inner radii and unequal radii, and coaxially fit each of the standard constraint rings onto the cylinder, with each of the standard constraint rings located within the field of view of the three-dimensional line laser sensor camera; Step 2, Synchronous Rotation Scanning Acquisition: Control the rotation mechanism to drive the cylinder to rotate uniformly around its theoretical axis for a complete revolution; the rotation angle is acquired in real time by an angle encoder coaxially connected to the rotation mechanism; the PLC is used to synchronously trigger the three-dimensional line laser sensor to capture laser stripe images and lock the corresponding rotation angles at preset equal angle intervals, thereby acquiring a one-to-one corresponding image sequence and synchronous angle sequence. Step 3: Spatiotemporal Fusion 3D Point Cloud Reconstruction: Process the image sequence frame by frame, extract the pixel coordinates of the laser stripe center line in a single frame image, and distinguish and separate the laser line segments corresponding to the cylinder body, the first standard constraint ring, and the second standard constraint ring; based on the pre-calibrated 3D line laser sensor parameters, use triangulation to reconstruct the pixel coordinates of each laser line segment into a 3D spatial point set; bind the rotation angle corresponding to the acquisition to each 3D spatial point, and generate three sets of dense 3D point clouds carrying rotation angle information, namely the cylinder body point cloud, the first constraint ring point cloud, and the second constraint ring point cloud; Step 4: Construct a global nonlinear optimization objective function: Parametrically construct the spatial axis L of the cylinder to be solved. The spatial axis L is represented by a reference point and a unit direction vector. The variables to be optimized include the reference point, the unit direction vector, and the unknown radius of the cylinder. Construct an objective function, which includes three sums of squared errors. The first term is the sum of the squares of the differences between the Euclidean distances from all points on the cylinder to the spatial axis L and the unknown radius of the cylinder. The second term is the sum of the squares of the differences between the Euclidean distances from all points on the first constraint loop to the spatial axis L and the known radius of the first constraint loop. The third term is the sum of the squares of the differences between the Euclidean distances from all points on the second constraint loop to the spatial axis L and the known radius of the second constraint loop. Step 5: Optimization and Result Output: The objective function is solved iteratively using a nonlinear least squares algorithm until the objective function converges. The high-precision spatial axis equation composed of the optimal reference point and the optimal unit direction vector is output, and the optimal cylinder body radius is output simultaneously.

2. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, The rotating mechanism is a servo motor or a stepper motor, and the angle encoder is rigidly connected to the rotating mechanism on the same axis for real-time, high-precision output of the real-time rotation angle of the cylinder. .

3. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, The preset equal angle interval of the PLC is selected from 0.1° or 0.5°, and the acquisition time for the cylinder to complete one rotation is controlled within 1 second.

4. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, The standard constraint ring is a detachable rigid ring. The inner cylindrical surface of each standard constraint ring is calibrated with high precision by a measuring device. The known radius of the first constraint ring is not equal to the known radius of the second constraint ring. After each standard constraint ring is fitted onto the cylinder, its inner cylindrical surface is forced to be collinear with the axis to be calibrated of the cylinder.

5. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, In step 3, the laser line segments are distinguished according to the spatial contour shape of the laser stripes. When the cylinder is stationary, the three-dimensional line laser sensor projects straight laser stripes. During the rotation scanning process of the cylinder, the laser stripes completely sweep across the entire outer surface of the cylinder and the inner surfaces of the two standard constraint rings. In the collected point cloud of the cylinder body, the point cloud of the first constraint ring, and the point cloud of the second constraint ring, each three-dimensional point in each group of point clouds is bound to the rotation angle corresponding to the acquisition time.

6. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, The equation for the spatial axis L parameter is: Any point in space To the spatial axis Euclidean distance It is obtained by calculating the cross product of vectors.

7. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, In step 5, the initialization method for the optimized variables is as follows: Principal Component Analysis (PCA) is used to coarsely fit the point cloud of the cylindrical body to obtain the initial value of the unit direction vector; the initial estimated value of the unknown radius of the cylindrical body is set according to the nominal size of the workpiece.

8. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 7, characterized in that, The complete mathematical expression for the objective function is: + + In the formula: For any three-dimensional point within the point cloud of the cylindrical body, Let be any three-dimensional point within the first constraint loop point cloud. For any three-dimensional point within the second constraint loop point cloud, Representative point To the spatial axis Euclidean distance, The radius of the first constraint loop is known. The radius of the second constraint loop is known. The radius of the cylinder is unknown.

9. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, The three sets of dense 3D point cloud storage formats are as follows: Point cloud of cylindrical body: First constraint loop point cloud: Second constraint loop point cloud: in, These represent the rotation angles of the cylinder during the acquisition of the corresponding 3D points.

10. The calibration method for a three-dimensional line laser sensor and a cylindrical rotation axis according to claim 1, characterized in that, The optimal cylindrical body radius output after calibration is used simultaneously for the detection of cylindrical workpiece form and position tolerances and outer diameter, realizing integrated output of axis calibration and workpiece size measurement.

Images