Helicopter pose determination method used before automatic paint spraying

Abstract

The invention belongs to the technical field of helicopter surface spraying, and particularly relates to a helicopter pose determination method used before automatic paint spraying, which comprises the following steps: step 1, according to the overall appearance structure characteristics of a helicopter, dividing the helicopter into a front-section fuselage and a rear-section fuselage according to the space, and selecting feature point positions on the front-section fuselage and the rear-section fuselage; 2, establishing a workshop unified coordinate system, and determining the pose of the robot in the workshop coordinate system; 3, scanning the feature point positions of the front helicopter body, and fitting the feature point positions with the feature point positions at the corresponding positions in the original helicopter digital model; 4, scanning, correcting and compensating feature point positions of a rear-section helicopter body, and fitting the feature point positions with feature point positions at corresponding positions in the original helicopter digital model; 5, a spraying track is generated, and dynamic correction is carried out; according to the method, for the specificity of a helicopter product, feature positioning points are selected, coordinate system iteration calibration is carried out in a segmented recognition and fitting mode, the actual posture of the helicopter is accurately recognized, and data support is provided for spraying track correction of a paint spraying robot.

Description

Technical Field

[0001] This invention belongs to the field of helicopter surface spraying technology, specifically relating to a method for determining the helicopter's attitude before automatic painting. Background Technology

[0002] Currently, the painting of the entire exterior surface of helicopters in China is done manually. However, this method suffers from problems such as unstable paint quality, poor appearance consistency, low work efficiency, low paint utilization, and harsh working environment. It can no longer meet the high requirements of the helicopter market for product appearance, performance, batch production, and operator occupational health. Therefore, an automated whole-aircraft painting system has emerged.

[0003] The spraying trajectory of an automated painting system is programmed in a specific coordinate system based on a standard digital model. During construction, coordinate system transformation is required based on the actual position of the object to be painted to achieve precise correction of the spraying trajectory. Currently, a common method for pose determination is to scan the entire machine body using visual recognition equipment to obtain the position and shape information of the object to be painted. However, this method suffers from drawbacks such as long processing time, high computational load, and low efficiency. Therefore, a more convenient, faster, and more accurate pose recognition method is urgently needed to improve the operational efficiency of the entire automated painting system. Summary of the Invention

[0004] The technical problem to be solved by the present invention is that the current method of determining the pose in helicopter painting is to scan the entire fuselage with a visual recognition device to obtain the position and shape information of the object to be painted, which has the disadvantages of long time, large amount of calculation and low efficiency. The purpose of this invention is to address the challenges of determining the helicopter's pose before automated painting, considering the complex freeform surface shape of the helicopter's overall surface and the limited space required for automated painting of local curved surfaces. This invention addresses the requirements of high-speed, high-efficiency measurement of large-scale helicopters, dynamic pose recognition and positioning of production units, and rapid identification and 3D reconstruction of the work surface. Specifically, considering the unique characteristics of helicopters, feature positioning points are selected, and a segmented recognition and fitting approach is used for iterative coordinate system calibration. This aims to accurately identify the actual helicopter pose and provide data support for the trajectory correction of the painting robot.

[0005] Technical solution: A method for determining the helicopter's pose before automatic painting includes the following steps: Step 1: Based on the overall shape and structural characteristics of the helicopter, divide the helicopter into a front fuselage and a rear fuselage according to space, and select characteristic points on the front fuselage and the rear fuselage; Step 2: Establish a unified coordinate system for the factory and determine the robot's pose within that coordinate system; Step 3: Scan the feature points of the front fuselage and fit them with the feature points at the corresponding positions in the original helicopter digital model; Step 4: Scan, correct, and compensate the feature points of the rear fuselage, and fit them with the feature points of the corresponding positions in the original helicopter digital model; Step 5: Generate the spraying trajectory and perform dynamic correction.

[0006] Furthermore, in step 1, the helicopter is divided into a front section and a rear section based on the boundary position at the connection between the helicopter cockpit fuselage and the tail boom; a transition feature point is selected in the front section and the rear section respectively, and another feature point is selected near the boundary between the front and rear sections.

[0007] Furthermore, the characteristic point at the front of the helicopter is the outline of the nose radome; the characteristic point at the rear is the outline of the vertical tail; and the transition characteristic point is the outline of the end of the air intake.

[0008] Furthermore, step 2 specifically includes the following steps: Step 2.1: Establish the factory coordinate system for the actual painting of the helicopter; Step 2.1.1: Select several reference points in the factory building. The reference points are evenly distributed and cover the main area of ​​the factory building. Step 2.1.2: Measure the three-dimensional coordinates of each reference point in the laser tracker coordinate system sequentially using a laser tracker; Step 2.1.3: Define the unit vector of the x-axis of the factory coordinate system, denoted as... The unit vector along the y-axis of the factory coordinate system is denoted as... The unit vector along the z-axis of the factory coordinate system is denoted as... The coordinates of the origin of the factory coordinate system are denoted as... ; Step 2.1.4: Transform the three-dimensional coordinates of the reference point in the laser coordinate system to the defined factory coordinate system using the transformation relation matrix H, thus completing the establishment of the factory coordinate system.

[0009] Step 2.2: Calibrate the robot's pose in the factory coordinate system: Step 2.2.1: Establish a tool coordinate system on the end flange of the painting robot; Step 2.2.2: Adjust the spatial pose of the flange multiple times, and after each spatial pose adjustment, record the coordinates of the tool coordinate system origin in the robot coordinate system, forming a point set containing N sampling points. Simultaneously, the laser tracker captures the spatial coordinates of the N sampling points in the factory coordinate system in real time, constructing a point set. ; Step 2.2.3: Based on point sets P and Q, construct an optimization model of rotation matrix R and translation vector t. Solve the transformation relationship between the robot coordinate system and the factory coordinate system through iterative calculation. The robot's pose in the factory coordinate system can be obtained through the transformation relationship.

[0010] Furthermore, step 2.2.3 iteratively calculates and solves the transformation relationship between the robot coordinate system and the factory coordinate system, including: Step 2.2.3.1: Calculate the geometric centroid of point set P respectively. Coordinates, geometric centroid of point set Q coordinate;

[0011]

[0012] Step 2.2.3.2: Find the geometric centroid of point set P. Geometric centroid of point set Q Alignment: Based on the centroid-aligned point sets P and Q, construct the cross-coordinate system covariance matrix H:

[0013] In the formula, T represents the transpose operation; Step 2.2.3.3: For the covariance matrix Perform singular value decomposition to obtain its orthogonal basis and characteristic structure:

[0014] in, and It is an orthogonal matrix. It is a diagonal matrix; Step 2.2.3.4: Based on the covariance matrix The singular value decomposition result is used to obtain the optimal rotation matrix R through orthogonal matrix combination: ; Step 2.2.3.5: Calculate the translation vector t using the centroid coordinates of point sets P and Q and the rotation matrix:

[0015] Step 2.2.3.6: The rigid transformation relationship between the robot base coordinate system and the factory coordinate system is expressed as follows: : .

[0016] Furthermore, step 3 specifically includes: Step 3.1: A 3D camera is mounted on the painting robot. Using the camera's coordinate system, multiple points are scanned to determine the front and transition feature points of the helicopter in its actual parking position, thus obtaining the relative positional relationship between the robot flange and the robot base. Then, based on the transformation relationship between the robot base and the factory coordinate system... And the relative positions of the camera and the flange. Solve for the relative positional relationship between the camera and the factory building. :

[0017] Step 3.2: Convert the point cloud data of the helicopter's front feature points and transition feature points scanned in the camera coordinate system to the factory coordinate system; Step 3.3: Using the centroid translation vector and rotation matrix, the cloud data of the helicopter's front section feature points and transition feature points in the factory coordinate system are fitted with the feature points at the corresponding positions in the original helicopter digital model. Through error iteration optimization, a coordinate transformation relationship that can fit both the helicopter's front section feature points and transition feature points is obtained.

[0018] Furthermore, step 3.3 specifically includes: Step 3.3.1: Calculate the centroid P1 of the overall contour of the front feature points of the helicopter and the centroid P2 of the overall contour of the transition feature points; Step 3.3.2: Calculate the overall center of gravity P0 of the front section of the computer body using P1 and P2; Step 3.3.3: Calculate the overall center of gravity of the original forward fuselage of the original helicopter digital model, denoted as P0. * ; Step 3.3.4: Through P0 and P0 * Translation vector of the center of gravity of the front part of the computer This yields the translational transformation relationship of the front section of the fuselage; Step 3.3.5: The centroid vector is formed by the centroid P1 of the overall contour of the helicopter's forward feature points and the centroid P2 of the overall contour of the transition feature points. And the centroid vector of the feature point at the corresponding position in the original helicopter digital model. Calculate the centroid vector centroid vector The included angle θ and the cross product Determine the rotation axis and rotation angle of the centroid vector offset, and then obtain the rotation transformation relation matrix N; Step 3.3.6: Transfer the centroid P1 of the overall contour of the helicopter's forward feature points and the centroid P2 of the overall contour of the transition feature points using the centroid translation vector. The rotation transformation relation matrix N is transformed to the original helicopter digital model coordinate system to obtain P1' and P2'. P1' and P2' are then calculated respectively. * P2' and P2 * The error between them, and between P1' and P1 * The error between P2' and P2 * The sum of errors, i.e., the total error E; Step 3.3.7: Employ the gradient descent optimization method, using the total error E to optimize the centroid translation vector. The rotation transformation relation matrix N is iteratively optimized to continuously reduce the total error E until the total error E converges to a sufficiently small value, thus obtaining the optimal centroid translation transformation relation. With the rotation transformation relationship matrix N', the feature points at the front and the transition feature points of the helicopter reach the best fitting state with the corresponding feature points in the original helicopter digital model.

[0019] Furthermore, step 4 specifically includes: Step 4.1: The 3D camera is installed on the painting robot. The camera coordinates are used to scan the rear feature points and transition feature points of the helicopter in the actual parking position. The scan data is then corrected and compensated. Step 4.2: Convert the point cloud data of the helicopter's rear feature points and transition feature points scanned in the camera coordinate system to the factory coordinate system; Step 4.3: Using the centroid translation vector and rotation matrix, the cloud data of the helicopter's rear section feature points and transition feature points in the factory coordinate system are fitted with the feature points at the corresponding positions in the original helicopter digital model. Through error iteration optimization, a coordinate transformation relationship that can fit both the helicopter's rear section feature points and transition feature points is obtained.

[0020] Furthermore, step 4.3 specifically includes: Step 4.3.1: Calculate the centroid P3 of the overall contour of the feature points at the front and rear sections of the helicopter and the centroid P2 of the overall contour of the transition feature points; Step 4.3.2: Calculate the overall center of gravity P of the rear section of the computer using P3 and P2. 01 ; Step 4.3.3: Calculate the original overall center of gravity of the rear fuselage of the original helicopter digital model, denoted as P. 01 * ; Step 4.3.4: via P 01 and P 01 * Computer rear centroid translation vector This yields the translational transformation relationship of the rear section of the fuselage; Step 4.3.5: The centroid vector is formed by the centroid P3 of the overall contour of the helicopter's rear feature points and the centroid P2 of the overall contour of the transition feature points. And the centroid vector of the feature point at the corresponding position in the original helicopter digital model. Calculate the centroid vector centroid vector The included angle θ' and cross product Determine the rotation axis and rotation angle of the centroid vector offset, and then obtain the rotation transformation relation matrix T; Step 4.3.6: Transfer the centroid P3 of the overall contour of the front and rear feature points of the helicopter and the centroid P2 of the overall contour of the transition feature points using the centroid translation vector. The rotation transformation relation matrix T is used to transform the system to the original helicopter digital model coordinate system, resulting in P3' and P2'. P3' and P2' are then calculated respectively. * P2' and P2 * The error between them, and between P3' and P3 * The error between P2' and P2 * The sum of the errors, i.e., the total error E'; Step 4.3.7: Use the gradient descent optimization method and the total error E' to optimize the centroid translation vector. The rotation transformation relation matrix T is iteratively optimized to continuously reduce the total error E' until the total error E' converges to a sufficiently small value, thus obtaining the optimal centroid translation transformation relation. With the rotation transformation relationship matrix T', the feature points of the rear section and the transition feature points of the helicopter reach the best fitting state with the corresponding feature points in the original helicopter digital model.

[0021] Furthermore, step 5 specifically includes: (1) In the theoretical coordinate system, the i-th spraying pose of the helicopter's front spraying trajectory is set as follows: ,in Euler angles represent rotations about the Z-axis, Y-axis, and X-axis of the theoretical coordinate system in sequence; (2) The i-th spraying position of the front spraying trajectory of the helicopter is transformed by the rotation transformation matrix N' and the centroid translation transformation matrix N'. After transformation, the adjusted spraying position coordinates are obtained. ;

[0022] (3) Adjust the i-th spraying attitude of the helicopter's front spraying trajectory: (3.1) Calculate the attitude matrix , ,in, Represents the rotation matrix about the Z, Y, and X axes; (3.2) The attitude matrix Converted to Euler angles in the theoretical coordinate system ZYX axis order Then the transformed spray pose of the i-th spray pose of the helicopter's front spray trajectory is: ; Follow steps (1) to (3) to complete the adjustment of all spraying positions of the front spraying trajectory of the helicopter; (4) In the theoretical coordinate system, the j-th spraying pose of the helicopter's rear spraying trajectory is set as follows: ,in Euler angles represent rotations about the Z-axis, Y-axis, and X-axis of the theoretical coordinate system in sequence; (5) The j-th spraying position of the rear spraying trajectory of the helicopter is transformed by the rotation transformation matrix T' and the center of gravity translation transformation matrix. After transformation, the adjusted spraying position coordinates are obtained. ;

[0023] (6) Adjust the j-th spraying posture of the front and rear spraying trajectories of the helicopter: (6.1) Calculate the attitude matrix , ,in, Represents the rotation matrix about the Z, Y, and X axes; (6.2) The attitude matrix Converted to Euler angles in the theoretical coordinate system ZYX axis order Then the spray pose after the transformation of the j-th spray pose of the helicopter's rear spray trajectory is: ; Follow steps (4) to (6) to complete the adjustment of all spraying positions of the rear spraying trajectory of the helicopter; Based on the piecewise coordinate transformation relationship, the theoretical spraying trajectory of the front and rear sections of the helicopter is mapped to the actual fuselage coordinate system to form the actual spraying trajectory of the front and rear sections of the helicopter fuselage. The actual spraying trajectories of the front and rear sections are smoothly connected through a sinusoidal transition function, forming a gradual fusion zone near the interface. Then, the offset of selected feature points on the front and rear sections of the machine is captured in real time by a laser tracker, and the transformation parameters are dynamically updated by a recursive estimation algorithm to achieve closed-loop correction of the spraying trajectory during the spraying process.

[0024] Beneficial effects This invention proposes a method for determining the helicopter's attitude before automatic painting. Based on the helicopter's fuselage shape characteristics, it adopts a segmented measurement, fitting, and iterative correction approach to achieve helicopter attitude positioning and paint trajectory calibration before automatic painting. The main advantages are as follows: (1) Improve fitting accuracy By using two-point fitting and error iteration optimization, the accuracy of feature point registration and fitting can be effectively improved.

[0025] (2) Reduce computational complexity Segmented calculation can process data from different points simultaneously, and downsample each feature point independently, reducing the number of points to be registered in a single operation and effectively accelerating convergence.

[0026] (3) Adapting to deformation due to differences in fuselage structure To address the unique external characteristics of the helicopter fuselage, the fuselage is divided into two sections, front and rear, at the junction of the fuselage and tail boom, significantly reducing errors caused by deformation due to differences in structural rigidity. Attached Figure Description

[0027] Figure 1 This is a schematic diagram of the feature points and segment positions of the present invention. Detailed Implementation The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. The described embodiments are only some embodiments of the present invention, and not all embodiments. 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.

[0028] The present invention will now be described in further detail with reference to the accompanying drawings and specific examples: The present invention proposes a helicopter pose determination method for automatic painting, which achieves self-calibration and dynamic mutual calibration through an optical sensing system and video stream-based artificial intelligence technology, automatically completes the three-dimensional scanning model of the aircraft, and achieves fast and high-precision positioning within the absolute workspace, thereby effectively dividing the workspace of the robotic arm and efficiently measuring its surface geometric information.

[0029] Specifically, the panoramic infrastructure designed using CCTAG markers enables pairwise extrinsic parameter calibration between multiple cameras and multiple LiDARs in a single capture, improving calibration efficiency and accuracy. In large-scale measurement fields, a unified reference coordinate system is established to ensure all equipment is positioned within the same spatial coordinate system. By constructing the relative motion relationships between the equipment, the operation of each device is coordinated, achieving global consistency.

[0030] Advanced point cloud data simplification algorithms, such as those based on RGB and contour preservation, are employed to reduce data volume and improve processing efficiency. A distributed processing architecture is utilized to distribute data processing tasks to various sensor nodes, enabling preliminary processing locally using edge computing technology. This reduces data transmission volume and improves system real-time performance and response speed.

[0031] To determine the relative position of the helicopter target, a rapid acquisition strategy with low computational cost is proposed. This strategy optimizes sensor resource allocation through rapid decision-making based on structured information, thereby improving measurement accuracy. By intelligently fusing information from multiple sensors, positioning and production resources are rationally allocated, providing real-time feedback during dynamic operation monitoring and decision support for subsequent operation planning. Ultimately, rapid positioning and high-precision measurement of typical geometric features on the work surface are achieved, enabling accurate geometric reconstruction.

[0032] Based on the 3D reconstruction of the working surface using features on the aircraft (the front apex of the nose radome, the lower edge of the cockpit door window, the outline of the air intake end, the side line of the horizontal stabilizer, the upper edge of the tail boom, etc.), the working surface is reconstructed using point cloud simplification, point cloud matching, and point cloud stitching methods, and high-precision local reconstructed surface information is fitted. During the implementation of the measurement field project, factors such as sensor basic parameters, factory building obstruction, explosion-proof requirements, and sensor parameter calibration requirements are comprehensively considered. The overall measurement field planning is completed, including sensor layout, factory construction requirements, power and gas supply layout, data transmission layout, and marker board layout, and a unified coordinate system for the factory is constructed.

[0033] The helicopter attitude determination method proposed in this invention for automatic painting includes the following steps: Step 1: Based on the overall shape and structural characteristics of the helicopter, select unique structural points on one side of the helicopter as product feature points to determine the actual position and posture of the helicopter after entering the painting area; divide the helicopter into a front fuselage and a rear fuselage according to space, and select feature points on the front fuselage and rear fuselage based on the helicopter's large curvature and many protruding structures on the outer surface. In step 1, the helicopter is divided into a front section and a rear section based on the boundary between the connection between the helicopter cockpit fuselage and the tail boom. A transition feature point is selected in both the front and rear sections, and another feature point is selected near the boundary between the front and rear sections.

[0034] Preferably, the characteristic point at the front of the helicopter is the outline of the nose radome; the characteristic point at the rear is the outline of the vertical tail; and the transition characteristic point is the outline of the end of the air intake, such as... Figure 1 As shown.

[0035] Based on the unique aerodynamic shape and complex curved surface features of helicopters, a structure with significant geometric characteristics is selected as the positioning reference. Using the connection surface between the cockpit and tail boom as the dividing plane, the fuselage is divided into front and rear sections, with transition feature points added in the dividing area to enhance continuity. For the front section, the outline of the nose radome is selected as the main feature point, and features are defined along its axial symmetry line according to the parametric equation r( θ )= R ( θ )[cos θ sin θ ,0] T ( θ ∈[− π / 12, π

[12] Continuously collect point cloud data; lock the end contour line of the air intake in the transition area, and evenly distribute set points around the lip; in the rear section, take the leading edge contour line of the vertical tail as the core feature, and extract cross-sectional points at fixed intervals along the height direction. At the same time, four corner auxiliary points are added to the bottom surface of the cockpit door hinge mounting base to improve the attitude sensitivity by utilizing its high-precision installation characteristics. By constructing a spatial topology network of the front, rear and transition sections, a feature point monitoring system covering the entire aircraft is formed.

[0036] Step 2: Establish a unified coordinate system for the factory and determine the robot's pose within that coordinate system; Step 2 specifically includes the following steps: Step 2.1: Establish a factory coordinate system for actual spraying of helicopters. The purpose of establishing the factory coordinate system is to obtain a unified reference coordinate system and unify the camera coordinate system, aircraft coordinate system, and robot coordinate system into the factory coordinate system. Step 2.1.1: Select several reference points in the factory building. The reference points are evenly distributed and cover the main area of ​​the factory building. Step 2.1.2: Measure the three-dimensional coordinates of each reference point in the laser tracker coordinate system sequentially using a laser tracker; Step 2.1.3: Define the unit vector of the x-axis of the factory coordinate system, denoted as... The unit vector along the y-axis of the factory coordinate system is denoted as... The unit vector along the z-axis of the factory coordinate system is denoted as... The coordinates of the origin of the factory coordinate system are denoted as... ; Step 2.1.4: Transform the three-dimensional coordinates of the reference point in the laser coordinate system to the defined factory coordinate system using the transformation relation matrix H, thus completing the establishment of the factory coordinate system.

[0037] Step 2.2: Calibrate the robot's pose in the factory coordinate system: Step 2.2.1: Utilize the high-precision spatial positioning capability of the laser tracker to establish a tool coordinate system on the robot's end flange as a reference for subsequent calibration processes; Step 2.2.2: Adjust the spatial pose of the flange using the robot control system to achieve diverse spatial configurations within the robot base coordinate system. During multiple adjustments of the flange's spatial pose, after each adjustment, record the coordinates of the tool coordinate system origin in the robot coordinate system using the robot teach pendant, forming a point set containing N sampling points. Simultaneously, the laser tracker captures the spatial coordinates of the N sampling points in the factory coordinate system in real time, constructing a point set. ; Step 2.2.3: Based on point sets P and Q, construct an optimization model of rotation matrix R and translation vector t using the least squares method. Solve the transformation relationship between the robot coordinate system and the factory coordinate system through iterative calculation. The robot's pose in the factory coordinate system can be obtained through the transformation relationship.

[0038] This step involves finding the optimal rotation matrix R and translation vector t in three-dimensional space such that the transformed point set... With the target point The error is minimized, therefore the objective function can be established:

[0039] Specifically, the following steps are included: Step 2.2.3.1: Calculate the geometric centroid of point set P respectively. Coordinates, geometric centroid of point set Q coordinate;

[0040]

[0041] Step 2.2.3.2: Find the geometric centroid of point set P. Geometric centroid of point set Q Alignment: Based on the centroid-aligned point sets P and Q, a cross-coordinate system covariance matrix H is constructed to quantify the spatial distribution correlation between the two point cloud sets.

[0042] In the formula, T represents the transpose operation; Step 2.2.3.3: For the covariance matrix Perform singular value decomposition to obtain its orthogonal basis and characteristic structure:

[0043] in, and It is an orthogonal matrix. It is a diagonal matrix; Step 2.2.3.4: Based on the covariance matrix The singular value decomposition result is used to obtain the optimal rotation matrix R through orthogonal matrix combination: ; Step 2.2.3.5: Calculate the translation vector t using the centroid coordinates of point sets P and Q and the rotation matrix:

[0044] Step 2.2.3.6: The rigid transformation relationship between the robot base coordinate system and the factory coordinate system is expressed as follows: : .

[0045] Step 3: Scan the feature points of the forward fuselage and fit them with the corresponding feature points in the original helicopter digital model: This step involves mounting a 3D camera to the robot's end effector via a quick-change device, and using a structured light sensor to determine the actual parking pose of the helicopter. Figure 1 The feature points (1) and (2) shown are scanned at multiple points. The relative positional relationship between the robot flange and the robot base can be obtained through the robot teach pendant. Then, based on the transformation relationship between the robot base and the factory coordinate system... And the relative positions of the camera and the flange. (Assuming the solution is obtained in advance), the relative positional relationship between the camera and the factory building can be determined. :

[0046] Therefore, point cloud data scanned in the camera coordinate system can be converted to the factory coordinate system, thereby achieving point cloud stitching.

[0047] The measured coordinates of the front feature point (1) and the transition feature point (2) are fitted with the corresponding feature points of the standard digital model stored in the system using the centroid translation vector and rotation matrix. Through error iteration optimization, a coordinate transformation relationship that fits well with both feature points (1) and (2) is obtained.

[0048] Step 3 specifically includes the following steps: Step 3.1: A 3D camera is mounted on the painting robot. Using the camera's coordinate system, multiple points are scanned to determine the front and transition feature points of the helicopter in its actual parking position, thus obtaining the relative positional relationship between the robot flange and the robot base. Then, based on the transformation relationship between the robot base and the factory coordinate system... And the relative positions of the camera and the flange. Solve for the relative positional relationship between the camera and the factory building. :

[0049] Step 3.2: Convert the point cloud data of the helicopter's front feature points and transition feature points scanned in the camera coordinate system to the factory coordinate system; Step 3.3: Using the centroid translation vector and rotation matrix, the cloud data of the helicopter's front section feature points and transition feature points in the factory coordinate system are fitted with the feature points at the corresponding positions in the original helicopter digital model. Through error iteration optimization, a coordinate transformation relationship that can fit both the helicopter's front section feature points and transition feature points is obtained.

[0050] Step 3.3 specifically includes: Step 3.3.1: Calculate the centroid P1 of the overall contour of the front feature points of the helicopter and the centroid P2 of the overall contour of the transition feature points; The calculation method for the centroid P1(x1,y1,z1) of the overall contour of the feature point (1) is as follows.

[0051]

[0052] in, The feature point (1) represents the number of points in the point cloud data. , , The coordinates of the i-th point in the point cloud data (1) are respectively represented by the coordinates of the i-th point in x, y, and z.

[0053] Similarly, the centroid P2(x2,y2,z2) of the overall contour of the feature point (2) can be calculated.

[0054] Step 3.3.2: Calculate the overall center of gravity P0 of the front section of the computer body via P1 and P2; Calculate the overall center of gravity P0 of the front section of the computer body via P1 and P2 ( , , ); Step 3.3.3: Using the same method as in Step 3.3.2, calculate the overall center of gravity of the original forward fuselage of the original helicopter digital model, denoted as P0. * , denoted as P0 * ( , , ); where the centroid P1 (x1) of the overall outline of the feature points at the front of the helicopter * y1 * z1 * ） The centroid P1 of the overall contour of the front feature points of the original helicopter digital model * The centroid P2 of the overall contour of the transition feature points corresponds to the centroid P2 of the overall contour of the transition feature points in the original helicopter digital model. * (x2) * y2 * z2 * (Corresponding position) Step 3.3.4: Through P0 and P0 * Translation vector of the center of gravity of the front part of the computer This yields the translational transformation relationship of the forward section of the fuselage; the translation vector of the center of gravity. = P0 * - P0 = (m x ,m y ,m z That is, the translation transformation relationship of the front section of the fuselage is M(x,y,z) = (x+m) / ( ... x ,y+m y ,z+m z ).

[0055] Step 3.3.5: The centroid vector is formed by the centroid P1 of the overall contour of the helicopter's forward feature points and the centroid P2 of the overall contour of the transition feature points. , = (x2-x1, y2-y1, z2-z1), and the centroid vector of the feature point at the corresponding position in the original helicopter digital model. , = (x2) * -x1 * ,y2 * -y1 * z2 * -z1 * ); Calculate the centroid vector centroid vector The included angle θ and the cross product , = × Determine the rotation axis and rotation angle of the centroid vector offset, and then obtain the rotation transformation relation matrix N; Step 3.3.6: Transfer the centroid P1 of the overall contour of the helicopter's forward feature points and the centroid P2 of the overall contour of the transition feature points using the centroid translation vector. The rotation transformation relation matrix N is transformed to the original helicopter digital model coordinate system to obtain P1' and P2'. P1' and P2' are then calculated respectively. * P2' and P2 * The error between them, and between P1' and P1 * The error between P2' and P2 * The sum of errors, i.e., the total error E; P1' and P1 * P2' and P2 * The errors between them are denoted as e1= e2= The total error is E = e1 + e2.

[0056] Step 3.3.7: Employ the gradient descent optimization method, using the total error E to optimize the centroid translation vector. The rotation transformation relation matrix N is iteratively optimized to continuously reduce the total error E until the total error E converges to a sufficiently small value, thus obtaining the optimal centroid translation transformation relation. With the rotation transformation relationship matrix N', the feature points at the front and the transition feature points of the helicopter reach the best fitting state with the corresponding feature points in the original helicopter digital model.

[0057] Step 4: The same method as in Step 3 is used, namely, scanning, correcting, and compensating the feature points of the rear fuselage, and fitting them with the corresponding feature points in the original helicopter digital model; specifically including: Step 4.1: The 3D camera is installed on the painting robot. The camera coordinates are used to scan the rear feature points and transition feature points of the helicopter in the actual parking position. The scan data is then corrected and compensated. Step 4.2: Convert the point cloud data of the helicopter's rear feature points and transition feature points scanned in the camera coordinate system to the factory coordinate system; Step 4.3: Using the centroid translation vector and rotation matrix, the cloud data of the helicopter's rear section feature points and transition feature points in the factory coordinate system are fitted with the feature points at the corresponding positions in the original helicopter digital model. Through error iteration optimization, a coordinate transformation relationship that can fit both the helicopter's rear section feature points and transition feature points is obtained.

[0058] Step 4.3 specifically includes: Step 4.3.1: Calculate the centroid P3 of the overall contour of the feature points at the front and rear sections of the helicopter and the centroid P2 of the overall contour of the transition feature points; Step 4.3.2: Calculate the overall center of gravity P of the rear section of the computer using P3 and P2. 01 ; Step 4.3.3: Calculate the original overall center of gravity of the rear fuselage of the original helicopter digital model, denoted as P. 01 * ; Step 4.3.4: via P 01 and P 01 * Computer rear centroid translation vector This yields the translational transformation relationship of the rear section of the fuselage; Step 4.3.5: The centroid vector is formed by the centroid P3 of the overall contour of the helicopter's rear feature points and the centroid P2 of the overall contour of the transition feature points. And the centroid vector of the feature point at the corresponding position in the original helicopter digital model. Calculate the centroid vector centroid vector The included angle θ' and cross product Determine the rotation axis and rotation angle of the centroid vector offset, and then obtain the rotation transformation relation matrix T; Step 4.3.6: Transfer the centroid P3 of the overall contour of the front and rear feature points of the helicopter and the centroid P2 of the overall contour of the transition feature points using the centroid translation vector. The rotation transformation relation matrix T is used to transform the system to the original helicopter digital model coordinate system, resulting in P3' and P2'. P3' and P2' are then calculated respectively. * P2' and P2 * The error between them, and between P3' and P3 * The error between P2' and P2 * The sum of the errors, i.e., the total error E'; Step 4.3.7: Use the gradient descent optimization method and the total error E' to optimize the centroid translation vector. The rotation transformation relation matrix T is iteratively optimized to continuously reduce the total error E' until the total error E' converges to a sufficiently small value, thus obtaining the optimal centroid translation transformation relation. With the rotation transformation relationship matrix T', the feature points of the rear section and the transition feature points of the helicopter reach the best fitting state with the corresponding feature points in the original helicopter digital model.

[0059] The helicopter's rear-stage scanning employs adaptive trajectory planning, dynamically adjusting the scanning speed based on the curvature of the vertical tail surface. Speed ​​is reduced in steep areas to ensure point cloud density, while speed is increased in flat areas to optimize efficiency. Multi-source data from the air intake tip and vertical tail feature points are integrated to construct a weighted registration model, using regularization terms to constrain the consistency between rotational transformations and the direction of gravity. To address thermal deformation caused by temperature fluctuations in the painting workshop environment, an infrared thermal imager is deployed to monitor the fuselage temperature field in real time, dynamically correcting the point cloud coordinates based on the material's thermal expansion coefficient. A multi-physics coupling compensation mechanism effectively suppresses cumulative errors caused by environmental factors, ensuring the spatial alignment accuracy between the rear-stage feature point cloud and the standard digital model.

[0060] Step 5: Generate the spraying trajectory and perform dynamic correction; specifically including: (1) In the theoretical coordinate system, the i-th spraying pose of the helicopter's front spraying trajectory is set as follows: ,in Euler angles represent rotations about the Z-axis, Y-axis, and X-axis of the theoretical coordinate system in sequence; (2) The i-th spraying position of the front spraying trajectory of the helicopter is transformed by the rotation transformation matrix N' and the centroid translation transformation matrix N'. After transformation, the adjusted spraying position coordinates are obtained. ;

[0061] (3) Adjust the i-th spraying attitude of the helicopter's front spraying trajectory: (3.1) Calculate the attitude matrix , ,in, Represents the rotation matrix about the Z, Y, and X axes; (3.2) The attitude matrix Converted to Euler angles in the theoretical coordinate system ZYX axis order Then the transformed spray pose of the i-th spray pose of the helicopter's front spray trajectory is: ; Follow steps (1) to (3) to complete the adjustment of all spraying positions of the front spraying trajectory of the helicopter; (4) In the theoretical coordinate system, the j-th spraying pose of the helicopter's rear spraying trajectory is set as follows: ,in Euler angles represent rotations about the Z-axis, Y-axis, and X-axis of the theoretical coordinate system in sequence; (5) The j-th spraying position of the rear spraying trajectory of the helicopter is transformed by the rotation transformation matrix T' and the center of gravity translation transformation matrix. After transformation, the adjusted spraying position coordinates are obtained. ;

[0062] (6) Adjust the j-th spraying posture of the front and rear spraying trajectories of the helicopter: (6.1) Calculate the attitude matrix , ,in, Represents the rotation matrix about the Z, Y, and X axes; (6.2) The attitude matrix Converted to Euler angles in the theoretical coordinate system ZYX axis order Then the spray pose after the transformation of the j-th spray pose of the helicopter's rear spray trajectory is: ; Follow steps (4) to (6) to complete the adjustment of all spraying positions of the rear spraying trajectory of the helicopter; Based on the piecewise coordinate transformation relationship, the theoretical spraying trajectory of the front and rear sections of the helicopter is mapped to the actual fuselage coordinate system, forming the actual spraying trajectory of the front and rear sections of the helicopter fuselage. The actual spraying trajectories of the front and rear sections are smoothly connected through a sinusoidal transition function, forming a gradual fusion zone near the interface to avoid abrupt changes in coating thickness.

[0063] Theoretical Trajectory Piecewise transformation mapped to the actual coordinate system:

[0064] in d This represents the length of the transition region. The online correction system recursively updates the transform parameters using a Kalman filter.

[0065] Spray gun process parameters mapped to Q = kv −0.7 (Inverse power law relationship between flow rate and velocity) and h =150mm+0.3∥∇ z ∥ (Height-Curvature Coupling Model) ensures coating uniformity on complex curved surfaces.

[0066] During trajectory remapping, the spray gun's moving speed and output are dynamically adjusted based on the surface curvature. In complex areas, the speed is reduced and the paint flow rate is increased to ensure uniformity. The online monitoring system captures feature point offsets in real time using a laser tracker and dynamically updates transformation parameters using a recursive estimation algorithm to achieve closed-loop correction during the spraying process. To address the explosion-proof requirements of the factory, a dual-redundancy safety mechanism is designed: automatic rescanning is triggered when the point cloud density or curvature continuity falls below a threshold; and the pneumatic spraying mode is immediately switched when combustible gas concentration exceeds the standard. The final generated correction trajectory achieves sub-millimeter-level continuous control in the transition zone, and the overall coating thickness uniformity meets aerospace process standards.

[0067] This invention has many specific applications. The above description is only a preferred embodiment of this patent and does not limit the implementation method and protection scope of this patent. For those skilled in the art, any solution obtained by making equivalent substitutions and obvious changes under the premise of the principle of this patent should be included in the protection scope of the patent.

Claims

1. A method for automatic pre-painting pose determination of a helicopter, characterized in that, The method comprises the following steps: Step 1: according to the overall structure characteristics of the helicopter, the helicopter is divided into front and rear fuselages according to space, and feature points are selected on the front and rear fuselages; Step 2: a unified coordinate system of the factory building is established, and the pose of the robot in the coordinate system of the factory building is determined; Step 3: the feature points of the front fuselage are scanned, and the feature points at the corresponding positions in the original helicopter numerical model are fitted; Step 4: the feature points of the rear fuselage are scanned, corrected and compensated, and the feature points at the corresponding positions in the original helicopter numerical model are fitted; Step 5: a spraying track is generated, and dynamic correction is performed.

2. The helicopter pose determination method according to claim 1, characterized in that, In step 1, the helicopter is divided into front and rear sections at the junction between the cockpit fuselage and the tail beam; a transition feature point is selected on the front and rear sections, and another feature point is selected near the junction between the front and rear sections.

3. The helicopter pose determination method according to claim 2, characterized in that, The feature point of the front section of the helicopter is the contour line of the radar dome; the feature point of the rear section is the contour line of the vertical tail; and the transition feature point is the contour line of the air inlet end.

4. The helicopter pose determination method according to claim 3, characterized in that, Step 2 specifically comprises the following steps: Step 2.1: a coordinate system of the factory building for actually spraying the helicopter is established; Step 2.1.1: a plurality of reference points are selected in the factory building, and the reference points are uniformly distributed and cover the main area of the factory building; Step 2.1.2: the three-dimensional coordinates of each reference point in the coordinate system of the laser tracker are measured in sequence by the laser tracker; Step 2.1.3: Define the unit vector of the factory coordinate system x-axis, denoted as , the unit vector of the factory coordinate system y-axis, denoted as , the unit vector of the factory coordinate system z-axis, denoted as , and the origin coordinates of the factory coordinate system, denoted as ; Step 2.1.4: The reference point three-dimensional coordinates in the laser coordinate system are converted to the defined factory coordinate system through the transformation relationship matrix H, and the factory coordinate system is established, wherein, Step 2.2: the pose of the robot in the coordinate system of the factory building is calibrated: Step 2.2.1: a tool coordinate system is established on the flange plate at the end of the spraying robot; Step 2.2.2: Adjust the spatial pose of the flange plate multiple times, and record the coordinate value of the origin of the tool coordinate system in the robot coordinate system after each spatial pose adjustment to form a point set containing N sampling points ; at the same time, the laser tracker captures the spatial coordinate values of the N sampling points in the factory coordinate system in real time to construct a point set ; Step 2.2.3: based on the point sets P and Q, an optimization model of the rotation matrix R and the translation vector t is constructed, the conversion relationship between the robot coordinate system and the factory building coordinate system is solved by iterative calculation, and the pose of the robot in the coordinate system of the factory building can be obtained through the conversion relationship.

5. The helicopter pose determination method according to claim 4, characterized in that, Its features are that, In step 2.2.3, the conversion relationship between the robot coordinate system and the factory building coordinate system includes: Step 2.2.3.1: Calculate the geometric centroid of the point set P, respectively coordinates, the geometric centroid of the point set Q coordinates; Step 2.2.3.2: Compute the geometric centroid of the point set P , the geometric centroid of the point set Q Align, based on the centroid aligned point sets P, Q, construct the covariance matrix H across coordinate systems: In the formula, T is the transpose operation; Step 2.2.3.3: Covariance matrix Perform singular value decomposition to obtain its orthonormal basis and eigenstructure: wherein and is an orthogonal matrix, is a diagonal matrix; Step 2.2.3.4: Based on the covariance matrix The singular value decomposition results in an optimal rotation matrix R by combining the orthogonal matrices ； Step 2.2.3.5: the translation vector t is calculated through the centroid coordinates of the point sets P and Q and the rotation matrix: Step 2.2.3.6: The rigid transformation relationship between the robot base coordinate system and the plant coordinate system is represented as : 。 6. The helicopter pose determination method according to claim 5, characterized in that, Step 3 specifically comprises: Step 3.1: A three-dimensional camera is installed on the spraying robot, and the front segment feature points of the helicopter actual parking pose and the transition feature points are scanned in the camera coordinates to obtain the relative position relationship between the robot flange and the robot base According to the conversion relationship between the robot base and the factory coordinate system And the relative position relationship between the camera and the flange , the relative position relationship between the camera and the factory is solved : Step 3.2: the feature point cloud data of the front section feature point and the transition feature point of the helicopter scanned in the camera coordinate system are converted to the coordinate system of the factory building; Step 3.3: through the barycentric translation vector and the rotation matrix, the feature point cloud data of the front section feature point and the transition feature point of the helicopter converted to the coordinate system of the factory building are fitted with the feature points at the corresponding positions in the original helicopter numerical model, and the coordinate transformation relationship that can fit the front section feature point and the transition feature point of the helicopter is obtained through error iterative optimization.

7. The helicopter pose determination method according to claim 6, characterized in that, Step 3.3 specifically comprises: Step 3.3.1: the barycenter P1 of the overall contour of the front section feature point of the helicopter and the barycenter P2 of the overall contour of the transition feature point are calculated; Step 3.3.2: the overall barycenter P0 of the front section of the fuselage is calculated through P1 and P2; Step 3.3.3: Calculate the original fuselage forebody overall center of gravity of the original helicopter number module, denoted as P0 * ; Step 3.3.4: The translation vector of the center of gravity of the front section of the fuselage is calculated by P0 and P0 * The translation vector of the center of gravity of the front section of the fuselage , that is, the translation transformation relationship of the front section of the fuselage is obtained. Step 3.3.5: The centroid vector is formed by the centroid P1 of the overall contour of the helicopter's forward feature points and the centroid P2 of the overall contour of the transition feature points. And the centroid vector of the feature point at the corresponding position in the original helicopter digital model. Calculate the centroid vector centroid vector The included angle θ and the cross product Determine the rotation axis and rotation angle of the centroid vector offset, and then obtain the rotation transformation relation matrix N; Step 3.3.6: Transfer the centroid P1 of the overall contour of the helicopter's forward feature points and the centroid P2 of the overall contour of the transition feature points using the centroid translation vector. The rotation transformation relation matrix N is transformed to the original helicopter digital model coordinate system to obtain P1' and P2'. P1' and P2' are then calculated respectively. * P2' and P2 * The error between them, and between P1' and P1 * The error between P2' and P2 * The sum of errors, i.e., the total error E; Step 3.3.7: Using the gradient descent optimization method, the total error E is used to iteratively optimize the translation vector of the barycenter and the rotation transformation relationship matrix N, constantly reducing the total error E, until the total error E converges to a small enough value, obtaining the optimal translation transformation relationship of the barycenter and the rotation transformation relationship matrix N', at this time, the feature points of the front section of the helicopter and the transition feature points respectively achieve the best fitting state with the corresponding feature points in the original helicopter model.

8. The helicopter pose determination method of claim 5, wherein, Step 4 specifically comprises: Step 4.1: A three-dimensional camera is installed on the spraying robot, and the helicopter rear segment feature points and transition feature points of the actual parking pose of the helicopter are scanned in the camera coordinates, and the scanning data is corrected and compensated; Step 4.2: The helicopter rear segment feature point and transition feature point cloud data scanned in the camera coordinate system are converted into the factory coordinate system; Step 4.3: Through the center of gravity translation vector and rotation matrix, the helicopter rear segment feature point and transition feature point cloud data converted into the factory coordinate system are fitted with the corresponding feature points in the original helicopter model, and the coordinate transformation relationship that can fit the helicopter rear segment feature points and transition feature points is obtained through error iteration optimization.

9. The helicopter pose determination method of claim 8, wherein, Step 4.3 specifically includes: Step 4.3.1: Calculate the center of gravity P3 of the overall contour of the helicopter front and rear segment feature points and the center of gravity P2 of the overall contour of the transition feature points; Step 4.3.2: Calculate the overall center of gravity P of the fuselage rear section by P3 and P2 01 ; Step 4.3.3: Calculate the original overall center of gravity of the original helicopter fuselage rear section, denoted as P 01 * ; Step 4.3.4: The translation vector of the rear segment of the fuselage is calculated by P 01 and P 01 * Translation vector of the rear segment of the fuselage , the translation transformation relationship of the rear segment of the fuselage is obtained. Step 4.3.5: The centroid vector is formed by the centroid P3 of the overall contour of the helicopter's rear feature points and the centroid P2 of the overall contour of the transition feature points. And the centroid vector of the feature point at the corresponding position in the original helicopter digital model. Calculate the centroid vector centroid vector The included angle θ' and cross product Determine the rotation axis and rotation angle of the centroid vector offset, and then obtain the rotation transformation relation matrix T; Step 4.3.6: The center of gravity P3 of the overall contour of the front and rear section feature points of the helicopter and the center of gravity P2 of the overall contour of the transition feature points are translated by the center of gravity translation vector , and the rotation transformation relationship matrix T is converted to the original helicopter digital model coordinate system to obtain P3' and P2'. The error between P3' and P3 * , the error between P2' and P2 * , and the sum of the error between P3' and P3 * and the error between P2' and P2 * , that is, the total error E' are calculated. Step 4.3.7: Using the gradient descent optimization method, the total error E' is used to iteratively optimize the translation vector of the barycenter and the rotation transformation relationship matrix T, constantly reducing the total error E', until the total error E' converges to a small enough value, obtaining the optimal barycenter translation transformation relationship and the rotation transformation relationship matrix T', at this time, the helicopter rear feature points and the transition feature points respectively reach the best fitting state with the corresponding feature points in the original helicopter model.

10. The helicopter pose determination method of claim 9, wherein, Step 5 specifically includes: (1) In the theoretical coordinate system, the i-th spraying pose of the helicopter front spraying trajectory is set as wherein represents the Euler angle of rotation around the Z-axis, Y-axis and X-axis of the theoretical coordinate system in sequence. (2) the i-th spraying position of the spraying track of the front section of the helicopter is transformed by the rotation transformation relationship matrix N' and the barycentric translation transformation relationship After transformation, the adjusted spraying position coordinates are obtained ; (3) Adjust the i-th spraying pose of the helicopter front segment spraying trajectory: (3.1) Compute pose matrix , , where, denotes the rotation matrix around the Z, Y, X axes; (3.2) The pose matrix is converted into the Euler angle form of the theoretical coordinate system ZYX axis sequence The spraying pose of the i-th spraying position of the helicopter front spraying track after conversion is ; According to steps (1)-(3), adjust all spraying poses of the helicopter front segment spraying trajectory; (4) In the theoretical coordinate system, the jth spraying pose of the helicopter rear section spraying trajectory is set as wherein represents the Euler angle of rotation around the Z axis, Y axis and X axis of the theoretical coordinate system in sequence. (5) the jth spraying position of the spraying track of the rear section of the helicopter is transformed by the rotation transformation relationship matrix T' and the barycentric translation transformation relationship After transformation, the adjusted spraying position coordinates are obtained ; (6) Adjust the j-th spraying pose of the helicopter front and rear segment spraying trajectory: (6.1) Compute pose matrix , , denotes the rotation matrix around the Z, Y, X axes; (6.2) The pose matrix is converted into the Euler angle form of the theoretical coordinate system ZYX axis sequence The jth spraying pose of the helicopter rear section spraying trajectory after conversion is ; According to steps (4)-(6), adjust all spraying poses of the helicopter rear segment spraying trajectory; Based on the segmented coordinate transformation relationship, the helicopter front and rear segment theoretical spraying trajectories are mapped into the actual fuselage coordinate system to form the helicopter fuselage front and rear segment actual spraying trajectories, and the front and rear segment actual spraying trajectories are smoothly connected through a sinusoidal transition function to form a gradual fusion area near the interface; Then, the laser tracker is used to capture the displacement of the selected feature points on the front and rear fuselages in real time, and the recursive estimation algorithm is used to dynamically update the transformation parameters to realize the closed-loop correction of the spraying trajectory during spraying.

Images