Steel structure virtual pre-splicing method based on laser radar and inertial measurement unit

Abstract

The invention discloses a steel structure virtual pre-splicing method based on a laser radar and an inertial measurement unit. The method comprises the following steps: scanning a steel structure by using a laser radar to obtain point cloud data; using an IMU to record IMU data of the steel structure in the space; carrying out filtering processing on the collected point cloud data; calibrating the IMU data by using IMU data processing software; registering the point cloud data collected at different time points; the IMU data and the point cloud data are fused, and the IMU data is utilized to assist accurate positioning and model construction of the point cloud data; automatically identifying key structure elements of the steel structure by using geometric features in the point cloud data, and constructing a three-dimensional model of the steel structure by using the key structure elements; in the assembling process, constructors can accurately adjust the positions and directions of the steel structure components according to information fed back in real time in the virtual pre-assembling process of the steel structure, and the assembling precision and accuracy are ensured.

Description

Technical Field

[0001] The invention relates to the technical field of three-dimensional modeling and virtual simulation, and in particular to a virtual pre-assembly method of a steel structure based on a laser radar and an inertial measurement unit. Background Art

[0002] In the field of construction and engineering, the precise assembly of large steel structures is the key to ensuring the quality and safety of the project. Traditional steel structure assembly methods rely on manual measurement and empirical judgment, which has problems of low efficiency and large errors. The reasons why traditional methods are difficult to accurately and quickly measure steel structures mainly include the following aspects: Complexity of steel structures: Large steel structures usually have complex shapes and structures, such as various special-shaped components, multi-curved structures, etc. It is difficult to fully and accurately obtain the three-dimensional information of these complex shapes using traditional measurement methods such as tape measures, levels, and theodolites. Some key parts may be missed, or the data may be inaccurate due to human operation errors during the measurement process.

[0003] On-site environmental interference: At construction sites, there are various interference factors, such as insufficient light, dust, obstacles, etc. These factors will affect the use effect and measurement accuracy of traditional measurement equipment. For example, insufficient light may make visual measurement difficult, and dust may adhere to the lens of the measuring instrument and affect observation.

[0004] Inefficiency: Traditional manual measurement methods require construction workers to measure and record each part, which is time-consuming and difficult to achieve real-time monitoring and feedback. For large-scale steel structure assembly projects, this inefficient measurement method will slow down the construction progress and increase construction costs.

[0005] Data accuracy is difficult to guarantee: Traditional measurement methods rely on manual reading and recording, which is prone to reading errors, recording errors, and other problems, and some minor deformations or displacements may be difficult to detect. Summary of the invention

[0006] The present invention provides a virtual pre-assembly method of a steel structure based on a laser radar and an inertial measurement unit, so as to solve the problem that it is difficult to accurately and quickly measure the steel structure in the prior art.

[0007] The present invention provides a virtual pre-assembly method for steel structure based on laser radar and inertial measurement unit, comprising: Use laser radar to scan the steel structure to obtain point cloud data; at the same time, use IMU to record the IMU data of the steel structure in space, and the IMU data includes the motion trajectory and direction change of the steel structure; Filter the collected point cloud data to remove noise and abnormal points; use IMU data processing software to calibrate the IMU data to ensure that the IMU data is synchronized with the point cloud data; A point cloud registration algorithm is used to register the point cloud data collected at different time points to ensure that the point cloud data is in a unified coordinate system; wherein, the registration includes coarse registration and fine registration; coarse registration includes: denoising and filtering the point cloud data to eliminate errors and redundant information in the data; extracting feature points or descriptors from the point cloud data to match corresponding points in the source point cloud and the target point cloud; using a feature matching algorithm to find corresponding point pairs in the source point cloud and the target point cloud for preliminary registration; preliminary registration uses a random sampling consistency algorithm to remove false matches and estimate the initial transformation matrix; fine registration includes: adjusting the initial transformation matrix through an iterative optimization algorithm to minimize the alignment error between the source point cloud and the target point cloud; in the fine registration process, defining an objective function to measure the alignment error between the point clouds; judging whether the iterative optimization converges by judging whether the alignment error meets a preset threshold or reaches the maximum number of iterations; if the iterative optimization converges, the point cloud data registration is completed; if the iterative optimization does not converge, the iterative optimization continues; Fuse IMU data with point cloud data, and use IMU data to assist in precise positioning and model building of point cloud data; The key structural elements of the steel structure are automatically identified by using the geometric features in the point cloud data, and a three-dimensional model of the steel structure is constructed using the key structural elements.

[0008] Furthermore, the IMU data is fused with the point cloud data, including: Clarify the coordinate systems of the IMU and LiDAR, as well as their coordinate system conversion relationships with the global coordinate system or the robot chassis coordinate system; based on the determined coordinate system conversion relationship, convert the IMU data and point cloud data to the same coordinate system; Read the angular velocity and linear acceleration data of the IMU, integrate the angular velocity data of the IMU to get the change of the rotation angle, integrate the linear acceleration data to get the change of the velocity and displacement, and apply these changes to the position and attitude of the point cloud data; In data acquisition and processing, a real-time data processing loop is established; in each cycle, the latest IMU data and point cloud data are obtained, and after time synchronization and coordinate conversion, the position and posture of the point cloud data are updated using the selected data fusion algorithm.

[0009] Furthermore, the IMU data is fused with the point cloud data, and also includes: Build error models for IMU and LiDAR to identify and compensate for sensor errors; apply error compensation algorithms to reduce model bias due to sensor errors.

[0010] Furthermore, the IMU data is fused with the point cloud data, and also includes: Verify the fusion results of IMU data and point cloud data; Adjust fusion parameters based on verification results; Output the fused point cloud data and IMU data for subsequent 3D modeling and virtual pre-assembly; Record all parameters and algorithms used in the fusion process.

[0011] Furthermore, the fusion results of IMU data and point cloud data are verified, including: Verify the accuracy of the fusion results by comparing with known control points or reference models; For matched control points, compare their coordinate differences in the scanned data and the design model; calculate the deviation values ​​of each control point in the x, y, and z directions, so as to obtain the position deviation between the actual construction and the design of the steel structure at different positions; compare the geometric dimensions of the main components of the steel structure; measure the dimensions of these components in the scanned data and the design model, and calculate the difference; Use a surface or entity-based comparison method to compare the scanned steel structure surface or entity with the design model to obtain a color cloud map of shape deviation; analyze the deviation distribution in the shape deviation cloud map to determine the area where shape deviation exists; Based on the deviation of the control points and the shape deviation, the installation quality of the entire steel structure and its compliance with the design are evaluated.

[0012] The present invention has the following beneficial effects: a virtual pre-assembly method of steel structure based on laser radar and inertial measurement unit of the present invention uses laser radar to scan the steel structure and obtain point cloud data; at the same time, uses IMU to record IMU data of the steel structure in space, and the IMU data includes the motion trajectory and direction change of the steel structure; the collected point cloud data is filtered to remove noise and abnormal points; the IMU data is calibrated using IMU data processing software to ensure that the IMU data is synchronized with the point cloud data; the point cloud registration algorithm is used to register the point cloud data collected at different time points to ensure that the point cloud data is in a unified coordinate system; the IMU data is merged with the point cloud data, and the IMU data is used to assist the precise positioning and model construction of the point cloud data; the geometric features in the point cloud data are used to automatically identify the key structural elements of the steel structure, and the key structural elements are used to build a three-dimensional model of the steel structure. During the assembly process, the construction personnel can accurately adjust the position and direction of the steel structure components according to the real-time feedback information in the virtual pre-assembly of the steel structure to ensure the precision and accuracy of the assembly. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solution of the present invention, the drawings required for use in the embodiments are briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work.

[0014] Figure 1 The present invention provides a flow chart of a virtual pre-assembly method for steel structure based on laser radar and inertial measurement unit. DETAILED DESCRIPTION

[0015] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution of the present invention will be clearly and completely described in conjunction with the specific embodiments of the present invention and the corresponding drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. The technical solutions provided by the embodiments of the present invention are described in detail below in conjunction with the drawings.

[0016] See also Figure 1 , a virtual pre-assembly method of steel structure based on laser radar and inertial measurement unit provided by the present invention, comprising: Step 1: Use laser radar to scan the steel structure to obtain point cloud data; at the same time, use IMU to record the IMU data of the steel structure in space, and the IMU data includes the motion trajectory and direction change of the steel structure.

[0017] Specifically, laser radar is used to scan various components of the steel structure to obtain its precise three-dimensional shape and position information. Laser radar can quickly and comprehensively scan steel structures and collect a large amount of point cloud data per second, which contains detailed geometric features of the steel structure surface. For example, for components such as steel beams and steel columns of bridges, laser radar can accurately capture information such as the contour and curvature of their surfaces.

[0018] The Inertial Measuring Unit (IMU) is installed on construction equipment or steel structure components to measure the acceleration, angular velocity, direction and other information of the equipment or components in real time. When assembling the steel structure of a bridge, when the steel beams and other components are being hoisted and moved, the IMU can continuously monitor the changes in their motion state, such as translation, rotation and other information.

[0019] The motion types included in the motion trajectory and the data collected for each motion type are as follows: Translational motion: Record the distance the IMU structure moves along the x, y, and z axes in three-dimensional space. For example, measure how far a steel structure moves forward, left or right, up or down during movement. In addition to displacement, the velocity and acceleration information of the IMU structure also need to be recorded. Velocity represents the rate of change of displacement per unit time, and acceleration represents the rate of change of velocity.

[0020] Rotational motion: IMU can measure the change in the rotation angle of the steel structure itself around the x, y, and z axes. Angular velocity is a physical quantity that describes the speed of rotation, that is, the change in the rotation angle per unit time. If an IMU structure rotates around a certain axis at a constant angular velocity, the rotation angle can be obtained by integrating the angular velocity.

[0021] Orientation changes include Euler angles. Euler angles, such as pitch, yaw, and roll, are used to describe the orientation of the IMU structure in space.

[0022] Step 2: Filter the collected point cloud data to remove noise and outliers; use IMU data processing software to calibrate the IMU data to ensure that the IMU data is synchronized with the point cloud data.

[0023] Specifically, ensure that the IMU data and the lidar data are synchronized in terms of timestamps for subsequent fusion.

[0024] Step three: Use the point cloud registration algorithm to register the point cloud data collected at different time points to ensure that the point cloud data is in a unified coordinate system.

[0025] Specifically, the registration includes coarse registration: denoising and filtering the point cloud data to eliminate errors and redundant information in the data; extracting feature points or descriptors from the point cloud data to match corresponding points in the source point cloud and the target point cloud; common features include Scale-Invariant Feature Transform (SIFT), Speeded Up Robust Features (SURF), FAST key point detection and BRIEF feature description (Oriented FAST and Rotated BRIEF, ORB), etc. Use the feature matching algorithm to find the corresponding point pairs in the source point cloud and the target point cloud for preliminary registration; the preliminary registration uses a random sampling consistency algorithm to remove false matches and estimate the initial transformation matrix.

[0026] The registration also includes fine registration: adjusting the initial transformation matrix through an iterative optimization algorithm to minimize the alignment error between the source point cloud and the target point cloud; optional methods include iterative closest point algorithm and its variants. In the fine registration process, an objective function is defined to measure the alignment error between point clouds; optional objective functions include minimizing the Euclidean distance from point to point or the distance from point to surface. Whether the iterative optimization converges is determined by judging whether the alignment error meets a preset threshold or reaches the maximum number of iterations; if the iterative optimization converges, the point cloud data registration is completed; if the iterative optimization does not converge, the iterative optimization continues.

[0027] Step 4: Fuse the IMU data with the point cloud data, and use the IMU data to assist in the precise positioning and model building of the point cloud data.

[0028] Specifically, the IMU data is fused with the point cloud data, including: Clarify the coordinate systems of the IMU and LiDAR, as well as their coordinate system conversion relationships with the global coordinate system or the robot chassis coordinate system; for example, the IMU has its own coordinate system, and the point cloud data is collected based on the LiDAR coordinate system. Determine the conversion relationship between these coordinate systems, such as translation vectors and rotation matrices, through information such as measurement or sensor installation parameters. According to the determined coordinate system conversion relationship, convert the IMU data and point cloud data to the same coordinate system; in this way, they can be fused and processed in a unified coordinate space. For example, if the IMU data is the angular velocity and acceleration in its own coordinate system, it needs to be converted to the coordinate system where the point cloud data is located through the rotation matrix and translation vector. Read the angular velocity and linear acceleration data of the IMU, integrate the angular velocity data of the IMU to obtain the change in rotation angle, integrate the linear acceleration data to obtain the change in velocity and displacement, and apply these changes to the position and attitude of the point cloud data; establish a real-time data processing loop in data acquisition and processing; in each cycle, obtain the latest IMU data and point cloud data, perform time synchronization and coordinate conversion, and use the selected data fusion algorithm to update the position and attitude of the point cloud data. According to the needs of the actual application and the characteristics of the data, dynamically adjust the parameters or weights of the data fusion algorithm as needed. For example, if the noise of the IMU data is large, its weight in the fusion algorithm can be appropriately reduced; if the update frequency of the point cloud data is low, the prediction weight of the IMU data can be increased to maintain the continuity of the pose estimation.

[0029] Establish error models for IMU and LiDAR to identify and compensate for sensor errors; apply error compensation algorithms to reduce model deviations caused by sensor errors. Verify the fusion results of IMU data and point cloud data; adjust fusion parameters based on the verification results; output the fused point cloud data and IMU data for subsequent 3D modeling and virtual pre-assembly; record all parameters and algorithms used in the fusion process.

[0030] Specifically, the fusion results of IMU data and point cloud data are verified, including: Verify the accuracy of the fusion results by comparing with known control points or reference models; Among them, control point matching can be done by manual matching and automatic matching.

[0031] For known key control points, manually identify the corresponding points in the scanned data and the design model. For example, find the actual scanned point corresponding to the center point of the column foot in the design model in the scanned steel structure image. Record the coordinate information of these matching points as the basis for subsequent precise matching and comparison. Use professional 3D model comparison software, such as GeomagicControl, to perform automatic control point matching algorithms based on the manually matched initial points. These algorithms will find more matching control points between the scanned data and the design model based on the geometric characteristics of the points, distance relationships, etc.

[0032] For matched control points, compare their coordinate differences in the scanned data and the design model; calculate the deviation values ​​of each control point in the x, y, and z directions to obtain the position deviation between the actual construction and the design of the steel structure at different locations; compare the geometric dimensions of the main components of the steel structure; measure the dimensions of these components in the scanned data and the design model and calculate the difference; use a surface or solid-based comparison method to compare the scanned steel structure surface or solid with the design model to obtain a color cloud map of shape deviation; analyze the deviation distribution in the shape deviation cloud map to determine the area where the shape deviation exists; evaluate the installation quality of the entire steel structure and its compliance with the design based on the deviation and shape deviation of the control points. The fusion of IMU data and lidar data can provide a more accurate and stable three-dimensional model of the steel structure, thereby predicting and solving potential assembly problems during virtual pre-assembly.

[0033] The present invention fuses the point cloud data collected by the lidar and the motion data measured by the IMU, and comprehensively analyzes and processes this data. For example, the initial position and attitude of the steel structure components can be accurately determined using the point cloud data, while combining the motion data of the IMU can track the dynamic changes of the components in real time during the assembly process, and promptly detect and correct possible deviations. In this way, during the assembly process, the construction personnel can accurately adjust the position and direction of the steel structure components according to the real-time feedback information to ensure the accuracy and precision of the assembly. For example, when the steel beam is hoisted near the designated position, through data fusion analysis, the relative position relationship between the steel beam and the installed part can be accurately judged, and then fine-tuning can be carried out to control the installation accuracy of the connection part within the range required by the design.

[0034] Step Five: Automatically identify the key structural elements of the steel structure using the geometric features in the point cloud data, and construct a three-dimensional model of the steel structure using the key structural elements.

[0035] For the surfaces of columns and beams, since they usually have large planar parts, the Random Sample Consensus (RANSAC) algorithm can be used for plane fitting. Randomly extract a small number of points from the point cloud, such as 3 points; assume a plane equation, and then calculate the number of points in the point cloud within a certain distance range of the plane, and repeat this process multiple times to find the plane that contains the most points. The direction and position of the structural components can be preliminarily judged based on the normal vector and position information of the fitted plane. For example, the normal vector of the column plane is usually perpendicular to the ground, and there is a specific angular relationship between the normal vector of the beam plane and the normal vector of the column plane.

[0036] Calculate the local curvature of the point cloud. Determine the curvature by calculating the geometric shape change of the neighborhood of each point, such as the k-nearest neighbors. For the node area, due to the complexity of its structural shape, the curvature is usually large; while the curvature of the main parts of beams and columns is relatively small. Different curvature thresholds can be set to distinguish different structural components. For example, the points with a curvature greater than C1 may be the points in the node area, and the points with a curvature less than C2 (C2 < C1) may belong to the surfaces of beams or columns.

[0037] For the identification of columns, look for structural components with a relatively large height and a basically consistent cross-sectional shape in the vertical direction. The fitted plane can be sliced in the vertical direction, and the shape and area changes of each slice can be analyzed. If the shape and area changes within a certain height range, such as the H shape and area changes are within the allowable range, such as setting the relative change rate as p, then it can be determined as a column. The identification of beams can be based on their slender shape characteristics. By calculating the length and width in the extending direction of the point cloud, if the ratio of the length to the width is greater than the set threshold L / W, such as L / W > 5, then it may be a beam.

[0038] Nodes are the connection points between beams and columns. By analyzing the connection relationship of the point cloud, we can find the areas that are connected to multiple structural components with different directions at the same time. For example, we first identify beams and columns, and then for each point, we count the number of points that are adjacent to it and belong to different structural components. If the number is greater than the set connection number threshold, such as n, the area where the point is located may be a node. And use these features to build a 3D model of the steel structure.

[0039] Simulate the assembly process of steel structures in a computer to perform virtual pre-assembly, check the fit of various components, and predict possible problems and conflicts. Analyze the results of virtual pre-assembly, identify design or manufacturing defects, and optimize the structural design based on the analysis results. Test the optimized model in a real environment to verify the accuracy of virtual pre-assembly. Prepare detailed reports and documents to record all steps and problems found during the pre-assembly process.

[0040] In summary, the present invention realizes real-time recording of the motion state of the structure, improves the real-time nature of the data, reduces human errors through automated data processing and model building, and can reduce rework and costs in actual construction by discovering problems in advance, thereby improving engineering efficiency.

[0041] The above-described embodiments of the present invention do not limit the protection scope of the present invention.

Claims

1. A virtual pre-assembly method for steel structure based on laser radar and inertial measurement unit, characterized in that: include: Use laser radar to scan the steel structure to obtain point cloud data; at the same time, use IMU to record the IMU data of the steel structure in space, and the IMU data includes the motion trajectory and direction change of the steel structure; Filter the collected point cloud data to remove noise and abnormal points; use IMU data processing software to calibrate the IMU data to ensure that the IMU data is synchronized with the point cloud data; A point cloud registration algorithm is used to register the point cloud data collected at different time points to ensure that the point cloud data is in a unified coordinate system; wherein, the registration includes coarse registration and fine registration; coarse registration includes: denoising and filtering the point cloud data to eliminate errors and redundant information in the data; extracting feature points or descriptors from the point cloud data to match corresponding points in the source point cloud and the target point cloud; using a feature matching algorithm to find corresponding point pairs in the source point cloud and the target point cloud for preliminary registration; preliminary registration uses a random sampling consistency algorithm to remove false matches and estimate the initial transformation matrix; fine registration includes: adjusting the initial transformation matrix through an iterative optimization algorithm to minimize the alignment error between the source point cloud and the target point cloud; in the fine registration process, defining an objective function to measure the alignment error between the point clouds; judging whether the iterative optimization converges by judging whether the alignment error meets a preset threshold or reaches the maximum number of iterations; if the iterative optimization converges, the point cloud data registration is completed; if the iterative optimization does not converge, the iterative optimization continues; Fuse IMU data with point cloud data, and use IMU data to assist in precise positioning and model building of point cloud data; The key structural elements of the steel structure are automatically identified by using the geometric features in the point cloud data, and a three-dimensional model of the steel structure is constructed using the key structural elements.

2. The method for virtual pre-assembly of steel structures based on laser radar and inertial measurement unit according to claim 1, characterized in that: Fusion of IMU data with point cloud data, including: Clarify the coordinate systems of the IMU and LiDAR, as well as their coordinate system conversion relationships with the global coordinate system or the robot chassis coordinate system; based on the determined coordinate system conversion relationship, convert the IMU data and point cloud data to the same coordinate system; Read the angular velocity and linear acceleration data of the IMU, integrate the angular velocity data of the IMU to get the change in rotation angle, integrate the linear acceleration data to get the change in velocity and displacement, and apply these changes to the position and attitude of the point cloud data; In data acquisition and processing, a real-time data processing loop is established; in each cycle, the latest IMU data and point cloud data are obtained, and after time synchronization and coordinate conversion, the position and posture of the point cloud data are updated using the selected data fusion algorithm.

3. The method for virtual pre-assembly of steel structures based on laser radar and inertial measurement unit according to claim 2, characterized in that: Fusion of IMU data with point cloud data also includes: Build error models for IMU and LiDAR to identify and compensate for sensor errors; apply error compensation algorithms to reduce model bias due to sensor errors.

4. The method for virtual pre-assembly of steel structures based on laser radar and inertial measurement unit according to claim 3, characterized in that: Fusion of IMU data with point cloud data also includes: Verify the fusion results of IMU data and point cloud data; Adjust fusion parameters based on verification results; Output the fused point cloud data and IMU data for subsequent 3D modeling and virtual pre-assembly; Record all parameters and algorithms used in the fusion process.

5. The steel structure virtual pre-assembly method based on laser radar and inertial measurement unit according to claim 4, characterized in that: Verify the fusion results of IMU data and point cloud data, including: Verify the accuracy of the fusion results by comparing with known control points or reference models; For matched control points, compare their coordinate differences in the scanned data and the design model; calculate the deviation values ​​of each control point in the x, y, and z directions, so as to obtain the position deviation between the actual construction and the design of the steel structure at different positions; compare the geometric dimensions of the main components of the steel structure; measure the dimensions of these components in the scanned data and the design model, and calculate the difference; Use a surface or entity-based comparison method to compare the scanned steel structure surface or entity with the design model to obtain a color cloud map of shape deviation; analyze the deviation distribution in the shape deviation cloud map to determine the area where shape deviation exists; Based on the deviation of the control points and the shape deviation, the installation quality of the entire steel structure and its compliance with the design are evaluated.

Images