Error correction method and system suitable for binocular vision system of bridge erecting machine

Abstract

The application belongs to the technical field of high-precision positioning of large objects, and particularly relates to an error correction method and system suitable for a binocular vision system of a bridge erecting machine, comprising: coplanarly installing a binocular vision system and at least three laser sensors at selected positions on the bridge erecting machine, and defining a sensor plane as an installation reference plane; setting a mark plane on a bridge that has been erected as a measurement reference; measuring depth information from each laser sensor to the mark plane by using the laser sensors; dynamically calculating a spatial angle deviation between the sensor plane and the mark plane in real time based on the depth information; generating a correction matrix for correcting coordinates according to the spatial angle deviation; and measuring a target object by using the binocular vision system to obtain original coordinates, and applying the correction matrix to calibrate the original coordinates. The application fundamentally eliminates systematic errors caused by non-parallel installation, and significantly improves the accuracy of final positioning and the stability of results.

Description

TECHNICAL FIELD

[0001] The present application belongs to the technical field of high-precision positioning of large objects, and specifically relates to an error correction method and system suitable for a binocular vision system of a bridge erecting machine. BACKGROUND

[0002] In the construction process of modern fabricated bridges, the bridge erecting machine is responsible for accurately hoisting the prefabricated box girder to the designated position. In order to achieve millimeter-level precise butt joint of the box girder and the reinforcement of the pier cap or the pier sleeve, binocular vision positioning technology is widely used. This technology installs a binocular vision camera on the bridge erecting machine, measures the relative three-dimensional coordinates of the bridge erecting machine and the target position (such as the box girder or pier column that has been erected) in real time, and thus guides the precise movement of the bridge erecting machine.

[0003] However, the measurement accuracy of the existing binocular vision positioning technology is highly dependent on a stringent prerequisite: the installation plane of the binocular vision camera must be strictly physically parallel to the bridge deck (or calibration plane) serving as the measurement reference. In complex field construction sites, due to installation condition limitations and operation errors, it is extremely difficult to achieve perfect parallel installation of the camera and the bridge deck. Any slight installation angle deviation, such as pitch angle or roll angle deviation, will be directly introduced into the three-dimensional coordinate calculation process, forming systematic measurement errors. This error will be amplified as the measurement distance increases, ultimately leading to substandard positioning accuracy, affecting construction quality, and even causing safety hazards.

[0004] Currently, there is a lack of an effective and convenient means to detect and compensate for this systematic error caused by non-parallel installation. Construction personnel often have to rely on experience to perform repeated debugging, which is time-consuming and labor-intensive and the effect is difficult to guarantee. Therefore, there is an urgent need for a new technical solution to break away from the dependence on high-precision physical installation and fundamentally reduce or eliminate the impact of installation errors on measurement results. SUMMARY

[0005] The present application aims to address the shortcomings of the prior art by providing an error correction method and system suitable for a binocular vision system of a bridge erecting machine to solve the problem of reduced positioning accuracy caused by non-parallel installation of the binocular camera.

[0006] The technical solution of the present application is an error correction method and system suitable for a binocular vision system of a bridge erecting machine, which comprises the following steps:

[0007] S1: installing a binocular vision system and at least three laser sensors on a selected position of the bridge erecting machine in a coplanar manner, defining the installation reference plane as the sensor plane, and ensuring that the measurement directions of the at least three laser sensors are parallel to each other and perpendicular to the sensor plane;

[0008] S2: setting a mark plane as a measurement reference on the erected bridge and within the measurement range of the binocular vision system and the at least three laser sensors;

[0009] S3: measuring depth information from each laser sensor to the mark plane by using the at least three laser sensors respectively;

[0010] S4: calculating a spatial angle deviation between the sensor plane and the mark plane based on the depth information;

[0011] S5: generating a correction matrix for correcting coordinates in real time according to the spatial angle deviation;

[0012] S6: measuring a target object by using the binocular vision system to obtain original coordinates, and applying the correction matrix to calibrate the original coordinates to obtain corrected coordinates.

[0013] Further to the above technical solution, the binocular vision system is a binocular camera, and the laser sensor is three.

[0014] In order to simplify the model and facilitate subsequent calculation, the three laser sensors are arranged in a right triangle at the periphery of the binocular camera. In order to ensure that the binocular camera and the laser sensor are in the same plane, a common mounting substrate of the camera and the laser sensor is processed by using a whole piece of 7075-T6 aluminum alloy, and the thickness is 15mm-20mm. The substrate plane is ground and processed, and the flatness reaches the ISO1101 standard P level (≤3μm / 100mm). The camera and the sensor are mounted by reserving fixing screw holes. The surface of the substrate is scanned by using a laser interferometer to generate a height cloud map for initial leveling calibration. An electronic level and a micrometer are used for fine calibration, and the sensor screw is adjusted until the flatness is ≤1μm / 100mm, so as to ensure that the binocular camera and the laser sensor form a plane parallel.

[0015] Further, the mark plane in step S2 is laid on the top edge of the box girder of the erected bridge, and the flatness of the mark plane is required to be less than 1mm, and the material adopts matte anodized aluminum.

[0016] Further, the process of calculating the spatial angle deviation in step S4 includes:

[0017] The plane where the binocular camera is located is defined as a reference plane: ;

[0018] The mark plane is modeled as a target plane: , wherein, is a slope, is an intercept;

[0019] Determine the position information of three laser sensors based on a reference plane and obtain the depth information measured by the three laser sensors respectively ;

[0020] Construct a matrix equation: ;

[0021] Solve the slope by least square method : ,

[0022] wherein, , , is the transpose matrix of ;

[0023] Calculate the spatial angle deviation based on the solved slope , : . .

[0024] Further, substitute the spatial angle deviation into the Rodrigues rotation formula to construct to obtain the correction matrix,

[0025] Define the correction matrix R: ,

[0026] wherein, is an identity matrix, is a skew-symmetric matrix, is an outer product matrix;

[0027] Based on the correction matrix R, obtain:

[0028] Correction matrix for rotation around x-axis ;

[0029] Correction matrix for rotation around y-axis ;

[0030] Correction matrix for rotation around z-axis .

[0031] Further, assume that the original coordinates obtained by using the binocular camera to measure a target object in three dimensions are wherein is a plane coordinate, is an original depth value;

[0032] Calculate the correction matrix and its transpose matrix ;

[0033] Through matrix operation transforming the original coordinates into corrected coordinates wherein is the corrected planar coordinate, is the corrected real depth value.

[0034] Further, the method further comprises a step S6 of sending the corrected coordinates to a programmable logic controller for controlling the action of the bridge erecting machine through socket communication of the transmission control protocol. The feature point coordinates on the to-be-erected box girder are corrected according to the coordinate correction value, and a control instruction is generated by comparing the feature point coordinates of the erected box girder, so as to complete the action instruction of the bridge erecting machine.

[0035] The application further provides a system for implementing the error correction method of the binocular vision system of the bridge erecting machine.

[0036] a mark plane laid on the top edge of the erected box girder of the bridge;

[0037] a mounting base plate;

[0038] a binocular vision system fixed on the mounting base plate and used for three-dimensional measurement of a target object to obtain original coordinates;

[0039] three laser sensors fixed on the mounting base plate and arranged in a right-angled triangle at the periphery of the binocular vision system, each laser sensor measuring direction being parallel to each other and used for measuring the depth information from itself to the mark plane;

[0040] a processor configured to perform the following operations: receiving the depth information measured by the three laser sensors; based on the depth information, calculating in real time the spatial angle deviation between the sensor plane defined by the mounting positions of the binocular vision system and the three laser sensors and the mark plane; generating in real time a correction matrix for correcting the coordinates according to the spatial angle deviation; and applying the correction matrix to calibrate the original coordinates obtained by the binocular vision system to obtain corrected coordinates, so as to correct the coordinates in real time and dynamically.

[0041] Advantages: compared with the prior art, the application has the advantages that: the application calibrates the mounting posture of the binocular vision system in real time through the independent laser sensors, and compensates the angle deviation in a mathematical way, so as to fundamentally eliminate the systematic error caused by the non-parallel installation, and significantly improve the accuracy of the final positioning and the stability of the results.

[0042] The application no longer requires the binocular vision system to be installed in a harsh physical parallel manner, and allows a certain installation error. This greatly simplifies the installation and debugging process on site, shortens the preparation time, reduces the requirement for the skills of the construction personnel, and improves the overall construction efficiency.

[0043] The correction method adopted by the present application makes the whole positioning system have stronger adaptability to slight changes of installation position or attitude changes caused by equipment vibration, the system robustness is significantly improved, and the system can better adapt to complex construction environment. The error detection and compensation process is completely completed automatically by the system without manual intervention, realizing the automatic calibration of the positioning system, and ensuring the objectivity and reliability of the measurement results. BRIEF DESCRIPTION OF DRAWINGS

[0044] Figure 1 is a flowchart of the error correction method for the binocular vision system of the bridge erecting machine provided by the present application.

[0045] Figure 2 is a top view of the positional relationship of the binocular camera, the laser sensor and the marker plane in the present application.

[0046] Figure 3 is a side view of the positional relationship of the binocular camera, the laser sensor and the marker plane in the present application.

[0047] In the drawings, the reference signs are: 1, mounting base plate; 2, binocular camera; 3, laser sensor; 4, marker plane. DETAILED DESCRIPTION

[0048] The technical solutions of the present application will be described in detail below with reference to the drawings, but the protection scope of the present application is not limited to the described embodiments.

[0049] Embodiment 1: Referring to Figure 1 The present application provides an error correction method for a binocular vision system of a bridge erecting machine, specifically comprising the following steps:

[0050] Step 1: System installation and reference setting

[0051] A marker plane is laid on the top edge of the erected bridge box girder, the marker plane range is large enough and within the effective measurement range of the binocular vision system and the laser sensor, the marker plane requires a flatness of less than 1mm, and the material adopts dull anodized aluminum to suppress mirror glare.

[0052] The device integrated with the binocular vision system and the three laser sensors is fixed at a proper position of the bridge erecting machine, and the binocular vision system adopts a binocular camera. In order to simplify the model and facilitate subsequent calculation, the three laser sensors are installed in a right triangle arrangement around the binocular camera, and the distance between each two is known, such as Figure 2 、 Figure 3As shown, the plane where the laser sensor is located is defined as the sensor plane, the laser sensor emits three laser beams to the marker plane, the three laser beams are configured to be parallel to each other, and each is perpendicular to the sensor plane. To ensure that the binocular camera and the laser sensor are in the same plane, a common mounting substrate for mounting the binocular camera and the three laser sensors is machined from a whole piece of 7075-T6 aluminum alloy, with a thickness of 15-20 mm, the mounting substrate is plane ground processing, the flatness reaches ISO 1101 standard P level (≤3 μm / 100 mm), and the mounting substrate is reserved with fixing screw holes for mounting the binocular camera and the laser sensor. The surface of the mounting substrate is scanned by a laser interferometer to generate a height cloud map for initial leveling calibration; fine calibration is performed by using an electronic level and a micrometer, and the laser sensor screw is adjusted until the flatness is ≤1 μm / 100 mm, to ensure that the plane formed by the binocular camera and the laser sensor is parallel.

[0053] Step 2: Depth information measurement

[0054] After starting the system, the laser sensor emits three laser beams to the marker plane, and the plane formed by the three laser spots is strictly parallel to the bridge deck where the marker plane is located. The vertical distance from each laser spot on the marker plane, i.e. the depth information, is sent to the processor (industrial computer) in real time through a serial communication protocol.

[0055] Step 3: Spatial angle deviation calculation

[0056] The industrial computer processes the depth information measured by the laser sensor to detect whether the sensor plane and the marker plane are parallel: if ( is the error threshold, and ), then they are parallel, and the coordinate measurement is directly performed. If they are not parallel, the included angle between them is calculated.

[0057] Define the plane where the binocular camera is located as the reference plane: ;

[0058] Model the marker plane as the target plane: , (( , is the slope, is the slope distance) ;

[0059] Measure the distance of each point to the marker plane using the three laser sensors .

[0060] Construct a matrix equation: ;

[0061] Solve by least squares method: , , .

[0062] Formula of calculating the angle: .

[0063] Step 4: Generation of Correction Matrix

[0064] Calculate the spatial angle deviation Then, a correction matrix is constructed by using the angle information and the Rodrigues' rotation formula.

[0065] Let the vector be rotated by the angle to get the new vector The Rodrigues' formula expresses the rotated vector by vector operation and matrix transformation as: .

[0066] This formula can be transformed into the correction matrix Linear transformation of the vector: .

[0067] The expression of the correction matrix is: , is the unit matrix, is the skew-symmetric matrix satisfying , is the outer product matrix.

[0068] Then, the correction matrix of rotation around the x-axis is ;

[0069] The correction matrix of rotation around the y-axis is ;

[0070] The correction matrix of rotation around the z-axis is .

[0071] Step 5: Coordinate Measurement and Calibration

[0072] When performing actual positioning, the binocular camera captures the target object (such as the feature points on the bridge pier), and calculates the original three-dimensional coordinates of the target through its own stereo vision algorithm. Before using the coordinates for subsequent calculations or control, the processor first uses the correction matrix generated in step 4 to rotate the original coordinates to a virtual coordinate system completely parallel to the marker plane, thereby eliminating the effects of installation errors and obtaining accurate corrected coordinates.

[0073] Let the coordinates of a point on the original plane be ,

[0074] Then the corresponding coordinates on the rotated plane are , , is a correction matrix. If rotating around the x-axis by an angle of is a correction matrix ,

[0075] is a transpose matrix ,

[0076] transformed coordinates: .

[0077] Step 6: Control instruction issuing

[0078] The processor calculates the direction and distance that the bridge girder needs to move according to the difference between the corrected accurate coordinates and the target position coordinates. Then, the accurate control instructions are sent to the programmable logic controller (PLC) of the bridge girder through the socket communication mode of the transmission control protocol (TCP). After receiving the instructions, the PLC drives the hydraulic or electrical system of the bridge girder to perform the corresponding fine adjustment action until the positioning is completed.

[0079] Through the above steps, the present application constitutes a complete closed-loop control system, realizing automatic detection, automatic compensation and precise positioning guidance of the installation error of the binocular vision system.

[0080] Embodiment 2: The present embodiment provides an error correction device suitable for the binocular vision system of the bridge girder, which is integrated on the bridge girder and used for high-precision positioning when hoisting the box girder.

[0081] The device mainly comprises a binocular vision system, at least three laser sensors and a processor.

[0082] In a preferred embodiment, the binocular vision system is composed of a binocular camera, which is installed on the bridge girder and faces the target area to be measured.

[0083] The laser sensor is preferably three high-precision laser sensors, and the installation layout of the sensor is preferably a right-angled triangle arrangement around the binocular camera, and the laser emission directions of the three laser sensors are parallel to each other and perpendicular to the installation plane. This layout simplifies the subsequent spatial geometric calculation.

[0084] The processor is usually an industrial computer, which is configured to perform the following operations: receiving depth information measured by the at least three laser sensors; based on the depth information, calculating in real time the spatial angle deviation between the sensor plane defined by the installation positions of the at least three laser sensors and the mark plane; generating a correction matrix for correcting coordinates in real time according to the spatial angle deviation; and applying the correction matrix to calibrate the original coordinates obtained by the binocular vision system to obtain corrected coordinates, so as to dynamically correct the coordinates in real time.

[0085] While the application has been described and illustrated with reference to specific preferred embodiments, it is not intended that it be limited to these particulars. Various changes in form and detail can be made without departing from the spirit and scope of the application as defined by the appended claims.

Claims

1. An error correction method for a binocular vision system of a bridge erection machine, characterized in that: The following steps are involved: S1: Install a binocular vision system and at least three laser sensors coplanarly at selected locations on the bridge erection machine, define the installation reference plane as the sensor plane, and ensure that the measurement directions of the at least three laser sensors are parallel to each other and perpendicular to the sensor plane; S2: On the erected bridge, within the measurement range of the binocular vision system and the at least three laser sensors, a marking plane is set as a measurement reference; S3: using the at least three laser sensors, respectively measuring depth information from each laser sensor to the marking plane; S4: Calculating a spatial angle deviation between the sensor plane and the marker plane based on the depth information; S5: Generating a correction matrix for correcting coordinates in real time according to the spatial angle deviation; S6: Using the binocular vision system to measure the target object to obtain original coordinates, and applying the correction matrix to calibrate the original coordinates to obtain corrected coordinates.

2. The error correction method for the binocular vision system of a bridge erection machine according to claim 1 is characterized in that: The binocular vision system is a binocular camera, and there are three laser sensors, which are arranged in a right triangle along the periphery of the binocular camera.

3. The error correction method for the binocular vision system of a bridge erection machine according to claim 2 is characterized in that: The binocular camera and three laser sensors are installed together on a mounting base plate. The mounting base plate is made of 7075-T6 aluminum alloy material, the surface of which is plane-ground and has a thickness of 15 mm to 20 mm.

4. The error correction method for the binocular vision system of a bridge erection machine according to claim 3 is characterized in that: The mounting substrate undergoes multi-stage leveling calibration through the following leveling steps to ensure that the coplanarity of the mounting reference surfaces of the binocular camera and the three laser sensors meets a preset standard: scanning the surface of the mounting substrate using a laser interferometer for initial leveling; And using an electronic level and a micrometer, fine calibration is performed by adjusting a fixing piece that fixes the laser sensor.

5. The error correction method for the binocular vision system of a bridge erection machine according to claim 1 is characterized in that: In step S2, the marking plane is laid on the edge of the top surface of the erected bridge box girder. The flatness of the marking plane is required to be less than 1 mm, and the material is matte anodized aluminum.

6. The error correction method for the binocular vision system of a bridge erection machine according to claim 2, characterized in that: The process of calculating the spatial angle deviation in step S4 includes: Define the plane where the binocular camera is located as the reference plane: ; Model the landmark plane as the target plane: ,in, is the slope, is the intercept; Determine the position information of three laser sensors based on the reference plane And obtain the depth information obtained by measuring the sign plane with three laser sensors ; Construct the matrix equation: ; Solving for the slope using the least squares method : , in, , , yes The transposed matrix of Based on the slope obtained 、 Calculate the spatial angle deviation : .

7. The error correction method for the binocular vision system of a bridge erection machine according to claim 6, characterized in that: The spatial angle deviation Substitute into the Rodriguez rotation formula to construct the correction matrix, Define the correction matrix R: , in, is the identity matrix, is an antisymmetric matrix, is the outer product matrix; Based on the correction matrix R, we get: Correction matrix for rotation around the x-axis ; Correction matrix for rotation around the y-axis ; Correction matrix for rotation around the z-axis .

8. The error correction method for the binocular vision system of a bridge erection machine according to claim 7, characterized in that: Assume that the original coordinates obtained by using the binocular camera to perform three-dimensional measurement of the target object are ,in is the plane coordinate, is the original depth value; Calculate the correction matrix and its transposed matrix ; Through matrix operations , the original coordinates Transform to corrected coordinates ,in is the corrected plane coordinate, is the corrected true depth value.

9. The error correction method for the binocular vision system of a bridge erection machine according to claim 7, characterized in that: The method further includes step S6: sending the corrected coordinates to a programmable logic controller for controlling the action of the bridge erection machine via a socket of a transmission control protocol.

10. A system for implementing the error correction method for a binocular vision system of a bridge erection machine according to claim 1, characterized in that: include: The marking plane is laid on the top edge of the erected bridge box girder; Install the baseboard; A binocular vision system, fixed on the mounting base plate, for performing three-dimensional measurement of the target object to obtain original coordinates; Three laser sensors are fixed on the mounting base and arranged in a right triangle around the binocular vision system, with the measurement directions of the laser sensors being parallel to each other and used to measure the depth information from the laser sensors themselves to the marking plane; The processor is configured to perform the following operations: receive depth information measured by the three laser sensors; and calculate, based on the depth information, in real time a spatial angle deviation between a sensor plane defined by the binocular vision system and the installation positions of the three laser sensors and the marking plane; According to the spatial angle deviation, a correction matrix for correcting coordinates is generated in real time; and the correction matrix is ​​applied to calibrate the original coordinates obtained by the binocular vision system to obtain corrected coordinates, and the coordinates are dynamically corrected in real time.

Images