A full-feature detection device and method for open freight vehicles

Abstract

The present invention discloses a full-feature detection device and method for an open freight vehicle, the device comprising: a laser scanner, which uses a laser to scan a target; a drive mechanism, which is used to drive the laser scanner to rotate and form a scanning surface; a control system, which is used to control the laser scanner and the drive mechanism to complete the scanning, and process the data collected by the laser scanner. The detection device and detection method with the above structure have the advantages of high precision, fast imaging and strong anti-interference ability, and can maintain a stable detection effect in a heavy dust environment. It is suitable for the more complex loading site working environment of cement, mineral materials, fertilizers and other production enterprises. The full-feature detection of the vehicle to be loaded can effectively improve the adaptability of the automatic loading model, and at the same time can effectively improve the safety of the loading process.

Description

Technical Field

[0001] The invention relates to a device and a method for detecting three-dimensional geometric information of an open freight vehicle. Background Art

[0002] With the acceleration of the pace of replacing people with automated machines, the demand for accurate three-dimensional detection in the automation industry has become increasingly prominent. In the application fields of fully automatic material loading and unmanned vehicle detection, the rapid and high-precision three-dimensional geometric information detection of vehicles has become very urgent. The working environment of the loading site of cement, mineral materials, fertilizers and other production enterprises is relatively complex, and the geometric dimensions and vehicle structures of the vehicles to be loaded are diverse. The realization of full-feature detection of the vehicles to be loaded can effectively improve the adaptability of the automatic loading models, and at the same time can effectively improve the safety of the loading process.

[0003] The existing three-dimensional geometric information detection of open freight vehicles generally adopts image-based special city detection and size detection. However, image detection is greatly affected by environmental interference, and the accuracy is difficult to guarantee. It has great limitations in practical applications. Therefore, a detection method with high accuracy, fast imaging and strong anti-interference ability is urgently needed. Summary of the invention

[0004] In view of this, the present invention provides an open freight vehicle full-feature detection device and method, which can perform full-feature detection of the open freight vehicle with high accuracy, fast imaging and strong anti-interference ability.

[0005] In order to solve the above technical problems, the technical solution of the present invention is: an open freight vehicle full-feature detection device, including: a laser scanner, which uses laser to scan the target; a driving mechanism, which is used to drive the laser scanner to rotate to form a scanning surface; a control system, which is used to control the laser scanner and the driving mechanism to complete the scanning, and process the data collected by the laser scanner.

[0006] As an improvement, an electrically controlled rotating table is used to drive the laser scanner to rotate 360°; an incremental encoder is used to detect the rotation angle of the laser scanner.

[0007] As an improvement, the control system is an embedded system, including: a control unit for controlling the electric control rotary table, the measuring encoder and the laser scanner; a data acquisition and processing unit for processing the data collected by the laser scanner and the incremental encoder; and an external interface for connecting to an external system for power supply and network communication.

[0008] The present invention also provides a full-feature detection method for an open freight vehicle, including: performing point cloud calibration to obtain correction parameters; performing installation calibration to obtain installation parameters; performing point cloud scanning on the target; introducing correction parameters and installation parameters to perform point cloud computing; classifying and extracting point clouds to complete the acquisition of three-dimensional geometric information.

[0009] As an improvement, the point cloud calibration includes: performing point cloud scanning on a reference object; inputting scanner cross-sectional data, 360° rotation angle data, and point cloud calibration parameters, and performing point cloud computing on the scanned reference object point cloud; adjusting correction parameters based on the result of comparing the point cloud computing with the actual reference object, and determining the correction parameters when the point cloud is not misaligned.

[0010] As an improvement, the point cloud calibration parameter is C = [σ r , σ p , Δ x , Δ y ],σ r is the roll angle deviation of the laser scanner installation, σ p is the pitch angle deviation of the laser scanner installation, Δ x and Δ y It is the installation eccentricity of the laser scanner on the plane of the electric control rotating table.

[0011] As an improvement, the point cloud computing of the scanned reference object point cloud is as follows:

[0012] The corrected installation eccentricity is calculated by formula II:

[0013]

[0014] Construct the matrix R using the turntable rotation angle β z ,

[0015]

[0016] Take the installation angle calibration parameters to construct the matrix R x , R y ,

[0017]

[0018]

[0019] Using formula I to perform point cloud computing,

[0020]

[0021] As an improvement, the installation calibration includes: performing point cloud scanning on the lane; determining the laser scanner installation heading angle; determining the laser scanner installation pitch angle; determining the laser scanner installation roll angle; determining the laser scanner installation height parameters; and outputting the installation parameters.

[0022] As an improvement, the step of determining the laser scanner installation heading angle includes:

[0023] Take the left and right boundary point clouds P A and P V Fitting a straight line L in the XY plane A and L V , take the average of the slopes of the two straight lines to calculate the orientation angle α of the lane in the scanner coordinate system M , and then construct the orientation calibration rotation matrix M c ;

[0024] Determining the installation pitch angle of the laser scanner comprises:

[0025] Journey to the West c After rotation, it satisfies -σ x <x<σ x , -σ y <y<σ y , σ y <σ x Points, fit the XZ plane straight line, take the slope of the straight line to calculate the installation pitch angle β M , construct the pitch calibration matrix M P ;

[0026] Determining the installation roll angle of the laser scanner comprises:

[0027] Journey to the West P After rotation, it satisfies -σ x <x<σ x , -σ y <y<σ y , σ x <σ y Point cloud, fit the YZ plane straight line, calculate the roll angle γ, and construct the roll calibration matrix M R ;

[0028] Determining the installation height of the laser scanner includes:

[0029] Using M R Correct the installation roll angle, select the points within the square with a side length of ε directly below the corrected point cloud, and take the average elevation as the equipment installation height parameter.

[0030] As an improvement, the classification and extraction of point clouds includes: extracting the point cloud of the bottom plate in the carriage; extracting the point cloud of the carriage railing; extracting the point cloud of the carriage pillar; extracting the point cloud of the carriage reinforcement; and extracting the point cloud of the carriage cylinder.

[0031] As an improvement, the extraction of the point cloud of the floor in the carriage includes:

[0032] Point cloud elevation grid, take the maximum density elevation H max ;

[0033] Take the elevation in [H max -Δ h , H max +Δh] range of point set P F ;

[0034] Get the point cloud of the floor inside the carriage and calculate the outer contour of the carriage;

[0035] Determine whether the outer contour line is empty. If so, move the contour line inward and continue to judge until it is not empty. If not, determine whether the point cloud on the contour line is continuous. If it is discontinuous, move the contour line inward and judge until it is continuous. If it is continuous, locate the inner floor contour of the car corresponding to the contour line.

[0036] As an improvement, the full-feature detection method for an open freight vehicle is characterized in that the calculation of the outer contour line of the vehicle compartment includes:

[0037] Point cloud P F Projection in multiple different directions, fitting 5 straight lines,

[0038]

[0039] Among them, L xz P F The straight line formed by the projection on the XZ plane, L xyL , L xyR , L yxF and L yxT P F The left and right and front and back maximum outer contour lines after projection on the XY plane.

[0040] As an improvement, the extraction of the point cloud of the carriage railing and the extraction of the point cloud of the carriage pillar include:

[0041] Take the plane projection point cloud of the guardrail, where the left and right guardrails are the projection point clouds on the XZ plane, and the front and rear guardrails are the projection point clouds on the YZ plane;

[0042] The projection point cloud of the left and right guardrails on the XZ plane is segmented along the X axis, and the projection point cloud of the front and rear guardrails on the YZ plane is segmented along the Y axis, and the lowest elevation H of the point cloud in each segment is calculated. M，Min and the highest elevation H M，Max ;

[0043] With elevation at H M，Min The points within the range of ±ΔH fit the contour line L under the guardrail ref ;

[0044] Take L ref The slope is taken as the reference value of the upper contour slope, and H M，Max The maximum value of the mean elevation is used as the intercept of the XZ plane line to construct the upper edge line L Uedg , construct straight lines downward until they are continuous.

[0045] L Uedg The straight line is used as the reference to divide the railing point cloud into point clouds P below the edge line. Below , point cloud P on the edge On And the point cloud P above the edge Beyond Three categories;

[0046] Take P Below and P On Points are taken as the vehicle guardrail point cloud, and P Beyond Point cloud of vehicle pillars;

[0047] Take P Beyond The point cloud of the vehicle pillar is segmented. Beyond The area along the X and Y axes is divided into multiple non-continuous intervals, and the elevation difference of the point cloud in each interval is judged, and the point cloud sets that meet the elevation difference threshold are used as different column point clouds.

[0048] As an improvement, the extraction of the point cloud of the carriage reinforcement and the extraction of the point cloud of the carriage oil cylinder include:

[0049] Take the point cloud set P in the car inner XY plane projection, with the rear railing as the reference line L ref , extract several upper boundary lines with intervals and continuous in the Y direction along the X axis, and make elevation judgment for each upper boundary point cloud set;

[0050] For each point set P in the segment interval i , judge according to the following rules

[0051] If P i If the point is close to the front of the vehicle in the X-axis direction and the span of the continuous interval in the Y-axis direction is less than 1 / 2 of the vehicle width, the point is judged to be the cylinder;

[0052] If P iClose to the front of the vehicle in the X-axis direction, and the span of the continuous interval in the Y-direction is greater than 1 / 2 of the vehicle width, P i The collective elevation is close to the height of the carriage floor, so the point is considered invalid;

[0053] If P i In the X-axis direction, it is far away from the front of the vehicle, and the span of the continuous interval in the Y direction is greater than 1 / 2 of the vehicle width. i The collective elevation is greater than the height of the car floor, and this point is judged to be a tie rod.

[0054] The benefits of the present invention lie in that the detection equipment and detection method with the above-mentioned structure have the advantages of high precision, rapid imaging and strong anti-interference ability, can maintain a stable detection effect in a heavy dust environment, and are suitable for the relatively complex loading site working environment of cement, mineral materials, fertilizer and other production enterprises. The full feature detection of the vehicles to be loaded can effectively improve the adaptability of the automatic loading vehicle model, and at the same time can effectively improve the safety of the loading process. BRIEF DESCRIPTION OF THE DRAWINGS

[0055] Figure 1 It is a schematic diagram of the appearance structure of the present invention.

[0056] Figure 2 It is a schematic diagram of the present invention.

[0057] Figure 3 This is a flow chart of point cloud calibration in the present invention.

[0058] Figure 4 Flow chart of installation and calibration in the present invention.

[0059] Figure 5 Schematic diagram of the global coordinate system.

[0060] Figure 6 Flowchart for extracting point cloud from car floor.

[0061] Figure 7 Flowchart for point cloud extraction of parapets and columns.

[0062] Markings in the figure: 1 laser scanner, 2 electric control rotary table, 3 incremental encoder, 4 embedded system, 41 external interface. DETAILED DESCRIPTION

[0063] In order to enable those skilled in the art to better understand the technical solution of the present invention, the present invention is further described in detail below in conjunction with specific implementation methods.

[0064] like Figure 1 , Figure 2As shown, the present invention provides a full-feature detection device for an open freight vehicle, including: a laser scanner 1, which uses laser to scan a target; a driving mechanism, which is used to drive the laser scanner to rotate to form a scanning surface; a control system, which is used to control the laser scanner and the driving mechanism to complete the scanning, and process the data collected by the laser scanner.

[0065] Specifically, the driving mechanism includes: an electrically controlled rotating table 2 for driving the laser scanner to rotate 360°; and an incremental encoder 3 for detecting the rotation angle of the laser scanner.

[0066] In this embodiment, the control system is an embedded system 4, including: a control unit for controlling the electric-controlled rotating table 2, the measuring encoder 3 and the laser scanner 1; a data acquisition and processing unit for processing the data collected by the laser scanner 1 and the incremental encoder 3; an external interface 41, connected to an external system for power supply and network communication.

[0067] The present invention also provides a method for detecting full features of an open freight vehicle, comprising:

[0068] S1 performs point cloud calibration to obtain correction parameters;

[0069] S2 performs installation calibration and obtains installation parameters;

[0070] S3 performs point cloud scanning on the target; introduces correction parameters and installation parameters for point cloud computing;

[0071] S4 classifies and extracts the point cloud to obtain three-dimensional geometric information.

[0072] In an ideal situation, the polar coordinates of the laser scanner are expressed as (α, d), where α is the emission angle, d is the measurement distance, and the angle of the electric control turntable β is i Derived from the incremental encoder. In the standard three-dimensional coordinate system, the point cloud is calculated as:

[0073]

[0074] However, in actual situations, factors such as installation errors and processing errors will introduce errors. The main sources of errors include:

[0075] (1) The relative attitude difference between the scanner and the electronically controlled turntable, including the roll and pitch angle deviation of the scanner;

[0076] (2) The offset between the scanner center point and the rotation center.

[0077] See also Figure 3 In order to correct the errors introduced in production assembly, point cloud verification must be performed. The specific steps are as follows:

[0078] S11 performs point cloud scanning on the reference object;

[0079] S12 inputs the scanner section data, 360° rotation angle data, and point cloud calibration parameters, and performs point cloud computing on the scanned reference object point cloud.

[0080] S121 introduces the calibration parameter C = [σ r , σ p , Δ x , Δ y ],σ r and σ p is the roll and pitch angle deviation of the laser scanner installation, Δ x and Δ y is the installation eccentricity of the scanner on the turntable plane. After the point cloud calibration parameters are introduced, the point cloud computing method is:

[0081] S122 is calculated by the formula P s , correct the installation eccentricity:

[0082]

[0083] S123 uses the turntable rotation angle β to construct the matrix R z :

[0084]

[0085] S124 takes the installation angle calibration parameters to construct the matrix R x , R y :

[0086]

[0087]

[0088] S125 then brings in the above parameters and uses formula I to perform point cloud computing.

[0089]

[0090] S13 adjusts the correction parameters according to the result of comparing the point cloud computing with the actual reference object, and determines the correction parameters when there is no misalignment in the point cloud.

[0091] The calibration parameters are calculated by trial and error. Walls and corners are selected as reference features. The calibration parameters are determined by binary search. The main basis for parameter adjustment includes:

[0092] (1) The thickness of the plane point cloud is small;

[0093] (2) The two point clouds within the turntable rotation angles of 0 to π and π to 2π overlap.

[0094] Installation and calibration completes the calculation of installation parameters of the detection equipment on site. To ensure that there is no blind spot scanning in the vehicle compartment, the detection equipment is installed above the lane, and its projection point on the lane plane is located in the vehicle compartment. The global coordinate system is defined as follows Figure 5 As shown, the system calibration parameters include:

[0095] (1) The heading angle of the laser scanner installation, that is, the azimuth angle relative to the boundaries of the lane;

[0096] (2) The laser scanner installation elevation angle, i.e., the elevation angle relative to the lane plane;

[0097] (3) The roll angle of the laser scanner installation, i.e., the roll angle relative to the lane plane;

[0098] (4) Vertical installation height;

[0099] See also Figure 4 The specific steps for calculating the installation and calibration parameters include:

[0100] S21 performs point cloud scanning of the lane;

[0101] S22 determines the laser scanner installation heading angle; obtains the left and right boundary point clouds P A and P V Fitting a straight line L in the XY plane A and L V , take the average of the slopes of the two straight lines to calculate the orientation angle α of the lane in the scanner coordinate system M , and then construct the orientation calibration rotation matrix M c ;

[0102] S23 determines the installation pitch angle of the laser scanner; c Satisfaction after rotation

[0103] -σ x <x<σ x , -σ y <y<σ y , σ y <σ x Points, fit the XZ plane straight line, take the slope of the straight line to calculate the installation pitch angle β M , construct the pitch calibration matrix M P ;

[0104] S24 determines the roll angle of the laser scanner installation; P Satisfaction after rotation

[0105] -σ x <x<σ x , -σ y <y<σ y , σ x <σy Point cloud, fit the YZ plane straight line, calculate the roll angle γ, and construct the roll calibration matrix M R ;

[0106] S25 determines the laser scanner installation height parameters; using M R Correct the installation roll angle, select the points within the square with a side length of ε directly below the corrected point cloud, and take the average elevation as the equipment installation height parameter.

[0107] S26 outputs installation parameters.

[0108] After scanning the point cloud of the freight vehicle, the correction parameters and installation parameters are introduced to perform point cloud computing. After the point cloud computing is completed, the calculated point cloud needs to be classified and extracted, including:

[0109] S41 extracts the point cloud of the floor in the carriage;

[0110] S42 extracts the point cloud of the carriage railing; extracts the point cloud of the carriage pillar;

[0111] S43 extracts the point cloud of the car body reinforcement and the point cloud of the car body cylinder.

[0112] Specifically, extracting the point cloud of the floor in the carriage includes the following steps:

[0113] S411 point cloud elevation grid, take the maximum density elevation H max ;

[0114] S412 takes the elevation at [H max -Δh, H max +Δh] range of point set P F ;

[0115] S413 obtains the point cloud of the bottom plate in the carriage and calculates the outer contour of the carriage; F Projection in multiple different directions, fitting 5 straight lines,

[0116]

[0117] Among them, L xz P F The straight line formed by the projection on the XZ plane, L xyL , L xyR , L yxF and L yxT P F The left and right and front and back maximum outer contour lines after projection on the XY plane.

[0118] S414 determines whether the outer contour line is empty. If so, the contour line is moved inward and the determination is continued until it is not empty. If not, the point cloud on the contour line is determined to be continuous. If it is not continuous, the contour line is moved inward and the determination is continued until it is continuous. If it is continuous, the contour line is located at the inner floor contour of the vehicle compartment corresponding to the contour line.

[0119] For the specific steps of extracting the point cloud of the carriage railing and the point cloud of the carriage pillar, see Figure 6 include:

[0120] S421 obtains the plane projection point cloud of the guardrail, wherein the left and right guardrails are the projection point clouds on the XZ plane, and the front and rear guardrails are the projection point clouds on the YZ plane;

[0121] S422 divides the projection point cloud of the left and right guardrails on the XZ plane into segments along the X axis, divides the projection point cloud of the front and rear guardrails on the YZ plane into segments along the Y axis, and calculates the lowest elevation H of the point cloud in each segment. M，Min and the highest elevation H M，Max ;

[0122] S423 is located at H M，Min The points within the +ΔH range fit the lower contour line L of the guardrail ref ;

[0123] S424 takes L ref The slope is taken as the reference value of the upper contour slope, and H M，Max The maximum value of the mean elevation is used as the intercept of the XZ plane line to construct the upper edge line L Uedg , construct straight lines downward until they are continuous.

[0124] S425 with L Uedg The straight line is used as the reference to divide the railing point cloud into point clouds P below the edge line. Below , point cloud P on the edge On And the point cloud P above the edge Beyond Three categories;

[0125] S426 Take P Below and P On Points are taken as the vehicle guardrail point cloud, and P Beyond Point cloud of vehicle pillars;

[0126] S427 Take P Beyond The point cloud of the vehicle pillar is segmented. Beyond The area along the X and Y axes is divided into multiple non-continuous intervals, and the elevation difference of the point cloud in each interval is judged, and the point cloud sets that meet the elevation difference threshold are used as different column point clouds.

[0127] After removing the point clouds of the bottom plate, railings and pillars, the point cloud inside the car includes the oil cylinder and internal reinforcement at the front end of the vehicle. The oil cylinder is fixed and connected to the front railing; the reinforcement is distributed inside the car, forming multiple approximately parallel straight lines. The specific steps for extracting the point cloud of the reinforcement and the point cloud of the oil cylinder include:

[0128] S431 Get the point cloud set P in the car inner XY plane projection, with the rear railing as the reference line L ref , extract several upper boundary lines with intervals and continuous in the Y direction along the X axis, and make elevation judgment for each upper boundary point cloud set;

[0129] S432 For each point set P in the segment interval i , judge according to the following rules

[0130] 1. If P i If the point is close to the front of the vehicle in the X-axis direction and the span of the continuous interval in the Y-axis direction is less than 1 / 2 of the vehicle width, the point is judged to be the cylinder;

[0131] 2. If P i Close to the front of the vehicle in the X-axis direction, and the span of the continuous interval in the Y-direction is greater than 1 / 2 of the vehicle width, P i The collective elevation is close to the height of the carriage floor, so the point is considered invalid;

[0132] 3. If P i In the X-axis direction, it is far away from the front of the vehicle, and the span of the continuous interval in the Y direction is greater than 1 / 2 of the vehicle width. i The collective elevation is greater than the height of the car floor, and this point is judged to be a tie rod.

[0133] Rule 1 distinguishes between the oil cylinder and the tie rod, while Rule 2 mainly restricts the bottom of the car body and its irregularities (front and rear warping), preventing the bottom of the car body from being identified as tie rods.

[0134] At this point, all features of the open freight vehicle compartment have been detected, identified and extracted, and the geometric parameters of each feature of the vehicle have been calculated. The geometric parameters can be used as a reference for the loader's automatic obstacle avoidance.

[0135] The above are only preferred embodiments of the present invention. It should be noted that the above preferred embodiments should not be regarded as limiting the present invention, and the protection scope of the present invention should be based on the scope defined by the claims. For ordinary technicians in this technical field, several improvements and modifications can be made without departing from the spirit and scope of the present invention, and these improvements and modifications should also be regarded as the protection scope of the present invention.

Claims

1. An open freight vehicle full-feature detection device, characterized in that include: Laser scanner, which uses laser to scan the target; A driving mechanism, used for driving the laser scanner to rotate and form a scanning surface; A control system, used to control the laser scanner and the drive mechanism to complete the scanning, and to process the data collected by the laser scanner; The method for detecting using the open freight vehicle full-feature detection device includes: Perform point cloud calibration to obtain correction parameters; including: Perform point cloud scanning on the reference object; Input the scanner section data, 360° rotation angle data, and point cloud calibration parameters to perform point cloud computing on the scanned reference point cloud; the point cloud calibration parameter is C = [σ r ,σ p ,Δ x ,Δ y ],σ r is the roll angle deviation of the laser scanner installation, σ p is the pitch angle deviation of the laser scanner installation, Δ x and Δ y The installation eccentricity of the laser scanner on the plane of the electric-controlled rotating table; The point cloud computing of the scanned reference point cloud is as follows: The corrected installation eccentricity is calculated by formula II: Construct the matrix R using the turntable rotation angle β z , Take the installation angle calibration parameters to construct the matrix R x , R y , Using formula I to perform point cloud computing, Adjust the correction parameters based on the comparison between the point cloud computing and the actual reference object, and determine the correction parameters when there is no misalignment in the point cloud; Perform installation verification and obtain installation parameters, including: Perform point cloud scanning of the lane; Determine the laser scanner installation heading angle; take the left and right boundary point clouds P A and P V Fitting a straight line L in the XY plane A and L V , take the average of the slopes of the two straight lines to calculate the orientation angle α of the lane in the scanner coordinate system M , and then construct the orientation calibration rotation matrix M c ; Determine the laser scanner installation pitch angle; learn from M c After rotation, it satisfies -σ x <x<σ x , -σ y <y<σ y ,σ y <σ x Points, fit the XZ plane straight line, take the slope of the straight line to calculate the installation pitch angle β M , construct the pitch calibration matrix M P ; Determine the roll angle of the laser scanner installation; learn from M P After rotation, it satisfies -σ x <x<σ x ,-σ y <y<σ y ,σ x <σ y Point cloud, fit the YZ plane straight line, calculate the roll angle γ, and construct the roll calibration matrix M R ; Determine the laser scanner installation height parameters; use M R Correct the installation roll angle, take the points within the square with a side length of ε directly below the corrected point cloud, and take the average elevation as the equipment installation height parameter; Output installation parameters; Perform point cloud scanning on the target; Introduce correction parameters and installation parameters for point cloud computing; The point cloud is classified and extracted to obtain three-dimensional geometric information.

2. The full-feature detection device for an open freight vehicle according to claim 1 is characterized in that The driving mechanism comprises: An electrically controlled rotating table is used to drive the laser scanner to rotate 360°; Incremental encoder for detecting the rotation angle of the laser scanner.

3. The full-feature detection device for an open freight vehicle according to claim 2 is characterized in that The control system is an embedded system, comprising: A control unit for controlling the electrically controlled rotary table, the measuring encoder and the laser scanner; A data acquisition and processing unit, used to process the data collected by the laser scanner and the incremental encoder; External interface, connected to external system for power supply and network communication.

4. A method for detecting full features of an open freight vehicle, characterized in that include: Perform point cloud calibration to obtain correction parameters; including: Perform point cloud scanning on the reference object; Input the scanner section data, 360° rotation angle data, and point cloud calibration parameters to perform point cloud computing on the scanned reference point cloud; the point cloud calibration parameter is C = [σ r ,σ p ,Δ x ,Δ y ],σ r is the roll angle deviation of the laser scanner installation, σ p is the pitch angle deviation of the laser scanner installation, Δ x and Δ y The installation eccentricity of the laser scanner on the plane of the electric-controlled rotating table; The point cloud computing of the scanned reference point cloud is as follows: The corrected installation eccentricity is calculated by formula II: Construct the matrix R using the turntable rotation angle β z , Take the installation angle calibration parameters to construct the matrix R x , R y , Using formula I to perform point cloud computing, Adjust the correction parameters based on the comparison between the point cloud computing and the actual reference object, and determine the correction parameters when there is no misalignment in the point cloud; Perform installation verification and obtain installation parameters, including: Perform point cloud scanning of the lane; Determine the laser scanner installation heading angle; take the left and right boundary point clouds P A and P V Fitting a straight line L in the XY plane A and L V , take the average of the slopes of the two straight lines to calculate the orientation angle α of the lane in the scanner coordinate system M , and then construct the orientation calibration rotation matrix M c ; Determine the laser scanner installation pitch angle; learn from M c After rotation, it satisfies -σ x <x<σ x , -σ y <y<σ y ,σ y <σ x Points, fit the XZ plane straight line, take the slope of the straight line to calculate the installation pitch angle β M , construct the pitch calibration matrix M P ; Determine the roll angle of the laser scanner installation; learn from M P After rotation, it satisfies -σ x <x<σ x ,-σ y <y<σ y ,σ x <σ y Point cloud, fit the YZ plane straight line, calculate the roll angle γ, and construct the roll calibration matrix M R ; Determine the laser scanner installation height parameters; use M R Correct the installation roll angle, take the points within the square with a side length of ε directly below the corrected point cloud, and take the average elevation as the equipment installation height parameter; Output installation parameters; Perform point cloud scanning on the target; Introduce correction parameters and installation parameters for point cloud computing; The point cloud is classified and extracted to obtain three-dimensional geometric information.

5. A method for detecting full features of an open freight vehicle according to claim 4, characterized in that The classification and extraction of point clouds comprises: Extract the point cloud of the floor in the carriage; Extract the point cloud of the carriage railing; Extract the point cloud of the carriage pillars; Extract the point cloud of the car body reinforcement; Extract the point cloud of the carriage cylinder.

6. A method for detecting full features of an open freight vehicle according to claim 5, characterized in that The extraction of the point cloud of the floor in the carriage includes: Point cloud elevation grid, take the maximum density elevation H max ; Take the elevation in [H max -Δh,H max +Δh] range of point set P F ; Get the point cloud of the floor inside the carriage and calculate the outer contour of the carriage; Determine whether the outer contour line is empty. If so, move the contour line inward and continue to determine until it is not empty. If not, determine whether the point cloud on the contour line is continuous. If it is not continuous, move the contour line inward and determine until it is continuous. If it is continuous, locate the inner floor contour of the car corresponding to the contour line.

7. A method for detecting full features of an open freight vehicle according to claim 6, characterized in that The calculation of the carriage outer contour line comprises: Point cloud P F Projection in multiple different directions, fitting 5 straight lines, Among them, L xz P F The straight line formed by the projection on the XZ plane, L xyL , L xyR , L yxF and L yxT P F The left and right and front and back maximum outer contour lines after projection on the XY plane.

8. A method for detecting full features of an open freight vehicle according to claim 5, characterized in that The extraction of the point cloud of the carriage railing and the extraction of the point cloud of the carriage pillar include: Take the plane projection point cloud of the guardrail, where the left and right guardrails are the projection point clouds on the XZ plane, and the front and rear guardrails are the projection point clouds on the YZ plane; The projection point cloud of the left and right guardrails on the XZ plane is segmented along the X axis, and the projection point cloud of the front and rear guardrails on the YZ plane is segmented along the Y axis, and the lowest elevation H of the point cloud in each segment is calculated. M,Min and the highest elevation H M,Max ; With elevation at H M,Min The points within the range of ±ΔH fit the contour line L under the guardrail ref ; Take L ref The slope is taken as the reference value of the upper contour slope, and H M,Max The maximum value of the mean elevation is used as the intercept of the XZ plane line to construct the upper edge line L Uedg , construct straight lines downward until they are continuous; L Uedg The straight line is used as the reference to divide the railing point cloud into point clouds P below the edge line. Below , point cloud P on the edge On And the point cloud P above the edge Beyond Three categories; Take P Below and P On Points are taken as the vehicle guardrail point cloud, and P Beyond Point cloud of vehicle pillars; Take P Beyond The point cloud of the vehicle pillar is segmented into the point cloud of the pillar; Beyond The area along the X and Y axes is divided into multiple non-continuous intervals, and the elevation difference of the point cloud in each interval is judged, and the point cloud sets that meet the elevation difference threshold are used as different column point clouds.

9. A method for detecting full features of an open freight vehicle according to claim 5, characterized in that The extraction of the car reinforcement point cloud and the extraction of the car cylinder point cloud include: Take the point cloud set P in the car inner XY plane projection, with the rear railing as the reference line L ref , extract several upper boundary lines with intervals and continuous in the Y direction along the X axis, and make elevation judgment for each upper boundary point cloud set; For each point set P in the segment interval i , judge according to the following rules If P i If the point is close to the front of the vehicle in the X-axis direction and the span of the continuous interval in the Y-axis direction is less than 1 / 2 of the vehicle width, the point is judged to be the cylinder; If P i Close to the front of the vehicle in the X-axis direction, and the span of the continuous interval in the Y-direction is greater than 1 / 2 of the vehicle width, P i The collective elevation is close to the height of the carriage floor, so the point is considered invalid; If P i In the X-axis direction, it is far away from the front of the vehicle, and the span of the continuous interval in the Y direction is greater than 1 / 2 of the vehicle width. i The collective elevation is greater than the height of the car floor, and this point is judged to be a tie rod.

Images