CN115615356A - 3D gluing detection analysis method based on template track - Google Patents

Abstract

The invention discloses a 3D gluing detection and analysis method based on template tracks, which comprises the following steps: step S1: calculating hand-eye conversion matrix and T of each laser receiver and the tail end of the moving mechanism by a line laser hand-eye calibration method ₁ 、T ₂ ……T _n (ii) a Step S2: template data are recorded in advance for different to-be-detected glued workpieces; and step S3: under the same space coordinate system, poor track points are adjusted according to the recorded robot TCP pose and the scanning glue path point cloud; and step S4: calculating a cutting and slicing plane equation of each point by using the adjusted track points; step S5: from the post-slicing data. By the method, the calculated amount is reduced, the segmentation data does not need to be recalculated once during each operation, and the data acquired during each detection is transferred to the coordinate system of the alignment template according to the offset obtained by the positioning equipmentAnd the pose is high in reliability and more accurate in segmentation in the relatively stable industrial environment.

Description

3D gluing detection analysis method based on template track

Technical Field

The application relates to the technical field of gluing detection, in particular to a 3D gluing detection analysis method based on template tracks.

Background

In the current industrial field, a common method for gluing detection is 2D detection, most of the 3D gluing detection schemes on the market are foreign brands, the hand-eye relationship between a gluing detection sensor and a glue gun is not usually calibrated, a complete glue path contour point cloud is obtained by manually adjusting splicing tracks of different sensors, a glue path slicing mode is adopted for analyzing a glue path, but the position angle of a slice still depends on manually adjusting the glue path slicing center to extract a glue section slice;

in the actual working process, the scheme has the following problems:

1. due to the relative relationship between the sensorless and the robot, accurate and precise rubber road 3D data are obtained by accurately splicing point clouds through a robot end running track;

2. the segmentation of the glue path completely depends on manual work, and the result is greatly interfered by manual operation;

3. the debugging process is complex and has unsatisfactory effect, and long-time continuous debugging is needed.

Disclosure of Invention

Aiming at the problems, the invention provides a 3D gluing detection analysis method based on a template track, which comprises the following steps:

step S1: calculating hand-eye conversion matrix and T of each laser receiver and the tail end of the moving mechanism by a line laser hand-eye calibration method ₁ 、T ₂ ……T _n ；

Step S2: for different to-be-detected gluing workpieces, recording template data in advance;

and step S3: under the same space coordinate system, poor track points are adjusted according to the recorded robot TCP pose and the scanning glue path point cloud;

and step S4: calculating a cutting slicing plane equation of each point by using the adjusted track points;

step S5: analyzing each segment of data according to the sliced data, checking the segmentation correctness, if the segmentation correctness is not good, adjusting track points, repeatedly testing, and modifying to be completely correct;

step S6: saving the point set and a cutting slicing plane equation thereof as a template;

step S7: and (4) carrying out test acquisition flow work by utilizing the template, and finally realizing the processing and analysis of the rubber road slice data.

Further, the step S2 is used for recording template data, and the specific method steps are as follows:

1) Placing a normal defect-free glued workpiece to a detection position;

2) Calibrating the position and posture of the gluing TCP point by the robot according to a 6-point method;

3) Completing the compiling of the gluing detection scanning track of the robot;

4) Starting the robot and the gluing detection sensor, and recording the position and posture data of the robot and the gluing detection sensor acquisition data in the whole scanning process;

5) And finishing point cloud splicing according to the relation between each hand and each eye.

Further, in the step S3, the method for adjusting the bad track point specifically includes the following steps:

i) taking the actual running direction of the glued road point cloud as a track point direction reference, and taking a deviated track point as a poor track;

II) directly adjusting coordinates of the poor track points to enable the poor track points to accord with the trend.

Further, the adjusting and calculating method of the trace point in the step S3 includes the following steps:

a) Acquiring and recording to obtain an original track point set of the robot, wherein the original track point set is P _{ORI_1} 、P _{ORI_2} 、P _{ORI_3} ……P _{ORI_n} ；

B) Observe the track point atManually adjusting the track points deviating from the glue path according to the distribution condition of the glue point cloud to obtain an adjusted track point set, wherein the adjusted track point set is P _{FIX_1} 、P _{FIX_2} 、P _{FIX_3} ……P _{FIX_n} ；

C) The track points are spatially ordered points according to a set chord length L _chord Segmenting the point set, and establishing a quintic spline equation system for calculating segmentation;

d) According to the resampling distance L _resample And performing point resampling on the obtained segmented quintic spline curve equation set to obtain a slice point set P distributed at uniform intervals ₁ 、P ₂ 、P ₃ ……P _n 。

Further, the method for calculating the cutting and slicing plane equation of each point in the step S4 comprises the following steps;

i) pair slice point set P ₁ 、P ₂ 、P ₃ ……P _n The derivation of the corresponding quintic spline curve equation of each point is carried out to obtain the tangent normal vector N of each point on the curve where each point is located ₁ 、N ₂ 、N ₃ ……N _n ；

II) according to point P _n And its normal vector N _n The passing point P can be obtained _n And the normal vector is N _n Equation A of the space plane _n X+B _n Y+C _n Z =0,A, B, C are plane equation coefficients.

Further, the data analysis method in step S5 includes the following steps:

a) Point P of glue road integral point cloud _n Where the set size is L length and H height _n Width W of _n Filtering the bounding box, and filtering point cloud data outside the bounding box;

b) Projecting the filtered data to a plane A _n X+B _n Y+C _n Z =0, the dimension is reduced from 3D data X, Y, Z to 2D data X, Y, and the processed point set is the slice data at the point.

Further, the length L and the height H _n And width W _n 0.5mm, 40mm and 50mm respectively.

Further, the test acquisition process in S7 specifically includes the following steps:

firstly), a workpiece to be detected is in place, and a spatial position deviation matrix T of the workpiece in the template manufacturing position is calculated in a four-corner positioning or 3D vision mode _offset ；

II): starting the robot and the gluing detection sensor, and recording the position and posture data of the robot and the gluing detection sensor acquisition data in the whole scanning process;

thirdly): completing the point cloud splicing of the glue path to be detected according to the relation of each hand and the eye to obtain the point cloud PCL _new ；

Fourthly): according to a deviation matrix T _offset Point cloud PCL _new Inner point P _{new_n} Converting to the corresponding point P in the template state _{model_n} ＝T _offset *P _{new_n} Obtaining the point cloud PCL after conversion _model ；

Fifthly), the steps of: dividing the converted point cloud PCL by using the established template track and using each point of the point set in the template track and the cutting fragment plane equation thereof _model And obtaining the glue road data of each frame.

Has the beneficial effects that:

according to the technical scheme, the hand-eye relation between the robot and the gluing detection laser sensor is calibrated in advance, splicing data are converted by the hand-eye relation instead of acquiring workpiece data of each project, the splicing point cloud angle corresponding to each frame is obtained through manual correction, data splicing is more accurate, and splicing precision is inevitably high after a calibration flow is normally completed.

The method has the advantages that the precision is not interfered by a debugging worker, the point cloud can be spliced randomly only by completing calibration once, the splicing relation of each workpiece is not needed to be corrected and debugged once, the preset track is used for segmenting the rubber road data, the calculated amount can be reduced, the segmentation data does not need to be recalculated every time of operation, meanwhile, the data is detected and collected every time, the offset is obtained according to the positioning equipment, the pose under the template coordinate system is shifted and aligned, and the method is high in reliability and accurate in segmentation under the relatively stable industrial environment.

Drawings

FIG. 1 is a flow chart of template track recording of a 3D gluing detection analysis method based on template track according to the present invention;

FIG. 2 is a flowchart of a method for calculating an overall glue path trajectory slice in a 3D gluing detection and analysis method based on a template trajectory according to the present invention;

fig. 3 is a flow of actual acquisition processing performed in the template trajectory-based 3D glue detection and analysis method provided by the present invention.

Detailed Description

In order to make the invention more comprehensible to those skilled in the art, the invention is described below with reference to the following embodiments and the accompanying drawings, in which reference is made to fig. 1 to 3.

In order to realize the content of the invention, the invention designs a 3D gluing detection analysis method based on a template track, which comprises the following steps:

step S1: calculating hand-eye conversion matrix and T of each laser receiver and the tail end of the moving mechanism by a line laser hand-eye calibration method ₁ 、T ₂ ……T _n ；

Step S2: pre-recording template data of different glued workpieces to be detected, placing the glued workpieces with normal defects to a detection position when recording the template data, calibrating the poses of gluing TCP points by a robot according to a 6-point method, then completing writing of gluing detection scanning tracks of the robot, starting the robot and gluing detection sensing, recording the pose data of the robot and the data collected by the gluing detection sensing in the whole scanning process, and finally completing point cloud splicing according to the relation of hands and eyes;

and step S3: under the same space coordinate system, poor track points are adjusted according to the recording robot TCP pose and the scanning glue path point cloud, firstly, the actual running direction of the glue path point cloud is used as the track point direction reference, the greater distance deviating from the track points is used as a poor track, then the coordinates of the poor track points are directly adjusted to enable the poor track points to be in line with the trend, under the same space coordinate system, the poor track points are finely adjusted according to the recording robot TCP pose and the scanning glue path point cloud to enable the poor track points to be approximate to the gluing track, the point set is segmented according to the set chord length of 20mm, a calculating segmented quintic spline curve equation set is established, and according to the resampling distance of 0.1mm and the obtained segmented quintic spline curve equation setGrouping, point resampling is carried out to obtain a slice point set P with uniform interval distribution ₁ 、P ₂ 、P ₃ ……P _n ；

And step S4: calculating the cutting slicing plane equation of each point by using the adjusted track points, and collecting P slicing points ₁ 、P ₂ 、P ₃ ……P _n The derivation of the corresponding quintic spline curve equation of each point is carried out to obtain the tangent normal vector N of each point on the curve where each point is located ₁ 、N ₂ 、N ₃ ……N _n According to the point P and its normal vector N _n The passing point P can be obtained _n And the normal vector is N _n Equation A of the space plane _n X+B _n Y+C _n Z =0,A, B, C are the coefficients of the plane equation;

step S5: analyzing each segment of data according to the sliced data, checking the segmentation correctness, adjusting track points if the segmentation correctness is not good, repeating the test, and modifying the point cloud of the whole glue path to the point P _n According to the size of the glue, setting a bounding box with the size of 0.5mm in length, 40mm in height and 50mm in width for filtering, filtering point cloud data outside the bounding box, and projecting the filtered data to a plane A _n X+B _n Y+C _n Z =0, reducing dimension from 3D data X, Y, Z to 2D data X, Y, and the point set after processing is point P _n The slice data of the position is checked and analyzed for template data, and fine adjustment is optimized.

Step S6: saving the point set and a cutting slicing plane equation thereof as a template;

step S7: the template is utilized to carry out testing and collecting flow work, finally processing and analyzing of the rubber road section data are achieved, the testing and collecting flow comprises that a workpiece to be detected is in place, and a space position deviation matrix T of the workpiece at the template manufacturing position is calculated in a four-corner positioning mode or a 3D vision mode _offset Starting the robot and the glue coating detection sensor, recording the robot position and attitude data and the glue coating detection sensor acquisition data in the whole scanning process, completing the point cloud splicing of the glue path to be detected according to the relation between hands and eyes to obtain the point cloud PCL _new From deviation matrix T _offset PCL point cloud _new Inner point P _{new_n} Conversion to corresponding points in the template stateP _{model_n} ＝T _offset *P _{new_n} Obtaining the point cloud PCL after conversion _model Dividing the converted point cloud PCL by using the established template track and using each point of the point set in the template track and the cutting fragment plane equation thereof _model And obtaining the glue road data of each frame.

In conclusion, the method and the device have the advantages that the preset track is used for segmenting the rubber road data, the calculated amount can be reduced, the segmentation data does not need to be recalculated every time of operation, the data are detected and collected every time, the offset is obtained according to the positioning equipment, the pose under the template coordinate system is transferred and aligned, and the method and the device are high in reliability and accurate in segmentation under the relatively stable industrial environment.

Claims (8)

1. A3D gluing detection analysis method based on template tracks is characterized by comprising the following steps;

step S1: calculating hand-eye conversion matrix and T of each laser receiver and the tail end of the moving mechanism by a line laser hand-eye calibration method ₁ 、T ₂ ……T _n ；

Step S2: for different to-be-detected gluing workpieces, recording template data in advance;

and step S3: under the same space coordinate system, poor track points are adjusted according to the recorded robot TCP pose and the scanning glue path point cloud;

and step S4: calculating a cutting slicing plane equation of each point by using the adjusted track points;

step S5: analyzing each segment of data according to the sliced data, checking the segmentation correctness, if the segmentation correctness is not good, adjusting track points, repeatedly testing, and modifying to be completely correct;

step S6: saving the point set and the cutting fragment plane equation thereof as a template;

step S7: and (4) performing test acquisition flow work by using the template, and finally realizing the processing and analysis of the glue road slice data.

2. The template trajectory-based 3D gluing detection and analysis method according to claim 1, wherein the step S2 is used for recording template data, and the specific method steps are as follows:

1) Placing a normal defect-free glued workpiece to a detection position;

2) Calibrating the position and posture of the gluing TCP point by the robot according to a 6-point method;

3) Completing the compiling of the gluing detection scanning track of the robot;

4) Starting the robot and the gluing detection sensor, and recording the position and posture data of the robot and the gluing detection sensor acquisition data in the whole scanning process;

5) And finishing point cloud splicing according to the relation between each hand and each eye.

3. The template trajectory-based 3D gluing detection and analysis method according to claim 2, wherein poor trajectory points are adjusted in the step S3, and the specific method comprises the following steps:

i) taking the actual running direction of the glued road point cloud as a track point direction reference, and taking a deviated track point as a poor track;

II) directly adjusting coordinates of the poor track points to enable the poor track points to accord with the trend.

4. The template trajectory-based 3D gluing detection and analysis method according to claim 3, wherein the adjustment calculation method of the trajectory points in the step S3 comprises the following steps:

a) Acquiring and recording to obtain an original track point set of the robot, wherein the original track point set is P _{ORI_1} 、P _{ORI_2} 、P _{ORI_3} ……P _{ORI_n} ；

B) Observing the distribution condition of the track points on the gluing point cloud, manually adjusting the track points deviating from the gluing path to obtain an adjusted track point set, wherein the adjusted track point set is P _{FIX_1} 、P _{FIX_2} 、P _{FIX_3} ……P _{FIX_n} ；

C) The track points are spatially ordered points according to a set chord length L _chord Segmenting the point set, and establishing a quintic spline equation system for calculating segmentation;

d) According to the resampling distance L _resample And the obtained piecewise quintic spline curve equationGrouping, point resampling is carried out to obtain a slice point set P with uniform interval distribution ₁ 、P ₂ 、P ₃ ……P _n 。

5. The template trajectory-based 3D gluing detection and analysis method according to claim 4, wherein the method for calculating the cutting and slicing plane equations of each point in the step S4 comprises the following steps;

i) To slice point set P ₁ 、P ₂ 、P ₃ ……P _n The derivation of the corresponding quintic spline curve equation of each point is carried out to obtain the tangent normal vector N of each point on the curve where each point is located ₁ 、N ₂ 、N ₃ ……N _n ；

II) according to point P _n And its normal vector N _n The passing point P can be obtained _n And the normal vector is N _n Equation A of the space plane _n X+B _n Y+C _n Z＝0，A _n 、B _n 、C _n Are the plane equation coefficients.

6. The template trajectory-based 3D gluing detection and analysis method according to claim 5, wherein the data analysis method in the step S5 comprises the following steps:

a) Point P of glue road integral point cloud _n Where the set size is L length and H height _n Width W of _n Filtering the bounding box, and filtering point cloud data outside the bounding box;

b) Projecting the filtered data to a plane A _n X+B _n Y+C _n Z =0, the dimension is reduced from 3D data X, Y, Z to 2D data X, Y, and the processed point set is the slice data at the point.

7. The template trajectory-based 3D gluing detection and analysis method according to claim 6, wherein the length L and the height H are _n And width W _n 0.5mm, 40mm and 50mm respectively.

8. The template trajectory-based 3D gluing detection and analysis method according to claim 7, wherein the test acquisition process in S7 specifically comprises the following steps:

firstly), a workpiece to be detected is in place, and a spatial position deviation matrix T of the workpiece in the template manufacturing position is calculated in a four-corner positioning or 3D vision mode _offset ；

II): starting the robot and the gluing detection sensor, and recording the position and posture data of the robot and the gluing detection sensor acquisition data in the whole scanning process;

thirdly): completing the point cloud splicing of the glue path to be detected according to the relation of each hand and the eye to obtain the point cloud PCL _new ；

Fourthly), the steps of: according to a deviation matrix T _offset Point cloud PCL _new Inner point P _{new_n} Converting to the corresponding point P in the template state _{model_n} ＝T _offset *P _{new_n} Obtaining the converted point cloud PCL _model ；

Fifthly), the steps of: dividing the converted point cloud PCL by using the established template track and using each point of the point set in the template track and the cutting fragment plane equation thereof _model And obtaining the glue road data of each frame.

Images