CN102645219B - Welding and locating method of welding seam and method of obtaining welding seam offset for visual navigation system of wall climbing robot for weld inspection - Google Patents

Abstract

The visual navigation system of the wall-climbing robot used for weld seam detection, the welding positioning method of the weld seam, and the welding seam offset acquisition method, the present invention relates to the visual navigation system of the wall-climbing robot used for weld seam detection and the welding seam The positioning method and the method of obtaining the offset of the welding seam. In order to solve the problem that the existing technology in places with harsh natural environments cannot realize the inspection and maintenance work of the tower in the way of teaching and reproduction. The charge-coupled camera and cross laser emitter of the visual navigation system are fixed at the front end of the robot head, and the cross laser emitter is directly above the charge-coupled camera, and the laser emitted by the cross laser emitter irradiates the surface of the workpiece to be welded to form a cross spot , the charge-coupled camera is used to photograph the cross spot on the workpiece surface, and the data output end of the charge-coupled camera is connected with the data input end of the computer. The invention is applicable to the field of inspection and maintenance of tower tubes.

Description

Welding seam welding positioning method and welding seam offset acquisition method of the visual navigation system of the wall-climbing robot used for welding seam detection

technical field

The invention is applicable to the field of inspection and maintenance of tower tubes.

Background technique

In recent years, the wind power industry has developed vigorously, and the number of wind power towers has increased sharply. Correspondingly, the demand for maintenance of towers has also greatly increased. The height of a general wind power tower is between 50 and 100 meters. The tower is divided into several sections, and each section is welded by a single steel plate coil. Due to the influence of welding process and welding precision, problems such as weak welding, pores, and slag inclusions will inevitably occur in the weld part, which will easily cause further cracks during the use of the wind field. Since most of the wind farms are built in places with harsh natural environments such as seas, canyons, and mountain passes, the above problems have become a great safety hazard in the use of wind turbines. For a long time, the overhaul and maintenance of wind power towers have been carried out by means of hanging operation, which has high risks. Therefore, there is an urgent need for machines that work in extreme modes to replace manual operations. Under such extreme conditions, it is difficult to realize the training of the robot's working process through teaching and reproduction. It is necessary for the robot to have a more intelligent control system to adapt to the working environment and perform related tasks.

Contents of the invention

In order to solve the problem that the existing technology in the harsh natural environment cannot realize the inspection and maintenance of the tower tube by the teaching and reproduction method, a welding positioning method and a welding seam welding positioning method for the visual navigation system of the wall-climbing robot for welding seam detection are provided. Seam offset acquisition method.

The visual navigation system of the wall-climbing robot used for welding seam detection according to the present invention includes a charge-coupled camera, a cross laser emitter and a computer,

The charge-coupled camera and the cross laser emitter are fixed on the front end of the robot head, and the cross laser emitter is directly above the charge-coupled camera, and the laser emitted by the cross laser emitter forms a cross spot on the surface of the workpiece to be welded. The charge-coupled camera is used to shoot the cross spot on the surface of the workpiece, and the angle between the optical axis of the laser beam emitted by the cross laser transmitter and the optical axis of the camera head of the charge-coupled camera is 45°, and the data output terminal of the charge-coupled camera is connected to the computer's Data input connection.

Welding location method, it comprises the steps:

Step 1. The cross laser emitter emits a red cross-shaped laser beam. The laser beam is irradiated on the surface of the workpiece to be welded to form a cross spot, and the laser beam is used to scan the surface of the workpiece. During the scanning process, a charge-coupled camera is used to collect the surface of the workpiece video information, go to step 2;

Step 2, the computer receives the video information collected by the charge-coupled camera;

Step 3, the computer processes each frame of image in the video information as follows:

Step 31, the frame image is divided into pixels, and the value of the pixel represents the value of the red, yellow and blue colors of the image at this point;

Step 32, storing the values of all red, yellow and blue color components of the frame image in three arrays respectively;

Step 33, the computer obtains the optimal threshold value of the gray value corresponding to the red component value by the two-dimensional maximum entropy segmentation method for the array of the red component corresponding to the frame image;

Step three and four, compare the gray value of all pixels of the frame image with the optimal threshold obtained in step three and three, set the gray value of the pixel higher than the threshold to 255, and set the gray value of the pixel lower than the threshold Set the value to 0 to convert the frame image into a binary image;

Steps 3 and 5, using the Canny operator to perform edge detection on the binary image obtained in steps 3 and 4, to obtain edge information of the binary image, which is shown as the edge of the cross spot on the frame image;

Steps three and six, extracting the edge information of steps three and five, taking the midline of the edge of the cross spot as the skeleton, and obtaining a smooth skeleton curve with a single pixel width;

Step 37: Carry out the Hough transformation of the straight line according to the skeleton curve obtained in step 36 to obtain two straight lines on both sides of the arc, mark the two endpoints of the arc, find the arc part, and detect the weld in the frame image position in .

The welding seam offset acquisition method, after the welding seam positioning, extracts the two endpoints of the arc of the welding seam and the center point of the crosslight three feature points, the distance between the two endpoints is a, and the distance between the center point and the welding seam The distance between the end points adjacent to the seam is b, and the measured values of a and b are a', b', and the offset angle can be obtained as:

$\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arccos</span>}\frac{\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; b</span>}}{{\mathrm{<span\; class="notranslate">\; a</span>}}^{<span\; class="notranslate">\; \text{'}</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; b</span>}}^{<span\; class="notranslate">\; \text{'}</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arccos</span>}\frac{{\mathrm{<span\; class="notranslate">\; b</span>}}^{<span\; class="notranslate">\; \text{'}</span>}}{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span>$

The position offset is:

w=b'-b, (15)

When a'=a, b'=b, it is a normal situation, and the robot is moving normally; when a'>a, b'≤b, the robot shifts to the left; when a'>a, b'≥b, the robot right offset.

The present invention uses the visual navigation system of the wall-climbing robot for weld seam detection to enable the robot to search for weld seams by itself through the visual navigation system; determine the weld seam position through the welding seam welding positioning method, and track the welding seam through the welding seam offset acquisition method Seam direction, so as to ensure that the flaw detection equipment can always detect along the weld seam, and achieve the purpose of not deviating from the weld seam.

Description of drawings

Fig. 1 is a schematic structural diagram of the visual navigation system of a wall-climbing robot used for weld detection; Fig. 2 is a two-dimensional histogram of gray-neighborhood gray-scale mean; Fig. 3 is a schematic diagram of a weld; Fig. 4 is a laser projection and The projection diagram of the offset of the intersection part of the weld; Figure 5 is the comparison between the laser projection and the offset of the intersection of the weld and the standard situation; Figure 6 is the equivalent diagram of Figure 5.

Detailed ways

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS 1. This embodiment is specifically described in conjunction with FIG. 1. The visual navigation system of the wall-climbing robot used for weld seam detection described in this embodiment includes a charge-coupled camera 1, a cross laser emitter 2 and a computer.

The charge-coupled camera 1 and the cross laser emitter 2 are fixed on the front end of the robot head, and the cross laser emitter 2 is directly above the charge-coupled camera 1, and the laser emitted by the cross laser emitter 2 is irradiated on the workpiece to be welded A cross spot is formed on the surface, and the charge-coupled camera 1 is used to photograph the cross spot on the workpiece surface, and the angle between the optical axis of the laser beam emitted by the cross laser emitter 2 and the optical axis of the camera head of the charge-coupled camera 1 is 45°, and the charge-coupled The data output end of the camera 1 is connected with the data input end of the computer.

Specific embodiment two, based on the welding positioning method of the weld seam of the visual navigation system of the wall-climbing robot that is used for weld seam detection described in specific embodiment one, it comprises the following steps:

Step 1. The cross laser transmitter 2 emits a red cross-shaped laser beam. The laser beam is irradiated on the surface of the workpiece to be welded to form a cross spot, and the laser beam is used to scan the surface of the workpiece. During the scanning process, the charge-coupled camera 1 is used to collect For the video information on the surface of the workpiece, perform step 2;

Step 2, the computer receives the video information collected by the charge-coupled camera;

Step 3, the computer processes each frame of image in the video information as follows:

Step 31, the frame image is divided into pixels, and the value of the pixel represents the value of the red, yellow and blue colors of the image at this point;

Step 32, storing the values of all red, yellow and blue color components of the frame image in three arrays respectively;

Step 33, the computer obtains the optimal threshold value of the gray value corresponding to the red component value by the two-dimensional maximum entropy segmentation method for the array of the red component corresponding to the frame image;

Step three and four, compare the gray value of all pixels of the frame image with the optimal threshold obtained in step three and three, set the gray value of the pixel higher than the threshold to 255, and set the gray value of the pixel lower than the threshold Set the value to 0 to convert the frame image into a binary image;

Steps 3 and 5, using the Canny operator to perform edge detection on the binary image obtained in steps 3 and 4, to obtain edge information of the binary image, which is shown as the edge of the cross spot on the frame image;

Steps three and six, extracting the edge information of steps three and five, taking the midline of the edge of the cross spot as the skeleton, and obtaining a smooth skeleton curve with a single pixel width;

Step 37: Carry out the Hough transformation of the straight line according to the skeleton curve obtained in step 36 to obtain two straight lines on both sides of the arc, mark the two endpoints of the arc, find the arc part, and detect the weld in the frame image position in .

Hough transformation in step 37, the specific steps to obtain the two straight lines on both sides of the arc are:

Step 371. Quantize (ρ, θ) space: the skeleton curve ρ=xcosθ+ysinθ obtains a two-dimensional matrix A(ρ,θ), and initializes A(ρ,θ) as an all-zero matrix;

Step 372: Calculate the corresponding ρ from each quantized value of θ for each non-zero gray value pixel coordinate (x, y) of the image, let a(i,j)+1→a(i,j ), a(i,j) is the element value of the i-th row j column of the matrix A(ρ,θ);

Step 373: After processing all non-zero gray value pixel coordinates (x, y), analyze A(ρ, θ), if A(ρ, θ) is greater than the threshold T, then there is a line segment, (ρ, θ) θ) is the fitting parameter of the line segment, T is a non-negative integer, determined by the prior knowledge of the scene in the image;

Step 374: Determine the line segment in the image jointly by (ρ, θ) and (x, y), and connect the broken parts to obtain a straight line segment.

Specific embodiment three, this embodiment is specifically described in conjunction with FIG. 2 , the difference between this embodiment and specific embodiment two is that the optimal threshold method for obtaining the gray value corresponding to the red component value described in step three or three is:

The four pixels on the top, bottom, left, and right adjacent to a pixel on the image and its four diagonal adjacent pixels constitute the neighborhood information of the eight pixels of the pixel, and the image is obtained by the two-dimensional maximum entropy segmentation method. The two-dimensional histogram of the gray value of the point and the gray value of the neighborhood information of the pixel, using the maximum two-dimensional entropy to obtain the optimal threshold f(x, y) of the gray value corresponding to the value of the red component;

Among them, (x, y) is the coordinate of the pixel point,

Areas A and B distributed along the diagonal of the two-dimensional histogram represent the target and background, respectively, and areas C and D away from the diagonal represent the boundary and noise. On areas A and B, use point grayscale-area gray The degree-mean two-dimensional maximum entropy method determines the optimal threshold, which can maximize the amount of information that truly represents the target and background. The two-dimensional entropy is:

$\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$

The discriminant function of entropy is:

f(x ₀ ,y ₀ )=log ₂ [P _A (1-P _A )]+H _A /P _A +(H _L -H _A )/(1-P _A ), (2)

The optimal threshold is:

f(x,y)=max{f(x ₀ ,y ₀ )}, (3)

in,

${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; A</span>}}\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\overset{\mathrm{<span\; class="notranslate">\; t</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; A</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\overset{\mathrm{<span\; class="notranslate">\; t</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$

${\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; L</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$

The number of gray levels of the image is L, the total number of pixels is N (m×n), g _{i, j} is the number of pixels whose gray level is i in the image and the average gray level of the area is j, p _{i, j} is the probability of occurrence of the pair (i, j) of point gray level - area gray level mean value, that is: p _{i, j} = g _{i, j} /N, where N (m×n) is the total number of pixels of the image, p _{i, j} , {i, j=1, 2, ..., L} represent the two-dimensional histogram of the image about the point gray level - the average value of the area gray level.

Embodiment 4. The difference between this embodiment and Embodiment 2 is that the specific steps for edge detection according to the binary image in steps 3 and 5 are:

Step 351, use the Gaussian filter to convolve the optimal threshold f(x, y) of the image:

g(x,y)=h(x,y,σ)*f(x,y) (7)

$\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&sigma;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;\&sigma;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$

Among them, g(x,y) is the convolved image, h(x,y,σ) is the Gaussian filter function, σ is the standard deviation, and * represents convolution;

Step 352, use the first-order partial derivative finite difference method to obtain the local gradient M(x,y) and edge direction θ(x,y) of the convolved image g(x,y),

g' _x (x,y)≈G _x (x,y)=[g(x+1,y)-g(x,y)+g(x+1,y+1)-g(x,y +1)]/2 (9)

g' _y (x,y)≈G _y (x,y)=[g(x,y+1)-g(x,y)+g(x+1,y+1)-g(x+1 ,y)]/2 (10)

$\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\sqrt{{\mathrm{<span\; class="notranslate">\; G</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; G</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$

θ(x,y)=arctan(G _x (x,y)/G _y (x,y)) (12)

Among them, G _x (x, y) represents the horizontal gradient of the image g(x, y); G _y (x, y) represents the vertical gradient of the image g(x, y); M(x, y) represents the image g The local gradient of (x,y); θ(x,y) represents the edge direction of the image g(x,y),

Step 353: Comparing the local gradient M(x,y) of the neighborhood information on each pixel of the image with the two pixels along the gradient line, if the local gradient M(x,y) of the neighborhood information is smaller than If the pixel along the gradient line, set M(x,y) to zero, the gradient value is an edge point; if the local gradient M(x,y) of the neighborhood information is greater than or equal to the pixel along the gradient line, the gradient value is not Edge points will not be processed;

Step 354, select two thresholds of the two-dimensional histogram, and act on the image obtained by non-maximum suppression, set the gray value of the pixel whose gradient value is smaller than the smaller threshold to 0, and obtain image 1; then Set the gray value of the pixel whose gradient value is smaller than the larger threshold to 0 to obtain image 2; based on image 2, use image 1 as a supplement to link to obtain the edge of the cross spot in the image.

Embodiment 5. The difference between this embodiment and Embodiment 4 is that the specific steps for linking and obtaining the edge of the cross spot in the image in steps 3, 5, and 4 are:

Step 3541, image 2 is scanned, and when a pixel point with a non-zero gray value is encountered, the contour line starting from it is traced until the end point of the contour line;

Step 3542, investigate the neighborhood information of the 8 pixel points of the pixel corresponding to the end position of the contour line in image 1 in image 1, if there is a non-zero gray in the neighborhood of 8 pixel points of this point If the pixel point of degree value exists, it is included in image 2 as a new starting point, and then steps 3641 are repeated until neither image 1 nor image 2 can continue;

Step 3543, after completing the link to the contour line containing a certain pixel point, mark this contour line as visited;

Repeat step 3541, step 3542 and step 3543 until no new contour line can be found in image 2.

Embodiment 6. The difference between this embodiment and Embodiment 2 is that the skeleton curves described in steps 3 and 6 are:

ρ=xcosθ+ysinθ; (13)

Among them, ρ=xcosθ+ysinθ indicates that a point (x, y) in the image space corresponds to a sinusoidal curve in (ρ, θ) space, x indicates the abscissa of the pixel, y indicates the ordinate of the pixel, and the origin of the image coordinates reaches the line The distance is ρ, and the angle between the normal of the line and the x-axis is θ.

Specific Embodiment 7. This embodiment is specifically described in conjunction with FIGS. 3 , 4 , 5 and 6 , based on the welding positioning of the weld seam of the visual navigation system of the wall-climbing robot for weld seam detection described in Embodiment 2 The welding seam offset acquisition method of the method, after the welding seam positioning, extracts the two endpoints of the welding seam arc and the three feature points of the center point of the crosslight, the distance between the two endpoints is a, and the center point to The distance from the adjacent endpoint is b, and the measured values of a and b are a', b', and the offset angle can be obtained as:

$\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arccos</span>}\frac{\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; b</span>}}{{\mathrm{<span\; class="notranslate">\; a</span>}}^{<span\; class="notranslate">\; \text{'}</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; b</span>}}^{<span\; class="notranslate">\; \text{'}</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arccos</span>}\frac{{\mathrm{<span\; class="notranslate">\; b</span>}}^{<span\; class="notranslate">\; \text{'}</span>}}{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span>$

The position offset is:

w=b'-b, (15)

When a'=a, b'=b, it is a normal situation, and the robot is moving normally; when a'>a, b'≤b, the robot shifts to the left; when a'>a, b'≥b, the robot right offset.

Figure 5 shows the situation where the robot is biased to the right. In Figure 6, a comparison is made between the offset state and the standard state. The included angle θ is the offset angle.

Whether it is a crawler wall-climbing robot or a wheeled wall-climbing robot, it is impossible for a sudden lateral displacement when crawling vertically, that is, the situation of a'=a&b'≥b cannot happen instantaneously. Therefore, Figure 5 can be equivalent to the result of Figure 6,

offset angle $\mathrm{\&theta;}=\arccos \frac{a+b}{a^{\text{'}}+b^{\text{'}}}=\arccos \frac{b^{\text{'}}}{b}$

When the robot crawls along the horizontal weld, it may slide down slightly due to the gravity factor. At this time, a'=a&b'≥b will appear, and the position offset w=b'-b.

The calculated w and θ are input to the PID control part of the robot as the feedback quantity, and the real-time correction of the running track can be realized.

Claims (6)

1. The welding positioning method of the welding seam of the visual navigation system of the wall-climbing robot used for welding seam detection, the method is realized by the following device, and the device includes a charge-coupled camera (1), a cross laser emitter (2) and computer, The charge-coupled camera (1) and the cross laser emitter (2) are fixed at the front end of the robot head, and the cross laser emitter (2) is directly above the charge-coupled camera (1), and the cross laser emitter ( 2) The emitted laser irradiates on the surface of the workpiece to be welded to form a cross spot. The charge-coupled camera (1) is used to photograph the cross spot on the surface of the workpiece, and the optical axis of the laser beam emitted by the cross laser emitter (2) is aligned with the charge-coupled camera ( 1) The included angle of the optical axis of the camera is 45°, and the data output end of the charge-coupled camera (1) is connected to the data input end of the computer; It is characterized in that the welding positioning method of the welding seam of the visual navigation system of the wall-climbing robot used for welding seam detection comprises the following steps: Step 1. The cross laser transmitter (2) emits a red cross-shaped laser beam. The laser beam is irradiated on the surface of the workpiece to be welded to form a cross spot, and the laser beam is used to scan the surface of the workpiece. During the scanning process, a charge-coupled camera is used to (1) Collect video information on the surface of the workpiece, and perform step 2; Step 2, the computer receives the video information collected by the charge-coupled camera; Step 3, the computer processes each frame of image in the video information as follows: Step 31, the frame image is divided into pixels, and the value of the pixel represents the value of the red, yellow and blue colors of the image at this point; Step 32, storing the values of all red, yellow and blue color components of the frame image in three arrays respectively; Step 33, the computer obtains the optimal threshold value of the gray value corresponding to the red component value by the two-dimensional maximum entropy segmentation method for the array of the red component corresponding to the frame image; Step three and four, compare the gray value of all pixels of the frame image with the optimal threshold obtained in step three and three, set the gray value of the pixel higher than the threshold to 255, and set the gray value of the pixel lower than the threshold Set the value to 0 to convert the frame image into a binary image; Steps 3 and 5, using the Canny operator to perform edge detection on the binary image obtained in steps 3 and 4, to obtain edge information of the binary image, which is shown as the edge of the cross spot on the frame image; Steps three and six, extracting the edge information of steps three and five, taking the midline of the edge of the cross spot as the skeleton, and obtaining a smooth skeleton curve with a single pixel width; Step 37: Carry out the Hough transformation of the straight line according to the skeleton curve obtained in step 36 to obtain two straight lines on both sides of the arc, mark the two endpoints of the arc, find the arc part, and detect the weld in the frame image position in . 2. the welding positioning method of the weld seam of visual navigation system according to claim 1, is characterized in that, the optimum threshold value method of obtaining the corresponding gray value of red component numerical value described in step 33 is: The four pixels on the top, bottom, left, and right adjacent to a pixel on the image and its four diagonal adjacent pixels constitute the neighborhood information of the eight pixels of the pixel, and the image is obtained by the two-dimensional maximum entropy segmentation method. The two-dimensional histogram of the gray value of the point and the gray value of the neighborhood information of the pixel, using the maximum two-dimensional entropy to obtain the optimal threshold f(x, y) of the gray value corresponding to the value of the red component; Among them, (x, y) is the coordinate of the pixel point, Areas A and B distributed along the diagonal of the two-dimensional histogram represent the target and background, respectively, and areas C and D away from the diagonal represent the boundary and noise. On areas A and B, use point grayscale-area gray The degree-mean two-dimensional maximum entropy method determines the optimal threshold, which can maximize the amount of information that truly represents the target and background. The two-dimensional entropy is: $\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 1</span>}<span\; class="notranslate">\; )</span>$ The discriminant function of entropy is: f(x ₀ ,y ₀ )=log ₂ [P _A (1-P _A )]+H _A /P _A +(H _L -H _A )/(1-P _A ), (2) The optimal threshold is: f(x,y)=max{f(x ₀ ,y ₀ )}, (3) in, ${\mathrm{<span\; class="notranslate">\; P</span>}}_{\mathrm{<span\; class="notranslate">\; A</span>}}\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\overset{\mathrm{<span\; class="notranslate">\; t</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 4</span>}<span\; class="notranslate">\; )</span>$ ${\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; A</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\overset{\mathrm{<span\; class="notranslate">\; the\; s</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\overset{\mathrm{<span\; class="notranslate">\; t</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 5</span>}<span\; class="notranslate">\; )</span>$ ${\mathrm{<span\; class="notranslate">\; h</span>}}_{\mathrm{<span\; class="notranslate">\; L</span>}}<span\; class="notranslate">\; =</span><span\; class="notranslate">\; -</span>\underset{\mathrm{<span\; class="notranslate">\; i</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}\underset{\mathrm{<span\; class="notranslate">\; j</span>}}{\overset{\mathrm{<span\; class="notranslate">\; L</span>}}{\mathrm{<span\; class="notranslate">\; \&Sigma;</span>}}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}{\mathrm{<span\; class="notranslate">\; log</span>}}_{\mathrm{<span\; class="notranslate">\; 2</span>}}{\mathrm{<span\; class="notranslate">\; p</span>}}_{\mathrm{<span\; class="notranslate">\; i</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; j</span>}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 6</span>}<span\; class="notranslate">\; )</span>$ The number of gray levels of the image is L, the total number of pixels is N (m×n), g _{i, j} is the number of pixels whose gray level is i in the image and the average gray level of the area is j, p _{i, j} is the probability of occurrence of the pair (i, j) of point gray level - area gray level mean value, that is: p _{i, j} = g _{i, j} /N, where N (m×n) is the total number of pixels of the image, p _{i, j} , {i, j=1, 2, ..., L} represent the two-dimensional histogram of the image about the point gray level - the average value of the area gray level. 3. the welding positioning method of the weld seam of visual navigation system according to claim 1, is characterized in that, the concrete steps that carry out edge detection according to binary image in step 35 are: Step 351, use the Gaussian filter to convolve the optimal threshold f(x, y) of the image: g(x,y)=h(x,y,σ)*f(x,y) (7) $\mathrm{<span\; class="notranslate">\; h</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; \&sigma;</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\frac{\mathrm{<span\; class="notranslate">\; 1</span>}}{{\mathrm{<span\; class="notranslate">\; 2</span>}\mathrm{<span\; class="notranslate">\; \&pi;\&sigma;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; e</span>}}^{<span\; class="notranslate">\; -</span>\frac{{\mathrm{<span\; class="notranslate">\; x</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; the\; y</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}{\mathrm{<span\; class="notranslate">\; 2</span>}{\mathrm{<span\; class="notranslate">\; \&sigma;</span>}}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 8</span>}<span\; class="notranslate">\; )</span>$ Among them, g(x,y) is the convolved image, h(x,y,σ) is the Gaussian filter function, σ is the standard deviation, and * represents convolution; Step 352, use the first-order partial derivative finite difference method to obtain the local gradient M(x,y) and edge direction θ(x,y) of the convolved image g(x,y), g' _x (x,y)≈G _x (x,y)=[g(x+1,y)-g(x,y)+g(x+1,y+1)-g(x,y +1)]/2 (9) g' _y (x,y)≈G _y (x,y)=[g(x,y+1)-g(x,y)+g(x+1,y+1)-g(x+1 ,y)]/2 (10) $\mathrm{<span\; class="notranslate">\; m</span>}<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span><span\; class="notranslate">\; =</span>\sqrt{{\mathrm{<span\; class="notranslate">\; G</span>}}_{\mathrm{<span\; class="notranslate">\; x</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; G</span>}}_{\mathrm{<span\; class="notranslate">\; the\; y</span>}}{<span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; x</span>}<span\; class="notranslate">\; ,</span>\mathrm{<span\; class="notranslate">\; the\; y</span>}<span\; class="notranslate">\; )</span>}^{\mathrm{<span\; class="notranslate">\; 2</span>}}}<span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 11</span>}<span\; class="notranslate">\; )</span>$ θ(x,y)=arctan(G _x (x,y)/G _y (x,y)) (12) Among them, G _x (x, y) represents the horizontal gradient of the image g(x, y); G _y (x, y) represents the vertical gradient of the image g(x, y); M(x, y) represents the image g The local gradient of (x,y); θ(x,y) represents the edge direction of the image g(x,y), Step 353: Comparing the local gradient M(x,y) of the neighborhood information on each pixel of the image with the two pixels along the gradient line, if the local gradient M(x,y) of the neighborhood information is smaller than If the pixel along the gradient line, set M(x,y) to zero, the gradient value is an edge point; if the local gradient M(x,y) of the neighborhood information is greater than or equal to the pixel along the gradient line, the gradient value is not Edge points will not be processed; Step 354, select two thresholds of the two-dimensional histogram, and act on the image obtained by non-maximum suppression, set the gray value of the pixel whose gradient value is smaller than the smaller threshold to 0, and obtain image 1; then Set the gray value of the pixel whose gradient value is smaller than the larger threshold to 0 to obtain image 2; based on image 2, use image 1 as a supplement to link to obtain the edge of the cross spot in the image. 4. the welding positioning method of the weld seam of visual navigation system according to claim 3, is characterized in that, the specific steps of linking the edge of the cross spot in the image obtained in step 354 are: Step 3541, image 2 is scanned, and when a pixel point with a non-zero gray value is encountered, the contour line starting from it is traced until the end point of the contour line; Step 3542, investigate the neighborhood information of the 8 pixel points of the pixel corresponding to the end position of the contour line in image 1 in image 1, if there is a non-zero gray in the neighborhood of 8 pixel points of this point If there is a pixel with a degree value, include it in Image 2 as a new starting point, and then repeat steps 3, 5, 4, and 1 until neither Image 1 nor Image 2 can continue; Step 3543, after completing the link to the contour line containing a certain pixel point, mark this contour line as visited; Repeat step 3541, step 3542 and step 3543 until no new contour line can be found in image 2. 5. the welding positioning method of the weld seam of visual navigation system according to claim 1, is characterized in that, the skeleton curve described in step three or six is: ρ=xcosθ+ysinθ; (13) Among them, ρ=xcosθ+ysinθ indicates that a point (x, y) in the image space corresponds to a sinusoidal curve in (ρ, θ) space, x indicates the abscissa of the pixel, y indicates the ordinate of the pixel, and the origin of the image coordinates reaches the line The distance is ρ, and the angle between the normal of the line and the x-axis is θ. 6. The seam offset acquisition method based on the welding positioning method of the seam according to claim 1 is characterized in that, after the seam is positioned, two endpoints of the arc of the seam and the central point three of the crosslight are extracted feature point, the distance between the two end points is a, the distance between the center point and the adjacent end point is b, and the measured values of a and b are a', b', and the offset angle can be obtained as : $\mathrm{<span\; class="notranslate">\; \&theta;</span>}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arccos</span>}\frac{\mathrm{<span\; class="notranslate">\; a</span>}<span\; class="notranslate">\; +</span>\mathrm{<span\; class="notranslate">\; b</span>}}{{\mathrm{<span\; class="notranslate">\; a</span>}}^{<span\; class="notranslate">\; \text{'}</span>}<span\; class="notranslate">\; +</span>{\mathrm{<span\; class="notranslate">\; b</span>}}^{<span\; class="notranslate">\; \text{'}</span>}}<span\; class="notranslate">\; =</span>\mathrm{<span\; class="notranslate">\; arccos</span>}\frac{{\mathrm{<span\; class="notranslate">\; b</span>}}^{<span\; class="notranslate">\; \text{'}</span>}}{\mathrm{<span\; class="notranslate">\; b</span>}}<span\; class="notranslate">\; ,</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; -</span><span\; class="notranslate">\; (</span>\mathrm{<span\; class="notranslate">\; 14</span>}<span\; class="notranslate">\; )</span>$ The position offset is: w=b'-b, (15) When a'=a, b'=b, it is normal, and the robot is moving normally; when a'>a, b'≤b, the robot shifts to the left; when a'>a, b'≥b, the robot moves toward right offset.