A trajectory planning method of a weld seam detection robot for a through-penetration

Abstract

The application discloses a kind of trajectory planning methods of through-piece weld seam detection robot, comprising: after robot enters biological shielding ring, global camera acquires through-piece image;Image is preprocessed, and three-dimensional coordinates of through-piece in world coordinate system are estimated based on image;Optimal path of robot detection through-piece is calculated using genetic algorithm, and the robot detects all through-pieces one by one from initial position to terminal position;After robot reaches the below of to-be-detected through-piece, local camera acquires to-be-detected through-piece image, and to-be-detected through-piece is detected by image preprocessing, combined with transformation matrix, adjusting the pose of robot.The application can meet the efficient autonomous inspection of nuclear power pressure vessel top cover detection robot.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of robot trajectory planning, in particular to a trajectory planning method of a penetration weld detection robot. BACKGROUND

[0002] With the continuous development of nuclear power technology, nuclear energy industry is gradually entering people's life. Nuclear energy, as an economic and clean new energy, is widely used in power generation, military and medical fields. The safety of nuclear power equipment operation has always been an important concern of nuclear energy development in various countries. The safety and reliability of the reactor pressure vessel top cover, as an important part of the nuclear reactor, is particularly important. The top cover and penetration are affected by temperature, pressure, thermal expansion and many other factors, which can easily induce stress corrosion cracks. Once the cracks and corrosion lead to nuclear leakage, not only will it cause significant economic losses, but also will threaten human nuclear safety. Therefore, it is necessary to collect relevant data by a robot for the staff to check and ensure safety.

[0003] In the existing technology, the staff aligns the penetration according to experience through the camera on the robot, and inaccurate positioning may cause accidental collision between the mechanical arm and the penetration, resulting in damage to the penetration. At the same time, there are a large number of penetrations in each top cover, which need to be recorded separately according to the number of each penetration, time and image for analysis and comparison. Unreasonable trajectory planning of the robot will greatly increase the workload and consume a lot of manpower and resources. SUMMARY

[0004] In view of the problems existing in the prior art, the embodiments of the present application provide a trajectory planning method of a penetration weld detection robot, so that the robot can independently position and align the penetration, and complete the detection task of the penetration with a reasonable trajectory.

[0005] To achieve the above-mentioned purpose, the present application provides a trajectory planning method of a penetration weld detection robot, comprising the following steps:

[0006] After the robot enters the biological shielding ring, the global camera collects the penetration image;

[0007] The image is preprocessed, and the three-dimensional coordinates of the penetration in the world coordinate system are estimated based on the image;

[0008] The genetic algorithm is used to calculate the optimal path of the robot for detecting the penetration, and the robot detects all penetrations one by one from the initial position to the terminal position;

[0009] After the robot reaches the position below the penetration to be detected, the local camera collects the image of the penetration to be detected, and through image preprocessing, the conversion matrix and translation vector of the camera coordinate system and the mechanical arm coordinate system are combined to adjust the pose of the robot and detect the penetration to be detected.

[0010] The global camera is installed on a carrying trolley, has a long working distance and a large working field of view, and is used for global observation and positioning.

[0011] According to the size M*N of the site and in combination with the working field of view X*Y of the global camera, a plurality of points are selected by calculation to shoot the through member, so as to ensure that the images shot by the camera fully cover the through member and prevent missing detection.

[0012] The collected global image is preprocessed, including: removing the noise interference of the through member image by using a guided filter, obtaining the profile of the through member in the image by using an edge detection operator, screening out the profiles that do not meet the conditions according to the elliptical concave-convex criterion, retaining the outer profile of the elliptical port of the through member, and estimating the center of the ellipse by using a weighted gray centroid method, and the calculation formula is:

[0013]

[0014] wherein (x c ,y c ) is the image pixel coordinate of the center of the ellipse, (u,v) is the image pixel point of the elliptical region, S is the elliptical region, I(u,v) is the gray value, and w(u,v) is the weight.

[0015] The three-dimensional coordinates of the center of the through member in the world coordinate system are calculated by calculating the relationship between the image pixel coordinate system and the world coordinate system, and the calculation formula is:

[0016]

[0017] wherein (x c ,y c ) is the image pixel coordinate system, K1 is the camera internal parameter, and K2 is the camera external parameter.

[0018] The genetic algorithm is used to calculate the optimal path of the robot for detecting the through member, including: determining the initial position and the terminal position of the robot, taking the through member number as the chromosome of the genetic algorithm, taking the detection sequence of the through member as the biological individual of the genetic algorithm, generating a cross individual by crossing the chromosomes of the biological individual, generating a mutation individual by mutating the chromosomes of the biological individual, calculating the fitness of all individuals, selecting the individual with the maximum fitness as the optimal individual, and obtaining the optimal detection path.

[0019] The crossing method of the chromosome is to select the parent by using a roulette method, exchange the chromosome fragments at the same position of the parent 1 and the parent 2, generate a mapping relationship of the exchanged chromosomes, replace the repeated chromosomes in the non-crossing part and the exchanged part according to the mapping relationship, and generate a cross individual; the mutation method of the chromosome is exchange mutation, no new chromosome is generated due to the through member number, and the positions of two chromosomes of the biological individual are randomly exchanged.

[0020] The individual fitness is the required time of the robot from the initial position to the end position, passing through all the through-pieces, and the calculation formula is:

[0021]

[0022] Wherein, p i is the position information of the i th through-piece, v max is the maximum speed of the robot arm, and v is the maximum speed of the robot arm in a path.

[0023] The local camera is installed on the robot arm, has a short working distance and a small working field of view, and is used for accurate alignment at a short distance.

[0024] The center of the circle is calculated based on the through-piece image, and the three-dimensional coordinates of the center of the through-piece in the camera coordinate system are calculated through the left and right camera projection matrices M1 and M2 and the pixel coordinates (x1, y1) and (x2, y2) of the center of the circle in the left and right cameras, and the calculation formula is:

[0025]

[0026] Wherein is an element of the left and right camera projection matrix, and (X, Y, Z) is the three-dimensional coordinates of the center of the through-piece in the camera coordinate system.

[0027] The detection of the to-be-detected through-piece is that the robot arm is lifted to the detection probe on the end effector by about 10mm above the through-piece in the vertical direction, the sixth axis of the robot arm is rotated to make the detection probe revolve around the to-be-detected through-piece, the detection probe shoots the weld seam image of the through-piece and the top cover, after the weld seam information is collected, the robot arm is lowered to the initial height in the vertical direction, and the detection of the next through-piece is prepared.

[0028] Before the trajectory planning of the robot arm, the working range of the robot arm in the through-piece plane also needs to be calculated, and the position reached by the carrying trolley under the condition of detecting all the through-pieces is calculated through the greedy algorithm combined with the through-piece coordinates.

[0029] Compared with the prior art, the application has the following advantages:

[0030] The application provides a trajectory planning method of a through-piece weld seam detection robot, which comprises the following steps: firstly, the through-pieces are positioned by using a global camera to obtain the rough positions of all the through-pieces; then, the time-optimal detection sequence of the through-pieces in the reachable range of the robot arm is calculated through a genetic algorithm; and after the alignment of the to-be-detected through-piece by using a local camera before detection, the detection of the through-pieces is completed.

[0031] The application adopts a visual guidance method to coarsely position and finely position the through-penetration under the pressure vessel top cover, and obtains the relative position of the through-penetration and the mechanical arm, thereby improving the positioning accuracy and positioning speed compared with manual positioning. The application adopts a genetic algorithm to sort the detection sequence of the through-penetration under the pressure vessel top cover, and obtains a time-optimal trajectory, thereby improving the detection efficiency. The nuclear power pressure vessel top cover detection robot realizes efficient autonomous inspection. BRIEF DESCRIPTION OF DRAWINGS

[0032] Figure 1 A method flowchart of one embodiment of a trajectory planning method of a through-penetration weld detection robot.

[0033] Figure 2 An industrial camera installation position schematic diagram of a trajectory planning method of a through-penetration weld detection robot.

[0034] Figure 3 An image information schematic diagram collected by a global camera of a trajectory planning method of a through-penetration weld detection robot.

[0035] Figure 4 An image information schematic diagram collected by a local camera of a trajectory planning method of a through-penetration weld detection robot.

[0036] Figure 5 A working state schematic diagram of a trajectory planning method of a through-penetration weld detection robot. DETAILED DESCRIPTION

[0037] In order to enable personnel in the technical field to better understand the scheme of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only some of the embodiments of the present application, not all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by a person of ordinary skill in the art without creative labor fall within the scope of protection of the present application.

[0038] The embodiment of the present application provides a trajectory planning method of a through-penetration weld detection robot, and solves the problem that the through-penetration weld detection robot cannot autonomously patrol.

[0039] Figure 1 A method flowchart of one embodiment of a trajectory planning method of a through-penetration weld detection robot is shown in Figure 1 , which includes: Figure 1

[0040] A. After the robot enters the biological shielding ring, the global camera collects the through-penetration image;

[0041] ​B. Preprocess the image, estimate the three-dimensional coordinates of the through-penetration in the world coordinate system based on the image;

[0042] C. Calculate the optimal path of the robot detecting the through-penetration by using a genetic algorithm, wherein the robot detects all through-penetrations one by one from the initial position to the terminal position;

[0043] D. After the robot reaches below the through-penetration to be detected, the local camera collects the image of the through-penetration to be detected, adjusts the pose of the robot by image preprocessing and combining the transformation matrix, and detects the through-penetration to be detected;

[0044] The application adopts a visual guidance method to perform coarse positioning and fine positioning on the through-penetration under the top cover of the pressure vessel, obtains the relative position of the through-penetration and the mechanical arm, and improves the positioning accuracy and positioning speed compared with manual positioning. The application adopts a genetic algorithm to sort the detection sequence of the through-penetration under the top cover of the pressure vessel, obtains the time-optimal trajectory, and thus improves the detection efficiency. The nuclear power pressure vessel top cover detection robot realizes efficient autonomous inspection.

[0045] The implementation process of the embodiment of the application will be described in detail as follows:

[0046] A. After the robot enters the biological shielding ring, the global camera collects the image of the through-penetration;

[0047] In specific implementation, an industrial camera is installed on the robot, and the industrial camera is installed at a position as shown in Figure 2 , wherein the global camera is installed on the trolley body to ensure that the images of multiple through-penetrations can be collected, and the image information collected by the global camera is as shown in Figure 3 . The local camera is installed on the end effector of the mechanical arm to ensure that a single through-penetration can be positioned, and the image information collected by the local camera is as shown in Figure 4 .

[0048] It should be noted that after the industrial camera is installed, the industrial camera is calibrated to obtain the transformation matrix of the intrinsic parameters, extrinsic parameters, and the pose of the end of the mechanical arm and the pose of the camera.

[0049] In specific implementation, according to the size MxN of the site, in combination with the working field XxY of the global camera, a plurality of points are selected by calculation to shoot the through-penetrations to ensure that the images shot by the camera fully cover the through-penetrations to prevent missed detection.

[0050] B. Preprocess the image, estimate the three-dimensional coordinates of the through-penetration in the world coordinate system based on the image;

[0051] In specific implementation, the following steps can be adopted:

[0052] B1. Remove the noise interference of the through-penetration image by using a guide filter;

[0053] It should be noted that the guided filtering needs a guide map, and the image of the through part itself is taken as the guide map.

[0054] B2. The edge detection operator preliminarily obtains the contour of the through part in the image;

[0055] B3. The contour that does not meet the condition is screened out according to the elliptical concave-convex criterion;

[0056] It should be noted that the contour that does not meet the condition is first screened out through the contour area, and the condition S min <S<S max is met, and then when the convexity feature of the contour is judged, the minimum bounding box of the contour is first found, the concave-convexity of the through part contour is represented by judging the concave-convexity of the bounding box, and the screening of the elliptical end surface contour is realized.

[0057] B4. The center of the ellipse is estimated by the weighted gray centroid method;

[0058] It should be noted that the thresholding operation is first performed on the through part contour image to obtain a binary image, and then the elliptical center coordinates are calculated through the formula, and the specific calculation formula is:

[0059]

[0060] where (x c ,y c ) is the image pixel coordinate of the center of the ellipse, (u, v) is the image pixel point of the elliptical region, S is the elliptical region, I(u, v) is the gray value, and w(u, v) is the weight.

[0061] C. The genetic algorithm is used to calculate the optimal path of the robot for detecting the through part, and the robot detects all the through parts one by one from the initial position to the terminal position;

[0062] It should be noted that in the case where the coordinates of the through part are known, the range of the mechanical arm at the working height is calculated, and through the greedy algorithm, the minimum number of stopping positions required by the carrying trolley under the condition of detecting all the through parts is calculated.

[0063] In specific implementation, the through part number is taken as the genetic algorithm chromosome, and the through part detection sequence is taken as the genetic algorithm biological individual. The crossover of the biological individual chromosome generates a crossover individual, the mutation of the biological individual chromosome generates a mutation individual, the fitness of all individuals is calculated, the individual with the maximum fitness is selected as the optimal individual, and the optimal detection path is obtained.

[0064] It should be noted that the individual fitness is the time required for the robot to pass all the through parts from the initial position to the terminal position, and the detection time of the through part is not included, and the calculation formula is:

[0065]

[0066] wherein, p i is the position information of the i-th through-piece number, v max is the maximum speed of the mechanical arm, v is the maximum speed that the mechanical arm can reach in a path.

[0067] It should be noted that the crossover method of the chromosome is to select the parent by using the roulette method, and since the through-piece number is determined, no new chromosome is generated, the chromosome fragments at the same position of parent 1 and parent 2 are exchanged, the exchanged chromosome is generated mapping relationship, according to the mapping relationship, the repeated chromosomes of the un-crossed part and the exchanged part are replaced, and the crossed individual is generated.

[0068] It should be noted that the mutation method of the chromosome is to randomly exchange the positions of two chromosomes of the biological individual.

[0069] D. After the robot reaches below the through-piece to be detected, the local camera collects the image of the through-piece to be detected, and through image preprocessing, the conversion matrix and translation vector of the camera coordinate system and the mechanical arm coordinate system are combined to adjust the pose of the robot and detect the through-piece to be detected.

[0070] It should be noted that the image preprocessing method is consistent with the above step B, and the conversion matrix and translation vector are obtained through the camera calibration in step A.

[0071] It should be noted that the trajectory for detecting the through-piece to be detected is: the mechanical arm rises in the vertical direction to the detection probe on the end effector by about 10mm above the through-piece, rotates the sixth axis of the mechanical arm to make the detection probe around the through-piece to be detected for one turn, the detection probe shoots the weld seam image of the through-piece and the top cover, after collecting the weld seam information, the mechanical arm descends in the vertical direction to the initial height, and is ready to detect the next through-piece.

[0072] The above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for part of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A trajectory planning method of a weld seam detection robot for a through-penetration, characterized by, The method comprises the following steps: A. After the robot enters the biological shielding ring, a global camera collects a through-piece image; B. The image is preprocessed, and three-dimensional coordinates of the through-piece in a world coordinate system are estimated based on the image; C. A genetic algorithm is used to calculate an optimal path of the robot for detecting the through-piece, and the robot detects all through-pieces one by one from an initial position to a terminal position; D. After the robot reaches below a through-piece to be detected, a local camera collects an image of the through-piece to be detected, the robot pose is adjusted through image preprocessing and a transformation matrix, and the through-piece to be detected is detected. In the step A, the global camera is installed on a bearing trolley, has a long working distance and a large working field of view, and is used for global observation and positioning; In the step B, the image preprocessing comprises the following steps: removing noise interference of the through-piece image by using a guide filter, acquiring a through-piece contour in the image by using an edge detection operator, screening out contours that do not meet the condition according to an elliptical concave-convex criterion, retaining an outer contour of an elliptical through-piece opening, and estimating an elliptical center; In the step B, the three-dimensional coordinates of the through-piece center in the world coordinate system are calculated according to a relationship between an image pixel coordinate system and the world coordinate system, and a calculation formula is as follows: ； wherein (u, v) is an image pixel coordinate system, is a camera intrinsic parameter, is a camera extrinsic parameter; In the step C, the genetic algorithm is used to calculate the optimal path of the robot for detecting the through-piece, and the method comprises the following steps: determining an initial position and a terminal position of the robot, taking through-piece numbers as genetic algorithm chromosomes, taking through-piece detection sequences as genetic algorithm biological individuals, generating a cross individual by crossing the biological individual chromosomes, generating a mutation individual by mutating the biological individual chromosomes, calculating all individual fitness, selecting an individual with the maximum fitness as an optimal individual, and obtaining a detection optimal path; In the step D, the local camera is installed on a mechanical arm, has a short working distance and a small working field of view, and is used for close-range accurate alignment; In the step D, the circle center is calculated based on the through-penetration image pose estimation through the following steps: 、 and the pixel coordinates of the circle center in the left and right cameras 、 The three-dimensional coordinates of the through-penetration circle center in the camera coordinate system are calculated according to the following formula: ； wherein are elements of the left and right camera projection matrices, and (X, Y, Z) are the three-dimensional coordinates of the center of the through-penetrating piece in the camera coordinate system; In the step D, a transformation matrix and a translation vector of a camera coordinate system and a mechanical arm coordinate system are combined; In the step D, the detection of the through-piece to be detected comprises the following steps: the mechanical arm is raised along a vertical direction to a detection probe on an end effector to exceed the through-piece by about 10 mm, the mechanical arm is rotated along a sixth axis to make the detection probe rotate around the through-piece to be detected, the detection probe shoots a weld seam image of the through-piece and a top cover, after the weld seam information is collected, the mechanical arm is lowered along the vertical direction to an initial height, and the next through-piece is prepared for detection.

2. The method of claim 1, wherein the genetic algorithm is used to calculate an optimal path for a robot to detect the penetrant, and The crossing method of the chromosomes is a roulette wheel method for selecting a parent, exchanging chromosome fragments at the same position of the parent 1 and the parent 2, generating a mapping relationship of the exchanged chromosomes, replacing repeated chromosomes of the un-crossed part and the exchanged part according to the mapping relationship, and generating a cross individual; the mutation method of the chromosomes is a swap mutation, no new chromosome is generated due to the through-piece number, and two chromosome positions of the biological individual are randomly exchanged.

3. The method of claim 2, wherein the genetic algorithm is configured to: The individual fitness is a required time of the robot from the initial position to the terminal position and passing through all through-pieces, and a calculation formula is as follows: ； wherein, is the position information for the i-th through-penetration, is the maximum speed of the robot arm, is the maximum speed of the robot arm over a path segment.

Images