Brake disc machine vision defect detection method based on robot collaborative operation

Through robot collaborative operation and machine vision technology, efficient and precise defect detection and classification of automobile brake discs are achieved, the problem of insufficient efficiency and accuracy of traditional manual inspection is solved, and labor costs are reduced and intelligence is improved.

CN118616361BActive Publication Date: 2024-11-15SHANDONG LUCIDE MACHINERY
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202410894999.2
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2024-07-05
Publication Date
2024-11-15
Estimated Expiration
2044-07-05

AI Technical Summary

Technical Problem

Traditional manual inspections cannot meet the high efficiency and high accuracy requirements of automotive brake disc defect detection, resulting in high labor costs and low intelligence.

Method used

Using the brake disc machine vision defect detection method based on robot collaborative operation, the brake disc to be detected is loaded into the detection area through the robot, and the detection posture is adjusted, and the material is discharged to the corresponding defect classification area according to the detection results, combined with machine vision technology Defect detection.

Benefits of technology

It significantly reduces labor costs, improves the efficiency and intelligence of brake disc defect detection, and achieves efficient and accurate defect detection and classification.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118616361B_ABST
    Figure CN118616361B_ABST
Patent Text Reader

Abstract

The present invention provides a brake disc machine vision defect detection method based on robot collaborative operation, comprising: controlling a robot to load a brake disc to be detected to a detection area; based on machine vision, performing defect detection on the brake disc to be detected in the detection area; in the process of performing defect detection on the brake disc to be detected in the detection area, controlling the robot to adjust the detection posture of the brake disc to be detected in the detection area; after the defect detection of the brake disc to be detected in the detection area is completed, controlling the robot to unload the brake disc to be detected in the detection area to a corresponding defect classification area according to the defect detection result. The present invention greatly reduces the labor cost and improves the defect detection efficiency of the brake disc, and is also very intelligent.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the field of machine vision technology, and in particular to a brake disc machine vision defect detection method based on robot collaborative operation. Background Art

[0002] With the rapid advancement of science and technology and the continuous development of industrial production, quality control plays a vital role in the manufacturing industry. However, in large-scale production processes, defects and failures are inevitable, which have a negative impact on product quality and production efficiency. Traditional manual inspection cannot meet the requirements of high-efficiency and high-precision quality control, so machine vision defect detection technology came into being.

[0003] At present, automobile brake discs need to be inspected for defects after they are manufactured and before they leave the factory. Generally, when inspecting brake discs for defects, an inspection area is set up. The brake discs need to reach the inspection area before the machine vision inspection device can perform the corresponding defect inspection. Therefore, the brake discs need to be manually loaded to the inspection area. Secondly, after the brake discs arrive at the inspection area, since each three-dimensional surface of the brake disc needs to be inspected for defects, it is necessary to manually adjust the inspection posture of the brake discs to be inspected in the inspection area, that is, flip the brake discs, etc. In addition, after the defect inspection of the brake discs is completed, it is necessary to manually unload the brake discs to be inspected in the inspection area to the corresponding defect classification area to complete the classification of brake discs with different defects.

[0004] However, manually loading the brake discs to the inspection area, manually adjusting the inspection posture of the brake discs to be inspected in the inspection area, and manually unloading the brake discs to be inspected in the inspection area to the corresponding defect classification area will result in excessive labor costs, further reduce the efficiency of brake disc defect detection, and the degree of intelligence is seriously insufficient.

[0005] Therefore, a solution is urgently needed. Summary of the invention

[0006] One of the purposes of the present invention is to provide a machine vision defect detection method for brake discs based on robot collaborative operation, wherein the robot is controlled to load the brake disc to be inspected to the inspection area. During the process of defect detection of the brake disc to be inspected located in the inspection area, the robot is controlled to adjust the detection posture of the brake disc to be inspected located in the inspection area. After the defect detection of the brake disc to be inspected located in the inspection area is completed, the robot is controlled to unload the brake disc to be inspected located in the inspection area to the corresponding defect classification area according to the defect detection result, which greatly reduces the labor cost and improves the defect detection efficiency of the brake disc. At the same time, it is also very intelligent.

[0007] The brake disc machine vision defect detection method based on robot collaborative operation provided by the embodiment of the present invention includes:

[0008] Control the robot to load the brake disc to be inspected to the inspection area;

[0009] Based on machine vision, defect detection is performed on the brake disc to be inspected in the inspection area;

[0010] In the process of performing defect detection on the brake disc to be detected in the detection area, the robot is controlled to adjust the detection posture of the brake disc to be detected in the detection area;

[0011] After the defect detection of the brake disc to be inspected located in the detection area is completed, the control robot unloads the brake disc to be inspected located in the detection area to the corresponding defect classification area according to the defect detection result.

[0012] Preferably, before the control robot loads the brake disc to be inspected to the inspection area, the process further includes:

[0013] Control the robot to clean the foreign matter on the brake disc to be inspected.

[0014] Preferably, the defect detection of the brake disc to be detected located in the detection area based on machine vision includes:

[0015] Control the light source to illuminate the brake disc to be tested;

[0016] Acquire a brake disc image of the brake disc to be inspected by using an industrial camera;

[0017] Locating the brake disc area in the brake disc image;

[0018] Locate the working surface area in the brake disc area;

[0019] Pre-treat the work surface area;

[0020] Perform defect extraction on the preprocessing results.

[0021] Preferably, the locating the brake disc area in the brake disc image includes:

[0022] Based on the binarization algorithm, the brake disc area in the brake disc image is located.

[0023] Preferably, the working surface area in the positioning brake disc area includes:

[0024] The circle fitting algorithm based on the Canny operator is used to locate the working surface area in the brake disc area.

[0025] Preferably, the pre-processing of the working surface area includes:

[0026] The working surface area is processed by median filtering and image enhancement is performed using the Laplace operator.

[0027] Preferably, the extracting defects from the preprocessing results includes:

[0028] Based on the OTSU algorithm, the scratch area is extracted from the preprocessing results;

[0029] Detect scratch lines in the scratch area based on the improved Canny algorithm;

[0030] Based on the probabilistic Hough transform method, the scratch lines are connected at breakpoints to obtain the defect extraction results.

[0031] Preferably, it also includes:

[0032] Control the robot to detect the brake discs temporarily sent for inspection in the first area;

[0033] Plan the appropriate inspection area and inspection posture of the inspection object in the second area;

[0034] Control the robot to prompt the inspection object to enter the inspection area and prepare the brake disc for inspection in the inspection posture to wait for picking up and loading;

[0035] Control the robot to enter the inspection area, pick up the inspection object, prepare the brake disc for inspection in the inspection posture, and load it to the inspection area;

[0036] Wherein, the first area is a circular area formed by the first circular area and the second circular area; the second area is a second circular area;

[0037] The first circular area is a circular area on the ground with the current position of the robot as the center and a radius of the first radius value; the second circular area is a circular area on the ground with the current position of the robot as the center and a radius of the second radius value; the second radius value is smaller than the first radius value.

[0038] Preferably, the planning of a suitable inspection area and inspection posture of the inspection object in the second area includes:

[0039] Determine a semicircular area in the second area; the line between the current position of the inspection object and the current position of the robot bisects the semicircular area;

[0040] Control the robot to detect multiple free areas within the semicircular area;

[0041] Traversing each free area in turn, and predicting a first moving route of the inspection object from the current position of the inspection object to the traversed free area and a second moving route of the robot from the current position of the robot to the traversed free area during each traversal;

[0042] Determine a first interval position at every first interval value on the first moving route, and determine a second interval position at every second interval value on the second moving route;

[0043] Determine whether the vector angles between the i-th first target vector and the j-th second target vector are both greater than or equal to the vector angle threshold, and if so, use the traversed idle area as the target idle area;

[0044] After traversing each idle area, count the number of areas in the target idle area;

[0045] When the number of areas is unique, the target idle area is used as the inspection area;

[0046] When the number of areas is not unique, each target free area is traversed in turn, and each time the first moving time of the inspection object from the current position of the inspection object to the traversed target free area and the second moving time of the robot from the current position of the robot to the traversed target free area are predicted respectively;

[0047] Determine whether the second moving time is greater than the first moving time, and if so, use the second moving time as an associated value and associate it with the traversed target idle area;

[0048] After traversing each target free area, the target free area associated with the smallest association value is used as the inspection area;

[0049] The inspection posture is determined based on a second movement route of the robot from the current position of the robot to the inspection area.

[0050] Preferably, the absolute value of i minus j is less than or equal to a first absolute value threshold;

[0051] The i-th first target vector is a direction vector constructed based on the i-th first interval position on the first moving route and the straight line direction from the i-th first interval position to the m-th second interval position on the second moving route; the absolute value of m minus i is less than or equal to the second absolute value threshold;

[0052] The j-th second target vector is a direction vector constructed based on the j-th second interval position on the second moving route and the straight line direction from the j-th second interval position to the n-th first interval position on the first moving route; the absolute value of n minus j is less than or equal to the third absolute value threshold;

[0053] i, n = 1, 2, 3, ..., K1; K1 is the total number of first interval positions on the first moving route;

[0054] j,m=1,2,3,…,K2; K2 is the total number of second interval positions on the second moving route.

[0055] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures particularly pointed out in the written description, claims, and drawings.

[0056] The technical solution of the present invention is further described in detail below through the accompanying drawings and embodiments. BRIEF DESCRIPTION OF THE DRAWINGS

[0057] The accompanying drawings are used to provide a further understanding of the present invention and constitute a part of the specification. Together with the embodiments of the present invention, they are used to explain the present invention and do not constitute a limitation of the present invention. In the accompanying drawings:

[0058] Figure 1 Schematic diagram of a brake disc machine vision defect detection method based on robot collaborative operation in an embodiment of the present invention;

[0059] Figure 2 A schematic diagram of an implementation process of locating a brake disc area in a brake disc image in an embodiment of the present invention;

[0060] Figure 3 is a schematic diagram of a brake disc image after grayscale conversion in an embodiment of the present invention;

[0061] Figure 4 is a schematic diagram of a brake disc image after binarization processing in an embodiment of the present invention;

[0062] Figure 5 It is a schematic diagram of an implementation process of locating a working surface area in a brake disc area in an embodiment of the present invention;

[0063] Figure 6 Schematic diagram of double thresholds of a circle fitting algorithm based on the Canny operator in an embodiment of the present invention;

[0064] Figure 7 Schematic diagram of the extraction result of the working surface area in an embodiment of the present invention;

[0065] Figure 8 A schematic diagram of the steps of using the Canny edge detection operator in an embodiment of the present invention;

[0066] Fig. 9 Schematic diagram of defect detection results in an embodiment of the present invention. DETAILED DESCRIPTION

[0067] The preferred embodiments of the present invention are described below in conjunction with the accompanying drawings. It should be understood that the preferred embodiments described herein are only used to illustrate and explain the present invention, and are not used to limit the present invention.

[0068] The embodiment of the present invention provides a brake disc machine vision defect detection method based on robot collaborative operation, such as Figure 1 As shown, including:

[0069] S1. Control the robot to load the brake disc to be inspected to the inspection area;

[0070] S2. Based on machine vision, defect detection is performed on the brake disc to be inspected in the inspection area;

[0071] S3. During defect detection of the brake disc to be detected in the detection area, controlling the robot to adjust the detection posture of the brake disc to be detected in the detection area;

[0072] S4. After the defect detection of the brake disc to be detected in the detection area is completed, the robot is controlled to unload the brake disc to be detected in the detection area to the corresponding defect classification area according to the defect detection result.

[0073] A light source and an industrial camera are aligned with the inspection area. The light source is used to shoot fill light for the brake disc to be inspected in the inspection area. The industrial camera is used to shoot the brake disc image of the brake disc to be inspected in the inspection area. A robot is set up and controlled to load the brake disc to be inspected into the inspection area. At this time, the industrial camera shoots the brake disc image and performs a defect detection based on machine vision. When the detection is completed, the robot is notified to adjust the inspection posture of the brake disc to be inspected in the inspection area, that is, flip it once so that the unphotographed surface of the brake disc to be inspected falls into the shooting range of the industrial camera. The industrial camera shoots the brake disc image again and then performs defect detection, and so on, until the defect detection of each three-dimensional surface of the brake disc to be inspected is completed, so as to realize the coordinated operation of the defect detection of the machine vision and the robot. After the defect detection of the brake disc to be inspected in the inspection area is completed, the robot is controlled to unload the brake disc to be inspected in the inspection area to the corresponding defect classification area according to the defect detection result. The defect classification area can be set to an area such as no defect, scratch defect, etc.

[0074] The present application controls a robot to load the brake disc to be inspected into the inspection area. During defect inspection of the brake disc to be inspected in the inspection area, the control robot adjusts the inspection posture of the brake disc to be inspected in the inspection area. After the defect inspection of the brake disc to be inspected in the inspection area is completed, the control robot unloads the brake disc to be inspected in the inspection area to the corresponding defect classification area according to the defect inspection result, which greatly reduces the labor cost and improves the defect detection efficiency of the brake disc. At the same time, it is also very intelligent.

[0075] In one embodiment, before controlling the robot to load the brake disc to be inspected to the inspection area, the method further includes:

[0076] Control the robot to clean the foreign matter on the brake disc to be inspected.

[0077] Before the brake disc to be inspected is loaded, foreign matter may be attached to its surface. The robot can be controlled to clean the foreign matter to avoid defect detection errors caused by the foreign matter.

[0078] In one embodiment, the defect detection of the brake disc to be detected in the detection area based on machine vision includes:

[0079] Control the light source to illuminate the brake disc to be tested;

[0080] Acquire a brake disc image of the brake disc to be inspected by using an industrial camera;

[0081] Locating the brake disc area in the brake disc image;

[0082] Locate the working surface area in the brake disc area;

[0083] Pre-treat the work surface area;

[0084] Perform defect extraction on the preprocessing results.

[0085] Control the light source to illuminate the brake disc to be inspected to achieve fill light; the brake disc image contains not only the brake disc, but also the local environment of the brake disc; therefore, locate the brake disc area in the brake disc image, and then the working surface area in the brake disc area, pre-process the working surface area, and extract defects from the pre-processing results.

[0086] In one embodiment, locating the brake disc area in the brake disc image includes:

[0087] Based on the binarization algorithm, the brake disc area in the brake disc image is located.

[0088] The implementation process of locating the brake disc area in the brake disc image is as follows Figure 2 shown.

[0089] The weighted average method is used to grayscale the brake disc. The formula is as follows. Figure 3 shown.

[0090] Gray(i,j)=0.299R(i,j)+0.578G(i,j)+0.114B(i,j)

[0091] Among them, R(i,j), G(i,j), B(i,j) are the pixel values ​​of the red, green, and blue color channels respectively, and Gray(i,j) is the grayscale value.

[0092] Then use the threshold operator to binarize the image and obtain the outermost brake disc contour. Figure 4 shown.

[0093] In one embodiment, the working surface area in the positioning brake disc area includes:

[0094] The circle fitting algorithm based on the Canny operator is used to locate the working surface area in the brake disc area.

[0095] Accurately extracting the working surface can reduce the complexity of subsequent inspection and improve the inspection speed.

[0096] The process of locating the working surface area in the brake disc area is as follows Figure 5 shown.

[0097] The circle fitting algorithm based on the Canny operator is used to extract the working surface. The Canny operator is also called the optimal edge detection algorithm. It uses a double threshold method to determine the potential boundary, that is, setting an upper threshold and a lower threshold, such as Figure 6 shown.

[0098] Afterwards, due to uneven illumination and interference problems that cannot be solved by preprocessing of the extracted grayscale image itself, an accurate circle fitting algorithm helps to separate the working area from the non-working area. At present, random Hough transform circle detection, least squares circle fitting, random sampling consistent circle fitting, minimum circumscribed circle fitting, etc. are all commonly used circle fitting methods. Since there are fewer free interfering pixels on the edge of the disk, this application uses the minimum circumscribed circle fitting method to perform circle fitting on the threshold segmented image, mask processing, and extract the working surface.

[0099] This is a recursive randomization algorithm. The specific steps of the algorithm are as follows: Assume that there is a set of points {P1, P2, …, P n}, where P i =(x i ,y i ) represents a point on the plane. The center of the smallest circumscribed circle (x c ,y c ), with radius R. Then:

[0100]

[0101] The algorithm eventually finds the minimum circumscribed circle of the point set and returns the center and radius. It then selects the required circle based on the radius, sets the mask area, and locates and extracts the working surface of the brake disc. The extraction results are shown in Figure 1. Figure 7 shown.

[0102] In one embodiment, the preprocessing of the working surface area includes:

[0103] The working surface area is processed by median filtering and image enhancement is performed using the Laplace operator.

[0104] The Laplace operator is defined as:

[0105]

[0106] In the x-direction:

[0107]

[0108] In the y direction:

[0109]

[0110] So, the difference of the Laplace operator is approximately:

[0111]

[0112] In one embodiment, the step of extracting defects from the preprocessing results includes:

[0113] Based on the OTSU algorithm, the scratch area is extracted from the preprocessing results;

[0114] Detect scratch lines in the scratch area based on the improved Canny algorithm;

[0115] Based on the probabilistic Hough transform method, the scratch lines are connected at breakpoints to obtain the defect extraction results.

[0116] The segmented scratches are broken and need to be connected. Through experiments, it is found that connecting broken scratches through traditional morphological operations has a certain degree of erosion on the scratches and cannot connect the scratches well. Therefore, based on morphological operations, this application proposes to use an improved Canny operator for scratch detection, combine the probabilistic Hough transform method to connect scratches, and introduce bidirectional connections to shorten the detection time of the Canny algorithm.

[0117] The basic steps of the Canny edge detection operator are: first use a Gaussian filter to smooth the image; then use the finite difference of the first-order partial derivative to calculate the amplitude and direction of the gradient; then perform non-maximum suppression on the gradient amplitude; and finally use a double threshold algorithm to detect and connect edges. In the traditional Canny operator, a hysteresis threshold connection is used to connect strong edge pixels with adjacent weak edge pixels to form a continuous edge line. The implementation process is as follows Fig. 9 shown.

[0118] However, using only forward connections may result in incomplete connections of some edge lines. Therefore, the present application introduces an improvement of bidirectional connections in the Canny operator. In bidirectional connections, first starting from a strong edge pixel, it is connected to the adjacent weak edge pixel along the edge direction and marked as a strong edge pixel. Then, it continues to connect in the opposite direction starting from the new strong edge pixel to further enhance the continuity of the edge line. In this way, weak edge pixels can be connected from two directions to form a complete edge line, which improves the accuracy of Canny edge detection and reduces false positives.

[0119] Specifically: G x and G y are the gradients of image I in the x and y directions respectively, and the gradient is calculated using the Sobel operator, which can be expressed mathematically as follows: x =Sobel x (I); G y =Sobel y (I). The magnitude M and direction D of the gradient can be calculated by the following formula:

[0120]

[0121] D = arctan2(G y ,G x )

[0122] The gradient is calculated as M, D = calculate_gradient(I), and then non-maximum suppression is performed: S = non_max_suppression(M, D). The gradient strength of the current pixel is compared with the gradient strength of the adjacent pixels along the positive and negative gradient directions. If it is the largest, the pixel is retained as an edge point. If it is not the largest, it is suppressed. Then the high threshold and low threshold are defined for double threshold processing, trong_threshold = max(S) × threshold2;

[0123] weak_threshold=strong_threshold×threshold1, pixels with gradient strength lower than the low threshold are suppressed; otherwise, they are strong edges and are retained as edge points. Finally, bidirectional connection is implemented, edges[strong_i,strong_j]=255, if a weak edge pixel is adjacent to a strong edge pixel in its 3x3 neighborhood, it is promoted to a strong edge and added to strong_i and strong_j. edges[weak_i,weak_j]=100, if a weak edge pixel is adjacent to a pixel marked as 100 in its 3x3 neighborhood, it is marked as 100, which represents reverse connection, ensuring connectivity and robustness in the opposite direction. In addition, Canny edge detection involves computationally intensive tasks such as gradient calculation and non-maximum suppression. ThreadPoolExecutor is used to execute calculate_gradient and non_max_suppression in parallel. Parallel processing aims to speed up the entire process by executing these tasks in parallel, and fully utilizes the multi-core architecture to effectively utilize available resources.

[0124] The improved algorithm captures the information of scratches more comprehensively and effectively, reducing missed detections. This improvement has practical significance for the application of scratch detection, and can more reliably identify scratches, providing a more reliable basis for further applications.

[0125] Although the standard Hough transform can detect straight lines in an image, it needs to traverse the entire space and perform calculations at each point, which leads to high computational complexity. Therefore, this application adopts a new Hough transform, namely the probabilistic Hough transform.

[0126] D={d i ={x,y i}|i=1,2,…}

[0127] Randomly select a pair of feature points (d i ,d j ), where d i =(x i ,y i ), d j =(x j ,y j ), so the parameters (a, b) can be obtained by the following formula.

[0128]

[0129] The array H(a,b) is accumulated continuously according to the obtained results. When the calculation continues until a certain array exceeds a specified maximum value, the largest H(a,b) array is the detected straight line.

[0130] By proposing an improved Canny edge detection algorithm, introducing bidirectional connection, and combining the probabilistic Hough transform method to connect scratches, it can better handle complex images and broken edges, improve the accuracy and stability of scratch connection, and suppress the interference of the brake disc surface's own texture.

[0131] In addition, the present application introduces parallel processing based on the improvement of the Canny operator, fully utilizing the performance of multi-core processors to accelerate the efficiency of algorithm execution. It can be seen that the algorithm of the present application can better connect the fracture scratches and suppress the interference of other information. Fig. 9 The defect detection results are shown.

[0132] In one embodiment, the brake disc machine vision defect detection method based on robot collaborative operation also includes:

[0133] Control the robot to detect the brake discs temporarily sent for inspection in the first area;

[0134] Plan the appropriate inspection area and inspection posture of the inspection object in the second area;

[0135] Control the robot to prompt the inspection object to enter the inspection area and prepare the brake disc for inspection in the inspection posture to wait for picking up and loading;

[0136] Control the robot to enter the inspection area, pick up the inspection object, prepare the brake disc for inspection in the inspection posture, and load it to the inspection area;

[0137] Wherein, the first area is a circular area formed by the first circular area and the second circular area; the second area is a second circular area;

[0138] The first circular area is a circular area on the ground with the current position of the robot as the center and a radius of the first radius value; the second circular area is a circular area on the ground with the current position of the robot as the center and a radius of the second radius value; the second radius value is smaller than the first radius value.

[0139] A brake disc temporarily submitted for inspection refers to a brake disc that needs to be temporarily inspected for defects. For example, after a slight scratch appears on a brake disc, a staff member / grinding robot grinds the brake disc on site, and then the brake disc needs to be temporarily inspected for defects immediately. The object to be inspected is a staff member / grinding robot, etc. When the object to be inspected is located in the first area far away from the robot, a suitable inspection area and inspection posture are planned for it in the second area. The inspection area is an area suitable for the object to enter, and the inspection posture is a body posture suitable for the object to submit the brake disc to the robot for inspection. The robot is controlled to prompt the object to enter the inspection area and prepare the brake disc for inspection in the inspection posture for picking up and loading. When prompting, the robot can be controlled to project an indication aperture in the inspection area, and can also use voice broadcast to prompt it how to make the inspection posture. The robot is controlled to enter the inspection area to pick up the brake disc prepared by the object to be inspected in the inspection posture, and load it to the inspection area. The first radius value can be, for example, 20 meters; the second radius value can be, for example, 10 meters. Generally, when performing defect inspection on brake discs, there may be temporary brake discs for inspection on site. At this time, the temporary brake discs need to be queued up for immediate defect inspection. Normally, in this case, the control system needs to stop running, and the object of inspection places the temporary brake discs in the inspection area, and then starts the system, which is more cumbersome. Alternatively, the object of inspection can put the temporary brake discs into the brake discs waiting for loading, and wait for the robot to load and transport them, which is even more cumbersome. Secondly, the object of inspection cannot approach the robot directly, because the object of inspection does not know the future operation of the robot and may collide with the robot. The embodiments of the present invention can perfectly solve these problems: a first area far away from the robot is set to detect the inspection object in the first area, and a second area close to the robot is set to immediately determine the inspection area and the inspection posture in the second area when the inspection object is detected. At this time, the inspection object can enter the inspection area in the inspection posture according to the inspection area and inspection posture indicated by the robot, thereby avoiding direct approach to the robot that may cause a collision with the robot. The brake disc temporarily inspected can also be immediately picked up and loaded by the robot for defect detection, which greatly improves the applicability of the system.

[0140] In one embodiment, the planning of a suitable inspection area and inspection posture of the inspection object in the second area includes:

[0141] Determine a semicircular area in the second area; the line between the current position of the inspection object and the current position of the robot bisects the semicircular area;

[0142] Control the robot to inspect multiple free areas within the semicircular area; the free area refers to the currently free area within the semicircular area where there are no other inspection objects, no objects temporarily placed, and no machines or equipment occupied;

[0143] Traverse each idle area in turn, and during each traversal, respectively predict a first moving route of the inspection object from the current position of the inspection object to the traversed idle area and a second moving route of the robot from the current position of the robot to the traversed idle area; when predicting the first moving route and the second moving route, a three-dimensional model can be performed on the site, and the first moving route and the second moving route can be planned on the model after the modeling;

[0144] Determine a first interval position at a first interval value on the first moving route, and determine a second interval position at a second interval value on the second moving route; the first interval value may be, for example, 0.3 meters; the second interval value may be, for example, 0.15 meters;

[0145] Determine whether the vector angles between the i-th first target vector and the j-th second target vector are both greater than or equal to the vector angle threshold. If so, use the traversed free area as the target free area. The vector angle threshold may be, for example, 130 degrees. If the vector angles between the i-th first target vector and the j-th second target vector are both greater than or equal to the vector angle threshold, it means that if the object to be inspected moves to generate the i-th first target vector and the robot moves to generate the j-th second target vector, the object to be inspected can completely / closely face the direction of the robot's movement, and can know where the robot is going to move. The object to be inspected can immediately dodge in case of an accident / necessity.

[0146] After traversing each idle area, count the number of areas in the target idle area;

[0147] When the number of areas is unique, the target idle area is used as the inspection area. When the object to be inspected moves to the target idle area, the robot's future movement can be clearly understood during the movement, thereby improving the movement safety of the object to be inspected. Therefore, when the number of areas is unique, the target idle area is used as the inspection area.

[0148] When the number of areas is not unique, each target free area is traversed in turn. During each traversal, the first moving time of the inspection object from the current position of the inspection object to the traversed target free area and the second moving time of the robot from the current position of the robot to the traversed target free area are predicted respectively; when predicting the first moving time and the second moving time, the prediction can be made based on the historical moving speeds of the inspection object and the robot combined with the distance between their respective current positions and the target free area;

[0149] Determine whether the second moving time is greater than the first moving time. If so, use the second moving time as an associated value and associate it with the traversed target free area. When the second moving time is greater than the first moving time, the robot arrives at the target free area later than the inspection object, so that the inspection object can arrive at the target free area in advance and make an inspection gesture, waiting for the robot to pick up the brake disc for inspection, so as to avoid the situation where the robot arrives at the inspection object first and hands out the brake disc for inspection before arriving at the target free area, which does not meet the constraints of the inspection area and the inspection gesture and reduces safety.

[0150] After traversing each target free area, the target free area associated with the smallest correlation value is used as the inspection area; the smaller the second moving time is, the shorter the time it takes for the robot and the inspection object to load the brake disc for inspection, so the target free area associated with the smallest correlation value is used as the inspection area;

[0151] Based on the second moving route of the robot from the current position of the robot to the inspection area, the inspection posture is determined; the inspection posture can be determined according to the second moving route of the robot from the current position of the robot to the inspection area, for example: the last moving position and the last moving direction of the robot can be determined, and the inspection posture is facing the last moving position of the robot in the opposite direction of the last moving direction, waiting for the brake disc to be inspected;

[0152] The absolute value of i minus j is less than or equal to a first absolute value threshold; the first absolute value threshold may be, for example, 3;

[0153] The i-th first target vector is a direction vector constructed based on the i-th first interval position on the first moving route and the straight line direction from the i-th first interval position to the m-th second interval position on the second moving route; the absolute value of m minus i is less than or equal to the second absolute value threshold; the second absolute value threshold may be, for example, 2;

[0154] The j-th second target vector is a direction vector constructed based on the j-th second interval position on the second moving route and the straight line direction from the j-th second interval position to the n-th first interval position on the first moving route; the absolute value of n minus j is less than or equal to the third absolute value threshold; the third absolute value threshold may be, for example, 4;

[0155] i, n = 1, 2, 3, ..., K1; K1 is the total number of first interval positions on the first moving route;

[0156] j,m=1,2,3,…,K2; K2 is the total number of second interval positions on the second moving route.

[0157] By setting the relationship between i, j, m, and n, it can be ensured as much as possible that when the inspection object moves to the target idle area, the future movement of the robot can be clearly understood during the movement process, thereby improving the movement safety of the inspection object.

[0158] Obviously, those skilled in the art can make various changes and modifications to the present invention without departing from the spirit and scope of the present invention. Thus, if these modifications and variations of the present invention fall within the scope of the claims of the present invention and their equivalents, the present invention is also intended to include these modifications and variations.

Claims

1. Brake disc machine vision defect detection method based on robot collaborative operation, It is characterized in that include: Control the robot to load the brake disc to be inspected to the inspection area; Based on machine vision, defect detection is performed on the brake disc to be inspected in the inspection area; In the process of performing defect detection on the brake disc to be detected in the detection area, the robot is controlled to adjust the detection posture of the brake disc to be detected in the detection area; After the defect detection of the brake disc to be detected in the detection area is completed, the control robot unloads the brake disc to be detected in the detection area to the corresponding defect classification area according to the defect detection result; Also includes: Control the robot to detect the brake discs temporarily sent for inspection in the first area; Plan the appropriate inspection area and inspection posture of the inspection object in the second area; Control the robot to prompt the inspection object to enter the inspection area and prepare the brake disc for inspection in the inspection posture to wait for picking up and loading; Control the robot to enter the inspection area, pick up the inspection object, prepare the brake disc for inspection in the inspection posture, and load it to the inspection area; Wherein, the first area is a circular area formed by the first circular area and the second circular area; the second area is a second circular area; The first circular area is a circular area on the ground with the current position of the robot as the center and a radius length of the first radius value; the second circular area is a circular area on the ground with the current position of the robot as the center and a radius length of the second radius value; the second radius value is smaller than the first radius value; The planned inspection object's suitable inspection area and inspection posture in the second area include: Determine a semicircular area in the second area; the line between the current position of the inspection object and the current position of the robot bisects the semicircular area; Control the robot to detect multiple free areas within the semicircular area; Traversing each free area in turn, and predicting a first moving route of the inspection object from the current position of the inspection object to the traversed free area and a second moving route of the robot from the current position of the robot to the traversed free area during each traversal; Determine a first interval position at every first interval value on the first moving route, and determine a second interval position at every second interval value on the second moving route; Determine whether the vector angles between the i-th first target vector and the j-th second target vector are both greater than or equal to the vector angle threshold, and if so, use the traversed idle area as the target idle area; After traversing each free area, count the number of areas in the target free area; When the number of areas is unique, the target idle area is used as the inspection area; When the number of areas is not unique, each target free area is traversed in turn, and each time the first moving time of the inspection object from the current position of the inspection object to the traversed target free area and the second moving time of the robot from the current position of the robot to the traversed target free area are predicted respectively; Determine whether the second moving time is greater than the first moving time, and if so, use the second moving time as an associated value and associate it with the traversed target idle area; After traversing each target free area, the target free area associated with the smallest association value is used as the inspection area; The inspection posture is determined based on a second movement route of the robot from the current position of the robot to the inspection area.

2. The brake disc machine vision defect detection method based on robot collaborative operation as claimed in claim 1, It is characterized in that Before controlling the robot to load the brake disc to be inspected to the inspection area, it also includes: Control the robot to clean the foreign matter on the brake disc to be inspected.

3. The brake disc machine vision defect detection method based on robot collaborative operation as claimed in claim 1, It is characterized in that The method of performing defect detection on the brake disc to be detected in the detection area based on machine vision includes: Control the light source to illuminate the brake disc to be tested; Acquire a brake disc image of the brake disc to be inspected by using an industrial camera; Locating the brake disc area in the brake disc image; Locate the working surface area in the brake disc area; Pre-treat the work surface area; Perform defect extraction on the preprocessing results.

4. The brake disc machine vision defect detection method based on robot collaborative operation as claimed in claim 3, It is characterized in that The locating the brake disc area in the brake disc image includes: Based on the binarization algorithm, the brake disc area in the brake disc image is located.

5. The brake disc machine vision defect detection method based on robot collaborative operation as claimed in claim 3, It is characterized in that The working surface area in the positioning brake disc area includes: The circle fitting algorithm based on the Canny operator is used to locate the working surface area in the brake disc area.

6. The brake disc machine vision defect detection method based on robot collaborative operation as claimed in claim 3, It is characterized in that The pre-processing of the working surface area includes: The working surface area is processed by median filtering and image enhancement is performed using the Laplace operator.

7. The brake disc machine vision defect detection method based on robot collaborative operation as claimed in claim 3, It is characterized in that The defect extraction of the preprocessing result includes: Based on the OTSU algorithm, the scratch area is extracted from the preprocessing results; Detect scratch lines in the scratch area based on the improved Canny algorithm; Based on the probabilistic Hough transform method, the scratch lines are connected at breakpoints to obtain the defect extraction results.

8. The brake disc machine vision defect detection method based on robot collaborative operation as claimed in claim 1, It is characterized in that The absolute value of i minus j is less than or equal to the first absolute value threshold; The i-th first target vector is a direction vector constructed based on the i-th first interval position on the first moving route and the straight line direction from the i-th first interval position to the m-th second interval position on the second moving route; the absolute value of m minus i is less than or equal to the second absolute value threshold; The j-th second target vector is a direction vector constructed based on the j-th second interval position on the second moving route and the straight line direction from the j-th second interval position to the n-th first interval position on the first moving route; the absolute value of n minus j is less than or equal to the third absolute value threshold; i,n=1,2,3,…,K 1 ; K 1 is the total number of first interval positions on the first moving route; j,m=1,2,3,…,K 2 ; K 2 is the total number of second interval positions on the second moving route.