A robot repeatability accuracy detection method and device

Abstract

The application discloses a kind of robot repeated positioning precision detection method and device, by obtaining the first positioning image and the second positioning image of robot, wherein the top center area of robot is provided with first label and second label;First positioning image and the second positioning image are input into label detection model, to make label detection model output first label position and second label position in first positioning image, and third label position and fourth label position in second positioning image;Based on first label position, second label position, third label position and fourth label position, the positioning angle deviation and displacement pixel distance of first positioning and second positioning are calculated;Based on displacement pixel distance, obtain displacement precision, and based on positioning angle deviation, obtain deviation angle precision;Compared with prior art, the technical scheme of the application can improve the applicability and accuracy of robot repeated positioning detection.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of image recognition processing, in particular to a robot repeated positioning precision detection method and device. BACKGROUND

[0002] Robot repeated positioning precision detection refers to testing and analyzing the repeated movement of a robot to evaluate its precision and stability. With the increasing automation and intelligence of production, power industry gradually applies inspection robots, which greatly improves the inspection efficiency of lines. With the increasing number of robots, it is increasingly important to evaluate the repositioning precision of robots to ensure that they can accurately position and move when performing tasks. Insufficient repositioning precision may cause errors or delays in the performance of robots, thereby affecting production efficiency. Regularly evaluating the repositioning precision of robots can ensure that they can always perform tasks with the highest accuracy.

[0003] The power inspection robots currently applied in power grids include three types: transmission line inspection robots, substation inspection robots and power tunnel inspection robots. The types of robots applied in the future will be more diverse. Due to the different shapes, sizes and heights of robots produced by different manufacturers, it brings certain challenges to the measurement of the repositioning precision of robots. Therefore, there is an urgent need for a robot repeated positioning precision detection method that can be applied to robots produced by different manufacturers to improve the applicability and accuracy of robot repeated positioning detection. SUMMARY

[0004] The technical problem to be solved by the application is to provide a robot repeated positioning precision detection method and device to improve the applicability and accuracy of robot repeated positioning detection.

[0005] To solve the above technical problems, the application provides a robot repeated positioning precision detection method, which comprises the following steps:

[0006] When it is detected that the robot completes the first positioning, the first positioning image of the robot is acquired, and when it is detected that the robot completes the second positioning, the second positioning image of the robot is acquired, wherein a first label and a second label are arranged on the top center area of the robot;

[0007] The first positioning image and the second positioning image are input into a pre-trained label detection model, so that the label detection model outputs the first label position and the second label position in the first positioning image and the third label position and the fourth label position in the second positioning image;

[0008] calculate a positioning angle deviation and a displacement pixel distance between the first positioning and the second positioning based on the first label position, the second label position, the third label position and the fourth label position;

[0009] obtain a displacement accuracy based on the displacement pixel distance and obtain an angle deviation accuracy based on the positioning angle deviation.

[0010] In a possible implementation, the calculating of the positioning angle deviation and the displacement pixel distance between the first positioning and the second positioning based on the first label position, the second label position, the third label position and the fourth label position specifically includes:

[0011] calculate a first label center point coordinate based on the first label position, a second label center point coordinate based on the second label position, a third label center point coordinate based on the third label position and a fourth label center point coordinate based on the fourth label position;

[0012] obtain a first positioning vector according to the first label center point coordinate and the second label center point coordinate, and obtain a second positioning vector according to the third label center point coordinate and the fourth label center point coordinate;

[0013] calculate the positioning angle deviation between the first positioning and the second positioning according to the first positioning vector and the second positioning vector;

[0014] and calculate the displacement pixel distance between the first positioning and the second positioning according to the first label center point coordinate and the third label center point coordinate.

[0015] In a possible implementation, the calculating of the first label center point coordinate based on the first label position specifically includes:

[0016] obtain the first label position, wherein the first label position includes a first label top-left corner coordinate and a first label bottom-right corner coordinate;

[0017] input the first label top-left corner coordinate and the first label bottom-right corner coordinate into a preset label center point coordinate calculation formula to obtain a first label center point coordinate corresponding to the first label position;

[0018] wherein the preset label center point coordinate calculation formula is as follows:

[0019] centerx = (left + right) / 2; centery = (top + bottom) / 2;

[0020] In the formula, (centerx, centery) is the first label center point coordinate, (left, top) is the first label upper left corner coordinate, and (right, bottom) is the first label lower right corner coordinate.

[0021] In a possible implementation, the displacement precision is obtained based on the displacement pixel distance, and specifically includes:

[0022] The displacement pixel distance is substituted into a preset real displacement pixel distance calculation formula to calculate a real displacement pixel distance, and the real displacement pixel distance is taken as the displacement precision.

[0023] The real displacement pixel distance calculation formula is as follows:

[0024] dist_real = (labels_dist_real / labels_dist_pix) * dist_pix.

[0025] In the formula, dist_real is the real displacement pixel distance, labels_dist_real is the actual label center distance, labels_dist_pix is the label center distance, and dist_pix is the displacement pixel distance.

[0026] In a possible implementation, when it is detected that the robot completes first positioning, a first positioning image of the robot is acquired, and when it is detected that the robot completes second positioning, a second positioning image of the robot is acquired, and specifically includes:

[0027] A positioning area is set, when it is detected that the robot completes first positioning, the positioning area is first shot based on a camera to obtain a first positioning image;

[0028] When it is detected that the robot completes second positioning, the positioning area is second shot based on the camera to obtain a second positioning image.

[0029] In a possible implementation, the pre-training process of the label detection model specifically includes:

[0030] A plurality of positioning sample images of the robot are acquired, each positioning sample image is respectively marked with a first label rectangular frame and a second label rectangular frame to obtain a positioning sample image dataset;

[0031] A training sample dataset is divided from the positioning sample image dataset according to a preset proportion, positioning sample images in the training sample dataset are taken as model inputs, first label rectangular frame labels and second label rectangular frame labels in the training sample dataset are taken as model outputs, model training is performed on the initial label detection model until the model converges or a preset training number is reached, and a label detection model is obtained.

[0032] The application further provides a precision detection device for robot repeated positioning, comprising a positioning image acquisition module, a model detection module, a positioning data calculation module and a positioning precision determination module.

[0033] The positioning image acquisition module is configured to acquire a first positioning image of a robot when detecting that the robot completes first positioning, and acquire a second positioning image of the robot when detecting that the robot completes second positioning, wherein a top center region of the robot is provided with a first label and a second label.

[0034] The model detection module is configured to input the first positioning image and the second positioning image into a pre-trained label detection model, so that the label detection model outputs a first label position and a second label position in the first positioning image, and a third label position and a fourth label position in the second positioning image.

[0035] The positioning data calculation module is configured to calculate a positioning angle deviation and a displacement pixel distance between the first positioning and the second positioning based on the first label position, the second label position, the third label position and the fourth label position.

[0036] The positioning precision determination module is configured to obtain a displacement precision based on the displacement pixel distance, and obtain a deviation angle precision based on the positioning angle deviation.

[0037] In a possible implementation, the positioning data calculation module is configured to calculate the positioning angle deviation and the displacement pixel distance between the first positioning and the second positioning based on the first label position, the second label position, the third label position and the fourth label position, and specifically includes:

[0038] The first label position is used to calculate a first label center point coordinate, the second label position is used to calculate a second label center point coordinate, the third label position is used to calculate a third label center point coordinate, and the fourth label position is used to calculate a fourth label center point coordinate.

[0039] According to the first label center point coordinates and the second label center point coordinates, a first positioning vector is obtained, and according to the third label center point coordinates and the fourth label center point coordinates, a second positioning vector is obtained;

[0040] According to the first positioning vector and the second positioning vector, a positioning angle deviation between the first positioning and the second positioning is calculated;

[0041] And according to the first label center point coordinates and the third label center point coordinates, a displacement pixel distance between the first positioning and the second positioning is calculated.

[0042] In a possible implementation, the positioning data calculation module is configured to calculate first label center point coordinates based on the first label position, and specifically includes:

[0043] The first label position is obtained, wherein the first label position includes first label top-left corner coordinates and first label bottom-right corner coordinates;

[0044] The first label top-left corner coordinates and the first label bottom-right corner coordinates are input into a preset label center point coordinates calculation formula to obtain first label center point coordinates corresponding to the first label position;

[0045] The preset label center point coordinates calculation formula is as follows:

[0046] centerx = (left + right) / 2; centery = (top + bottom) / 2;

[0047] In the formula, (centerx, centery) is the first label center point coordinates, (left, top) is the first label top-left corner coordinates, and (right, bottom) is the first label bottom-right corner coordinates.

[0048] In a possible implementation, the positioning accuracy determination module is configured to obtain displacement accuracy based on the displacement pixel distance, and specifically includes:

[0049] The displacement pixel distance is substituted into a preset real displacement pixel distance calculation formula to calculate a real displacement pixel distance, and the real displacement pixel distance is taken as the displacement accuracy;

[0050] The real displacement pixel distance calculation formula is as follows:

[0051] dist_real = (labels_dist_real / labels_dist_pix) * dist_pix;

[0052] In the formula, dist real is a real displacement pixel distance, labels dist real is an actual label center distance, labels dist pix is a label center distance, and dist pix is a displacement pixel distance.

[0053] In a possible implementation, the positioning image acquisition module is configured to acquire a first positioning image of the robot when detecting that the robot completes the first positioning, and acquire a second positioning image of the robot when detecting that the robot completes the second positioning, and specifically includes:

[0054] setting a positioning area, performing first shooting on the positioning area based on the camera when detecting that the robot completes the first positioning, and obtaining the first positioning image;

[0055] performing second shooting on the positioning area based on the camera when detecting that the robot completes the second positioning, and obtaining the second positioning image.

[0056] In a possible implementation, the pre-training process of the label detection model specifically includes:

[0057] acquiring a plurality of positioning sample images of the robot, performing first label rectangular frame labeling and second label rectangular frame labeling on each positioning sample image respectively, and obtaining a positioning sample image dataset;

[0058] dividing a training sample dataset from the positioning sample image dataset according to a preset ratio, taking the positioning sample images in the training sample dataset as model inputs, taking the first label rectangular frame labeling and the second label rectangular frame labeling in the training sample dataset as model outputs, performing model training on an initial label detection model until the model converges or a preset training number is reached, and obtaining the label detection model.

[0059] The application further provides a terminal device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, and the processor implements the robot repeated positioning precision detection method according to any one of the above when executing the computer program.

[0060] The application further provides a computer readable storage medium including a stored computer program, wherein the computer readable storage medium controls a device where the computer readable storage medium is located to execute the robot repeated positioning precision detection method according to any one of the above when the computer program runs.

[0061] The robot repeated positioning precision detection method and device according to the embodiments of the application have the following beneficial effects compared with the prior art.

[0062] When it is detected that the robot completes the first positioning, a first positioning image of the robot is acquired, when it is detected that the robot completes the second positioning, a second positioning image of the robot is acquired, wherein a first label and a second label are arranged in a top center area of the robot; the first positioning image and the second positioning image are input into a pre-trained label detection model, so that the label detection model outputs a first label position and a second label position in the first positioning image, and a third label position and a fourth label position in the second positioning image; based on the first label position, the second label position, the third label position and the fourth label position, a positioning angle deviation and a displacement pixel distance of the first positioning and the second positioning are calculated; based on the displacement pixel distance, a displacement accuracy is obtained, and based on the positioning angle deviation, an angle deviation accuracy is obtained; compared with the prior art, the technical scheme of the present application can realize the detection of the repeated positioning accuracy of robots of different manufacturers and different shapes by arranging a first label and a second label in the top center area of the robot, and performing label recognition and positioning data calculation on the positioning image obtained by each positioning based on the label detection model, improve the applicability of the detection, and at the same time, when the repeated positioning accuracy of the robot is detected, the detection is performed from two aspects of the angle deviation accuracy and the displacement accuracy, and the accuracy of the detection is improved. BRIEF DESCRIPTION OF DRAWINGS

[0063] Figure 1 is a flowchart of an embodiment of a robot repeated positioning accuracy detection method provided by the present application;

[0064] Figure 2 is a structural schematic diagram of an embodiment of a robot repeated positioning accuracy detection device provided by the present application;

[0065] Figure 3 is a shape schematic diagram of a first label and a second label of an embodiment provided by the present application;

[0066] Figure 4 is a model monitoring output schematic diagram of a first positioning image and a second positioning image of an embodiment provided by the present application;

[0067] Figure 5 is a positioning vector schematic diagram of an embodiment provided by the present application. DETAILED DESCRIPTION

[0068] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0069] Example 1, see Figure 1 , Figure 1 This is a flowchart illustrating an embodiment of the accuracy detection method for repetitive robot positioning provided by the present invention, as shown below. Figure 1 As shown, the method includes steps 101-104, as detailed below:

[0070] Step 101: When the robot completes the first positioning, the first positioning image of the robot is acquired. When the robot completes the second positioning, the second positioning image of the robot is acquired. The top center area of ​​the robot is provided with a first label and a second label.

[0071] In one embodiment, a positioning area is set. When the robot completes the first positioning, the positioning area is photographed for the first time using a camera to obtain a first positioning image. When the robot completes the second positioning, the positioning area is photographed for the second time using the camera to obtain a second positioning image.

[0072] Specifically, when testing the positioning accuracy of a robot indoors, a positioning area is set at the positioning endpoint, and a camera is installed on top of the positioning area, with the camera pointing vertically downwards to capture the image so that the camera's view covers the positioning endpoint and the positioning area it occupies. When the robot performs a positioning operation and moves to the positioning endpoint, i.e., the positioning area, the camera can capture the robot's image.

[0073] In one embodiment, to adapt to various robot situations, a uniform self-adhesive label is affixed to the top of the robot for AI target detection. Specifically, a first label and a second label are set in the central area of ​​the top of the robot, wherein the first label and the second label are circular labels; preferably, the first label is a red circular label, the second label is a blue circular label, and the center of the label is a white circle with a radius of 2mm. Similarly, the circular label can also be other colors; for example... Figure 3 As shown, Figure 3 This is a schematic diagram of the shapes of the first and second labels.

[0074] In an embodiment, when it is detected that the robot completes the first positioning, if the robot reaches the positioning area, a first positioning image of the robot is captured based on the camera; when it is detected that the robot completes the first positioning, if the robot does not reach the positioning area, it is considered that the current detection accuracy deviation is too large, and the current accuracy detection process is ended based on the fact that the camera cannot capture the first positioning image of the robot.

[0075] In an embodiment, the camera is connected to the front-end controller and the algorithm server through a router or a switch, respectively.

[0076] In an embodiment, the front-end controller includes a front-end display interface, and the front-end display interface includes an image display area and an image acquisition button area, wherein the image display area is used to display the robot positioning image acquired by the camera, and the image acquisition button area is used to receive the image acquisition instruction of the on-site staff; when it is detected that the image acquisition button area is triggered, the positioning image currently captured by the camera is acquired, and the acquired positioning image is saved.

[0077] In an embodiment, the same positioning endpoint is set for the two positioning operations of the robot, so that the robot positions the same target in the two positioning operations; after it is determined that the first positioning image and the second positioning image are acquired, the first positioning image and the second positioning image are sent to the algorithm server, so that the positioning accuracy of the robot is detected based on the first positioning image and the second positioning image by the algorithm server.

[0078] Step 102: input the first positioning image and the second positioning image into the pre-trained label detection model, so that the label detection model outputs the first label position and the second label position in the first positioning image, and the third label position and the fourth label position in the second positioning image.

[0079] In an embodiment, the pre-trained label detection model is stored in the algorithm server, and when the first positioning image and the second positioning image are sent to the algorithm server, the first positioning image and the second positioning image are input into the pre-trained label detection model.

[0080] In an embodiment, the pre-training process of the label detection model is performed by obtaining a plurality of positioning sample images of the robot, performing first label rectangular frame labeling and second label rectangular frame labeling on each positioning sample image respectively to obtain a positioning sample image dataset; a training sample dataset is divided from the positioning sample image dataset according to a preset ratio; the positioning sample images in the training sample dataset are used as model input, the first label rectangular frame labeling and the second label rectangular frame labeling in the training sample dataset are used as model output, and model training is performed on an initial label detection model until the model converges or a preset training number is reached, thereby obtaining the label detection model.

[0081] Specifically, in the plurality of positioning sample image acquisition stage of the robot, the camera simulates the actual environment and photographs the robot from various angles to obtain positioning sample images of different angles, wherein each positioning sample image contains a first label and a second label, and the plurality of positioning sample images are saved to a designated image directory, such as / dataset / image. Preferably, the number of the plurality of positioning sample images is more than 1000.

[0082] Specifically, in the label rectangular frame labeling stage, the first label in each positioning sample image is labeled with a first label rectangular frame, and the second label in each positioning sample image is labeled with a second label rectangular frame; preferably, the bbox of the red circular label and the bbox of the blue circular label in the positioning sample image are manually labeled, and a yolo label format dataset is generated, with each picture xxx.jpg corresponding to an annotation file xxx.txt, saved to a designated annotation directory, such as / dataset / labels.

[0083] Preferably, the yolo annotation format is as follows: the data meaning of each line is classification+bbox position representation 1, 0.503516, 0.3875, 0.0289063, 0.0527778, and 0, 0.542969, 0.497222, 0.028125, 0.05, wherein 1 and 0 are classifications, and the bbox contains the upper left label coordinates and the lower right label coordinates of the label.

[0084] Specifically, in the dividing stage of the positioning sample image dataset, the training sample dataset, the test sample dataset and the validation sample dataset are divided from the positioning sample image dataset according to a preset ratio; wherein the preset ratio is 7:2:1.

[0085] In an embodiment, Yolo5 is selected as an initial label detection model. When setting the model parameters of the initial label detection model, the number of classifications is set to 2, when the classification is 0, it corresponds to a red circular label, and when the classification is 1, it corresponds to a blue circular label. At the same time, the number of training rounds is set to 30 rounds.

[0086] In an embodiment, after the pre-training of the label detection model is completed, the first positioning image and the second positioning image are input into the label detection model, the first label and the second label in the first positioning image are detected based on the label detection model, and the first label position and the second label position are correspondingly obtained, and the first label and the second label in the second positioning image are detected based on the label detection model, and the third label position and the fourth label position are correspondingly obtained; as Figure 4 Figure 4 is shown, the model monitoring output schematic diagram of the first positioning image and the second positioning image; Figure 4 In an embodiment, t1 is the first positioning image photographed at t1, t2 is the second positioning image photographed at t2, t1 red is the first label rectangular frame label, that is, the first label position of the red circular label, t1 blue is the second label rectangular frame label, that is, the second label position of the blue circular label, t2 red is the third label rectangular frame label, that is, the third label position of the red circular label, and t2 blue is the fourth label rectangular frame label, that is, the fourth label position of the blue circular label.

[0087] Step 103: Based on the first label position, the second label position, the third label position and the fourth label position, the positioning angle deviation and the displacement pixel distance of the first positioning and the second positioning are calculated.

[0088] In an embodiment, based on the first label position, the first label center point coordinate is calculated, based on the second label position, the second label center point coordinate is calculated, based on the third label position, the third label center point coordinate is calculated, and based on the fourth label position, the fourth label center point coordinate is calculated.

[0089] Specifically, the first label position is obtained, wherein the first label position includes a first label upper left corner coordinate and a first label lower right corner coordinate; the first label upper left corner coordinate and the first label lower right corner coordinate are input into a preset label center point coordinate calculation formula to obtain the first label center point coordinate corresponding to the first label position.

[0090] The preset label center point coordinate calculation formula is as follows:

[0091] ​centerx = (left + right) / 2; centery = (top + bottom) / 2;

[0092] In the formula, (centerx, centery) is the first label center point coordinate, (left, top) is the first label upper left corner coordinate, and (right, bottom) is the first label lower right corner coordinate.

[0093] Specifically, the second label position is obtained, wherein the second label position includes a second label upper left corner coordinate and a second label lower right corner coordinate; the second label upper left corner coordinate and the second label lower right corner coordinate are input into a preset label center point coordinate calculation formula to obtain a second label center point coordinate corresponding to the second label position.

[0094] Specifically, the third label position is obtained, wherein the third label position includes a third label upper left corner coordinate and a third label lower right corner coordinate; the third label upper left corner coordinate and the third label lower right corner coordinate are input into a preset label center point coordinate calculation formula to obtain a third label center point coordinate corresponding to the first label position.

[0095] Specifically, the fourth label position is obtained, wherein the fourth label position includes a fourth label upper left corner coordinate and a fourth label lower right corner coordinate; the fourth label upper left corner coordinate and the fourth label lower right corner coordinate are input into a preset label center point coordinate calculation formula to obtain a fourth label center point coordinate corresponding to the fourth label position.

[0096] In an embodiment, a first positioning vector is obtained according to the first label center point coordinate and the second label center point coordinate, and a second positioning vector is obtained according to the third label center point coordinate and the fourth label center point coordinate; as shown in Figure 5 Figure 5 is a positioning vector schematic diagram, Figure 5 In the formula, t1 vec is the first positioning vector, and t2 vec is the second positioning vector.

[0097] Specifically, when the first positioning vector is obtained, the second label center point coordinate is subtracted from the first label center point coordinate to obtain a first center point coordinate difference vector, a first length of the first center point coordinate difference vector is calculated based on a Euclidean distance formula, and the first center point coordinate difference vector is divided by the first length to obtain a first point positioning vector.

[0098] ​Specifically, when the second position vector is obtained, the fourth label center point coordinate is subtracted from the third label center point coordinate to obtain a second center point coordinate difference vector, a second length of the second center point coordinate difference vector is calculated based on a Euclidean distance formula, and the second center point coordinate difference vector is divided by the second length to obtain the second position vector.

[0099] In an embodiment, a positioning angle deviation between the first positioning and the second positioning is calculated according to the first positioning vector and the second positioning vector.

[0100] Specifically, the first positioning vector and the second positioning vector are input into a preset positioning angle deviation calculation formula to obtain the positioning angle deviation between the first positioning and the second positioning, wherein the preset positioning angle deviation calculation formula is as follows:

[0101]

[0102] wherein, θ is the positioning angle deviation, t1 vec is the first positioning vector, t2 vec is the second positioning vector, |t1 vec is a module of the first positioning vector, |t2 vec is a module of the second positioning vector.

[0103] In an embodiment, a displacement pixel distance between the first positioning and the second positioning is calculated according to the first label center point coordinate and the third label center point coordinate.

[0104] Specifically, the first label center point coordinate and the third label center point coordinate are substituted into a preset displacement pixel distance calculation formula to obtain the displacement pixel distance between the first positioning and the second positioning; wherein the displacement pixel distance calculation formula is as follows:

[0105]

[0106] wherein, dist_pix is the displacement pixel distance, (x1, y1) is the first label center point coordinate, and (x2, y2) is the second label center point coordinate.

[0107] Step 104: a displacement precision is obtained based on the displacement pixel distance, and a deviation angle precision is obtained based on the positioning angle deviation.

[0108] In an embodiment, the displacement pixel distance is substituted into a preset real displacement pixel distance calculation formula to calculate a real displacement pixel distance, and the real displacement pixel distance is taken as the displacement precision.

[0109] In an embodiment, the real displacement pixel distance calculation formula is as follows:

[0110] dist_real=(labels_dist_real / labels_dist_pix)*dist_pix;

[0111] In the formula, dist_real is the real displacement pixel distance, labels_dist_real is the actual label center distance, labels_dist_pix is the label center distance, and dist_pix is the displacement pixel distance.

[0112] Preferably, the label center distance labels_dist_pix is the distance between the first label center point of the first label and the second label center point of the second label, which is calculated by the distance formula between two points.

[0113] Preferably, the actual label center distance labels_dist_real is a constant, which is measured when the label is designed; preferably, the actual label center distance is 38 mm.

[0114] In an embodiment, after the positioning angle deviation is obtained, the positioning angle deviation is directly used as the deviation angle precision between the first positioning of the robot and the second positioning.

[0115] In summary, the precision detection method for robot repetitive positioning provided by the embodiment of the present application can satisfy the detection needs of robots of different manufacturers and different shapes by detecting the label during repetitive positioning detection, thereby improving the applicability of repetitive positioning detection; and the process of repetitive positioning precision detection only involves a camera, a label, and a label detection model, so the equipment is simple, easy to operate, and low in detection cost; and the repetitive positioning precision detection of the robot is performed from two aspects of deviation angle precision and displacement precision, thereby improving the accuracy of detection.

[0116] Embodiment 2 is shown in Figure 2 , Figure 2 is a structural schematic diagram of an embodiment of a precision detection device for robot repetitive positioning provided by the present application, as shown in Figure 2 The device includes a positioning image acquisition module 201, a model detection module 202, a positioning data calculation module 203, and a positioning precision determination module 204, and specifically as follows:

[0117] The positioning image acquisition module 201 is configured to acquire a first positioning image of the robot when it is detected that the robot completes a first positioning, and acquire a second positioning image of the robot when it is detected that the robot completes a second positioning, wherein a top center region of the robot is provided with a first label and a second label.

[0118] The model detection module 202 is configured to input the first positioning image and the second positioning image into a pre-trained label detection model, so that the label detection model outputs a first label position and a second label position in the first positioning image, and a third label position and a fourth label position in the second positioning image.

[0119] The positioning data calculation module 203 is configured to calculate a positioning angle deviation and a displacement pixel distance between the first positioning and the second positioning based on the first label position, the second label position, the third label position and the fourth label position.

[0120] The positioning precision determination module 204 is configured to obtain a displacement precision based on the displacement pixel distance, and obtain a deviation angle precision based on the positioning angle deviation.

[0121] In an embodiment, the positioning data calculation module 203 is configured to calculate a positioning angle deviation and a displacement pixel distance between the first positioning and the second positioning based on the first label position, the second label position, the third label position and the fourth label position, specifically, calculate a first label center point coordinate based on the first label position, calculate a second label center point coordinate based on the second label position, calculate a third label center point coordinate based on the third label position, and calculate a fourth label center point coordinate based on the fourth label position; obtain a first positioning vector according to the first label center point coordinate and the second label center point coordinate, and obtain a second positioning vector according to the third label center point coordinate and the fourth label center point coordinate; calculate the positioning angle deviation between the first positioning and the second positioning according to the first positioning vector and the second positioning vector; and calculate the displacement pixel distance between the first positioning and the second positioning according to the first label center point coordinate and the third label center point coordinate.

[0122] In an embodiment, the positioning data calculation module 203 is configured to calculate a first label center point coordinate based on the first label position. Specifically, the first label position is obtained, wherein the first label position comprises a first label top-left corner coordinate and a first label bottom-right corner coordinate; and the first label top-left corner coordinate and the first label bottom-right corner coordinate are input into a preset label center point coordinate calculation formula to obtain the first label center point coordinate corresponding to the first label position.

[0123] In an embodiment, the preset label center point coordinate calculation formula is as follows:

[0124] centerx = (left + right) / 2; centery = (top + bottom) / 2;

[0125] In the formula, (centerx, centery) is the first label center point coordinate, (left, top) is the first label top-left corner coordinate, and (right, bottom) is the first label bottom-right corner coordinate.

[0126] In an embodiment, the positioning accuracy determination module 204 is configured to obtain a displacement accuracy based on the displacement pixel distance. Specifically, the displacement pixel distance is substituted into a preset real displacement pixel distance calculation formula to calculate a real displacement pixel distance, and the real displacement pixel distance is taken as the displacement accuracy.

[0127] In an embodiment, the real displacement pixel distance calculation formula is as follows:

[0128] dist_real = (labels_dist_real / labels_dist_pix) * dist_pix;

[0129] In the formula, dist_real is the real displacement pixel distance, labels_dist_real is an actual label center distance, labels_dist_pix is a label center distance, and dist_pix is the displacement pixel distance.

[0130] In an embodiment, the positioning image acquisition module 201 is configured to acquire a first positioning image of a robot when it is detected that the robot completes a first positioning, and acquire a second positioning image of the robot when it is detected that the robot completes a second positioning. Specifically, a positioning area is set, a first positioning image is obtained by performing a first shooting on the positioning area based on a camera when it is detected that the robot completes the first positioning, and a second positioning image is obtained by performing a second shooting on the positioning area based on the camera when it is detected that the robot completes the second positioning.

[0131] In an embodiment, the pre-training process of the label detection model includes: obtaining a plurality of positioning sample images of the robot, performing first label rectangular frame labeling and second label rectangular frame labeling on each positioning sample image respectively to obtain a positioning sample image dataset; dividing a training sample dataset from the positioning sample image dataset according to a preset proportion; taking the positioning sample images in the training sample dataset as model input, taking the first label rectangular frame labeling and the second label rectangular frame labeling in the training sample dataset as model output, performing model training on an initial label detection model until the model converges or a preset training number is reached to obtain the label detection model.

[0132] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described device can refer to the corresponding process in the foregoing method embodiments, which will not be repeated here.

[0133] It should be noted that the embodiments of the robot repeat positioning precision detection device described above are only illustrative, and the modules described as separate components can or can not be physically separated, and the components displayed as modules can or can not be physical units, that is, they can be located in one place or distributed on multiple network units. According to actual needs, part or all of the modules can be selected to achieve the purpose of the embodiment.

[0134] On the basis of the above-mentioned embodiments of the robot repeat positioning precision detection method, another embodiment of the present application provides a robot repeat positioning precision detection terminal device. The robot repeat positioning precision detection terminal device comprises a processor, a memory and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the robot repeat positioning precision detection method of any embodiment of the present application is realized.

[0135] For example, in this embodiment, the computer program can be divided into one or more modules, which are stored in the memory and executed by the processor to complete the present application. The one or more modules can be a series of computer program instruction segments that can complete a specific function, which are used to describe the execution process of the computer program in the robot repeat positioning precision detection terminal device.

[0136] The robot repeat positioning precision detection terminal device can be a desktop computer, a notebook, a palm computer and a cloud server, etc. The robot repeat positioning precision detection terminal device can include, but is not limited to, a processor and a memory.

[0137] The processor can be a central processing unit (CPU), and can also be other general-purpose processors, digital signal processors (DSP), application specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gates or transistor logic devices, discrete hardware components, etc. The general-purpose processor can be a microprocessor or the processor can also be any conventional processor. The processor is a control center of the robot repetitive positioning precision detection terminal device, and is connected with various parts of the robot repetitive positioning precision detection terminal device through various interfaces and lines.

[0138] The memory can be used to store the computer program and / or modules, and the processor realizes various functions of the robot repetitive positioning precision detection terminal device by running or executing the computer program and / or modules stored in the memory, and calling the data stored in the memory. The memory can mainly include a program storage area and a data storage area. The program storage area can store an operating system, at least one application program required by a function, etc.; and the data storage area can store data created according to the use of the mobile phone, etc. In addition, the memory can include a high-speed random access memory, and can also include a nonvolatile memory, such as a hard disk, a memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state memory devices.

[0139] On the basis of the above-mentioned embodiment of the robot repetitive positioning precision detection method, another embodiment of the present application provides a storage medium including a stored computer program, wherein when the computer program runs, the device where the storage medium is located is controlled to execute the robot repetitive positioning precision detection method of any one of the embodiments of the present application.

[0140] In this embodiment, the storage medium described above is a computer readable storage medium, the computer program includes computer program code, which can be in the form of source code, object code, executable files or some intermediate forms, etc. The computer readable medium can include any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), electrical carrier signal, telecommunication signal and software distribution medium, etc. It should be noted that the contents contained in the computer readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction, for example, in some jurisdictions, according to legislation and patent practice, the computer readable medium does not include electrical carrier signals and telecommunication signals.

[0141] To sum up, the present application provides a kind of robot precision detection method and device of repeated positioning, by obtaining the first positioning image and the second positioning image of robot, wherein the top center area of robot is provided with first label and second label;First positioning image and the second positioning image are input into label detection model, to make label detection model output first label position and second label position in first positioning image, and third label position and fourth label position in second positioning image;Based on first label position, second label position, third label position and fourth label position, the positioning angle deviation and displacement pixel distance of first positioning and second positioning are calculated;Based on displacement pixel distance, displacement precision is obtained, and based on positioning angle deviation, deviation angle precision is obtained;Compared with prior art, the technical scheme of the present application can improve the applicability and accuracy of robot repeated positioning detection.

[0142] The above only describes the preferred embodiments of the present application, and it should be noted that for ordinary skilled in the art, without departing from the technical principles of the present application, a number of improvements and replacements can be made, and these improvements and replacements should be considered as the protection scope of the present application.

Claims

1. A method for detecting accuracy of robot repeatability, characterized in that, The method comprises the following steps: When it is detected that the robot completes the first positioning, a first positioning image of the robot is acquired, and when it is detected that the robot completes the second positioning, a second positioning image of the robot is acquired, wherein a top center area of the robot is provided with a first label and a second label; The first positioning image and the second positioning image are input into a pre-trained label detection model, so that the label detection model outputs a first label position and a second label position in the first positioning image and a third label position and a fourth label position in the second positioning image; Based on the first label position, the second label position, the third label position and the fourth label position, a positioning angle deviation and a displacement pixel distance between the first positioning and the second positioning are calculated, specifically as follows: Based on the first label position, a first label center point coordinate is calculated, based on the second label position, a second label center point coordinate is calculated, based on the third label position, a third label center point coordinate is calculated, and based on the fourth label position, a fourth label center point coordinate is calculated; According to the first label center point coordinate and the second label center point coordinate, a first positioning vector is obtained, and according to the third label center point coordinate and the fourth label center point coordinate, a second positioning vector is obtained; According to the first positioning vector and the second positioning vector, a positioning angle deviation between the first positioning and the second positioning is calculated; And according to the first label center point coordinate and the third label center point coordinate, a displacement pixel distance between the first positioning and the second positioning is calculated; Based on the displacement pixel distance, a displacement accuracy is obtained, and based on the positioning angle deviation, an angle deviation accuracy is obtained.

2. The method of claim 1, wherein, Based on the first label position, a first label center point coordinate is calculated, specifically as follows: The first label position is acquired, wherein the first label position comprises a first label upper left corner coordinate and a first label lower right corner coordinate; The first label upper left corner coordinate and the first label lower right corner coordinate are input into a preset label center point coordinate calculation formula to obtain a first label center point coordinate corresponding to the first label position; The preset label center point coordinate calculation formula is as follows: centerx = (left + right) / 2; centery = (top + bottom) / 2; In the formula, (centerx, centery) is the first label center point coordinate, (left, top) is the first label upper left corner coordinate, and (right, bottom) is the first label lower right corner coordinate.

3. The method of claim 1, wherein, Based on the displacement pixel distance, a displacement accuracy is obtained, specifically as follows: The displacement pixel distance is substituted into a preset real displacement pixel distance calculation formula to calculate a real displacement pixel distance, and the real displacement pixel distance is taken as the displacement accuracy; The real displacement pixel distance calculation formula is as follows: dist_real = (labels_dist_real / labels_dist_pix) * dist_pix; In the formula, dist_real is the real displacement pixel distance, labels_dist_real is the actual label center distance, labels_dist_pix is the label center distance, and dist_pix is the displacement pixel distance.

4. The method of claim 1, wherein, When it is detected that the robot completes the first positioning, a first positioning image of the robot is acquired, and when it is detected that the robot completes the second positioning, a second positioning image of the robot is acquired, specifically comprising: A positioning area is set, and when it is detected that the robot completes the first positioning, the positioning area is first photographed based on the camera to obtain the first positioning image; When it is detected that the robot completes the second positioning, the positioning area is second photographed based on the camera to obtain the second positioning image.

5. The method of claim 1, wherein, The pre-training process of the label detection model specifically comprises: A plurality of positioning sample images of the robot are acquired, and each positioning sample image is labeled with a first label rectangular frame and a second label rectangular frame to obtain a positioning sample image dataset; A training sample dataset is divided from the positioning sample image dataset according to a preset ratio, the positioning sample images in the training sample dataset are used as model inputs, the first label rectangular frame labeling and the second label rectangular frame labeling in the training sample dataset are used as model outputs, an initial label detection model is trained until the model converges or a preset training number is reached, and a label detection model is obtained.

6. A device for detecting the accuracy of repeated positioning of a robot, characterized in that Comprise: A positioning image acquisition module, a model detection module, a positioning data calculation module, and a positioning precision determination module; The positioning image acquisition module is configured to acquire a first positioning image of the robot when it is detected that the robot completes the first positioning, and acquire a second positioning image of the robot when it is detected that the robot completes the second positioning, wherein a top center area of the robot is provided with a first label and a second label. The model detection module is configured to input the first positioning image and the second positioning image into a pre-trained label detection model, so that the label detection model outputs a first label position and a second label position in the first positioning image, and a third label position and a fourth label position in the second positioning image. The positioning data calculation module is configured to calculate a positioning angle deviation and a displacement pixel distance of the first positioning and the second positioning based on the first label position, the second label position, the third label position, and the fourth label position, specifically comprising: The first label center point coordinate is calculated based on the first label position, the second label center point coordinate is calculated based on the second label position, the third label center point coordinate is calculated based on the third label position, and the fourth label center point coordinate is calculated based on the fourth label position. According to the first label center point coordinate and the second label center point coordinate, a first positioning vector is obtained, and according to the third label center point coordinate and the fourth label center point coordinate, a second positioning vector is obtained; According to the first positioning vector and the second positioning vector, a positioning angle deviation between the first positioning and the second positioning is calculated; And according to the first label center point coordinate and the third label center point coordinate, a displacement pixel distance between the first positioning and the second positioning is calculated; The positioning accuracy determination module is configured to obtain a displacement accuracy based on the displacement pixel distance, and obtain an angle deviation accuracy based on the positioning angle deviation.

7. A terminal device, characterized by comprising: The computer readable storage medium comprises a computer program stored therein, wherein the computer program, when executed, controls a device in which the computer readable storage medium is located to perform the robot repeated positioning accuracy detection method according to any one of claims 1 to 5.

8. A computer-readable storage medium, characterized in that, The computer readable storage medium comprises a computer program stored therein, wherein the computer program, when executed, controls a device in which the computer readable storage medium is located to perform the robot repeated positioning accuracy detection method according to any one of claims 1 to 5.

Images