Photographing-path adaptive planning method and apparatus for appearance inspection

Through the adaptive planning method of the photo path, the target photo points are screened and the network topology is trained using the self-organizing mapping algorithm, which solves the problem of unreasonable shooting path of the fly shooting device, and shortens the shooting time and improves the efficiency of the robot arm.

WO2024259731A1PCT designated stage expired Publication Date: 2024-12-26SHANGHAI MICROINTELLIGENCE CO LTD

Patent Information

Application Number
PCT/CN2023/102245
Authority / Receiving Office
WO · WO
Patent Type
Applications
Current Assignee / Owner
Priority Date
2023-06-20
Filing Date
2023-06-25
Publication Date
2024-12-26

AI Technical Summary

Technical Problem

The existing flying camera device has unreasonable shooting paths during the shooting process, resulting in the long shooting time of the robotic arm and low shooting efficiency.

Method used

Adaptive planning method for taking photos is adopted to filter out target photography points through inverse solution calculation of robotic arm, and train the target network topology using self-organization mapping algorithm to obtain the target network topology, plan the shortest target shooting path, and smooth the Bezier curve to optimize the path.

Benefits of technology

It reduces the shooting time of the robotic arm, improves the shooting efficiency, reduces the shooting consumption of the robotic arm, and ensures the speed and accuracy of image acquisition.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN2023102245_26122024_PF_FP_ABST
    Figure CN2023102245_26122024_PF_FP_ABST
Patent Text Reader

Abstract

A photographing-path adaptive planning method and apparatus for appearance inspection. The method comprises: respectively performing manipulator inverse solution calculation for several initial photographing point positions, so as to obtain an inverse solution calculation result of each initial photographing point position (S101); performing screening on the initial photographing point positions on the basis of the inverse solution calculation results of the initial photographing point positions, so as to obtain several target photographing point positions which have inverse solutions (S102); on the basis of each of the target photographing point position, performing training by using a self-organizing map algorithm, so as to obtain a target network topology structure which is used for representing the positional relationship among the target photographing point position (S103); and at least on the basis of the position sequence of the target photographing point positions in the target network topology structure, performing planning to obtain a target photographing path (S104). The method in the present application can rationally and accurately plan an optimal photographing path, such that the photographing time of a manipulator can be shortened, thereby improving the photographing efficiency.
Need to check novelty before this filing date? Find Prior Art

Description

A method and device for adaptive planning of photographing paths for appearance inspection Technical Field

[0001] The present invention relates to the technical field of industrial quality inspection, and in particular to a method, device, medium and equipment for adaptive planning of a photographing path for appearance inspection. Background Art

[0002] In the field of industrial visual inspection, detecting surface defects on workpieces requires multi-angle photography. However, traditional single, fixed camera locations often fail to meet the appearance inspection requirements of quality inspection systems. This is why flying camera devices have emerged. By utilizing these devices, a flying camera arm can be controlled to drive a camera to capture images in real-world scenarios, collecting data from multiple angles of the workpiece, ensuring the safety of subsequent appearance inspections.

[0003] However, the existing flying shooting device has the problem of unreasonable shooting path during the shooting process, which results in a long shooting time and low shooting efficiency of the robotic arm.

[0004] Summary of the Invention

[0005] In view of this, the present invention provides a method and device for adaptive planning of photographing paths for appearance inspection, the main purpose of which is to solve the problem that the shooting path of the flying shooting device is unreasonable during the shooting process, which in turn causes the robotic arm to take a long time and have low shooting efficiency.

[0006] To solve the above problems, the present application provides a method for adaptively planning a photographing path for appearance inspection, comprising:

[0007] Perform inverse calculations on the manipulator for several initial photographing points to obtain inverse calculation results for each initial photographing point;

[0008] Screening each of the initial photographing points based on the inverse solution calculation results of each of the initial photographing points to obtain a number of target photographing points for which an inverse solution exists;

[0009] Based on the target photographing points, a self-organizing map algorithm is used to train and obtain a target network topology structure for representing the positional relationship of each target photographing point;

[0010] A target shooting path is planned and obtained based at least on the position sequence of each target shooting point in the target network topology structure.

[0011] Optionally, the method further includes obtaining a number of initial photographing points, specifically including:

[0012] Based on the position information of the object to be detected, several initial shooting points are obtained at the end of the robotic arm.

[0013] Optionally, performing inverse solution calculations for several initial photographing points to obtain inverse solution calculation results for each initial photographing point specifically includes:

[0014] Based on the initial photographing points, determining the end-of-arm posture corresponding to each photographing point;

[0015] Based on the pre-established kinematic model of the robotic arm, the inverse solution calculation is performed on the end position of each robotic arm to obtain the robotic arm inverse solution corresponding to each robotic arm end position, so as to obtain the inverse solution calculation result of each initial shooting point.

[0016] Optionally, the method of using a self-organizing map algorithm to train the target network topology structure for characterizing the positional relationship of each target photographing point based on each target photographing point includes:

[0017] Based on each of the target photographing points, performing Euclidean distance calculation on any two photographing points respectively to obtain a number of Euclidean distances corresponding to each target photographing point;

[0018] Based on several Euclidean distances corresponding to each photo point, the positions of the network nodes corresponding to each photo point in the initial network topology are adjusted to train and obtain the target network topology structure used to characterize the positional relationship of each target photo point.

[0019] Optionally, the planning and obtaining of the target shooting path based at least on the position sequence of each target shooting point in the target network topology structure specifically includes:

[0020] Determine a subpath between any two adjacent target photographing points based on the position sequence of each target photographing point in the target network topology structure to obtain a plurality of subpaths;

[0021] The target shooting path is obtained based on the sub-path planning.

[0022] Optionally, determining a subpath between any two adjacent target photographing points specifically includes:

[0023] Perform collision detection on any two adjacent target photo points and obtain the detection results;

[0024] If the detection result shows that no collision occurs, determining, based on the two adjacent target photographing points, a linear path between the two target photographing points as a subpath;

[0025] If the detection result indicates a collision, any nonlinear path between the two adjacent target photo points is determined as a subpath based on the two adjacent target photo points. Alternatively, a number of substitute photo points are determined based on the adjacent target photo points, and a self-organizing map algorithm is used to train a subnetwork topology structure for characterizing the positional relationship of each substitute photo point based on each substitute photo point, and the nonlinear subpath is determined based on the positional sequence of each substitute photo point in the subnetwork topology structure.

[0026] Optionally, after obtaining the target path, the method further includes:

[0027] The target shooting path is smoothed based on the Bezier curve to obtain the target shooting path.

[0028] To solve the above problems, the present application provides a device for adaptively planning a photographing path for appearance inspection, comprising:

[0029] A calculation module is used to perform inverse solution calculations of the robotic arm for a number of initial photographing points, and obtain inverse solution calculation results for each initial photographing point;

[0030] A screening module is used to screen each initial photographing point based on the inverse solution calculation result of each initial photographing point to obtain a number of target photographing points for which an inverse solution exists;

[0031] A training module is used to train, based on each of the target photographing points, a self-organizing map algorithm to obtain a target network topology structure for characterizing the positional relationship between each of the target photographing points;

[0032] The planning module is used to plan and obtain a target shooting path based on at least the position sequence of each target shooting point in the target network topology structure.

[0033] To solve the above problems, the present application provides a storage medium, which stores a computer program. When the computer program is executed by a processor, it implements the steps of any of the above-mentioned methods for adaptive planning of photographing paths for appearance detection.

[0034] To solve the above problems, the present application provides an electronic device, which includes at least a memory and a processor, wherein a computer program is stored on the memory, and when the processor executes the computer program on the memory, it implements the steps of the adaptive planning method of photographing path for appearance detection described in any of the above items.

[0035] The adaptive planning method and device for photographing paths for appearance inspection in the present application obtain target photographing points for which there are inverse solutions of the robotic arm through screening, and adopt a self-organizing mapping algorithm to perform unsupervised training using each target photographing point, thereby obtaining a target network topology structure that characterizes the positional relationship of each target photographing point, that is, each photographing point is mapped to a node / neuron in the target network topology structure, thereby accurately determining the positional sequence between each target photographing point, and thus quickly and reasonably planning to obtain the shortest target photographing path containing each target photographing point, laying the foundation for subsequent control of the robotic arm movement based on the target photographing path to quickly capture images of the object to be inspected, reducing the robotic arm shooting time, improving the shooting efficiency, and reducing the robotic arm shooting consumption.

[0036] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0038] FIG1 is a flow chart of a method for adaptively planning a photographing path for appearance inspection according to an embodiment of the present application;

[0039] FIG2 is a structural block diagram of a device for adaptively planning a photographing path for appearance inspection according to another embodiment of the present application;

[0040] FIG3 is a structural block diagram of an electronic device according to another embodiment of the present application. DETAILED DESCRIPTION

[0041] Various aspects and features of the present application are described herein with reference to the accompanying drawings.

[0042] It should be understood that various modifications may be made to the embodiments of the present application. Therefore, the above description should not be considered as limiting, but merely as an example of an embodiment. Other modifications within the scope and spirit of the present application will occur to those skilled in the art.

[0043] The accompanying drawings, which are incorporated in and constitute a part of the specification, illustrate embodiments of the present application and, together with the general description of the present application given above and the detailed description of the embodiments given below, serve to explain the principles of the present application.

[0044] These and other characteristics of the present application will become apparent from the following description of a preferred form of embodiment given as a non-limiting example with reference to the accompanying drawings.

[0045] It should also be understood that although the present application has been described with reference to certain specific examples, those skilled in the art will readily be able to implement many other equivalent forms of the present application.

[0046] The above and other aspects, features and advantages of the present application will become more apparent from the following detailed description when taken in conjunction with the accompanying drawings.

[0047] Specific embodiments of the present application will be described hereinafter with reference to the accompanying drawings; however, it should be understood that the embodiments described are merely examples of the present application and may be implemented in a variety of ways. Familiar and / or repetitive functions and structures are not described in detail to avoid obscuring the present application with unnecessary or redundant details. Therefore, the specific structural and functional details described herein are not intended to be limiting, but rather serve merely as a basis and representative basis for the claims to teach those skilled in the art to variously utilize the present application with substantially any suitable detailed structure.

[0048] This specification may use the phrases "in one embodiment," "in another embodiment," "in yet another embodiment," or "in other embodiments," which may all refer to one or more of the same or different embodiments according to the present application.

[0049] The present application embodiment provides a method for adaptive planning of a photographing path for appearance inspection, which can be applied to electronic devices such as servers and terminals. As shown in FIG1 , the method in this embodiment includes the following steps:

[0050] Step S101, performing inverse solution calculation of the robot arm for each of the initial photographing points to obtain the inverse solution calculation results of each initial photographing point;

[0051] During the specific implementation of this step, several initial photographing points at the end of the robotic arm can be obtained in advance based on the position information of the object to be detected.

[0052] When performing the inverse solution calculation for the robotic arm, an inverse solution calculation is performed for each of the initial photographing points based on a pre-established kinematic model of the robotic arm to determine whether a corresponding robotic arm inverse solution exists for each of the photographing points. In other words, based on each of the initial photographing points, the robotic arm end position corresponding to each photographing point can be determined, and then an inverse solution calculation is performed on each robotic arm end position to obtain the inverse solution calculation result for the initial photographing point.

[0053] In the specific implementation process of this step, the process of finding the inverse solution is as follows: According to the DH (Denavit-Hartenberg) convention, the forward solution equation of the rigid manipulator is established, and θ i ,a i ,d i ,α i Represents the parameters related to link i and joint i, that is, θ i Indicates the connecting rod length, a i Indicates the connecting rod torsion, d i Indicates the link offset, α i Represents the joint angle. Homogeneous matrix A i A represents the transformation matrix of link i rotating around joint i, i It can be obtained by multiplying four basic transformations, as shown below:

[0054] In this step, the forward kinematic equation of the robotic arm can be expressed as Where Θ=[θ1, θ2, θ3,…θ n ] represents the angle value of the n joint axes of the robotic arm.

[0055] The problem of finding the inverse solution of the robotic arm can be expressed as a nonlinear least squares problem. The L2 norm of F(Θ)-Y is minimized by optimizing the angle value Θ of the joint axis. The optimization function is as follows:

[0056] Among them, Y is the homogeneous matrix representation of the end-arm pose corresponding to the photographing point.

[0057] Step S102: screening each of the initial photographing points based on the inverse solution calculation results of each of the initial photographing points to obtain a number of target photographing points for which an inverse solution exists;

[0058] During the specific implementation of this step, if the initial photographing point has a corresponding robotic arm inverse solution, the initial photographing point is retained as the target photographing point; if the initial photographing point does not have a corresponding robotic arm inverse solution, the initial photographing point can be deleted, or the initial photographing point can be adjusted until the adjusted initial photographing point has a corresponding robotic arm inverse solution, thereby obtaining several target photographing points.

[0059] Step S103, based on the target photographing points, a self-organizing map algorithm is used to train and obtain a target network topology structure for representing the positional relationship of each target photographing point;

[0060] In the specific implementation of this step, it is assumed that a set of target shooting points is given, represented as [β1,β2,β3,…β n ], where β i ∈Rd , is the coordinate in the robot arm configuration space. The robot arm is required to move to each photo point to take a photo, and finally return to the starting point, and make the total length of the path the shortest. Therefore, for these n photo points, the total number of feasible routes is The number of routes will increase rapidly as the number of photo points increases, so the traditional method cannot achieve a good path determination effect. Therefore, in this step, the self-organizing map algorithm is used to perform unsupervised training using each target photo point, thereby obtaining the target network topology that represents the positional relationship between each target photo point. That is, first assume that the robot arm is at any two photo points β i and β j There will be no collision when moving linearly between them. At this time, the Euclidean distance / Euclidean distance d between the two camera points can be directly calculated. ij , according to the properties of Euclidean distance / Euclidean distance, it can be easily concluded that d ji =d ij , and then the unsupervised model training can be performed according to the Euclidean distance to train and obtain the target network topology.

[0061] Step S104: planning and obtaining a target shooting path based at least on the position sequence of each target shooting point in the target network topology.

[0062] In the specific implementation process of this step, after obtaining the target network topology, the positional relationship of the target photographing points corresponding to each neuron can be determined according to the positional relationship of each neuron / node in the target network topology, thereby achieving the target photographing points [β1, β2, β3, ... β n ] to sort and obtain the sorted photo point sequence

[0063] The adaptive planning method for photographing paths for appearance inspection in this embodiment obtains target photographing points for which there are inverse solutions for the robotic arm by screening, and adopts a self-organizing mapping algorithm to perform unsupervised training using each target photographing point, thereby obtaining a target network topology structure that represents the positional relationship between each target photographing point, that is, each photographing point is mapped to a node / neuron in the target network topology structure, thereby accurately determining the positional sequence between each target photographing point, and thus quickly and reasonably planning the shortest target photographing path containing each target photographing point, laying the foundation for subsequently controlling the movement of the robotic arm based on the target photographing path to quickly capture images of the object to be inspected, reducing the robotic arm shooting time, improving the shooting efficiency, and reducing the robotic arm shooting consumption.

[0064] Based on the above embodiment, another embodiment of the present application provides a method for adaptively planning a photographing path for appearance inspection, comprising the following steps:

[0065] Step S201, obtaining a number of initial photographing points at the end of the robotic arm based on the position information of the object to be detected;

[0066] Step S202: determining the end-of-arm posture corresponding to each initial photographing point based on the initial photographing point;

[0067] Step S203: Based on the pre-established kinematic model of the manipulator, inverse calculation is performed on the posture of each manipulator end to obtain the manipulator inverse solution corresponding to each manipulator end posture, so as to obtain the inverse solution calculation result of each initial photographing point;

[0068] Step S204: Based on each of the target photographing points, perform Euclidean distance calculation on any two photographing points to obtain a number of Euclidean distances corresponding to each of the target photographing points;

[0069] Step S205: Based on a number of Euclidean distances corresponding to each photographing point, the positions of the network nodes corresponding to each photographing point in the initial network topology are adjusted to obtain a target network topology structure for representing the positional relationship between each target photographing point.

[0070] Step S206, based on the position sequence of each target photographing point in the target network topology, determining a subpath between any two adjacent target photographing points to obtain a plurality of subpaths;

[0071] In this step, when determining a subpath between any two adjacent target photographing points, a collision detection may be first performed on the any two adjacent target photographing points to obtain a detection result; if the detection result shows no collision, a linear path between the two target photographing points is determined as a subpath based on the two adjacent target photographing points; if the detection result shows a collision, any nonlinear path between the two target photographing points is determined as a subpath based on the two adjacent target photographing points; or, based on the adjacent target photographing points, a number of substitute photographing points are determined, and a self-organizing map algorithm is used to train a subnetwork topology structure for characterizing the positional relationship of each substitute photographing point based on each substitute photographing point, and the nonlinear subpath is determined based on the positional sequence of each substitute photographing point in the subnetwork topology structure.

[0072] That is, for two adjacent photo points and if and If there is no collision between the target shooting points, the linear path between them can be determined as a subpath. and If there is a collision between and Find a feasible path between and calculate the total distance along this path as the new d i,i+1 , if in and If a feasible path cannot be found between i,i+1 =∞, then and Find t points p1, p2, p3, ... p nearby t , that is, t substitute photo points are obtained. At this time, t is small, and a route from p1 to p can be quickly obtained through greedy algorithm or branch definition. t Alternatively, a self-organizing map algorithm is used based on these t substitute photo points to train and obtain a sub-network topology structure for characterizing the position relationship of each substitute photo point, so as to determine a nonlinear sub-path based on the position sequence of each substitute photo point in the sub-network topology structure, and finally the path is used as and subpaths between them.

[0073] Among them, when performing collision detection, the specific collision detection principle is as follows: determine the position of each point on the robotic arm in space. When the robotic arm moves, the position of the point on the robotic arm in space will also move accordingly. After calculating the position of the point on the robotic arm in space, it can be calculated whether it overlaps with the obstacle in space. If there is an overlap, it means a collision has occurred. If there is no overlap, it means no collision has occurred.

[0074] Step S206 obtains the target shooting path based on the sub-path planning.

[0075] In this step, after obtaining several sub-paths, the sub-paths can be spliced / combined to form a butterfly target shooting path.

[0076] In the specific implementation of this embodiment, after obtaining the target shooting path, the target shooting path can also be smoothed based on the Bezier curve to obtain the target shooting path. By smoothing the shooting path, it is beneficial to control the smooth movement of the robot arm, avoid large fluctuations in the robot arm displacement due to excessive fluctuations in the movement path, minimize the time and energy consumption of the robot arm movement, and at the same time ensure the quality of the captured image and the detection accuracy.

[0077] The adaptive planning method for photographing paths for appearance inspection in this embodiment obtains target photographing points for which there are inverse solutions for the robotic arm by screening, and adopts a self-organizing mapping algorithm to perform unsupervised training using each target photographing point, thereby obtaining a target network topology structure that represents the positional relationship between each target photographing point, that is, each photographing point is mapped to a node / neuron in the target network topology structure, thereby accurately determining the positional sequence between each target photographing point, and thus quickly and reasonably planning the shortest target photographing path containing each target photographing point, laying the foundation for subsequently controlling the movement of the robotic arm based on the target photographing path to quickly capture images of the object to be inspected, reducing the robotic arm shooting time, improving the shooting efficiency, and reducing the robotic arm shooting consumption.

[0078] The method in this embodiment can complete tasks such as path planning and collision detection for the robotic arm's photo points, thereby enabling multi-angle photography of the workpiece under test, effectively improving the efficiency and accuracy of quality inspection. In the field of industrial manufacturing, the application of this invention will greatly promote the development of quality control technology and provide more reliable and efficient quality inspection methods for industrial production.

[0079] Another embodiment of the present application provides a device for adaptively planning a photographing path for appearance inspection, as shown in FIG2 , including:

[0080] The calculation module 11 is used to perform inverse solution calculation of the robot arm for each of the initial photographing points to obtain the inverse solution calculation results of each initial photographing point;

[0081] A screening module 12 is configured to screen each of the initial photographing points based on the inverse solution calculation results of the initial photographing points to obtain a number of target photographing points for which an inverse solution exists;

[0082] A training module 13 is configured to use a self-organizing map algorithm to train, based on each of the target photographing points, a target network topology structure for representing the positional relationship between the target photographing points;

[0083] The planning module 14 is configured to plan and obtain a target shooting path based at least on the position sequence of each target shooting point in the target network topology.

[0084] During the specific implementation of this embodiment, the adaptive planning device for photographing paths for appearance detection also includes an acquisition module for acquiring a number of initial photographing points. The acquisition module is specifically used to: acquire a number of initial photographing points at the end of the robotic arm based on the position information of the object to be detected.

[0085] During the specific implementation of this embodiment, the calculation module is specifically used to: determine the robot arm end posture corresponding to each initial photographing point based on each of the initial photographing points; based on the pre-established kinematic model of the robot arm, perform inverse solution calculation on each of the robot arm end postures to obtain the robot arm inverse solution corresponding to each robot arm end posture, so as to obtain the inverse solution calculation result of each initial photographing point.

[0086] During the specific implementation of this embodiment, the training module is specifically used to: calculate the Euclidean distance between any two photo points based on each of the target photo points to obtain a number of Euclidean distances corresponding to each target photo point; and adjust the positions of the network nodes corresponding to each photo point in the initial network topology structure based on the number of Euclidean distances corresponding to each photo point, so as to train and obtain a target network topology structure for representing the positional relationship between each target photo point.

[0087] During the specific implementation of this embodiment, the planning module specifically includes an acquisition unit and a planning unit. The acquisition unit is used to: determine the sub-path between any two adjacent target photo points based on the position sequence of each target photo point in the target network topology structure to obtain a plurality of sub-paths; the planning unit is used to: obtain the target shooting path based on each of the sub-path plans.

[0088] During the specific implementation of this embodiment, the obtaining unit is specifically configured to: perform collision detection on any two adjacent target photographing points to obtain a detection result; if the detection result shows that no collision has occurred, determine, based on the two adjacent target photographing points, a linear path between the two target photographing points as a subpath; if the detection result shows that a collision has occurred, determine, based on the two adjacent target photographing points, any nonlinear path between the two target photographing points as a subpath; or, determine, based on the adjacent target photographing points, a number of substitute photographing points, and use a self-organizing map algorithm to train, based on each of the substitute photographing points, obtain a subnetwork topology structure for characterizing the positional relationship of each substitute photographing point, and determine a nonlinear subpath based on the positional sequence of each substitute photographing point in the subnetwork topology structure.

[0089] In the specific implementation process of this embodiment, the photographing path adaptive planning device for appearance detection also includes a smoothing processing module, which is used to: after obtaining the target path, smooth the target shooting path based on the Bezier curve to obtain the target shooting path.

[0090] The adaptive planning device for photographing paths for appearance inspection in this embodiment obtains target photographing points for which there are inverse solutions for the robotic arm by screening, and adopts a self-organizing mapping algorithm to perform unsupervised training using each target photographing point, thereby obtaining a target network topology structure that represents the positional relationship between each target photographing point, that is, each photographing point is mapped to a node / neuron in the target network topology structure, thereby accurately determining the positional sequence between each target photographing point, and thus quickly and reasonably planning the shortest target photographing path containing each target photographing point, laying the foundation for subsequently controlling the movement of the robotic arm based on the target photographing path to quickly capture images of the object to be inspected, reducing the robotic arm shooting time, improving the shooting efficiency, and reducing the robotic arm shooting consumption.

[0091] Another embodiment of the present application provides a storage medium storing a computer program. When the computer program is executed by a processor, the following method steps are implemented:

[0092] Step 1: Perform inverse calculations on the robot arm for each of the initial photographing points to obtain inverse calculation results for each of the initial photographing points.

[0093] Step 2: Screening each of the initial photographing points based on the inverse solution calculation results of each of the initial photographing points to obtain a number of target photographing points for which an inverse solution exists;

[0094] Step 3: Based on the target photographing points, a self-organizing map algorithm is used to train and obtain a target network topology structure for representing the positional relationship of each target photographing point;

[0095] Step 4: Plan and obtain a target shooting path based at least on the position sequence of each target shooting point in the target network topology structure.

[0096] The specific implementation process of the above method steps can be found in any of the above embodiments of the photographing path adaptive planning method for appearance detection, and this embodiment will not be repeated here.

[0097] The storage medium in the present application obtains target photographing points for which there are inverse solutions of the robotic arm through screening, and adopts the self-organizing mapping algorithm to perform unsupervised training using each target photographing point, thereby obtaining a target network topology structure that characterizes the positional relationship of each target photographing point, that is, each photographing point is mapped to a node / neuron in the target network topology structure, thereby accurately determining the position sequence between each target photographing point, so that the shortest target photographing path containing each target photographing point can be obtained through quick and reasonable planning, laying the foundation for subsequent control of the movement of the robotic arm based on the target photographing path to quickly capture images of the object to be detected, reducing the shooting time of the robotic arm, improving the shooting efficiency, and reducing the shooting consumption of the robotic arm.

[0098] Another embodiment of the present application provides an electronic device, as shown in FIG3 , including at least a memory 1 and a processor 2. The memory 1 stores a computer program, and the processor 2 implements the following method steps when executing the computer program in the memory 1:

[0099] Step 1: Perform inverse calculations on the robot arm for each of the initial photographing points to obtain inverse calculation results for each of the initial photographing points.

[0100] Step 2: Screening each of the initial photographing points based on the inverse solution calculation results of each of the initial photographing points to obtain a number of target photographing points for which an inverse solution exists;

[0101] Step 3: Based on the target photographing points, a self-organizing map algorithm is used to train and obtain a target network topology structure for representing the positional relationship of each target photographing point;

[0102] Step 4: Plan and obtain a target shooting path based at least on the position sequence of each target shooting point in the target network topology structure.

[0103] The specific implementation process of the above method steps can be found in any of the above embodiments of the photographing path adaptive planning method for appearance detection, and this embodiment will not be repeated here.

[0104] The storage medium in the present application obtains target photographing points for which there are inverse solutions of the robotic arm through screening, and adopts the self-organizing mapping algorithm to perform unsupervised training using each target photographing point, thereby obtaining a target network topology structure that characterizes the positional relationship of each target photographing point, that is, each photographing point is mapped to a node / neuron in the target network topology structure, thereby accurately determining the position sequence between each target photographing point, so that the shortest target photographing path containing each target photographing point can be obtained through quick and reasonable planning, laying the foundation for subsequent control of the movement of the robotic arm based on the target photographing path to quickly capture images of the object to be detected, reducing the shooting time of the robotic arm, improving the shooting efficiency, and reducing the shooting consumption of the robotic arm.

[0105] The above embodiments are merely exemplary embodiments of the present application and are not intended to limit the scope of the present application. The scope of protection of the present application is defined by the claims. Those skilled in the art may make various modifications or equivalent substitutions to the present application within the essence and scope of protection of the present application, and such modifications or equivalent substitutions shall also be deemed to fall within the scope of protection of the present application.

Claims

1. A method for adaptive planning of photographing paths for appearance detection, characterized in that: include: Perform inverse calculation of the robot arm for several initial photographing points respectively to obtain the inverse calculation results of each initial photographing point; Based on the inverse solution calculation results of each initial photographing point, each of the initial photographing points is screened to obtain a number of target photographing points for which an inverse solution exists; Based on each of the target photographing points, a self-organizing mapping algorithm is used to train and obtain a target network topology structure for characterizing the positional relationship of each of the target photographing points; A target shooting path is planned and obtained based at least on the position sequence of each of the target shooting points in the target network topology.

2. The method according to claim 1, characterized in that The method further includes obtaining a number of initial photographing points, specifically including: Based on the position information of the object to be detected, several initial photographing points at the end of the robotic arm are obtained.

3. The method according to claim 1, characterized in that The inverse solution calculation is performed for each of the initial photographing points to obtain the inverse solution calculation result of each of the initial photographing points, specifically including: Based on the initial photographing points, determine the end position of the robotic arm corresponding to each photographing point; Based on the pre-established kinematic model of the robotic arm, the inverse solution calculation is performed on the end position of each robotic arm to obtain the robotic arm inverse solution corresponding to each robotic arm end position, so as to obtain the inverse solution calculation result of each initial shooting point.

4. The method according to claim 1, characterized in that Based on each of the target photographing points, a target network topology structure for characterizing the positional relationship of each of the target photographing points is obtained by using a self-organizing mapping algorithm for training, including: Based on each of the target photographing points, respectively perform Euclidean distance calculation on any two photographing points to obtain a number of Euclidean distances corresponding to each of the target photographing points; Based on the Euclidean distances corresponding to each photo point, the initial network topology structure is The positions of the network nodes corresponding to the photo points are adjusted to train and obtain the target network topology structure used to characterize the positional relationship of each target photo point.

5. The method according to claim 1, characterized in that The step of planning and obtaining a target shooting path based at least on the position sequence of each target shooting point in the target network topology structure specifically includes: Based on the position sequence of each target photographing point in the target network topology structure, a subpath between any two adjacent target photographing points is determined to obtain a plurality of subpaths; The target shooting path is obtained based on the sub-path planning.

6. The method according to claim 5, characterized in that Determining the subpath between any two adjacent target photographing points specifically includes: Perform collision detection on any two adjacent target photo points to obtain the detection results; When the detection result is that no collision occurs, based on the two adjacent target photographing points, a linear path between the two target photographing points is determined as a subpath; When the detection result is a collision, any nonlinear path between the two target photo points is determined as a subpath based on the two adjacent target photo points; or, a number of substitute photo points are determined based on the adjacent target photo points, and a subnetwork topology structure for characterizing the positional relationship of each substitute photo point is obtained based on each of the substitute photo points using a self-organizing mapping algorithm for training, so as to determine a nonlinear subpath based on the position sequence of each substitute photo point in the subnetwork topology structure.

7. The method according to claim 5, characterized in that After obtaining the target path, the method further includes: The target shooting path is smoothed based on the Bezier curve to obtain the target shooting path.

8. A photographing path adaptive planning device for appearance detection, characterized in that: include: A calculation module is used to perform inverse solution calculations of the robot arm for a number of initial photographing points to obtain inverse solution calculation results of each initial photographing point; A screening module, used to screen each of the initial photographing points based on the inverse solution calculation results of each of the initial photographing points, and obtain a number of target photographing points for which an inverse solution exists; A training module, for obtaining a target network topology structure for characterizing the positional relationship of each target photographing point by using a self-organizing mapping algorithm based on each target photographing point; The planning module is used to plan and obtain a target shooting path based on at least the position sequence of each target shooting point in the target network topology structure.

9. A storage medium, characterized in that: The storage medium stores a computer program, and when the computer program is executed by the processor, the steps of the method for adaptively planning a photographing path for appearance detection described in any one of claims 1 to 6 are implemented.

10. An electronic device, characterized in that: At least includes a memory and a processor, wherein a computer program is stored in the memory, and when the processor executes the computer program in the memory, the steps of the method for adaptive planning of a photographing path for appearance detection described in any one of claims 1 to 6 are implemented.

Citation Information

Patent Citations

  • Single-joint fault space manipulator pose reaching space solving method

    CN107598929A

  • Automatic planning method for three-coordinate measurement of vehicle body

    CN110059879A

  • Automatic planning method of robot photographing measurement system for thin plate

    CN111609847A

  • Flying shooting control method and device

    CN112511759A

  • Mechanical arm splicing path planning method and device

    CN113650011A

Cited By

  • Detection camera path planning method and device, computer equipment and storage medium

    CN121007845A