Robot-based cable plugging method, device, electronic device and storage medium

Abstract

The present application relates to the field of robotics, and in particular to a robot-based cable plugging method, device, electronic device, and storage medium, wherein the method comprises: obtaining an input image containing a cable connector, and extracting a connector area of ​​the cable connector from the input image; identifying the connector area, obtaining connector features of the cable connector, and matching an optimal registration strategy based on the connector features, and generating a transformation relationship between the connector area and a target connector area based on the optimal registration strategy, and obtaining a posture adjustment amount of the cable connector based on the transformation relationship; adjusting the posture of the cable connector according to the posture adjustment amount and inserting the cable connector into a target socket. Thus, the problem of plugging multiple types of cable systems based on robots is solved, and the system has the advantages of simple assembly and strong versatility.

Description

Technical Field

[0001] The present application relates to the field of robotics technology, and in particular to a robot-based cable plugging method, device, electronic device, and storage medium. Background Art

[0002] The insertion of cable connectors is a crucial component of cable assembly, ensuring the mechanical and electrical connection between the cable and the device, and is crucial to the quality of cable assembly. Conventional cable insertion is often performed manually. Automated cable insertion using robots can effectively leverage the accuracy of repetitive operations, reduce labor intensity, and improve assembly efficiency.

[0003] As the performance of my country's aircraft models improves and production demand increases, the number of cables such as power and data will increase significantly, and the layout will become more complex. Therefore, improving the technical level of aircraft cable assembly has become an urgent problem to be solved in the field of aircraft assembly. Using robots for automated assembly of cables is an effective way. Plugging is an important operation in cable assembly. This process ensures the formation of cable connection paths, and the quality of plugging directly affects the performance of the aircraft.

[0004] Research and development of robotic cable plugging systems is relatively limited both domestically and internationally. Existing research has primarily focused on the processing and manufacturing of electronic cables and wire harnesses, often using specialized mechanisms. Overall, current automated cable assembly technology is difficult to apply to cable plugging tasks in aircraft environments, and there are currently no reports on automated aircraft cable assembly systems. Summary of the Invention

[0005] The present application provides a robot-based cable plugging method, device, electronic device and storage medium to solve the plugging problem of multi-type cable systems based on robots, and has the advantages of simple system assembly and strong versatility.

[0006] The first aspect of the present application provides a robot-based cable plugging method, comprising the following steps: acquiring an input image containing a cable connector, and extracting a connector area of ​​the cable connector from the input image; identifying the connector area, obtaining connector features of the cable connector, and matching an optimal registration strategy based on the connector features, generating a transformation relationship between the connector area and a target connector area based on the optimal registration strategy, and obtaining a posture adjustment amount of the cable connector based on the transformation relationship; and adjusting the posture of the cable connector based on the posture adjustment amount to insert the cable connector into the target socket. Optionally, in some embodiments, matching an optimal registration strategy based on the connector features, and generating a transformation relationship between the connector area and the target connector area based on the optimal registration strategy, further comprises: when the optimal registration strategy is an overall area registration strategy, the first transformation relationship between the connector area and the target connector area is:

[0007]

[0008] in, For the extracted cable connector area, The cable connector area for the target state, is the pixel coordinate of each point, is the homogeneous transformation matrix, It is the cable connector area after the transformation; Respectively for regions and The center coordinates of is the weight coefficient.

[0009] Optionally, in some embodiments, matching the connector features with the optimal registration strategy and generating a transformation relationship between the connector region and the target connector region based on the optimal registration strategy further includes: when the optimal registration strategy is a global feature descriptor registration strategy, a second transformation relationship between the connector region and the target connector region is:

[0010] ;

[0011] in, They correspond to the matching feature descriptor pairs Input and target feature point coordinates; are the corresponding decentering coordinates, Matching feature descriptor sets The coordinates of the center of mass of is the transformation matrix between coordinates; The pose detection result of the solution.

[0012] Optionally, in some embodiments, matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: when the best registration strategy is an undirected isomorphism recognition primitive registration strategy, a third transformation relationship between the connector region and the target connector region is:

[0013] ;

[0014] in, are the sets of recognition primitives for the input image The set of identification primitives for the center and target connectors the center of The pose detection result of the solution.

[0015] Optionally, in some embodiments, matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: when the best registration strategy is a directed isomorphism recognition primitive registration strategy, a fourth transformation relationship between the connector region and the target connector region is:

[0016] ;

[0017] in, are the recognition primitives of the input image Identification primitives of the center and target connectors the center of is the transformation matrix of position and attitude; The pose detection result of the solution.

[0018] Optionally, in some embodiments, matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: when the best registration strategy is a heterogeneous recognition primitive registration strategy, a fifth transformation relationship between the connector region and the target connector region is:

[0019] ;

[0020] in, In the observation connector and target connector respectively i The class identifies the center coordinates of the primitive; are the corresponding decentered coordinates; Calculate the weights for each type of normalized recognition primitives, i The more reliable the related identification and calculation results of the class identification primitives are, the more reliable the corresponding The bigger; The pose detection result of the solution.

[0021] A second aspect of the present application provides a robot-based cable plugging device, including: an acquisition module for acquiring an input image containing a cable connector and extracting a connector area of ​​the cable connector from the input image; a matching module for identifying the connector area, obtaining connector features of the cable connector, matching an optimal alignment strategy based on the connector features, generating a transformation relationship between the connector area and a target connector area based on the optimal alignment strategy, and obtaining a posture adjustment amount of the cable connector based on the transformation relationship; and an adjustment module for adjusting the posture of the cable connector according to the posture adjustment amount to insert the cable connector into the target socket.

[0022] Optionally, in some embodiments, the matching module further includes: a first transformation unit, wherein when the optimal registration strategy is the overall region registration strategy, a first transformation relationship between the connector region and the target connector region is:

[0023]

[0024] in, For the extracted cable connector area, The cable connector area for the target state, is the pixel coordinate of each point, is the homogeneous transformation matrix, It is the cable connector area after the transformation; Respectively for regions and The center coordinates of is the weight coefficient.

[0025] Optionally, in some embodiments, the matching module further includes: a second transformation unit, wherein when the optimal registration strategy is the overall feature descriptor registration strategy, the second transformation relationship between the connector area and the target connector area is:

[0026] ;

[0027] in, They correspond to the matching feature descriptor pairs Input and target feature point coordinates; are the corresponding decentering coordinates, Matching feature descriptor sets The coordinates of the center of mass of is the transformation matrix between coordinates; The pose detection result of the solution.

[0028] Optionally, in some embodiments, the matching module further includes: a third transformation unit, wherein when the optimal registration strategy is an undirected isomorphic recognition primitive registration strategy, a third transformation relationship between the connector region and the target connector region is:

[0029] ;

[0030] in, are the sets of recognition primitives for the input image The set of identification primitives for the center and target connectors the center of The pose detection result of the solution.

[0031] Optionally, in some embodiments, the matching module further includes: a fourth transformation unit, wherein when the optimal registration strategy is a directed isomorphic recognition primitive registration strategy, a fourth transformation relationship between the connector region and the target connector region is:

[0032] ;

[0033] in, are the recognition primitives of the input image Identification primitives of the center and target connectors the center of is the transformation matrix of position and attitude; The pose detection result of the solution.

[0034] Optionally, in some embodiments, the matching module further includes: a fifth transformation unit, wherein when the optimal registration strategy is the heterogeneous recognition primitive registration strategy, a fifth transformation relationship between the connector region and the target connector region is:

[0035] ;

[0036] in, In the observation connector and target connector respectively i The class identifies the center coordinates of the primitive; are the corresponding decentered coordinates; Calculate the weights for each type of normalized recognition primitives, i The more reliable the related identification and calculation results of the class identification primitives are, the more reliable the corresponding The bigger; The pose detection result of the solution.

[0037] The third aspect of the present application provides an electronic device, comprising: a memory, a processor, and a computer program stored on the memory and executable on the processor, wherein the processor executes the program to implement the robot-based cable plugging method as described in the above embodiment.

[0038] The fourth aspect of the present application provides a computer-readable storage medium having a computer program stored thereon, which is executed by a processor to implement the robot-based cable splicing method as described in the above embodiment.

[0039] Thus, by acquiring an input image containing a cable connector, extracting the connector region of the cable connector from the input image, and identifying the connector region, the connector features of the cable connector are obtained. An optimal registration strategy is then matched based on the connector features. Based on the optimal registration strategy, a transformation relationship between the connector region and the target connector region is generated. The position adjustment amount of the cable connector is obtained based on the transformation relationship, and the position of the cable connector is adjusted based on the position adjustment amount to insert the cable connector into the target socket. This solves the problem of plugging in multiple types of cable systems based on robots, and has the advantages of simple system assembly and strong versatility.

[0040] Additional aspects and advantages of the present application will be given in part in the description below, and in part will become apparent from the description below, or will be learned through practice of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0041] The above and / or additional aspects and advantages of the present application will become apparent and easily understood from the following description of the embodiments in conjunction with the accompanying drawings, in which:

[0042] Figure 1 A flowchart of a robot-based cable plugging method according to an embodiment of the present application;

[0043] Figure 2 Schematic diagram of different connectors provided according to one embodiment of the present application;

[0044] Figure 3 A schematic diagram of a connector modeling process provided according to one embodiment of the present application;

[0045] Figure 4 A schematic diagram of a connector modeling principle provided according to an embodiment of the present application;

[0046] Figure 5 A schematic diagram of a typical structure of a recognition element provided according to one embodiment of the present application;

[0047] Figure 6 A schematic diagram of the transformation relationship of a connector provided according to an embodiment of the present application;

[0048] Figure 7 A schematic diagram of a transformation relationship of a connector provided according to another embodiment of the present application;

[0049] Figure 8 A schematic diagram of the structure of a robot-based cable plugging system provided according to one embodiment of the present application;

[0050] Figure 9 A schematic diagram of a robot-based cable plugging principle provided according to one embodiment of the present application;

[0051] Figure 10 A schematic diagram of a target state and an actual state of a connector provided according to an embodiment of the present application;

[0052] Figure 11 A schematic diagram of an operation error state when contact occurs between a connector and a socket according to an embodiment of the present application;

[0053] Figure 12 A schematic diagram of the design principle of a cable robot plugging system and a connector detection algorithm provided according to one embodiment of the present application;

[0054] Figure 13 Schematic block diagram of a robot-based cable plugging device according to an embodiment of the present application;

[0055] Figure 14 A schematic diagram of the structure of an electronic device provided according to an embodiment of the present application. DETAILED DESCRIPTION

[0056] The following describes in detail embodiments of the present application, examples of which are shown in the accompanying drawings, wherein the same or similar reference numerals throughout represent the same or similar elements or elements having the same or similar functions. The embodiments described below with reference to the accompanying drawings are exemplary and are intended to be used to explain the present application, and should not be construed as limiting the present application.

[0057] The following describes the robot-based cable plugging method, device, electronic device and storage medium of the embodiment of the present application with reference to the accompanying drawings. In response to the problem mentioned in the above background technology that the current cable automation assembly technology is difficult to apply to the cable plugging task in the aircraft environment, the present application provides a robot-based cable plugging method, in which an input image containing a cable connector is obtained, and the connector area of ​​the cable connector is extracted from the input image, and the connector area is identified to obtain the connector features of the cable connector, and the best alignment strategy is matched according to the connector features, and the transformation relationship between the connector area and the target connector area is generated based on the best alignment strategy, and the posture adjustment amount of the cable connector is obtained according to the transformation relationship, and the posture of the cable connector is adjusted according to the posture adjustment amount to insert the cable connector into the target socket. In this way, the plugging problem of multi-type cable systems based on robots is solved, and it has the advantages of simple system assembly and strong versatility.

[0058] Specifically, Figure 1 A schematic flow chart of a robot-based cable splicing method provided in an embodiment of the present application.

[0059] like Figure 1 As shown, the robot-based cable plugging method includes the following steps:

[0060] In step S101 , an input image containing a cable connector is acquired, and a connector region of the cable connector is extracted from the input image.

[0061] Specifically, before performing posture detection of the cable connector, the embodiment of the present application needs to identify and extract the area of ​​the target connector in the input image. The requirement of this part is to obtain the area containing the complete cable connector or all recognition primitives from the original image.

[0062] In step S102, the connector area is identified, the connector features of the cable connector are obtained, and the optimal registration strategy is matched according to the connector features. Based on the optimal registration strategy, a transformation relationship between the connector area and the target connector area is generated, and the posture adjustment amount of the cable connector is obtained according to the transformation relationship.

[0063] It is understandable that cable connectors in production have a variety of shapes and structural characteristics. Since existing cable assembly operations are mostly performed manually, there is currently no set of connector classification, definition standards and modeling methods for automated operations.

[0064] Existing connector descriptions are primarily based on product category, with two basic classification methods: 1. By physical structure: circular or rectangular (based on the connector's cross-section); 2. By operating frequency: low frequency or high frequency (with 3 MHz as the boundary). This description method is simple and intuitive, but it lacks guidance for cable plug-in system design. This module will abstract and define cable connectors from the perspective of automated operation and establish a connector identification model.

[0065] From the perspective of target recognition and operation action output, connectors can be divided into: connectors with infinite posture directions, connectors with limited posture directions, and connectors with only one posture direction (such as Figure 2 shown).

[0066] The orientation of the connector is defined by the rotational symmetry of the connector end face around the insertion direction. The insertion direction is the z-axis. The rotation angle of the mating part of the plug-in assembly relative to the defined initial posture. The area of ​​the plug-in mating part after decentering is expressed in the cylindrical coordinate system as ,in, They are the cylindrical coordinates of the rotation angle, projection distance and height respectively.

[0067] Then the connector with infinite pose directions, the connector with finite pose directions, and the connector with unique pose direction have an uncertain rotation angle, a finite number of rotation angles, and zero rotation angle, respectively:

[0068]

[0069] Where, A small angle increment.

[0070] The set of the above rotation angles is defined as , then the connector posture direction set can be described as:

[0071] ;

[0072] Where, is a predefined initial connector end face posture; The rotation angle is The rotation matrix of .

[0073] Among them, most industrial cable connectors have a unique posture direction and can be further described according to the number of cores, type arrangement, etc. Figure 3 shown.

[0074] In addition to the insertion normal, the output of the detection algorithm can be determined based on the posture direction of the connector. For connector objects with infinite posture directions, the detection algorithm only needs to detect the spatial position; for connector objects with limited posture directions, the detection algorithm needs to give a feasible posture and optimize the action based on the posture before insertion, and output the target posture corresponding to the minimum adjustment amount; for connector objects with a unique posture direction, the recognition algorithm needs to achieve accurate and reliable detection of the target posture of the connector. That is, the error between the output connector posture of the detection algorithm and the target connector posture is in the form of:

[0075] ;

[0076] in, The error between the output posture of the detection algorithm and the target posture corresponds to the posture adjustment required for plugging; The connector posture is output by the detection algorithm; is a posture direction of the target connector; is the set of attitude directions of the target connector; is a symbolic function; is the cost function of the robot motion (usually a function of joint state and time), which measures the robot's movement from posture to Movement to posture Without considering the optimization of the robot trajectory in the operation space, .

[0077] In subsequent research, for each cable connector to be plugged in, the posture description method of the target connector is first determined, and the adjustment action of the robot is determined in the planning.

[0078] Modeling based on the operation process can realize the preliminary design of system hardware and algorithms according to the target connector object. The following will summarize and model the cable connectors from the perspective of target detection to realize the auxiliary design of the connector detection algorithm.

[0079] Because the mating male and female connectors have complementary mating portions and similar structures, only the cable-end connector is defined and described below. Calculations convert quantities related to the fixed-end connector to equivalent quantities related to the cable-end connector.

[0080] Connector modeling methods based on target detection such as Figure 4As shown in the figure, based on the detection algorithm's focus area, cable connectors can be divided into two categories: connectors based on overall morphology detection and connectors based on internal structure detection. Overall morphology-based detection primarily targets connectors without obvious internal structure. The key to overall morphology-based detection lies in defining a robust connector pose descriptor and reliably obtaining the connector's outline and overall area.

[0081] In the embodiment of the present application, the typical structural unit inside the connector is called an identification primitive, that is, an identification primitive is a collection of connector regions. A specific subset of , which is defined as follows:

[0082] ;

[0083] in, It is a collection of decentered connector regions in a description space (three-dimensional space, two-dimensional image space, or feature space); The connector area boundary Neighborhood, Curve is the boundary of the connector area; For the i A set of regions of recognition primitives, To identify primitives regional centres; To identify primitives the regional boundaries, is the recognition threshold; It is the modulus of the regional gradient under the filter kernel function (such as Gaussian kernel and Laplacian kernel); is the area filling function, so that the obtained is a simply connected region.

[0084] The identification primitives defined in the invention reflect the areas within the connector with local extrema in the description space, and these areas can be used to describe the connector's position. The internal structure-based detection method mainly utilizes the characteristics and distribution patterns of the connector's identification primitives.

[0085] The identification primitives defined above have the following properties:

[0086] 1. Direction attribute. Similar to the description in the classification standard, the directed and undirected attributes of the recognition primitives can be defined based on whether the recognition primitives themselves need to be described in terms of posture.

[0087] 2. Type attributes: Based on whether all the identification primitives defined in the connector are the same, the homogeneous and heterogeneous attributes of the identification primitives can be defined.

[0088] It should be noted that the same connector can have both directed and undirected recognition primitives. In addition, the directional attributes of the recognition primitives themselves cannot directly determine the orientation of the connector as a whole.

[0089] Typical structures that can be defined as identification primitives in actual connectors are as follows Figure 5 shown. Figure 5 The connector in (a) is a certain type of aviation plug. Figure 5 The connector in (b) is a two-phase charger and a national standard three-phase power plug. Figure 5 The connector in (c) is the plug of a USB data cable. The boxed area is the identification element of the connector.

[0090] Before detecting the connector pose, it is necessary to identify and extract the target connector area in the input image. The requirement of this part is to obtain the area containing the complete connector or all recognition primitives from the original image, that is, to define a window area of ​​the image. (usually rectangular or circular), with:

[0091] ;

[0092] in, is the pixel coordinate of the point in the region; is the set of all identified primitive regions; is the boundary envelope area of ​​the connector; Window area the boundaries; is the relaxation boundary distance.

[0093] Target recognition and detection can effectively improve the performance of the algorithm by separating the area of ​​interest from the background and identifying the set of primitive regions. The description of the internal structure of the connector is realized. It should be noted that it is often not necessary to use All elements in , using only The subset of can realize the posture detection of the connector. The basic principle of using recognition primitives is to make all equations in the detection algorithm positive or over-determined.

[0094] After obtaining a series of recognition primitives, we need to calculate the plug's posture based on this information. Based on the above cable connector modeling method, the following introduces the posture detection method for different types of connectors.

[0095] Based on target detection-based modeling and the detection algorithm's focus area, cable connector pose detection methods can be divided into those based on overall morphology detection and those based on internal structure detection. Considering both cost and performance, planar cameras have a wider range of applications in industry, and subsequent research will be conducted based on these industrial cameras.

[0096] Due to the diverse nature of connectors, it's often difficult to define a universal pose description method for overall morphology-based inspection. This requires analysis and design based on the specific connector being inspected. Furthermore, most cable connectors have recognizable internal structures. Here, we present two design ideas for connector inspection methods based on overall morphology.

[0097] Optionally, in some embodiments, the optimal registration strategy is matched according to the connector features, and the transformation relationship between the connector area and the target connector area is generated based on the optimal registration strategy, further comprising: when the optimal registration strategy is the overall region registration strategy, the basic idea based on the overall region registration strategy is to maximize the overlap of the connector area in the input image after operation with the connector area in the target image, that is, to find the transformation relationship between the input connector area and the target connector area so that the two have the largest area intersection-union ratio and the smallest center offset. The process is as follows: Figure 6 The connector in the figure is a Type-C USB interface.

[0098] The specific method based on overall region registration is described as follows. The connector region extracted from the input image is denoted as ; The connector area of ​​the target state is ; The pixel coordinates of each point are ; The current state of the connector can be estimated by solving the following optimization problem:

[0099] ;

[0100] in, For the extracted cable connector area, The cable connector area for the target state, is the pixel coordinate of each point, is the homogeneous transformation matrix, It is the cable connector area after the transformation; Respectively for regions and The center coordinates of is the weight coefficient.

[0101] According to the solution , the corresponding pose adjustments can be calculated. Methods based on global region registration require that the connector's extraction area is sufficient to describe the connector's pose and accurately extract the entire connector area. This method is more suitable for connectors where a singly connected region can be extracted. When it is difficult to obtain a complete connector outline, the algorithm's accuracy will be significantly affected.

[0102] Optionally, in some embodiments, the best registration strategy is matched according to the connector features, and the transformation relationship between the connector area and the target connector area is generated based on the best registration strategy, further comprising: when the best registration strategy is the overall feature descriptor registration strategy, the basic idea based on the overall feature descriptor registration strategy is to first define the feature descriptor for describing the connector in the input space, then extract the feature descriptor of the connector in the input image, and find the corresponding feature descriptor in the target connector, and calculate the transformation relationship between the connector in the image and the target connector based on the relationship between the matching feature descriptors. The process is as follows: Figure 7 As shown, Figure 7 The connector object in the figure is the crystal head part of the network cable.

[0103] The specific method of overall feature descriptor registration is described as follows. The descriptor set extracted from the connector area of ​​the input image is denoted as ; The descriptor set extracted from the target connector region is ; Then solve the following feature point matching and ICP problems in turn:

[0104] ;

[0105] in, for Feature descriptors in ; is the set threshold; The above formula pairs the feature descriptors, and the relationship between the input connector and the target connector can be calculated by using the relationship between the matching feature points.

[0106] ;

[0107] in, They correspond to the matching feature descriptor pairs Input and target feature point coordinates; are the corresponding decentering coordinates, Matching feature descriptor sets The coordinates of the center of mass of is the transformation matrix between coordinates; The pose detection result of the solution.

[0108] Based on the above solution, the corresponding pose adjustment can be calculated. The holistic feature descriptor-based registration method requires the defined feature descriptor to be sufficiently stable and robust. Generally, feature descriptors are calculated based on the gradient of the input features, so this method usually requires the connector contour to have a sudden change in curvature.

[0109] In general, the whole-region registration method is suitable for connectors where a singly connected region can be extracted. It performs well when the extracted connector region is a convex set. However, when it is difficult to obtain the complete connector boundary, the algorithm accuracy will be greatly affected. The whole-feature descriptor registration method is suitable for situations where feature descriptors can be extracted from the connector contour. This usually requires the presence of a sudden change in curvature on the connector contour. When the number of extracted descriptors is too small, the algorithm accuracy will be greatly affected.

[0110] Internal structure-based detection methods primarily leverage the connector's inherent recognition primitives and design corresponding algorithms based on their distribution patterns. This section describes the basic connector description and detection methods based on the directional attributes (undirected, directed) and type attributes (homogeneous, heterogeneous) of the recognition primitives.

[0111] Optionally, in some embodiments, the best registration strategy is matched according to the connector features, and the transformation relationship between the connector area and the target connector area is generated based on the best registration strategy, further comprising: when the best registration strategy is an undirected isomorphic recognition primitive registration strategy, the configuration strategy of the undirected isomorphic recognition primitive is similar to the idea of ​​the registration strategy based on the overall feature descriptor. Figure 5 The undirected, isomorphic recognition primitives shown in (a) first need to extract the region of each recognition primitive, which is recorded as Due to the undirected and isomorphic characteristics of the recognition primitives, each recognition primitive can be used to The center of . Then the recognition primitive center of the input image is Identification motif center with target connector Matching is performed and the conversion relationship between the two is calculated based on the matching results. The above calculation process can be transformed into a similar matching and ICP problem:

[0112] ;

[0113] in, are the sets of recognition primitives for the input image The set of identification primitives for the center and target connectors the center of The pose detection result of the solution.

[0114] Optionally, in some embodiments, the best registration strategy is matched according to the connector features, and the transformation relationship between the connector area and the target connector area is generated based on the best registration strategy, further comprising: when the best registration strategy is a directed isomorphic recognition primitive registration strategy, the directed isomorphic recognition primitive registration strategy is a high-dimensional extension of the undirected isomorphic recognition primitive detection method. Figure 5 The directed, isomorphic recognition primitive shown in (b) of Extraction, since each recognition primitive has a direction attribute, at this time only the primitive center It is not possible to fully describe the recognition primitives, so it is necessary to The shape defines the unit direction vector of each primitive , that is, the descriptor of the recognition primitive is At this time, the distance function between the identification primitives is Defined as:

[0115] ;

[0116] in, To identify the posture distance function between primitives, when Both can realize the recognition of primitives i When matching, take the above formula (The recognition primitive has a rotation angle of 180 degrees, such as Figure 5 (b) shows the rectangular primitive), whereas only Recognition primitives can be realized i Remove the matching (180 degrees is not the rotation angle for identifying primitives); Is a weight coefficient. Identify the direction vector of the primitive Typically, a specific direction may be chosen to identify the axis of symmetry of the primitive region.

[0117] Using the distance definition above The corresponding optimization algorithm problem can be realized:

[0118] ;

[0119] in, are the recognition primitives of the input image Identification primitives of the center and target connectors the center of is the transformation matrix of position and attitude; The pose detection result of the solution.

[0120] Optionally, in some embodiments, the best registration strategy is matched according to the connector features, and the transformation relationship between the connector area and the target connector area is generated based on the best registration strategy, further comprising: when the best registration strategy is a heterogeneous recognition primitive registration strategy, the registration strategy of the heterogeneous recognition primitive is a generalization and synthesis of the homogeneous recognition primitive registration strategy. Figure 5 The recognition primitives with heterogeneous attributes shown in (c) also need to extract different types of recognition primitives separately. m Different recognition primitives are defined as follows according to the directional properties of the primitives:

[0121] ;

[0122] Then, the recognition primitives in each category are matched and correlated with the corresponding recognition primitives of the target connector. Finally, the results of all categories are combined to obtain the final calculation result:

[0123] ;

[0124] in, In the observation connector and target connector respectively i The class identifies the center coordinates of the primitive; are the corresponding decentered coordinates; Calculate the weights for each type of normalized recognition primitives, i The more reliable the related identification and calculation results of the class identification primitives are, the more reliable the corresponding The bigger; The pose detection result of the solution.

[0125] In step S103 , the posture of the cable connector is adjusted according to the posture adjustment amount and the cable connector is inserted into the target socket.

[0126] Specifically, after obtaining the posture adjustment amount of the cable connector according to the transformation relationship, the embodiment of the present application can adjust the posture of the cable connector according to the posture adjustment amount and insert the cable connector into the target socket.

[0127] It should be noted that, based on the process flow required for cable plugging, cable connectors can be divided into two categories: those requiring fixed operation and those not. Connectors that do not require fixed operation simply require direct insertion to connect active devices, while connectors that do require fixed operation require methods such as threaded fastening to achieve connection between active devices. Those that do not require fixed operation primarily include conventional cables such as network cables, USB cables, audio cables, and power cables, while those that do require fixed operation primarily include various industrial cables used in the automotive, aviation, and aerospace industries.

[0128] In this embodiment, the actuator requirements can be determined based on whether a fixation operation is required, thereby implementing a preliminary design of the operation action. For objects that do not require fixation, the actuator only needs to achieve reliable clamping of the target connector object; for connectors that require fixation, the actuator also needs to achieve appropriate connector fixation or have an auxiliary fixation mechanism.

[0129] In order to enable relevant technical personnel in this field to further understand the robot-based cable plugging method of the embodiment of the present application.

[0130] The embodiment of the present application utilizes a rotatable industrial camera and an industrial robot to build a cable plugging operation platform. In view of the task characteristics of the robot performing aircraft cable connector plugging (hereinafter referred to as cable plugging), the plugging process can be divided into four typical stages: grasping, observation, insertion, and fixation after insertion. A complete set of robot plugging system design methods for aviation cable plugs was invented, the plugging process and plugging performance requirements of cable connectors were analyzed, and a universal cable automatic plugging process was designed in combination with the characteristics of robot automatic operation. At the same time, the functions and precision requirements of the plugging system were analyzed to adapt to the automatic plugging tasks of different types of cable connectors. In view of the characteristics of many types and different shapes of cable connectors, various types of connectors are analyzed and abstracted from the perspective of robot automatic plugging, and a universal detection method for different types of connectors is proposed. The embodiment of the present application can solve the problem of multi-type cable system design based on robots, and has the advantages of simple system assembly and strong versatility.

[0131] The robot-based cable plugging system mainly implements the following task processes:

[0132] (1) Observe and calibrate the position of the target socket;

[0133] (2) Identify and grab the connector portion of the cable to be operated from the cable placement area;

[0134] (3) Adjust the robot so that the cable connector is close to and aligned with the socket after grasping;

[0135] (4) Control the robot to insert the grabbing cable connector into the target socket;

[0136] (5) Fix the cable connector and socket to achieve electrical path.

[0137] The basic objectives of the cable plugging task can be summarized as follows:

[0138] (1) During the plugging process, there is no obstruction between the cable connector and the socket, that is, the relative position error between the two is less than the design allowable value;

[0139] (2) During the plugging process, the plugging force generated between the cable connector and the socket is less than the design allowable value.

[0140] The above design allowable values ​​are determined by the matching tolerance between the contact surface material of the connector and the socket.

[0141] To complete the above cable plugging tasks, the cable plugging system needs to have the following basic functions:

[0142] (1) Realize the identification and grasping of cable connectors;

[0143] (2) Realize the detection and control of the relative position between the cable connector and the socket;

[0144] (3) Realize the detection and control of the plug-in operation force.

[0145] Figure 8 The system represents the structure of a robot-based cable plugging system. A dual-arm robot is used in the system to implement cable plugging. This solves the posture adjustment problem caused by the limited flexible working space of the robot. Considering the versatility and portability of the system, an industrial array camera is used to achieve clear acquisition of the original image. The design process of the system is as follows: Figure 9 As shown, it mainly includes four main modules: S1 system process design, S2 system error analysis and tolerance design, S3 cable connector identification model establishment and S4 connector detection algorithm design.

[0146] For system process design, the operation process of cable plugging by the robot can be divided into the following four stages according to the focus of the cable plugging task. The task objectives and design ideas of each operation stage are as follows:

[0147] 1. Crawl stage.

[0148] The primary goal of the grasping phase is to reliably grasp the target cable at the appropriate location and ensure that the connector's position is fully controllable after grasping. This phase requires selecting the appropriate grasping location and designing a reliable end-grip mechanism. This ensures that after grasping the connector, the robot can fully control the cable connector. The grasping phase also includes detection and identification of the target grasping location, as well as pre-positioning of the connector.

[0149] 2. Observation phase.

[0150] The primary objective of the observation phase is to use sensors to observe the target connector and receptacle (calibrating and inspecting the actual fixed-end connector), obtaining accurate pose information and calculating the required robot adjustments to adjust the robot to the desired mating pose. The observation phase is divided into a pre-observation phase (performed before mating) and a target observation phase (performed during mating). The pre-observation phase targets the fixed-end receptacle. Through observation and correction, manufacturing and assembly errors introduced during production, processing, and assembly are compensated to achieve the accurate target pose. This pre-observation phase only needs to be performed once, after the fixed-end processing is complete. The pre-observation phase also requires defining the relative pose of the target cable connector and receptacle and designing a corresponding target detection algorithm. The target observation phase targets the cable connector during mating, calculating its pose through observation. The target observation phase also requires resolving the observed and calculated results into robot control variables.

[0151] 3. Insertion phase.

[0152] The main goal of the insertion stage is to use control methods to compensate for the errors generated during the plugging operation, that is, starting from the desired plugging posture point, use sensors to identify and analyze the contact status between the target connector and the socket and perform corresponding adjustment actions so that the connector can be smoothly inserted into the target end. The insertion stage can be divided into the pre-insertion adjustment stage and the insertion control stage according to the insertion status of the connector. Among them, the pre-insertion adjustment stage responds to situations where large operational errors occur. The goal of this process is to reduce the pre-insertion error and control the connector from a stuck state to a pluggable state. The insertion control stage reduces the plugging force and improves the flexibility of the insertion process through flexible control, and completes the judgment of the final state of the plugging.

[0153] 4. Post-insertion fixation stage.

[0154] The primary goal of the post-insertion fixation phase is to securely connect the inserted connector to the target end through the end effector. This phase requires designing the appropriate actuators or operational procedures to achieve the required fixation, and also considers the coordination and coordination between devices.

[0155] Of the four stages described above, the grasping stage is the foundation for implementing a cable plugging system, the post-insertion securing stage ensures reliable cable plugging operations, and the observation and insertion stages are key to successful cable plugging. These stages are present in all types of cable connectors and require specific design considerations. The post-insertion securing stage can be omitted depending on the specific connector.

[0156] In order to design a robot cable plug-in system, it is necessary to study the conditions for successful connector insertion. Figure 10 The connector state shown in (a) can be abstractly defined as follows Figure 10 The operating error shown in (b). Figure 10 The target state in is the desired state of the connector under observation, which corresponds to the zero error state that needs to be adjusted to the fixed plug end. Figure 11 This defines the operational error state when the connector and receptacle make contact. In the figure, the inclined arrow area and the vertical arrow area represent the contact surface of the current cable connector and the contact surface of the fixed end receptacle, respectively. and They represent the normal direction of the socket surface and the actual insertion direction respectively; represents the deflection error in the insertion direction; represents the rotational error around the insertion direction; and It represents the translation error in the plane of the socket.

[0157] Order set and Representing the points on the contact surface of the cable connector and the contact surface of the fixed end socket, respectively, the following necessary and sufficient conditions for the collision-free initial plugging stage can be obtained:

[0158] ;

[0159] in, They are The point in Between two points Norm distance; This is the tolerance during the insertion phase. This value depends on the material of the contact part and the stiffness of the robot system. The greater the stiffness of the two, the smaller the tolerance value.

[0160] The above equation provides a criterion for collision-free initial insertion. It is primarily used to determine the initial insertion state and evaluate performance in simulation. For a few connectors with simple structures, this equation can also be used directly in the control loop. However, in general, in most connector cases, explicit analytical form cannot be expressed and used. Considering the controller capabilities during the insertion phase, we can derive the following necessary conditions for a successful initial insertion phase:

[0161] ;

[0162] in, and These tolerances represent the translational tolerance at the initial insertion stage, the rotational tolerance around the insertion direction, and the angular tolerance in the insertion direction. These tolerance parameters can be calculated independently from the structural dimensions without considering the effects of dynamics and deformation during contact.

[0163] This formula facilitates performance evaluation of the inspection algorithm and the initial insertion phase. Tolerance parameters can be derived from CAD models, physical measurements, and robotic insertion experiments. These insertion tolerances are influenced by connector clearance and robot control capabilities. Larger clearances and superior controller performance result in larger initial insertion tolerances.

[0164] It should be noted that the above indicators are not sufficient to determine a successful insertion. The measured tolerance parameter values ​​are often the lower limit of the system's required performance. When both translational and rotational errors exist, the values ​​of these insertion tolerance parameters will be smaller than when only translational or rotational errors exist. Therefore, in subsequent designs, it is desirable for the system's performance indicators to have a larger margin than the measured tolerance parameters (a tolerance margin of at least 50% of the measured results). Larger clearances and higher-performance controllers can allow for even smaller tolerance margins.

[0165] For a given cable connector object, the design steps of the connector detection algorithm for plugging are as follows:

[0166] (1) First, the plug-in operation process is designed according to the modeling standard based on the operation process, the end effector is designed accordingly, and the error output form of the detection algorithm is determined.

[0167] (2) Then, the basic connector detection method (whole-based method and internal structure-based method) is determined according to the target detection-based modeling standard. The target connector is described accordingly according to the aforementioned algorithm requirements and the corresponding detection algorithm is designed.

[0168] (3) Verify the algorithm performance indicators and optimize the algorithm accordingly based on the results.

[0169] In summary, the general cable robot plug-in system and connector detection algorithm design process is as follows Figure 12 It should be noted that in most cases, connector objects have obvious internal structural characteristics. Detection methods based on internal structure are more common and have better performance. For specific connector objects, the methods and implementation methods within the design framework can be optimized according to their characteristics.

[0170] As a result, the problem that the current automated cable assembly technology is difficult to apply to cable plugging tasks in aircraft environments is solved, and the automation level of the manufacturing industry is improved. The embodiment of the present application is based on the automatic plugging of cable connectors by robots, which has long-term significance for promoting the development of robot-based automated cable assembly technology.

[0171] According to the robot-based cable plugging method proposed in the embodiment of the present application, an input image containing a cable connector is obtained, and the connector area of ​​the cable connector is extracted from the input image, and the connector area is identified to obtain the connector features of the cable connector. The optimal registration strategy is matched according to the connector features, and a transformation relationship between the connector area and the target connector area is generated based on the optimal registration strategy. The position adjustment amount of the cable connector is obtained according to the transformation relationship, and the position of the cable connector is adjusted according to the position adjustment amount to insert the cable connector into the target socket. In this way, the plugging problem of multi-type cable systems based on robots is solved, and the advantages of simple system assembly and strong versatility are achieved.

[0172] Next, a robot-based cable plugging device proposed according to an embodiment of the present application will be described with reference to the accompanying drawings.

[0173] Figure 13 It is a block diagram of a robot-based cable plugging device according to an embodiment of the present application.

[0174] like Figure 13 As shown, the robot-based cable plugging device 10 includes: an acquisition module 100 , a matching module 200 and an adjustment module 300 .

[0175] The acquisition module 100 is configured to acquire an input image containing a cable connector and extract a connector region of the cable connector from the input image.

[0176] The matching module 200 is used to identify the connector area, obtain the connector features of the cable connector, match the optimal alignment strategy based on the connector features, generate a transformation relationship between the connector area and the target connector area based on the optimal alignment strategy, and obtain the position adjustment amount of the cable connector based on the transformation relationship.

[0177] The adjustment module 300 is configured to adjust the posture of the cable connector according to the posture adjustment amount and insert the cable connector into the target socket.

[0178] Optionally, in some embodiments, the matching module 200 further includes: a first transformation unit, wherein when the optimal registration strategy is the overall region registration strategy, the first transformation relationship between the connector region and the target connector region is:

[0179]

[0180] in, For the extracted cable connector area, The cable connector area for the target state, is the pixel coordinate of each point, is the homogeneous transformation matrix, It is the cable connector area after the transformation; Respectively for regions and The center coordinates of is the weight coefficient.

[0181] Optionally, in some embodiments, the matching module 200 further includes: a second transformation unit, wherein when the optimal registration strategy is the overall feature descriptor registration strategy, the second transformation relationship between the connector region and the target connector region is:

[0182] ;

[0183] in, They correspond to the matching feature descriptor pairs Input and target feature point coordinates; are the corresponding decentering coordinates, Matching feature descriptor sets The coordinates of the center of mass of is the transformation matrix between coordinates; The pose detection result of the solution.

[0184] Optionally, in some embodiments, the matching module 200 further includes: a third transformation unit, wherein when the optimal registration strategy is the undirected isomorphism recognition primitive registration strategy, the third transformation relationship between the connector region and the target connector region is:

[0185] ;

[0186] in, are the sets of recognition primitives for the input image The set of identification primitives for the center and target connectors the center of The pose detection result of the solution.

[0187] Optionally, in some embodiments, the matching module 200 further includes: a fourth transformation unit, wherein when the optimal registration strategy is the directed isomorphism recognition primitive registration strategy, the fourth transformation relationship between the connector region and the target connector region is:

[0188] ;

[0189] in, are the recognition primitives of the input image Identification primitives of the center and target connectors the center of is the transformation matrix of position and attitude; The pose detection result of the solution.

[0190] Optionally, in some embodiments, the matching module 200 further includes: a fifth transformation unit, wherein when the optimal registration strategy is the heterogeneous recognition primitive registration strategy, the fifth transformation relationship between the connector region and the target connector region is:

[0191] ;

[0192] in, In the observation connector and target connector respectively i The class identifies the center coordinates of the primitive; are the corresponding decentered coordinates; Calculate the weights for each type of normalized recognition primitives, i The more reliable the related identification and calculation results of the class identification primitives are, the better i Class corresponding The bigger; The pose detection result of the solution.

[0193] It should be noted that the aforementioned explanation of the embodiment of the robot-based cable plugging method is also applicable to the robot-based cable plugging device of this embodiment, and will not be repeated here.

[0194] According to the robot-based cable plug-in device proposed in the embodiment of the present application, an input image containing a cable connector is obtained, and the connector area of ​​the cable connector is extracted from the input image, and the connector area is identified to obtain the connector features of the cable connector. The optimal registration strategy is matched according to the connector features, and a transformation relationship between the connector area and the target connector area is generated based on the optimal registration strategy. The position adjustment amount of the cable connector is obtained according to the transformation relationship, and the position of the cable connector is adjusted according to the position adjustment amount to insert the cable connector into the target socket. In this way, the plug-in problem of multi-type cable systems based on robots is solved, and the advantages of simple system assembly and strong versatility are achieved.

[0195] Figure 14 This is a schematic diagram of the structure of an electronic device provided in an embodiment of the present application. The electronic device may include:

[0196] Memory 1401 , processor 1402 , and computer programs stored in the memory 1401 and executable on the processor 1402 .

[0197] When the processor 1402 executes the program, the robot-based cable plugging method provided in the above embodiment is implemented.

[0198] Furthermore, the electronic device further includes:

[0199] The communication interface 1403 is used for communication between the memory 1401 and the processor 1402 .

[0200] The memory 1401 is used to store computer programs that can be run on the processor 1402 .

[0201] The memory 1401 may include a high-speed RAM (Random Access Memory) memory, and may also include a non-volatile memory, such as at least one disk memory.

[0202] If the memory 1401, processor 1402, and communication interface 1403 are implemented independently, the communication interface 1403, memory 1401, and processor 1402 can be connected to each other via a bus and communicate with each other. The bus can be an ISA (Industry Standard Architecture) bus, a PCI (Peripheral Component Interconnect) bus, or an EISA (Extended Industry Standard Architecture) bus. Buses can be divided into address buses, data buses, control buses, etc. For ease of representation, Figure 14 Only one thick line is used in the diagram, but this does not mean that there is only one bus or one type of bus.

[0203] Optionally, in a specific implementation, if the memory 1401, the processor 1402 and the communication interface 1403 are integrated on a chip, the memory 1401, the processor 1402 and the communication interface 1403 can communicate with each other through an internal interface.

[0204] The processor 1402 may be a CPU (Central Processing Unit), or an ASIC (Application Specific Integrated Circuit), or one or more integrated circuits configured to implement the embodiments of the present application.

[0205] An embodiment of the present application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-mentioned robot-based cable splicing method.

[0206] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present application. In this specification, the schematic representations of the above terms do not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, those skilled in the art can combine and combine different embodiments or examples described in this specification and features of different embodiments or examples without contradiction.

[0207] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of technical features indicated. Thus, a feature specified as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of this application, "N" means at least two, for example, two, three, etc., unless otherwise specifically defined.

[0208] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, fragment or portion of code comprising one or more executable instructions for implementing the steps of a custom logical function or process, and the scope of the preferred embodiments of the present application includes alternative implementations in which functions may be performed out of the order shown or discussed, including performing functions in a substantially simultaneous manner or in reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present application belong.

[0209] It should be understood that various parts of the present application can be implemented using hardware, software, firmware, or a combination thereof. In the above embodiment, the N steps or methods can be implemented using software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented using hardware, as in another embodiment, it can be implemented using any one or a combination of the following technologies known in the art: a discrete logic circuit having a logic gate circuit for implementing a logic function on a data signal, an application-specific integrated circuit having a suitable combination of logic gate circuits, a programmable gate array, a field programmable gate array, etc.

[0210] Those skilled in the art will understand that all or part of the steps in the method of the above embodiment can be completed by instructing related hardware through a program, and the program can be stored in a computer-readable storage medium. When the program is executed, it includes one or a combination of the steps of the method embodiment.

[0211] Although the embodiments of the present application have been shown and described above, it can be understood that the above embodiments are exemplary and cannot be understood as limitations on the present application. Ordinary technicians in this field can change, modify, replace and modify the above embodiments within the scope of the present application.

Claims

1. A robot-based cable plugging method, characterized in that: The following steps are involved: Acquire an input image containing a cable connector, and extract a connector region of the cable connector from the input image; Identifying the connector region, obtaining connector features of the cable connector, matching an optimal registration strategy based on the connector features, generating a transformation relationship between the connector region and a target connector region based on the optimal registration strategy, and obtaining a posture adjustment amount of the cable connector based on the transformation relationship; as well as The cable connector is inserted into a target socket by adjusting the cable connector's posture according to the posture adjustment amount. The identifying the connector area, obtaining the connector features of the cable connector, and matching the optimal registration strategy according to the connector features, includes: From the perspective of target recognition and operation action output, connectors are divided into: connectors with infinite posture directions, connectors with limited posture directions, and connectors with only one posture direction; The output of the detection algorithm can be determined based on the orientation of the connector. The error between the output connector orientation of the detection algorithm and the target connector orientation is expressed as: ； in, The error between the output posture of the detection algorithm and the target posture corresponds to the posture adjustment required for plugging; The connector posture is output by the detection algorithm; is a posture direction of the target connector; is the set of attitude directions of the target connector; is a symbolic function; is the cost function of the robot motion, which measures the robot's movement from posture Movement to posture Without considering the optimization of the robot trajectory in the operation space, ; Based on the focus area of ​​the detection algorithm, cable connectors can be divided into two categories: connectors based on overall morphology detection and connectors based on internal structure detection. Among them, overall morphology-based detection is mainly aimed at connector objects without obvious internal structure. The key to overall morphology-based detection is to define a robust connector pose descriptor and reliably obtain the connector outline and overall connector area. The detection method based on internal structure mainly utilizes the characteristics of connectors in identifying primitives and designs corresponding algorithms based on distribution rules to realize the direction attributes and type attributes of the identification primitives. The direction attributes include undirected and directed, and the type attributes include homogeneous and heterogeneous. The optimal registration strategy is an overall region registration strategy, an overall feature descriptor registration strategy, an undirected isomorphic recognition primitive registration strategy, a directed isomorphic recognition primitive registration strategy, or a heterogeneous recognition primitive registration strategy.

2. The method according to claim 1, characterized in that The method of matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: When the optimal registration strategy is the overall region registration strategy, the first transformation relationship between the connector region and the target connector region is: in, For the extracted cable connector area, The cable connector area for the target state, is the pixel coordinate of each point, is the homogeneous transformation matrix, It is the cable connector area after the transformation; Respectively for regions and The center coordinates of is the weight coefficient.

3. The method according to claim 1, characterized in that The method of matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: When the optimal registration strategy is the overall feature descriptor registration strategy, the second transformation relationship between the connector region and the target connector region is: ； in, They correspond to the matching feature descriptor pairs Input and target feature point coordinates; are the corresponding decentering coordinates, Matching feature descriptor sets The coordinates of the center of mass of is the transformation matrix between coordinates; The pose detection result of the solution.

4. The method according to claim 1, wherein The method of matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: When the optimal registration strategy is the undirected isomorphism recognition primitive registration strategy, the third transformation relationship between the connector region and the target connector region is: ； in, are the sets of recognition primitives for the input image The set of identification primitives for the center and target connectors the center of The pose detection result of the solution.

5. The method according to claim 1, wherein The method of matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: When the optimal registration strategy is the directed isomorphism recognition primitive registration strategy, the fourth transformation relationship between the connector region and the target connector region is: ； in, are the recognition primitives of the input image Identification primitives of the center and target connectors the center of is the transformation matrix of position and attitude; The pose detection result of the solution.

6. The method according to claim 1, characterized in that The method of matching the best registration strategy according to the connector features and generating a transformation relationship between the connector region and the target connector region based on the best registration strategy further includes: When the optimal registration strategy is the heterogeneous recognition primitive registration strategy, the fifth transformation relationship between the connector region and the target connector region is: ； in, In the observation connector and target connector respectively i The class identifies the center coordinates of the primitive; are the corresponding decentered coordinates; Calculate the weights for each type of normalized recognition primitives, i The more reliable the related identification and calculation results of the class identification primitives are, the better i Class corresponding The bigger; The pose detection result of the solution.

7. A robot-based cable plugging device, characterized in that: include: an acquisition module, configured to acquire an input image containing a cable connector, and extract a connector region of the cable connector from the input image; a matching module, configured to identify the connector region, obtain connector features of the cable connector, match an optimal registration strategy based on the connector features, generate a transformation relationship between the connector region and a target connector region based on the optimal registration strategy, and obtain a posture adjustment amount of the cable connector based on the transformation relationship; as well as An adjustment module is configured to adjust the posture of the cable connector according to the posture adjustment amount and insert the cable connector into a target socket, wherein: The identifying the connector area, obtaining the connector features of the cable connector, and matching the optimal registration strategy according to the connector features, includes: From the perspective of target recognition and operation action output, connectors are divided into: connectors with infinite posture directions, connectors with limited posture directions, and connectors with only one posture direction; The output of the detection algorithm can be determined based on the orientation of the connector. The error between the output connector orientation of the detection algorithm and the target connector orientation is expressed as: ； in, The error between the output posture of the detection algorithm and the target posture corresponds to the posture adjustment required for plugging; The connector posture is output by the detection algorithm; is a posture direction of the target connector; is the set of attitude directions of the target connector; is a symbolic function; is the cost function of the robot motion, which measures the robot's movement from posture Movement to posture Without considering the optimization of the robot trajectory in the operation space, ; Based on the focus area of ​​the detection algorithm, cable connectors can be divided into two categories: connectors based on overall morphology detection and connectors based on internal structure detection. Among them, overall morphology-based detection is mainly aimed at connector objects without obvious internal structure. The key to overall morphology-based detection is to define a robust connector pose descriptor and reliably obtain the connector outline and overall connector area. The detection method based on internal structure mainly utilizes the characteristics of connectors in identifying primitives and designs corresponding algorithms based on distribution rules to realize the direction attributes and type attributes of the identification primitives. The direction attributes include undirected and directed, and the type attributes include homogeneous and heterogeneous. The optimal registration strategy is an overall region registration strategy, an overall feature descriptor registration strategy, an undirected isomorphic recognition primitive registration strategy, a directed isomorphic recognition primitive registration strategy, or a heterogeneous recognition primitive registration strategy.

8. The device according to claim 7, characterized in that The matching module further includes: The first transformation unit, when the optimal registration strategy is the overall region registration strategy, the first transformation relationship between the connector region and the target connector region is: in, For the extracted cable connector area, The cable connector area for the target state, is the pixel coordinate of each point, is the homogeneous transformation matrix, It is the cable connector area after the transformation; Respectively for regions and The center coordinates of is the weight coefficient.

9. An electronic device, characterized in that: include: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the program to implement the robot-based cable splicing method according to any one of claims 1 to 6.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: The program is executed by a processor to implement the robot-based cable plugging method according to any one of claims 1 to 6.

Images