Automobile door assembling, adjusting and tightening method and device, computer equipment, readable storage medium and program product
By acquiring the surface difference of the gap before and after tightening and the robot's pose, a target assembly and adjustment model is established. The tightening changes are learned and embedded into the assembly and adjustment process, which solves the problem of low accuracy in the door tightening process and achieves multi-vehicle adaptation and improved assembly and adjustment effect.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SPEEDBOT ROBOTICS CO LTD
- Filing Date
- 2026-03-06
- Publication Date
- 2026-05-19
AI Technical Summary
Existing automatic assembly and adjustment technology has difficulty handling multiple vehicle models and various tightening variations during the door tightening process, resulting in low assembly and adjustment accuracy.
By acquiring the gap surface difference before and after tightening of the previous assembly vehicle, matching the preset tightening change model, establishing the target assembly model, and combining the initial gap surface difference and robot pose, the door assembly and tightening process is carried out. The tightening changes are learned and embedded into the assembly process to ensure the accuracy after tightening.
It improves the accuracy and success rate of door installation and adjustment, adapts to multiple vehicle models and various tightening variations, and ensures the aesthetics and balance requirements of the door and the vehicle body.
Smart Images
Figure CN122059025A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of computer technology, and in particular to a method, apparatus, computer device, computer-readable storage medium, and computer program product for tightening automobile doors. Background Technology
[0002] With the development of automotive manufacturing and automation technologies, the use of vision algorithms to guide industrial robots in tasks such as parts stamping and assembly is gradually replacing traditional manual manufacturing, offering advantages such as high manufacturing efficiency, strong stability, and low labor costs. The automatic assembly and adjustment function for installing car doors and hoods onto the vehicle body also possesses these advantages. However, because process errors cannot be completely eliminated, installing doors and hoods along fixed tracks can lead to misalignment between the doors / hoods and the vehicle body, resulting in a lack of aesthetic appeal and requiring significant manual adjustments.
[0003] However, the principle behind automatic adjustment technology is to use a camera to capture real-time gaps and surface differences between the door cover and the vehicle body, then use algorithms to calculate the required movement of the robot and guide its movement to correct the door cover's installation position. Theoretically, doors installed using automatic adjustment technology can ensure that the gaps and surface differences between the door cover and the vehicle body meet aesthetic and balance requirements. However, in practice, because the door hinges need to be tightened after automatic adjustment, the unobservable physical changes can cause shifts in the overall door position, reducing the actual installation effect when the vehicle leaves the workstation.
[0004] To address this issue, the current common practice is to use a reverse fixed-value compensation method. This method involves setting a reverse compensation value based on manually observed trends and states of the vehicle after tightening. This value is executed at the end of the assembly process to preemptively counteract the effects of subsequent hinge tightening. However, this method uses a single compensation value, making it unsuitable for handling multiple vehicle models and various tightening variations, thus compromising assembly accuracy. Summary of the Invention
[0005] Therefore, it is necessary to provide a method, apparatus, computer equipment, computer-readable storage medium, and computer program product for accurately adjusting and tightening automobile doors, in order to address the aforementioned technical problems.
[0006] In a first aspect, this application provides a method for adjusting and tightening an automobile door, including:
[0007] Obtain the clearance surface difference between the door and body of the previous assembled vehicle before tightening and the clearance surface difference after tightening;
[0008] Based on the clearance surface difference before tightening and the clearance surface difference after tightening, a target tightening variation model is matched in the preset tightening variation model set. The tightening variation model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening variation model in the preset tightening variation model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data.
[0009] Obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference of the assembly and adjustment vehicle before tightening, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in historical data.
[0010] Obtain the initial gap surface difference between the door and the body of the vehicle being assembled and adjusted, and the initial pose of the assembly and adjustment robot;
[0011] Based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, the door assembly and adjustment processing and door tightening processing of the current assembly and adjustment vehicle are carried out through the target assembly and adjustment model to obtain the door assembly and adjustment and tightening processing results of the current assembly and adjustment vehicle.
[0012] In one embodiment, the method further includes:
[0013] Obtain the clearance surface difference before tightening and the corresponding clearance surface difference after tightening of the vehicle in historical data;
[0014] Training data for the tightening model is constructed based on the clearance surface difference before tightening and the corresponding clearance surface difference after tightening.
[0015] The initial tightening variation model is trained using the tightening model training data to obtain a preset set of tightening variation models.
[0016] In one embodiment, the training data for constructing the tightening model based on the clearance surface difference before tightening and the corresponding clearance surface difference after tightening includes:
[0017] Determine the sets of values consisting of the clearance surface difference before tightening and the corresponding clearance surface difference after tightening;
[0018] For each set of values, a first vector is constructed based on the clearance surface difference before tightening, and a second vector is constructed based on the clearance surface difference after tightening, thus obtaining the vector set corresponding to the set of values.
[0019] The training data for the tightening model is constructed based on the vector group.
[0020] In one embodiment, for the Nth tightening variation model in the preset tightening variation model set, when N=1, the training data of the tightening variation model is the tightening model training data, and when N is greater than 1, the training data of the tightening variation model is the external point data of the relationship between the gap surface difference before tightening and the gap surface difference after tightening as defined by the previous tightening model training data.
[0021] In one embodiment, it further includes:
[0022] Retrieve historical data on the assembly and adjustment records of vehicles being assembled and tightened;
[0023] Based on the initial gap surface difference in the assembly and adjustment record data and the corresponding historical pose of the assembly and adjustment robot, training data for the assembly and adjustment model is constructed.
[0024] Based on the assembly and adjustment model training data, the tightening variation model, and the pose mapping function, the initial assembly and adjustment model is trained to obtain the assembly and adjustment model corresponding to the tightening variation model. The pose mapping function is used to establish the mapping relationship between the pose of the assembly and adjustment robot and the gap surface difference.
[0025] In one embodiment, based on the assembly and adjustment model training data, the tightening variation model, and the pose mapping function, the initial assembly and adjustment model is trained to obtain the assembly and adjustment model corresponding to the tightening variation model, including:
[0026] Input the training data of the assembly and adjustment model into the initial assembly and adjustment model to obtain the predicted target pose output by the initial assembly and adjustment model;
[0027] Based on the pose mapping function, determine the pre-tightening clearance surface difference corresponding to the predicted target pose;
[0028] Based on the tightening change model, determine the clearance surface difference after tightening corresponding to the clearance surface difference before tightening;
[0029] The model loss is determined based on the surface difference of the clearance after tightening;
[0030] The parameters of the initial assembly and adjustment model are updated by backpropagation based on the model loss, resulting in the assembly and adjustment model corresponding to the tightening variation model.
[0031] Secondly, this application also provides an automotive door mounting and tightening device, comprising:
[0032] The historical data acquisition module is used to acquire the gap surface difference between the door and the body of the previous assembled vehicle before tightening and the gap surface difference after tightening.
[0033] The tightening model matching module is used to match a target tightening variation model in a preset tightening variation model set based on the clearance surface difference before tightening and the clearance surface difference after tightening. The tightening variation model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening variation model in the preset tightening variation model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data.
[0034] The assembly and adjustment model matching module is used to obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference before tightening of the assembly and adjustment tightening vehicle, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in historical data.
[0035] The assembly and adjustment data acquisition module is used to acquire the initial gap surface difference between the door and the body of the current assembly and adjustment vehicle and the initial pose of the assembly and adjustment robot.
[0036] The assembly and adjustment processing module is used to perform door assembly and adjustment processing and door tightening processing on the current assembly and adjustment vehicle based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, and obtain the door assembly and adjustment tightening processing results of the current assembly and adjustment vehicle.
[0037] Thirdly, this application also provides a computer device, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to perform the following steps:
[0038] Obtain the clearance surface difference between the door and body of the previous assembled vehicle before tightening and the clearance surface difference after tightening;
[0039] Based on the clearance surface difference before tightening and the clearance surface difference after tightening, a target tightening variation model is matched in the preset tightening variation model set. The tightening variation model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening variation model in the preset tightening variation model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data.
[0040] Obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference of the assembly and adjustment vehicle before tightening, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in historical data.
[0041] Obtain the initial gap surface difference between the door and the body of the vehicle being assembled and adjusted, and the initial pose of the assembly and adjustment robot;
[0042] Based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, the door assembly and adjustment processing and door tightening processing of the current assembly and adjustment vehicle are carried out through the target assembly and adjustment model to obtain the door assembly and adjustment and tightening processing results of the current assembly and adjustment vehicle.
[0043] Fourthly, this application also provides a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, performs the following steps:
[0044] Obtain the clearance surface difference between the door and body of the previous assembled vehicle before tightening and the clearance surface difference after tightening;
[0045] Based on the clearance surface difference before tightening and the clearance surface difference after tightening, a target tightening variation model is matched in the preset tightening variation model set. The tightening variation model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening variation model in the preset tightening variation model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data.
[0046] Obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference of the assembly and adjustment vehicle before tightening, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in historical data.
[0047] Obtain the initial gap surface difference between the door and the body of the vehicle being assembled and adjusted, and the initial pose of the assembly and adjustment robot;
[0048] Based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, the door assembly and adjustment processing and door tightening processing of the current assembly and adjustment vehicle are carried out through the target assembly and adjustment model to obtain the door assembly and adjustment and tightening processing results of the current assembly and adjustment vehicle.
[0049] Fifthly, this application also provides a computer program product, including a computer program that, when executed by a processor, performs the following steps:
[0050] Obtain the clearance surface difference between the door and body of the previous assembled vehicle before tightening and the clearance surface difference after tightening;
[0051] Based on the clearance surface difference before tightening and the clearance surface difference after tightening, a target tightening variation model is matched in the preset tightening variation model set. The tightening variation model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening variation model in the preset tightening variation model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data.
[0052] Obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference of the assembly and adjustment vehicle before tightening, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in historical data.
[0053] Obtain the initial gap surface difference between the door and the body of the vehicle being assembled and adjusted, and the initial pose of the assembly and adjustment robot;
[0054] Based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, the door assembly and adjustment processing and door tightening processing of the current assembly and adjustment vehicle are carried out through the target assembly and adjustment model to obtain the door assembly and adjustment and tightening processing results of the current assembly and adjustment vehicle.
[0055] The aforementioned automotive door assembly, adjustment, and tightening method, apparatus, computer equipment, computer-readable storage medium, and computer program product acquire the pre-tightening clearance surface difference and post-tightening clearance surface difference between the door and body of the previous assembly vehicle; based on the pre-tightening clearance surface difference and post-tightening clearance surface difference, match a target tightening change model in a preset tightening change model set; acquire the target assembly and adjustment model corresponding to the target tightening change model; acquire the initial clearance surface difference between the door and body of the current assembly vehicle and the initial pose of the assembly and adjustment robot; based on the initial clearance surface difference and the initial pose of the assembly and adjustment robot, perform door assembly and adjustment processing and door tightening processing on the current assembly vehicle through the target assembly and adjustment model, and obtain the door assembly and adjustment tightening processing result of the current assembly vehicle. This application learns tightening changes through tightening change models, and then establishes the correlation between tightening change models and assembly and adjustment models, embedding tightening changes into the assembly and adjustment process, thereby enabling the assembly and adjustment model to perform assembly and adjustment while ensuring the success rate after tightening, and ensuring the accuracy of automotive door assembly and adjustment. Attached Figure Description
[0056] To more clearly illustrate the technical solutions in the embodiments of this application or related technologies, the drawings used in the description of the embodiments of this application or related technologies will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort.
[0057] Figure 1 This is a diagram illustrating the application environment of a car door assembly and tightening method in one embodiment.
[0058] Figure 2 This is a flowchart illustrating a method for installing, adjusting, and tightening a car door in one embodiment.
[0059] Figure 3 This is a schematic diagram of the car door's posture before and after the door cover is tightened and adjusted in one embodiment.
[0060] Figure 4 This is a schematic diagram of the door posture before and after tightening according to reverse compensation in one embodiment;
[0061] Figure 5 This is a schematic diagram illustrating the tightening of the front and rear door postures based on a network model in one embodiment;
[0062] Figure 6 This is a structural block diagram of a car door mounting and tightening device in one embodiment;
[0063] Figure 7 This is an internal structural diagram of a computer device in one embodiment;
[0064] Figure 8 This is a diagram of the internal structure of a computer device in another embodiment. Detailed Implementation
[0065] To make the objectives, technical solutions, and advantages of this application clearer, the following detailed description is provided in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the scope of this application.
[0066] The automotive door installation and tightening method provided in this application embodiment can be applied to, for example... Figure 1 In the application environment shown, the car body-in-white requiring door assembly is positioned at a designated location along a conveyor line and secured by clamps. The server 102, responsible for assembly control, uses a vision system to capture images of the gap between the door and the car body during the assembly and tightening process. Simultaneously, the assembly robot 104, responsible for door assembly, returns real-time pose data to the server 102. During the assembly process, the server 102 acquires the pre-tightening and post-tightening gap surface differences between the door and the car body of the previous assembled vehicle. Based on these differences, it matches a target tightening variation model from a pre-set set of tightening variation models. Then, it obtains the target assembly model corresponding to the target tightening variation model. Based on the captured images, the initial gap difference between the door and the body of the vehicle being assembled is obtained, along with the initial pose fed back by the assembly robot 104. Using the initial gap difference and the robot's initial pose, the assembly robot 104 is controlled by the target assembly model to perform door assembly and tightening processes on the vehicle being assembled, yielding the door assembly and tightening results. The gap between the processed door and the body is then collected by the vision system and used as input for the next vehicle assembly process. The server 102 can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services.
[0067] In one exemplary embodiment, such as Figure 2 As shown, a method for adjusting and tightening an automotive door is provided, which can be applied to... Figure 1 Taking server 104 as an example, the explanation includes the following steps 201 to 209. Wherein:
[0068] Step 201: Obtain the clearance surface difference between the door and the body of the previously assembled vehicle before tightening and the clearance surface difference after tightening.
[0069] The "previous assembly vehicle" refers to the vehicle on the current production line where the doors have been assembled, adjusted, and tightened. This application specifically uses data from the previous assembly vehicle to guide the assembly and adjustment process of the current vehicle. Regarding the gap difference before and after tightening, the door assembly and adjustment process requires the assembly and adjustment robot 104 to automatically assemble and adjust the door to the corresponding position on the vehicle body. After the automatic assembly and adjustment is complete, the door hinges are tightened and fixed. The gap difference before and after tightening refers to the real-time gap difference values between the door cover and the vehicle body before and after the door hinges are tightened and fixed.
[0070] For example, this application is specifically used for tightening and adjusting control between a car door and the car body on an assembly line. Figure 3 As shown, during the assembly and adjustment of a car door cover, the door's posture is generally adjusted before tightening the bolts. However, due to the tightening action and differences in physical materials, there will be some fluctuations in tightening (the changes are exaggerated in the diagram to show the actual changes; in reality, they are only sub-millimeter or millimeter level, but they can still cause the door's posture to exceed the process requirements after tightening). During the car assembly and adjustment process, although each car is theoretically produced using the same process, in the actual production line, there are many slowly changing systemic factors, such as fixture wear or thermal deformation, changes in ambient temperature and humidity, equipment status drift, and batch differences in materials. These factors do not change randomly but have temporal continuity and sustainability. These factors usually remain relatively stable for a short period (from a few minutes to a few hours), so adjacent vehicles are in highly similar "production environments." Therefore, the data from previous vehicles can be used to guide the tightening of the current car door assembly and adjustment. Therefore, in the solution of this application, when adjusting and tightening the door of the current vehicle, the server 102 first obtains the gap surface difference between the door and the body of the previous vehicle before tightening and the gap surface difference after tightening as guidance data. In a specific embodiment, after the automatic door adjustment step is completed, the database of the server 102 records the value of the gap surface difference between the door and the body taken at the last time after the adjustment movement, recorded as the data before tightening x (each data is a one-dimensional ordered vector containing the measured values of all measuring points arranged in a certain order). After the bolts are tightened, the camera at the same position takes another picture, recorded as the data after tightening y (the format and order are the same as the data before tightening x).
[0071] Step 203: Based on the clearance surface difference before tightening and the clearance surface difference after tightening, match the target tightening change model in the preset tightening change model set. The tightening change model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening change model in the preset tightening change model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle in historical data.
[0072] The preset tightening variation model set contains multiple tightening variation models. The tightening variation model is specifically used to predict the gap surface difference after tightening based on the gap surface difference before tightening. Each tightening variation model in the set is trained based on the gap surface difference before tightening and the gap surface difference after tightening of the vehicle in historical data. The specific training can be carried out in stages, with each part of the historical data corresponding to a tightening variation model.
[0073] For example, this application can specifically model the data relationship between the clearance surface difference before and after tightening using a tightening variation model. Specifically, a tightening variation model is obtained by pre-training a neural network model using a dataset composed of the clearance surface differences before and after tightening. The dataset before tightening is denoted as X1, and the dataset after tightening is denoted as Y1. Using the data {X1, Y1}, a neural network model g1 is learned, such that g1(X1) = Y1. In practical applications, a target tightening variation model that can be used to model the changes in data before and after tightening can be matched based on the clearance surface differences before and after tightening.
[0074] Step 205: Obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference of the assembly and adjustment vehicle before tightening, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model.
[0075] The target assembly and adjustment model is a model for automatically assembling and adjusting the doors of the current assembly and adjustment vehicle. Given the specific deformation patterns guided by the target tightening variation model, it learns how to proactively 'reverse compensate' for this deformation, ensuring the door achieves the ideal gap 'after tightening'. Specifically, it receives the actual state of the current door (gap difference before tightening + current robot pose) as input and outputs a pre-compensated target pose, ensuring the result meets the standard even if known deformation occurs during the door tightening process. However, this model is trained using historical data on the gap difference before tightening of assembly and tightening vehicles, the historical poses of the assembly and adjustment robot, and the gap difference after tightening predicted by the target tightening variation model. It is only effective for production environments that conform to the description of the target tightening variation model and lacks general applicability.
[0076] For example, this application learns tightening variations through a tightening variation model, and then establishes a correlation between the tightening variation model and the assembly and adjustment model, embedding the tightening variations into the assembly and adjustment process. This allows the assembly and adjustment model to perform assembly and adjustment while ensuring a high success rate after tightening. Therefore, for each tightening variation model, an assembly and adjustment model is pre-trained to guide the assembly and adjustment process. After determining the target tightening variation model, the entire assembly and adjustment process can be guided by obtaining the target assembly and adjustment model corresponding to the target tightening variation model.
[0077] Step 207: Obtain the initial gap surface difference between the door and the body of the vehicle being assembled and the initial pose of the assembly robot.
[0078] Step 209: Based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, the door assembly and adjustment processing and door tightening processing of the current assembly and adjustment vehicle are carried out through the target assembly and adjustment model to obtain the door assembly and adjustment tightening processing result of the current assembly and adjustment vehicle.
[0079] For example, after determining the target assembly and adjustment model, the door assembly and tightening process of the current vehicle can be guided by the target assembly and adjustment model. At this time, the initial gap surface difference between the door and the body of the current vehicle and the initial pose of the assembly and adjustment robot can be obtained as initial input data. By inputting the initial gap surface difference and the initial pose of the assembly and adjustment robot into the target assembly and adjustment model, the target assembly and adjustment model will output the target robot pose. Then, the server 102 controls the assembly and adjustment robot 104 to move to the target robot pose to complete the assembly and adjustment. Then, the vision system collects the data after the door assembly and adjustment, thereby recording the data before tightening. Then, the server controls the assembly and adjustment robot to perform the tightening operation and records the data after tightening for use in the decision-making of the next vehicle.
[0080] against Figure 3 The tightening changes shown are generally performed as follows: Figure 4 As shown, by collecting the average value of the door's posture after tightening, the value is back-compensated to the posture before tightening, ensuring that the expected value after tightening is within the process requirements. This application, however, does... Figure 5 As shown, by collecting data on changes before and after tightening, multiple network models of tightening changes are trained, and corresponding assembly and adjustment models are trained. This enables the selection of the most suitable assembly and adjustment model based on the vehicle body and door conditions during assembly and adjustment, thereby increasing the tightening effect within the theoretical process value and ensuring the tightening and adjustment effect.
[0081] The aforementioned automotive door assembly, adjustment, and tightening method, apparatus, computer equipment, computer-readable storage medium, and computer program product acquire the pre-tightening clearance surface difference and post-tightening clearance surface difference between the door and body of the previous assembly vehicle; based on the pre-tightening clearance surface difference and post-tightening clearance surface difference, match a target tightening change model in a preset tightening change model set; acquire the target assembly and adjustment model corresponding to the target tightening change model; acquire the initial clearance surface difference between the door and body of the current assembly vehicle and the initial pose of the assembly and adjustment robot; based on the initial clearance surface difference and the initial pose of the assembly and adjustment robot, perform door assembly and adjustment processing and door tightening processing on the current assembly vehicle through the target assembly and adjustment model, and obtain the door assembly and adjustment tightening processing result of the current assembly vehicle. This application learns tightening changes through tightening change models, and then establishes the correlation between tightening change models and assembly and adjustment models, embedding tightening changes into the assembly and adjustment process, thereby enabling the assembly and adjustment model to perform assembly and adjustment while ensuring the success rate after tightening, and ensuring the accuracy of automotive door assembly and adjustment.
[0082] In an exemplary embodiment, the method further includes: acquiring the pre-tightening clearance surface difference and the corresponding post-tightening clearance surface difference of the assembled and tightened vehicle from historical data; constructing tightening model training data based on the pre-tightening clearance surface difference and the corresponding post-tightening clearance surface difference; and training the initial tightening variation model using the tightening model training data to obtain a preset tightening variation model set.
[0083] For example, this application also includes a training process for a preset set of tightening variation models. This process requires first collecting historical data on the pre-tightening clearance surface difference and the corresponding post-tightening clearance surface difference of the assembled and tightened vehicle. Then, this data is used as training data to train the initial tightening variation model, resulting in various tightening variation models in the preset set of tightening variation models. In this embodiment, by collecting historical data on the pre-tightening clearance surface difference and the corresponding post-tightening clearance surface difference of the assembled and tightened vehicle, historical data can be effectively collected to complete the training process of the models in the preset set of tightening variation models, thereby effectively modeling the change in clearance surface difference before and after tightening and ensuring the accuracy of door assembly and tightening.
[0084] Furthermore, the training data for constructing the tightening model based on the clearance surface difference before tightening and the corresponding clearance surface difference after tightening includes: determining each numerical group consisting of the clearance surface difference before tightening and the corresponding clearance surface difference after tightening; for each numerical group, constructing a first vector based on the clearance surface difference before tightening and a second vector based on the clearance surface difference after tightening, thus obtaining the vector group corresponding to the numerical group; and constructing the training data for the tightening model based on the vector group.
[0085] For example, for the data on the clearance surface difference before tightening, each data point is a set of values, which can be transformed into a one-dimensional ordered vector containing the measured values of all measuring points arranged in a certain order. Similarly, the set of values for the clearance surface difference after tightening can also be transformed into a second vector, with the same format and order as the data before tightening. This results in vector sets corresponding to the sets of values for the clearance surface difference before tightening and the corresponding clearance surface difference after tightening. These vector sets can then be used as training data for the tightening model.
[0086] Furthermore, for the Nth tightening variation model in the preset tightening variation model set, when N=1, the training data of the tightening variation model is the tightening model training data; when N is greater than 1, the training data of the tightening variation model is the external point data of the relationship between the gap surface difference before tightening and the gap surface difference after tightening as defined by the previous tightening model training data.
[0087] For example, after collecting a certain amount of data, the data set before tightening is denoted as X1, and the data set after tightening is denoted as Y1. The data {X1, Y1} combination is used as the training data for the tightening model. Then, a neural network model g1 is learned by training a neural network. That is, when N=1, the training data for the tightening variation model is the tightening model training data, such that g1(X1)=Y1. After this stage of training is completed, it can be checked that the subset {X2, Y2} in the training data {X1, Y1} combination set belongs to the outside point of the relationship g1(X1)=Y1, that is, ||g1(X2)-Y2|| is greater than a certain preset value ε. The training data {X2, Y2} is used to train the second network model g2. That is, when N is greater than 1, the training data for the tightening variation model is the outside point data in the previous tightening model training data, relative to the relationship between the gap surface difference before tightening and the gap surface difference after tightening defined by the previous tightening variation model. This process continues until multiple network models {g1, g2, g3...} are learned that cover the initial training set. That is, for any {x, y} belonging to {X1, Y1}, there exists a g belonging to {g1, g2, g3...} such that ||g(x) - y|| < ε. In this embodiment, a segmented training method is used to train each tightening variation model in the preset tightening variation model set. This effectively covers various correspondences between the gap surface difference before and after tightening, thus ensuring the accuracy of the tightening process relationship identification.
[0088] In an exemplary embodiment, the method further includes: acquiring assembly and adjustment record data of the vehicle in historical data; constructing assembly and adjustment model training data based on the initial gap surface difference and the corresponding historical pose of the assembly and adjustment robot in the assembly and adjustment record data; and performing model training processing on the initial assembly and adjustment model based on the assembly and adjustment model training data, the tightening change model and the pose mapping function to obtain the assembly and adjustment model corresponding to the tightening change model, wherein the pose mapping function is used to establish the mapping relationship between the pose of the assembly and adjustment robot and the gap surface difference.
[0089] Specifically, the pose mapping function is used to establish the mapping relationship between the pose of the assembly robot and the gap surface difference. Given any robot coordinate b in the known area of the standard installation position of the door in space, it is possible to capture and obtain the gap surface difference value a at each location. That is, there exists an injective relationship h(b) = a in this area. This injective relationship h(b) = a corresponds to the pose mapping function.
[0090] For example, before applying the assembly and adjustment model to control the assembly and adjustment process, it is necessary to complete the training process of the assembly and adjustment model. At this time, the assembly and adjustment record data of the tightening vehicle in the historical data can be obtained first. Then, the initial gap surface difference and the corresponding historical pose of the assembly and adjustment robot contained therein can be used as model input to construct the assembly and adjustment model training data. During the subsequent training process, the assembly and adjustment model training data, the tightening change model, and the pose mapping function training base data can be used. Among them, the tightening change model and the pose mapping function are fixed during the training process. By inputting the assembly and adjustment model training data into the initial assembly and adjustment model, the parameters of the initial assembly and adjustment model are adjusted to obtain the assembly and adjustment model corresponding to the required tightening change model. For each tightening change model in the preset tightening change model set, a corresponding assembly and adjustment model can be trained in the above manner. During the assembly and adjustment process, the assembly and adjustment model can be used to control the assembly and adjustment robot's assembly and adjustment process with the tightened gap surface difference as the posterior, so as to ensure the assembly and adjustment effect.
[0091] In an exemplary embodiment, the assembly and adjustment model is trained based on the assembly and adjustment model training data, the tightening variation model, and the pose mapping function to obtain the assembly and adjustment model corresponding to the tightening variation model. This process includes: inputting the assembly and adjustment model training data into the initial assembly and adjustment model to obtain the predicted target pose output by the initial assembly and adjustment model; determining the pre-tightening clearance surface difference corresponding to the predicted target pose according to the pose mapping function; determining the post-tightening clearance surface difference corresponding to the pre-tightening clearance surface difference according to the tightening variation model; determining the model loss based on the post-tightening clearance surface difference; and performing backpropagation parameter update processing on the initial assembly and adjustment model based on the model loss to obtain the assembly and adjustment model corresponding to the tightening variation model.
[0092] For example, in the model training process, the parameters of the tightening variation model and the pose mapping function are fixed. First, the training data of the assembly and adjustment model can be input into the initial assembly and adjustment model, which predicts the robot's pose data during the assembly and adjustment process, obtaining the predicted target pose output by the initial assembly and adjustment model. Then, since the pose mapping function establishes the mapping relationship between the robot's pose and the clearance surface difference, the pre-tightening clearance surface difference corresponding to the predicted target pose before tightening can be determined through the pose mapping function after assembly and adjustment. The tightening variation model models the data relationship between the pre-tightening clearance surface difference and the post-tightening clearance surface difference; therefore, the post-tightening clearance surface difference corresponding to the pre-tightening clearance surface difference can be determined through the tightening variation model. Then, by comparing the predicted actual post-tightening clearance surface difference with the ideal post-tightening clearance surface difference, the model loss can be obtained; finally, the parameters of the initial assembly and adjustment model are updated through backpropagation based on the model loss, resulting in the assembly and adjustment model corresponding to the tightening variation model. For this process:
[0093] In the past, without considering the tightening deformation, the assembly and adjustment model input was the current clearance surface difference a1 and the robot coordinate value b1. It would optimize and reduce the loss value of the deviation between the clearance surface difference a2=h(b2) and the theoretical clearance surface difference a0 when the robot moves to coordinate b2, that is, min||a2-a0||=min||h(b2)-a0||=min||h(f(a1,b1))-a0||.
[0094] Considering the tightening deformation, the tightening variation model is substituted into the equation. To optimize the deviation between the tightened clearance surface difference and the theoretical clearance surface difference, for each g belonging to {g1,g2,g3...}, the corresponding assembly and adjustment model {f1,f2,f3...} can be trained to minimize ||g(h(f(a1,b1)))-a0||, thereby ensuring the tightening effect. In this embodiment, the predicted target pose is mapped to the clearance surface difference before tightening using a pose mapping function. Then, the tightening variation model is used to determine the clearance surface difference after tightening, thereby estimating the model loss and completing the model backpropagation process, thus effectively ensuring the efficiency and effectiveness of model training.
[0095] It should be understood that although the steps in the flowcharts of the above embodiments are shown sequentially according to the arrows, these steps are not necessarily executed in the order indicated by the arrows. Unless explicitly stated herein, there is no strict order restriction on the execution of these steps, and they can be executed in other orders. Moreover, at least some steps in the flowcharts of the above embodiments may include multiple steps or multiple stages. These steps or stages are not necessarily completed at the same time, but can be executed at different times. The execution order of these steps or stages is not necessarily sequential, but can be performed alternately or in turn with other steps or at least some of the steps or stages of other steps.
[0096] Based on the same inventive concept, this application also provides an automobile door assembly and tightening device for implementing the automobile door assembly and tightening method described above. The solution provided by this device is similar to the solution described in the above method. Therefore, the specific limitations of one or more embodiments of the automobile door assembly and tightening device provided below can be found in the limitations of the automobile door assembly and tightening method described above, and will not be repeated here.
[0097] In one exemplary embodiment, such as Figure 6 As shown, a car door mounting and tightening device is provided, comprising:
[0098] The historical data acquisition module 601 is used to acquire the gap surface difference between the door and the body of the previous assembled vehicle before tightening and the gap surface difference after tightening.
[0099] The tightening model matching module 603 is used to match a target tightening variation model in a preset tightening variation model set based on the clearance surface difference before tightening and the clearance surface difference after tightening. The tightening variation model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening variation model in the preset tightening variation model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data.
[0100] The assembly and adjustment model matching module 605 is used to obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference before tightening of the assembly and adjustment tightening vehicle, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in historical data.
[0101] The assembly and adjustment data acquisition module 607 is used to acquire the initial gap surface difference between the door and the body of the current assembly and adjustment vehicle and the initial pose of the assembly and adjustment robot.
[0102] The assembly and adjustment processing module 609 is used to perform door assembly and adjustment processing and door tightening processing on the current assembly and adjustment vehicle based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, and to obtain the door assembly and adjustment tightening processing result of the current assembly and adjustment vehicle.
[0103] In one embodiment, the system further includes a tightening variation model training module, which is used to: acquire the pre-tightening clearance surface difference and the corresponding post-tightening clearance surface difference of the assembled and tightened vehicle in historical data; construct tightening model training data based on the pre-tightening clearance surface difference and the corresponding post-tightening clearance surface difference; and train the initial tightening variation model using the tightening model training data to obtain a preset tightening variation model set.
[0104] In one embodiment, the tightening variation model training module is specifically used to: determine each numerical group consisting of the clearance surface difference before tightening and the corresponding clearance surface difference after tightening; for each numerical group, construct a first vector based on the clearance surface difference before tightening and construct a second vector based on the clearance surface difference after tightening to obtain the vector group corresponding to the numerical group; and construct tightening model training data based on the vector group.
[0105] In one embodiment, for the Nth tightening variation model in the preset tightening variation model set, when N=1, the training data of the tightening variation model is the tightening model training data; when N is greater than 1, the training data of the tightening variation model is the external point data of the relationship between the pre-tightening clearance surface difference and the post-tightening clearance surface difference defined by the previous tightening model training data.
[0106] In one embodiment, the system further includes an assembly and adjustment model training module, which is used to: acquire assembly and adjustment record data of the vehicle being assembled and tightened from historical data; construct assembly and adjustment model training data based on the initial gap surface difference and the corresponding historical pose of the assembly and adjustment robot in the assembly and adjustment record data; and perform model training processing on the initial assembly and adjustment model based on the assembly and adjustment model training data, the tightening change model, and the pose mapping function to obtain the assembly and adjustment model corresponding to the tightening change model. The pose mapping function is used to establish the mapping relationship between the pose of the assembly and adjustment robot and the gap surface difference.
[0107] In one embodiment, the assembly and adjustment model training module is specifically used for: inputting the assembly and adjustment model training data into the initial assembly and adjustment model to obtain the predicted target pose output by the initial assembly and adjustment model; determining the pre-tightening clearance surface difference corresponding to the predicted target pose according to the pose mapping function; determining the post-tightening clearance surface difference corresponding to the pre-tightening clearance surface difference according to the tightening change model; determining the model loss based on the post-tightening clearance surface difference; and performing backpropagation parameter update processing on the initial assembly and adjustment model according to the model loss to obtain the assembly and adjustment model corresponding to the tightening change model.
[0108] The various modules in the aforementioned automotive door mounting and tightening device can be implemented entirely or partially through software, hardware, or a combination thereof. These modules can be embedded in or independent of the processor in a computer device, or stored in the memory of a computer device as software, so that the processor can call and execute the corresponding operations of each module.
[0109] In one exemplary embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as follows: Figure 7 As shown, the computer device includes a processor, memory, input / output (I / O) interfaces, and a communication interface. The processor, memory, and I / O interfaces are connected via a system bus, and the communication interface is also connected to the system bus via the I / O interfaces. The processor provides computational and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system, computer programs, and a database. The internal memory provides the environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The database stores various data related to the installation, adjustment, and tightening of automotive doors. The I / O interfaces are used for information exchange between the processor and external devices. The communication interface is used for communication with external terminals via a network connection. When the computer program is executed by the processor, it implements a method for installing, adjusting, and tightening automotive doors.
[0110] In one exemplary embodiment, a computer device is provided, which may be a terminal, and its internal structure diagram may be as follows: Figure 8As shown, the computer device includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, Near Field Communication (NFC), or other technologies. When the computer program is executed by the processor, it implements a method for tightening and adjusting a car door. The display unit is used to form a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device of the computer device can be a touch layer covering the display screen, or buttons, trackballs, or touchpads set on the casing of the computer device, or external keyboards, touchpads, or mice, etc.
[0111] Those skilled in the art will understand that Figure 7 and Figure 8 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.
[0112] In one embodiment, a computer device is also provided, including a memory and a processor, wherein the memory stores a computer program, and the processor executes the computer program to implement the steps in the above method embodiments.
[0113] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, implements the steps in the above method embodiments.
[0114] In one embodiment, a computer program product or computer program is provided, the computer program product or computer program including computer instructions stored in a computer-readable storage medium. A processor of a computer device reads the computer instructions from the computer-readable storage medium, and executes the computer instructions, causing the computer device to perform the steps in the above method embodiments.
[0115] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, data stored, data displayed, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of the relevant data must comply with relevant regulations.
[0116] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The computer program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, databases, or other media used in the embodiments provided in this application can include at least one of non-volatile memory and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetic random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM). The databases involved in the embodiments provided in this application may include at least one type of relational database and non-relational database. Non-relational databases may include, but are not limited to, blockchain-based distributed databases. The processors involved in the embodiments provided in this application may be general-purpose processors, central processing units, graphics processing units, digital signal processors, programmable logic devices, artificial intelligence (AI) processors, etc., and are not limited to these.
[0117] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this application.
[0118] The embodiments described above are merely illustrative of several implementation methods of this application, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this application should be determined by the appended claims.
Claims
1. A method for adjusting and tightening a car door, characterized in that, The method includes: Obtain the clearance surface difference between the door and body of the previous assembled vehicle before tightening and the clearance surface difference after tightening; Based on the clearance surface difference before tightening and the clearance surface difference after tightening, a target tightening change model is matched in a preset tightening change model set. The tightening change model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening change model in the preset tightening change model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data. Obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference of the assembly and adjustment tightening vehicle before tightening, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in the historical data. Obtain the initial gap surface difference between the door and the body of the vehicle being assembled and adjusted, and the initial pose of the assembly and adjustment robot; Based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, the door assembly and adjustment process and door tightening process of the current assembly and adjustment vehicle are performed through the target assembly and adjustment model to obtain the door assembly and adjustment and tightening process result of the current assembly and adjustment vehicle.
2. The method according to claim 1, characterized in that, The method further includes: Obtain the clearance surface difference before tightening and the corresponding clearance surface difference after tightening of the vehicle from historical data; Based on the clearance surface difference before tightening and the corresponding clearance surface difference after tightening, tightening model training data is constructed; The initial tightening variation model is trained using the tightening model training data to obtain a preset tightening variation model set.
3. The method according to claim 2, characterized in that, The training data for constructing the tightening model based on the clearance surface difference before tightening and the corresponding clearance surface difference after tightening includes: Determine the sets of values consisting of the clearance surface difference before tightening and the corresponding clearance surface difference after tightening; For each set of values, a first vector is constructed based on the clearance surface difference before tightening, and a second vector is constructed based on the clearance surface difference after tightening, thus obtaining the vector set corresponding to the set of values. Training data for the tightening model is constructed based on the vector set.
4. The method according to claim 1, characterized in that, For the Nth tightening variation model in the preset tightening variation model set, when N=1, the training data of the tightening variation model is the tightening model training data. When N is greater than 1, the training data of the tightening variation model is the external point data of the relationship between the gap surface difference before tightening and the gap surface difference after tightening as defined by the previous tightening model training data.
5. The method according to any one of claims 1 to 4, characterized in that, Also includes: Retrieve historical data on the assembly and adjustment records of vehicles being assembled and tightened; Based on the initial gap surface difference in the assembly and adjustment record data and the corresponding historical pose of the assembly and adjustment robot, the assembly and adjustment model training data is constructed. Based on the assembly and adjustment model training data, the tightening variation model, and the pose mapping function, the initial assembly and adjustment model is trained to obtain the assembly and adjustment model corresponding to the tightening variation model. The pose mapping function is used to establish the mapping relationship between the pose of the assembly and adjustment robot and the gap surface difference.
6. The method according to claim 5, characterized in that, The step of training the initial assembly and adjustment model based on the assembly and adjustment model training data, the tightening variation model, and the pose mapping function to obtain the assembly and adjustment model corresponding to the tightening variation model includes: The training data of the assembly and adjustment model is input into the initial assembly and adjustment model to obtain the predicted target pose output by the initial assembly and adjustment model; Based on the pose mapping function, determine the pre-tightening clearance surface difference corresponding to the predicted target pose; Based on the tightening variation model, determine the gap surface difference after tightening corresponding to the gap surface difference before tightening; The model loss is determined based on the gap surface difference after tightening; The parameters of the initial assembly and adjustment model are updated by backpropagation based on the model loss to obtain the assembly and adjustment model corresponding to the tightening variation model.
7. A car door mounting and tightening device, characterized in that, The device includes: The historical data acquisition module is used to acquire the gap surface difference between the door and the body of the previous assembled vehicle before tightening and the gap surface difference after tightening. The tightening model matching module is used to match a target tightening variation model in a preset tightening variation model set based on the clearance surface difference before tightening and the clearance surface difference after tightening. The tightening variation model is used to predict the clearance surface difference after tightening based on the clearance surface difference before tightening. Each tightening variation model in the preset tightening variation model set is trained based on the clearance surface difference before tightening and the clearance surface difference after tightening of the vehicle being assembled and tightened in historical data. The assembly and adjustment model matching module is used to obtain the target assembly and adjustment model corresponding to the target tightening change model. The target assembly and adjustment model is trained based on the gap surface difference before tightening of the assembly and adjustment tightening vehicle, the historical pose of the assembly and adjustment robot, and the gap surface difference after tightening predicted by the target tightening change model in the historical data. The assembly and adjustment data acquisition module is used to acquire the initial gap surface difference between the door and the body of the current assembly and adjustment vehicle and the initial pose of the assembly and adjustment robot. The assembly and adjustment processing module is used to perform door assembly and adjustment processing and door tightening processing on the current assembly and adjustment vehicle based on the initial gap surface difference and the initial pose of the assembly and adjustment robot, and to obtain the door assembly and adjustment tightening processing result of the current assembly and adjustment vehicle.
8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that, When the processor executes the computer program, it implements the steps of the method according to any one of claims 1 to 6.
9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.
10. A computer program product, comprising a computer program, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 6.