Video generation method and electronic device

Abstract

The application discloses a video generation method and an electronic device, and relates to the technical field of operation and maintenance, comprising the following steps: creating a component index table according to bill of materials data, binding model nodes of an initial three-dimensional model with component identifiers, reorganizing the structure of the initial three-dimensional model and calibrating a motion reference based on structure standardization rules and coordinate correction rules, and finally fusing disassembly rules, control instructions and disassembly process driving target three-dimensional models to execute precise disassembly actions, so as to solve the problems of model penetration, action deviation, dependence on artificial repeated debugging and data island unable to close loop caused by model and data separation in related technologies, realize full-process automation from configuration to video generation, reduce the operation threshold of non-professionals, ensure that the disassembly actions conform to industrial assembly logic and spatial physical constraints, and significantly improve the efficiency, consistency and reusability of video production.

Description

Technical Field

[0001] This application relates to the field of product operation and maintenance technology, and in particular to a video generation method and electronic device. Background Technology

[0002] As industrial products become increasingly sophisticated and assembly processes become more complex, the installation alignment compensation in related technologies employs static visual pre-positioning before the installation tool contacts the workpiece. This requires visual photography, deviation calculation, and coordinate correction of the installation tool before contact, with the entire process planned and executed manually for each compensation step.

[0003] However, the static pre-positioning of related technologies cannot cope with the microscopic deformation and slippage of the workpiece after contact, and lacks dynamic changes in the contact process. Summary of the Invention

[0004] This application provides a video generation method and electronic device to at least solve the problems in the related art of disassembly and assembly video production, which rely on manual segmentation, are inefficient, cannot be used by non-professionals, and are difficult to generate in batches of high-quality videos that meet industrial assembly specifications.

[0005] This application provides a video generation method, including: Based on the configuration information of the target product's operation and maintenance requirements, obtain the bill of materials data of the target product from the bill of materials system, and create a component index table corresponding to the target product based on the bill of materials data; Based on the component identifier and component hierarchy in the component index table, the initial 3D model corresponding to the target product is retrieved from the model library, and the model nodes in the initial 3D model are bound to the component identifier; Based on structural standardization rules and coordinate correction rules, the initial 3D model is structurally reorganized and motion reference is calibrated to obtain a target 3D model that matches the product structure of the target product. Based on the disassembly and assembly rules of the target product, as well as the control instructions and disassembly and assembly procedures for the operation and maintenance requirements of the target product, the target 3D model is controlled to perform disassembly and assembly actions to generate disassembly and assembly videos of the target product.

[0006] This application also provides an electronic device, including: a memory for storing a computer program; and a processor for executing the computer program to implement the steps of any of the above-described video generation methods.

[0007] This application solves the problems of model separation, motion deviation, reliance on repeated manual debugging, and data silos that occur in related technologies due to the separation of model and data. It achieves full automation from configuration to video generation, reduces the operating threshold for non-professionals, ensures that the disassembly and assembly actions conform to industrial assembly logic and spatial physical constraints, and significantly improves the efficiency, consistency, and reusability of video production. By creating a component index table based on bill of materials data and binding model nodes with component identifiers based on structural standardization rules and coordinate correction rules, and finally integrating disassembly and assembly rules, user control commands, and disassembly and assembly processes to drive the model to perform precise disassembly and assembly actions, it can solve the problems of model separation, motion deviation, reliance on repeated manual debugging, and data silos that cannot be closed-loop in the entire process. Attached Figure Description

[0008] To more clearly illustrate the embodiments of this application, the accompanying drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 A flowchart illustrating the first video generation method provided in this application embodiment; Figure 2 A flowchart illustrating the second video generation method provided in this application embodiment; Figure 3 A flowchart illustrating the third video generation method provided in this application embodiment; Figure 4 This is a schematic diagram of the structure of a specific video generation system provided in an embodiment of this application; Figure 5 A schematic diagram illustrating a specific video generation method provided in an embodiment of this application; Figure 6 This is a schematic diagram of the structure of a video generation device provided in an embodiment of this application. Detailed Implementation

[0010] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the protection scope of this application.

[0011] It should be noted that, in the description of this application, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. The terms "first," "second," etc., in this application are used to distinguish similar objects and are not used to describe a specific order or sequence.

[0012] With the rapid iteration of industrial products and the increasing number of detachable components, there is an urgent need for standardized disassembly and assembly videos. Currently, the production of disassembly and assembly videos for related technologies relies entirely on manual, segmented operations: technicians must manually extract the disassembly and assembly steps from process documents, animators complete model import, layer organization, material and animation production, and rendering output in 3D software, and voice-over and subtitles are added later. The entire process involves many steps and a long cycle, making it difficult to meet the industrial demand for rapid, batch video generation. Furthermore, non-professionals cannot perform video production, and scalability is poor.

[0013] Specifically, taking server products as an example, the production and release of disassembly and assembly videos in related technologies also rely on the aforementioned manual segmentation process: technicians extract disassembly and assembly steps from process documents, animators complete model organization, animation production and rendering in 3D software, and then add voice-over and subtitles using post-production software; the videos need to be manually uploaded to PLM for review, and only after approval can they be manually uploaded to the official website for release. Furthermore, the visualization canvas platform in related technologies can only perform simple text and image layout, does not link with 3D software, and the generated content must rely on external images, unable to drive 3D models. AI natural language control, due to the lack of integration with coordinate, assembly, and collision rules, can only simply recognize actions, resulting in large execution deviations and requiring repeated attempts.

[0014] To address the shortcomings of the aforementioned related technologies, this application establishes an index mapping between bill of materials data and 3D model nodes to achieve precise binding between component identities and model nodes; it also performs structural reorganization and motion benchmark calibration on the 3D model to eliminate hierarchical confusion and coordinate deviations; and it integrates disassembly and assembly rules, user natural language control commands, and disassembly and assembly processes, driving the 3D model to perform precise disassembly and assembly actions that conform to physical constraints through a rule engine. This achieves full-process automation from project configuration and model processing to video generation, effectively solving the problems of low production efficiency, large molding deviations, data silos, and inability for non-professionals to operate in related technologies.

[0015] It is understood that the applicability of the video generation method in this application is not limited to the operation and maintenance field, i.e., the operation and maintenance field of the target product, such as server operation and maintenance. This application, by replacing the underlying bill of materials database and 3D model library, can be equivalently extended to precision mechanical assembly fields such as medical surgical instrument calibration, aerospace engine repair, new energy battery pack disassembly, and smart home appliance after-sales guidance, demonstrating broad cross-industry applicability. Similarly, the disassembly and assembly videos in this application are not limited to planar videos; they can also be synchronized in real-time to AR, VR, or MR devices. Through virtual-real integration technology, the disassembly and assembly videos are precisely superimposed on real physical entities, realizing real-time guided maintenance operations based on augmented reality, improving the intuitiveness and accuracy of on-site operations.

[0016] To enable those skilled in the art to better understand the present application, the present application will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0017] Figure 1 This is a schematic flowchart illustrating the first video generation method provided in an embodiment of this application. Figure 1 As shown, the method includes the following steps: Step 101: Based on the configuration information of the target product's operation and maintenance requirements, obtain the bill of materials data of the target product from the bill of materials system, and create a component index table corresponding to the target product based on the bill of materials data.

[0018] In some embodiments, this application provides a unified target workbench (such as an infinite canvas workbench) as the user's operation entry point. The user first enters relevant information about the maintenance project into the target workbench, including product model, video type, project manager, etc.; after the information is entered completely and passes verification, the system confirms the maintenance project creation is complete. Subsequently, the user uploads configuration information for the target product (e.g., model specifications, component selection list, etc.) through the target workbench; the system initiates a precise query to the bill of materials system based on this configuration information to obtain the corresponding complete bill of materials data. Finally, the system creates a component index table based on the bill of materials data. This process ensures that all subsequent operations are based on accurate product data bound to the current maintenance project, avoiding data mismatch or omissions.

[0019] Among them, the target workbench is an integrated operation platform that provides a visual interactive interface and supports functions such as project information entry, configuration upload, model preview, command input and video generation.

[0020] A Bill of Material (BOM) system is a system that stores data on the material composition of a product, providing information such as product structure, component attributes, and assembly relationships.

[0021] Configuration information is data describing the specific hardware composition of the target product, used to accurately match the corresponding BOM version in the bill of materials system, and includes at least the complete machine model and a list of optional components.

[0022] The bill of materials (BOM) data is a set of data describing the composition structure of the target product, including but not limited to the unique code of the whole machine, the material codes of each level, the standard name of the component, the component type, the material attributes, the quantity, the installation location, the assembly relationship, the parent-child relationship, the disassembly and assembly sequence, the disassembly and assembly constraints, the standard installation torque, whether it is a vulnerable part, and whether it supports hot-swapping, etc.

[0023] The component index table is a structured data table generated based on the bill of materials data, with component identifiers (such as unique IDs) as the primary key. It is used to establish a fast mapping relationship between material information and 3D model nodes.

[0024] Step 102: Based on the component identifier and component hierarchy in the component index table, retrieve the initial 3D model corresponding to the target product from the model library. The model nodes in the initial 3D model are bound to the component identifier.

[0025] In some embodiments, based on the component identifiers and hierarchical relationships in the component index table, the initial 3D model (typically in engineering formats such as STP or ASM) corresponding to the target product is located and retrieved from the model library. Since the part objects (model nodes) in the initial 3D model are not associated with the components in the target product, this application can bind each model node to a component identifier, so that each model node has a corresponding component identifier, laying the foundation for subsequent precise control and animation generation.

[0026] The initial 3D model is a raw engineering model obtained directly from the model library without standardization. Its model tree hierarchy may be chaotic, the coordinate reference may be inconsistent, and it is not associated with the components in the target product.

[0027] Model nodes correspond to the digital representation of each component (such as a screw, a hard drive, or a fan) in the target product in a 3D environment. Each model node not only records the name of its corresponding component, its spatial position in the whole machine, and its current rotation angle and other geometric attributes, but also reflects the assembly hierarchy between components through a tree-like hierarchical structure (for example, multiple screw nodes are attached under a heatsink node), thereby realizing a one-to-one mapping and precise control between physical components and their 3D digital models.

[0028] Understandably, after obtaining the initial 3D model, this application can load the initial 3D model into the central display area of ​​the target workbench at once, and through a real-time two-way synchronization mechanism, completely synchronize the entire model to the 3D software environment. The synchronization includes all model nodes, hierarchical structure, coordinate information, display status, and layer information, ensuring that the target workbench and the 3D software present a completely consistent model form. Users do not need to manually open the 3D software, manually import the model, or adjust the view; both ends remain consistent at all times. Furthermore, the system establishes a one-to-one mapping relationship between component identifiers in the component index table and corresponding model nodes in the 3D software, thereby enabling quick location and operation of the corresponding model nodes through component identifiers.

[0029] Step 103: Based on the structural standardization rules and coordinate correction rules, the initial three-dimensional model is restructured and the motion reference is calibrated to obtain a target three-dimensional model that matches the product structure of the target product.

[0030] In some embodiments, the initial 3D model may have issues such as redundant objects, disordered hierarchy, lack of parent-child constraints, offset coordinate center, and undefined rotation pivot points, making it unsuitable for direct generation of automated disassembly and assembly videos. Therefore, this application uses structural standardization rules and coordinate correction rules to restructure and calibrate the motion reference of the bound initial 3D model. This ensures that the adjusted target 3D model has a clear hierarchical structure, a unified motion reference, and clear component relationships, enabling it to be directly invoked by the motion rule engine to generate disassembly and assembly videos without clipping or offset.

[0031] Specifically, regarding structural standardization rules, this application can restructure the initial 3D model based on structural standardization rules. Structural restructuring includes at least one of the following: cleaning up redundant objects, reconstructing the model hierarchy, setting a rigid structure for the model, and establishing subordinate constraint relationships between model nodes.

[0032] Cleaning up redundant objects can be achieved by reviewing all model nodes, labels, constraints, and layers in the initial 3D model. It can automatically identify and delete empty model nodes, invalid labels, incorrect constraints, garbage nodes, and unnecessary auxiliary model nodes, thereby reducing the system's computational burden and avoiding interference with subsequent identification.

[0033] The model hierarchy can be reorganized according to a dedicated structure for disassembly and assembly videos. This involves uniquely designating the entire machine assembly as the top-level root node, retaining only one root node globally, and extending all structures, components, and parts downwards to ensure unified management of the entire machine structure. First-level child nodes are independent, fully functional component modules that can be disassembled and installed individually, such as hard drive modules, PCIe adapter modules, fan modules, power supplies, and memory. First-level nodes are parallel and not nested or subordinate to each other, with no overlapping levels. Their business relationships are established solely through spatial location and assembly sequence, without hierarchical subordination. Second-level child nodes are scattered parts attached to first-level nodes, cannot be separated independently, and belong to auxiliary connectors / fasteners / small auxiliary parts.

[0034] Setting the model to a rigid structure can automatically set non-disassembly structures (such as shells and shielding components) that are not involved in operation and maintenance movement as fixed bodies, excluding them from motion calculations.

[0035] Establishing subordinate constraint relationships between model nodes can automatically create subordinate (also known as parent-child) binding relationships for components and parts with subordinate relationships, such as heat sinks and fixing screws, fans and clips, hard drives and screws, to ensure synchronous movement during disassembly and assembly, and to prevent abnormalities such as separation, misalignment, or loss.

[0036] Specifically, regarding coordinate correction rules, this application can perform motion datum calibration on the initial 3D model based on the coordinate correction rules. The motion datum calibration includes at least one of resetting the coordinate center of the model node, aligning the local coordinate axes of the model node to the global coordinate system, and resetting the rotation center point of the rotating model node.

[0037] Resetting the coordinate center of model nodes allows for a unified coordinate center reset operation on all model nodes, discarding the original modeling coordinates and redefining the component pivot points based on the requirements of assembly and disassembly operations. All model nodes corresponding to static fixed structures, shells without movement, and fixed supports will have their coordinate centers uniformly reset to the coordinates corresponding to the component center, ensuring consistent and standardized coordinates for static components. Resetting the rotation center point of rotating model nodes (e.g., model objects with rotational motion attributes such as screws, shafts, and hinges, which require setting a unique rotation pivot point) refers to resetting the rotation center point of each model node corresponding to all moving parts with rotational motion, flipping, opening, or swinging motion. This application can automatically extract the rotation reference point as the unique rotation center based on the bill of materials data and the coaxial hole positions and shaft structures of the initial 3D model, preventing motion eccentricity. For model nodes corresponding to rotating parts such as screws and knobs, the center of the coaxial assembly hole, the center of the cylindrical mating axis, and the hinge mounting reference point are used as the unique rotation center point. For model nodes corresponding to unilateral opening, closing, or swinging parts, the assembly edge line on the fixed side of the part and the center point of the hinge fixed end are defined as the rotation pivot. For model nodes corresponding to rotating components without a clear shaft structure, the fixed rotation reference point of the component's reciprocating motion is calculated based on the overall assembly limit relationship and disassembly / assembly trajectory, and used as the custom rotation center. Once the rotation center point is determined, it is permanently fixed and bound to the model node of the corresponding component, serving as the sole motion reference for subsequent rotation, angle adjustment, torque simulation, and opening and closing actions, preventing rotational eccentricity, action misalignment, and rotational offset.

[0038] Aligning the local coordinate axes of model nodes to the global coordinate system means that the model nodes of all components in the entire machine are uniformly axially corrected according to the standard direction of the global world coordinate system. The X, Y, and Z axes of the components are completely aligned with the global world coordinate system, and the axial orientation of the model nodes of all components is consistent, with no local custom skewed axes. For horizontal plug-in and front-back pull-out components, the model nodes uniformly use the ±X and ±Z axes of the world coordinate system as the main directions of movement; for vertical lifting and vertical disassembly / assembly components, they uniformly match the ±X axis direction of the world coordinate system. During the correction process, only the attitude and coordinate center point of the local coordinate axes of the model nodes are adjusted, without changing the actual spatial position, assembly posture, external structure, and installation position of the components. Only the motion reference and axial logic are corrected.

[0039] Step 104: Based on the disassembly and assembly rules of the target product, as well as the control instructions and disassembly and assembly process for the operation and maintenance requirements of the target product, control the target 3D model to perform disassembly and assembly actions to generate a disassembly and assembly video of the target product.

[0040] In some embodiments, users can input control commands and product technical documents for the target product through the target workbench. Since the disassembly / assembly video does not necessarily require control of all model nodes corresponding to all components of the target product, this application first needs to determine the structured control information required for disassembly / assembly based on the disassembly / assembly process in the product technical documents and the control commands input by the user. This structured control information includes at least the target component, action type, motion direction, and motion parameters. Subsequently, according to the disassembly / assembly rules and the structured control information, the model nodes in the target 3D model corresponding to the target component are controlled to perform corresponding disassembly / assembly actions, ultimately generating the disassembly / assembly video.

[0041] Among them, the disassembly and assembly rules are a set of predefined logical rules used to constrain the disassembly and assembly behavior of components, which may include at least one of the following: fixed component rotation rules, movement distance rules, disassembly and assembly sequence rules, and collision avoidance rules.

[0042] Product technical documents are documents that describe the product assembly and disassembly methods, operating procedures, and safety requirements, and include at least one or more of the following: process documents, user manuals, and product development plans.

[0043] The control commands are text information input by the user that describes the required disassembly or assembly actions, such as "loosen the four fixing screws on the radiator" or "lift the radiator vertically upwards by 10 centimeters". These control commands can be input in natural language, or through gesture recognition, motion capture, and voice semantic recognition, etc., and are not limited in this embodiment.

[0044] Disassembly and assembly operations include basic motion units such as rotational actions (e.g., tightening, loosening) and translational actions (e.g., inserting, pulling out, lifting, placing), as well as composite operations composed of multiple basic actions (e.g., disassembling the entire module).

[0045] In summary, this application establishes a complete workflow from product configuration to automatic generation of disassembly and assembly videos: first, a component index table is constructed using bill of materials data; then, 3D model nodes are bound to component identifiers; next, the model undergoes structural standardization and motion benchmark calibration; finally, disassembly and assembly rules, user-inputted control commands, and disassembly and assembly procedures are integrated, and a rule engine drives the model to perform precise disassembly and assembly, generating disassembly and assembly videos. This effectively solves the problems of low production efficiency, large molding deviations, data silos, and inability for non-professionals to operate in related technologies.

[0046] Figure 2 A flowchart of the second video generation method proposed in this application is further shown. Based on... Figure 1 The illustrated embodiment further explains step 101. Figure 2 This may include the following steps.

[0047] Step 201: Based on the configuration information of the target product's operation and maintenance requirements, obtain the bill of materials data of the target product from the bill of materials system, and create a component index table corresponding to the target product based on the bill of materials data.

[0048] In some embodiments, this application provides a unified target workbench as a user operation entry point, enabling users to create operation and maintenance projects for the target product, and obtain configuration information, product technical documents, and control instructions for the target product's operation and maintenance requirements after the operation and maintenance project is created. The target workbench can be deployed based on a cloud-based web architecture or a local client architecture, employing a visual graphical interface with capabilities such as unlimited scaling, drag-and-drop functionality, component-based editing, real-time preview, and two-way data synchronization.

[0049] Specifically, users can launch the target workbench deployed based on a cloud-based web architecture or a local client architecture. On the target workbench's homepage, users select the "Create New Project" command, and the system displays a project information entry interface through the information input module. Users then input basic project information through this interface, selecting the server model, video type, and project manager in sequence.

[0050] For server models, users can select a standard model from the system drop-down list; if a model is not listed, it means the project team has not yet submitted a requirement, and it cannot be selected in this case.

[0051] For video types, these can include component replacement videos and complete machine rack installation videos, which are selected by the user through interface options. The project version number is automatically generated by the system according to rules, with a default format of Vx.y, where x represents a major version and y represents a minor revision. Specific change timing and rules are as follows: When a project undergoes fundamental changes that affect the core content, structure, assembly relationships, or device configuration of the video, the major version number x is incremented by 1, and the minor version number y is reset to zero. When a project only undergoes minor modifications that do not affect the core structure, change the assembly / disassembly logic, or replace the device model, the major version number x remains unchanged, and the minor version number y is increased by 1.

[0052] The project manager can be manually filled in by the user; the project name does not need to be manually entered by the user, and is automatically generated by the system based on the information selected by the user. The generation rule is fixed as: server model + video type + project version number.

[0053] For example, the user selects the following information: Server model: ABCDE; Video type: Parts replacement video; Project version number: V1.0 (automatically generated by the system); Project manager: Li XX. The system automatically generates the project name based on the rules: ABCDE Parts Replacement Video V1.0; Once the user completes the basic information for the project and clicks the confirmation button, the system determines that the project corresponding to the current target product (i.e., the product model of the current product) has been created. All subsequent work, such as video production, model processing, video generation, review and release, will be executed independently within this project, and the data will be stored in a closed loop.

[0054] Once the project is created, the information input module immediately performs project initialization operations in the background, automatically creating an independent project storage space, automatically allocating file storage paths, automatically binding the current user's operation permissions, automatically completing identity verification and connection initialization with the BOM system interface, automatically preloading form templates related to the PLM review process, and automatically establishing a communication link with the 3D software.

[0055] After the above operations are completed, the information input module will display the product configuration entry area on the Infinite Canvas main interface. This area uses a visual form format, allowing users to completely enter the configuration information of the target product for which a disassembly / assembly video needs to be produced, using methods such as selection, checkmarks, dropdowns, and input. Users can choose to enter the complete configuration or select the mainstream configuration information of the target product that has already been pre-defined, and the system will automatically fill in the standard configuration.

[0056] After the configuration information is entered, this application can perform real-time verification of the entered content through the information input module, and provide prompts for missing, conflicting, and non-standard items to ensure that the target product configuration is complete, effective, and compatible.

[0057] After the target product configuration is entered, the information input module displays the document upload area on the canvas. This area supports three upload modes: drag-and-drop, click-to-upload, and paste-to-upload, and is compatible with various common file formats such as DOCX, PDF, PPT, and TXT. Following the system prompts, users upload product technical documents (such as process documents, product development plans, and user manuals). These technical documents contain key textual information about the target product's component assembly and disassembly procedures, operating specifications, precautions, tool requirements, safety constraints, and assembly sequence. This provides the original textual basis for subsequent parsing of control instructions. During the document upload process, the information input module can display the upload progress, file size, page count, and parsing status in real time, and automatically intercepts and alerts users to duplicate, empty, or corrupted files to ensure that the uploaded documents are valid and usable.

[0058] In this application, to address issues such as overlapping content, inconsistent descriptions, and differing operational expressions among uploaded product technical documents, a multi-source text intelligent integration and processing mechanism can be implemented through the information input module. The system automatically compares the text paragraphs, operational steps, parameter requirements, and constraint descriptions of each document in the product technical documents in the background, merging and deduplicating duplicate content and eliminating redundant expressions. When different documents have content conflicts or differing expressions regarding the same disassembly / assembly steps, operational requirements, or structural constraints, the system strictly adheres to preset priority rules for acceptance, with the specific priority order being: User Manual > Process Documents > Product Development Plan. The parsed text needs to be structured and organized, using an NLP model to extract key elements such as disassembly / assembly objects, operational actions, and movement directions to form a standardized action sequence (i.e., disassembly / assembly process), which is then passed to subsequent modules, achieving data input once and reuse throughout the entire process.

[0059] After project creation, product configuration, and document upload are completed, the system enters the data interaction and model invocation phase. This system is integrated with the BOM (Bill of Materials). Once the information input module completes product configuration entry, it immediately sends a data request command to the BOM data interaction module. Upon receiving the command, the BOM data interaction module establishes a stable, secure, and real-time data connection with the BOM system. After the connection is established, the BOM data interaction module uses the user-entered configuration information as the sole retrieval condition to initiate a precise query to the BOM system, retrieving the complete machine bill of materials data that perfectly corresponds to that configuration.

[0060] The BOM data interaction module receives, parses, verifies, cleans, and restructures the returned results. The extracted BOM data includes, but is not limited to: unique machine code, material codes at each level, standard component names, component types, material properties, quantities, installation locations, assembly relationships, parent-child relationships, disassembly / assembly sequence, disassembly / assembly constraints, standard installation torque, whether a component is a consumable part, and whether it is hot-swappable. This ensures that subsequent model matching, model materials, animation generation, and motion control fully comply with product design specifications.

[0061] Step 202: Perform configuration information consistency verification, model mapping feasibility verification, and assembly logic conformity verification on the bill of materials data respectively, and obtain at least one verification result.

[0062] In some embodiments, after obtaining the bill of materials data, this application first performs triple verification on the data to ensure that the data relied upon for subsequent model matching, video generation, and motion control is accurate, complete, and executable. Specifically, this includes the following three verifications: Configuration Information Consistency Verification: This verifies whether the configuration information entered by the user is consistent with the overall machine structure in the Bill of Materials (BOM) system. For example, it checks whether the BOM version corresponding to the selected model is correct, and whether the component options in the configuration (such as whether it includes a specific hard drive, power module, etc.) match the BOM records. This verification aims to prevent subsequent retrieval of incorrect models or missing components due to mismatches between the configuration and the BOM.

[0063] Model mapping feasibility verification: This verifies whether each component name in the bill of materials data can be found in the 3D model library as a corresponding model node. The system compares the component names with the naming conventions of model nodes in the 3D model to determine if any components cannot be mapped. This verification ensures that during subsequent binding operations, each component identifier can be located in a unique model node in the 3D model, avoiding situations where there is material but no model.

[0064] Assembly logic conformity verification: This verifies whether the disassembly / assembly sequence, parent-child relationships, and assembly constraints recorded in the bill of materials (BOM) data conform to the actual assembly logic of the target product. For example, it checks whether the outer casing must be disassembled before the internal modules can be disassembled, and whether the relationship between the heat sink and screws is correct. This verification aims to identify potential logical errors or version discrepancies in the BOM data, ensuring that the subsequently generated disassembly / assembly animations conform to the assembly specifications of the actual product.

[0065] After performing the above triple verification, the system will obtain a pass or fail result for each verification item. Only when all verifications pass will the subsequent model retrieval and binding steps be allowed; if any verification fails, the system will issue a prompt to the user and provide correction suggestions, thereby ensuring the reliability of video generation from the data source.

[0066] Step 203: If at least one verification result indicates that the verification has passed, create a component index table based on the index table creation rules and the bill of materials data. The index table creation rules include at least one of the following: hierarchical index rules, unique identifier rules, sequential increment rules, and spatial order rules.

[0067] In some embodiments, when inconsistencies occur during the triple verification, the BOM data interaction module will automatically prompt the user and provide correction suggestions, such as pointing out components whose configurations do not match the BOM, components whose names cannot be found in the corresponding models, or steps with assembly logic errors. After the user corrects the issues according to the prompts, the system will re-execute the verification.

[0068] Once all validations pass, the BOM data interaction module creates a component index table from the bill of materials data. This creation process must adhere to preset index table creation rules (i.e., naming and indexing rules), uniquely identifying and binding each component. Specifically, each component is assigned a unique code, which is simultaneously bound to the component name, material number, and hierarchical path, thus forming a unique index to ensure that subsequent model matching and motion control can accurately locate each component.

[0069] Index table creation rules include at least one of the following: Hierarchical indexing rules: The whole machine model code is used as the first-level root directory, and functional components (such as CPU modules and power supply modules) are used as the second-level indexes, thereby establishing an index hierarchy corresponding to the model tree hierarchy, which facilitates fast retrieval according to the product composition structure.

[0070] Unique Identification Rule: Each component is assigned a globally unique code, which is simultaneously bound to the component's material number, its hierarchical path in the BOM, and its parent-child relationship ID. For example, the unique code of a screw can be associated with the parent ID of its heatsink, ensuring that the parent-child relationship is clearly traceable in the index table.

[0071] Sequential Incremental Numbering Rule: A sequential coding logic is used to assign incremental numbers to components according to the actual order of disassembly. For example, the first component disassembled has a lower number, and subsequent components have progressively higher numbers. This rule ensures that the disassembly and assembly logic can be called sequentially, avoiding animation errors caused by inconsistent numbering.

[0072] Spatial Order Rule: For multiple components of the same type (such as multiple hard drives or multiple screws), the system prioritizes numbering them according to a pre-defined slot order, such as "Hard Drive 0", "Hard Drive 1", and "Hard Drive 2". If no pre-defined order exists, standard numbers are automatically generated based on the spatial distribution of the components within the machine, following the principle of "front to back, left to right, and top to bottom", such as "Screw 1", "Screw 2", and "Screw 3". This rule ensures that the numbering of components of the same type is deterministic and repeatable, facilitating user understanding and system recall.

[0073] The component index table created by the above index table creation rules establishes the mapping foundation between material data and 3D models, and provides structured data support for subsequent disassembly and assembly sequence control and action rule execution.

[0074] In summary, this application provides a unified target workbench, enabling users to easily create maintenance projects, input product configurations, upload product technical documents, and automatically perform multi-source text integration and standardized action sequence extraction. Based on this, the system acquires bill of materials data and performs triple verification (configuration consistency, model mapping feasibility, and assembly logic conformity) to ensure data accuracy and integrity. After successful verification, a component index table is created based on rules such as hierarchical indexing, unique identifiers, sequential increment, and spatial order. Each component is assigned a unique code and its material information and hierarchical path are bound. Thus, this application achieves full-process automation from project configuration and data verification to structured indexing, significantly improving the reliability and traceability of data processing. It provides a precise and unified data foundation for subsequent 3D model binding, disassembly and assembly control, and video generation, effectively solving problems such as data silos, error-prone manual operation, and poor scalability in existing technologies.

[0075] Figure 3 A flowchart of the third video generation method proposed in this application is further shown. Based on... Figure 1 The illustrated embodiment further explains step 104. Figure 3 This may include the following steps.

[0076] Step 301: Generate the disassembly and assembly process based on the product technical documents that meet the operation and maintenance requirements of the target product.

[0077] In some embodiments, after the information input module completes the uploading of product technical documents (such as process documents, product development plans, user manuals, etc.), the intelligent parsing and action control module simultaneously performs fully automatic text processing on these documents.

[0078] Specifically, the system uses a Natural Language Processing (NLP) model to extract key elements from product technical documents, including the objects to be disassembled / assembled, operational actions, tools used, direction of movement, distance traveled or number of rotations, torque requirements, constraints, safety warnings, and the order of operations. Based on this extracted information, the system generates a standardized and executable sequence of disassembly / assembly steps (i.e., the disassembly / assembly process).

[0079] After the disassembly and assembly process is obtained through parsing, the system can display the generated disassembly and assembly steps on the canvas of the target workbench. Users can view, confirm, or adjust the order of the steps, but there is no need to re-edit or manually write the step content, thus ensuring the accuracy and efficiency of the disassembly and assembly process.

[0080] Step 302: Based on the disassembly and assembly process, the control commands are parsed to obtain structured control information including the target component, the action type of the target component, the direction of movement of the target component, and the motion parameters of the target component.

[0081] In some embodiments, the system first parses the user-input control command to obtain parsed information, which includes at least the name (full name, abbreviation, or alias) of the component to be controlled by the command. Subsequently, to avoid inaccurate use of the parsed information, this application can calculate the text similarity between the name to be controlled and the names of each component in the disassembly / assembly process (i.e., the sequence of disassembly / assembly steps) using the system's intelligent parsing and motion control module. Based on the different similarities, the system executes different processing logic: When the similarity is greater than or equal to a preset first threshold (e.g., 90%), the system automatically locks the component as the target component and highlights it on the canvas; When the similarity is between the second threshold and the first threshold (e.g., 70% to 90%), the system displays the top three candidate parts with the highest similarity for the user to choose from, and identifies the part selected by the user as the target part. When the similarity is less than or equal to the second threshold (e.g., 70%), the system prompts that no matching part was found and guides the user to re-enter the information.

[0082] After locking onto the target component, the system further breaks down the control commands from multiple dimensions, extracting a four-element structure of "action type + target object + direction of movement + motion parameters". For example: For the instruction "Loosen the 4 fixing screws on the heatsink", the system extracts the following: action type is "loosen", target object is "fixing screws on the heatsink", movement direction is "counterclockwise" (default right-hand screws), and movement parameter is "4 screws". For the instruction "Lift the radiator vertically upward by 10 centimeters", the system extracts the following: action type is "lift", target object is "radiator", movement direction is "vertical upward", and movement parameter is "10 centimeters". For the instruction "insert the memory vertically downward into the slot until the latch closes", the system extracts the following: action type is "insertion", target object is "memory", movement direction is "vertical downward", and movement parameter is "until the latch closes" (termination condition). For the instruction "pull the power module out horizontally", the system extracts the following: action type is "pull out", target object is "power module", movement direction is "horizontally outward", and movement parameters are empty (the pulling distance is automatically determined by the system based on path pre-scanning).

[0083] Finally, the system integrates the above analysis results into structured control information, which includes target components, action types, motion directions, and motion parameters, providing precise input for subsequent calls to disassembly and assembly rules to drive model actions.

[0084] Step 303: Based on the structured manipulation information and disassembly / assembly rules, control the target model nodes corresponding to the target components in the target 3D model to perform disassembly / assembly actions to generate a disassembly / assembly video.

[0085] In some embodiments, the disassembly and assembly rules include at least one of the following: fixed component rotation rules, movement distance rules, disassembly and assembly sequence rules, and collision avoidance rules. Based on the action type, target component, movement direction, and movement parameters in the structured manipulation information, and in conjunction with the above rules, the system drives the target model node to complete precise disassembly and assembly actions, and renders and generates video in real time.

[0086] 1) Rules for rotating fixed components When the target component in the structured control information is a fastener (such as a screw or bolt), the system first identifies the thread direction of the fastener. The thread direction can be identified via AI vision: if it is a right-hand screw, counter-clockwise is for loosening and clockwise is for tightening; if it is a left-hand screw, the reverse action is performed (clockwise to loosen, counter-clockwise to tighten). The system executes its action logic based on a standard right-hand screw.

[0087] For the loosening action: The system drives the target model node to rotate continuously counterclockwise at a preset fixed angular velocity (e.g., 60° / second), while simultaneously detecting the thread engagement and end-face contact status between the screw and nut (or screw hole). Combined with 3D spatial collision detection, when it is detected that the threaded segment of the fastener has completely disengaged from the engagement structure, or the gap between the end face of the fastener and the mounting surface is less than a preset gap threshold and the axial displacement reaches the locking position (i.e., the screw threaded segment has completely disengaged from the nut engagement structure, and the thread engagement relationship has been completely released), it is determined to be fully unlocked, and the rotation stops immediately, completing the screw loosening.

[0088] For the tightening action: the system drives the target model node to rotate continuously in the opposite direction (clockwise) at the same fixed angular velocity (e.g., 60° / second), while simultaneously applying dual judgment conditions: First, the gap distance between the screw end face and the mounting contact surface is detected in real time; the condition is met when the end faces are in contact and there is no obvious gap. Second, the thread is detected to be fully engaged, the axial feed stroke reaches the assembly locking position, and the component has no further downward displacement. When either of the above conditions is met, the tightening rotation automatically stops to prevent over-tightening from causing the model to interlock and to ensure that the animation matches the actual assembly locking state.

[0089] This application employs a fixed-component rotation rule, combining thread direction recognition, constant angular velocity drive, and physical state-based stop determination. This eliminates the open-loop control methods in related technologies that rely on a fixed number of turns or time, achieving closed-loop precise control of screw assembly and disassembly. This rule ensures that the screw stops just as it disengages from the thread when loosening and stops just as it engages when tightening. This avoids model interlacing or visual distortion caused by excessive rotation, and also prevents incomplete assembly or disassembly due to insufficient rotation, significantly improving the realism, certainty, and industrial usability of the assembly and disassembly animation.

[0090] 2) Movement Distance Rules For linear movements such as translation, lifting, and pulling out, the system adopts a dual-mode determination mechanism to distinguish between two scenarios: one with a specified distance and the other without.

[0091] For commands specifying a distance: When the control command contains a clear movement distance parameter (e.g., "lift 10 centimeters"), the system strictly follows the three-dimensional coordinate difference to perform precise positioning and movement, using the target coordinate point as the endpoint, and completes the directional displacement at a uniform speed to ensure that the action size is completely consistent with the command requirements.

[0092] For unspecified distances: When the control command does not specify a movement distance parameter, the system performs adaptive dynamic calculations based on the actual spatial structure of the entire machine model, the positioning of surrounding components, assembly gaps, and motion interference boundaries. The specific execution is as follows: Before the action is officially executed, the system performs a full-domain pre-scan of the motion path: taking the target model node to be moved as an independent detection object, the system divides all surrounding structural components such as the chassis shell, other functional modules, fixed brackets, and limiting structures into opposing collision objects; it scans all occlusion boundaries, entity contours, assembly limits, and spatial interference areas on the preset motion trajectory layer by layer to fully identify the effective spatial range in the direction of motion in order to determine the safe travel distance (i.e., the maximum barrier-free travel distance).

[0093] During the motion path pre-scanning process, a reasonable pull-out or lifting distance can be calculated proportionally based on the insertion depth, embedding distance, or assembly engagement length of the target model node itself, to ensure that the component is completely separated from the slot, socket, limit buckle, and assembly limit structure, thus achieving complete separation.

[0094] Within the safe formation range, this application can also automatically reduce the reserved safety buffer gap, ensuring that the component is completely removed from the installation position and meets the visual effect of disassembly and assembly demonstration, while not touching any surrounding structural components, and without collision, squeezing, or mold penetration throughout the process.

[0095] Simultaneously, this application can also perform real-time interference detection when controlling the target model nodes to perform disassembly and assembly actions based on safe travel and structured control information. Once the critical position of completely detaching from the assembly structure without spatial interference is reached, the displacement is immediately stopped. If the space is small and the boundaries are limited, the system automatically adapts to shorten the travel or tilt the parts before moving and removing them, with avoiding collisions as the first priority; if the space is ample, the second priority is to achieve a reasonable disassembly and display effect.

[0096] All adaptive displacement movements can use a fixed speed (e.g., 25mm / s) to maintain a smooth rhythm and ensure that the entire disassembly and assembly animation is coordinated and the visuals are natural and coherent.

[0097] This application employs a movement distance rule and a dual-mode judgment mechanism that distinguishes between specified and unspecified distances. When the user explicitly provides the movement distance parameter, it achieves precise fixed-length control. When the user does not provide the movement distance parameter, it achieves intelligent movement control without manual preset through motion path pre-scanning, insertion depth calculation, and adaptive safety stroke planning. Furthermore, this application combines real-time interference detection, safety buffer reservation, and automatic stroke adaptation in space-constrained situations to ensure that components can completely detach from the assembly structure during removal or lifting without colliding with or passing through surrounding components. Unified fixed speed control further ensures the coordination and visual continuity of the animation rhythm. It effectively solves the problems of repeated manual trial and error in movement distance calculations, and the tendency for incomplete movement or excessive collisions in related technologies, significantly improving the automation, spatial adaptability, and physical realism of assembly and disassembly animations.

[0098] 3) Disassembly and assembly sequence rules Before the assembly / reassembly operation, the system automatically verifies the removal status of external model nodes (such as those corresponding to peripheral protection and shielding components) based on the model hierarchy and the hierarchical constraints between model nodes. For internal components, all external shielding components must be removed before the disassembly / reassembly of internal components can be performed. If the prerequisite disassembly / reassembly conditions are not met (e.g., attempting to disassemble internal modules without removing the outer shell), the system prohibits subsequent actions and provides a text prompt. During the reassembly / reassembly phase, the system follows reverse sequence constraints to prevent skipping levels of disassembly / reassembly or unauthorized operations, ensuring the rationality and standardization of the disassembly / reassembly process.

[0099] This application employs a disassembly / assembly sequence rule. By leveraging the hierarchical structure of the model and the subordinate constraints between nodes, it automatically verifies the removal status of peripheral obstructing components and applies precondition constraints to the disassembly and assembly of internal components, thus achieving automated logical verification of the disassembly / assembly process. This rule mandates the removal of outer components first during the disassembly phase and automatically imposes reverse constraints on the installation sequence during the reassembly phase, effectively preventing logical errors such as skipping levels in disassembly / assembly or improper operations, and avoiding animation clipping issues or physical impossibilities caused by incorrect sequence. Simultaneously, the system promptly outputs text prompts when conditions are not met, guiding users to operate according to the standardized process, improving interactivity and educational guidance value. Consequently, the generated disassembly / assembly videos strictly conform to the actual assembly logic of the product, enhancing the accuracy, safety, and standardization of maintenance guidance.

[0100] 4) Collision and obstacle avoidance rules During the process of controlling the target model node to perform disassembly and assembly actions, the system calculates the distance between the target model node and the surrounding model nodes in real time. The moving target model node is treated as an independent object, and the other components are uniformly regarded as obstacles.

[0101] When the interval distance is less than a preset threshold (e.g., 2mm) and the interval distance is greater than or equal to a preset distance (e.g., 1mm), a collision risk is determined. The system automatically controls the moving parts to deviate a preset distance (e.g., 1mm) in a preset offset direction (e.g., lateral) to avoid obstacles, while maintaining the original direction of movement and speed unchanged.

[0102] When the interval distance is less than the preset threshold, but the interval distance is less than the preset distance (i.e., there is occlusion in multiple directions at the same time and no offset space), the system immediately pauses the operation and prompts the user, suggesting that the disassembly and assembly sequence be adjusted or that the interfering parts be removed first.

[0103] When the interval distance throughout the movement is greater than or equal to a preset threshold (e.g., ≥2mm) and there are no obstructions during the movement, the path is deemed safe, and the action is executed directly according to the predetermined parameters.

[0104] In this application, the collision avoidance mechanism only makes minor positional corrections and does not change the component's attitude or final assembly position. This rule, combined with the aforementioned pre-determined safe travel range in the movement distance rule, ensures that the component's movement is collision-free, clipping-free, with reasonable travel, and smooth and compliant operation.

[0105] This application employs collision avoidance rules, monitoring the distance between model nodes and surrounding obstacles in real time during movement, and using a tiered response mechanism (controllable offset within a preset threshold and pause prompts when there is no offset space) to achieve dynamic autonomous obstacle avoidance during movement. Combined with the pre-path pre-scanning in the movement distance rules, this forms a dual-protection system of pre-planning safe routes and real-time fine-tuning of obstacle avoidance. This rule only makes minor corrections to component positions without changing their posture or final assembly position, ensuring that obstacle avoidance behavior does not interfere with normal assembly / disassembly logic. Therefore, it effectively avoids model interlacing and visual distortion caused by minor gaps or path deviations during animation generation, significantly improving the smoothness, physical realism, and robustness of assembly / disassembly actions. Furthermore, timely prompts guide users to optimize assembly / disassembly strategies, further enhancing usability and industrial adaptability.

[0106] After all the above actions are completed, the system automatically records the action trajectory, time nodes, and coordinate changes, forming complete animation data. Combined with automatic rendering, speech synthesis, and subtitle generation, the final output is a disassembly and assembly video that conforms to industrial assembly standards.

[0107] Furthermore, to enhance the viewing experience of the disassembly and assembly videos, this application also includes rendering and post-production compositing. Specifically, this application can determine the target component based on the component name corresponding to the target model node, and obtain the component type and material of the target component. Based on the component type, material, and model hierarchy, the application performs adaptive switching of the camera perspective and intelligent rendering of material lighting and shadows in the disassembly and assembly videos. This improves the viewing experience and educational guidance effect of the disassembly and assembly videos, achieving professional-grade camera language and image quality without manual intervention.

[0108] Specifically, after receiving the disassembly and assembly video, the intelligent analysis and motion control module can send it to the audio-visual generation and rendering module. Upon receiving the video, the module automatically adds keyframes and controls the movement speed of the target model nodes to ensure the video rhythm matches the actual operation, avoiding unnatural phenomena such as jumps, stutters, or abrupt changes. Simultaneously, based on the type of the target model node, the system automatically selects the optimal camera view and switches accordingly: when operating large component model nodes (such as the chassis cover or the entire unit's air duct), the system automatically uses the overall main view; when operating small part model nodes (such as screws, clips, gold fingers, or terminals), it automatically switches to a close-up view. The entire view switching process is smooth, jitter-free, and without abrupt jumps, ensuring a clear, intuitive, and professional viewing experience.

[0109] Once the viewpoint is determined, the module automatically adjusts the scene lighting, intelligently adapting the intensity, angle, and shadow status of the main light, fill light, and ambient light. Based on the model's hierarchical structure and the component names corresponding to each model node, the system automatically identifies model nodes for different components such as metal, sheet metal, plastic parts, circuit boards, and screws. It also accurately determines the material type, surface roughness, reflectivity, and optical properties of various component model nodes by obtaining material information from the bill of materials data. Combined with the currently selected overall video style (e.g., futuristic, minimalist, high-tech), the system custom-adjusts the lighting brightness, shadow levels, highlight intensity, and shadow texture, matching the lighting logic of the corresponding style. It defaults to realistic rendering standards, fully restoring the target product's true materials and appearance details; users can also upload reference images through the target workbench, and the system automatically adapts the lighting parameters based on the reference image's color tone and atmosphere. The system optimizes lighting, color, and rendering in sync with the art style, ensuring clear component outlines, complete details, no overexposure, no black areas, no visual obstruction, balanced overall brightness, accurate material representation, and a unified and harmonious visual texture throughout the entire film.

[0110] After the animation, perspective, and art style settings are completed, the module enters the audio-visual synthesis stage. This involves automatically generating voiceover based on the disassembly and assembly steps in the video. Users can choose to use any large voice model, supporting multiple timbres, multiple languages, and adjustable speech rate. The voiceover is strictly time-aligned with the animation actions, achieving "action occurs, voice is synchronized." Subtitles are generated according to the video disassembly and assembly steps, with fixed display positions and uniform font sizes. Key constraints such as direction, distance, and number of rotations are automatically highlighted in yellow; safety warnings such as power outages, anti-static measures, and prohibition of live operation are automatically highlighted in red to enhance the reminder effect. Users can also save the set subtitle styles for one-click recall in subsequent operations.

[0111] After compositing, the module performs fully automatic video rendering according to system default or user-preset parameters, supporting 1080P and 4K resolutions, with the universal MP4 encoding format, compatible with all scenarios including web and mobile devices. The rendering process runs in the background, with the progress and estimated remaining time displayed in real time on the canvas. Once rendering is complete, the system automatically loads the final disassembly and assembly video onto the target workbench canvas, allowing users to directly play, pause, drag the progress bar, or preview in full screen, confirming the final effect without the need for third-party post-production software.

[0112] Furthermore, this application enables automatic review and publication after obtaining the final disassembly and assembly video. That is, after the disassembly and assembly video is produced, there is no need for local saving or separate rendering and export. Users can click the "Submit for PLM (Product Lifecycle Management) Review" button in the project management and publication module. The project management and publication module is deeply integrated with the PLM system, initiating the review process by calling the final disassembly and assembly video within the current project, without any secondary export or file transfer. If the review is rejected, the system clearly displays the reason for rejection. Users can directly modify steps and adjust actions within the canvas, reusing existing project data to regenerate relevant content, and then resubmit for review with one click. When the PLM review status shows "Passed," the canvas automatically lights up the "Publish to Official Website" button and indicates that publication is possible. After the user clicks, the project management and publication module continues to call the final video of this project and automatically executes the official website publication process without additional rendering and export. After publication, the system initiates a fully automatic archiving process, uniformly naming and backing up models, scripts, coordinate tables, videos, and version logs, supporting one-click retrospective of historical versions and cross-project material reuse.

[0113] In summary, this application automatically extracts the disassembly and assembly process from product technical documents using natural language processing technology. It then combines this with semantic parsing of user control commands to generate structured control information. Based on rules governing component rotation, movement distance, disassembly / assembly sequence, and collision avoidance, it precisely drives the 3D model nodes to perform disassembly and assembly actions. Simultaneously, it automatically handles perspective switching, material and lighting rendering, speech synthesis, and subtitle generation. Furthermore, it integrates PLM review and official website publishing processes, achieving full automation from project creation, model processing, motion control, video synthesis, to review and publication. This application effectively solves problems in related technologies, such as reliance on manual segmentation in video production, severe data silos, and the tendency for AI-generated videos to exhibit clipping and motion deviations. It significantly lowers the operational threshold for non-professionals, ensures that disassembly and assembly videos strictly conform to industrial assembly logic and spatial physical constraints, and greatly improves the efficiency, consistency, and reusability of video production.

[0114] Furthermore, for ease of understanding, such as Figure 4 As shown, this application provides a schematic diagram of the structure of a specific video generation system.

[0115] Reference Figure 4The system uses the Infinite Canvas Workbench 1 (i.e., the target workbench) as its core interaction layer, exchanging data bidirectionally with six functional modules: Information Input Module 2, BOM Data Interaction Module 3, 3D Model Processing Module 4, Intelligent Parsing and Motion Control Module 5, Audio / Video Generation and Rendering Module 6, and Project Management and Publishing Module 7. The BOM Data Interaction Module 3 further connects to the BOM system 8 to initiate precise bill of materials data queries; the 3D Model Processing Module 4 further connects to the server's 3D model library 9 to initiate precise model calls. These modules are independent yet closely collaborative, collectively covering the entire process of data input, data interaction, model processing, intelligent decision-making, audio / video generation, and project and workflow management. This architecture, through modular division of labor and bidirectional data synchronization, achieves a unified closed-loop operation from project configuration, bill of materials acquisition, 3D model standardization, instruction parsing and motion control, to video rendering and compositing, review and publishing.

[0116] Furthermore, this application, based on a distributed architecture of an infinite canvas workbench, can support multiple users (such as engineers) to collaborate remotely within the same canvas space. The system can automatically handle spatial interference and logical conflicts between disassembly and assembly actions defined by different users, and synthesize a globally unified sequence of project actions in real time, thereby achieving efficient parallel operation and version consistency management in team collaboration scenarios.

[0117] The specific execution process of each module can be found in the following references. Figures 1 to 3 The embodiments shown will not be described in detail here.

[0118] Furthermore, based on Figure 4 The video generation system shown is as follows: Figure 3 The present application further provides a schematic diagram of a specific video generation process.

[0119] Reference Figure 5 The process begins with a video production request initiated from the infinite canvas workbench, and proceeds sequentially through the input phase, processing phase, calculation phase, output phase, and review and release phase.

[0120] Input Phase: Users create maintenance projects in the wireless canvas workbench, enter the configuration information of the target product, and upload product technical documents (such as process documents and user manuals). The system retrieves the corresponding bill of materials data from the BOM system based on the configuration information. Simultaneously, using the configuration information and product technical documents shown in the diagram, users can also input control commands for the target product through the wireless canvas workbench.

[0121] Processing Phase: A dual-track parallel processing approach is adopted, namely, a model standardization track and a motion analysis track. For the model standardization track, the bill of materials data is restructured to generate a component index table; based on the component index table, the initial 3D model is retrieved from the model library, and model nodes are bound to component identifiers; automatic coordinate axis correction and motion reference calibration are performed to obtain a standardized target 3D model. For the motion analysis track, the semantic understanding of user-input control commands is performed through natural language processing (NLP) and product technical documents, extracting the four-element structure of "motion type-target component-motion direction-motion parameters," and compiling it into machine-executable control commands.

[0122] Calculation phase: The rule engine provides precise control over the motion: it initiates motion path pre-scanning, calculates safe travel distance and performs adaptive obstacle avoidance; it continuously detects the distance between parts during motion, automatically performs minor offsets to avoid obstacles and prevents mold penetration; it performs stop determination based on physical constraints (such as thread engagement state, end face fit, coordinate difference) to ensure that tightening / loosening actions are accurate and in place.

[0123] Output stage: TTS audio and synchronized subtitles are automatically generated based on the disassembly and assembly steps, achieving millisecond-level alignment of audio, text, and visuals, and finally synthesizing a complete disassembly and assembly video.

[0124] Review and Release Phase: Submit the generated video to the PLM system for review. Once approved, publish it to the official website with one click and automatically archive project data (model, script, coordinate table, video and version log), supporting historical version backtracking and material reuse.

[0125] The above process is completed in an integrated manner within the Infinite Canvas workbench, realizing a closed-loop process from data input, model and instruction processing, rule-based calculation, to audio and video synthesis, review and release.

[0126] The specific implementation process for each stage can be referred to Figures 1 to 3 The embodiments shown will not be described in detail here.

[0127] Through the above description of the embodiments, those skilled in the art can clearly understand that the methods according to the above embodiments can be implemented by means of software plus necessary general-purpose hardware platforms. Of course, they can also be implemented by hardware, but in many cases the former is a better implementation method.

[0128] Embodiments of this application also provide a video generation apparatus 600. Figure 6 This is a schematic diagram of the structure of a video generation device provided in an embodiment of this application, as shown below. Figure 6 As shown, it includes: Create unit 610, which is used to obtain the bill of materials data of the target product from the bill of materials system according to the configuration information of the target product's operation and maintenance requirements, and create the component index table corresponding to the target product based on the bill of materials data; The retrieval unit 620 is used to retrieve the initial 3D model corresponding to the target product from the model library based on the component identifier and component hierarchy in the component index table. The model nodes in the initial 3D model are bound to the component identifier. The structural optimization unit 630 is used to perform structural reorganization and motion datum calibration on the initial three-dimensional model based on structural standardization rules and coordinate correction rules, so as to obtain a target three-dimensional model that matches the product structure of the target product. The control unit 640 is used to control the target 3D model to perform disassembly and assembly actions based on the disassembly and assembly rules of the target product, as well as the control instructions and disassembly and assembly process of the target product's operation and maintenance requirements, so as to generate a disassembly and assembly video of the target product.

[0129] In one possible implementation of this application embodiment, the structural optimization unit 630 is used for: Based on coordinate correction rules, motion datum calibration is performed on the initial 3D model to obtain the corrected 3D model. Motion datum calibration includes at least one of resetting the coordinate center of the model node, aligning the local coordinate axes of the model node to the global coordinate system, and resetting the rotation center point of the rotating model node. Based on structural standardization rules, the corrected 3D model is restructured to obtain the target 3D model. The structural restructuring includes at least one of the following: cleaning up redundant objects, reconstructing the model hierarchy, setting the model rigid structure, and establishing subordinate constraint relationships between model nodes.

[0130] In one possible implementation of this application embodiment, the creation unit 610 is used to: perform configuration information consistency verification, model mapping feasibility verification, and assembly logic conformity verification on the bill of materials data respectively, and obtain at least one verification result; If at least one verification result indicates that the verification passed, a component index table is created based on the index table creation rules and the bill of materials data. The index table creation rules include at least one of the following: hierarchical index rules, unique identifier rules, sequential increment rules, and spatial order rules.

[0131] In one possible implementation of this application embodiment, the control unit 640 is used for: Generate disassembly and assembly procedures based on the product technical documents that require the operation and maintenance of the target product; The control commands are parsed based on the disassembly and assembly process to obtain structured control information that includes the target component, the action type of the target component, the direction of movement of the target component, and the motion parameters of the target component. Based on structured manipulation information and disassembly / assembly rules, the target model nodes corresponding to the target components in the target 3D model are controlled to perform disassembly / assembly actions to generate disassembly / assembly videos.

[0132] In one possible implementation of this application embodiment, the control unit 640 is used for: The control commands are parsed to obtain parsing information, which includes the name of the controllable component corresponding to the control command. Determine the similarity between the name to be controlled and the names of the parts to be disassembled and assembled in the disassembly and assembly process; If the similarity is greater than the first similarity threshold, the disassembled / assembled component is identified as the target component. Based on the target disassembly and assembly steps and analysis information corresponding to the target components in the disassembly and assembly process, structured control information is obtained.

[0133] In one possible implementation of this application embodiment, the disassembly and assembly rules include fixing member rotation rules, and the control unit 640 is used for: When the target component in the structured control information is a fixed part, identify the thread direction of the fixed part; The rotation direction of the fixed component is determined based on the thread direction and the action type in the structured control information. Control the target model node corresponding to the fastener in the target 3D model to rotate along the rotation direction at a preset rotation speed, and obtain the thread engagement state or end face contact state of the target model node during the rotation process; If the thread engagement state and / or end face contact state of the target model node meet the preset stop conditions, control the target model node to stop rotating.

[0134] In one possible implementation of this application embodiment, the disassembly and assembly rules include movement distance rules, and the control unit 640 is used for: When the motion parameters in the structured control information do not include the travel distance parameter, the target model nodes are pre-scanned based on the structured control information and the target 3D model to determine the safe travel distance of the target 3D model. Based on safe travel and structured control information, control the target model nodes to perform disassembly and assembly actions.

[0135] In one possible implementation of this application embodiment, the disassembly and assembly rules include disassembly and assembly sequence rules, and the control unit 640 is used for: Based on the hierarchical structure of the target 3D model and the subordinate constraints between model nodes, determine the removal status of external model nodes associated with the target model nodes; When the dismantling status indicates that the dismantling has been completed, control the target model node to perform dismantling and assembly actions.

[0136] In one possible implementation of this application embodiment, the disassembly and assembly rules include collision and obstacle avoidance rules, and the control unit 640 is used for: During the process of controlling the target model node to perform disassembly and assembly actions, the distance between the target model node and the surrounding model nodes is obtained; If the interval distance is less than the preset distance threshold, or if the interval distance is greater than or equal to the preset distance, the target model node is controlled to shift to the preset offset direction by the preset distance.

[0137] For a description of the features in the embodiment corresponding to the video generation device, please refer to the relevant description in the embodiment corresponding to the video generation method, which will not be repeated here.

[0138] Embodiments of this application also provide an electronic device, including a memory and a processor, wherein the memory stores a computer program and the processor is configured to run the computer program to perform the steps in any of the above-described video generation method embodiments.

[0139] Embodiments of this application also provide a computer-readable storage medium storing a computer program, wherein the computer program is configured to execute the steps in any of the above-described video generation method embodiments when it is run.

[0140] In one exemplary embodiment, the aforementioned computer-readable storage medium may include, but is not limited to, various media capable of storing computer programs, such as a USB flash drive, read-only memory (ROM), random access memory (RAM), portable hard disk, magnetic disk, or optical disk.

[0141] Embodiments of this application also provide a computer program product, which includes a computer program that, when executed by a processor, implements the steps in any of the above-described video generation method embodiments.

[0142] Embodiments of this application also provide another computer program product, including a non-volatile computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps in any of the above-described video generation method embodiments.

[0143] Any of the components, modules, units, parts, methods, and operations described herein can be implemented using software, firmware, hardware (e.g., fixed logic circuitry), manual processing, or any combination thereof. Alternatively or additionally, any functionality described herein can be executed at least in part by one or more hardware logic components, such as, but not limited to, a central processing unit (CPU), a field-programmable gate array (FPGA), an application-specific integrated circuit (ASIC), an application-specific standard product (ASSP), a system-on-a-chip (SoC), a complex programmable logic device (CPLD), a microprocessor (MCU), etc. The terms "system," "computing device," or "apparatus" as used herein encompass various means, devices, and machines for processing data, including, for example, one or more programmable processors, computers, SoCs, or combinations thereof. The apparatus may also include code that creates an execution environment for the computer program in question, such as code constituting processor firmware, a protocol stack, a database management system, an operating system, a cross-platform runtime environment, a virtual machine, or one or more combinations thereof. The aforementioned computer program (also known as a program, software, software application, app, script, or code) can be written in any form of programming language, including compiled or interpreted languages, declarative or procedural languages, and can be deployed in any form, including as a standalone program or as a module, component, subroutine, object, or other unit suitable for a computing environment.

[0144] Those skilled in the art will further recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, computer software, or a combination of both. To clearly illustrate the interchangeability of hardware and software, the components and steps of the various examples have been generally described in terms of functionality in the foregoing description. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0145] The video generation method provided in this application has been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this application. The descriptions of the embodiments above are only for the purpose of helping to understand the method and its core ideas. It should be noted that those skilled in the art can make several improvements and modifications to this application without departing from the principles of this application, and these improvements and modifications also fall within the protection scope of the claims of this application.

Claims

1. A method of video generation, the method comprising: include: Based on the configuration information of the target product's operation and maintenance requirements, obtain the bill of materials data of the target product from the bill of materials system, and create a component index table corresponding to the target product based on the bill of materials data; Based on the component identifiers and component hierarchy in the component index table, the initial 3D model corresponding to the target product is retrieved from the model library, and the model nodes in the initial 3D model are bound to the component identifiers; Based on structural standardization rules and coordinate correction rules, the initial three-dimensional model is structurally reorganized and motion reference is calibrated to obtain a target three-dimensional model that matches the product structure of the target product. Based on the disassembly and assembly rules of the target product, as well as the control instructions and disassembly and assembly procedures for the operation and maintenance requirements of the target product, the target 3D model is controlled to perform disassembly and assembly actions to generate a disassembly and assembly video of the target product.

2. The method according to claim 1, characterized in that, The process of structural reorganization and motion datum calibration of the initial 3D model based on structural standardization rules and coordinate correction rules to obtain a target 3D model that matches the product structure of the target product includes: Based on the coordinate correction rules, the initial three-dimensional model is subjected to motion reference calibration to obtain the corrected three-dimensional model. The motion reference calibration includes at least one of resetting the coordinate center of the model node, aligning the local coordinate axes of the model node to the global coordinate system, and resetting the rotation center point of the rotational model node. Based on the structural standardization rules, the corrected 3D model is restructured to obtain the target 3D model. The structural restructuring includes at least one of the following: cleaning up redundant objects, reconstructing the model hierarchy, setting a rigid structure for the model, and establishing subordinate constraint relationships between model nodes.

3. The method according to claim 1, characterized in that, The step of creating the component index table corresponding to the target product based on the bill of materials data includes: The bill of materials data is subjected to configuration information consistency verification, model mapping feasibility verification, and assembly logic conformity verification respectively, and at least one verification result is obtained. If at least one verification result indicates that the verification is successful, the component index table is created based on the index table creation rules and the bill of materials data. The index table creation rules include at least one of the following: hierarchical index rules, unique identifier rules, sequential increment rules, and spatial order rules.

4. The method according to claim 1, characterized in that, The process of controlling the target 3D model to perform disassembly and assembly actions based on the disassembly and assembly rules of the target product, as well as the control instructions and disassembly and assembly procedures for the operation and maintenance requirements of the target product, to generate a disassembly and assembly video of the target product includes: The disassembly and assembly process is generated based on the product technical documents that specify the operation and maintenance requirements of the target product. Based on the disassembly and assembly process, the control commands are parsed to obtain structured control information that includes the target component, the action type of the target component, the movement direction of the target component, and the movement parameters of the target component. Based on the structured manipulation information and the disassembly and assembly rules, the target model nodes corresponding to the target components in the target 3D model are controlled to perform the disassembly and assembly actions to generate the disassembly and assembly video.

5. The method according to claim 4, characterized in that, The process of parsing the control commands based on the disassembly and assembly process to obtain structured control information including the target component, the action type of the target component, the movement direction of the target component, and the movement parameters of the target component includes: The control command is parsed to obtain parsing information, which includes the name of the controllable component corresponding to the control command. Determine the similarity between the name to be manipulated and the name of the component to be disassembled in the disassembly and assembly process; If the similarity is greater than a first similarity threshold, the disassembled component is determined to be the target component; Based on the target disassembly and assembly steps corresponding to the target component in the disassembly and assembly process and the parsed information, the structured control information is obtained.

6. The method according to claim 4, characterized in that, The disassembly and assembly rules include fixed component rotation rules. The step of controlling the target model nodes corresponding to the target component in the target 3D model to perform the disassembly and assembly actions based on the structured manipulation information and the disassembly and assembly rules includes: When the target component in the structured manipulation information is a fastener, the thread direction of the fastener is identified; The rotation direction of the fastener is determined based on the thread direction and the action type in the structured control information. Control the target model node in the target 3D model corresponding to the fastener to rotate along the rotation direction at a preset rotation speed, and obtain the thread engagement state or end face contact state of the target model node during the rotation process; If the thread engagement state and / or end face contact state of the target model node meet the preset stop conditions, the target model node is controlled to stop rotating.

7. The method according to claim 4, characterized in that, The disassembly / assembly rules include movement distance rules. The step of controlling the target model nodes corresponding to the target component in the target 3D model to perform the disassembly / assembly action based on the structured manipulation information and the disassembly / assembly rules includes: If the motion parameters in the structured control information do not include the travel distance parameter, a motion path pre-scan is performed on the target model nodes based on the structured control information and the target 3D model to determine the safe travel distance of the target 3D model; Based on the safety travel and the structured control information, the target model node is controlled to perform the disassembly and assembly actions.

8. The method according to claim 4, characterized in that, The disassembly and assembly rules include disassembly and assembly sequence rules. The step of controlling the target model nodes corresponding to the target component in the target 3D model to perform the disassembly and assembly actions based on the structured manipulation information and the disassembly and assembly rules includes: Based on the hierarchical structure of the target 3D model and the subordinate constraint relationships between model nodes, determine the removal status of the external model nodes associated with the target model nodes; When the dismantling status indicates that dismantling has been completed, the target model node is controlled to perform the dismantling and assembly actions.

9. The method according to claim 4, characterized in that, The disassembly / assembly rules include collision and obstacle avoidance rules. The step of controlling the target model nodes corresponding to the target component in the target 3D model to perform the disassembly / assembly action based on the structured manipulation information and the disassembly / assembly rules includes: During the process of controlling the target model node to perform the disassembly and assembly actions, the distance between the target model node and the surrounding model nodes is obtained; When the interval distance is less than a preset distance threshold, or when the interval distance is greater than or equal to a preset distance, the target model node is controlled to shift by the preset distance in a preset offset direction.

10. An electronic device, characterized in that, include: Memory, used to store computer programs; A processor, configured to implement the steps of the video generation method as described in any one of claims 1 to 9 when executing the computer program.

Images