Cross-platform data synchronization method for assisting assembly of automobile rearview mirror

Abstract

This invention discloses a cross-platform data synchronization method for assisted assembly of automotive rearview mirrors, relating to the field of multi-source sensing data synchronization technology. Specifically, it includes: collecting assembly process data through multi-source sensing devices; using a redundant timing module and a high-precision clock synchronization protocol, employing an adaptive data fusion algorithm that integrates real-time error feedback and deviation tracing to generate a workstation status data packet containing complete assembly elements; after an assembly event is triggered, the system pushes the data packet to relevant heterogeneous controllers; each controller, combining adaptive processing cycle estimation and dynamic delay prediction models, generates a precise execution timestamp based on the event timestamp; and through a publish-subscribe architecture, sends control commands to the actuators and synchronously pushes assembly guidance information and execution status to the display terminal, ensuring that the synchronization error between information display and actual action is within a set threshold, effectively improving assembly accuracy and production efficiency.

Description

Technical Field

[0001] This invention belongs to the field of multi-source sensing data synchronization technology, specifically relating to a cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors. Background Technology

[0002] As the automotive industry transforms towards intelligent and flexible production, the assembly process of rearview mirrors, as precision components integrating optics, electronic control, and heating functions, faces increasingly stringent requirements for precision, efficiency, and consistency. Currently, automotive rearview mirror assembly lines are gradually upgrading from traditional manual operation to automated and semi-automated modes, widely introducing industrial robots, multi-source sensing devices, and heterogeneous controllers to form a multi-station collaborative production architecture. However, existing cross-platform data synchronization technology in automotive rearview mirror assembly lines still faces the following prominent issues: Insufficient accuracy of multi-source sensing data fusion: During rearview mirror assembly, industrial cameras are susceptible to interference from mirror reflections, six-dimensional force sensors experience data fluctuations due to instantaneous impacts during pressing, and temperature and humidity changes also affect measurement accuracy. Existing technologies mostly employ fixed-weight fusion algorithms, failing to consider the real-time operating status of sensors and assembly error feedback, resulting in distorted multi-source data fusion results. This fails to generate a "digital snapshot" reflecting the true assembly state, thus affecting the accuracy of subsequent control decisions. Poor synchronization of heterogeneous controllers: Different controllers (such as PLCs and robot controllers) have significant differences in processing speed and response latency, and the transmission latency of industrial networks fluctuates dynamically due to data volume and equipment load. Existing technologies typically achieve coordination based on fixed delay compensation or simple timestamp alignment, without establishing a dynamic delay prediction mechanism. This leads to disordered execution timing of control commands, such as the robot moving into position not being synchronized with the screwdriver tightening action, causing assembly deviations or equipment collision risks.

[0003] Therefore, there is an urgent need for a cross-platform data synchronization method for the auxiliary assembly of automotive rearview mirrors to solve the above problems. Summary of the Invention

[0004] The purpose of this invention is to provide a cross-platform data synchronization method for the auxiliary assembly of automotive rearview mirrors, which solves the technical problems in the prior art, such as insufficient accuracy of multi-source fusion, reliance on fixed delay compensation or simple timestamp alignment, resulting in disordered execution timing of control commands and asynchronous coordination of multiple actuators.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: Cross-platform data synchronization methods for automotive rearview mirror auxiliary assembly include: Step S1: For the auxiliary assembly system framework that includes multiple workstations, multiple heterogeneous controllers and a central server, each workstation collects multi-source sensing data in real time during the assembly process by deploying multi-source sensing devices. Step S2: The central server receives and parses the sensing data uploaded by each workstation, aligns the timestamps of the multi-source data based on a preset unified time reference, and fuses the spatially related multi-source sensing data based on a unified spatial coordinate reference to generate a workstation status data packet with a unique identifier. The fusion adopts an adaptive weighted fusion algorithm, which dynamically adjusts the weight coefficients according to the real-time error index of the sensor. Step S3: In response to the triggering of the assembly event, the central server pushes the workstation status data packet to the heterogeneous controller cluster associated with the workstation. Step S4: After receiving the workstation status data packet, each controller in the heterogeneous controller cluster performs parallel processing, generates corresponding control instructions according to its own control logic, and adds an execution timestamp associated with the triggering event to the control instructions based on a unified spatiotemporal reference. The execution timestamp is calibrated and generated based on the event timestamp that triggers the assembly event. Based on the event timestamp that triggers the assembly event, the processing cycle, the preset fixed delay compensation item, and the dynamic delay compensation item generated in real time by the LSTM delay prediction model are accumulated to obtain the target execution time. Step S5: Each controller sends the control command carrying the execution timestamp to the corresponding actuator, and pushes the assembly guidance information and the execution status of the control command to the display terminal of each workstation in real time through a cross-platform display synchronization protocol, so as to realize the information display and decision synchronization of the assembly process.

[0006] Furthermore, unifying the spatiotemporal benchmark specifically includes: Based on the dual-redundant clock consisting of the redundant timing module on the central server and the local crystal oscillator, the clock calibration of sensing devices and heterogeneous controllers at each workstation is achieved through the protocol of time-sensitive network, ensuring that the clock synchronization accuracy meets the assembly sequence requirements.

[0007] Furthermore, the fusion process employs an adaptive weighted fusion algorithm, dynamically adjusting the weighting coefficients based on the real-time error indicators of the sensors. These weighting coefficients are dynamically adjusted based on the measurement accuracy of each sensing device. The specific method is as follows: When the measurement error of a certain sensing device exceeds the preset allowable range, the weight coefficient is dynamically adjusted according to the sensor's real-time error index. When the variance of the sensor's measurement values ​​for multiple consecutive frames exceeds the threshold, its weight is reduced, and the reduced weight is redistributed to other sensors according to the accuracy ratio.

[0008] Furthermore, in response to the triggering of the assembly event, the specific method is as follows: Event triggering conditions are configured through a threshold matrix and logical rules preset by the central server. A targeted push mechanism is adopted. The central server uses a pre-stored workstation-controller association mapping table to push workstation status data packets only to the associated heterogeneous controllers participating in the assembly event, in order to save network bandwidth resources.

[0009] Furthermore, the event triggering conditions are configured through a threshold matrix and logical rules preset by the central server, specifically as follows: Assembly events are configured with a priority grading mechanism. Equipment fault alarm events and assembly parameter deviation events have higher priority than assembly start events and key process completion events. The central server prioritizes the targeted push tasks of workstation status data packets corresponding to high-priority events to ensure rapid response and collaborative processing of abnormal events.

[0010] Furthermore, after receiving the workstation status data packet, each controller in the heterogeneous controller cluster performs parallel processing, specifically as follows: Each controller independently receives the workstation status data packet, parses out the data subset associated with its execution logic, and performs control logic operations in parallel based on the data subset and a preset control algorithm to generate control instructions that conform to the controller's interface protocol.

[0011] Furthermore, the execution timestamp is generated based on the event timestamp that triggered the assembly event, specifically as follows: When an assembly event is triggered, the central server or event source device records the instant of the assembly event and defines it as an event timestamp. The assembly event information carrying the event timestamp is pushed to the heterogeneous controller cluster along with the workstation status data packet. After receiving the workstation status data packet, the heterogeneous controller cluster parses the event timestamp, calculates the execution timestamp of the control command to be executed based on the processing cycle required by its own control logic and the preset execution delay compensation parameters, and appends the execution timestamp to the control command.

[0012] Furthermore, based on the processing cycle required by each control logic and the preset execution delay compensation parameters, the execution timestamp at which the control instruction should be executed is calculated. The specific method is as follows: The processing cycle is dynamically determined by the controller through real-time monitoring of its own data parsing and logical operation time, or by calling the pre-stored baseline value of the processing cycle of the same type of task and adaptively adjusting it in combination with the current data complexity. The execution delay compensation parameters include fixed delay compensation items and dynamic delay compensation items. The fixed delay compensation items are set based on the controller hardware performance and the response characteristics of the actuator. The dynamic delay compensation items are generated by a pre-trained delay prediction model based on historical processing delay data, current network transmission status and assembly condition parameters. The calculation formula for the execution timestamp is: Execution timestamp = Event timestamp + Processing cycle + Fixed delay compensation item + Dynamic delay compensation item. During the calculation process, if the processing cycle is detected to exceed the preset fluctuation range in real time, the dynamic delay compensation item is automatically updated in real time to ensure the accuracy of the execution timestamp calculation and the timing synchronization of the control command execution.

[0013] Furthermore, the implementation method of the cross-platform display synchronization protocol is as follows: The cross-platform display synchronization protocol adopts a publish-subscribe architecture. Each controller acts as a publisher, encapsulating the control command execution status and assembly guidance information in a unified data format and publishing it to the protocol bus. Each workstation display terminal acts as a subscriber, subscribing to associated data based on its own workstation identifier. The protocol supports data compression transmission and breakpoint resumption, reduces the load of a single transmission through data fragmentation technology, and configures a data verification mechanism to ensure transmission integrity. After receiving the data, the display terminal calibrates its local display clock based on a unified spatiotemporal reference, ensuring that the display time of the assembly guidance information and execution status is aligned with the control command execution timestamp.

[0014] In summary, due to the adoption of the above technical solution, the beneficial effects of the present invention are: 1. This invention solves the problem of multi-source data fusion distortion caused by sensor interference and lack of error feedback in the existing technology by integrating an adaptive weight adjustment mechanism that combines real-time error monitoring and deviation tracing closed-loop optimization. It realizes the generation of digital snapshots that can accurately reflect the real state of assembly, provides reliable data support for control decisions, significantly reduces assembly accuracy deviation caused by data distortion, and ensures the consistency of key assembly indicators. 2. This invention solves the problem of poor collaborative synchronization caused by differences in processing speed of heterogeneous controllers and dynamic fluctuations in network latency by using redundant time synchronization and microsecond-level clock synchronization protocol, combined with adaptive processing cycle estimation and dynamic delay prediction model. It realizes high-precision timing collaboration of multiple devices and multiple workstations, avoids control command execution errors, reduces assembly deviation and equipment collision risks, and improves collaborative stability and reliability. 3. This invention solves the problems of asynchronous display and execution status and information lag in the prior art by using a unified spatiotemporal reference and a publish-subscribe architecture cross-platform synchronization protocol. It achieves precise synchronization between assembly guidance information and the actions of the actuator, greatly reduces information lag error, lowers the operator error rate, fully leverages the auxiliary value of advanced display equipment, and improves assembly efficiency and operational safety. Attached Figure Description

[0015] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0016] Figure 1 The diagram illustrates the steps of the cross-platform data synchronization method for the auxiliary assembly of automotive rearview mirrors according to the present invention. Figure 2 The flowchart of the adaptive weighted data fusion based on a unified spatiotemporal benchmark of the present invention is shown; Figure 3 The flowchart of the parallel control instruction generation and execution timing calibration based on event timestamps of the present invention is shown. Detailed Implementation

[0017] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0018] like Figure 1 The cross-platform data synchronization method for automotive rearview mirror auxiliary assembly shown includes the following steps: Step S1: For the auxiliary assembly system framework that includes multiple workstations, multiple heterogeneous controllers and a central server, each workstation collects multi-source sensing data in real time during the assembly process by deploying multi-source sensing devices. Multi-source sensing devices at each workstation collect assembly process data in real time according to a preset sampling frequency, as specifically implemented as follows: An industrial camera captures the positioning features of the rearview mirror housing and the edge position of the lens, outputs image data, and transmits it to the vision controller via the CameraLink interface, simultaneously carrying a hardware timestamp. The laser displacement sensor measures the gap distance between the lens and the housing, and outputs an analog signal which is converted into digital data by an AD converter. A six-dimensional force sensor monitors torque data during the pressing process and provides real-time feedback on the pressing contact status; The RFID reader reads the RFID tags on the workpiece tray to obtain identification information such as workpiece model, process completion status, and serial number. Temperature and humidity sensors collect temperature and humidity data of the workstation environment to provide environmental parameters for assembly accuracy compensation; The bolt tightening torque sensor collects real-time torque data during the bolt tightening process and records the entire tightening process curve; All sensed data undergoes local preprocessing at the workstation edge nodes (including data noise reduction, outlier removal, and format standardization) before being uploaded to the central server via a TSN switch. During data transmission, the hardware timestamp is preserved to ensure the time-series traceability of the original data.

[0019] After each workpiece completes assembly at its current station, it proceeds to the downstream inspection station. Inspection equipment (such as a vision measurement system or an electrical tester) collects the actual assembly deviation value of the workpiece (e.g., the deviation of the lens installation position, the difference between the measured bolt torque value and the target value). This measured value is fed back to the central server in real time via the TSN network.

[0020] Step S2: The central server receives and parses the sensing data uploaded by each workstation, aligns the timestamps of the multi-source data based on a preset unified time reference, and fuses the spatially related multi-source sensing data based on a unified spatial coordinate reference to generate a workstation status data packet with a unique identifier. The workstation status data packet contains complete digital snapshot information of all assembly elements of the workstation at a specific time. like Figure 2As shown, after receiving multi-source sensing data, the central server completes the time alignment of the data based on a unified time reference. The server has a built-in dual-redundant time synchronization module with primary GPS time synchronization, backup Beidou time synchronization, and a local temperature-controlled crystal oscillator. It periodically broadcasts the master clock time to all network devices through the TSN's gPTP generalized precise time synchronization protocol. After the slave clocks of each sensing device and controller are synchronized and calibrated, the clock synchronization accuracy of the entire network can meet the microsecond-level timing requirements of the assembly process. On this basis, the central server executes adaptive fusion logic for generating complete digital snapshots to deal with the unique problems of rearview mirror assembly, such as visual data fluctuations caused by mirror reflection, sensor interference from instantaneous impact during pressing, and the impact of environmental temperature and humidity changes on measurement accuracy. First, the hardware timestamp of the data packet is extracted, and the data within the same workstation and the same time window are grouped to complete spatiotemporal alignment and association. On the basis of time alignment, the spatial association of the data is further completed based on a unified spatial coordinate reference. Based on the pre-calibrated workstation spatial coordinate system, the measurement values ​​from different sources are mapped to the same reference system, establishing the correspondence between assembly elements such as the lens center position, shell edge contour, pressing force vector, and bolt torque value and multi-source observation values. For the same physical quantity that needs to be fused, the system adopts an initial weight allocation method based on precision perception, assuming that there are n sensor measurements for the same physical quantity. (i=1, 2, ..., n), corresponding standard deviation of measurement accuracy Pre-calibration, such as in the lens positioning process, where the laser displacement sensor is less affected by ambient light, and the standard deviation of position measurement. Industrial cameras are susceptible to interference from specular reflections, and their standard deviation is... The corresponding basic weights are based on The system continuously monitors the variance of measurements from each sensor and defines a real-time error index for sensor i. When the variance of positioning results in three consecutive frames exceeds a threshold (e.g., 0.1mm) due to lens reflection from industrial cameras or other sensors, this threshold is set based on the nominal accuracy of the industrial camera (0.05mm) and the principle of three times the standard deviation to ensure no misjudgment within the normal fluctuation range. The sensor is then deemed unreliable, and its corresponding weighting coefficient is adjusted accordingly. Adjusted to and reduce the weight Redistribute the data according to the accuracy ratio of other sensors, using the formula. Determine the redistribution method, where Let j represent sensor j. Based on assembly error feedback adjustment, the assembly results of subsequent processes are introduced as feedback. If subsequent visual inspection finds a systematic deviation in the lens position, the central server performs reverse trace analysis to determine the rationality of the weight allocation and defines a deviation feedback factor. ,in The assembly deviation values ​​measured in subsequent processes are collected in real time by subsequent visual inspection or electrical testing equipment. The preset assembly tolerance threshold is set according to the product assembly process requirements. In this embodiment, the process tolerance threshold is set to 1. If δ>1, the process tolerance threshold is determined by the formula... Correcting the preset weight baseline of the vision sensor, where This represents the corrected weight baseline for the visual sensor. This indicates that the baseline weight of the visual sensor before correction is equal to the base weight of the visual sensor in the initial weight allocation. The learning rate for the preset fusion model is determined based on offline simulation.

[0021] If subsequent visual inspection reveals a systematic deviation in the lens position, the central server reverse-engineers the weight allocation for rationality. If δ>1 (the preset assembly allowable deviation threshold) and it is confirmed that the visual sensor weight is too high, causing the deviation to be amplified, the above-mentioned weight baseline correction operation is performed to achieve closed-loop optimization of the fusion model.

[0022] Each workstation generates a workstation status data packet with a unique UUID identifier after completing one data fusion cycle (the fusion cycle is preset based on the sensor sampling frequency and system real-time requirements). The data packet contains the system timestamp at the time of fusion. RFID-read workpiece model and serial number, lens position coordinates Pressing force value Bolt torque Visually guided feature point matching results, digital snapshots of environmental temperature and humidity assembly elements, sensor operating status, and current weighting coefficients. The real-time error index sensor status flag data packets are encapsulated in JSON format, transmitted via TSN stream to the central server database for archiving and real-time push.

[0023] The data fusion cycle is set based on the highest sampling frequency of the sensors in the workstation (100fps for industrial cameras, 10ms cycle) and the minimum cycle of the control command (10ms), ensuring that each cycle contains at least the latest sampled values ​​of each sensor and meets the real-time control requirements.

[0024] Step S3: In response to the triggering of the assembly event, the central server pushes the workstation status data packet to the heterogeneous controller cluster associated with the workstation. The central server pre-stores an event configuration table containing a threshold matrix and logical rules, defining four types of assembly events, namely the assembly start event triggered when the RFID reads a new workpiece and the workstation is idle; Pressing force reaches And maintain the critical process completion event triggered at time t1. The preset pressing force target threshold is set based on the workpiece assembly and fastening requirements, and t1 is the stable holding time to ensure that the pressing is in place; Lens position deviation or Assembly parameter deviation event triggered at the time, The reference position coordinates for the lens are determined by tooling calibration; 0.2mm is the position deviation threshold in this embodiment. The lower limit threshold for torque is set based on the product technical specifications. The device fault alarm event is triggered when the controller reports an alarm from the servo drive or when the sensor self-test fails. All events are configured with a priority tiering mechanism. Fault alarm events are of the highest priority (P0), parameter deviation events are of the highest priority (P1), and process completion and start events are of the highest priority (P2 and P3), respectively. The central server maintains the event queue and prioritizes high-priority events. The central server also maintains a workstation-controller association mapping table (the mapping table is preset based on the association between workstation functions and controller responsibilities) to record the list of controllers that each workstation needs to notify. After an event is triggered, the server immediately retrieves the mapping table and pushes the workstation status data packet, which contains only the subset of data required by the associated controller, to the corresponding controller via a TSN stream. The push uses UDP multicast, and each controller listens to a designated multicast address to reduce network bandwidth consumption. High-priority events can be notified to the controller for immediate processing via an interrupt signal.

[0025] Step S4: After receiving the workstation status data packet, each controller in the heterogeneous controller cluster performs parallel processing, generates corresponding control instructions according to its own control logic, and adds an execution timestamp associated with the triggering event to the control instructions based on a unified spatiotemporal reference. The execution timestamp is calibrated and generated based on the event timestamp that triggers the assembly event. like Figure 3 As shown, each controller in the heterogeneous controller cluster independently receives workstation status data packets and parses out a subset of data that matches its own execution logic. The robot controller, for example, parses the lens position coordinates. With angle The torque controller analyzes the target torque value. With current torque feedback The vision controller analyzes the image feature point matching results and the deviation. The machine-to-machine (M2M) and AR display controllers analyze the process completion status, alarm information, and operation instructions. Each controller performs parallel logical operations based on a preset control algorithm to generate control commands that conform to its own interface protocol, including the robot's target position. With speed Position control commands ( Generated by kinematic programming algorithm, (Based on assembly efficiency and stability requirements) Target bolt tightening torque With slope control parameters Torque control command ( As a preset value, The visual guidance instructions include the torque rise slope (determined based on bolt material and thread specifications), updating the visual template or triggering a photo, and displaying operation instructions. To address the coordination challenges posed by differences in processing speeds and network latency fluctuations among heterogeneous controllers, this invention employs an execution timestamp generation mechanism based on event sourcing. Each control instruction is assigned a future execution time bound to its original triggering event. At the moment the event is triggered, the central server records a high-precision event timestamp. This data is encapsulated along with the workstation status data packet as metadata, unifying the time base of all control actions to the physical occurrence time of the event. After receiving the data packet, each heterogeneous controller determines the historical average processing cycle baseline value. , The complexity factor is generated by statistically analyzing historical processing data of similar tasks stored locally on the controller, and then combining the size and complexity of the data subset. , Dynamically calculate based on the size of the data subset and task complexity (such as point cloud density, number of faces in the 3D model, etc.). and The product of these is used as the processing cycle assigned to each control instruction bound to the original triggering event, i.e. At the same time, a high-precision timer is activated to record the actual processing time. If the actual time taken deviates from the estimated time by more than the pre-approval threshold, such as This 15% threshold is determined based on historical processing cycle fluctuation statistics, and will then... Feedback is sent to the controller's local storage for updating. value; Calculate the delay compensation term. Delay compensation is divided into fixed compensation term and dynamic delay compensation term. Fixed compensation term Based on the physical inertia of the actuator (such as a robot controller) =2ms, servo driver response latency =1ms), dynamic delay compensation item Generated by a lightweight LSTM latency prediction model from the controller or edge node; The LSTM model is pre-trained using historical factory operation data, and the model is input with the current end-to-end delay sample value of the TSN network. Historical trend of controller CPU utilization Variance of processing cycle for similar tasks Current temperature of the implementing agency Each heterogeneous controller and the central server achieve nanosecond-level clock synchronization via the gPTP protocol. In each control cycle (10ms), the network interface card (NIC) hardware inserts a precise transmission timestamp Tt into the data packets sent by the controller to the central server at the physical layer. Similarly, the central server records the reception timestamp Tr upon reception. The difference between the two, Tt-Tr, is the current end-to-end latency sampling value of the TSN network. The controller system monitoring module samples the CPU utilization rate every t1 time interval (determined according to actual needs), and calculates the moving average of the two sample values ​​within the past 2t1 time intervals. Based on the last 10 similar tasks The sample variance is calculated by updating it once for each control command generated. The temperature data is collected every 200ms by a PT100 temperature sensor installed on the actuator (robot joint, servo motor). .

[0026] A single-layer LSTM network is used, with 64 hidden layer neurons. The input sequence length is 10 (using feature vectors from the past 10 time steps), and the input feature dimension is 4. The LSTM network is followed by a fully connected layer (output dimension 1), the activation function is tanh, and the loss function is mean squared error (MSE).

[0027] Model training: Training set: Collects 30 consecutive days of historical operation data from the factory, containing more than 1 million feature vectors and corresponding labels; Tag definition: ,in This represents the actual total delay, which is the measured time from instruction generation to the start of the execution mechanism's response; Training parameters: Adam optimizer was used, initial learning rate was 0.001, batch size was 64, number of training epochs was 100, and an early stopping strategy was adopted during training (training was stopped if the validation set loss did not decrease for 5 consecutive epochs). Model deployment: After training is complete, the model weights are stored as binary files and deployed locally on the controller; Inference process: The model infers once every inference cycle (the inference cycle of the LSTM model can be adaptively adjusted according to the operating conditions, usually set to 5-10 times the control cycle, for example, between 50ms and 200ms). The controller obtains the feature sequences of the latest 10 time steps, inputs them into the model for backpropagation, and outputs... This meets the real-time requirements.

[0028] After training, the model is deployed on the controller or edge nodes. During real-time operation, the input sequence is updated every T1 time interval, and the forward propagation yields... Determine the event timestamp Based on this timestamp, by accumulating and The product of the processing cycle, fixed compensation term, and dynamic delay compensation term yields the execution timestamp of the control command. .

[0029] Set event timestamp The controller monitors the actual processing time in real time by counting microseconds from the system's baseline time to the time the event occurs. (Unit: microseconds), fixed delay compensation item (Unit: microsecond) Based on the physical inertia preset of the actuator, dynamic delay compensation. (Unit: microseconds) is generated by the LSTM delay prediction model. The total delay time is then... The execution timestamp at which the instruction should be executed. (Still starting from the same reference, the microsecond count) is determined by the following formula: +

[0030] Through this mechanism, although the PLC, robot, and AR controller have vastly different processing speeds, the generated instructions all carry the same... Execution timestamps used as a benchmark to accurately compensate for differences in processing paths This enables the synchronized execution of robot physical motion, screwdriver torque output, and AR glasses guidance information.

[0031] Step S5: Each controller sends control commands carrying execution timestamps to the corresponding actuators, and pushes the assembly guidance information and the execution status of the control commands to the display terminals of each workstation in real time through a cross-platform display synchronization protocol, so as to realize the information display and decision synchronization of the assembly process.

[0032] Each controller carries control commands with execution timestamps and sends them to corresponding actuators such as robot servo drives, electric screwdrivers, and cylinders via fieldbuses such as EtherCAT and Profinet. The actuators then perform precise actions according to the command target values ​​and execution timestamps, enabling multi-mechanism collaborative operation. To achieve real-time synchronous display of assembly guidance information and control command execution status, this invention adopts a publish-subscribe cross-platform display synchronization protocol. Each controller acts as a publisher, displaying control command execution status, assembly guidance information, and event timestamps. With execution timestamp After being encapsulated in the Protocol Buffers unified data format, it is published via the TSN protocol bus, supporting Snappy compression algorithm compression (compression rate of approximately 70%, based on typical assembly data statistics) and 512KB chunked transmission of data larger than 1MB (512KB is the chunk size threshold, balancing transmission efficiency and reliability); each workstation display terminal acts as a subscriber, registering its own workstation identifier upon startup. It subscribes to related message topics, verifies data integrity using CRC32 (CRC32 is a preset data verification algorithm) after receiving messages, requests retransmission via breakpoint resume when packets are lost, and aligns message timestamps with the local clock based on a unified time base after receiving messages. Then, it dynamically displays assembly guidance information and execution status according to the execution timestamp sequence, such as overlaying virtual operation arrows in AR glasses and highlighting the OK status when the torque is up to standard, ensuring that the information seen by the operator is synchronized with the robot's actual actions.

[0033] The above formulas are all dimensionless calculations, and the preset parameters in the formulas should be set by those skilled in the art according to the actual situation.

[0034] The above description is only a preferred embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any equivalent substitutions or modifications made by those skilled in the art within the scope of the technology disclosed in the present invention, based on the technical solution and inventive concept of the present invention, should be covered within the scope of protection of the present invention.

[0035] The preferred embodiments of the present invention disclosed above are merely illustrative of the invention. These preferred embodiments do not exhaustively describe all details, nor do they limit the invention to specific implementations. Clearly, many modifications and variations can be made based on the content of this specification. This specification selects and specifically describes these embodiments to better explain the principles and practical applications of the invention, thereby enabling those skilled in the art to better understand and utilize the invention. The invention is limited only by the claims and their full scope and equivalents.

Claims

1. A cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors, characterized in that, include: Step S1: For the auxiliary assembly system framework that includes multiple workstations, multiple heterogeneous controllers and a central server, each workstation collects multi-source sensing data in real time during the assembly process by deploying multi-source sensing devices. Step S2: The central server receives and parses the sensing data uploaded by each workstation, aligns the timestamps of the multi-source data based on a preset unified time reference, and fuses the spatially related multi-source sensing data based on a unified spatial coordinate reference to generate a workstation status data packet with a unique identifier. The fusion adopts an adaptive weighted fusion algorithm, which dynamically adjusts the weight coefficients according to the real-time error index of the sensor. Step S3: In response to the triggering of the assembly event, the central server pushes the workstation status data packet to the heterogeneous controller cluster associated with the workstation. Step S4: After receiving the workstation status data packet, each controller in the heterogeneous controller cluster performs parallel processing, generates corresponding control instructions according to its own control logic, and adds an execution timestamp associated with the triggering event to the control instructions based on a unified spatiotemporal reference. The execution timestamp is calibrated and generated based on the event timestamp that triggers the assembly event. Based on the event timestamp that triggers the assembly event, the processing cycle, the preset fixed delay compensation item, and the dynamic delay compensation item generated in real time by the LSTM delay prediction model are accumulated to obtain the target execution time. Step S5: Each controller sends the control command carrying the execution timestamp to the corresponding actuator, and pushes the assembly guidance information and the execution status of the control command to the display terminal of each workstation in real time through a cross-platform display synchronization protocol, so as to realize the information display and decision synchronization of the assembly process.

2. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 1, characterized in that, Generate a workstation status data packet with a unique identifier, specifically including: The unique identifier is composed of a unique workstation code, an assembly event type identifier, and a timestamp, ensuring that the data packet is globally unique. The data packet is encapsulated in a standardized format and includes a unique serial number of the assembly object, real-time assembly core parameters, multi-source sensor status data, environmental perception parameters, and a data integrity check code. The real-time assembly core parameters cover positioning coordinates, assembly forces, tightening torques, and assembly gaps. The data integrity check code is generated using a preset hash algorithm and is used to verify the integrity and tamper-proof nature of the data packet during transmission and storage.

3. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 1, characterized in that, The fusion process employs an adaptive weighted fusion algorithm, dynamically adjusting the weighting coefficients based on the real-time error indicators of the sensors. These weighting coefficients are dynamically adjusted according to the measurement accuracy of each sensing device. The specific method is as follows: When the measurement error of a certain sensing device exceeds the preset allowable range, the weight coefficient is dynamically adjusted according to the sensor's real-time error index. When the variance of the sensor's measurement values ​​for multiple consecutive frames exceeds the threshold, its weight is reduced, and the reduced weight is redistributed to other sensors according to the accuracy ratio.

4. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 1, characterized in that, In response to the assembly event, the specific method is as follows: Event triggering conditions are configured through a threshold matrix and logical rules preset by the central server. A targeted push mechanism is adopted. The central server uses a pre-stored workstation-controller association mapping table to push workstation status data packets only to the associated heterogeneous controllers participating in the assembly event, in order to save network bandwidth resources.

5. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 4, characterized in that, Event triggering conditions are configured through a threshold matrix and logical rules preset by the central server. The specific method is as follows: Assembly events are configured with a priority grading mechanism. Equipment fault alarm events and assembly parameter deviation events have higher priority than assembly start events and key process completion events. The central server prioritizes the targeted push tasks of workstation status data packets corresponding to high-priority events to ensure rapid response and collaborative processing of abnormal events.

6. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 1, characterized in that, After receiving the workstation status data packet, each controller in the heterogeneous controller cluster processes it in parallel. The specific method is as follows: Each controller independently receives the workstation status data packet, parses out the data subset associated with its execution logic, and performs control logic operations in parallel based on the data subset and a preset control algorithm to generate control instructions that conform to the controller's interface protocol.

7. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 1, characterized in that, The execution timestamp is generated based on the event timestamp that triggered the assembly event. The specific method is as follows: When an assembly event is triggered, the central server or event source device records the instant of the assembly event and defines it as an event timestamp. The assembly event information carrying the event timestamp is pushed to the heterogeneous controller cluster along with the workstation status data packet. After receiving the workstation status data packet, the heterogeneous controller cluster parses the event timestamp, calculates the execution timestamp of the control command to be executed based on the processing cycle required by its own control logic and the preset execution delay compensation parameters, and appends the execution timestamp to the control command.

8. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 7, characterized in that, Based on the processing cycle required by each control logic and the preset execution delay compensation parameters, the execution timestamp at which the control command should be executed is calculated. The specific method is as follows: The processing cycle is dynamically determined by the controller through real-time monitoring of its own data parsing and logical operation time, or by calling the pre-stored baseline value of the processing cycle of the same type of task and adaptively adjusting it in combination with the current data complexity. The execution delay compensation parameters include fixed delay compensation items and dynamic delay compensation items. The fixed delay compensation items are set based on the controller hardware performance and the response characteristics of the actuator. The dynamic delay compensation items are generated by a pre-trained delay prediction model based on historical processing delay data, current network transmission status and assembly condition parameters. The calculation formula for the execution timestamp is: Execution timestamp = Event timestamp + Processing cycle + Fixed delay compensation item + Dynamic delay compensation item. During the calculation process, if the processing cycle is detected to exceed the preset fluctuation range in real time, the dynamic delay compensation item is automatically updated in real time to ensure the accuracy of the execution timestamp calculation and the timing synchronization of the control command execution.

9. The cross-platform data synchronization method for auxiliary assembly of automotive rearview mirrors according to claim 1, characterized in that, The implementation method of the cross-platform display synchronization protocol is as follows: The cross-platform display synchronization protocol adopts a publish-subscribe architecture. Each controller acts as a publisher, encapsulating the control command execution status and assembly guidance information in a unified data format and publishing it to the protocol bus. Each workstation display terminal acts as a subscriber, subscribing to associated data based on its own workstation identifier. The protocol supports data compression transmission and breakpoint resumption, reduces the load of a single transmission through data fragmentation technology, and configures a data verification mechanism to ensure transmission integrity. After receiving the data, the display terminal calibrates its local display clock based on a unified spatiotemporal reference, ensuring that the display time of the assembly guidance information and execution status is aligned with the control command execution timestamp.

Images