Construction robot remote monitoring and virtual-real synchronous control platform based on digital twinning

Abstract

The present application relates to the technical field of robot remote monitoring, in particular to a construction robot remote monitoring and virtual-real synchronous control platform based on digital twinning, comprising: a data acquisition and preprocessing unit; a digital twinning construction and mapping unit; a remote monitoring and virtual-real synchronous control unit, adopting a UKF algorithm for predicting motion state by introducing an environmental confidence factor, a dynamic time window for compensating delay, a dynamic safety threshold and a simulation check to ensure the safety of the instruction, and a virtual-real consistency check unit. The present application is aimed at the construction scene of dust obscuration interference and network fluctuation, and effectively improves the stability of construction robot motion state prediction, virtual-real synchronous accuracy and instruction issuing safety by predicting motion state through the UKF algorithm for introducing an environmental confidence factor, compensating delay through a dynamic time window, ensuring the safety of the instruction through a dynamic safety threshold and a simulation check, and triggering a calibration warning when the virtual-real deviation exceeds the limit.

Description

Technical Field

[0001] This invention relates to the field of robot remote monitoring technology, and more specifically, to a construction robot remote monitoring and virtual-real synchronous control platform based on digital twins. Background Technology

[0002] With the rapid development of digital twins, the Internet of Things, and intelligent control technologies, remote monitoring and virtual operation of construction robots has become an important research direction in the field of building automation. By constructing a real-time mapping relationship between physical robots and virtual models, precise control and status monitoring of construction equipment can be achieved remotely, effectively reducing the direct operational risks to personnel in complex and hazardous environments.

[0003] Patents related to digital twin and remote control technologies for construction robots have been published. Invention patent CN202510884570.X discloses a method and system for optimizing and monitoring multi-robot collaborative operations of construction robots based on digital twins. This method constructs a digital twin containing an environmental perception model and a kinematic / dynamic model based on a building information model and robot physical parameters; performs multi-robot task allocation and path planning including static / dynamic obstacle avoidance in a virtual environment; synchronizes physical environment data in real time through multi-modal sensors; dynamically optimizes operation parameters and achieves closed-loop control using genetic algorithms or particle swarm optimization algorithms; generates safety warnings and emergency response plans based on machine learning; and innovatively integrates Markov decision-making path planning based on reinforcement learning with multi-agent game-based task allocation, employing multi-source data fusion of LiDAR, vision, and inertial navigation. Invention patent CN201910720426.7 discloses a construction robot based on virtual reality control, including a robotic arm. The hinge of the robotic arm corresponds to the joint state of a human arm. Sensors capture the motion position information of the joint of the human arm. The acquired motion position information is optimized by a joint motion angle estimation method based on H∞ filtering recursion to achieve human-computer interaction control of the robot.

[0004] However, the aforementioned existing technologies still have the following technical defects: in construction scenarios with environmental interference such as dust and obstructions, as well as network latency fluctuations, it is impossible to ensure the stable prediction of the construction robot's motion state by the remote control platform, the accurate synchronization of virtual and real actions, and the safe issuance of remote commands; although invention patent CN202510884570.X constructs a digital twin and adopts multi-source data fusion, its virtual-real synchronization mainly relies on the real-time transmission of conventional data, without considering the quantitative impact of environmental interference on the confidence of state prediction, nor dynamically adjusting the synchronization compensation strategy according to environmental changes, resulting in increased state prediction deviation and inaccurate virtual-real synchronization in harsh environments; although invention patent CN201910720426.7 uses H∞ filtering to improve the robustness of joint angle estimation, its solution is geared towards on-site human-machine interaction rather than remote virtual control, and cannot cope with the virtual-real time sequence offset problem caused by remote network latency, and does not provide active safety verification methods before issuing remote commands, making it difficult to guarantee the safety of remote control in complex construction environments. In view of this, we propose a construction robot remote monitoring and virtual-real synchronization control platform based on digital twins. Summary of the Invention

[0005] The purpose of this invention is to provide a remote monitoring and virtual-real synchronous control platform for construction robots based on digital twins, in order to solve the problems mentioned in the background art, such as the inability to ensure stable prediction of the movement state of construction robots, accurate synchronization of virtual and real actions, and safe issuance of remote commands by the remote control platform in construction scenarios with environmental interference such as dust and obstruction, as well as network latency fluctuations.

[0006] To address the aforementioned technical problems, the present invention aims to provide a remote monitoring and virtual-real synchronous control platform for construction robots based on digital twins, comprising: The data acquisition and preprocessing unit collects the operation data and environmental perception data of the construction robot, cleans and converts them according to protocols to obtain standardized operation data. A digital twin construction and mapping unit receives standardized operational data, constructs a digital twin of the construction robot, and realizes the state mapping between the physical entity and the virtual model; The remote monitoring and virtual-real synchronization control unit receives remote control commands, and based on the digital twin, adopts an unscented Kalman filter that integrates environmental perception information and a dynamic time window adaptive virtual-real synchronization control algorithm to perform dynamic safety threshold command verification and delay compensation operations. The verified commands are then sent to the construction robot, and the execution status is synchronously fed back to the digital twin. The virtual-real consistency verification unit compares the virtual state data of the digital twin with the physical operation data of the construction robot and performs a consistency verification.

[0007] As a further improvement to this technical solution, the data acquisition and preprocessing unit includes a data acquisition module, a data cleaning module, a protocol conversion module, and a data normalization and output module, wherein: The data acquisition module is used to collect the operation data and environmental perception data of the construction robot, and transmit the collected raw data to the data cleaning module in real time. The operation data includes position, posture, speed, joint angle and equipment status data, and the environmental perception data includes construction environment images, laser point clouds and obstacle distance data. The data cleaning module is used to remove and complete outliers and missing values ​​in the original data, and then transmit the cleaned data to the protocol conversion module. The protocol conversion module is used to convert the custom device protocol of the construction robot into a unified standard communication protocol, so as to achieve data format unification, obtain preliminary standardized data, and transmit it to the data straightening and output module. The data normalization output module is used to uniformly encode and time-series align the preliminary standardized data to obtain standardized operational data, and transmit the standardized operational data to the digital twin construction and mapping unit.

[0008] As a further improvement to this technical solution, the digital twin construction and mapping unit includes an initial model construction module, a data-driven twin calibration module, and a virtual-real state synchronization mapping module, wherein: The initial model building module is used to build an initial virtual model of the construction robot and transmit it to the data-driven twin calibration module; The data-driven twin calibration module is used to receive standardized operating data, calibrate the parameters of the initial virtual model of the construction robot, obtain the digital twin of the construction robot, and transmit it to the virtual-real state synchronization mapping module. The virtual-real state synchronization mapping module is based on the digital twin of the construction robot to realize real-time state mapping between the physical entity and the virtual model, and provides a basis for virtual simulation and state prediction for remote control.

[0009] As a further improvement to this technical solution, the data-driven twin calibration module includes a model parameter extraction submodule, a state deviation calculation submodule, and a model parameter update submodule, wherein: The model parameter extraction submodule is used to extract the inherent parameters of the initial virtual model of the construction robot and transmit them to the state deviation calculation submodule; The state deviation calculation submodule calculates the state deviation between the physical entity of the construction robot and the initial virtual model based on standardized operating data and transmits it to the model parameter update submodule. The model parameter update submodule updates the parameters and state covariance information of the initial virtual model based on the state deviation to obtain a digital twin of the construction robot.

[0010] As a further improvement to this technical solution, the remote monitoring and virtual-real synchronization control unit includes an instruction receiving and parsing module, a virtual-real status synchronization module, an instruction virtual verification module, and an instruction issuing and status feedback module, wherein: The instruction receiving and parsing module is used to receive remote control instructions, parse the remote control instructions, and transmit them to the instruction virtual verification module. The virtual-real state synchronization module is based on a digital twin and incorporates confidence factors for dust and occlusion sensing. An improved unscented Kalman filter algorithm is used to predict the physical state of a construction robot, and the corrected prediction confidence level is then applied. Used to dynamically adjust the synchronization time window, generate an adaptively adjusted synchronization time window, and complete the alignment of virtual and real states and delay compensation; The instruction virtual verification module is based on a digital twin, the compensated virtual-real synchronization state, and the corrected prediction confidence level. Perform multi-step instruction simulation verification and base the results on the corrected prediction confidence level. The system dynamically adjusts the security threshold and transmits verified instructions to the instruction issuance and status feedback module. When instruction verification fails, it generates a verification failure receipt and the reason for the failure and returns it to the remote control terminal. The instruction issuance and status feedback module is used to issue verified instructions to the construction robot and simultaneously feed back the execution status of the construction robot to the digital twin.

[0011] As a further improvement to this technical solution, the virtual-real state synchronization module includes an environment perception submodule, a network latency dynamic monitoring submodule, an improved unscented Kalman filter state prediction submodule, and a dynamic time window synchronization submodule, wherein: The environmental perception submodule is used to acquire image data of the construction environment in real time and calculate the confidence factors for dust and occlusion perception. And transmit it to the improved unscented Kalman filter state prediction submodule; The network latency dynamic monitoring submodule is used to monitor network transmission latency in real time and generate latency statistics. The improved unscented Kalman filter state prediction submodule is based on state data from a digital twin and confidence factors for dust and occlusion perception. Predict the physical state of the construction robot and calculate the corrected prediction confidence level. When the corrected prediction confidence level When the value falls below a preset minimum safety threshold, the system security protection mechanism is triggered, prohibiting the issuance of new remote control commands and sending an environmental anomaly warning to the remote control terminal, and adjusting the predicted confidence level. Simultaneously transmitted to the dynamic time window synchronization submodule and the instruction virtual verification module; The dynamic time window synchronization submodule is used to combine delay statistics with the corrected prediction confidence. Generate an adaptive synchronization time window This completes the alignment of virtual and real states and delay compensation.

[0012] As a further improvement to this technical solution, the process of predicting the physical state of the construction robot and calculating the corrected prediction confidence in the improved unscented Kalman filter state prediction submodule includes the following steps: S32.1 Extracting the state vector of the construction robot corresponding to the digital twin. and covariance matrix Determine the state dimension parameters; S32.2 Generate a finite set of Sigma points based on the unscented Kalman filter mechanism to adapt to the nonlinear motion characteristics of the construction robot; S32.3. Perform nonlinear mapping transformation on the Sigma point based on the nonlinear kinematic model of the construction robot; S32.4. Perform weighted fusion processing on the transformed Sigma points to output the predicted physical state of the construction robot. and the updated covariance matrix ; S32.5, Based on the updated covariance matrix The trace is used to calculate the original prediction confidence based on the unscented Kalman filtering mechanism. ; S32.6, Introduce the dust and occlusion sensing confidence factors output by the environmental perception submodule. The original prediction confidence level is corrected to obtain the corrected prediction confidence level. .

[0013] As a further improvement to this technical solution, the process of generating an adaptive synchronization time window based on delay statistics and corrected prediction confidence and achieving virtual-real synchronization in the dynamic time window synchronization submodule includes the following steps: S32.7 Continuously collect multiple sets of network delay samples This forms a sliding delay sample set; S32.8 Perform statistical operations on the delay sample set to obtain the mean delay. With delay standard deviation ; S32.9, Based on statistical characteristics, the corrected prediction confidence level and the preset minimum time window Maximum time window and nonlinear adjustment coefficient Construct an adaptive synchronization time window ; S32.10, with adaptive synchronization time window Based on this, the state of the digital twin is aligned with the predicted physical state of the construction robot to complete timing calibration and delay compensation.

[0014] As a further improvement to this technical solution, the instruction virtual verification module includes a multi-step kinematics simulation submodule, a collision detection simulation submodule, and a dynamic safety threshold comparison submodule, wherein: The multi-step kinematics simulation submodule is based on a digital twin and a corrected predicted physical state. To make future remote control commands Kinematic simulation of steps The value range is 3-5; The collision detection simulation submodule is used to detect whether there is a risk of spatial collision in each simulated action within the digital twin virtual scene; The dynamic security threshold comparison submodule is based on the corrected prediction confidence level. Dynamically adjust the safety threshold, and the safety threshold is related to the corrected prediction confidence. A positive correlation is established, and the results of multiple simulation steps are compared with the corresponding safety thresholds. Commands that pass the verification are selected and output. When any simulation result exceeds the safety threshold, the command verification is deemed to have failed and the corresponding failure reason is generated.

[0015] As a further improvement to this technical solution, the virtual-real consistency verification unit includes a status data receiving module, a virtual-real data comparison module, a consistency determination module, and a verification result output module, wherein: The status data receiving module is used to acquire the virtual status data of the digital twin and the physical operation data of the construction robot, and transmit them to the virtual-real data comparison module. The virtual-real data comparison module is used to compare virtual state data and physical operation data item by item and transmit the comparison results to the consistency judgment module. The consistency determination module is used to perform virtual-real state consistency verification based on the comparison results. When the virtual-real deviation exceeds the preset threshold, a twin weight calibration instruction is generated and transmitted to the digital twin construction and mapping unit. At the same time, an operation warning instruction is generated and transmitted to the remote monitoring and virtual-real synchronization control unit, and the verification determination result is transmitted to the verification result output module. The verification result output module is used to organize and output the consistency verification results, supporting the platform's virtual and real status monitoring and anomaly identification.

[0016] Compared with the prior art, the beneficial effects of the present invention are as follows: By collecting operational data and environmental perception data of a construction robot and constructing a digital twin, an unscented Kalman filter algorithm incorporating dust and occlusion perception confidence factors is used to predict the robot's motion state. A dynamic time window is used to adaptively compensate for network latency, thereby improving the prediction stability of motion state in construction scenarios with environmental interference and network fluctuations. Based on the corrected prediction confidence, a safety threshold is dynamically adjusted. Multi-step kinematic simulation and collision detection are combined to actively verify remote control commands for safety. Verified commands are sent to the construction robot with synchronized feedback on execution status, enhancing the security of remote command issuance. By comparing virtual state data with physical operational data, twin weight calibration and control warnings are triggered when the virtual-real deviation exceeds a preset threshold, ensuring the synchronization accuracy of virtual and real actions. This invention can improve the prediction stability of the construction robot's motion state, the synchronization accuracy of virtual and real actions, and the security of remote command issuance in construction scenarios with environmental interference such as dust and occlusion, as well as network latency fluctuations. Attached Figure Description

[0017] Figure 1 This is a schematic diagram of the overall system framework of the present invention; Figure 2 This is a schematic diagram of the technical process of remote monitoring and virtual-real synchronous control in this invention; The meanings of the labels in the diagram are as follows: 1. Data Acquisition and Preprocessing Unit; 11. Data Acquisition Module; 12. Data Cleaning Module; 13. Protocol Conversion Module; 14. Data Normalization and Output Module; 2. Digital twin construction and mapping unit; 21. Initial model construction module; 22. Data-driven twin calibration module; 23. Virtual-real state synchronization mapping module; 3. Remote monitoring and virtual-real synchronous control unit; 31. Command receiving and parsing module; 32. Virtual-real status synchronization module; 33. Command virtual verification module; 34. Command issuance and status feedback module; 4. Virtual-to-real consistency verification unit; 41. Status data receiving module; 42. Virtual-to-real data comparison module; 43. Consistency judgment module; 44. Verification result output module. Detailed Implementation

[0018] 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 of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] like Figures 1-2 As shown, this embodiment provides a remote monitoring and virtual-real synchronous control platform for construction robots based on digital twins, including: Data acquisition and preprocessing unit 1: Data acquisition and preprocessing unit 1 acquires the operation data and environmental perception data of the construction robot, performs cleaning and protocol conversion, and obtains standardized operation data; In this embodiment, the data acquisition and preprocessing unit 1 includes a data acquisition module 11, a data cleaning module 12, a protocol conversion module 13, and a data normalization output module 14. The specific working process is as follows: The data acquisition module 11 is used to collect the operation data and environmental perception data of the construction robot, and transmit the collected raw data to the data cleaning module 12 in real time. The operation data includes position, posture, speed, joint angle and equipment status data, and the environmental perception data includes construction environment images, laser point clouds and obstacle distance data. Specifically, the data acquisition module 11 collects real-time data on the robot's own operation and the construction environment using various types of sensors mounted on the construction robot and in the construction scene. Among these: Operational data is collected through the robot's built-in inertial measurement unit (IMU), joint encoders, displacement sensors, and status detectors. Specifically, it includes the robot's three-dimensional position, spatial posture, movement speed, rotation angle of each joint, motor speed, load status, fault codes, and other core equipment status data. The environmental perception data is collected by the robot using high-definition industrial cameras, lidar, millimeter-wave radar, and distance sensors. Specifically, it includes environmental parameters such as real-time images of the construction environment, 3D laser point cloud data, obstacle distances, and site boundary information.

[0020] All the collected raw data are generated in real time with a millisecond sampling frequency and transmitted to the data cleaning module 12 through the robot's built-in high-speed communication interface. This comprehensively covers the robot's status and construction environment information, avoiding data loss that could lead to deviations in subsequent twin modeling and control commands.

[0021] The data cleaning module 12 is used to remove and complete outliers and missing values ​​in the original data, and then transmit the cleaned data to the protocol conversion module 13. Specifically, the 3σ principle outlier removal algorithm and the linear interpolation missing value completion algorithm are used to verify the original data frame by frame: outlier data that exceeds the normal value range is directly removed, and missing data caused by short-term communication interruption is completed by linear interpolation based on the adjacent valid data, while filtering out high-frequency noise from the sensor.

[0022] After the above cleaning process, invalid interference information is removed from the data, and the true and valid state and environmental data are retained. The data is then transmitted to the protocol conversion module 13 to ensure the accuracy of subsequent data processing.

[0023] Protocol conversion module 13 is used to convert the custom device protocol of the construction robot into a unified standard communication protocol, realize the data format unification, obtain preliminary standardized data, and transmit it to data straightening output module 14; Specifically, the protocol conversion module 13 has a built-in multi-type device protocol parsing library, which is compatible with the custom protocols of mainstream construction robots. It converts heterogeneous device private protocols into industrial standard communication protocols such as MQTT, ModbusTCP, and OPCUA, unifies the data frame structure, data type and transmission rules, and obtains preliminary standardized data.

[0024] By converting protocols, communication barriers between different devices are broken down, enabling compatible access to data from multiple brands of construction robots. After processing, the preliminary standardized data is transmitted to the data regularization output module 14.

[0025] The data normalization output module 14 is used to uniformly encode and time-series align the preliminary standardized data to obtain standardized operating data, and transmit the standardized operating data to the digital twin construction and mapping unit 2.

[0026] Specifically, to ensure the temporal consistency and efficient transmission of data within the platform, the data normalization output module 14 further normalizes the initially standardized data: On the one hand, the Protobuf lightweight encoding format is used to uniformly encode the data, reducing the data transmission volume and improving transmission efficiency; On the other hand, based on a globally unified clock, timestamp alignment and timing calibration are performed on the data collected from multiple sensors and multiple data sources to eliminate data misalignment caused by sampling time differences between different sensors, and finally generate standardized operating data with unified format, accurate timing, and standardized coding.

[0027] Finally, the standardized operational data is transmitted stably and in real time to the digital twin construction and mapping unit 2 via the platform's internal data bus.

[0028] Digital twin construction and mapping unit 2 receives standardized operation data, constructs a digital twin of the construction robot, and realizes the state mapping between the physical entity and the virtual model; In this embodiment, the digital twin construction and mapping unit 2 includes an initial model construction module 21, a data-driven twin calibration module 22, and a virtual-real state synchronization mapping module 23, wherein: The initial model building module 21 is used to build the initial virtual model of the construction robot and transmit it to the data-driven twin calibration module 22; Specifically, based on the 3D CAD engineering drawings, mechanical structure parameters, kinematic constraints, and physical properties of the construction robot, an initial virtual model containing geometric, kinematic, dynamic, and constraint models is constructed using Unity3D, Unreal Engine, or ROS-Gazebo simulation engines.

[0029] The initial virtual model fully replicates the theoretical attributes of the construction robot, such as its external structure, joint ratio, range of motion, and load characteristics, and can simulate the robot's basic motion logic. However, it does not take into account the actual operating state of the physical entity, resulting in a deviation between theoretical parameters and actual working conditions.

[0030] After the initial model building module 21 completes the initial virtual model building, it transmits the complete data, including model files, structural parameters, and kinematic DH parameters, to the data-driven twin calibration module 22, providing a basic carrier for subsequent model calibration.

[0031] The data-driven twin calibration module 22 receives standardized operational data, calibrates the parameters of the initial virtual model of the construction robot, obtains a digital twin of the construction robot, and transmits it to the virtual-real state synchronization mapping module 23. The data-driven twin calibration module 22 includes a model parameter extraction submodule, a state deviation calculation submodule, and a model parameter update submodule, wherein: The model parameter extraction submodule is used to extract the inherent parameters of the initial virtual model of the construction robot and transmit them to the state deviation calculation submodule. Specifically, it extracts the robot's inherent parameters from the initial virtual model, including immutable theoretical parameters such as geometric dimensions, joint DH parameters, moment of inertia, rated speed, initial attitude, and motion constraint threshold. At the same time, it extracts the current virtual state parameters of the model. After unifying and organizing the above parameters, they are transmitted to the state deviation calculation submodule to provide model-side reference data for deviation calculation.

[0032] The state deviation calculation submodule calculates the state deviation between the physical entity of the construction robot and the initial virtual model based on standardized operating data and transmits it to the model parameter update submodule. Specifically, it receives the standardized operating data output by the data acquisition and preprocessing unit 1, compares the real-time operating data of the physical robot (position, posture, speed, joint angles, etc.) with the theoretical state parameters of the initial virtual model dimension by dimension, calculates the position deviation using Euclidean distance, calculates the posture and joint angle deviation using angle difference, generates a comprehensive state deviation vector, accurately quantifies the degree of difference between the model and the entity, and transmits the deviation results to the model parameter update submodule.

[0033] The model parameter update submodule updates the parameters and state covariance information of the initial virtual model based on the state deviation, thus obtaining a digital twin of the construction robot. Specifically, based on the state deviation vector, an adaptive parameter correction algorithm combining least squares and unscented Kalman filtering is used to dynamically correct the variable parameters of the initial virtual model and synchronously update the model's state covariance matrix. The state covariance matrix converges synchronously as the deviation decreases, and is used to quantify the model's prediction reliability.

[0034] The update stops when the state deviation meets the preset accuracy threshold (position deviation < 5mm, angle deviation < 0.5°). At this point, the initial virtual model is corrected into a digital twin that can truly reflect the actual working conditions of the physical robot. This twin has the same motion characteristics, state response and environmental adaptability as the physical entity.

[0035] Finally, the calibrated digital twin is transmitted to the virtual-real state synchronization mapping module 23.

[0036] The virtual-real state synchronization mapping module 23 is based on the digital twin of the construction robot to realize real-time state mapping between the physical entity and the virtual model, and provides a basis for virtual simulation and state prediction for remote control.

[0037] Specifically, the virtual-real state synchronization mapping module 23 continuously receives standardized operating data and maps the real-time position, posture, joint movements, equipment status and environmental perception information of the physical robot to the digital twin at a synchronization frequency of 10ms to 50ms, driving the virtual model to make action responses that are completely consistent with the physical entity.

[0038] Meanwhile, the virtual-real state synchronization mapping module 23, based on the simulation capabilities of the digital twin, predicts and extrapolates the robot's motion state in the near future, providing virtual simulation scenarios and state prediction data for the remote monitoring and virtual-real synchronized control unit, ensuring that remote control commands can be pre-rehearsed and verified in the virtual scene.

[0039] Through real-time synchronous mapping, the digital twin can completely replicate the operating state of the physical robot and changes in the construction environment, forming a virtual-real linkage effect of "physical entity moving → virtual model moving synchronously", providing core guarantees for the safety and accuracy of remote control.

[0040] The remote monitoring and virtual-real synchronization control unit 3 receives remote control commands. Based on a digital twin, it employs an unscented Kalman filter that integrates environmental perception information and a dynamic time window adaptive virtual-real synchronization control algorithm to perform dynamic safety threshold command verification and delay compensation operations. Commands that pass verification are then sent to the construction robot, and the execution status is synchronously fed back to the digital twin. The remote monitoring and virtual-real synchronization control unit 3 includes a command receiving and parsing module 31, a virtual-real status synchronization module 32, a command virtual verification module 33, and a command sending and status feedback module 34, wherein: In this embodiment, the instruction receiving and parsing module 31 is used to receive remote control instructions, parse the remote control instructions, and transmit them to the instruction virtual verification module 33. Specifically, the instruction receiving and parsing module 31 is used to receive remote control instructions. The instruction receiving and parsing module 31 performs parsing processing such as format verification, field extraction, and data normalization on the remote control instructions. The instruction receiving and parsing module 31 transmits the parsed standardized remote control instructions to the instruction virtual verification module 33. Furthermore, the instruction receiving and parsing module 31 establishes a stable connection with the remote control terminal through dual-link redundant communication via 5G private network and industrial Ethernet. The remote control instructions received by the instruction receiving and parsing module 31 include data on the construction robot's position movement, posture adjustment, joint rotation, operation execution, and equipment start / stop control. The instruction receiving and parsing module 31 first performs frame header verification, data integrity verification, and checksum matching verification on the remote control instructions, eliminating invalid instructions that are incomplete, garbled, or have illegal formats. The instruction receiving and parsing module 31 then breaks down the valid instructions into the target position, movement range, running speed, and core control fields of the operation parameters. The instruction receiving and parsing module 31 converts non-standard remote control instructions into standardized instruction data in a unified platform encoding format. The instruction receiving and parsing module 31 stably transmits the standardized instruction data to the instruction virtual verification module 33, providing standard instruction input for the instruction virtual verification module 33 to perform multi-step instruction simulation verification and dynamic safety threshold determination.

[0041] In this embodiment, the virtual-real state synchronization module 32 is based on a digital twin and employs a confidence factor that incorporates dust and occlusion sensing. An improved unscented Kalman filter algorithm is used to predict the physical state of a construction robot, and the corrected prediction confidence level is then applied. The virtual-real state synchronization module 32 is used to dynamically adjust the synchronization time window, generate an adaptively adjusted synchronization time window, and complete the virtual-real state alignment and delay compensation. It includes an environment perception submodule, a network delay dynamic monitoring submodule, an improved unscented Kalman filter state prediction submodule, and a dynamic time window synchronization submodule, among which: The environmental perception submodule is used to acquire image data of the construction environment in real time and calculate the confidence factors for dust and occlusion perception. It is then transmitted to the improved unscented Kalman filter state prediction submodule; the specific operating logic is as follows: The environmental perception submodule receives RGB image data from a high-definition industrial camera mounted on the construction robot at a sampling frequency of 20Hz and three-dimensional point cloud data from a lidar at a sampling frequency of 10Hz. The environmental perception submodule uses the YOLOv8 semantic segmentation algorithm to perform pixel-level segmentation of the RGB image, identify dusty areas and obstacle-occluded areas in the image, and calculate the proportion of dusty areas. Percentage of the obscured area ; The environmental perception submodule uses a statistical filtering algorithm to process the laser point cloud data and calculate the point cloud efficiency. (The ratio of the number of valid point clouds to the total number of point clouds); Based on the three features mentioned above, the environmental perception submodule calculates the confidence factors for dust and occlusion perception using a weighted fusion formula. The calculation formula is: ; in, The value ranges from 0 to 1, assuming the environment is free of dust and obstructions, and the point cloud is complete. When dust completely covers the field of vision, obstacles completely block the view, and point clouds are all ineffective. ; Finally, the environmental perception submodule will calculate the confidence factors for dust and occlusion perception. The data is transmitted in real time to the improved unscented Kalman filter state prediction submodule, providing a basis for environmental interference correction for state prediction.

[0042] The network latency dynamic monitoring submodule is used to monitor network transmission latency in real time and generate latency statistics. Specifically, the network latency dynamic monitoring submodule sends heartbeat detection packets at a period of 10ms, records data transmission and reception time nodes, tracks the latency change characteristics of network transmission in real time, and transmits the generated latency statistics completely to the dynamic time window synchronization submodule, providing raw network status data support for the timing calibration of the dynamic time window synchronization submodule.

[0043] The improved unscented Kalman filter state prediction submodule is based on state data from a digital twin and confidence factors for dust and occlusion sensing. Predict the physical state of the construction robot and calculate the corrected prediction confidence level. When the corrected prediction confidence level When the value falls below a preset minimum safety threshold, the system security protection mechanism is triggered, prohibiting the issuance of new remote control commands and sending an environmental anomaly warning to the remote control terminal, and adjusting the predicted confidence level. Simultaneously transmitted to the dynamic time window synchronization submodule and the instruction virtual verification module 33; the input data of the improved unscented Kalman filter state prediction submodule includes the state data of the digital twin output by the digital twin construction and mapping unit 2, and the dust and occlusion perception confidence factors output by the environmental perception submodule. The improved output data of the unscented Kalman filter state prediction submodule includes the predicted physical state of the construction robot. The updated covariance matrix Corrected prediction confidence Furthermore, the first output terminal establishes a data transmission link with the input terminal of the dynamic time window synchronization submodule, and the second output terminal establishes a data transmission link with the input terminal of the instruction virtual verification module 33.

[0044] In the improved unscented Kalman filter state prediction submodule, the process of predicting the physical state of the construction robot and calculating the corrected prediction confidence includes the following steps: S32.1 Extracting the state vector of the construction robot corresponding to the digital twin. and covariance matrix Determine the state dimension parameters; In this step, the improved unscented Kalman filter state prediction submodule extracts the construction robot state vector corresponding to the digital twin. and covariance matrix Determine the state dimension parameters ; where, state vector , The coordinates of the robot's three-dimensional position (unit: m). The robot's Euler angle pose (unit: rad). The three-dimensional linear velocity of the robot (unit: m / s). The robot's three-dimensional angular velocity (unit: rad / s); covariance matrix. It is a 12×12 symmetric positive definite matrix used to quantify the uncertainty of each component of the state vector; S32.2 Generate a finite set of Sigma points based on the unscented Kalman filter mechanism to adapt to the nonlinear motion characteristics of the construction robot; In this step, the improved unscented Kalman filter state prediction submodule generates based on the unscented Kalman filter mechanism. Group Sigma points To adapt to the nonlinear motion characteristics of construction robots; the formula for generating Sigma points is: ; in, Let be the proportionality coefficient, and take . , Fixed parameters for the UKF algorithm; Indicates the first The Sigma points are sampling points used in unscented Kalman filtering to approximate the state probability distribution, totaling [number missing]. One, used to transmit the mean and covariance information of the state distribution; This represents the mean sampling point in the Sigma point set, i.e., the central sampling point of the state distribution, which is directly taken as the estimated value of the current state vector.

[0045] S32.3, Perform a nonlinear mapping transformation on the Sigma point based on the nonlinear kinematic model of the construction robot; specifically, based on the nonlinear kinematic model of the construction robot... A nonlinear mapping transformation is performed on the Sigma points to obtain the predicted Sigma point set. .

[0046] S32.4. Perform weighted fusion processing on the transformed Sigma points to output the predicted physical state of the construction robot. and the updated covariance matrix ; In this step, the transformed Sigma points are weighted and fused to output the predicted physical state of the construction robot. and the updated covariance matrix The weighted fusion formula is: ; In the formula: Represents the mean weight, which is the weight of the first... The weighting coefficients of each Sigma point when calculating the mean of the predicted state; Represents the covariance weight, which is the first... The weighting coefficients of each Sigma point when calculating the predicted covariance; This represents the outer product of the deviations of the Sigma points from the mean of the predicted state, used to quantify the distributional differences among the Sigma points. in, Weighted by mean, The covariance weights are calculated using the following formula: ; In the formula: This represents the mean weight of the 0th Sigma point (mean point); This represents the covariance weight of the 0th Sigma point (mean point); This represents the Sigma point index, with a value range of [value range missing]. ,correspond One Sigma point; The optimal parameters are those for the Gaussian distribution. Let be the process noise covariance matrix, and let be a diagonal matrix with diagonal elements ranging from 1 to 2. .

[0047] S32.5, Based on the updated covariance matrix The trace is used to calculate the original prediction confidence based on the unscented Kalman filtering mechanism. ; In this step, based on traces Calculate the original prediction confidence level. The calculation formula is: ; in, This is the normalization constant; The value range is 0~1. The smaller the value, the lower the prediction uncertainty. The larger.

[0048] S32.6, Introduce the dust and occlusion sensing confidence factors output by the environmental perception submodule. The original prediction confidence level is corrected to obtain the corrected prediction confidence level. .

[0049] In this step, we introduce right After weighted adjustment, the corrected prediction confidence level is obtained. The corrected formula is: .

[0050] The dynamic time window synchronization submodule is used to combine delay statistics with the corrected prediction confidence. Generate an adaptive synchronization time window The process of aligning virtual and real states and compensating for delays is completed. In the dynamic time window synchronization submodule, the process of generating an adaptive synchronization time window based on delay statistics and corrected prediction confidence, and then achieving virtual-real synchronization, includes the following steps: S32.7 Continuously collect multiple sets of network delay samples This forms a sliding delay sample set; In this step, the dynamic time window synchronization submodule continuously collects multiple sets of network latency samples based on the instruction sending timestamp and data receiving timestamp in the network latency statistics. ; Build length of the dynamic time window synchronization submodule Fixed-length sliding delay sample set Each time a new set of delayed samples is collected, it is added to the end of the sample set, while the oldest sample at the beginning is removed to maintain the real-time validity of the sample set; among which, For the first Group network transmission delay samples, in milliseconds. .

[0051] S32.8 Perform statistical operations on the delay sample set to obtain the mean delay. With delay standard deviation ; In this step, the sliding delay sample set is... Perform standard statistical operations and calculate the delayed mean using the arithmetic mean formula. The standard deviation of delay is calculated using the standard deviation formula. The specific calculation formula and process are as follows: Formula for calculating the mean delay: ; In the formula: This is the average network latency, expressed in milliseconds (ms), used to characterize the average time taken for network transmission. This is a summation operation on all delayed samples within the sample set.

[0052] Formula for calculating delayed standard deviation: ; In the formula: This represents the standard deviation of network latency, measured in milliseconds (ms), and is used to quantify the fluctuation range of network latency. A higher value indicates poorer network transmission stability; This is for square root operations; It is the square of the difference between a single delayed sample and the delay mean.

[0053] S32.9, Based on statistical characteristics, the corrected prediction confidence level and the preset minimum time window Maximum time window and nonlinear adjustment coefficient Construct an adaptive synchronization time window ; In this step, based on the mean delay Delay standard deviation Corrected prediction confidence An adaptive synchronization time window is constructed by combining system preset parameters. The specific preset parameters, calculation formulas, and calculation process are as follows: System preset parameter definitions: The minimum time window is the lower limit threshold of the synchronization time window, with a value of 50ms. The maximum time window is the upper limit threshold of the synchronization time window, with a value of 200ms. This is a nonlinear adjustment coefficient, which is the adjustment weight coefficient of network fluctuations on the synchronization window, and its value is 0.3. Adaptive synchronization time window calculation formula: ; In the formula: The time window is for adaptive synchronization, and the unit is ms. The adjustable interval length of the synchronization time window; This is a confidence level adjustment term; For network fluctuation adjustment; This represents the network latency fluctuation coefficient.

[0054] Calculation logic: Corrected prediction confidence The higher the value, the smaller the network latency fluctuation coefficient, and the better the adaptive synchronization time window. The closer to the minimum time window The real-time synchronization between virtual and real systems is stronger; the corrected prediction confidence level The lower the value, the greater the network latency fluctuation coefficient, and the more adaptive the synchronization time window. The adaptive increase in size results in higher stability of virtual-real synchronization.

[0055] S32.10, with adaptive synchronization time window Based on this, the state of the digital twin is aligned with the predicted physical state of the construction robot to complete timing calibration and delay compensation.

[0056] In this step, the dynamic time window synchronization submodule adaptively synchronizes the time window. Using the timing reference, perform virtual and real state timestamp alignment calculations to complete the timing calibration and end-to-end latency compensation between the digital twin and the physical construction robot. The specific calculation process is as follows: The dynamic time window synchronization submodule extracts the raw timestamps of the digital twin's state. Construction robots predict physical states timestamp ;in, For the digital twin's state timestamp before calibration. The physical state timestamps for the construction robot are predicted, and both are in milliseconds. Formula for calculating timestamp offset calibration: ; In the formula: The timestamp for the calibrated digital twin state, in milliseconds, is used to achieve precise alignment with the timestamp for the predicted physical state of the construction robot.

[0057] The dynamic time window synchronization submodule will synchronize the calibrated digital twin status timestamps. Construction robots predict physical states timestamp The dynamic time window synchronization submodule achieves complete alignment and end-to-end latency compensation between the digital twin and the physical construction robot. It eliminates the timing misalignment of virtual and real states caused by network transmission delay and algorithm calculation delay.

[0058] In this embodiment, the instruction virtual verification module 33 is based on the digital twin, the compensated virtual-real synchronization state, and the corrected prediction confidence. Perform multi-step instruction simulation verification and base the results on the corrected prediction confidence level. The system dynamically adjusts the safety threshold and transmits verified instructions to the instruction issuance and status feedback module 34. When instruction verification fails, it generates a verification failure receipt and the reason for failure and returns it to the remote control terminal. The instruction virtual verification module 33 includes a multi-step kinematics simulation submodule, a collision detection simulation submodule, and a dynamic safety threshold comparison submodule, wherein: The multi-step kinematics simulation submodule is based on digital twins and corrected predicted physical states. To make future remote control commands Kinematic simulation of steps The value range is 3-5; Specifically, the input data of the multi-step kinematics simulation submodule includes the digital twin output by the digital twin construction and mapping unit 2 and the predicted physical state of the construction robot output by the virtual-real state synchronization module 32. The standardized remote control commands output by the command receiving and parsing module 31; the output data of the multi-step kinematics simulation submodule is the future data corresponding to the remote control commands. Step kinematic motion simulation sequence.

[0059] Furthermore, the multi-step kinematics simulation submodule is based on a digital twin and a modified predicted physical state. To make future remote control commands Kinematic simulation of step movement, among which The value range is 3-5, and this embodiment preferably uses 5. The specific operational logic of the multi-step kinematics simulation submodule is as follows: The multi-step kinematics simulation submodule loads the kinematic constraint parameters of the digital twin, including joint range of motion, maximum motion speed, attitude deflection limit, and working radius parameters. The multi-step kinematics simulation submodule predicts the physical state of a construction robot. To simulate the initial baseline state, the standardized remote control commands are decomposed into continuous motion control quantities; The multi-step kinematics simulation submodule extrapolates frame by frame with equal time steps (50ms / step), sequentially calculating the future steps from step 1 to step 2. The robot's three-dimensional position, Euler angles, joint rotation angles, linear velocity, and angular velocity parameters are measured. The multi-step kinematics simulation submodule integrates the simulation state parameters of each step into a continuous sequence. The kinematic motion simulation sequence fully reproduces the entire motion trajectory of the remote control command execution. The multi-step kinematics simulation submodule will The kinematic motion simulation sequence is synchronously transmitted to the collision detection simulation submodule and the dynamic safety threshold comparison submodule, providing a simulation data basis for subsequent detection and comparison.

[0060] The collision detection simulation submodule is used to detect whether there is a risk of spatial collision for each simulated action in the digital twin virtual scene; Specifically, the input data for the collision detection simulation submodule is the output data from the multi-step kinematics simulation submodule. The output data of the step kinematic motion simulation sequence and the digital twin construction and mapping unit 2 is the digital twin virtual scene data; the output data of the collision detection simulation submodule is the collision risk detection result of each step of the simulation motion.

[0061] Furthermore, the specific operational logic of the collision detection simulation submodule is as follows: The collision detection simulation submodule loads static obstacle data from the digital twin virtual scene, including parameters of building structure, construction equipment, site boundaries, and pipeline facilities; The collision detection simulation submodule employs a collision detection algorithm combining AABB axis-aligned bounding boxes and OBB directed bounding boxes. Frame-by-frame detection of the kinematic motion simulation sequence; The spatial risk types detected by the collision detection simulation submodule include: rigid collisions between the robot body and scene obstacles, motion interference between robot joints and surrounding components, boundary crossing risks of the robot exceeding the construction site boundaries, and overturning risks of the robot's posture exceeding the safe range. The collision detection simulation submodule marks the detection result for each simulation action. No risk is marked as "normal", and risk is marked as the corresponding risk type. The collision detection simulation submodule transmits the frame-by-frame collision risk detection results to the dynamic safety threshold comparison submodule, which serves as one of the core bases for instruction verification.

[0062] The dynamic security threshold comparison submodule is based on the corrected prediction confidence level. Dynamically adjust the safety threshold, and the safety threshold is related to the corrected prediction confidence. A positive correlation is established, and the results of multiple simulation steps are compared with the corresponding safety thresholds. Commands that pass the verification are selected and output. When any simulation result exceeds the safety threshold, the command verification is deemed to have failed and the corresponding failure reason is generated.

[0063] Specifically, the input data for the dynamic safety threshold comparison submodule includes the output data from the multi-step kinematics simulation submodule. The simulation sequence of step kinematics, the frame-by-frame collision risk detection results output by the collision detection simulation submodule, and the corrected prediction confidence output by the virtual-real state synchronization module 32. The output data of the dynamic security threshold comparison submodule is either a verification pass instruction or an instruction verification failure receipt and the reason for the failure.

[0064] Furthermore, the specific operating logic and calculation formula of the dynamic security threshold comparison submodule are as follows: Dynamic security threshold calculation formula: Dynamic security threshold comparison submodule construction and A positively correlated dynamic security threshold model, with core security thresholds including location security thresholds. Attitude safety threshold Safe distance threshold The calculation formula is: ; In the formula: The dynamic safety threshold includes three types of safety thresholds: position, attitude, and distance. The basic safety threshold is a preset fixed value for the system, where the location basic threshold is... Attitude baseline threshold Basic threshold for safe distance ; Threshold comparison logic: The dynamic security threshold comparison submodule will The positional deviation, posture deviation, and obstacle distance parameters in the kinematic motion simulation sequence are compared one by one with the corresponding dynamic safety thresholds; at the same time, a comprehensive judgment is made in conjunction with the collision detection results. like All simulation steps showed no collision risk, and all parameters were less than or equal to the corresponding dynamic safety threshold, thus the instruction verification was successful. If any step of the simulation presents a collision risk, or if any parameter exceeds the corresponding dynamic safety threshold, the instruction verification is deemed to have failed. The failure reason generation rule dynamic security threshold comparison submodule generates standardized failure reasons based on the verification failure type, including: There is a risk of collision in the first step. Step position deviation exceeds limit, first Step attitude deviation exceeds limit, first Insufficient safe walking distance; The result output dynamic safety threshold comparison submodule transmits the verified instructions to the instruction issuance and status feedback module 34; and sends the verification failure receipt and failure reason back to the remote control terminal to remind the operator to adjust the control instructions.

[0065] In this embodiment, the instruction issuance and status feedback module 34 is used to issue verified instructions to the construction robot and simultaneously feed back the execution status of the construction robot to the digital twin.

[0066] Specifically, the input data of the instruction issuance and status feedback module 34 is the verification data output by the instruction virtual verification module 33, which verifies the remote control instruction and the real-time execution status data of the physical construction robot; the output data is the control instruction that the construction robot can recognize and the status update data of the digital twin; the first input terminal of the instruction issuance and status feedback module 34 is connected to the output terminal of the instruction virtual verification module 33, the first output terminal of the instruction issuance and status feedback module 34 is connected to the controller of the physical construction robot, and the second output terminal of the instruction issuance and status feedback module 34 is connected to the input terminal of the digital twin construction and mapping unit 2.

[0067] Furthermore, the specific operational logic of the instruction issuance and status feedback module 34 is as follows: Verification pass command issuance process: The instruction issuance and status feedback module 34 receives the verification pass instruction output by the instruction virtual verification module 33 and converts the standardized control instructions into CAN bus and Modbus TCP industrial control signals that can be recognized by the construction robot controller. The instruction issuance and status feedback module 34 sends control signals to the motion controller of the physical construction robot in real time through the 5G industrial communication link, driving the robot to perform corresponding actions, including position movement, posture adjustment, joint rotation, and task execution. The instruction issuance and status feedback module 34 records the instruction issuance timestamp and instruction execution start identifier to ensure the timing accuracy and traceability of instruction issuance.

[0068] Execution status synchronization feedback process: The instruction issuance and status feedback module 34 collects real-time instruction execution status data of the physical construction robot, including action execution progress, current position and posture, real-time joint angles, motor running status, task completion rate, and fault alarm information. The instruction issuance and status feedback module 34 standardizes and organizes the collected execution status data, removes noisy data and outliers, and ensures the authenticity and validity of the feedback data. The instruction issuance and status feedback module 34 synchronously transmits the normalized execution status data to the digital twin construction and mapping unit 2, driving the digital twin to update its status in real time, so that the digital twin and the physical construction robot are completely synchronized in action and status.

[0069] Virtual-Real Consistency Verification Unit 4 compares the virtual state data of the digital twin with the physical operation data of the construction robot and performs a consistency verification.

[0070] In this embodiment, the virtual-real consistency verification unit 4 includes a status data receiving module 41, a virtual-real data comparison module 42, a consistency determination module 43, and a verification result output module 44, wherein: The status data receiving module 41 is used to acquire the virtual status data of the digital twin and the physical operation data of the construction robot, and transmit them to the virtual-real data comparison module 42. Specifically, the status data receiving module 41 is used to synchronously acquire the virtual status data of the digital twin and the physical operation data of the construction robot. The virtual status data comes from the digital twin construction and mapping unit 2 and includes virtual simulation data such as the three-dimensional position, spatial posture, joint angle, movement speed, and equipment operation status of the digital twin. The physical operation data comes from the data acquisition and preprocessing unit 1 and includes the actual working condition data such as the three-dimensional position, spatial posture, joint angle, movement speed, and equipment operation status of the physical construction robot collected in real time.

[0071] The status data receiving module 41 performs unified format conversion and global timestamp alignment on the two data streams, removes invalid data with missing data or misaligned timing, and ensures that the virtual status data and physical operation data are dimensionally matched and time-synchronized. After processing, the normalized dual data streams are stably transmitted to the virtual-physical data comparison module 42.

[0072] The virtual-real data comparison module 42 is used to compare the virtual state data and the physical operation data item by item and transmit the comparison results to the consistency judgment module 43. Specifically, the virtual-real data comparison module 42 receives the virtual state data and physical operation data transmitted by the state data receiving module 41, and performs precise comparisons one by one according to the data dimensions. The comparison dimensions cover five core dimensions: three-dimensional position deviation, spatial posture deviation, joint angle deviation, motion speed deviation, and equipment state deviation.

[0073] Furthermore, the virtual-real data comparison module 42 uses Euclidean distance to calculate position deviation, angle difference to calculate posture and joint angle deviation, numerical difference to calculate speed deviation, and state matching degree to determine equipment state deviation, generating deviation values ​​and deviation comparison vectors dimension by dimension to fully reflect the differences between virtual state and physical operating state.

[0074] After comparison processing, the virtual and real data comparison module 42 transmits the comparison results, including the deviation values ​​of each dimension, to the consistency judgment module 43, providing a quantitative basis for subsequent consistency judgment.

[0075] The consistency determination module 43 is used to perform virtual-real state consistency verification based on the comparison results. When the virtual-real deviation exceeds the preset threshold, it generates a twin weight calibration command and transmits it to the digital twin construction and mapping unit 2. At the same time, it generates an operation warning command and transmits it to the remote monitoring and virtual-real synchronization control unit 3, and transmits the verification determination result to the verification result output module 44. Specifically, the consistency determination module 43 receives the comparison results transmitted by the virtual-real data comparison module 42, and performs a consistency verification judgment based on the platform's preset engineering-level virtual-real consistency verification thresholds. This set of thresholds is strictly set according to the general industry standards for virtual-real synchronization of digital twins of industrial robots and construction machinery, the experience of actual working conditions on construction sites, and the industry's conventional indicators of motion control accuracy of construction robots. It has a real and reliable engineering basis and industry universality. The specific thresholds and their values ​​for each dimension are as follows: Three-dimensional position consistency threshold: maximum allowable deviation ≤ 5mm; value basis - the industry-standard positioning accuracy of construction robots is millimeter level. The reasonable upper limit of error caused by dust, ground bumps, and conventional sensor interference at the construction site is 5mm. The synchronization of virtual and real positions of digital twins must meet this accuracy requirement. Exceeding this threshold will cause the virtual model and the physical entity to be misaligned, which will directly affect the safety of remote control and the accuracy of operation.

[0076] Spatial posture consistency threshold: maximum allowable deviation ≤ 0.5°; value basis - the posture control accuracy of construction robots is the industry standard range of 0.1°~0.5°. Posture deviation exceeding 0.5° can easily lead to risks such as operation trajectory deviation and body posture imbalance. The error between the digital twin virtual posture and the physical entity posture must be controlled within this range to meet the accuracy requirements of virtual-real synchronization of industrial-grade robots.

[0077] Joint angle consistency threshold: maximum allowable deviation ≤ 0.5°; the basis for the value is that the industry reasonable upper limit for encoder resolution and mechanical transmission error of multi-joint configuration of construction robots is 0.5°. If the virtual and physical joint angle deviation exceeds this value, it will cause the virtual joint movement to be mismatched with the physical joint movement, affecting the motion simulation and the execution effect of control commands.

[0078] Motion linear velocity consistency threshold: maximum allowable deviation ≤ 0.05m / s; basis for value selection - construction robots on construction sites are mostly in low-speed safe operation conditions. The allowable error for speed control accuracy in the industry is 0.05m / s. Excessive speed deviation will cause the virtual motion and physical motion to be out of sync, increasing the safety hazards of remote control.

[0079] Equipment operating status consistency judgment rules: The virtual status and the physical status must be completely matched; the basis for value selection is that the operating status of the equipment, such as start-up, shutdown, fault, standby, operation, etc., is discrete Boolean data with no intermediate transitional status. It must be completely consistent to determine the status is valid. This is the basic industry judgment criterion for the synchronization of digital twins and physical equipment.

[0080] The consistency determination module 43 compares the deviation values ​​of each dimension output by the virtual-real data comparison module 42 with the above-mentioned preset thresholds one by one: if all the deviation values ​​of each dimension are less than or equal to the corresponding preset thresholds, it is determined that the virtual twin and the construction robot are consistent in virtual and real states; if any one-dimensional or multi-dimensional deviation value exceeds the corresponding preset threshold, it is determined that the virtual and real states are inconsistent.

[0081] Furthermore, when the virtual-real deviation exceeds a preset threshold, the consistency judgment module 43 automatically generates a twin weight calibration instruction and transmits it to the digital twin construction and mapping unit 2, driving the digital twin to recalibrate parameters and correct its state based on the real-time running data of the physical robot; at the same time, it generates an operation warning instruction and transmits it to the remote monitoring and virtual-real synchronization control unit 3, triggering the platform's safety control mechanism, reminding operators and the system to control abnormal virtual-real synchronization states, and ensuring the safety of the remote control process.

[0082] Finally, the consistency determination module 43 synchronously transmits the final verification determination result, deviation information of each dimension, and instruction issuance record to the verification result output module 44.

[0083] The verification result output module 44 is used to organize and output the consistency verification results, supporting the platform's virtual and real status monitoring and anomaly identification. Specifically, the verification result output module 44 receives the verification judgment results transmitted by the consistency judgment module 43, integrates and organizes the consistency judgment conclusion, deviation values ​​of each dimension, anomaly alarm information, recalibration instruction records, and control early warning instruction records, and generates a standardized virtual and real consistency verification report.

[0084] Finally, the verification result output module 44 outputs the verification report to the platform monitoring terminal and data management unit in real time, intuitively displaying the virtual-real matching status of the digital twin and the construction robot, quickly locating deviation and abnormal points, supporting platform operation and maintenance personnel to monitor the virtual-real synchronization status in real time, handle abnormal problems in a timely manner, and ensure the long-term stable operation of the platform.

[0085] Those skilled in the art will understand that the process of implementing all or part of the steps of the above embodiments can be carried out by hardware or by a program instructing the relevant hardware.

[0086] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the claimed invention.

Claims

1. A remote monitoring and virtual-real synchronous control platform for construction robots based on digital twins, characterized in that, include: The data acquisition and preprocessing unit (1) collects the operation data and environmental perception data of the construction robot, performs cleaning and protocol conversion, and obtains standardized operation data. The digital twin construction and mapping unit (2) receives standardized operation data, constructs a digital twin of the construction robot, and realizes the state mapping between the physical entity and the virtual model; The remote monitoring and virtual-real synchronization control unit (3) receives remote control commands, and based on the digital twin, uses an unscented Kalman filter and a dynamic time window adaptive virtual-real synchronization control algorithm that integrates environmental perception information to perform dynamic safety threshold command verification and delay compensation operations. It then sends verified commands to the construction robot and synchronously feeds back the execution status to the digital twin. The remote monitoring and virtual-real synchronization control unit (3) includes a virtual-real status synchronization module (32) and a command virtual verification module (33), wherein: The virtual-real state synchronization module (32) is based on a digital twin and adopts a confidence factor that incorporates dust and occlusion perception. An improved unscented Kalman filter algorithm is used to predict the physical state of a construction robot, and the corrected prediction confidence level is then applied. The synchronization time window is used to dynamically adjust the synchronization time window, generate an adaptively adjusted synchronization time window, and complete the virtual-real state alignment and delay compensation; the virtual-real state synchronization module (32) includes an environment perception submodule, a network delay dynamic monitoring submodule, an improved unscented Kalman filter state prediction submodule, and a dynamic time window synchronization submodule, wherein: The environmental perception submodule is used to acquire image data of the construction environment in real time and calculate the confidence factors for dust and occlusion perception. And transmit it to the improved unscented Kalman filter state prediction submodule; The network latency dynamic monitoring submodule is used to monitor network transmission latency in real time and generate latency statistics. The improved unscented Kalman filter state prediction submodule is based on state data from a digital twin and confidence factors for dust and occlusion perception. Predict the physical state of the construction robot and calculate the corrected prediction confidence level. The corrected prediction confidence level Simultaneously transmitted to the dynamic time window synchronization submodule and the instruction virtual verification module (33); The dynamic time window synchronization submodule is used to combine delay statistics with the corrected prediction confidence. Generate an adaptive synchronization time window Complete the alignment of virtual and real states and delay compensation; The instruction virtual verification module (33) is based on the digital twin, the compensated virtual-real synchronization state, and the corrected prediction confidence. Perform multi-step instruction simulation verification and base the results on the corrected prediction confidence level. The security threshold is dynamically adjusted, and the instruction that passes the verification is transmitted to the instruction issuance and status feedback module (34); when the instruction verification fails, a verification failure receipt and the reason for failure are generated and returned to the remote control terminal. Virtual-real consistency verification unit (4) compares the virtual state data of the digital twin with the physical operation data of the construction robot and performs consistency verification.

2. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 1, characterized in that, The data acquisition and preprocessing unit (1) includes a data acquisition module (11), a data cleaning module (12), a protocol conversion module (13), and a data normalization output module (14), wherein: The data acquisition module (11) is used to collect the operation data and environmental perception data of the construction robot and transmit the collected raw data to the data cleaning module (12) in real time. The operation data includes position, posture, speed, joint angle and equipment status data, and the environmental perception data includes construction environment images, laser point clouds and obstacle distance data. The data cleaning module (12) is used to remove and complete outliers and missing values ​​in the original data, and to obtain the cleaned data and transmit it to the protocol conversion module (13). The protocol conversion module (13) is used to convert the custom device protocol of the construction robot into a unified standard communication protocol, realize the data format unification, obtain preliminary standardized data and transmit it to the data regularization output module (14). The data normalization output module (14) is used to uniformly encode and time-series align the preliminary standardized data to obtain standardized operating data, and transmit the standardized operating data to the digital twin construction and mapping unit (2).

3. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 1, characterized in that, The digital twin construction and mapping unit (2) includes an initial model construction module (21), a data-driven twin calibration module (22), and a virtual-real state synchronization mapping module (23), wherein: The initial model building module (21) is used to build the initial virtual model of the construction robot and transmit it to the data-driven twin calibration module (22). The data-driven twin calibration module (22) is used to receive standardized operating data, perform parameter calibration on the initial virtual model of the construction robot, obtain the digital twin of the construction robot, and transmit it to the virtual-real state synchronization mapping module (23). The virtual-real state synchronization mapping module (23) is based on the digital twin of the construction robot to realize the real-time state mapping between the physical entity and the virtual model, and provides a basis for virtual simulation and state prediction for remote control.

4. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 3, characterized in that, The data-driven twin calibration module (22) includes a model parameter extraction submodule, a state deviation calculation submodule, and a model parameter update submodule, wherein: The model parameter extraction submodule is used to extract the inherent parameters of the initial virtual model of the construction robot and transmit them to the state deviation calculation submodule; The state deviation calculation submodule calculates the state deviation between the physical entity of the construction robot and the initial virtual model based on standardized operating data and transmits it to the model parameter update submodule. The model parameter update submodule updates the parameters and state covariance information of the initial virtual model based on the state deviation to obtain a digital twin of the construction robot.

5. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 1, characterized in that, The remote monitoring and virtual-real synchronization control unit (3) further includes an instruction receiving and parsing module (31) and an instruction issuing and status feedback module (34), wherein: The instruction receiving and parsing module (31) is used to receive remote control instructions, parse the remote control instructions and transmit them to the instruction virtual verification module (33). The instruction issuance and status feedback module (34) is used to issue verified instructions to the construction robot and simultaneously feed back the execution status of the construction robot to the digital twin.

6. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 1, characterized in that, In the improved unscented Kalman filter state prediction submodule, the process of predicting the physical state of the construction robot and calculating the corrected prediction confidence includes the following steps: S32.1 Extracting the state vector of the construction robot corresponding to the digital twin. and covariance matrix Determine the state dimension parameters; S32.2 Generate a finite set of Sigma points based on the unscented Kalman filter mechanism to adapt to the nonlinear motion characteristics of the construction robot; S32.

3. Perform nonlinear mapping transformation on the Sigma point based on the nonlinear kinematic model of the construction robot; S32.

4. Perform weighted fusion processing on the transformed Sigma points to output the predicted physical state of the construction robot. and the updated covariance matrix ; S32.5, Based on the updated covariance matrix The trace is used to calculate the original prediction confidence based on the unscented Kalman filtering mechanism. ; S32.6, Introduce the dust and occlusion sensing confidence factors output by the environmental perception submodule. The original prediction confidence level is corrected to obtain the corrected prediction confidence level. .

7. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 1, characterized in that, In the dynamic time window synchronization submodule, the process of generating an adaptive synchronization time window based on delay statistics and corrected prediction confidence, and achieving virtual-real synchronization, includes the following steps: S32.7 Continuously collect multiple sets of network delay samples This forms a sliding delay sample set; S32.8 Perform statistical operations on the delay sample set to obtain the mean delay. With delay standard deviation ; S32.9, Based on statistical characteristics, the corrected prediction confidence level and the preset minimum time window Maximum time window and nonlinear adjustment coefficient Construct an adaptive synchronization time window ; S32.10, with adaptive synchronization time window Based on this, the state of the digital twin is aligned with the predicted physical state of the construction robot to complete timing calibration and delay compensation.

8. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 1, characterized in that, The instruction virtual verification module (33) includes a multi-step kinematics simulation submodule, a collision detection simulation submodule, and a dynamic safety threshold comparison submodule, wherein: The multi-step kinematics simulation submodule is based on a digital twin and a corrected predicted physical state. To make future remote control commands Kinematic simulation of steps The value range is 3-5; The collision detection simulation submodule is used to detect whether there is a risk of spatial collision in each simulated action within the digital twin virtual scene; The dynamic security threshold comparison submodule is based on the corrected prediction confidence level. Dynamically adjust the safety threshold, and the safety threshold is related to the corrected prediction confidence. A positive correlation is established, and the results of multiple simulation steps are compared with the corresponding safety thresholds. Commands that pass the verification are selected and output. When any simulation result exceeds the safety threshold, the command verification is deemed to have failed and the corresponding failure reason is generated.

9. The construction robot remote monitoring and virtual-real synchronous control platform based on digital twin as described in claim 1, characterized in that, The virtual-real consistency verification unit (4) includes a status data receiving module (41), a virtual-real data comparison module (42), a consistency determination module (43), and a verification result output module (44), wherein: The status data receiving module (41) is used to acquire the virtual status data of the digital twin and the physical operation data of the construction robot, and transmit them to the virtual-real data comparison module (42). The virtual-real data comparison module (42) is used to compare the virtual state data and the physical operation data item by item and transmit the comparison results to the consistency judgment module (43). The consistency determination module (43) is used to perform virtual-real state consistency verification based on the comparison results. When the virtual-real deviation exceeds the preset threshold, a twin weight calibration instruction is generated and transmitted to the digital twin construction and mapping unit (2). At the same time, an operation warning instruction is generated and transmitted to the remote monitoring and virtual-real synchronization control unit (3). The verification determination result is transmitted to the verification result output module (44). The verification result output module (44) is used to organize and output the consistency verification result, supporting the platform's virtual and real status monitoring and anomaly identification.

Images