CN119992019A - A method for constructing a virtual cultural tourism platform based on digital twins - Google Patents
A method for constructing a virtual cultural tourism platform based on digital twins Download PDFInfo
- Publication number
- CN119992019A CN119992019A CN202510472264.5A CN202510472264A CN119992019A CN 119992019 A CN119992019 A CN 119992019A CN 202510472264 A CN202510472264 A CN 202510472264A CN 119992019 A CN119992019 A CN 119992019A
- Authority
- CN
- China
- Prior art keywords
- virtual
- real
- federation
- parameters
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Landscapes
- Processing Or Creating Images (AREA)
Abstract
The invention discloses a virtual travel platform construction method based on digital twinning, which relates to the technical field of digital twinning and cultural travel and comprises the steps of dynamically distributing federal model precision weights through a federal learning framework based on a federal learning initialization instruction set, calibrating a real-time video stream with a three-dimensional model by utilizing a video twinning engine to generate a virtual-real scene bidirectional mapping protocol containing precision grading parameters and NPC control instructions, and generating a closed-loop interaction data set containing NPC behavior instructions and physical equipment control signals by detecting and scanning a physical space three-dimensional geometry through rays based on the virtual-real scene bidirectional mapping protocol and executing a mixed reality collision monitoring algorithm. According to the invention, the federal model precision weight is dynamically allocated, the real-time video stream frame rate fluctuation is combined to trigger the target detection algorithm to perform semantic segmentation on the key region, and the geometric alignment error value is calculated to divide the precision level, so that the virtual environment is in seamless butt joint with the real world.
Description
Technical Field
The invention relates to the technical field of digital twinning and cultural tourism, in particular to a method for constructing a virtual cultural tourism platform based on digital twinning.
Background
Traditional travel experience mainly depends on direct interaction between physical space of physical scenic spots and tourists, and the mode has remarkable limitations in information transmission efficiency, personalized service provision, real-time feedback mechanism and the like. In recent years, digital twin technology gradually reaches the brand-new angle, and high-precision simulation and dynamic update of an actual scene are realized by creating a digital mirror image of a physical environment. However, in existing travel applications, digital twinning techniques are mostly limited to static presentations or limited interactive functions, failing to fully exploit their potential to enhance guest experiences.
Currently, one of the major challenges facing the travel industry is how to achieve efficient data sharing and analysis to support personalized guest services while guaranteeing user privacy. Existing travel platforms often fail to efficiently integrate data from different sources, resulting in a lack of pertinence and flexibility in the services provided. Furthermore, while some advanced Virtual Reality (VR) and Augmented Reality (AR) technologies have been applied to enhance guest experiences, the application of these technologies is generally independent of actual operational management and decision support, limiting their potential in optimizing resource configurations, improving management efficiency.
Disclosure of Invention
The present invention has been made in view of the above-described problems occurring in the prior art.
Therefore, the invention provides a virtual travel platform construction method based on digital twinning, which solves the problem.
In order to solve the technical problems, the invention provides the following technical scheme:
the invention provides a virtual travel platform construction method based on digital twinning, which comprises the steps of collecting a real-time data stream of a travel domain, initializing a federation learning frame at an edge point to generate a federation learning initialization instruction set, dynamically distributing federation model precision weights through the federation learning initialization instruction set based on the federation learning initialization instruction set, calibrating a real-time video stream and a three-dimensional model through a video twinning engine to generate a virtual-real scene bidirectional mapping protocol containing precision grading parameters and NPC control instructions, scanning a three-dimensional geometric structure of a physical space through ray detection based on the virtual-real scene bidirectional mapping protocol and executing a mixed reality collision monitoring algorithm to generate a closed-loop interaction data set containing NPC behavior instructions and physical equipment control signals, executing federation model training at each edge node based on the closed-loop interaction data set, calculating federation model gradients and uploading the federation model gradients to a federation main node, updating global federation model parameters after the main node aggregates gradients, triggering dynamic navigation and risk early warning through the global federation model parameters, and synchronizing the optimized virtual-real scene bidirectional mapping protocol to a multi-scenic-region digital living body through the meta-universe protocol, and constructing a virtual travel platform of a cross-scene.
As an optimal scheme of the virtual travel platform construction method based on digital twinning, the travel universe real-time data stream comprises tourist behavior thermodynamic distribution data, environment parameters, cultural relic state data and real-time video stream data.
The method for constructing the virtual travel platform based on digital twinning is an optimal scheme, wherein the federation learning framework is initialized at the edge point to generate a federation learning initialization instruction set, the specific steps are as follows,
Initializing a federation learning framework at an edge node, deploying a lightweight federation learning client, loading a text travel global real-time data stream, and outputting a local training data set with aligned dimensions;
Based on the local training data set, dynamically generating a federal learning initialization instruction set containing federal model architecture templates, aggregation weight rules and communication protocol parameters through multi-modal learning.
The method for constructing the virtual travel platform based on digital twin is a preferable scheme, wherein the method is based on a federal learning initialization instruction set, dynamically distributes federal model precision weights through a federal learning framework, simultaneously calibrates a real-time video stream with a three-dimensional model by utilizing a video twin engine to generate a virtual-real scene bidirectional mapping protocol comprising precision grading parameters and NPC control instructions, and comprises the following specific steps of,
Extracting equipment performance parameters and data reliability indexes of the edge nodes based on the federation learning initialization instruction set, and dynamically distributing federation model precision weights;
when the video twin engine detects a real-time video stream frame rate fluctuation hyperstability threshold, triggering a target detection algorithm to perform semantic segmentation on a key region, and extracting the boundary frame coordinates of a moving object;
calculating a geometric alignment error value and dividing a precision level based on the boundary frame coordinates to generate precision grading parameters comprising grid density and algorithm selection rules;
Based on the precision grading parameters, evaluating the geometric alignment error value of the current scene and dynamically switching the NPC behavior engine containing logic complexity by using the federal model precision weight through scene consistency analysis;
And packaging the federal model precision weight, the geometric alignment error value and the NPC behavior engine into a structured data packet, synchronizing the structured data packet to an edge node and a cloud host node through a space-time unified interface, and generating a virtual-real scene bidirectional mapping protocol containing precision grading parameters and NPC control instructions.
The virtual travel platform construction method based on digital twinning is an optimal scheme, wherein the virtual and real scene bidirectional mapping protocol is used for scanning a physical space three-dimensional geometric structure through ray detection and executing a mixed reality collision monitoring algorithm to generate a closed loop interaction data set containing NPC behavior instructions and physical equipment control signals, and the method comprises the following specific steps of,
Based on a virtual-real scene bidirectional mapping protocol, scanning a physical space three-dimensional geometric structure through ray detection, and extracting surface point cloud data of a physical space object;
Aligning the three-dimensional model coordinate system based on the surface point cloud data, and executing a mixed reality collision detection algorithm to calculate the collision probability;
dynamically adjusting the path planning of the virtual NPC according to the collision probability, and generating an instruction set containing obstacle avoidance priority and moving speed;
And converting the instruction set of the obstacle avoidance priority and the moving speed into a driving control signal of the physical equipment, and packaging the driving control signal into a closed-loop interaction data set containing the NPC behavior instruction and the physical equipment control signal.
As an optimal scheme of the virtual travel platform construction method based on digital twin, the method is based on a closed-loop interaction data set, federal model training is executed at each edge node, and the method comprises the following specific steps,
Extracting NPC behavior instructions, equipment control signals and pose verification data by analyzing a closed-loop interaction data set, and generating a standardized training sample;
Performing federal model training based on the standardized training samples, and adding dynamic differential noise to the trained federal model;
Homomorphic encryption is carried out on the federation model added with noise, an encryption parameter packet is generated and uploaded to a cloud aggregation node, and the aggregation node performs weighted average on parameters of all nodes and generates global federation model parameters.
Homomorphic encryption is carried out on the federation model added with noise, an encryption parameter packet is generated and uploaded to a cloud aggregation node, and the aggregation node performs weighted average on parameters of all nodes and generates global federation model parameters.
The method for constructing the virtual travel platform based on digital twin is used as an optimal scheme, wherein the method calculates the gradient of the federation model and uploads the gradient to a federation main node, the main node aggregates the gradient and then updates the global federation model parameters, the specific steps are as follows,
Adding Laplace noise to the gradient calculated by the edge nodes to generate privacy protection gradients, and encrypting the gradients of the edge nodes by using Paillier;
and carrying out weighted summation based on the encryption gradient of each edge node, and updating the global federation model parameters through natural gradient descent.
The method for constructing the virtual travel platform based on the digital twin is a preferable scheme, wherein the method uses global federal model parameters to trigger dynamic navigation and risk early warning, synchronizes the optimized virtual-real scene bidirectional mapping protocol to the multi-scenic spot digital twin body through a meta-universe protocol, constructs the cross-scene virtual travel platform, and comprises the following specific steps,
Analyzing the tourist position, the scenic spot capacity and the environmental data in real time through the global federal model parameters, dynamically generating a guiding path, and calculating dynamic association parameters of the tourist distribution and the capacity risk of the scenic spot in real time;
Encoding the navigation path and the dynamic association parameters into bidirectional mapping parameters through a meta universe protocol, and synchronizing the bidirectional mapping parameters to the digital twin nodes of the scenic spots based on block chain consensus;
based on the synchronized digital twin nodes, a virtual travel platform with virtual-real interaction across scenes is constructed.
In a second aspect, the invention provides a computer device comprising a memory and a processor, the memory storing a computer program, wherein the computer program when executed by the processor implements any step of the method for constructing a digital twin based virtual travel platform according to the first aspect of the invention.
In a third aspect, the present invention provides a computer readable storage medium having a computer program stored thereon, wherein the computer program when executed by a processor implements any step of the method for constructing a digital twinning-based virtual travel platform according to the first aspect of the present invention.
The method has the advantages that the federal model precision weight is dynamically distributed, the real-time video stream frame rate fluctuation is combined to trigger the target detection algorithm to perform semantic segmentation on the key region, the geometric alignment error value is calculated to divide the precision level, the virtual environment is seamlessly butted with the real world, meanwhile, the NPC behavior logic complexity is dynamically switched based on the precision grading parameter, the reality and the interactivity of the virtual character are improved, and the calculation cost is reduced by optimizing the resource use.
Drawings
In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings required for the description of the embodiments will be briefly described below, it being obvious that the drawings in the following description are only some embodiments of the present invention, and that other drawings may be obtained according to these drawings without inventive effort for a person skilled in the art.
Fig. 1 is a flowchart of a virtual travel platform construction method based on digital twin in embodiment 1.
Fig. 2 is a flow chart of federal model weight distribution and virtual-real scene mapping of the virtual travel platform construction method based on digital twin in embodiment 1
Fig. 3 is a flowchart of collision detection and closed-loop interaction data set generation of the virtual travel platform construction method based on digital twinning in embodiment 1.
Fig. 4 is a flow chart of federation model training and global federation model parameter updating based on the digital twin-based virtual travel platform construction method in example 1.
Detailed Description
In order that the above-recited objects, features and advantages of the present invention will become more readily apparent, a more particular description of the invention will be rendered by reference to specific embodiments thereof which are illustrated in the appended drawings.
In the following description, numerous specific details are set forth in order to provide a thorough understanding of the present invention, but the present invention may be practiced in other ways other than those described herein, and persons skilled in the art will readily appreciate that the present invention is not limited to the specific embodiments disclosed below.
Further, reference herein to "one embodiment" or "an embodiment" means that a particular feature, structure, or characteristic can be included in at least one implementation of the invention. The appearances of the phrase "in one embodiment" in various places in the specification are not necessarily all referring to the same embodiment, nor are separate or alternative embodiments mutually exclusive of other embodiments.
Embodiment 1, referring to fig. 1 to fig. 4, the embodiment provides a virtual travel platform construction method based on digital twinning, which comprises the following steps:
s1, acquiring a text traveling domain real-time data stream, initializing a federation learning framework at an edge point, and generating a federation learning initialization instruction set.
S1.1, the text traveling domain real-time data stream comprises tourist behavior thermodynamic distribution data, environment parameters, cultural relic state data and real-time video stream data.
It should be noted that the tourist behavior thermodynamic distribution data can accurately capture the aggregation situation and the flow trend of tourists in different areas, provide basis for optimizing scenic spot layout and service facility configuration, the environmental parameters comprise information such as temperature, humidity, air quality and the like, are helpful for evaluating and improving the scenic spot environmental quality, the cultural heritage and the security monitoring and protection state of exhibits are related to support timely measures to prevent potential risks, and the real-time video stream data provides dynamic visual information, so that the system can be used for monitoring various activities occurring in the scenic spot.
S1.2, initializing a federation learning framework at an edge node, deploying a lightweight federation learning client, loading a text traveling domain real-time data stream, and outputting a local training data set with aligned dimensions.
It should be noted that after the edge node initializes the federation learning framework and deploys the lightweight federation learning client, by loading the global real-time data stream of the text, the global real-time data stream of the text is first cleaned and standardized to eliminate noise, fill in missing values and unify data formats, so as to ensure that data from different sources have consistent structure and semantic expression, and the preset feature engineering rule is a standard set based on historical data analysis and task requirements, and is used for guiding how to clean, standardize and select key features of multidimensional heterogeneous data, so as to ensure consistency and effectiveness of the data. The preset characteristic engineering rules are determined by analyzing the past data patterns and specific application targets so as to adapt to the characteristics and requirements of different edge nodes. Based on preset characteristic engineering rules, multidimensional heterogeneous data (such as guest behavior thermodynamic distribution data, environment parameters, cultural relic state data and real-time video stream data) are mapped into a unified characteristic space, and the most relevant characteristics of targets are screened out from the multidimensional heterogeneous data according to task requirements so as to ensure the effectiveness and pertinence of a data set and form a local training data set with consistent dimensions, in the process, the characteristic weights and the dimension distribution are dynamically adjusted so as to adapt to the computing capacity and the data characteristics of different edge nodes, and finally the output local training data set not only realizes dimension alignment, but also maintains the diversity and representativeness of the data, thereby laying a solid foundation for efficient training and global parameter aggregation of a subsequent federal model.
S1.3, dynamically generating a federal learning initialization instruction set containing federal model architecture templates, aggregation weight rules and communication protocol parameters through multi-modal learning based on the local training data set.
Furthermore, firstly, the characteristics of multi-source heterogeneous data in the local training data set are utilized, and the multi-mode learning method is adopted to extract key information capable of reflecting essential characteristics of the data, so that the data of different sources can be effectively processed under a unified frame. And then, constructing a federal model architecture template suitable for the current application scene according to the extracted key characteristics and task requirements, wherein the federal model architecture template defines the basic structure and algorithm logic of cooperation among all edge nodes in the federal learning process. Meanwhile, an aggregation weight rule is determined, corresponding weights are distributed according to factors such as data quantity, quality and computing capacity of each node, and contribution degree of each node can be fully considered when federal model aggregation is ensured. Finally, setting communication protocol parameters, defining details such as data transmission format, frequency and security mechanism among nodes, so as to ensure efficient and safe information exchange.
And S2, dynamically distributing the precision weight of the federal model through a federal learning framework based on the federal learning initialization instruction set, and simultaneously calibrating the real-time video stream with the three-dimensional model by utilizing a video twin engine to generate a virtual-real scene bidirectional mapping protocol containing precision grading parameters and NPC control instructions.
And S2.1, extracting performance parameters and data reliability indexes of the edge nodes based on the federation learning initialization instruction set, and dynamically distributing federation model precision weights.
It should be noted that, first, according to the requirements of the federal learning initialization instruction set, the device performance parameters including the processing speed, the memory capacity, the network bandwidth and other information are obtained from each edge node, and meanwhile, the data reliability indexes, such as the data integrity check result and the historical error rate, are collected.
Further, based on the obtained processing speed, memory capacity, network bandwidth, and data integrity check results and historical error rates. And setting scoring standards according to the historical data and task demands, and evaluating the performance and the data reliability of each node. The nodes with high performance and high reliability are evaluated more. The evaluation result is the comprehensive score of each node, reflects the potential contribution degree of each node, and is used for dynamically distributing the federal model precision weight and optimizing the federal model precision. According to the evaluation result, dynamically distributing federal model precision weights to each participating edge node, and ensuring that nodes with higher equipment performance and more reliable data obtain larger weights, thereby playing a more important role in the joint training process. The process ensures the effective utilization of resources and the maximization of model accuracy in the federal learning process, so that the final federal learning model can fully benefit from the advantages of each node. By the method, accurate consideration and reasonable utilization of edge node characteristics are realized, and overall learning efficiency and accuracy are improved.
And S2.2, when the video twin engine detects that the real-time video stream frame rate fluctuation exceeds a threshold value, triggering a target detection algorithm to carry out semantic segmentation on the key region, and extracting the boundary frame coordinates of the moving object.
The method comprises the steps that when a video twin engine detects that the real-time video stream frame rate fluctuation exceeds a preset threshold, a target detection algorithm is triggered to conduct semantic segmentation on a key region, and boundary frame coordinates of a moving object are extracted. The threshold here is a reference value set based on the frame rate fluctuation condition of the history video stream data for judging whether or not an abnormal fluctuation occurs in the current frame rate.
Specifically, the threshold is typically set to 10% above and below the average frame rate, i.e., if the frame rate of the real-time video stream fluctuates beyond this range (e.g., the average frame rate is 30 frames/second, the threshold range is 27 to 33 frames/second), further object detection and semantic segmentation operations are triggered to identify and mark semantic information within critical areas, such as guests, cultural relics, or dynamically changing portions of the environment, to achieve precise differentiation and localization of different objects in a complex scene. This process ensures that important information can be captured and processed in time when the frame rate abnormally fluctuates.
S2.3, calculating geometrical alignment error values and dividing precision levels based on boundary frame coordinates to generate precision grading parameters comprising grid density and algorithm selection rules, wherein the expression is as follows:
;
Wherein, Representing the value of the geometric alignment error,Representing the actual bounding box coordinates,Representing the coordinates of the mapped bounding box,Representing the reference dimensions of the federal model,Representing the dynamic adjustment coefficient of the dynamic adjustment,Representing the federal maximum weight value,Representing federal learning weights for the current node.
The method comprises the steps of calculating a geometric alignment error value based on the difference between the actual coordinates of a boundary frame and digital twin mapping coordinates, comparing the coordinate dimension offset of a target detection frame, carrying out normalization processing to obtain a comprehensive error evaluation result reflecting data mapping accuracy, dynamically dividing an accuracy hierarchy according to the comprehensive error evaluation result, adopting a fine-granularity grid and a high-accuracy deep learning registration algorithm when the error is low, simultaneously endowing nodes with higher federation learning weight to strengthen contribution to a global federation model, switching to a balance mode when the error is medium and medium, using a medium-grid density and lightweight ICP registration algorithm and distributing moderate weight, and starting a sparse grid and quick feature matching method and reducing weight to reduce noise influence when the error is high, and finally generating accuracy classification parameters comprising grid density configuration, algorithm selection rules and federation weight values to realize error self-adaptive resource allocation and federation collaborative optimization.
And S2.4, based on the precision grading parameters, evaluating the geometric alignment error value of the current scene and dynamically switching the NPC behavior engine containing logic complexity by using the federal model precision weight through scene consistency analysis.
The specific process includes that the precision grading parameters are utilized to carry out fine analysis on the current scene, deviation of actual objects and expected positions is analyzed, and geometric alignment error values are determined to reflect the scene consistency level. And combining the federal model precision weight, adjusting the contribution proportion of the edge nodes, and ensuring that the high-precision nodes occupy larger proportion in model training. And then, according to the scene consistency and the change of the error value, dynamically switching the NPC behavior engine mode, optimizing the logic processing mechanism of the NPC behavior engine mode, enabling the NPC behavior to be more natural and accord with the actual environment requirements, and adapting to the requirements of scenes with different complexity. And then, according to the calculated geometric alignment error value, combining the federal model precision weight, adjusting the contribution proportion of different edge nodes in federal learning, and ensuring that the high-precision nodes play a larger role in model training. And then, dynamically switching the working mode of the NPC behavior engine according to the scene consistency and the change condition of the error value so as to adapt to different logic complexity requirements. In this process, the NPC behavior engine adjusts the internal logic processing mechanism to optimize the NPC behavior according to the specific requirements of the current scenario.
Furthermore, by the precision grading parameter and scene fine analysis-based method, the geometric alignment error value is accurately evaluated, the federal model precision weight is effectively and dynamically adjusted, and meanwhile, the NPC behavior engine is ensured to be capable of flexibly coping with various complex scene changes.
And S2.5, packaging the federal model precision weight, the geometric alignment error value and the NPC behavior engine into a structured data packet, synchronizing the structured data packet to an edge node and a cloud host node through a space-time unified interface, and generating a virtual-real scene bidirectional mapping protocol containing precision grading parameters and NPC control instructions.
Further, federal model accuracy weights, geometric alignment error values, and NPC behavior engines are packaged into structured data packets. This process involves integrating the contribution ratio of each edge node in federal learning (i.e., federal model accuracy weight), the deviation between the actual object and the expected location in the current scene (i.e., geometric alignment error value), and the NPC behavior logic (i.e., NPC behavior engine) adjusted based on this information into a unified data structure. This is done to ensure that all relevant information can be transmitted and processed in a consistent and standardized manner. Next, synchronizing to the edge node and the cloud host node through the space-time unified interface. This means that the above structured data packets are sent to the respective edge nodes and cloud host nodes using a predefined communication protocol (instant air unified interface). The spatio-temporal unified interface ensures that all nodes receive the same information, both in the temporal and spatial dimensions, and can make corresponding adjustments or responses based on the information. Based on the above, generating virtual-real scene bidirectional mapping protocol containing precision grading parameter and NPC control instruction. Specifically, a rule set (i.e., virtual-to-real scene bi-directional mapping protocol) is formulated for guiding interactions between a virtual scene and the real world based on information collected from nodes (e.g., federal model accuracy weights, geometric alignment error values, etc.). The virtual-real scene bidirectional mapping protocol not only comprises how to adjust the data processing mode (namely precision grading parameters) according to different precision requirements, but also comprises a specific operation guide (namely NPC control instructions) aiming at the NPC so as to optimize the behavior of the NPC under different scenes.
The method comprises the steps of integrating key parameters, guaranteeing comprehensiveness and consistency of information, achieving efficient data synchronization through a standard interface, and finally generating a detailed interaction protocol based on the collected information, so that interaction of virtual world and real world is more accurate and effective. The series of steps work together to ensure that high-level data accuracy and NPC behavior rationality can be maintained in complex and changeable environments.
And S3, based on a virtual-real scene bidirectional mapping protocol, scanning a physical space three-dimensional geometric structure through ray detection and executing a mixed reality collision monitoring algorithm to generate a closed-loop interaction data set containing NPC behavior instructions and physical equipment control signals.
S3.1, scanning a physical space three-dimensional geometric structure through ray detection based on a virtual-real scene bidirectional mapping protocol, and extracting surface point cloud data of a physical space object.
The specific process comprises the steps that based on a virtual-real scene bidirectional mapping protocol, the process of scanning a physical space three-dimensional geometric structure through ray detection firstly depends on precision grading parameters in the protocol and the alignment relation between a virtual scene and a real environment. The ray detection emits a plurality of rays from the virtual camera or the sensor position, covers key points of the target area, and acquires the space coordinates of the physical space object surface according to the intersection information of the rays and the physical space object surface. The process combines the geometric alignment error value in the virtual-real scene bidirectional mapping protocol to ensure that the ray detection result can accurately reflect the geometric characteristics of the real physical space, and avoid the data deviation caused by error accumulation.
Specifically, after the ray detection is completed, the surface point cloud data of the physical space object is further extracted. And (3) removing noise points and retaining effective data by sampling and filtering the space coordinates of all ray intersection points to form a group of dense three-dimensional point sets. The surface point cloud data of the physical space object not only characterizes the surface morphology of the physical space object, but also provides basic support for subsequent mixed reality collision monitoring and virtual-real interaction. By combining the point cloud data with the precision grading parameters in the virtual-real scene bidirectional mapping protocol, the point cloud density and the sampling strategy can be dynamically adjusted, so that the calculation efficiency is optimized while the data precision is ensured.
S3.2, aligning the surface point cloud data with a three-dimensional model coordinate system, and executing a mixed reality collision detection algorithm to calculate collision probability, wherein the expression is as follows:
;
Wherein, The probability of collision is indicated and,The Sigmoid function is represented as a function,Representing the real-time pose of a real object in physical space,Representing the current pose of the NPC and objects in the virtual scene,A scene reference size is indicated and,A duration of the tracking time is indicated,Indicating the decay time constant and,A time decay function is represented and is shown,Indicating the speed of the relative movement of the two elements,Representing the current minimum euclidean distance,Indicating the angle of the movement direction.
The specific process comprises the steps of precisely aligning surface point cloud data acquired in a physical space with a three-dimensional model in a virtual scene, and ensuring a coordinate system of the surface point cloud data and the three-dimensional model, so that an accurate basis is provided for subsequent collision detection. Then, collision probability is calculated by using a Sigmoid function, and the Sigmoid function comprehensively considers real-time pose (namely position and pose) of a real object in a physical space and current pose of NPC and an object in a virtual scene, and meanwhile, the scene reference size is combined as a reference scale. In addition, the mixed reality collision detection algorithm continuously tracks time and decays time constant, and the change trend of collision probability along with time is adjusted through a time decay function to reflect the dynamic change of the collision probability in different time intervals. The relative motion velocity further affects the calculation of the collision probability, which describes the rate of approach between the physical object and the virtual character. Finally, the current minimum Euclidean distance is used to quantify the closest distance between the two, directly affecting the probability of collision occurrence.
And S3.3, dynamically adjusting the path planning of the virtual NPC according to the collision probability, and generating an instruction set containing obstacle avoidance priority and moving speed.
Further, when dynamically adjusting the path planning of the virtual NPC according to the collision probability, firstly, estimating the potential collision risk between the virtual NPC and the real object in the physical space according to the collision probability value calculated by the mixed reality collision detection algorithm. When the collision probability is high, a safety path far away from a high risk area is planned for the NPC preferentially, and obstacles are avoided by recalculating path points, and when the collision probability is low, the NPC is allowed to move along the original path, and meanwhile, certain flexibility is reserved to cope with emergency conditions. The process combines the real-time pose information and the geometric structure in the scene, and ensures that the path adjustment is efficient and meets the actual environment requirements.
Specifically, when an instruction set containing obstacle avoidance priority and moving speed is generated, the obstacle avoidance priority is set according to the collision probability, the high probability corresponds to the high priority, the NPC is required to take avoidance measures as soon as possible, and the low probability allows lower priority to be processed. Then, the movement speed of the NPC is dynamically adjusted in combination with the relative movement speed and the current minimum Euclidean distance, and the NPC is properly decelerated when approaching an obstacle to reduce the collision risk, and is accelerated in a safety area to improve the operation efficiency. Finally, the obstacle avoidance priority and the adjusted moving speed are integrated into a specific instruction set for guiding the behavior decision of the NPC, so that more intelligent and smoother virtual-real interaction experience is realized.
And S3.4, converting the instruction set of the obstacle avoidance priority and the moving speed into a driving control signal of the physical equipment, and packaging the driving control signal into a closed-loop interaction data set containing the NPC behavior instruction and the physical equipment control signal.
The specific process comprises the steps of according to the obstacle avoidance priority and the speed parameter defined in the instruction set, mapping the obstacle avoidance priority and the speed parameter into specific physical equipment operation instructions. For example, high priority obstacle avoidance commands may be translated into emergency braking or steering control signals, while adjustments in the speed of movement may correspond to motor speed or power adjustments of the actuators. Through a preset conversion rule and a device interface protocol, the logic instructions are further encoded into a language or signal format which is adapted to specific hardware, so that the physical device is directly driven to complete corresponding actions, and the virtual NPC behavior is ensured to be accurately executed in a real environment.
Further, the converted driving control signals and the NPC behavior instructions are integrated to form a unified data structure. The data structure not only comprises the path planning, obstacle avoidance strategies and other behavior information of the NPC, but also covers the actual control signals of the corresponding physical devices, wherein the actual control signals are directly generated by analyzing the global federal model parameters and the real-time data of the edge nodes and combining with the preset behavior instructions, and are used for guiding the physical devices to execute specific operations to realize the synchronism of virtual-real interaction so as to realize the synchronism of the virtual-real interaction. And then, marking the data structure through the time stamp and the scene identifier, ensuring that the transmission and analysis of the instruction set among different nodes have consistency and real-time performance, and finally generating a closed-loop interaction data set.
And S4, based on the closed-loop interaction data set, performing federation model training at each edge node, calculating federation model gradients, uploading the federation model gradients to a federation main node, and updating the federation model after the main node aggregates the gradients.
And S4.1, extracting NPC behavior instructions, equipment control signals and pose verification data by analyzing the closed-loop interaction data set, and generating a standardized training sample.
The method comprises the steps of analyzing a closed-loop interaction data set, and firstly classifying and extracting NPC behavior instructions, equipment control signals and pose verification data contained in the closed-loop interaction data set. Based on data fields and protocol rules, the process identifies and separates out behavior instructions related to NPC path planning and obstacle avoidance strategies and driving control signals corresponding to physical equipment actions, extracts real-time pose information of virtual roles and real objects in a scene, and further cleans the real-time pose information after preliminary screening to remove noise and redundant information, so that high accuracy and consistency of the extracted data are ensured, and a foundation is laid for subsequent training sample generation.
Furthermore, when the standardized training sample is generated, the extracted NPC behavior instruction, the equipment control signal and the pose verification data are encoded and normalized according to a uniform format. By mapping data of different dimensions to the same feature space and combining the time stamp and scene context information, a structured sample set is constructed. In addition, the standardized training samples are marked according to task requirements, such as collision risk level or behavior execution effect, so that standardized training samples with clear labels are formed.
And S4.2, performing federal model training based on the standardized training samples, and adding dynamic differential noise to the trained federal model.
It should be noted that, first, the federal model training is performed using a standardized training sample that contains a preprocessed and annotated data set, ensuring that each participating edge node can learn on the same basis. In the training process, each edge node updates its own local model according to the local data, and gathers the updated results into the global federation model to gradually optimize the federation model.
In particular, this process involves calculating the appropriate amount of noise and applying it to the federal model according to specific privacy protection requirements. By the method, potential privacy leakage risks can be effectively prevented on the premise of not significantly affecting the performance of the federal model. Finally, through the step of federation model training and dynamic differential noise adding based on the standardized training sample, the effective training and privacy protection of the federation model are realized, and the accuracy and the safety of the federation model in practical application are ensured.
S4.3, homomorphic encryption is carried out on the federation model added with noise, an encryption parameter packet is generated and uploaded to a cloud aggregation node, the aggregation node carries out weighted average on parameters of all nodes, and global federation model parameters are generated, wherein the expression is as follows:
;
Wherein, Representing the parameters of the global federal model,Representing a homomorphic decryption function,Representing the total number of edge nodes,Representing the index of the edge node,Representing a homomorphic encryption function,Represent the firstGlobal federal model parameters for the individual edge nodes,Representing a differential privacy sensitivity of the user,Representing the privacy budget of the user,Represent the firstThe federal weights of the individual edge nodes,Representing a standard laplace noise generator,Representing a homomorphic encryption public key,Representing a homomorphic encryption private key.
In the federation learning process, firstly, homomorphic encryption is carried out on global federation model parameters added with noise to generate an encryption parameter packet, so that the safety of data in the transmission process is ensured. After the encryption parameter package is uploaded to the cloud aggregation node, the aggregation node directly performs weighted average operation on the encryption parameter under the condition of not decrypting by utilizing the homomorphic encryption characteristic.
Specifically, the parameters of each edge node are weighted according to federation weights, the weights reflect the contribution degree of each edge node to the global federation model, and meanwhile, the differential privacy sensitivity and the privacy budget are combined, so that the privacy protection effect is ensured. And finally, restoring the aggregation result into a global federation model parameter through a homomorphic decryption function to finish updating the global federation model.
The global federation model parameters refer to unified parameter values generated after weighted average of global federation models of all edge nodes in federation learning.
S4.4, adding Laplace noise to the gradient calculated by the edge nodes to generate privacy protection gradients, and encrypting the gradient of each edge node by using Paillier, wherein the expression is as follows:
;
Wherein, Represent the firstThe edge node is at the firstThe noise gradient in the training of the round,A sequence number representing the training round,Represent the firstThe edge node is at the firstThe wheel is not noisy and the wheel is not noisy,Representing the gradient of the noise in the image,Representing warpThe sensitivity of the dynamic gradient of the wheel training,Representing the dynamic adjustment function of the device,Represent the firstThe federal weights of the individual edge nodes,Represent the firstThe real-time resource indicator of the individual edge nodes,Represents the value of the resource weight coupling function,The representation is based on training roundsIs a dynamic noise control function of (1),Representing the time-dependent turnsA dynamically decaying privacy budget function,Represents a distribution which is in a normal state,The position parameter is represented by a parameter of the position,The scale parameter is represented by a scale parameter,Representing the skewness parameter.
The specific process includes, in each training round of federal learning, first adding laplace noise to the gradient calculated by the edge node to generate a privacy preserving gradient. The process determines proper noise amount through a dynamic regulation function and a dynamic noise regulation function based on training rounds, and ensures that the performance of the global federal model is not influenced while protecting the data privacy.
Furthermore, the original gradient of each edge node is weighted according to federal weight, real-time resource index and resource weight coupling function, and then proper amount of Laplacian noise is added, so that the data contribution of a single node is difficult to be deduced by reverse engineering. To further enhance security, noisy gradients are encrypted using the Paillier encryption algorithm. The encrypted gradient is then uploaded to the cloud aggregation node, and the aggregation operation can be safely performed without revealing any private information. The method not only effectively protects the data privacy of the parties, but also promotes the efficient federal model training under multiparty collaboration.
And S4.5, carrying out weighted summation based on encryption gradients of all edge nodes, and updating global federation model parameters through natural gradient descent, wherein the expression is as follows:
;
Wherein, Represent the firstThe global federal model parameters after the round of updating,Represent the firstGlobal federal model parameters at the time of the round of training,The step size coefficient is represented as such,Representing the total number of edge nodes,Representing a homomorphic decryption operation,Represent the firstThe noise gradient of the individual edge nodes,Represent the firstThe federal weights of the individual edge nodes,Representing a homomorphic encryption private key,Representing an element-by-element multiplication,Matrix of information representing snow chargeIs a low-rank approximation of the inverse matrix,Representing a Riemann manifold regularization term.
The specific process comprises the steps of carrying out weighted summation based on encryption gradients of all edge nodes in each round of training, and updating global federal model parameters through natural gradient descent. Firstly, the gradient after encryption and noise addition processing is obtained from each edge node, and the encryption gradient is restored into a processable form by homomorphic decryption operation. Then, the encryption gradients are weighted and summed according to the federal weight of each node, so that nodes with larger contributions are ensured to have higher influence in the updating process.
And then, adopting a natural gradient descent method, combining step size coefficients to adjust the update amplitude, and optimizing the update direction by using a low-rank approximate inverse matrix of the Fisher information matrix, and simultaneously, keeping good geometric characteristics of the global federal model parameter space by using Riemann manifold regularization terms. By the method, not only is the global federation model parameters effectively updated, but also data privacy protection and efficient global federation model parameter optimization under multiparty cooperation are ensured. The process comprehensively considers the data contribution and the real-time resource condition of different nodes, and ensures that the global federal model parameters can continuously improve the performance on the premise of protecting the privacy.
And S5, triggering dynamic navigation and risk early warning by using global federation model parameters, synchronizing the optimized virtual-real scene bidirectional mapping protocol to a multi-scenic-region digital twin body through a meta-universe protocol, and constructing a cross-scene virtual travel platform.
S5.1, analyzing the tourist position, the scenic spot capacity and the environmental data in real time through the global federal model parameters, dynamically generating a guide path, and calculating dynamic association parameters of the tourist distribution and the capacity risk of the scenic spot in real time, wherein the expression is as follows:
;
Wherein, Represent the firstDynamic correlation parameters of the scenic spot guest distribution and capacity risk of the wheel,Representing a smoothed positive number activation function,Representing the mathematical expectation that the data will be,Representing the probability density function of the object,Representing the spatial coordinate vectors within the scenic spot,Is shown in the firstUpdated global federation model parameters after the secondary global aggregation,Representing a matrix of scenic spot capacity thresholds,The path gradient penalty coefficients are represented,Represent the firstWheel navigation pathIs used for the spatial gradient of (a),Representing an element-wise multiplication algorithm of the two vectors.
The specific process comprises the step of utilizing global federal model parameters updated by federation aggregation to evaluate the distribution situation of tourists in the current scenic spot and the influence of the distribution situation on the capacity of the scenic spot. The mathematical expectation and probability density functions are processed through a smooth positive number activation function, and the tourist density and potential capacity risk of each area are accurately calculated by combining the space coordinate vectors of the scenic spots. The process also considers the scenic spot capacity threshold matrix and the path gradient penalty coefficient, ensures that the generated tour guide path can guide tourists to avoid crowded areas and can optimize the whole tour experience.
Further, the path gradient penalty coefficient is used for adjusting the spatial gradient of the navigation path, so that the navigation path is prevented from being too concentrated or scattered, the people flow distribution in the scenic spot is effectively managed, and the risk of capacity overload is reduced. Finally, through comprehensive analysis and calculation, the goal of dynamically generating the optimal navigation path is realized, and the distribution and capacity risk of tourists in the scenic spot can be monitored and regulated in real time.
S5.2, encoding the navigation path and the dynamic association parameters into bidirectional mapping parameters through a meta-universe protocol, and synchronizing the navigation path and the dynamic association parameters to the multi-scenic spot digital twin nodes based on block chain consensus.
The specific process comprises the steps of carrying out structural integration on the generated guide path and dynamic association parameters such as scenic spot tourist distribution, capacity risk and the like, and ensuring that the mapping relation between virtual and real scenes is clear and consistent. The digital twin nodes of the multi-scenic spot are formed by combining the actual requirements of the physical scenic spot, guaranteeing the consistency and the safety of data by using a block chain technology and realizing the standardized processing of complex information by means of a metauniverse protocol. The distributed network is formed, so that each scenic spot can not only locally optimize the operation of the scenic spot, but also efficiently cooperate with other scenic spots, and the overall service quality is improved together. And then, converting the dynamic association parameters into a unified format adapting to the digital twin bodies of the multiple scenic spots by using standardized coding rules of the metauniverse protocol so as to realize seamless butt joint and collaborative management among different scenic spots. On the basis, the coded bidirectional mapping parameters are verified and synchronized through a block chain consensus mechanism, so that the digital twin nodes of all scenic spots can safely and reliably acquire and update data, real-time interaction and resource optimization configuration of cross-scenes are supported, and a foundation is laid for constructing a highly-linked virtual travel platform.
S5.2, constructing a virtual travel platform of virtual-real interaction across scenes based on the synchronized digital twin node parameters.
Further, the digital twin bodies of all scenic spots are updated by using the synchronized parameters, so that the high consistency and dynamic linkage of the virtual scene and the real environment are ensured. And then, correlating and fusing the digital twin node parameters of different scenic spots to form a unified cross-scene data frame, and supporting seamless switching and interaction of tourists among a plurality of scenic spots. Based on the method, virtual roles, navigation paths and physical equipment control signals are synchronized in real time by combining virtual-real mapping protocols and interaction logic, so that immersive tour experience is provided for tourists. By the method, not only is the efficient cooperation and optimal configuration of the resources of the scenic spots realized, but also the continuity and participation of tourists in the cross-scene tour are improved, and a highly intelligent virtual travel platform with strong interactivity is created.
The embodiment also provides computer equipment, which is suitable for the situation of the virtual travel platform construction method based on the digital twin, and comprises a memory and a processor, wherein the memory is used for storing computer executable instructions, and the processor is used for executing the computer executable instructions to realize the virtual travel platform construction method based on the digital twin, which is provided by the embodiment.
The computer device may be a terminal comprising a processor, a memory, a communication interface, a display screen and input means connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage media. The communication interface of the computer device is used for carrying out wired or wireless communication with an external terminal, and the wireless mode can be realized through WIFI, an operator network, NFC (near field communication) or other technologies. The display screen of the computer equipment can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer equipment can be a touch layer covered on the display screen, can also be keys, a track ball or a touch pad arranged on the shell of the computer equipment, and can also be an external keyboard, a touch pad or a mouse and the like.
The present embodiment also provides a storage medium having a computer program stored thereon, which when executed by a processor implements the virtual travel platform construction method based on digital twinning as proposed in the above embodiments, and the storage medium may be implemented by any type of volatile or non-volatile storage device or combination thereof, such as a static random access Memory (Static Random Access Memory, SRAM for short), an electrically erasable Programmable Read-Only Memory (ELECTRICALLY ERASABLE PROGRAMMABLE READ-Only Memory, EEPROM for short), an erasable Programmable Read-Only Memory (Erasable Programmable Read Only Memory, EPROM for short), a Programmable Read-Only Memory (PROM for short), a Read-Only Memory (ROM for short), a magnetic Memory, a flash Memory, a magnetic disk or an optical disk.
In conclusion, the method and the device dynamically allocate the federal model precision weight, trigger the target detection algorithm to carry out semantic segmentation on the key region by combining the real-time video stream frame rate fluctuation, calculate the geometric alignment error value to divide the precision level, enable the virtual environment to be in seamless butt joint with the real world, and dynamically switch the NPC behavior logic complexity based on the precision grading parameter, thereby improving the sense of reality and interactivity of the virtual character and reducing the calculation cost by optimizing the resource use.
It should be noted that the above embodiments are only for illustrating the technical solution of the present invention and not for limiting the same, and although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention may be modified or substituted without departing from the spirit and scope of the technical solution of the present invention, which is intended to be covered in the scope of the claims of the present invention.
Claims (10)
1. A virtual travel platform construction method based on digital twinning is characterized by comprising the following steps of,
Collecting a text traveling domain real-time data stream, initializing a federation learning framework at an edge point, and generating a federation learning initialization instruction set;
Based on a federation learning initialization instruction set, dynamically distributing federation model precision weights through a federation learning framework, and simultaneously calibrating a real-time video stream with a three-dimensional model by utilizing a video twin engine to generate a virtual-real scene bidirectional mapping protocol containing precision grading parameters and NPC control instructions;
based on a virtual-real scene bidirectional mapping protocol, scanning a physical space three-dimensional geometric structure through ray detection and executing a mixed reality collision monitoring algorithm to generate a closed-loop interaction data set containing NPC behavior instructions and physical equipment control signals;
Based on the closed-loop interaction data set, performing federation model training on each edge node, calculating federation model gradients and uploading the federation model gradients to a federation master node, and updating global federation model parameters after the master node aggregates the gradients;
And triggering dynamic navigation and risk early warning by using global federation model parameters, synchronizing the optimized virtual-real scene bidirectional mapping protocol to a multi-scenic-region digital twin body through a meta-universe protocol, and constructing a cross-scene virtual travel platform.
2. The method for constructing a virtual travel platform based on digital twinning as set forth in claim 1, wherein the travel domain real-time data stream comprises guest behavior thermodynamic distribution data, environmental parameters, cultural relic status data and real-time video stream data.
3. The method for constructing a virtual travel platform based on digital twin according to claim 2, wherein the step of initializing a federal learning frame at an edge point to generate a federal learning initialization instruction set comprises the following steps,
Initializing a federation learning framework at an edge node, deploying a lightweight federation learning client, loading a text travel global real-time data stream, and outputting a local training data set with aligned dimensions;
Based on the local training data set, dynamically generating a federal learning initialization instruction set containing federal model architecture templates, aggregation weight rules and communication protocol parameters through multi-modal learning.
4. The method for constructing a virtual travel platform based on digital twin, as set forth in claim 3, wherein the federal learning initialization instruction set dynamically distributes federal model precision weights through a federal learning framework, and simultaneously calibrates real-time video streams with a three-dimensional model by using a video twin engine to generate a virtual-real scene bidirectional mapping protocol comprising precision grading parameters and NPC control instructions,
Based on the federation learning initialization instruction set, extracting performance parameters and data reliability indexes of the edge nodes, and dynamically distributing federation model precision weights;
when the video twin engine detects a real-time video stream frame rate fluctuation hyperstability threshold, triggering a target detection algorithm to perform semantic segmentation on a key region, and extracting the boundary frame coordinates of a moving object;
calculating a geometric alignment error value and dividing a precision level based on the boundary frame coordinates to generate precision grading parameters comprising grid density and algorithm selection rules;
Based on the precision grading parameters, evaluating the geometric alignment error value of the current scene and dynamically switching the NPC behavior engine containing logic complexity by using the federal model precision weight through scene consistency analysis;
And packaging the federal model precision weight, the geometric alignment error value and the NPC behavior engine into a structured data packet, synchronizing the structured data packet to an edge node and a cloud host node through a space-time unified interface, and generating a virtual-real scene bidirectional mapping protocol containing precision grading parameters and NPC control instructions.
5. The method for constructing a virtual travel platform based on digital twin, as set forth in claim 4, wherein the method comprises generating a closed-loop interactive data set comprising NPC behavior instructions and physical device control signals by scanning a physical space three-dimensional geometry structure through ray detection and executing a mixed reality collision monitoring algorithm based on a virtual-real scene bidirectional mapping protocol,
Based on a virtual-real scene bidirectional mapping protocol, scanning a physical space three-dimensional geometric structure through ray detection, and extracting surface point cloud data of a physical space object;
Aligning the three-dimensional model coordinate system based on the surface point cloud data, and executing a mixed reality collision detection algorithm to calculate the collision probability;
dynamically adjusting the path planning of the virtual NPC according to the collision probability, and generating an instruction set containing obstacle avoidance priority and moving speed;
And converting the instruction set of the obstacle avoidance priority and the moving speed into a driving control signal of the physical equipment, and packaging the driving control signal into a closed-loop interaction data set containing the NPC behavior instruction and the physical equipment control signal.
6. The method for constructing a virtual travel platform based on digital twin according to claim 5, wherein the federal model training is performed at each edge node based on the closed-loop interaction data set by the steps of,
Extracting NPC behavior instructions, physical equipment control signals and pose verification data by analyzing a closed-loop interaction data set, and generating a standardized training sample;
Performing federal model training based on the standardized training samples, and adding dynamic differential noise to the trained federal model;
Homomorphic encryption is carried out on the federation model added with noise, an encryption parameter packet is generated and uploaded to a cloud aggregation node, and the aggregation node performs weighted average on parameters of all nodes and generates global federation model parameters.
7. The method for constructing a virtual travel platform based on digital twin according to claim 6, wherein the step of calculating the federation model gradient and uploading the federation model gradient to a federation master node, the master node aggregating the gradient and updating global federation model parameters comprises the following steps,
Adding Laplace noise to the gradient calculated by the edge nodes to generate privacy protection gradients, and encrypting the gradients of the edge nodes by using Paillier;
and carrying out weighted summation based on the encryption gradient of each edge node, and updating the global federation model parameters through natural gradient descent.
8. The method for constructing a virtual travel platform based on digital twin according to claim 7, wherein the method comprises triggering dynamic navigation and risk early warning by using global federal model parameters, synchronizing the optimized virtual-real scene bidirectional mapping protocol to a multi-scenic region digital twin through a meta-universe protocol, constructing a cross-scene virtual travel platform,
Analyzing the tourist position, the scenic spot capacity and the environmental data in real time through the global federal model parameters, dynamically generating a guiding path, and calculating dynamic association parameters of the tourist distribution and the capacity risk of the scenic spot in real time;
Encoding the navigation path and the dynamic association parameters into bidirectional mapping parameters through a meta universe protocol, and synchronizing the bidirectional mapping parameters to the digital twin nodes of the scenic spots based on block chain consensus;
based on the synchronized digital twin nodes, a virtual travel platform with virtual-real interaction across scenes is constructed.
9. A computer device comprises a memory and a processor, wherein the memory stores a computer program, and the computer program is characterized in that the processor realizes the steps of the method for constructing the virtual travel platform based on digital twinning according to any one of claims 1-8 when executing the computer program.
10. A computer readable storage medium having a computer program stored thereon, wherein the computer program when executed by a processor implements the steps of the method for constructing a digital twin-based virtual travel platform of any one of claims 1 to 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510472264.5A CN119992019B (en) | 2025-04-16 | 2025-04-16 | Virtual travel platform construction method based on digital twin |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202510472264.5A CN119992019B (en) | 2025-04-16 | 2025-04-16 | Virtual travel platform construction method based on digital twin |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN119992019A true CN119992019A (en) | 2025-05-13 |
| CN119992019B CN119992019B (en) | 2025-08-08 |
Family
ID=95628688
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202510472264.5A Active CN119992019B (en) | 2025-04-16 | 2025-04-16 | Virtual travel platform construction method based on digital twin |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN119992019B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN120316829A (en) * | 2025-06-16 | 2025-07-15 | 徐州医科大学 | A comprehensive evaluation method for medical big models based on digital twin medical continuum |
Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117669699A (en) * | 2023-11-28 | 2024-03-08 | 山东大学 | A semantic information federated learning method and system for digital twins in an industrial Internet of Things scenario |
| CN118822792A (en) * | 2024-08-02 | 2024-10-22 | 西安云图信息技术有限公司 | A smart cultural tourism digital twin interactive system |
| CN118890285A (en) * | 2024-07-05 | 2024-11-01 | 黑龙江大学 | Digital twin modeling method for privacy protection in cyber-physical energy system |
-
2025
- 2025-04-16 CN CN202510472264.5A patent/CN119992019B/en active Active
Patent Citations (3)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117669699A (en) * | 2023-11-28 | 2024-03-08 | 山东大学 | A semantic information federated learning method and system for digital twins in an industrial Internet of Things scenario |
| CN118890285A (en) * | 2024-07-05 | 2024-11-01 | 黑龙江大学 | Digital twin modeling method for privacy protection in cyber-physical energy system |
| CN118822792A (en) * | 2024-08-02 | 2024-10-22 | 西安云图信息技术有限公司 | A smart cultural tourism digital twin interactive system |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN120316829A (en) * | 2025-06-16 | 2025-07-15 | 徐州医科大学 | A comprehensive evaluation method for medical big models based on digital twin medical continuum |
Also Published As
| Publication number | Publication date |
|---|---|
| CN119992019B (en) | 2025-08-08 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN118283548B (en) | Intelligent security system and method for smart city based on 5G | |
| US11900536B2 (en) | Visual-inertial positional awareness for autonomous and non-autonomous tracking | |
| US10410328B1 (en) | Visual-inertial positional awareness for autonomous and non-autonomous device | |
| US10366508B1 (en) | Visual-inertial positional awareness for autonomous and non-autonomous device | |
| Weikersdorfer et al. | Event-based particle filtering for robot self-localization | |
| JP2021111385A (en) | System and method enabling collaborative 3d map data fusion platform and virtual world system thereof | |
| Wei et al. | Survey of connected automated vehicle perception mode: from autonomy to interaction | |
| CN106791613A (en) | A kind of intelligent monitor system being combined based on 3DGIS and video | |
| CN117671130A (en) | Digital twin intelligent fishing port construction and use method based on oblique photography | |
| CN119992019B (en) | Virtual travel platform construction method based on digital twin | |
| CN115103161A (en) | Patrol method, system, equipment and medium based on video propulsion | |
| CN111973979B (en) | Live management of the real world via a persistent virtual world system | |
| Kong et al. | The RoboSense challenge: Sense anything, navigate anywhere, adapt across platforms | |
| CN118393900A (en) | Automatic driving decision control method, device, system, equipment and storage medium | |
| CN117033727A (en) | Satellite terminal path calculation and display method based on OpenLayers | |
| Huang et al. | Vehicle-to-everything cooperative perception for autonomous driving | |
| Hassan et al. | Toward the creation of a digital twin authoring tool: A smart mobility perspective in smart cities | |
| CN120927018A (en) | Vehicle passing path generation method and device, vehicle and storage medium | |
| CN120633467B (en) | Pedestrian track prediction method, system, equipment and medium based on continuous learning | |
| CN120337284A (en) | Virtual reality multi-person interaction method and system | |
| CN120298460A (en) | A multi-target tracking system for highway scenes based on UAV visual perception | |
| CN115272909B (en) | An unmanned driving test monitoring system | |
| Zuo et al. | Method for acquiring passenger standing positions on subway platforms based on binocular vision | |
| Yu et al. | Pedestrian trajectory prediction using spatio-temporal vae | |
| Corradini et al. | Experimental evaluation of road-crossing decisions by autonomous wheelchairs against environmental factors |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |