A virtual reality-based consumer experience scenario simulation method and system

Abstract

The application discloses a kind of based on virtual reality's consumption experience scene simulation method and system, it is related to virtual reality simulation technical field, including, receiving user in virtual reality consumption scene through voice and text input natural language requirement description;Behavior data sequence generated when user and updated virtual reality consumption scene are interacted is collected and recorded;Behavior data sequence is input decision interference analysis mechanism, extract multiple scene factors from behavior data sequence, identify and quantify the interference factor leading to user decision process interruption from multiple scene factors;According to interference factor, generate dynamic optimization instruction, and according to dynamic optimization instruction, the information presentation density and interactive process complexity of updated virtual reality consumption scene are adjusted in real time.The application can accurately identify the interference factor leading to decision interruption by real-time analysis of user behavior data sequence through decision interference analysis mechanism.

Description

Technical Field

[0001] This invention relates to the field of virtual reality simulation technology, and in particular to a method and system for simulating consumer experience scenarios based on virtual reality. Background Technology

[0002] Virtual reality technology is increasingly being applied in the consumer experience field, and existing technologies can now construct immersive shopping environments through 3D modeling and real-time rendering. In terms of scene generation, deep learning-based cross-modal generation models can convert text into 3D models, while behavioral analysis technology optimizes user experience through eye tracking and interaction data collection. Related research mainly focuses on two directions: static scene construction and multimodal content generation, providing a technological foundation for the simulation of virtual consumption scenarios.

[0003] However, existing technologies lack the ability to dynamically perceive and adaptively optimize the user's decision-making process. Current methods mostly rely on preset rules or offline data analysis, which cannot identify interfering factors that cause decision interruption in real time during the interaction process. When users face information overload or excessively complex interaction processes, it is difficult to adjust scene parameters in a timely manner, resulting in a decline in user experience and reduced decision-making efficiency, which restricts the improvement of the simulation effect of virtual consumption scenarios. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a virtual reality-based consumer experience scenario simulation method to solve the problem of user decision-making process interruption caused by the lack of real-time decision interference identification and dynamic optimization mechanisms in virtual reality consumer scenarios.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for simulating a consumer experience scenario based on virtual reality, comprising: receiving a natural language demand description input by a user in a virtual reality consumer scenario via voice and text; constructing a cross-modal generative artificial intelligence model, inputting the natural language demand description into the cross-modal generative artificial intelligence model, interpreting the demand description semantics of the natural language demand description, and outputting a three-dimensional data stream that matches the demand description semantics and conforms to the virtual reality consumer scenario; instantiating the three-dimensional data stream into interactive virtual objects and dynamically integrating and rendering them into the virtual reality consumer scenario to generate an updated virtual reality consumer scenario; collecting and recording behavioral data sequences generated when the user interacts with the updated virtual reality consumer scenario; inputting the behavioral data sequences into a decision interference analysis mechanism, extracting multiple scenario factors from the behavioral data sequences, identifying and quantifying interference factors that cause interruptions in the user's decision-making process from the multiple scenario factors; generating dynamic optimization instructions based on the interference factors, and adjusting the information presentation density and interaction process complexity of the updated virtual reality consumer scenario in real time according to the dynamic optimization instructions.

[0008] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the natural language demand description includes the attribute description, functional description, and relationship description of the virtual object and the virtual reality consumer scenario.

[0009] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the specific steps for constructing the cross-modal generative artificial intelligence model are as follows:

[0010] A semantic planning layer is built using a multimodal alignment algorithm, and a three-dimensional execution layer is built using a conditional diffusion mechanism.

[0011] By integrating the semantic planning layer and the 3D execution layer through an iterative optimization mechanism, a cross-modal generative artificial intelligence model is constructed.

[0012] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the output is a three-dimensional data stream that semantically matches the demand description and conforms to the virtual reality consumer scenario. The specific steps are as follows:

[0013] Semantic parsing is performed on natural language requirement descriptions to extract attribute descriptions, functional descriptions, and relational descriptions, generating structured semantic feature vectors;

[0014] Multi-dimensional perception analysis is performed on virtual reality consumption scenarios to extract spatial scale features and lighting style features, and to generate scene environment feature vectors.

[0015] The structured semantic feature vector and the scene environment feature vector are fused at the feature level to generate a scene condition vector;

[0016] The scene condition vector is iteratively denoised using a conditional diffusion mechanism to generate preliminary 3D geometric data.

[0017] The initial 3D geometric data is validated by differentiable rendering, and the output is a 3D data stream that semantically matches the requirements description and conforms to the virtual reality consumption scenario.

[0018] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the specific steps for generating the updated virtual reality consumer scenario are as follows:

[0019] Physically instantiate the 3D data stream to generate interactive virtual objects;

[0020] Interactive virtual objects are integrated into virtual reality consumption scenarios through potential energy field optimization algorithms to determine the optimal position and posture.

[0021] By solving the rendering equation, pixel color values ​​are calculated and materials are fused to determine the optimal position and posture of the virtual object, generating a lighting-blended virtual object that is coordinated with the lighting environment of the virtual reality consumer scene.

[0022] Collect the viewpoint changes generated by the user during the interaction, and dynamically schedule the lighting blending virtual object at the level of detail based on the viewpoint changes to generate virtual objects with optimized level of detail.

[0023] Perform semantic consistency verification between the virtual objects optimized at the level of detail and the virtual reality consumption scene, and generate an updated virtual reality consumption scene.

[0024] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the specific steps for collecting and recording the behavioral data sequence generated when the user interacts with the updated virtual reality consumer scenario are as follows:

[0025] By synchronously collecting user behavior data in virtual reality consumption scenarios through a multimodal sensor array, a timestamp-aligned behavior data stream is established;

[0026] Topological feature extraction and entropy change analysis are performed on behavioral data streams to generate structured behavioral event sequences;

[0027] The structured behavioral event sequence is encoded and encapsulated to generate a behavioral data sequence.

[0028] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the behavioral data sequence includes the user's visual gaze trajectory, the interaction operation sequence with virtual objects, and the session duration.

[0029] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the specific steps for identifying and quantifying the interference factors that cause the interruption of the user's decision-making process from multiple scenario factors are as follows:

[0030] The user's visual gaze trajectory, the sequence of interactions with virtual objects, and the duration of the session are input into the decision interference analysis mechanism to extract multiple scene factors and generate a set of scene factors.

[0031] A causal discovery algorithm is used to mine the causal structure of the scene factor set and construct a causal network among the scene factors.

[0032] Interference propagation analysis is performed using causal networks among scenario factors to calculate the causal impact strength of each scenario factor on the interruption of the decision-making process.

[0033] Based on the strength of causal influence, counterfactual reasoning methods are applied to simulate potential interference scenarios and identify initial interference factors.

[0034] The impact of the initial interference factors is quantitatively analyzed to identify the interference factors that cause the interruption of the user's decision-making process.

[0035] As a preferred embodiment of the virtual reality-based consumer experience scenario simulation method of the present invention, the specific steps of generating dynamic optimization instructions based on interference factors and adjusting the information presentation density and interaction process complexity of the updated virtual reality consumer scenario in real time according to the dynamic optimization instructions are as follows:

[0036] Based on the interference factor, optimize the updated information density control parameters and interaction complexity parameters of the virtual reality consumption scenario, and generate dynamic optimization instructions;

[0037] The dynamic optimization command is input into the adaptive control mechanism, and the optimal control sequence is determined through the model predictive control algorithm. This drives the scene renderer to adjust the distribution of the detail level of virtual objects, and at the same time drives the interaction controller to adjust the state transition logic of the interaction process, thus completing the initial optimization and adjustment of the updated virtual reality consumption scenario.

[0038] Based on the initial optimization and adjustments, a real-time feedback verification mechanism was established to collect user eye movement trajectory and interaction response data, calculate optimization effect indicators, dynamically correct control parameters, and continuously optimize in a closed loop.

[0039] Secondly, the present invention provides a virtual reality-based consumer experience scenario simulation system, including a demand input module, a 3D generation module, a scenario integration module, a data acquisition module, an interference analysis module, and an optimization and adjustment module;

[0040] The demand input module is used to receive natural language demand descriptions input by users through voice and text in virtual reality consumption scenarios.

[0041] The 3D generation module is used to construct a cross-modal generative artificial intelligence model. It inputs the natural language demand description into the cross-modal generative artificial intelligence model, interprets the demand description semantics of the natural language demand description, and outputs a 3D data stream that matches the demand description semantics and conforms to the virtual reality consumption scenario.

[0042] The scene integration module is used to instantiate the three-dimensional data stream into interactive virtual objects and dynamically integrate and render them into the virtual reality consumption scene to generate an updated virtual reality consumption scene.

[0043] The data acquisition module is used to collect and record the behavioral data sequence generated when the user interacts with the updated virtual reality consumption scenario;

[0044] The interference analysis module is used to input the behavioral data sequence into the decision interference analysis mechanism, extract multiple scenario factors from the behavioral data sequence, and identify and quantify the interference factors that cause the interruption of the user's decision-making process from the multiple scenario factors.

[0045] The optimization and adjustment module is used to generate dynamic optimization instructions based on interference factors, and adjust the information presentation density and interaction process complexity of the updated virtual reality consumption scene in real time according to the dynamic optimization instructions.

[0046] The beneficial effects of this invention are as follows: By analyzing user behavior data sequences in real time through a decision interference analysis mechanism, it can accurately identify interference factors that lead to decision interruptions. Based on multi-dimensional scenario factor extraction and causal inference algorithms, the decision interference analysis mechanism can quantitatively assess the impact of each factor on the decision-making process and generate targeted dynamic optimization instructions. By adjusting the information presentation density and interaction process complexity in real time, it effectively reduces the user's cognitive load, ensures the continuity of the decision-making process, and enhances the user's immersion and decision-making efficiency in virtual consumption scenarios. Attached Figure Description

[0047] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0048] Figure 1 This is a flowchart of a method for simulating consumer experience scenarios based on virtual reality.

[0049] Figure 2This is a schematic diagram of a virtual reality-based consumer experience scenario simulation system.

[0050] Figure 3 This is a flowchart for outputting a 3D data stream.

[0051] Figure 4 A flowchart for identifying and quantifying interference factors. Detailed Implementation

[0052] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0053] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0054] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0055] Reference Figures 1-4 This is one embodiment of the present invention, which provides a method for simulating consumer experience scenarios based on virtual reality, including the following steps:

[0056] S1. Receive natural language descriptions of user needs input via voice and text in virtual reality consumption scenarios;

[0057] S1.1: Natural language requirements description includes the attribute description, functional description, and relationship description of virtual objects with virtual reality consumption scenarios.

[0058] Specifically, in virtual reality consumption scenarios, voice input devices and text input devices are activated. The voice input device is integrated into the microphone hardware in the virtual reality headset, and the text input device is integrated into the virtual keyboard or gesture recognition in the virtual reality controller.

[0059] When a user speaks through a voice input device, the microphone captures the analog voice signal and converts it into a digital audio data stream. When a user types through a text input device, the virtual keyboard or gesture recognition generates a sequence of text characters and stores it as a text data stream.

[0060] The digital audio data stream is converted into a text string using speech recognition software and then merged with the text data stream to form a natural language requirement description.

[0061] It should be noted that attribute descriptions refer to detailed descriptions of the physical characteristics of virtual objects (such as color, shape, and size).

[0062] Functional description refers to the purpose or function of a virtual object (such as defining it as a seat or tool).

[0063] Relationship description refers to the description of the position or interaction relationship between a virtual object and other virtual objects in a virtual reality consumption scenario.

[0064] S2. Construct a cross-modal generative artificial intelligence model. Input the natural language demand description into the cross-modal generative artificial intelligence model, interpret the demand description semantics of the natural language demand description, and output a three-dimensional data stream that matches the demand description semantics and conforms to the virtual reality consumption scenario.

[0065] S2.1: Use a multimodal alignment algorithm to build a semantic planning layer and use a conditional diffusion mechanism to build a three-dimensional execution layer;

[0066] Specifically, when building the semantic planning layer, a dataset containing text descriptions and corresponding 3D shapes is used for training. Each 3D shape in the dataset is encoded by a 3D encoder to generate a fixed-dimensional vector representation, namely the 3D shape embedding vector. The BERT architecture of Transformer is used to process natural language requirement descriptions. The text information is encoded through the self-attention mechanism in BERT to form text embedding vectors. In this process, a multimodal alignment algorithm is applied. Through contrastive learning training, the text embedding vectors and 3D shape embedding vectors are aligned in a shared feature space, forming the structured text feature vectors generated by the semantic planning layer. When building the 3D execution layer, a conditional diffusion mechanism is used. The structured text feature vectors generated by the semantic planning layer are used as conditional information. A denoising network with a U-Net structure is used to perform a stepwise denoising process from random noise to 3D data, forming the 3D execution layer.

[0067] S2.2: Through iterative optimization mechanisms, the semantic planning layer and the three-dimensional execution layer are collaboratively integrated to construct a cross-modal generative artificial intelligence model.

[0068] Specifically, the gradient descent optimization method is used to simultaneously update the BERT parameters in the semantic planning layer and the conditional diffusion parameters in the 3D execution layer during training. The structured text feature vectors generated by the semantic planning layer are passed to the 3D execution layer as conditional information, and the preliminary 3D geometric data generated by the 3D execution layer is used to evaluate the degree of matching with the text semantics. The parameters of BERT and conditional diffusion are adjusted according to the feedback signal generated by the degree of matching. After multiple iterations, the collaborative performance of the semantic planning layer and the 3D execution layer is improved, forming a cross-modal generative artificial intelligence model.

[0069] S2.3: Perform semantic parsing on the natural language requirement description, extract attribute descriptions, functional descriptions and relational descriptions, and generate structured semantic feature vectors;

[0070] Specifically, the semantics of natural language demand descriptions are parsed through BERT's self-attention mechanism in the semantic planning layer, identifying the features of attribute descriptions, the uses of functional descriptions, and the spatial relationships of relational descriptions. The identified descriptions are then converted into structured vector representations, forming structured semantic feature vectors.

[0071] S2.4: Perform multi-dimensional perception analysis on virtual reality consumption scenarios, extract spatial scale features and lighting style features, and generate scene environment feature vectors;

[0072] Specifically, virtual reality consumption scenarios exist in the form of point cloud data. The spatial structure of the point cloud data is analyzed through the PointNet architecture to capture the spatial scale characteristics of the virtual reality consumption scenario, including spatial size and object layout information. At the same time, the rendered images of the virtual reality consumption scenario are analyzed through image feature extraction algorithms to capture the lighting style characteristics of the virtual reality consumption scenario, including light intensity and color distribution. The spatial scale features and lighting style features are converted into feature representations and then combined to form a scene environment feature vector.

[0073] S2.5: Perform feature-level fusion of structured semantic feature vectors and scene environment feature vectors to generate scene condition vectors;

[0074] Specifically, the structured semantic feature vector contains attribute descriptions, functional descriptions, and relational descriptions extracted from natural language demand descriptions, while the scene environment feature vector contains spatial scale features and lighting style features extracted from virtual reality consumption scenarios. The structured semantic feature vector and the scene environment feature vector are combined through vector concatenation to form a concatenated feature vector. After the concatenated feature vector is adjusted in dimension by a fully connected layer, a scene condition vector is generated.

[0075] S2.6: Iteratively denoise the scene condition vector through a conditional diffusion mechanism to generate preliminary 3D geometric data;

[0076] Specifically, the scene condition vector serves as conditional information, guiding the denoising process of the conditional diffusion mechanism. The conditional diffusion mechanism starts with random noise and gradually reduces noise interference through a denoising network with a U-Net structure. Each denoising process combines the semantic requirements and environmental constraint information in the scene condition vector. After multiple iterations, the random noise is gradually transformed into point cloud data with geometric structure, generating preliminary three-dimensional geometric data.

[0077] S2.7: Perform differentiable rendering verification on the preliminary 3D geometric data, and output a 3D data stream that matches the semantics of the requirement description and conforms to the virtual reality consumption scenario.

[0078] Specifically, a differentiable renderer is used to convert the initial 3D geometric data into a 2D image view. By comparing the 2D image view with the attribute features specified in the natural language requirement description, the degree of matching between the initial 3D geometric data and the semantics of the requirement description is evaluated. At the same time, the coordination between the 2D image view and the lighting style features and spatial scale features of the virtual reality consumption scene is checked. Based on the evaluation results, the initial 3D geometric data is adjusted and optimized so that the 3D geometric data simultaneously meets the semantic matching requirements and the environmental adaptation requirements, generating a 3D data stream that matches the semantics of the requirement description and conforms to the virtual reality consumption scene.

[0079] S3. Instantiate the 3D data stream into an interactive virtual object and dynamically integrate and render it into the virtual reality consumption scene to generate an updated virtual reality consumption scene.

[0080] S3.1: Perform physical instantiation processing on the 3D data stream to generate interactive virtual objects;

[0081] Specifically, the 3D data stream contains 3D data in point cloud format. Based on the functional description extracted from the natural language requirement description, corresponding physical attribute parameters, including mass and friction attributes, are added to the 3D data stream. Using the 3D data in point cloud format from the 3D data stream, a corresponding collider structure is generated to ensure that the collider precisely matches the outline defined in the 3D data stream. At the same time, interactive behavior attributes, including graspable and movable attributes, are added to generate interactive virtual objects.

[0082] S3.2: Integrate interactive virtual objects into virtual reality consumption scenarios using potential energy field optimization algorithms to determine optimal positions and postures;

[0083] Specifically, the potential field optimization algorithm constructs a composite potential field containing attraction and repulsion based on the spatial scale characteristics of virtual reality consumption scenarios and the physical properties of interactive virtual objects. The attraction guides the interactive virtual objects to move toward the target position specified in the natural language demand description, while the repulsion prevents the interactive virtual objects from penetrating existing objects in the virtual reality consumption scenario. Under the combined action of the composite potential field, the interactive virtual objects automatically adjust their spatial position and rotational posture to stabilize at the optimal position and posture.

[0084] S3.3: By solving the rendering equation, pixel color values ​​are calculated and materials are fused for virtual objects with optimal positions and postures to generate lighting-blended virtual objects that are coordinated with the lighting environment of virtual reality consumer scenes.

[0085] Specifically, the lighting style features are used as the lighting environment. The lighting style features provide the ambient light intensity and the direction of the main light source. The base color and metallicity attributes of the virtual object material with the optimal position and posture are derived from the 3D data stream. The material blending coefficient is determined based on the metallicity attribute. The material blending coefficient is used to perform weighted blending of the diffuse color value of the virtual object with the ambient reflection color value. The diffuse color value of the virtual object with the optimal position and posture is determined by the base color, the dot product of the normal vector and the light source direction, and the ambient light intensity. The ambient reflection color value is obtained by sampling from the environment map generated by the lighting style features of the virtual reality consumer scene. Coordinated pixel color values ​​are generated for each surface point of the virtual object with the optimal position and posture, thereby generating a lighting blended virtual object that is coordinated with the lighting environment of the virtual reality consumer scene.

[0086] The pixel color values ​​and material fusion of virtual objects with optimal positions and poses are calculated by solving the rendering equation. The expression is as follows:

[0087] ;

[0088] ;

[0089] In the formula, Indicates pixel color; Indicates the material fusion coefficient; This represents the diffuse color value of the virtual object itself, indicating its optimal position and orientation. This represents the ambient light color value reflected from the surface of a virtual object, indicating its optimal position and orientation. The base color of the material of the virtual object whose optimal position and orientation have been determined; Represents the normal vector of a point on a virtual object whose optimal position and orientation are determined; This represents the direction vector from a virtual object surface point with a determined optimal position and orientation to the scene's main light source; This indicates the brightness value of the ambient light.

[0090] It should be noted that the material blending coefficient comes from the metallicity attribute of the virtual object material defined in the 3D data stream. The example value is 0.05 (for metallic materials such as silver jewelry) or 0.95 (for non-metallic materials such as plastic toys). The value is based on the characteristic that metallic materials in physical rendering mainly rely on environmental reflection rather than their own diffuse reflection.

[0091] S3.4: Collect the viewpoint changes generated by the user during the interaction, and dynamically schedule the lighting blending virtual object according to the viewpoint changes to generate a virtual object with optimized detail level;

[0092] Specifically, the system collects viewpoint change data generated by the user during interaction through the built-in tracking sensors of the virtual reality headset, including viewpoint position and gaze direction. Based on the distance between the viewpoint position and the center point of the lighting blending virtual object, as well as the angle between the gaze direction and the surface normal of the lighting blending virtual object, the required level of detail is determined. The dynamic scheduling mechanism of the level of detail selects the corresponding precision geometric representation from the lighting blending virtual object according to the determined level of detail. When the viewpoint position moves, the dynamic scheduling mechanism of the level of detail smoothly transitions between different precision geometric representations to avoid abrupt visual changes. Finally, a virtual object with optimized level of detail is generated.

[0093] S3.5: Perform semantic consistency verification between the virtual objects optimized at the level of detail and the virtual reality consumption scene, and generate an updated virtual reality consumption scene.

[0094] Specifically, the verification process includes spatial relationship verification and functional compatibility verification. Spatial relationship verification checks whether the position and posture of the virtual objects optimized at the level of detail in the virtual reality consumption scene conform to the relationship description specified in the natural language requirement description. Functional compatibility verification checks whether there is a conflict between the functional description of the virtual objects optimized at the level of detail and the functions of existing objects in the virtual reality consumption scene. When both spatial relationship verification and functional compatibility verification are passed, the virtual objects optimized at the level of detail are persistently integrated into the virtual reality consumption scene to generate an updated virtual reality consumption scene.

[0095] S4. Collect and record the sequence of behavioral data generated when users interact with the updated virtual reality consumption scene.

[0096] S4.1: Synchronously collect user behavior data in virtual reality consumption scenarios through a multimodal sensor array to establish a timestamp-aligned behavior data stream;

[0097] Specifically, the multimodal sensor array includes an eye tracker, an inertial measurement unit (IMU), and a handheld controller integrated into the virtual reality headset. The eye tracker records visual gaze sequences generated by eye movements at a specific frequency, the IMU records posture data generated by head movements at a specific frequency, and the controller records interactive operation sequences generated by hand operations at a specific frequency. All sensors are coordinated through a shared hardware synchronization signal. When the synchronization signal is triggered, each sensor simultaneously marks the currently recorded data frame with the same time reference. Based on the same time reference, the data streams from different sensors are aligned and combined in the time dimension to establish a timestamp-aligned behavioral data stream.

[0098] S4.2: Perform topological feature extraction and entropy change analysis on the behavioral data stream to generate a structured sequence of behavioral events;

[0099] Specifically, the spatial distribution pattern of the visual gaze point sequence is analyzed using a sliding window to extract the topological features of the clustering degree of the visual gaze point sequence distribution. The topological features of the clustering degree include the convex hull area of ​​the gaze point and the average nearest neighbor distance. By analyzing the temporal variation pattern of the interactive operation sequence, the regularity entropy change feature of the operation rhythm is evaluated. The regularity entropy change feature adopts the Shannon entropy of the interval time of the operation event. When the extracted gaze point convex hull area is lower than a preset area threshold or the Shannon entropy exceeds a preset Shannon entropy threshold, the start and end of the behavior event are marked. All marked behavior events are organized in chronological order to generate a structured behavior event sequence.

[0100] It should be noted that the area threshold is set based on the average display size of the virtual object in the scene, and the value ranges from 5% to 20% of the display area of ​​the virtual object; below 5% will cause normal scattered gaze to be misjudged as a convergence event, and above 20% will cause the real gaze convergence behavior to be unable to be effectively captured;

[0101] The Shannon entropy threshold is set based on the statistical distribution characteristics of user operation interval time, and the value ranges from 1.5 to 3. A value below 1.5 will cause the user's normal hesitant operation to be misjudged as a smooth operation, and a value above 3 will cause the user's occasional efficient operation to be misjudged as chaotic operation.

[0102] S4.3: Encode and encapsulate the structured behavioral event sequence to generate a behavioral data sequence.

[0103] Specifically, when encoding and encapsulating structured behavioral event sequences, the Protocol Buffers serialization protocol is used to define the data structure of the structured behavioral event sequence, including event type field, timestamp field, and feature description field. Each structured behavioral event is converted according to the Protocol Buffers message format, and the event type enumeration value, start and end timestamps, and feature vector array are filled into the corresponding message fields. All converted Protocol Buffers messages are arranged in chronological order and serialized into a continuous binary data stream to generate a behavioral data sequence.

[0104] S4.4: Behavioral data sequences include the user's visual gaze trajectory, the sequence of interactions with virtual objects, and the duration of the session.

[0105] It should be noted that the user's visual gaze trajectory refers to the sequence data of the three-dimensional coordinates and corresponding timestamps of continuous gaze points recorded by an eye tracker;

[0106] The sequence of interaction operations with virtual objects refers to the sequence data recorded by the controller of the user's operations such as grabbing, releasing, and clicking on virtual objects, and the corresponding target object identifiers;

[0107] Session duration refers to the total time from when a user begins to interact with the virtual reality consumption scenario to when the interaction ends, expressed as a timestamp difference in milliseconds.

[0108] S5. Input the behavioral data sequence into the decision interference analysis mechanism, extract multiple scenario factors from the behavioral data sequence, and identify and quantify the interference factors that cause the interruption of the user's decision-making process from the multiple scenario factors.

[0109] S5.1: Input the user's visual gaze trajectory, the sequence of interactions with virtual objects, and the duration of the session into the decision interference analysis mechanism to extract multiple scene factors and generate a set of scene factors;

[0110] Specifically, the user's visual gaze trajectory, the sequence of interactions with virtual objects, and the duration of the session are transmitted to the decision interference analysis mechanism. The decision interference analysis mechanism analyzes the spatial distribution characteristics of the visual gaze trajectory and extracts attention distraction factors, analyzes the continuity characteristics of the interaction sequence and extracts operation interruption factors, analyzes the deviation characteristics of the session duration from the expected duration and extracts decision delay factors, and combines the extracted attention distraction factors, operation interruption factors, and decision delay factors to form a set of scene factors.

[0111] S5.2: Use a causal discovery algorithm to mine the causal structure of the scene factor set and construct a causal network among the scene factors;

[0112] Specifically, the PC algorithm (Peter-Clark algorithm) is used to analyze the conditional independence relationship between scene factors in the scene factor set. Statistical tests are used to determine whether there is a dependency relationship between scene factors. When two scene factors remain correlated under the condition of other scene factors, a causal connection is established between them. According to the directional constraints of the causal connection, the causal relationship direction between scene factors is determined, forming a causal network structure composed of nodes and directional edges, thus constructing a causal network between scene factors.

[0113] S5.3: Utilize the causal network among scenario factors to perform interference propagation analysis, calculate the causal impact strength of each scenario factor on the interruption of the decision-making process, expressed as:

[0114] ;

[0115] In the formula, Indicates the first The causal influence strength value of each scenario factor; Represents the scene factor index; Indicates the first Regression coefficients of each scenario factor The absolute value; Represents the variable used for loop summation; This represents the total number of scene factors in the scene factor set; This indicates that in the process of summation, the first... Regression coefficients of each scenario factor The absolute value of.

[0116] It should be noted that the first Regression coefficients of each scenario factor This value is derived from the least squares estimate obtained through multiple linear regression analysis of user behavior data sequences. The example value is 0.5, based on the standardized coefficients obtained after fitting the user behavior data, reflecting the... The relative influence of each scenario factor.

[0117] S5.4: Based on the strength of causal influence, apply counterfactual reasoning methods to simulate potential interference scenarios and identify initial interference factors;

[0118] Specifically, based on the importance ranking of various scenario factors provided by the strength of causal influence, counterfactual reasoning is applied to simulate potential interference scenarios. Counterfactual reasoning is carried out by constructing assumptions. For scenario factors with high causal influence, it is assumed that they have not occurred or have weakened in intensity during actual user interaction, and the possible changes in the decision-making process under this assumption are deduced. By comparing the differences in decision-making process interruption indicators between the counterfactual scenarios and the actual recorded scenarios, the possibility of each scenario factor as a root cause of influence is assessed. When the absence of a certain scenario factor can improve the decision-making process, the scenario factor is identified as the initial interference factor that causes the interruption of the user's decision-making process.

[0119] Counterfactual reasoning example: Suppose that in a real-world scenario, a high value of the user's attention distraction factor leads to decision interruption. In counterfactual reasoning, we simulate a scenario with a low value of the attention distraction factor. If the simulation results show that the decision interruption index decreases, then the attention distraction factor is confirmed as the initial interference factor.

[0120] S5.5: Perform quantitative analysis of the impact of the initial interference factors to quantify the interference factors that cause the interruption of the user's decision-making process.

[0121] Specifically, the mean difference method under the counterfactual reasoning framework is used to quantify the impact of initial interference factors. The mean difference method compares the average value of the decision-making process interruption index in the actual observation scenario with the average value of the decision-making process interruption index after eliminating the initial interference factors in the counterfactual scenario. The difference between the two average values ​​is the quantified value of the impact of the initial interference factors. The larger the quantified value of the impact, the greater the contribution of the initial interference factors to the interruption of the user's decision-making process. Through quantitative analysis, the interference factors that cause the interruption of the user's decision-making process can be accurately identified from multiple initial interference factors.

[0122] S6. Generate dynamic optimization instructions based on interference factors, and adjust the information presentation density and interaction process complexity of the updated virtual reality consumption scene in real time according to the dynamic optimization instructions.

[0123] S6.1: Based on the interference factor, optimize the updated information density control parameters and interaction complexity parameters of the virtual reality consumption scenario, and generate dynamic optimization instructions;

[0124] Specifically, for the attention distraction factor, the upper limit of the number of virtual objects to be presented simultaneously and the text information density level in the information density control parameters are adjusted; for the operation interruption factor, the number of operation steps and the complexity of state transition conditions in the interaction complexity parameters are adjusted; based on the specific intensity value of the current interference factor, the adjustment direction and magnitude of the information density control parameters and interaction complexity parameters are determined by querying the preset fuzzy logic mapping table, and dynamic optimization instructions are generated.

[0125] It should be noted that the preset fuzzy logic mapping table is based on the assessment of the degree of influence of interference factors on decision interference in historical experimental data. The fuzzy logic mapping table establishes the correspondence between the intensity of interference factors and the adjustment range of information density control parameters and interaction complexity parameters.

[0126] S6.2: Input the dynamic optimization command into the adaptive control mechanism, determine the optimal control sequence through the model predictive control algorithm, drive the scene renderer to adjust the distribution of the detail level of virtual objects, and drive the interaction controller to adjust the state transition logic of the interaction process, thus completing the initial optimization and adjustment of the updated virtual reality consumption scene.

[0127] It should be noted that the model predictive control algorithm receives dynamic optimization instructions (including target information density and target interaction complexity), combines them with the real-time state of the updated virtual reality consumption scenario, predicts behavior over a future period, and provides the optimal sequence of control actions. This continuous optimization drives the scene renderer to precisely adjust the distribution of details of virtual objects (to regulate information presentation density) and simultaneously drives the interaction controller to adjust the state transition logic of the interaction process (to reduce the complexity of the interaction process). This ensures that the initial optimization process is not only dynamic but also forward-looking and optimizing, capable of coordinating multiple control variables (such as rendering details and interaction logic).

[0128] Specifically, the dynamic optimization instructions are passed to the adaptive control mechanism. The model predictive control algorithm in the adaptive control mechanism sets the target information density and target interaction complexity according to the dynamic optimization instructions. Combined with the current updated virtual reality consumption scene state, it determines the optimal control sequence for a period of time in the future. The optimal control sequence includes detail level adjustment instructions sent to the scene renderer and state logic simplification instructions sent to the interaction controller. The scene renderer adjusts the multi-resolution geometric representation distribution of virtual objects according to the detail level adjustment instructions, and the interaction controller adjusts the state transition conditions of the interaction process according to the state logic simplification instructions. Through these adjustments, the updated virtual reality consumption scene achieves initial optimization in terms of information presentation density and interaction process complexity.

[0129] S6.3: Based on the initial optimization and adjustments, establish a real-time feedback verification mechanism, collect user eye movement trajectory and interaction response data, calculate optimization effect indicators and dynamically correct control parameters for continuous closed-loop optimization.

[0130] Specifically, the real-time feedback verification mechanism continuously collects the spatial distribution pattern of gaze points in the user's eye movement trajectory and the operation result records in the interaction response data through a multimodal sensor array. Based on the comparison between the spatial distribution pattern of gaze points and the baseline gaze point distribution characteristics, it evaluates the improvement in attention stability; based on the comparison between the operation result records and the baseline operation success rate, it evaluates the improvement in operation efficiency; it combines the improvement in attention stability and operation efficiency with the gaze stability weighting coefficient and the operation efficiency weighting coefficient to form an optimization effect index; when the optimization effect index is lower than the expected target, it slightly corrects the information density control parameters and interaction complexity parameters according to the direction of the deviation in improvement. The corrected information density control parameters and interaction complexity parameters reguide the operation of the scene renderer and the interaction controller, realizing continuous closed-loop optimization of the updated virtual reality consumption scenario.

[0131] It should be noted that the expected targets refer to the specific quantitative indicators of the percentage increase in attention stability and the percentage increase in operational efficiency before the initial optimization and adjustment. The targets are set based on the assessment of the degree of influence of interference factors on decision-making.

[0132] Collect user eye-tracking trajectory and interaction response data, calculate the optimization effect index, and the expression is:

[0133] ;

[0134] In the formula, Indicates the optimization effect metrics; This represents the gaze stability weighting coefficient; This represents the current gaze distribution entropy; This represents the baseline gaze point distribution entropy; This represents the operational efficiency weighting coefficient; Indicates the current success rate of the operation; This indicates the baseline operation success rate.

[0135] It should be noted that the gaze stability weight coefficient is derived from the assessment of the importance of attention distraction factors to decision interference in historical experience and historical data. The example value is 0.6, and the basis for this value is to ensure that the improvement of attention stability occupies a dominant weight in the evaluation of optimization effect. The operational efficiency weight coefficient is derived from the assessment of the degree of influence of operational interruption factors on decision interference in historical experience and historical data. The example value is 0.4, and the basis for this value is to ensure that the improvement of operational efficiency occupies an appropriate but secondary weight in the evaluation of optimization effect.

[0136] This embodiment also provides a virtual reality-based consumer experience scenario simulation system, including: a demand input module, a 3D generation module, a scenario integration module, a data acquisition module, an interference analysis module, and an optimization and adjustment module; the demand input module is used to receive natural language demand descriptions input by users in a virtual reality consumer scenario via voice and text; the 3D generation module is used to construct a cross-modal generative artificial intelligence model, input the natural language demand descriptions into the cross-modal generative artificial intelligence model, interpret the demand description semantics of the natural language demand descriptions, and output a 3D data stream that matches the demand description semantics and conforms to the virtual reality consumer scenario; the scenario integration module is used to instantiate the 3D data stream. It is an interactive virtual object that is dynamically integrated and rendered into the virtual reality consumption scene to generate an updated virtual reality consumption scene; a data acquisition module is used to collect and record the behavioral data sequence generated when users interact with the updated virtual reality consumption scene; an interference analysis module is used to input the behavioral data sequence into the decision interference analysis mechanism, extract multiple scene factors from the behavioral data sequence, and identify and quantify the interference factors that cause the interruption of the user's decision-making process from the multiple scene factors; an optimization and adjustment module is used to generate dynamic optimization instructions based on the interference factors, and adjust the information presentation density and interaction process complexity of the updated virtual reality consumption scene in real time according to the dynamic optimization instructions.

[0137] This embodiment also provides a computer device applicable to the simulation method of consumer experience scenarios based on virtual reality, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the simulation method of consumer experience scenarios based on virtual reality as proposed in the above embodiment.

[0138] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0139] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the method for simulating a virtual reality-based consumer experience scenario as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0140] In summary, this invention utilizes a decision interference analysis mechanism to perform real-time analysis of user behavior data sequences, enabling precise identification of interference factors that disrupt decision-making. Based on multi-dimensional scenario factor extraction and causal inference algorithms, this mechanism can quantitatively assess the impact of each factor on the decision-making process and generate targeted dynamic optimization instructions. By adjusting the density of information presentation and the complexity of the interaction process in real time, it effectively reduces the user's cognitive load, ensures the continuity of the decision-making process, and enhances the user's immersion and decision-making efficiency in virtual consumption scenarios.

[0141] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for simulating consumer experience scenarios based on virtual reality, characterized in that: include, Receive natural language descriptions of user needs via voice and text input in virtual reality consumption scenarios; Construct a cross-modal generative artificial intelligence model, input natural language demand descriptions into the cross-modal generative artificial intelligence model, interpret the demand description semantics of natural language demand descriptions, and output a three-dimensional data stream that matches the demand description semantics and conforms to the virtual reality consumption scenario; The 3D data stream is instantiated into interactive virtual objects and dynamically integrated and rendered into the virtual reality consumption scene to generate an updated virtual reality consumption scene. Collect and record the sequence of behavioral data generated when users interact with the updated virtual reality consumption scenario; The behavioral data sequence is input into the decision interference analysis mechanism to extract multiple scenario factors from the behavioral data sequence, and to identify and quantify the interference factors that cause the interruption of the user's decision-making process from the multiple scenario factors. Based on the interference factor, dynamic optimization instructions are generated, and the information presentation density and interaction process complexity of the updated virtual reality consumption scenario are adjusted in real time according to the dynamic optimization instructions. The behavioral data sequence includes the user's visual gaze trajectory, the sequence of interactions with virtual objects, and the duration of the session. The specific steps for identifying and quantifying the interference factors that cause interruptions in the user's decision-making process from multiple scenario factors are as follows. The user's visual gaze trajectory, the sequence of interactions with virtual objects, and the duration of the session are input into the decision interference analysis mechanism to extract multiple scene factors and generate a set of scene factors. A causal discovery algorithm is used to mine the causal structure of the scene factor set and construct a causal network among the scene factors. Interference propagation analysis is performed using causal networks among scenario factors to calculate the causal impact strength of each scenario factor on the interruption of the decision-making process. Based on the strength of causal influence, counterfactual reasoning methods are applied to simulate potential interference scenarios and identify initial interference factors. The impact of the initial interference factors is quantitatively analyzed to identify the interference factors that cause the interruption of the user's decision-making process.

2. The virtual reality-based consumer experience scenario simulation method as described in claim 1, characterized in that: The natural language requirement description includes the attribute description, function description, and relationship description of the virtual object and the virtual reality consumption scenario.

3. The virtual reality-based consumer experience scenario simulation method as described in claim 1, characterized in that: The specific steps for constructing a cross-modal generative artificial intelligence model are as follows. A semantic planning layer is built using a multimodal alignment algorithm, and a three-dimensional execution layer is built using a conditional diffusion mechanism. By integrating the semantic planning layer and the 3D execution layer through an iterative optimization mechanism, a cross-modal generative artificial intelligence model is constructed.

4. The virtual reality-based consumer experience scenario simulation method as described in claim 1, characterized in that: The output is a 3D data stream that semantically matches the demand description and conforms to the virtual reality consumption scenario. The specific steps are as follows: Semantic parsing is performed on natural language requirement descriptions to extract attribute descriptions, functional descriptions, and relational descriptions, generating structured semantic feature vectors; Multi-dimensional perception analysis is performed on virtual reality consumption scenarios to extract spatial scale features and lighting style features, and to generate scene environment feature vectors. The structured semantic feature vector and the scene environment feature vector are fused at the feature level to generate a scene condition vector; The scene condition vector is iteratively denoised using a conditional diffusion mechanism to generate preliminary 3D geometric data. The initial 3D geometric data is validated by differentiable rendering, and the output is a 3D data stream that semantically matches the requirements description and conforms to the virtual reality consumption scenario.

5. The virtual reality-based consumer experience scenario simulation method as described in claim 1, characterized in that: The specific steps for generating the updated virtual reality consumption scenario are as follows: Physically instantiate the 3D data stream to generate interactive virtual objects; Interactive virtual objects are integrated into virtual reality consumption scenarios through potential energy field optimization algorithms to determine the optimal position and posture. By solving the rendering equation, pixel color values ​​are calculated and materials are fused to determine the optimal position and posture of the virtual object, generating a lighting-blended virtual object that is coordinated with the lighting environment of the virtual reality consumer scene. Collect the viewpoint changes generated by the user during the interaction, and dynamically schedule the lighting blending virtual object at the level of detail based on the viewpoint changes to generate virtual objects with optimized level of detail. Perform semantic consistency verification between the virtual objects optimized at the level of detail and the virtual reality consumption scene, and generate an updated virtual reality consumption scene.

6. The virtual reality-based consumer experience scenario simulation method as described in claim 1, characterized in that: The specific steps for collecting and recording the behavioral data sequences generated when users interact with the updated virtual reality consumption scenario are as follows. By synchronously collecting user behavior data in virtual reality consumption scenarios through a multimodal sensor array, a timestamp-aligned behavior data stream is established; Topological feature extraction and entropy change analysis are performed on behavioral data streams to generate structured behavioral event sequences; The structured behavioral event sequence is encoded and encapsulated to generate a behavioral data sequence.

7. The virtual reality-based consumer experience scenario simulation method as described in claim 1, characterized in that: The process of generating dynamic optimization instructions based on interference factors, and adjusting the information presentation density and interaction complexity of the updated virtual reality consumption scene in real time according to these instructions, involves the following specific steps: Based on the interference factor, optimize the updated information density control parameters and interaction complexity parameters of the virtual reality consumption scenario, and generate dynamic optimization instructions; The dynamic optimization command is input into the adaptive control mechanism, and the optimal control sequence is determined through the model predictive control algorithm. This drives the scene renderer to adjust the distribution of the detail level of virtual objects, and at the same time drives the interaction controller to adjust the state transition logic of the interaction process, thus completing the initial optimization and adjustment of the updated virtual reality consumption scenario. Based on the initial optimization and adjustments, a real-time feedback verification mechanism was established to collect user eye movement trajectory and interaction response data, calculate optimization effect indicators, dynamically correct control parameters, and continuously optimize in a closed loop.

8. A virtual reality-based consumer experience scenario simulation system, based on the virtual reality-based consumer experience scenario simulation method according to any one of claims 1 to 7, characterized in that: It includes a requirements input module, a 3D generation module, a scene integration module, a data acquisition module, an interference analysis module, and an optimization and adjustment module; The demand input module is used to receive natural language demand descriptions input by users through voice and text in virtual reality consumption scenarios. The 3D generation module is used to construct a cross-modal generative artificial intelligence model. It inputs the natural language demand description into the cross-modal generative artificial intelligence model, interprets the demand description semantics of the natural language demand description, and outputs a 3D data stream that matches the demand description semantics and conforms to the virtual reality consumption scenario. The scene integration module is used to instantiate the three-dimensional data stream into interactive virtual objects and dynamically integrate and render them into the virtual reality consumption scene to generate an updated virtual reality consumption scene. The data acquisition module is used to collect and record the behavioral data sequence generated when the user interacts with the updated virtual reality consumption scenario; The interference analysis module is used to input the behavioral data sequence into the decision interference analysis mechanism, extract multiple scenario factors from the behavioral data sequence, and identify and quantify the interference factors that cause the interruption of the user's decision-making process from the multiple scenario factors. The optimization and adjustment module is used to generate dynamic optimization instructions based on interference factors, and adjust the information presentation density and interaction process complexity of the updated virtual reality consumption scene in real time according to the dynamic optimization instructions.

Images