AR intelligent shopping guide interaction method and system applied to store management

Abstract

The invention provides an AR intelligent shopping guide interaction method and system applied to store management, and the method comprises the steps: firstly obtaining an interaction data set of a user in a store, covering a user position movement track, a commodity attention action sequence and voice consultation content, carrying out the user intention analysis of the interaction data set, and carrying out the user intention analysis of the interaction data set; the method comprises the steps of obtaining commodity demand characteristics and interaction preference characteristics, generating AR shopping guide interaction content, including commodity three-dimensional display information and an interaction guide instruction, adaptive to a user based on the commodity demand characteristics and the interaction preference characteristics, aligning and fusing the AR shopping guide interaction content with a store physical scene space, generating an AR shopping guide interaction scene capable of being displayed in real time, and displaying the AR shopping guide interaction scene in real time. And finally, responding to user interaction operation in the AR shopping guide interaction scene, generating feedback information and updating the interaction data set to trigger interactive process iterative optimization, thereby providing personalized and immersive shopping guide experience, and improving store operation efficiency and customer satisfaction.

Description

Technical Field

[0001] The present invention relates to the field of VR technology. Specifically, it relates to an AR intelligent shopping guide interaction method and system applied to store management. Background Art

[0002] In the retail industry, stores, as important places for commodity sales and customer service, their operational efficiency and customer experience directly affect the market competitiveness of enterprises. The traditional store shopping guide method mainly relies on manual services. Shopping guide personnel need to provide commodity information and recommendations to customers based on their own experience and knowledge. This method has obvious limitations. On the one hand, the service quality of manual shopping guides is affected by factors such as the personal ability and emotional state of shopping guide personnel, and it is difficult to ensure consistent and high-quality services for every customer. On the other hand, with the increasing richness of store commodity types and the diversification of customer needs, it is difficult for manual shopping guides to quickly and accurately understand the commodity needs and preferences of customers and cannot provide personalized shopping suggestions for customers.

[0003] In recent years, although some stores have introduced electronic shopping guide systems, most of these electronic shopping guide systems only provide simple commodity information query functions, lack in-depth analysis of customers' actual interaction behaviors in the store, and cannot dynamically adjust shopping guide content according to customers' real-time behaviors. At the same time, the integration degree of existing shopping guide systems with the physical store scenario is relatively low, and it is difficult for customers to obtain an immersive shopping experience, and the purchase desire of customers cannot be fully stimulated. Summary of the Invention

[0004] In view of the above-mentioned problems, in combination with the first aspect of the present invention, embodiments of the present invention provide an AR intelligent shopping guide interaction method applied to store management. The method includes: Obtaining an interaction data set of a user in a store, where the interaction data set includes a user location movement trajectory, a commodity attention action sequence, and voice consultation content; Performing user intention parsing processing on the interaction data set to obtain the commodity demand characteristics and interaction preference characteristics of the user; Generating AR shopping guide interaction content adapted to the user based on the commodity demand characteristics and interaction preference characteristics, where the AR shopping guide interaction content includes three-dimensional display information of commodities and interaction guiding instructions; Performing spatial alignment and fusion processing on the AR shopping guide interaction content and the physical store scenario to generate an AR shopping guide interaction scenario that can be displayed in real time; Responding to a user interaction operation in the AR shopping guide interaction scenario, generating interaction feedback information and updating the interaction data set to trigger iterative optimization of the interaction process.

[0005] In another aspect, an AR intelligent shopping guide interaction system for store management provided by an embodiment of the present invention includes a processor and a machine-readable storage medium. The machine-readable storage medium is connected to the processor. The machine-readable storage medium is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the machine-readable storage medium to implement the above method.

[0006] Based on the above aspects, in the embodiment of the present invention, by obtaining the interaction data set of the user in the store, including the user's position movement trajectory, the sequence of product attention actions and the voice consultation content, the behavior information of the user in the store is comprehensively captured, providing a rich data basis for subsequent user intention analysis. By performing user intention analysis processing on the interaction data set, the product demand characteristics and interaction preference characteristics of the user are obtained, and the shopping intention and personalized needs of the user can be accurately grasped. The AR shopping guide interaction content generated based on the product demand characteristics and interaction preference characteristics includes three-dimensional product display information and interaction guidance instructions, which not only intuitively displays the appearance and features of the product, but also provides clear interaction guidance for the user, enhancing the user's shopping experience. By performing spatial alignment and fusion processing on the AR shopping guide interaction content and the physical store scene, an AR shopping guide interaction scene that can be displayed in real time is generated, realizing the seamless combination of virtual shopping guide information and the real store environment, and providing an immersive shopping environment for the user. Finally, in response to the user interaction operation in the AR shopping guide interaction scene, interaction feedback information is generated and the interaction data set is updated to trigger the iterative optimization of the interaction process, so that the entire shopping guide interaction process can be continuously adjusted and optimized according to the real-time feedback of the user, improving the accuracy and effectiveness of the shopping guide, and thus enhancing the operation efficiency of the store and customer satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0007] Figure 1 is a schematic execution flow diagram of an AR intelligent shopping guide interaction method for store management provided by an embodiment of the present invention.

[0008] Figure 2 is a schematic diagram of exemplary hardware and software components of an AR intelligent shopping guide interaction system for store management provided by an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0009] The present invention will be specifically described below with reference to the accompanying drawings of the specification. Figure 1 is a schematic flow diagram of an AR intelligent shopping guide interaction method for store management provided by an embodiment of the present invention. The AR intelligent shopping guide interaction method for store management will be introduced in detail below.

[0010] Step S110: Obtain an interaction data set of the user in the store, where the interaction data set includes the user's position movement trajectory, the sequence of product attention actions, and the voice consultation content.

[0011] Before obtaining the user's interaction data set in the store, it is necessary to strictly comply with relevant laws and regulations and obtain various authorization permissions from the user. For example, at the store entrance, in prominent positions within the store, and in relevant mobile applications or mini-programs, clearly and understandably explain to the user the purpose, scope, and use of data collection. For example, inform the user that collecting their location movement trajectory is for better personalized product recommendations and navigation services; collecting the sequence of product attention actions is to understand the user's product interest preferences in order to provide more accurate shopping guide suggestions; collecting voice consultation content is to promptly respond to the user's questions and needs and improve service quality. At the same time, clearly state that the collected data will be strictly confidential and will not be used for other purposes without the user's consent.

[0012] Moreover, provide the user with multiple authorization options to facilitate the user to authorize according to their own wishes. Common authorization methods include: Online authorization: Before the user enters the store, if they use the store's mobile application or mini-program, they can set authorization options in the application or mini-program. The user can choose full authorization, partial authorization, or refusal to authorize. For example, the user can choose to only authorize the collection of the location movement trajectory and the sequence of product attention actions, while refusing to authorize the collection of voice consultation content.

[0013] Offline authorization: Set up an authorization terminal device at the store entrance. The user can perform authorization operations by touching the screen or scanning a QR code. Detailed authorization instructions and options will be displayed on the authorization terminal device, and the user can select according to their own needs. At the same time, store staff will also be on-site to provide assistance and explanations to ensure that the user clearly understands the content and meaning of the authorization.

[0014] After the user completes the authorization operation, the authorization status is fed back to the user in real time. If the user chooses full authorization, prompt the user "You have successfully authorized the collection of all interaction data, and we will provide you with more personalized services"; if the user chooses partial authorization, inform the user "You have authorized the collection of part of the interaction data, and we will provide corresponding services according to your authorization"; if the user refuses to authorize, prompt "You have refused the data collection authorization, which may affect the effect of the personalized services we provide for you, but you can still browse and shop normally." At the same time, the user can modify the authorization status at any time through online or offline methods.

[0015] During the data collection process, a strict supervision and management mechanism shall be established to ensure that the data collection activities are carried out strictly in accordance with the user's authorization. The system will monitor the collected data in real time to check whether the data source and type are consistent with the user's authorization. If any unauthorized situation is found, the data collection can be stopped immediately and an alarm shall be sent to the relevant management personnel. At the same time, the data collection activities shall be audited and evaluated regularly to ensure the compliance and security of data collection.

[0016] After obtaining the user's authorization permission, the data collection and subsequent processing can be carried out according to the steps described below. For example, when obtaining the user's location movement trajectory, the positioning sensor will start working on the premise of the user's authorization and record the user's location information within the store; when collecting the sequence of product attention actions and voice consultation content, the camera, motion sensor and microphone array will also collect data within the scope of the user's authorization. Through a strict authorization permission process, the privacy rights and interests of users are protected, and at the same time, the legality and compliance of data collection activities are ensured.

[0017] Even if the user's authorization is obtained, during the entire data collection and processing process, technical means such as encryption processing, anonymization processing, access control and security auditing shall still be adopted to ensure that the user's privacy-sensitive data is fully protected. For example, the authorized data collected is encrypted for storage and transmission to prevent the data from being stolen during the transmission process; during the data processing process, the user's identity information is anonymized to avoid the user's identity being identified due to data leakage. Through continuous privacy protection measures, users can participate in the AR intelligent shopping guide interaction activities with confidence.

[0018] In this embodiment, to achieve the AR intelligent shopping guide interaction applied to store management, first, it is necessary to obtain the set of interaction data of the user within the store. For the acquisition of the user's location movement trajectory, it is completed by means of positioning sensors deployed at different positions in the store. The positioning sensors adopt advanced positioning technologies, such as the technology that combines satellite positioning and indoor positioning. Satellite positioning can provide relatively accurate initial location information for users in areas near the doors and windows of the store, while the indoor positioning technology uses devices such as Bluetooth beacons and Wi-Fi access points distributed in the store to accurately determine the user's location indoors by analyzing the signal strength. When the user enters the store, the positioning sensor will keep working and record the user's location information at regular time intervals. For example, the sensor records the user's position coordinates (x, y, z) in the three-dimensional space at each relatively short time interval. As time goes by, a series of coordinate points 、 …… will form the user's location movement trajectory.

[0019] The collection of the product attention action sequence relies on multi-modal sensors installed in the product display area. Among them, the camera has high-definition shooting and motion capture functions, and can clearly record the user's hand movements, body postures, etc.; the pressure sensor is installed on the product display shelf and can detect the pressure changes generated when the product is picked up or put back; the gyroscope is used to detect the rotation angle and direction of the product. When the user operates in the product display area, the camera will capture the user's action pictures in real time, and the pressure sensor and gyroscope will synchronously record the state changes of the product. The system will sort and record the user's actions in chronological order according to the data of these sensors to form a product attention action sequence. For example, if the user first gently touches the product, then picks up the product and turns it to observe, and finally puts the product back to its original place, actions such as "touch the product", "pick up the product", "turn the product", and "put back the product" can be recorded in sequence, and the time when each action occurs can be marked.

[0020] The acquisition of voice consultation content depends on the microphone arrays distributed in various corners of the store. These microphone arrays adopt advanced noise reduction and beamforming technologies, and can accurately capture the user's voice signals in the noisy store environment. When the user issues a voice consultation, the microphone array will perform preliminary processing on the received voice signals, such as amplification, filtering, etc., to enhance the clarity of the voice signals. Then, the processed voice signals are transmitted to the backend voice processing server through wireless communication technology. The server uses a deep neural network model to identify and analyze the voice signals and convert them into text content. For example, when the user asks about the performance parameters and price of a certain electronic product, after the microphone array collects the voice signals, through the processing and recognition of the server, it is converted into the text "What are the performance parameters and price of this electronic product", so as to obtain the user's voice consultation content.

[0021] Step S120: Perform user intention parsing processing on the interaction data set to obtain the product demand characteristics and interaction preference characteristics of the user.

[0022] In this embodiment, after obtaining the interaction data set of the user in the store, it is necessary to perform in-depth analysis and processing on it to parse out the product demand characteristics and interaction preference characteristics of the user. Through the mining and correlation analysis of different types of interaction data, the needs and preferences of the user can be understood more accurately.

[0023] Step S121: Perform stay area recognition processing on the user's location movement trajectory, and extract the stay duration parameter and area coverage range of the user in the product display area.

[0024] The purpose of this step is to find out the stay situation of the user in the product display area through the analysis of the user's location movement trajectory, so as to extract relevant key parameters.

[0025] Step S1211: Split the user's location movement trajectory into a sequence of consecutive location points in timestamp order. The sequence of location points includes the store space coordinates corresponding to each timestamp.

[0026] After obtaining the user's location movement trajectory data, check the original location data, remove invalid or incorrect data points, and then sort the remaining data points in ascending order of timestamps. Each location point contains the corresponding timestamp and the store space coordinates (x, y, z). The timestamp is used to mark the time when the location point is recorded, and the coordinates (x, y, z) represent the specific location of the user in the three-dimensional space of the store.

[0027] Step S1212: Calculate the coordinate change amount between adjacent location points, and identify the continuous location point interval with a coordinate change amount less than a preset threshold as the candidate interval for the stay area.

[0028] After obtaining the sequence of consecutive location points, it is necessary to further analyze the user's movement state. By calculating the coordinate change amount between adjacent location points, it can be determined whether the user is in a moving state or a staying state. For two adjacent location points and , calculate the coordinate differences in the three dimensions of x, y, and z respectively , , . Then, compare the absolute values of these differences with the preset threshold. The preset threshold is set according to the actual situation of the store and the normal movement speed of the user, and it represents the minimum range of position change of the user in a short period of time. If the absolute value of the coordinate change amount between adjacent location points is less than the preset threshold in all three dimensions, it is considered that the user is in a staying state between these two location points. The system will continuously check the consecutive location points to find the continuous location point interval where the coordinate change amount is less than the preset threshold, and identify it as the candidate interval for the stay area. For example, in a sequence of consecutive location points, from the i-th location point to the j-th location point, the coordinate change amount between adjacent location points is less than the preset threshold, then the time period and spatial range corresponding to the i-th location point to the j-th location point constitute a candidate interval for the stay area.

[0029] Step S1213: Perform boundary fitting on the coordinates of all location points within the candidate interval for the stay area to generate a region coverage descriptor that includes the minimum bounding rectangle.

[0030] After determining the candidate interval of the stay area, it is necessary to further process the position points within this candidate interval of the stay area to determine the range of the user's stay area. Using the method of boundary fitting, by analyzing and calculating the coordinates of all position points within the candidate interval of the stay area, the smallest circumscribed rectangle that can enclose these position points is found. The specific boundary fitting algorithm will consider the distribution of the position points and continuously adjust the boundaries of the rectangle so that the area of the rectangle is the smallest and it can completely contain all the position points. For example, for a set of position points located on a plane, the algorithm will first find the maximum and minimum values of these position points on the x-axis and y-axis, and then determine the boundaries of the rectangle based on these extreme values. The four vertex coordinates of the finally generated smallest circumscribed rectangle constitute the region coverage range descriptor, which can accurately describe the range of the user's stay area.

[0031] Step S1214: Calculate the difference between the start timestamp and the end timestamp of the candidate interval of the stay area to obtain the stay duration parameter of the user in this area.

[0032] To obtain the stay duration of the user in the stay area, the start timestamp and the end timestamp of each candidate interval of the stay area can be recorded. The start timestamp is the time mark of the first position point within this interval, and the end timestamp is the time mark of the last position point within this interval. By calculating the difference between the end timestamp and the start timestamp, the stay duration parameter of the user in this area can be obtained. For example, if the start timestamp of a candidate interval of the stay area is , and the end timestamp is , then the stay duration parameter of the user in this area is . This stay duration parameter can reflect the degree of attention and the level of interest of the user in this area.

[0033] Step S1215: Screen the intervals in the candidate intervals of the stay area that cover the spatial positioning marker points of the commodity display area, and extract the corresponding stay duration parameter and the region coverage range descriptor as the stay duration parameter and the region coverage range of the user in the commodity display area.

[0034] Since the areas where users stay in the store may include different areas such as product display areas, rest areas, aisles, etc., in order to accurately extract the stay information of users in the product display area, it is necessary to screen the candidate intervals of the stay areas. Spatial positioning marker points are pre-set in the product display area of the store, and the coordinate information of these marker points is stored in the database of the system. The system will compare the area coverage descriptors of each candidate interval of the stay area with the spatial positioning marker points in the product display area. If the area coverage of a certain candidate interval of the stay area contains the spatial positioning marker points in the product display area, then this interval is considered as the stay area of the user in the product display area. The system will extract the corresponding stay duration parameter and area coverage descriptor of this interval as the stay duration parameter and area coverage of the user in the product display area. For example, after comparison, it is found that the minimum bounding rectangle of a certain candidate interval of the stay area covers a spatial positioning marker point in the product display area, then the stay duration parameter and area coverage descriptor of this interval are used as the effective stay information of the user in this product display area.

[0035] Step S122: Perform action pattern analysis and processing on the product attention action sequence to identify the picking frequency of the product by the user, the operation of adjusting the display angle, and the characteristics of the putting-back position.

[0036] After obtaining the product attention action sequence, it is necessary to perform action pattern analysis and processing on it to discover the operation characteristics of the user on the product. For the identification of the picking frequency, the "pick up the product" actions in the product attention action sequence can be counted. Whenever the system detects the action of the user picking up the product, the counter will be incremented by 1. For example, within a certain period of time, the system records that the user picks up the same product multiple times, then the picking frequency will increase accordingly. By analyzing the picking frequency, the degree of interest of the user in the product can be understood. The higher the picking frequency, the higher the possible attention of the user to the product.

[0037] The identification of the display angle adjustment operation is carried out by analyzing the user action pictures recorded by the camera and the product rotation data detected by the gyroscope. The camera can capture the actions of the user's hand, such as the action posture of rotating the product; the gyroscope can accurately measure the rotation angle and direction of the product. The system will judge whether the user has performed the operation of adjusting the display angle of the product based on this data, and record the adjusted angle and direction. For example, after the user picks up the product and rotates the product clockwise by a set angle, the size of the rotation angle and the clockwise direction information can be recorded.

[0038] The recognition of the put-back position feature mainly relies on the data from pressure sensors and position sensors. The pressure sensor can detect the pressure change generated when the commodity is put back on the display shelf, thereby determining whether the commodity is put back; the position sensor can detect the specific position coordinates where the commodity is put back. The system will analyze the relationship between the position where the commodity is put back and the pick-up position, as well as the specific position of the commodity on the display shelf based on this data. For example, when the user puts back the commodity, the position coordinates where the commodity is put back can be recorded and compared with the pick-up position coordinates to determine whether the user has put the commodity back to its original position or placed it in an adjacent position, etc.

[0039] Step S123: Perform semantic understanding processing on the voice consultation content, and extract the commodity attribute keywords, usage scenario descriptions, and price sensitivity expressions consulted by the user.

[0040] For the processing of voice consultation content, semantic understanding must be carried out first. Using natural language processing technology, operations such as word segmentation, part-of-speech tagging, and syntactic analysis are performed on the text content after voice conversion. By analyzing the word segmentation results, keywords related to commodity attributes are found. For example, in the user's voice consultation content, it is mentioned that "What are the screen size and processor performance of this mobile phone", and "screen size" and "processor performance" can be identified as commodity attribute keywords.

[0041] The extraction of the usage scenario description finds relevant statements in which the user describes the usage scenario of the commodity through in-depth semantic analysis. For example, when the user says "I want to use this camera during outdoor travel. How about its battery life", "outdoor travel" can be extracted as the usage scenario description.

[0042] The recognition of the price sensitivity expression focuses on the user's attitude and expression towards price in the voice consultation content. For example, when the user says "If the price of this commodity is appropriate, I will buy it", the system can judge that the user has a certain sensitivity to price and record the relevant information. By extracting and analyzing this information, the user's needs and concerns about the commodity can be understood more accurately.

[0043] Step S124: Input the residence duration parameter, area coverage, pick-up frequency, display angle adjustment operation, put-back position feature, commodity attribute keywords, usage scenario description, and price sensitivity expression into the intention correlation analysis module, and generate the user's commodity demand feature through context correlation rule matching processing. The commodity demand feature includes the target commodity type, core attribute preference, and usage scenario adaptation requirements.

[0044] For example, if a user stays in a certain type of product display area for a long time and has a high frequency of picking up products, and specific attributes and usage scenarios of this type of product are mentioned in the voice consultation, it can be determined that the user has a high demand for this type of product, and thus the user's product demand characteristics are generated. The target product type is determined according to the category to which the product concerned by the user belongs; the core attribute preference is judged according to the product attribute keywords consulted by the user and the operation behaviors on the product; the usage scenario adaptation requirement is determined according to the user's usage scenario description. For example, if the user stays in the electronics area for a long time, picks up a certain tablet computer many times, and the voice consultation mentions the screen resolution and battery life of the tablet computer, as well as the usage requirements in the office scenario, then the system will determine that the target product type is the tablet computer, the core attribute preference is the screen resolution and battery life, and the usage scenario adaptation requirement is the office scenario.

[0045] Step S125: Based on the path smoothness of the user location movement trajectory, the operation coherence of the product attention action sequence, and the expression clarity of the voice consultation content, through interactive behavior pattern recognition processing, generate the user's interactive preference characteristics, where the interactive preference characteristics include the information presentation rhythm requirement, the acceptance degree of interactive operation complexity, and the voice feedback response time requirement.

[0046] For the generation of the user's interactive preference characteristics, first analyze the path smoothness of the user location movement trajectory. The path smoothness can be evaluated by calculating the curvature and speed change of the user location movement trajectory. If the user's movement trajectory is relatively smooth and the speed change is small, it indicates a high path smoothness; otherwise, the path smoothness is low. For example, if the user moves along a relatively straight route in the store and the speed is relatively stable, then the path smoothness is high.

[0047] The operation coherence of the product attention action sequence is judged by analyzing the sequence and time interval of the user's actions. If the connection between the user's actions is close and the time interval is short, it indicates good operation coherence; otherwise, the operation coherence is poor. For example, after picking up a product, the user quickly observes and adjusts the angle, and then quickly puts the product back. The whole process is action - coherent and the time interval is short.

[0048] The expression clarity of the voice consultation content is evaluated by the accuracy of the voice recognition result and the clarity of the semantics. If the accuracy of the voice recognition is high, the semantic expression is clear and there is no ambiguity, it indicates a high expression clarity; otherwise, the expression clarity is low. For example, if the user clearly expresses the specific requirements for a certain product and the voice recognition result is accurate, then the expression clarity is high.

[0049] Using an interactive behavior pattern recognition algorithm, generate the interactive preference features of the user. The information presentation rhythm requirement is determined according to the path fluency and operation coherence. If the path fluency is high and the operation coherence is good, the user may be more inclined to a faster information presentation rhythm; otherwise, a slower information presentation rhythm may be required. The acceptance of the complexity of interactive operations is judged based on the user's operation behavior and the content of voice consultations. If the user can proficiently perform complex operations and shows a high degree of attention to the technical details of the product in the voice consultation, it indicates that the user has a high acceptance of the complexity of interactive operations; otherwise, the acceptance is low. The requirement for the timeliness of voice feedback response is determined according to the user's operation speed and the urgency of the voice consultation. If the user's operation speed is fast and the tone of the voice consultation is relatively urgent, it indicates that the user has a high requirement for the timeliness of voice feedback response; otherwise, the requirement is low.

[0050] Step S130: Generate AR shopping guide interactive content adapted to the user based on the product demand characteristics and interactive preference features, where the AR shopping guide interactive content includes three-dimensional display information of the product and interactive guidance instructions.

[0051] In this embodiment, after obtaining the product demand characteristics and interactive preference features of the user, it is necessary to generate AR shopping guide interactive content adapted to the user based on these features.

[0052] Step S131: According to the target product type and core attribute preferences in the product demand characteristics, retrieve the three-dimensional model data and attribute description information of the corresponding product from the store product database.

[0053] The store product database stores detailed information of all products, including three-dimensional model data and attribute description information. According to the target product type in the product demand characteristics, searches and filters can be performed in the database to find the corresponding product records. Then, according to the core attribute preferences, further determine the product attribute description information to be retrieved. For example, if the target product type is a sofa and the core attribute preferences are material and size, the three-dimensional model data of this sofa and the detailed description information about the material and size of the sofa can be retrieved from the database.

[0054] Step S132: Based on the usage scenario adaptation requirements in the product demand characteristics, perform scene rendering processing on the three-dimensional model data to generate virtual placement effect display information of the product in the user-described usage scenario.

[0055] To enable the user to better understand the effect of the product in the actual usage scenario, it is necessary to perform scene rendering processing on the three-dimensional model data.

[0056] Step S1321: Analyze the usage scenario description in the commodity demand characteristics, and extract the key scenario elements, where the key scenario elements include the scenario space size, lighting conditions, and the layout of existing furniture.

[0057] Conduct an in-depth analysis of the usage scenario description in the commodity demand characteristics, and extract the key scenario elements through natural language processing technology. For example, if the user describes the usage scenario as "used in a living room, the living room has a large area, and there is window lighting", the scenario space size can be extracted as the large living room area, the lighting condition as window lighting, and the layout of existing furniture needs to be further determined according to the actual living room situation.

[0058] Step S1322: Retrieve the virtual scenario template that matches the key scenario elements from the store scenario database, where the virtual scenario template includes space size parameters, lighting intensity distribution, and furniture model position coordinates.

[0059] The store scenario database stores a variety of different virtual scenario templates, and each template contains detailed information about the scenario. According to the extracted key scenario elements, the most similar virtual scenario template can be searched and matched in the database. For example, based on the general situation of the scenario space size, lighting conditions, and the layout of existing furniture, a living room scenario template is found, which includes the space size parameters of the living room, the distribution of lighting intensity in different areas, and the position coordinates of the furniture model.

[0060] Step S133: Retrieve the 3D model data and attribute description information of the corresponding commodity from the store commodity database according to the target commodity type and core attribute preferences in the commodity demand characteristics.

[0061] In this embodiment, the commodity entries in the store commodity database include the 3D model data and attribute description information of the commodity. The 3D model data is a virtual model of the commodity created by 3D modeling software, which can display the appearance and structure of the commodity from different angles and perspectives. The attribute description information includes various attributes of the commodity, such as color, material, size, function, etc. For example, if the target commodity type is a sofa and the core attribute preferences are leather, adjustable angle, and a specific size range, the sofa commodity entries that meet these conditions can be searched in the store commodity database. For the 3D model data, it may be stored in different file formats, such as OBJ, FBX, etc., and contains the geometric shape, texture, and material information of the commodity. By retrieving these 3D model data, the real appearance of the commodity can be presented to the user in subsequent processing. The attribute description information will be stored in the database in text form, describing in detail the characteristics of each attribute of the commodity.

[0062] Step S134: Based on the adaptation requirements of the usage scenarios in the commodity demand characteristics, perform scene rendering processing on the 3D model data to generate virtual placement effect display information of the commodity in the user-described usage scenario.

[0063] Step S1341: Analyze the usage scenario description in the commodity demand characteristics, and extract the key scene elements. The key scene elements include scene space size, lighting conditions, and the layout of existing furniture.

[0064] After obtaining the usage scenario description in the commodity demand characteristics, use natural language processing technology to deeply analyze it. First, perform word segmentation on the description statement to split the sentence into individual words. Then, through part-of-speech tagging and semantic analysis, identify the words related to the key scene elements. For the extraction of the scene space size, attention can be paid to the vocabulary related to size, area, volume, etc. in the description to infer the approximate range of the space size. The extraction of lighting conditions will focus on the words related to light and illumination to determine the intensity, type, and source of the light. The extraction of the layout of existing furniture requires analyzing the types and positional relationships of the furniture mentioned in the description to determine the types and relative positions of the furniture.

[0065] Step S1342: Retrieve a virtual scene template that matches the key scene elements from the store scene database. The virtual scene template includes space size parameters, lighting intensity distribution, and furniture model position coordinates.

[0066] The store scene database stores various types and specifications of virtual scene templates, and each template has its corresponding space size parameters, lighting intensity distribution, and furniture model position coordinates. After obtaining the key scene elements, they can be matched with the templates in the database. For the space size parameters, the extracted range of the scene space size can be compared with the space size in the template to find the closest template. The matching of the lighting intensity distribution will select a template with a lighting intensity and distribution that match the extracted lighting conditions. For example, if the extracted lighting condition is bright natural light, a template with strong natural light simulation and reasonable lighting distribution can be selected. For the furniture model position coordinates, a template with similar types and positional relationships of furniture can be found according to the description of the layout of existing furniture.

[0067] Step S1343: Import the 3D model data into the virtual scene template, and adjust the scale factor of the commodity 3D model according to the scene space size parameters to make the size of the commodity 3D model match the scene space size.

[0068] After retrieving a suitable virtual scene template, import the three-dimensional model data of the products previously obtained from the store product database into the template. To place the three-dimensional product models reasonably in the scene, it is necessary to adjust the scale factor of the three-dimensional product models according to the scene space size parameters. First, analyze the space range defined by the scene space size parameters to determine the specific numerical ranges of its length, width, and height. Then, measure the dimensions of the three-dimensional product models in each dimension. By comparing the dimensions of the three-dimensional product models with the scene space size, calculate the scale factor that needs to be adjusted. For example, if the scene space is relatively small and the dimensions of the three-dimensional product models are large, it is necessary to reduce the scale of the three-dimensional product models; conversely, if the scene space is large and the dimensions of the three-dimensional product models are relatively small, the scale can be appropriately enlarged. Through the above adjustments, ensure that the dimensions of the three-dimensional product models match the scene space size, so that the products present a natural and reasonable placement effect in the virtual scene.

[0069] Step S1344: Adjust the material reflectivity and diffuse color parameters of the three-dimensional product model according to the illumination condition to generate a rendering effect that conforms to the scene illumination intensity distribution.

[0070] After completing the adjustment of the scale of the three-dimensional product models, it is necessary to adjust the material reflectivity and diffuse color parameters of the three-dimensional product models according to the illumination condition of the scene. Different illumination conditions will have different effects on the appearance of the products. To make the products present a realistic illumination effect in the virtual scene, it is necessary to optimize the material parameters. For a scene with strong illumination intensity, appropriately increase the material reflectivity of the three-dimensional product models so that the product surfaces can reflect more light and present a bright luster; at the same time, adjust the diffuse color parameters according to the color temperature of the illumination to make the colors of the products more in line with the visual effects under the actual illumination. For example, if it is natural light illumination, the diffuse color can be biased towards warm tones; if it is artificial light illumination, make corresponding adjustments according to the color of the light. For a scene with weak illumination intensity, reduce the material reflectivity to make the products look softer. Through the above adjustments, combined with the illumination intensity distribution of the scene, generate a rendering effect that conforms to the actual situation, enabling users to more intuitively feel the appearance of the products under specific illumination conditions.

[0071] Step S1345: Based on the existing furniture layout, calculate the spatial position relationship between the three-dimensional product model and the furniture model to generate placement position coordinates that avoid model overlap.

[0072] Step S13451: Extract the bounding box coordinates of the existing furniture models in the virtual scene template, and the bounding box coordinates include the minimum x, y, z coordinates and the maximum x, y, z coordinates of the furniture models.

[0073] In the virtual scene template, each furniture model has its corresponding bounding box coordinates. The bounding box is the smallest cuboid that can completely enclose the furniture model. The position and range of the furniture model in three-dimensional space can be determined by its minimum x, y, z coordinates and maximum x, y, z coordinates. The system will traverse the existing furniture models in the virtual scene template and use a three-dimensional model processing algorithm to extract the bounding box coordinates of each furniture model. For example, for a wardrobe model, by analyzing its geometric shape and vertex coordinates, the minimum x, y, z coordinates and maximum x, y, z coordinates of the smallest cuboid that can enclose the wardrobe are calculated, and this coordinate information will be used for subsequent calculation of the positional relationship with the three-dimensional model of the commodity.

[0074] Step S13452: Extract the bounding box coordinates of the three-dimensional model of the commodity, where the bounding box coordinates include the minimum x, y, z coordinates and maximum x, y, z coordinates of the commodity model.

[0075] Similarly, for the three-dimensional model of the commodity, its bounding box coordinates also need to be extracted. By analyzing the geometric structure of the three-dimensional model of the commodity and using the same algorithm as for extracting the bounding box coordinates of the furniture model, the minimum x, y, z coordinates and maximum x, y, z coordinates of the smallest cuboid that can completely enclose the three-dimensional model of the commodity are calculated. For example, for a sofa model, its boundary range in three-dimensional space is determined to obtain the corresponding bounding box coordinates.

[0076] Step S13453: Calculate the coordinate overlap amounts of the bounding box of the three-dimensional model of the commodity and the bounding boxes of each existing furniture model in the three dimensions of x, y, and z.

[0077] After obtaining the bounding box coordinates of the three-dimensional model of the commodity and the existing furniture models, it is necessary to calculate their coordinate overlap amounts in the three dimensions of x, y, and z. For each dimension, the minimum and maximum coordinate values of the bounding box of the three-dimensional model of the commodity and the bounding box of the furniture model are compared respectively. For example, in the x dimension, calculate the difference between the minimum x coordinate of the bounding box of the three-dimensional model of the commodity and the maximum x coordinate of the bounding box of the furniture model, and the difference between the maximum x coordinate of the bounding box of the three-dimensional model of the commodity and the minimum x coordinate of the bounding box of the furniture model, and take the smaller value of these two differences as the coordinate overlap amount in the x dimension. The same method is applied to the y and z dimensions to obtain the coordinate overlap amounts in the three dimensions. Through the above calculation, it can be judged whether there is an overlapping situation between the three-dimensional model of the commodity and the furniture model in each dimension.

[0078] Step S13454: When the coordinate overlap amount in any dimension is greater than zero, translate the bounding box coordinates of the three-dimensional model of the commodity in the positive or negative direction of that dimension until the coordinate overlap amounts in all dimensions are less than or equal to zero.

[0079] If, when calculating the coordinate overlap, it is found that the coordinate overlap in any dimension is greater than zero, it indicates that there is an overlap between the 3D model of the product and the furniture model in that dimension. To avoid model overlap, a translation operation needs to be performed on the bounding box coordinates of the 3D model of the product. Depending on the overlap situation, choose to translate in the positive or negative direction of that dimension. For example, if there is an overlap in the x dimension and the maximum x coordinate of the bounding box of the 3D model of the product is greater than the minimum x coordinate of the bounding box of the furniture model, then translate the bounding box coordinates of the 3D model of the product along the negative x-axis by a set distance; otherwise, translate it along the positive x-axis. After each translation, recalculate the coordinate overlap in the three dimensions and repeat this process continuously until the coordinate overlap in all dimensions is less than or equal to zero, that is, the 3D model of the product and the furniture model no longer overlap.

[0080] Step S13455: Record the translated bounding box coordinates of the 3D model of the product as the placement position coordinates to avoid model overlap.

[0081] After the translation operation to ensure that the 3D model of the product does not overlap with the existing furniture model in each dimension, record the translated bounding box coordinates of the 3D model of the product, which represent the placement position of the 3D model of the product in the virtual scene to avoid overlapping with the existing furniture model. Use these placement position coordinates as the final result for subsequent rendering and display, so that the product can be placed in the virtual scene at a reasonable position, presenting a real and non-overlapping scene effect to the user.

[0082] Step S1346: Perform multi-angle rendering processing on the adjusted 3D model of the product to generate virtual placement effect display information including a front view, a side view, and a top view.

[0083] After completing the scale adjustment, material adjustment, and position determination of the 3D model of the product, perform multi-angle rendering processing on the adjusted 3D model of the product. Use a 3D rendering engine to render the 3D model of the product from different perspectives. Render from the front view, side view, and top view respectively to provide the appearance information of the product from different perspectives. The front view can display the front features and main appearance of the product, the side view can present the side structure and thickness information of the product, and the top view can display the top shape and layout of the product. Through the above multi-angle rendering, generate virtual placement effect display information including a front view, a side view, and a top view. These virtual placement effect display information will be stored in the form of images or videos so that users can observe the placement effect of the product in the usage scenario from multiple angles in subsequent AR shopping guide interactions.

[0084] Step S135: According to the information presentation rhythm requirements in the interaction preference characteristics, perform content stratification processing on the attribute description information to generate a multi-level information display structure including a basic attribute summary and extended attribute details.

[0085] According to the information presentation rhythm requirements in the interaction preference characteristics, the content stratification processing is carried out on the attribute description information retrieved from the database before. First, classify and screen the attribute description information to determine the basic attributes and extended attributes. The basic attributes are the most basic and crucial attribute information of the product, such as the name of the product, main functions, basic dimensions, etc. These information can quickly enable users to have a preliminary understanding of the product. The extended attributes are more detailed and in-depth attribute information, such as the technical parameters of the product, material composition, detailed description of special functions, etc. Then, according to the information presentation rhythm requirements, organize the basic attribute information into a basic attribute summary and present it to users in a concise and clear manner. The extended attribute information is organized into extended attribute details to provide more detailed content when users need to further understand the product. Through the above stratification processing, a multi-level information display structure including the basic attribute summary and extended attribute details is generated to meet the needs of different users at different information acquisition stages.

[0086] Step S136: Combine the acceptance of interaction operation complexity in the interaction preference characteristics to design the interaction trigger method of the AR shopping guide interaction content, and the interaction trigger method includes gesture sensing trigger, voice command trigger, and line-of-sight focus trigger.

[0087] Design the interaction trigger method of the AR shopping guide interaction content according to the acceptance of interaction operation complexity in the interaction preference characteristics. If the user has a high acceptance of interaction operation complexity, relatively complex and diverse interaction trigger methods can be designed. For example, for gesture sensing trigger, multiple different gesture actions can be designed to trigger different functions, such as spreading both hands to expand the detailed information of the product, and swiping with one hand to switch the product perspective. The voice command trigger can set a richer command set to allow users to express more complex requirements through natural language, such as "show the material composition and technical parameters of the product". The line-of-sight focus trigger can combine eye tracking technology to trigger different operations according to the position and time where the user's line of sight stays, such as automatically popping up the detailed introduction of a certain part when the line of sight stays on that part of the product for a long time.

[0088] If the user has a low acceptance of interaction operation complexity, a simple and easy-to-understand interaction trigger method is designed. For gesture sensing trigger, only a few commonly used gestures can be set, such as a click gesture to select the product, and a swipe-up gesture to view more information. The voice command trigger can adopt fixed simple commands, such as "view details", "return", etc. The line-of-sight focus trigger can reduce the trigger sensitivity to reduce the situation of false triggers. Through the above design, a suitable interaction trigger method is provided according to the user's acceptance of interaction operation complexity, improving the user's interaction experience.

[0089] Step S137: Based on the voice feedback response timeliness requirement in the interaction preference features, perform duration compression and key information enhancement processing on the voice commentary information of the AR shopping guide interaction content to generate a voice guidance instruction that meets the response timeliness requirement.

[0090] For example, if the user has a high requirement for the voice feedback response timeliness, duration compression needs to be performed on the voice commentary information. First, analyze the voice commentary content and remove redundant expressions and unnecessary information. For example, avoid repeated explanations and overly detailed background introductions. Then, refine and streamline the remaining key information, and use more concise language to express the same meaning. At the same time, appropriately adjust the speech rate, and under the premise of ensuring clear audibility of the voice, increase the speech rate to shorten the duration of the voice commentary.

[0091] While performing duration compression, it is also necessary to perform key information enhancement processing. By adjusting parameters such as the intonation, volume, and tone color of the voice, highlight the key information. For example, broadcast key product attributes, functional features, etc. with a higher volume and clearer intonation, so that users can more easily capture important information. Through the above duration compression and key information enhancement processing, generate a voice guidance instruction that meets the response timeliness requirement, ensuring that users can obtain key product information in a shorter time.

[0092] Step S138: Integrate the three-dimensional model data, multi-level information display structure, interaction trigger method, and voice guidance instruction after the scene rendering processing to generate AR shopping guide interaction content adapted to the user.

[0093] In this embodiment, first, associate the three-dimensional model data after the scene rendering processing with the multi-level information display structure. For example, while displaying the three-dimensional model, display the basic attribute summary in the form of a floating window next to the model, and when the user triggers the corresponding interaction operation, display the extended attribute details. Then, bind the interaction trigger method to the three-dimensional model and the information display structure. For example, set the gesture sensing trigger to switch the perspective of the three-dimensional model and display the corresponding information content at the same time; the voice command trigger can directly call up specific information or perform specific operations. Finally, integrate the voice guidance instruction into the entire interaction process, and play the voice guidance instruction in a timely manner when the user performs the interaction operation to provide operation tips and information commentary. Through the above content integration, generate AR shopping guide interaction content adapted to the user, providing the user with a complete and smooth AR shopping guide interaction experience.

[0094] Step S140: Perform spatial alignment and fusion processing on the AR shopping guide interaction content and the physical store scene to generate an AR shopping guide interaction scene that can be displayed in real time.

[0095] Step S141: Obtain the spatial positioning marker point data of the store's physical scene, where the spatial positioning marker point data includes the coordinates of the product display shelves, the boundary coordinates of the fitting rooms, and the center line coordinates of the aisles.

[0096] In this embodiment, for the product display shelves, devices such as 3D laser scanners are used to accurately measure their coordinate positions in the three-dimensional space. The display shelves are scanned from multiple angles to obtain the coordinate information of their respective vertices, thereby determining the accurate positions of the product display shelves. To obtain the boundary coordinates of the fitting rooms, it is necessary to measure the boundaries of the fitting rooms, including the coordinates of positions such as the walls and doors. Total stations and other measuring instruments are used to measure along the boundaries of the fitting rooms and record the coordinates of each point on the boundaries to determine the boundary range of the fitting rooms. The measurement of the center line coordinates of the aisles is carried out by setting measurement points in the aisles and using measuring tools to measure the position and orientation of the center line of the aisle and record the coordinates of each point on the center line. Through the above measurements and markings, the spatial positioning marker point data of the store's physical scene, including the coordinates of the product display shelves, the boundary coordinates of the fitting rooms, and the center line coordinates of the aisles, is obtained.

[0097] Step S142: Perform coordinate mapping processing on the 3D model data in the AR shopping guide interaction content, match and align the virtual coordinate origin of the 3D model with the spatial positioning marker point coordinates of the corresponding product display shelves, and generate model space alignment parameters.

[0098] In this embodiment, first, determine the virtual coordinate origin of the 3D model, which is usually the initial coordinate point of the model in the 3D modeling software. Then, match and align this virtual coordinate origin with the spatial positioning marker point coordinates of the corresponding product display shelves. Through transformation operations such as translation, rotation, and scaling, the virtual coordinate system of the 3D model is made to match the actual coordinate system of the store's physical scene. For example, calculate the offset between the virtual coordinate origin of the 3D model and the spatial positioning marker point coordinates of the product display shelves, and move the virtual coordinate origin of the model to the position that coincides with the spatial positioning marker point coordinates of the product display shelves through translation operations. At the same time, according to the actual direction and size of the product display shelves, rotate and scale the 3D model to make it conform to the actual scene in terms of direction and proportion. Through the above coordinate mapping processing, model space alignment parameters are generated, and these parameters will be used for subsequent rendering and fusion to ensure that the 3D model can be accurately placed at the corresponding position in the store's physical scene.

[0099] Step S143: Based on the basic attribute summary and extended attribute details in the multi-level information display structure, determine the display position of the information display window in the store's physical scene, where the display position includes the visible area directly in front of the product display shelves and the auxiliary area on the side of the user's current staying area.

[0100] In this embodiment, for the determination of the visible area directly in front of the merchandise display rack, the visual habits and line of sight range of users when browsing products need to be considered. First, analyze the spatial layout and orientation of the merchandise display rack to determine the general range directly in front of it. Then, in combination with the visual characteristics of the human eye, when an average person stands or walks normally, their line of sight will focus within a set height and angle range in the front. Taking the center position of the merchandise display rack as a reference, the set distances upward and downward, as well as the set angle ranges to the left and right, are determined as the visible area directly in front. For example, the height range set as a proportion from the top of the merchandise display rack downward, and the fan-shaped areas set at certain angles to the left and right with the center line of the display rack as the axis, all belong to the visible area directly in front. Setting an information display window within this visible area directly in front can ensure that users can easily notice the information in the window when viewing products.

[0101] For the determination of the side auxiliary area of the area where the user currently stays, the position information of the user needs to be obtained in real time. Based on the previously obtained user position movement trajectory data, determine the area where the user currently stays. At the side of the staying area, select a suitable position as the display position of the information display window. This display position should neither affect the normal activities of the user nor be inconvenient for the user to view the information. For example, if the user stays beside a certain merchandise display rack, then at a position on the left or right side of the display rack, at a set distance from the user's body and not blocked by other items, can be used as the side auxiliary area. Setting an information display window in this side auxiliary area can provide additional information support for the user. When the user needs to further understand the product, they can conveniently view details such as the extended attributes in the window.

[0102] Step S144: Perform trigger area calibration processing on the gesture sensing trigger, voice command trigger, and line of sight focus trigger in the interaction trigger method, associate and bind the trigger area range with the spatial positioning marker point coordinates of the merchandise display rack and the user's current position movement trajectory, and generate trigger area calibration parameters.

[0103] Step S1441: For the gesture sensing trigger, with the spatial positioning marker point coordinates of the merchandise display rack as the center, calculate the distance between the user and the display rack based on the coordinates of the nearest position point of the user's current position movement trajectory, and adjust the effective area radius of the gesture sensing trigger based on the distance to generate gesture trigger area calibration parameters.

[0104] In this embodiment, first, the coordinates of the nearest position point are extracted from the movement trajectory of the user's current position. Using the spatial distance calculation method, the distance between the coordinates of this position point and the coordinates of the spatial positioning marker point of the commodity display shelf is calculated. This distance reflects the relative position relationship between the user and the commodity display shelf. According to this distance, the radius of the effective area triggered by gesture sensing is adjusted. If the user is relatively close to the commodity display shelf, the radius of the effective area triggered by gesture sensing can be appropriately reduced because when the user operates at a close distance, their gesture movements are relatively easy to be accurately recognized, and a smaller effective area can reduce the possibility of false triggering. On the contrary, if the user is relatively far from the commodity display shelf, in order to ensure that the user can easily trigger the interaction, the radius of the effective area triggered by gesture sensing needs to be appropriately increased. Through the above adjustment, calibration parameters for the gesture trigger area are generated, and these parameters include the adjusted radius of the effective area, the coordinates of the center position, etc., which are used for subsequent gesture sensing trigger judgment.

[0105] Step S1442: For the trigger of the voice command, extract the coordinates of the voice collection device in the movement trajectory of the user's current position, calculate the relative position between the voice collection device and the coordinates of the spatial positioning marker point of the commodity display shelf, and based on the relative position, adjust the effective decibel threshold and the keyword recognition tolerance range triggered by the voice command, and generate calibration parameters for the voice trigger area.

[0106] In this embodiment, the voice collection device is usually a smart device carried by the user or a microphone installed at a specific position in the store. Calculating the relative position between the coordinates of the voice collection device and the coordinates of the spatial positioning marker point of the commodity display shelf includes information such as distance and direction. According to this relative position, the effective decibel threshold and the keyword recognition tolerance range triggered by the voice command are adjusted. If the voice collection device is relatively close to the commodity display shelf, the effective decibel threshold can be appropriately reduced because the voice signal is relatively strong at a close distance, and a lower threshold can more sensitively capture the user's voice command. At the same time, the keyword recognition tolerance range is narrowed to improve the recognition accuracy. If the voice collection device is relatively far from the commodity display shelf, in order to ensure that the user's voice command can be received, the effective decibel threshold needs to be appropriately increased, and the keyword recognition tolerance range is increased to adapt to the possible attenuation and interference of the voice signal. Through the above adjustment, calibration parameters for the voice trigger area are generated, and these parameters include the adjusted effective decibel threshold, the keyword recognition tolerance range, and the relative position information, etc.

[0107] Step S1443: For the trigger of the line-of-sight focus, obtain the eye movement tracking data of the user's current line-of-sight direction, calculate the included angle between the line-of-sight direction and the coordinates of the spatial positioning marker point of the commodity display shelf, and based on the included angle, adjust the effective angle range and the required duration of continuous fixation triggered by the line-of-sight focus, and generate calibration parameters for the line-of-sight trigger area.

[0108] In this embodiment, the eye movement tracking device can monitor the movement trajectory and fixation direction of the user's eyes in real time. Calculate the angle between the user's line of sight direction and the coordinates of the spatial positioning marker points on the product display shelf. This angle reflects the relative angular relationship between the user's line of sight and the product display shelf. According to this angle, adjust the effective angle range and the required duration of continuous fixation triggered by the line of sight focus. If the angle between the user's line of sight direction and the product display shelf is small, it means that the user is facing the product display shelf. At this time, the effective angle range can be appropriately reduced, and the required duration of continuous fixation can be shortened because the user's attention is more concentrated on the product and it is easier to trigger an interaction. If the angle is large, it means that the user's line of sight deviates from the product display shelf, and the effective angle range needs to be appropriately increased, and the required duration of continuous fixation needs to be extended to ensure that the interaction is triggered only when the user is truly interested in the product. Through the above adjustments, calibration parameters for the line of sight trigger area are generated, and these parameters include the adjusted effective angle range, the required duration of continuous fixation, and angle information, etc.

[0109] Step S1444: Integrate the gesture trigger area calibration parameters, voice trigger area calibration parameters, and line of sight trigger area calibration parameters to generate trigger area calibration parameters.

[0110] For example, organize information such as the effective area radius and center position coordinates of the gesture trigger area, the effective decibel threshold and keyword recognition tolerance range of the voice trigger area, and the effective angle range and required duration of continuous fixation of the line of sight trigger area according to the set format and structure. At the same time, considering the collaborative work between different trigger methods, ensure that there are no conflicts or contradictions between the various parameters. Through the above integration, a comprehensive trigger area calibration parameter is generated for subsequent interaction trigger judgment and control, enabling different interaction trigger methods to work collaboratively under the unified calibration parameters, improving the accuracy and reliability of the interaction.

[0111] Step S145: Match and adjust the audio output parameters of the voice guidance instruction with the sound field environment data of the store physical scene. The sound field environment data includes the ambient noise decibel value and the sound reflection coefficient, and generate voice output parameters that conform to the sound field environment.

[0112] In this embodiment, first, obtain the ambient noise decibel value and the sound reflection coefficient of the store physical scene. The ambient noise decibel value can be measured in real time by noise monitoring equipment installed in the store, which reflects the noise level in the store. The sound reflection coefficient is related to factors such as the building structure and decoration materials of the store and can be obtained through acoustic testing or simulation calculations, which represents the degree of sound reflection in the store.

[0113] Then, adjust the volume of the voice guidance instructions according to the ambient noise decibel value. If the ambient noise decibel value is high, it indicates that the store is relatively noisy, and it is necessary to appropriately increase the volume of the voice guidance instructions to ensure that the user can clearly hear the voice information. For example, when the ambient noise decibel value reaches a set level, increase the volume of the voice guidance instructions by a set proportion. At the same time, adjust the tone and quality of the voice according to the sound reflection coefficient. If the sound reflection coefficient is large, it means that there will be more reflections of the sound in the store, which may cause problems such as reverberation and echo of the sound. At this time, the voice can be processed to adjust its tone and quality, reduce the influence of reverberation and echo, and make the voice clearer and more audible. For example, through audio filtering technology, remove some unnecessary reflected sound components. Through the above matching and adjustment, generate voice output parameters that conform to the sound field environment, including adjusted volume, tone, quality and other parameters, to ensure that the voice guidance instructions can achieve the best playback effect in the actual sound field environment of the store.

[0114] Step S146: According to the model space alignment parameters, the display position of the information display window, the trigger area calibration parameters, and the voice output parameters, perform fusion rendering processing on the AR shopping guide interaction content and the physical store scene to generate a real-time displayable AR shopping guide interaction scene.

[0115] In this embodiment, first, according to the model space alignment parameters, accurately place the previously processed three-dimensional model of the commodity at the corresponding position in the physical store scene. Use a three-dimensional rendering engine to synthesize the model and the scene to ensure that the position, orientation, and scale of the model match the actual scene. For example, align the virtual coordinate system of the three-dimensional model of the commodity with the actual coordinate system of the physical store scene so that the model can seamlessly integrate into the scene.

[0116] Then, according to the display position of the information display window, render the information display window at the corresponding position in the physical store scene. Display information such as the basic attribute summary and extended attribute details in the multi-level information display structure in the window. At the same time, according to the trigger area calibration parameters, set the response mechanism for interaction triggering in the corresponding area. When the user performs interaction operations such as gestures, voice, or line of sight, it can accurately trigger the corresponding interaction functions, such as switching the model perspective, displaying extended information, etc.

[0117] Finally, according to the voice output parameters, play the voice guidance instructions in the physical store scene with appropriate volume, tone, and quality. Combine the rendered three-dimensional model and the information display window to provide the user with an all-round AR shopping guide interaction experience. Through the above fusion rendering processing, generate a real-time displayable AR shopping guide interaction scene. The user can see the fused scene in real time through the AR device in the store, interact with the commodity, and obtain commodity information.

[0118] Step S150: In response to the user interaction operation in the AR shopping guide interaction scenario, generate interaction feedback information and update the interaction data set to trigger the iterative optimization of the interaction process.

[0119] Step S151: Monitor the operation behavior of the user on the interaction triggering method in the AR shopping guide interaction scenario, where the operation behavior includes gesture triggering actions, voice command content, and the position where the line of sight focus stays.

[0120] In this embodiment, for the monitoring of gesture triggering actions, it is achieved through gesture recognition sensors installed in AR devices or in the store. These sensors can capture the movement trajectory and posture changes of the user's hand in real time. For example, when the user makes specific gestures, such as clicking, swiping, pinching, etc., the sensors will transmit this action information to the system in real time. The system will identify and classify these actions to determine which interaction function the user has triggered.

[0121] For the monitoring of voice command content, voice recognition technology is used to recognize the voice issued by the user in real time. The voice signal of the user is collected through the microphone on the AR device or the microphone array in the store, and then the signal is transmitted to the voice processing server for recognition and analysis. The server will convert the voice signal into text content and perform semantic understanding on the text content to determine the intention expressed by the user's voice command. For example, when the user says "View the detailed information of this product", it can be recognized that the user's command is to view the detailed information of the product.

[0122] For the monitoring of the position where the line of sight focus stays, it is achieved through eye tracking technology. The eye tracking device can monitor the movement trajectory and gaze direction of the user's eyeball in real time to determine the position where the line of sight focus stays. When the user's line of sight stays on a certain product or information display window for a set time, the stay position can be recorded, and it can be judged that the user may be interested in the content at that position. Through the above monitoring, the operation behavior information of the user in the AR shopping guide interaction scenario can be obtained in real time.

[0123] Step S152: Perform response status recognition processing on the operation behavior, judge whether the user has successfully triggered the information display or function call of the AR shopping guide interaction content, and generate an interaction success status identifier.

[0124] In this embodiment, for the gesture trigger action, according to the previously set gesture trigger rules, it can be determined whether the user's gesture action meets the corresponding trigger conditions. If the conditions are met and the system successfully executes the corresponding information display or function call operation, such as switching the perspective of the product model, displaying the detailed product information, etc., it is determined that the user has successfully triggered the interaction, and an interaction success status flag of "success" is generated; otherwise, if the gesture action does not meet the trigger conditions or an error occurs when the system executes the operation, an interaction success status flag of "failure" is generated.

[0125] For the voice command content, semantic analysis can be performed on the speech recognition result to determine whether the user's voice command matches the commands preset in the system. If the match is successful and the system can correctly execute the corresponding information display or function call operation, it is determined that the user has successfully triggered the interaction, and an interaction success status flag of "success" is generated; if the voice command cannot be recognized or does not match the preset command, resulting in the system being unable to execute the corresponding operation, an interaction success status flag of "failure" is generated.

[0126] For the position where the line of sight focus stays, according to the effective angle range triggered by the line of sight and the requirement of the continuous fixation duration, it can be determined whether the position and duration of the user's line of sight stay meet the trigger conditions. If the conditions are met and the system successfully triggers the corresponding interaction function, such as popping up the relevant information window of the product, it is determined that the user has successfully triggered the interaction, and an interaction success status flag of "success" is generated; otherwise, an interaction success status flag of "failure" is generated. Through the above response status recognition process, it can be accurately determined whether the user has successfully triggered the information display or function call of the AR shopping guide interaction content.

[0127] Step S153: When the interaction success status flag is "success", extract the information display level or function call type corresponding to the user's trigger operation, and record the access depth of the user to the multi-level information display structure and the selection preference for the function call type.

[0128] In this embodiment, for the extraction of the information display level, if the user triggers the display of the basic attribute summary, it is recorded that the user accesses the basic level; if the user triggers the display of the extended attribute details, it is recorded that the user accesses the extended level. Through the above records, the access depth of the user to the multi-level information display structure can be understood, that is, whether the user only views the basic information or further views the detailed information.

[0129] For records of function call types, analyze which function the user triggers, such as switching the perspective of the product model, viewing the video introduction of the product, comparing different products, etc. Record the function call type selected by the user to understand the user's preference for different functions. For example, if the user triggers the function of switching the perspective of the product model multiple times, it indicates that the user is more interested in viewing the product from different angles; if the user often selects the function of viewing the video introduction of the product, it indicates that the user prefers to understand the product through videos.

[0130] Step S154: When the interaction success status flag is failure, analyze the matching deviation between the operation behavior and the trigger area calibration parameter, and record the position offset and time delay parameter of the user's trigger operation.

[0131] In this embodiment, for gesture trigger actions, compare the difference between the actual position where the user makes the gesture and the effective range of the gesture trigger area, and calculate the position offset. For example, if the effective range of the gesture trigger area is a circular area centered on a certain point, and the user's gesture action occurs at a set distance outside the circular area, then this distance is the position offset. At the same time, record the time delay parameter from when the user makes the gesture action to the system response, and analyze whether there is a problem of excessive delay.

[0132] For voice command triggers, analyze the relative position relationship between the position where the user issues the voice command and the voice trigger area, and calculate the position offset. For example, if the voice trigger area is a set range centered on the product display shelf, and the user issues the voice command at a position far from the display shelf, it may result in poor reception of the voice signal. At the same time, record the time delay parameter from when the voice command is issued to the system recognition and response.

[0133] For sight focus triggers, compare the difference between the position where the user's sight focus stays and the effective angular range and position of the sight trigger area, and calculate the position offset. For example, if the sight trigger area is a specific angular and position range, and the user's sight focus deviates from this range, then the deviation angle and distance are the position offset. Similarly, record the difference between the sight focus stay time and the continuous fixation duration requirement, as well as the time delay parameter from sight trigger to system response.

[0134] Step S155: Based on the interaction success status flag, access depth, selection preference, position offset, and time delay parameter, generate interaction feedback information including the evaluation result of interaction effectiveness and the direction of optimization and improvement.

[0135] In this embodiment, first, the effectiveness of the interaction is evaluated. According to the interaction success status flag, if the number of successful times is large, it indicates that the effectiveness of the interaction is high; if the number of failed times is large, it indicates that there are problems with the interaction and improvement is needed. At the same time, combined with the access depth and selection preferences, the degree of interest and demand of the user for the interaction content is analyzed. If the user's access depth is relatively deep and there are obvious selection preferences for certain functions, it indicates that these interaction contents and functions are more popular among users and can be further optimized and expanded.

[0136] Then, according to the position offset and time delay parameters, the problems and deficiencies in the interaction process are found. If the position offset is large, it indicates that the calibration of the trigger area may be inaccurate and the range and position of the trigger area need to be adjusted; if the time delay parameter is long, it indicates that the response speed of the system is slow and the processing flow and algorithm of the system need to be optimized. Based on these analysis results, directions for optimization and improvement are proposed, such as adjusting the trigger area calibration parameters, optimizing the speech recognition and gesture recognition algorithms, and improving the response speed of the system. Through the above processing, interaction feedback information containing the evaluation results of interaction effectiveness and directions for optimization and improvement is generated.

[0137] Step S156: Supplement the operation behavior, interaction success status flag, access depth, selection preference, position offset, and time delay parameters to the interaction data set to form an updated interaction data set.

[0138] For example, record the user's new operation behavior in the product attention action sequence, and use information such as the interaction success status flag, access depth, and selection preference as new features to add to the user's interaction data.

[0139] Step S157: Input the updated interaction data set into the interaction process optimization module. Through the comparative analysis and processing of historical interaction data and current interaction data, adjust the association rules for user intention parsing processing, the generation logic of AR shopping guide interaction content, and the parameter configuration of spatial alignment and fusion processing, and trigger the iterative optimization of the interaction process.

[0140] For example, step S1571: Extract a historical interaction data subset and a current interaction data subset from the updated interaction data set. The historical interaction data subset contains the interaction data before this interaction, and the current interaction data subset contains the interaction data newly added in this interaction.

[0141] In this embodiment, the updated interaction data set can be divided into a historical interaction data subset and a current interaction data subset according to the timestamp information of the interaction data. The historical interaction data subset covers all interaction data before this interaction, including information such as the user's past location movement trajectory in the store, the sequence of product attention actions, and the content of voice consultations. These data reflect the user's past behavior patterns and preferences, and are of great significance for analyzing the user's long-term behavior trends. For example, the historical data may record the location and duration of the user's multiple stays in a certain type of product area, as well as the attention actions and voice consultation content for specific products. By analyzing, we can understand the user's long-term interest level in this type of product.

[0142] The current interaction data subset contains the newly added interaction data during this interaction, that is, the information such as the user's operation behavior, interaction success status flag, access depth, selection preference, position offset, and time delay parameter recorded from the start of this interaction to the current moment. These data reflect the user's latest behavior and feedback in the current interaction, and can capture the changes in the user's behavior in a timely manner. For example, the user's new attention actions and voice consultation content for a certain product during this interaction, as well as the success or failure situation during the interaction, all belong to the category of the current interaction data subset.

[0143] Step S1572: Compare and analyze the user's location movement trajectories in the historical interaction data subset and the current interaction data subset, identify the change trend of the user's movement path, and adjust the preset threshold for identifying the staying area in the user intention parsing process.

[0144] In this embodiment, the location movement trajectories of the user in different time periods can be compared point by point to observe the changes in aspects such as the shape, direction, and speed of the movement path. For example, compare the areas passed by the user in the past few interactions with the areas passed by the user in this interaction to determine whether new areas are involved or whether the access frequency to certain areas has changed. If it is found that the user frequently visits a product area that was less involved before during this interaction, this may mean that the user's interest range has expanded.

[0145] By identifying the change trend of the movement path, the system can further determine whether the user's behavior pattern has changed. If the user's movement path becomes more concentrated in certain specific areas, it indicates that the user may have a higher degree of attention to the products in these areas; on the contrary, if the movement path becomes more dispersed, it may indicate that the user's interests are more extensive.

[0146] Based on these analysis results, the preset threshold for identifying the staying area in user intention parsing processing can be adjusted. If the change in the user's movement path indicates that their behavior is more cautious and the movement speed is slower, the preset threshold for identifying the staying area may be appropriately reduced to more accurately capture the user's staying behavior; on the contrary, if the user's movement speed increases and the behavior is faster, the preset threshold may be appropriately increased to avoid misjudging some short stays as meaningful staying areas. The above adjustments can make the identification of the staying area more in line with the user's current behavior pattern and improve the accuracy of user intention parsing.

[0147] Step S1573: Compare and analyze the sequence of product attention actions in the historical interaction data subset and the current interaction data subset, identify the evolution law of the user's action pattern, and adjust the feature extraction weight of the action pattern analysis in the user intention parsing processing.

[0148] In this embodiment, the sequence of product attention actions of the user in different time periods can be disassembled and compared in detail, and features such as the type, order, and time interval of each action can be analyzed. For example, compare the frequency and order of actions such as picking up, observing, and putting back a certain type of product by the user in the past with the action pattern in the current interaction. If it is found that the user picks up the product more frequently and observes it for a longer time in the current interaction, this may indicate that the user's interest in the product has increased.

[0149] By comparing a large amount of historical data and current data, the system can identify the evolution law of the user's action pattern. For example, it may be found that as the number of interactions increases, the user's operation actions on the product become more proficient and diversified, or the preference for certain specific actions has changed.

[0150] According to the identified evolution law, the feature extraction weight of the action pattern analysis in the user intention parsing processing can be adjusted. If the frequency of a certain action significantly increases in the current interaction and is more closely related to the user's interest and needs, then the weight of the action feature will be increased in the action pattern analysis to highlight its importance; on the contrary, if the frequency of a certain action decreases and its role in judging the user's intention weakens, the weight of the action feature will be reduced. The above adjustments can make the action pattern analysis more accurately reflect the user's true intention and improve the reliability of user intention parsing.

[0151] Step S1574: Compare and analyze the voice consultation content in the historical interaction data subset and the current interaction data subset, identify the updated features of the user's semantic expression, and adjust the keyword matching rule of semantic understanding in the user intention parsing processing.

[0152] In this embodiment, semantic analysis can be performed on the voice consultation content, and it can be decomposed and compared in terms of keywords, sentence structures, semantic intentions, etc. For example, analyze whether there are differences between the keywords and expression methods frequently used by the user in past voice consultations and those in the current interaction. If it is found that the user uses some new keywords in the current interaction, or the expression of product attributes and requirements is more specific and clear, this may indicate that the user's needs have changed.

[0153] Through the comparison of a large amount of voice consultation content, the system can identify the updated features of the user's semantic expression. These features may include the emergence of new keywords, the shift of semantic focus, the simplification or complication of the expression method, etc. For example, in the past, the user may have simply asked about the price of the product, while in the current interaction, the user details the price range of the product, whether there are promotional activities, and the relationship between price and performance, etc. This indicates that the user's attention to price is more in-depth and the aspects of concern are more extensive.

[0154] According to the identified updated features, the keyword matching rules for semantic understanding in user intention parsing processing can be adjusted. If new keywords appear, they can be added to the keyword matching rules to ensure that the new needs of the user can be accurately identified; if the semantic weights of some keywords change, the weights of these keywords in the matching rules can be adjusted accordingly. For example, when the user's attention to the environmental protection performance of the product increases, the weights of keywords related to environmental protection in the matching rules can be increased so that the system can more accurately capture the user's demand for the environmental protection attributes of the product. The above adjustments can improve the accuracy of semantic understanding and better parse the intention of the user's voice consultation.

[0155] Step S1575: Compare and analyze the access depth of the AR shopping guide interaction content in the historical interaction data subset and the current interaction data subset, identify the hierarchical changes in the user's information needs, and adjust the hierarchical logic of the multi-level information display structure in the generation of the AR shopping guide interaction content.

[0156] In this embodiment, the access situation of the user to the AR shopping guide interaction content in different interactions can be recorded, including whether only the basic attribute summary is viewed, or the extended attribute details are further viewed, as well as the specific content and depth of the view. For example, in the historical interaction, the user may have simply browsed the basic attribute summary of the product, while in the current interaction, the user has delved into the extended attribute details such as the technical parameters and usage instructions of the product, indicating that the hierarchical level of the user's information needs has changed.

[0157] By comparing and analyzing the access depth, the system can identify the hierarchical change trend of user information needs. If it is found that the user gradually increases the frequency and depth of access to extended attribute details in multiple interactions, it means that the user has a deeper understanding of the product and may want to obtain more detailed information; conversely, if the user's access to extended attribute details decreases, it may mean that the user prefers to quickly obtain basic information.

[0158] According to the identified hierarchical change trends, the hierarchical logic of the multi-level information display structure in the generation of AR shopping guide interactive content can be adjusted. If the user's demand for extended attribute details increases, the display method of the extended attribute details can be optimized to make it more prominent and easy to access, and some new extended attribute information may be added; if the user pays more attention to the basic attribute summary, the content of the basic attribute summary can be further simplified and optimized to make it more concise and clear. For example, put some important basic attribute information in a more prominent position, or reformat the basic attribute summary to improve the readability of the information. The above adjustments can make the multi-level information display structure more in line with the user's information needs and improve the practicality of AR shopping guide interactive content.

[0159] Step S1576: Compare and analyze the interaction trigger operation deviations in the historical interaction data subset and the current interaction data subset, identify the habitual characteristics of the user's triggering behavior, and adjust the parameter configuration of the trigger area calibration in the spatial alignment fusion processing.

[0160] This embodiment can record information such as the position offset and time delay parameters when the user triggers the interactive function in different interactions, and compare the changes of these deviations in different time periods. For example, analyze whether the position offset and time delay of the user's gesture triggering in the past few interactions are the same as those in this interaction. If it is found that the position offset of the user's gesture triggering in this interaction is significantly reduced and the time delay is shortened, it means that the user has become more proficient in the triggering operation.

[0161] By comparing a large number of interaction trigger operation deviations, the habitual characteristics of user triggering behaviors can be identified, which may include user preferences for different triggering methods, the accuracy and stability of triggering operations, the regularity of triggering time, etc. For example, some users may be more accustomed to using gesture triggers, while some users prefer to use voice command triggers; some users' trigger operations are more accurate and stable, while some users may have large deviations.

[0162] According to the identified habitual features, the parameter configuration for trigger area calibration in spatial alignment and fusion processing can be adjusted. If the user's usage frequency of a certain triggering method increases, the trigger area calibration parameters for this triggering method can be optimized to better conform to the user's usage habits. For example, if the user has a high usage frequency of gesture triggering, the effective range and sensitivity of the gesture trigger area can be adjusted according to the position and action characteristics of the user's gesture trigger, improving the accuracy and reliability of gesture triggering; if the user's trigger operation time is relatively regular, the time setting of trigger response can be adjusted according to this rule to reduce time delay. The above adjustments can improve the accuracy of interactive triggering and the user experience, making the spatial alignment and fusion processing more in line with the actual needs of the user.

[0163] Step S1577: Complete the iterative optimization of the interaction process through the above adjustment operations.

[0164] After comparative analysis of aspects such as the user's position movement trajectory, product attention action sequence, voice consultation content, access depth of AR shopping guide interaction content, and deviation of interactive trigger operations, and corresponding adjustment of the association rules for user intention parsing processing, the generation logic of AR shopping guide interaction content, and the parameter configuration of spatial alignment and fusion processing, the iterative optimization of the interaction process is completed.

[0165] In subsequent interaction processes, work will be carried out with the optimized rules and parameters, which can more accurately parse the user's intention, generate AR shopping guide interaction content that better meets the user's needs, and achieve more accurate spatial alignment and fusion processing. For example, under the new association rules for user intention parsing, the user's demand and preference for products can be more accurately identified, providing more personalized product recommendations for the user; under the optimized generation logic of AR shopping guide interaction content, the multi-level information display structure is more reasonable, better meeting the user's information needs; under the adjusted parameter configuration of spatial alignment and fusion processing, the interactive trigger is more accurate, and the user can more conveniently interact with the AR shopping guide interaction content. Through the above iterative optimization, the quality and effect of AR intelligent shopping guide interaction are continuously improved, providing a better shopping experience for the user.

[0166] Figure 2 FIG. shows a schematic diagram of exemplary hardware and software components of an AR intelligent shopping guide interaction system 100 applicable to store management that can implement the idea of the present application. For example, the processor 120 can be used on the AR intelligent shopping guide interaction system 100 applicable to store management and is used to execute the functions in the present application.

[0167] The AR intelligent shopping guide interaction system 100 applied to store management can be a general-purpose server or a special-purpose server, both of which can be used to implement the AR intelligent shopping guide interaction method applied to store management in this application. Although only one server is shown in this application, for convenience, the functions described in this application can be implemented in a distributed manner on multiple similar platforms to balance the processing load.

[0168] For example, the AR intelligent shopping guide interaction system 100 applied to store management may include a network port 110 connected to a network, one or more processors 120 for executing program instructions, a communication bus 130, and different forms of storage media 140, such as disks, ROMs, or RAMs, or any combination thereof. Exemplarily, the AR intelligent shopping guide interaction system 100 applied to store management may also include program instructions stored in a ROM, a RAM, or other types of non-transitory storage media, or any combination thereof. The method of this application can be implemented according to these program instructions. The AR intelligent shopping guide interaction system 100 applied to store management also includes an I / O interface 150 between the computer and other input and output devices.

[0169] In addition, an embodiment of the present invention also provides a readable storage medium, in which computer-executable instructions are preset. When the processor executes the computer-executable instructions, the above-mentioned AR intelligent shopping guide interaction method applied to store management is implemented.

[0170] It should be noted that, in order to simplify the presentation of the disclosure of the present invention and thus help the understanding of one or more embodiments of the invention, in the foregoing description of the embodiments of the present invention, sometimes multiple features are merged into one embodiment, drawing, or description thereof.

Claims

1. An AR intelligent shopping guide interaction method applied to store management, characterized in that, The method includes: Obtaining an interaction data set of a user in a store, where the interaction data set includes the user's location movement trajectory, a sequence of product attention actions, and voice consultation content; Performing user intention parsing processing on the interaction data set to obtain the user's product demand characteristics and interaction preference characteristics; Generating AR shopping guide interaction content adapted to the user based on the product demand characteristics and interaction preference characteristics, where the AR shopping guide interaction content includes three-dimensional product display information and interaction guiding instructions; Performing spatial alignment and fusion processing on the AR shopping guide interaction content and the physical store scene to generate a real-time displayable AR shopping guide interaction scene; Responding to user interaction operations in the AR shopping guide interaction scene, generating interaction feedback information and updating the interaction data set to trigger iterative optimization of the interaction process.

2. The AR intelligent shopping guide interaction method applied to store management according to claim 1, wherein The performing user intention parsing processing on the interaction data set to obtain the user's product demand characteristics and interaction preference characteristics includes: Performing stay area recognition processing on the user's location movement trajectory, and extracting the stay duration parameter and area coverage range of the user in the product display area; Performing action pattern analysis processing on the sequence of product attention actions to identify the frequency of product picking up by the user, the operation of adjusting the display angle, and the characteristics of the putting-back position of the product; Performing semantic understanding processing on the voice consultation content, and extracting the product attribute keywords, usage scenario descriptions, and price sensitivity expressions consulted by the user; Inputting the stay duration parameter, area coverage range, picking-up frequency, display angle adjustment operation, putting-back position characteristics, product attribute keywords, usage scenario descriptions, and price sensitivity expressions into an intention correlation analysis module, and generating the user's product demand characteristics through context correlation rule matching processing. The product demand characteristics include the target product type, core attribute preferences, and usage scenario adaptation requirements; Based on the path smoothness of the user's location movement trajectory, the operation coherence of the sequence of product attention actions, and the expression clarity of the voice consultation content, generating the user's interaction preference characteristics through interaction behavior pattern recognition processing. The interaction preference characteristics include information presentation rhythm requirements, acceptance of interaction operation complexity, and voice feedback response time requirements.

3. The AR intelligent shopping guide interaction method applied to store management according to claim 2, wherein, The performing stay area recognition processing on the user's location movement trajectory and extracting the stay duration parameter and area coverage range of the user in the product display area includes: Dividing the user's location movement trajectory into a continuous sequence of location points in chronological order of timestamps, where the sequence of location points contains the store space coordinates corresponding to each timestamp; Calculating the coordinate change amount between adjacent location points, and identifying a continuous location point interval with a coordinate change amount less than a preset threshold as a stay area candidate interval; Performing boundary fitting processing on the coordinates of all location points within the stay area candidate interval to generate a region coverage range descriptor including a minimum bounding rectangle; Calculating the difference between the start timestamp and the end timestamp of the stay area candidate interval to obtain the stay duration parameter of the user in this area; Screen the intervals in the candidate intervals of the staying area that cover the spatial positioning marker points of the commodity display area, and extract the corresponding staying duration parameters and area coverage descriptors as the staying duration parameters and area coverage of the user in the commodity display area.

4. The AR intelligent shopping guide interaction method applied to store management according to claim 2, wherein, Generating AR shopping guide interaction content adapted to the user based on the commodity demand characteristics and interaction preference characteristics includes: According to the target commodity type and core attribute preference in the commodity demand characteristics, retrieve the three-dimensional model data and attribute description information of the corresponding commodity from the store commodity database; Based on the usage scenario adaptation requirements in the commodity demand characteristics, perform scene rendering processing on the three-dimensional model data to generate virtual placement effect display information of the commodity in the usage scenario described by the user; According to the information presentation rhythm requirements in the interaction preference characteristics, perform content stratification processing on the attribute description information to generate a multi-level information display structure including basic attribute summaries and extended attribute details; Combined with the acceptance of interaction operation complexity in the interaction preference characteristics, design the interaction trigger method of the AR shopping guide interaction content, and the interaction trigger method includes gesture sensing trigger, voice command trigger, and line-of-sight focus trigger; Based on the voice feedback response timeliness requirements in the interaction preference characteristics, perform duration compression and key information strengthening processing on the voice commentary information of the AR shopping guide interaction content to generate a voice guidance command that meets the response timeliness requirements; Integrate the three-dimensional model data, multi-level information display structure, interaction trigger method, and voice guidance command after the scene rendering processing to generate AR shopping guide interaction content adapted to the user.

5. The AR intelligent shopping guide interaction method applied to store management according to claim 4, wherein Based on the usage scenario adaptation requirements in the commodity demand characteristics, performing scene rendering processing on the three-dimensional model data to generate virtual placement effect display information of the commodity in the usage scenario described by the user includes: Analyze the usage scenario description in the commodity demand characteristics and extract the key scene elements, and the key scene elements include scene space size, lighting conditions, and existing furniture layout; Retrieve the virtual scene template matching the key scene elements from the store scene database, and the virtual scene template includes spatial size parameters, lighting intensity distribution, and furniture model position coordinates; Import the three-dimensional model data into the virtual scene template, and adjust the scale factor of the commodity three-dimensional model according to the scene space size parameters so that the size of the commodity three-dimensional model matches the scene space size; Adjust the material reflectivity and diffuse color parameters of the commodity three-dimensional model according to the lighting conditions to generate a rendering effect that conforms to the scene lighting intensity distribution; Based on the existing furniture layout, calculate the spatial position relationship between the commodity three-dimensional model and the furniture model to generate placement position coordinates that avoid model overlap; Perform multi-angle rendering processing on the adjusted commodity three-dimensional model to generate virtual placement effect display information including front view, side view, and top view.

6. The AR intelligent shopping guide interaction method applied to store management according to claim 5, characterized in that, Based on the existing furniture layout, calculating the spatial position relationship between the commodity three-dimensional model and the furniture model to generate placement position coordinates that avoid model overlap includes: Extract the bounding box coordinates of the existing furniture models in the virtual scene template, where the bounding box coordinates include the minimum x, y, z coordinates and the maximum x, y, z coordinates of the furniture models; Extract the bounding box coordinates of the 3D model of the commodity, where the bounding box coordinates include the minimum x, y, z coordinates and the maximum x, y, z coordinates of the commodity model; Calculate the coordinate overlap amounts of the bounding box of the 3D model of the commodity and the bounding boxes of each existing furniture model in the three dimensions of x, y, and z; When the coordinate overlap amount in any dimension is greater than zero, translate the bounding box coordinates of the 3D model of the commodity in the positive or negative direction of that dimension until the coordinate overlap amounts in all dimensions are less than or equal to zero; Record the translated bounding box coordinates of the 3D model of the commodity as the placement position coordinates to avoid model overlap.

7. The AR intelligent shopping guide interaction method applied to store management according to claim 4, characterized in that The spatial alignment and fusion processing of the AR shopping guide interaction content with the physical store scene to generate a real-time displayable AR shopping guide interaction scene includes: Obtain the spatial positioning marker point data of the physical store scene, where the spatial positioning marker point data includes the commodity display shelf coordinates, the fitting room boundary coordinates, and the center line coordinates of the passage; Perform coordinate mapping processing on the 3D model data in the AR shopping guide interaction content, match and align the virtual coordinate origin of the 3D model with the spatial positioning marker point coordinates of the corresponding commodity display shelf, and generate model spatial alignment parameters; Based on the basic attribute summary and the extended attribute details in the multi-level information display structure, determine the display position of the information display window in the physical store scene, where the display position includes the visible area directly in front of the commodity display shelf and the side auxiliary area of the user's current staying area; Perform trigger area calibration processing on the gesture sensing trigger, voice command trigger, and line of sight focus trigger in the interaction trigger mode, associate and bind the trigger area range with the spatial positioning marker point coordinates of the commodity display shelf and the user's current position movement trajectory, and generate trigger area calibration parameters; Match and adjust the audio output parameters of the voice guidance instruction with the sound field environment data of the physical store scene, where the sound field environment data includes the environmental noise decibel value and the sound reflection coefficient, and generate voice output parameters that conform to the sound field environment; According to the model spatial alignment parameters, the information display window display position, the trigger area calibration parameters, and the voice output parameters, perform fusion rendering processing on the AR shopping guide interaction content and the physical store scene to generate a real-time displayable AR shopping guide interaction scene.

8. The AR intelligent shopping guide interaction method applied to store management according to claim 7, wherein, The performing trigger area calibration processing on the gesture sensing trigger, voice command trigger, and line of sight focus trigger in the interaction trigger mode, associating and binding the trigger area range with the spatial positioning marker point coordinates of the commodity display shelf and the user's current position movement trajectory, and generating trigger area calibration parameters includes: For the gesture sensing trigger, with the spatial positioning marker point coordinates of the commodity display shelf as the center, calculate the distance between the user and the display shelf according to the coordinates of the nearest position point of the user's current position movement trajectory, and adjust the effective area radius of the gesture sensing trigger based on the distance to generate gesture trigger area calibration parameters; In response to the voice command trigger, the coordinates of the voice collection device of the user's current position movement trajectory are extracted, the relative position of the voice collection device and the coordinates of the spatial positioning mark point of the commodity display shelf are calculated, and the effective decibel threshold and keyword recognition tolerance range of the voice command trigger are adjusted based on the relative position to generate the voice trigger area calibration parameters; For the sight focus trigger, the eye tracking data of the user's current sight direction is obtained, the angle between the sight direction and the coordinates of the spatial positioning mark point of the commodity display shelf is calculated, and the effective angle range and continuous gaze duration requirement of the sight focus trigger are adjusted based on the angle to generate the sight trigger area calibration parameters; The gesture trigger area calibration parameters, the voice trigger area calibration parameters and the sight trigger area calibration parameters are integrated to generate trigger area calibration parameters.

9. The AR intelligent shopping guide interaction method applied to store management according to claim 7, characterized in that, The responding to the user interaction operation in the AR shopping guide interaction scene, generating interaction feedback information and updating the interaction data set to trigger iterative optimization of the interaction process includes: Monitoring the user's operation behavior of the interaction triggering mode in the AR shopping guide interaction scene, wherein the operation behavior includes gesture triggering action, voice command content and sight focus stay position; Perform response status recognition processing on the operation behavior, determine whether the user successfully triggers the information display or function call of the AR shopping guide interactive content, and generate an interaction success status identifier; When the interaction success status is marked as successful, extracting the information display level or function call type corresponding to the user trigger operation, and recording the user's access depth to the multi-level information display structure and the selection preference for the function call type; When the interaction success status is marked as failure, analyzing the matching deviation between the operation behavior and the trigger area calibration parameter, and recording the position offset and time delay parameter of the user trigger operation; Based on the interaction success status identifier, access depth, selection preference, position offset and time delay parameter, generating interaction feedback information including interaction effectiveness evaluation results and optimization improvement directions; Adding the operation behavior, interaction success status identifier, access depth, selection preference, position offset and time delay parameter to the interaction data set to form an updated interaction data set; The updated interaction data set is input into the interaction process optimization module, and through comparative analysis of historical interaction data and current interaction data, the association rules of the user intention analysis processing, the generation logic of the AR shopping guide interaction content and the parameter configuration of the spatial alignment fusion processing are adjusted to trigger iterative optimization of the interaction process.

10. An AR intelligent shopping guide interaction system applied to store management, characterized in that, It includes a processor and a memory, the memory is connected to the processor, the memory is used to store programs, instructions or codes, and the processor is used to execute the programs, instructions or codes in the memory to implement the AR smart shopping guide interaction method applied to store management as described in any one of claims 1 to 9 above.

Images