Virtual shopping mall navigation system based on VR technology

Abstract

The application discloses a virtual shopping mall navigation system based on VR technology and concretely relates to the technical field of virtual shopping mall navigation, wherein the real-time perception of the collision detection failure coefficient and the state synchronization delay coefficient in the virtual shopping mall navigation system is carried out, and a path segment level through-modeling hidden danger heat map is constructed based on this, potential through-modeling risks of each segment on the path are identified and quantified, and the initial path is iteratively corrected with the risk minimization as the target, which not only effectively avoids the cross through-modeling area, but also balances the path length and the navigation efficiency, realizes the whole-process closed-loop control from the path generation, the risk perception to the iterative optimization and the visual presentation, and thus can significantly improve the reliability and the sense of immersion of the navigation in the complex VR environment with multiple users and high concurrency.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual shopping mall navigation, more particularly, the present application relates to a virtual shopping mall navigation system based on VR technology. BACKGROUND

[0002] With the rapid development of virtual reality (VR) technology, virtual shopping malls, as an important carrier of immersive interactive experience, have been widely used in retail, exhibition, and shopping guide, etc. In such systems, users usually enter a highly simulated three-dimensional virtual shopping mall space through VR devices to interact with the environment and others in the form of avatars. In order to enhance the spatial realism and social immersion of users, virtual shopping mall systems generally use a multi-user synchronization mechanism to enable multiple user avatars to coexist and interact in the same virtual space.

[0003] However, in high concurrency or high load running scenarios, VR systems usually use a local priority optimization strategy to balance computing performance and bandwidth resources. Among them, the local user avatar is considered as the main view reference object, and its position and action information is kept at high precision rendering and high frequency updating; while other remote user avatars are usually "degraded" by the system, such as reducing the action refresh frequency, simplifying the bone posture calculation or using interpolation extrapolation for delay compensation. This optimization mechanism can alleviate network delay and rendering pressure to some extent, but due to the inconsistency of state synchronization delay and physical detection in a distributed environment, the state perception of different user local systems to other user avatars has time lag and spatial error, causing collision detection failure between avatars and state synchronization delay, resulting in "model crossing" phenomena such as crossing and overlapping of each other's virtual avatars. This phenomenon seriously damages the user's perception of spatial continuity and physical reality, and reduces the overall immersion experience quality of the system, especially in key application scenarios such as path navigation, close-range interaction or product recommendation. SUMMARY

[0004] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a virtual shopping mall navigation system based on VR technology to solve the problems raised in the background art.

[0005] To achieve the above-mentioned purpose, the present application provides the following technical solutions:

[0006] The virtual shopping mall navigation system based on VR technology comprises a user initialization module, a navigation path planning module, a model crossing hazard perception module, a path dynamic optimization module, and a navigation visualization module.

[0007] A user initialization module is configured to construct a virtual mall space based on a VR three-dimensional modeling technology, and to load a virtual avatar of a current online user in real time, and initialize spatial position, posture data, and motion data of the virtual avatar;

[0008] A navigation path planning module is configured to generate an initial navigation path from a current user position to a target commodity area in response to a commodity selection instruction of the user;

[0009] A module for sensing a threading risk is configured to acquire collision detection failure information of the user avatar and state synchronization delay information of a remote avatar and a local view during execution of the navigation path, the collision detection failure information including a collision detection failure coefficient, and the state synchronization delay information including a state synchronization delay coefficient;

[0010] A path dynamic optimization module is configured to construct a threading risk heat map according to the collision detection failure coefficient and the state synchronization delay coefficient, and to iteratively correct the initial navigation path;

[0011] A navigation visualization module is configured to present the final corrected navigation path to the user in a multi-modal manner.

[0012] In a preferred embodiment, the collision detection failure coefficient is acquired according to the following logic:

[0013] For each user virtual avatar i, at each timestamp t, the following is recorded:

[0014] A local observation state vector , wherein is a position vector of the user virtual avatar i at time t, is a velocity vector of the user virtual avatar i at time t, is a posture vector of the user virtual avatar i at time t;

[0015] A system prediction state vector , wherein is a time advance amount predicted by the system;

[0016] A local prediction deviation vector is calculated;

[0017] A global perception difference map is constructed , wherein is a user virtual avatar node set, each node being a user virtual avatar, is an edge set;

[0018] For any user virtual avatar node pair , a state difference degree of i from the perspective of j is calculated: , wherein is the state difference degree, is the system predicted state vector of i in the local perspective of user j;

[0019] calculating the similarity weight of user virtual avatar node edge: wherein is the similarity weight of user virtual avatar node edge;

[0020] calculating the node inconsistency entropy: wherein is the node inconsistency entropy, is the neighbor node set of user virtual avatar node i;

[0021] calculating the collision detection failure coefficient: wherein is the collision detection failure coefficient, is the user set that has spatial interaction with the current path, is the average node inconsistency entropy of user virtual avatar node g within the current path range, wherein is the node inconsistency entropy of user virtual avatar node g, , is the residence time of user virtual avatar node g within the current path range, is the minimum spatial distance from user virtual avatar node i to the midpoint of the current path, is a minimum constant to prevent division by zero.

[0022] In a preferred embodiment, the acquisition logic of the state synchronization delay coefficient is as follows:

[0023] For any user virtual avatar u, at system timestamp t, the cross-perspective residual tensor is calculated according to the received M remote user perspective synchronization state packets: wherein is the cross-perspective residual tensor of user virtual avatar u, is the state synchronization error vector of remote user virtual avatar v to local user virtual avatar u, is the local observation state vector of user virtual avatar u, is the system predicted state vector of user virtual avatar u in the perspective of user virtual avatar v after a delay ;

[0024] Acquiring the time decay weight of user synchronization state under a preset time window T: wherein is the time decay weight, ;

[0025] Performing time weighting on the states within the window: ,in These are the time-weighted cross-view residual tensor coefficients;

[0026] Environmental factors are introduced to model the interference in the synchronization state. These environmental factors include the number of other users within a unit spatial range in the current time window. An environmental coupling factor is defined as follows: ,in For the user's virtual avatar u, the environmental coupling factor, This represents the number of other users within a unit of space during the current time window.

[0027] Calculate the state synchronization delay coefficient: ,in This is the state synchronization delay coefficient.

[0028] In a preferred embodiment, a heat map of the potential for clipping is constructed based on the collision detection failure coefficient and the state synchronization delay coefficient, as follows:

[0029] Initial navigation path Divide into K equal path segments: ,in Let k be the kth path segment, where k = {1, 2, ..., K}, and K is the total number of path segments.

[0030] For each path segment Get the path segment A collection of all online user virtual avatars within a meter range : ,in Virtual avatars for users With path segment The shortest distance;

[0031] For each path segment Path segments are constructed based on the collision detection failure coefficient and state synchronization delay coefficient of the user's virtual avatar within the interaction area. Wear-through risk value: ,in For path segment The risk value of wearing out. Virtual avatars for users The collision detection failure coefficient, Virtual avatars for users State synchronization delay coefficient, These represent the preset proportional coefficients for the collision detection failure coefficient and the state synchronization delay coefficient, respectively. All are greater than 0. This is a very small constant to prevent division by zero;

[0032] Construct a threading risk hidden danger heat map according to the threading risk value of each path segment : .

[0033] In a preferred embodiment, the threading risk value in the threading risk hidden danger heat map is compared with a preset threading risk value threshold, and the initial navigation path is iteratively corrected as follows:

[0034] If the threading risk value is greater than the threading risk value threshold, an iterative correction signal is generated, the path search algorithm is called to regenerate the navigation path from the current user position to the target commodity area, and a minimization cost function is defined as the iterative condition for iteratively correcting the initial navigation path, and the minimization cost function is: , wherein is the path length of the navigation path, , respectively represent the preset proportion coefficients of the path length and the threading risk value, and , are both greater than 0;

[0035] If the threading risk value is less than or equal to the threading risk value threshold, the iterative correction signal does not need to be generated.

[0036] Technical effects and advantages of the present application:

[0037] 1. The present application realizes real-time perception of the collision detection failure coefficient and the state synchronization delay coefficient in the virtual shopping mall navigation system, refines the potential cross-threading problem caused by the inconsistency between avatars, and constructs a threading risk hidden danger heat map at the path segment level based on this, identifies and quantifies the potential threading risk of each segment on the path, and iteratively corrects the initial path with the goal of minimizing risk. Not only effectively avoids the cross-threading area, but also balances the path length and navigation efficiency, realizes the whole process closed loop control from path generation, risk perception to iterative optimization and visual presentation, and thus can significantly improve the reliability and immersion of navigation in a complex VR environment with multiple users and high concurrency. Specifically, with the help of the collision detection failure coefficient and the state synchronization delay coefficient, the system accurately identifies the potential threading high-risk area on each path segment of the user's movement, and dynamically avoids the initial path through multiple rounds of iterative correction, thereby effectively avoiding abnormal situations such as virtual avatar threading and overlapping that destroy the continuity of space. At the same time of improving the accessibility and robustness of VR navigation path, it guarantees the physical consistency of multi-user collaborative interaction, greatly enhances the user trust and stickiness of virtual shopping malls, and has obvious promoting effect on the immersive experience quality of retail, e-commerce, exhibition and other scenes. BRIEF DESCRIPTION OF DRAWINGS

[0038] For the convenience of those skilled in the art to understand, the present application will be further described below in conjunction with the drawings;

[0039] Figure 1 Flowchart of the system of the embodiment of the present application. DETAILED DESCRIPTION

[0040] The technical solutions in the embodiments of the present application will be described clearly and completely below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the present application.

[0041] Embodiment: The present application provides a virtual shopping mall navigation system based on VR technology as shown in the figure, which comprises a user initialization module, a navigation path planning module, a module for sensing hidden dangers in model penetration, a path dynamic optimization module and a navigation visualization module. Figure 1

[0042] The user initialization module is used to construct a virtual shopping mall space based on VR three-dimensional modeling technology, and to load a virtual avatar of a current online user in real time, and to initialize spatial position, attitude data and motion data of the virtual avatar.

[0043] The navigation path planning module is used to generate an initial navigation path from a current user position to a target commodity area in response to a commodity selection instruction of the user.

[0044] The module for sensing hidden dangers in model penetration is used to acquire collision detection failure information of the user avatar and state synchronization delay information of a remote avatar and a local view during execution of the navigation path, wherein the collision detection failure information comprises a collision detection failure coefficient, and the state synchronization delay information comprises a state synchronization delay coefficient.

[0045] The path dynamic optimization module is used to construct a hidden danger heat map in model penetration according to the collision detection failure coefficient and the state synchronization delay coefficient, and to iteratively correct the initial navigation path.

[0046] The navigation visualization module is used to present the final corrected navigation path to the user in a multi-modal manner.

[0047] The user initialization module is used to construct a virtual shopping mall space based on VR three-dimensional modeling technology, and to load a virtual avatar of a current online user in real time, and to initialize spatial position, attitude data and motion data of the virtual avatar, specifically as follows:

[0048] ​Load a pre-designed three-dimensional scene model (such as a shopping mall structure, shelves, goods, etc.), use a unified coordinate system (such as a right-handed coordinate system) for space definition; use Unity3D, Unreal Engine, etc. Engine to register the space of object nodes and establish the hierarchical structure tree of scene objects; process the navigation grid for passable areas, impassable areas, and display areas; mark the user's birth point (entry), guide point (path reference anchor point), and target product point (product interaction anchor point); assign a standard virtual avatar model to each new user, which includes skeleton nodes, animation components (Animator), and action state machines; position the virtual avatar of the new user to the shopping mall entrance area or the user's last login location, initialize the default orientation and initial view angle to ensure that the user faces the navigation area; obtain real-time motion data from VR devices (such as head-mounted displays, hand controllers, full-body motion capture devices), including: HMD position and attitude, left / right hand controller three-axis acceleration and direction, relative positions of each joint, bind the obtained motion data to the corresponding nodes of the virtual avatar, realize real-time pose mapping and action driving; bind user interaction behaviors (such as gestures, buttons, voice) and action events, such as "raising hands to greet" triggering social animations, "opening mouth" triggering voice animations, etc.; assign a unique virtual identity ID and communication channel to each user, bind the virtual avatar to the shopping mall scene, define its reachable area, interactive objects (such as clicking on products, approaching a promotional area to trigger a prompt), etc.

[0049] A navigation path planning module is used to respond to the user's product selection instruction, call the path search algorithm to generate an initial navigation path from the current user's location to the target product area, as follows:

[0050] The user selects a product by clicking, voice, gesture, etc. The system parses the unique ID and spatial anchor point coordinates of the target product from the product database; identifies the shelf number, floor location, and region label of the target product to assist in navigation; takes the spatial center point of the current virtual avatar (usually the foot or torso projection point) as the starting point of the path; according to the current user's location to the target product location, call the path search algorithm, including but not limited to: breadth-first search algorithm, depth-first search algorithm, Dijkstra algorithm, Floyd algorithm;

[0051] Convert the search result into a path point sequence , where is the midpoint vector of the path, including coordinates, angles, and movement patterns; that is, an initial navigation path from the current user's location to the target product area is obtained.

[0052] The module for perceiving the risk of model penetration is configured to acquire collision detection failure information and state synchronization delay information of the remote avatar and the local view during the execution of the navigation path, the collision detection failure information including a collision detection failure coefficient, and the state synchronization delay information including a state synchronization delay coefficient.

[0053] The collision detection failure coefficient is a measurement index for measuring the failure degree of the collision detection mechanism between the virtual avatars in the virtual mall, which reflects the frequency and severity of physical contact between the avatars during navigation or interaction that the system fails to accurately identify and process. The coefficient not only reflects the adaptability of the current collision detection mechanism to the multi-user high-concurrency scene, but also indirectly reveals the stability and accuracy of the system's physical simulation and synchronization mechanism. A larger collision detection failure coefficient indicates that the collision perception mechanism of the system in this area or path segment has serious lag or accuracy problems, the system fails to effectively capture physical contact events, resulting in significant "model penetration" between avatars, which destroys the coherence and realism of the space and thus affects the immersive experience and interaction willingness of the user. Especially in a virtual mall, which is a complex environment with multiple interactions and multi-person synchronization, a higher collision detection failure coefficient often means that model overlap, skeletal dislocation, and action penetration may occur frequently between users, which may lead to a lack of trust in the physical rules and interaction boundaries of the virtual environment and reduce the commercial application value of the system. Conversely, a smaller collision detection failure coefficient indicates that the system's spatial collision recognition and feedback mechanism between users is running well, can capture the approach, contact, and intersection behavior between virtual avatars in real time, and trigger reasonable avoidance reactions based on the physical engine or rule model, thereby effectively preventing the occurrence of "model penetration", maintaining the stability of the spatial topology, and enhancing the social authenticity and environmental trust between users. This is crucial for maintaining the continuity of behaviors such as shopping guides, product consultations, and social interactions in the mall. Based on the collision detection failure coefficient, the system can quantitatively evaluate the risk of model penetration of virtual avatars, predict the risk hotspots of model penetration during path planning, and dynamically adjust the path to avoid high-collision-detection-failure-coefficient areas. In addition, the coefficient can be used as a performance feedback index for the physical engine to drive the adaptive upgrade or elastic scheduling mechanism of the underlying algorithm (such as dynamic switching of frame rate priority), thereby improving the robustness of the system.

[0054] The acquisition logic of the collision detection failure coefficient is as follows:

[0055] For each user virtual avatar i, at each timestamp t, record:

[0056] The local observation state vector wherein is the position vector of the user virtual avatar i at time t, is the velocity vector of the user virtual avatar i at time t, a pose vector of the user virtual avatar i at time t;

[0057] a system predicted state vector wherein a time lead of the system prediction;

[0058] a local prediction bias vector ;

[0059] constructing a global awareness difference graph wherein a set of user virtual avatar nodes, each node being a user virtual avatar, a set of edges;

[0060] for any pair of user virtual avatar nodes , calculating a state difference degree of i from the perspective of j: wherein a state difference degree, a system predicted state vector of user j in its local perspective of i;

[0061] calculating a similarity weight of the edge of the user virtual avatar nodes: wherein a similarity weight of the edge of the user virtual avatar nodes;

[0062] calculating a node inconsistency entropy: wherein a node inconsistency entropy, a set of neighbor nodes of the user virtual avatar node i;

[0063] The greater the node inconsistency entropy, the greater the state difference of user i “seen” by different users, i.e., the more inconsistent the state awareness;

[0064] calculating a collision detection failure coefficient: wherein a collision detection failure coefficient, a set of users having spatial interaction with the current path, an average node inconsistency entropy of the user virtual avatar node g within the current path range, wherein a node inconsistency entropy of the user virtual avatar node g, , a residence time of the user virtual avatar node g within the current path range, a minimum spatial distance of the user virtual avatar node i to the midpoint of the current path, a minimum constant to prevent division by zero (generally taken );

[0065] It should be noted that the above formulas are all dimensionless values calculated, and the commonly used dimensionless methods include Min-Max normalization, Z-Score standardization, etc., which will not be repeated here;

[0066] The state synchronization delay coefficient in the application is a key index for quantifying the state synchronization lag degree of different user avatars in a network environment in a distributed virtual reality system. The coefficient reflects the time difference between the actual state of the remote user avatar observed by the local user and the synchronization state and the consistency deviation. A larger state synchronization delay coefficient indicates that the state data received by the remote user avatar at the current node has a significant time lag or interpolation error, resulting in a serious deviation between the virtual behavior and the real operation. A smaller state synchronization delay coefficient indicates that the remote state synchronization process is timely and accurate, and the virtual avatar rendered by the system is highly consistent with the user behavior, and the perception continuity is good. In the virtual mall navigation environment constructed based on VR technology, multiple user avatars need to perform collaborative navigation and interaction behaviors in the same virtual space, including parallel movement, path intersection, target guidance, etc. Due to the close space and high intersection of the moving paths between virtual avatars, the system has higher requirements for the synchronization frequency and accuracy of key state information such as position, speed and attitude. However, in a high-concurrency user scenario, the system often considers performance load optimization and only synchronizes the remote user state at a low frequency, or uses interpolation prediction to make up for network delay, thereby causing phenomena such as "position lag", "speed inconsistency" and "direction drift" in the local view of the user. The core cause of such phenomena is the distributed view difference caused by state synchronization delay, which is particularly obvious in near-field interaction and can easily cause "model penetration" between virtual avatars, i.e., two users do not detect contact in their local view, but due to state synchronization delay, the virtual bodies of each other inevitably overlap or intersect in space. By introducing the state synchronization delay coefficient, the application provides a measurable, feedbackable and predictable model penetration risk evaluation method. The state synchronization delay coefficient not only reflects the real-time performance index of the system distributed synchronization mechanism, but also becomes a risk prediction signal in the virtual interaction process between users. Its introduction changes the model penetration risk from "passive occurrence" to "perceptible and controllable" warning, providing an important technical guarantee for improving the stability, immersion and safety of the virtual mall navigation system, and has significant engineering application value and expansion potential.

[0067] The acquisition logic of the state synchronization delay coefficient is as follows:

[0068] For any user virtual avatar u, at system timestamp t, the cross-view residual error tensor is calculated according to the received M remote user view synchronization state packets: wherein a cross-view residual tensor for the user virtual avatar u, a state synchronization error vector for the remote user virtual avatar v to the local user virtual avatar u, a local observation state vector for the user virtual avatar u, a delayed system predicted state vector for the user virtual avatar u after the state synchronization;

[0069] a time decay weight for the user synchronization state in a preset time window T: wherein is the time decay weight, ;

[0070] performing time weighting on the states within the window: wherein is the time-weighted cross-view residual tensor coefficient;

[0071] introducing environmental factors to interfere with the synchronization state, the environmental factors including the number of other users within a unit space range within the current time window, defining an environmental coupling factor: wherein is the environmental coupling factor of the user virtual avatar u, is the number of other users within a unit space range within the current time window;

[0072] calculating a state synchronization delay coefficient: wherein is the state synchronization delay coefficient;

[0073] It should be noted that the above formulas are all dimensionless values, and common dimensionless methods include Min-Max normalization, Z-Score standardization, etc., which are not described here;

[0074] a path dynamic optimization module, configured to construct a penetration risk heat map according to the collision detection failure coefficient and the state synchronization delay coefficient, specifically as follows:

[0075] dividing an initial navigation path into K path segments: wherein is the kth path segment, k={1, 2,..., K}, and K is the total number of path segments;

[0076] for each path segment , obtaining a set of all online user virtual avatars within a range of meters of the path segment : wherein Virtual avatar for user and the shortest distance of the path segment ;

[0077] For each path segment , the penetration risk value of the path segment is constructed according to the collision detection failure coefficient and the state synchronization delay coefficient of the user virtual avatar in the interaction area: , wherein is the penetration risk value of the path segment , is the collision detection failure coefficient of the user virtual avatar , is the state synchronization delay coefficient of the user virtual avatar , respectively represent the preset proportion coefficients of the collision detection failure coefficient and the state synchronization delay coefficient, and both are greater than 0, is a very small constant (usually ) to prevent division by zero;

[0078] It should be noted that the preset proportion coefficients of the various indicators are set according to actual conditions, for example, expert weighting method is adopted, that is, experts in the relevant field are invited to determine the preset proportion coefficients of the various indicators through professional opinion investigation and comprehensive evaluation, for example, may be 0.5, 0.5;

[0079] The penetration risk hidden danger heat map is constructed according to the penetration risk value of each path segment: The penetration risk hidden danger heat map represents the risk intensity of the penetration risk hidden danger existing in each segment of the entire path;

[0080] The penetration risk value in the penetration risk hidden danger heat map is compared with the preset penetration risk value threshold, and the initial navigation path is iteratively corrected, as follows:

[0081] If the penetration risk value is greater than the penetration risk value threshold, it means that the penetration risk of the current path segment is too high, an iterative correction signal is generated, the path search algorithm is called to regenerate the navigation path from the current user location to the target commodity area, and a minimization cost function is defined as the iterative condition for iteratively correcting the initial navigation path, and the minimization cost function is: , wherein is the path length of the navigation path, , respectively represent the preset proportion coefficients of the path length and the penetration risk value, and , both are greater than 0;

[0082] It should be noted that when the iterative correction signal is generated, the path search algorithm is called to regenerate the navigation path from the current user position to the target commodity area, and the running process of the threading hidden danger perception module->path dynamic optimization module needs to be re-executed; the above formulas are all dimensionless values, and common dimensionless methods include Min-Max normalization, Z-Score standardization, etc., which will not be repeated here; 、 According to the actual situation, for example, the expert weighting method is adopted, that is, experts in the relevant field are invited to determine the preset proportion coefficient of each index through professional opinion investigation and comprehensive evaluation, for example, 、 It can be 0.5, 0.5;

[0083] If the threading risk value is less than or equal to the threading risk value threshold, it means that the threading risk of the current path segment is low and is within a controllable range, and no iterative correction signal needs to be generated.

[0084] The navigation visualization module is used to present the final corrected navigation path to the user in a multi-modal manner;

[0085] The multi-modal manner includes three-dimensional light trail guidance, ground virtual arrow, and field voice prompt.

[0086] Three-dimensional light trail guidance refers to a light trail band extending along the navigation path, that is, dynamic particles or light texture voxels are uniformly laid on the path curve to realize forward flowing animation, and the light trail color is visually encoded according to the threading risk intensity (such as red-yellow-green gradient).

[0087] The ground virtual arrow refers to dynamically generating a directional pointing arrow on the ground projection layer to help the user adjust the orientation in real time. For each path segment, a set of virtual arrows pointing to the end point are generated at the start point. When the user is too close, the arrow size is dynamically scaled to avoid visual obstruction.

[0088] The field voice prompt refers to providing time-sequenced voice navigation prompts based on the user's current view angle and path segment information.

[0089] Three-dimensional light trail is the main navigation channel, ground virtual arrow is used for immediate fine tuning, and field voice prompt undertakes key path event prompt.

[0090] The application can realize the whole-process closed-loop control from path generation, risk perception to iterative optimization and visual presentation, and can significantly improve the reliability and immersion of navigation in a complex VR environment with multiple users and high concurrency. Specifically, by means of the collision detection failure coefficient and the state synchronization delay coefficient, the system can accurately identify the high-risk areas of potential crossing on each path of user movement, and dynamically avoid the initial path through multiple rounds of iterative correction, thereby effectively avoiding abnormal situations such as avatar crossing and overlapping that destroy the sense of continuous space. In this way, the reachability and robustness of the VR navigation path are improved, the physical consistency of multi-user collaborative interaction is ensured, and the user trust and stickiness of the virtual mall are greatly enhanced, which has a significant promoting effect on the immersive experience quality of retail, e-commerce, exhibition and other scenes.

[0091] The above formulas are dimensionless numerical calculations, and the formulas are obtained by software simulation of a large amount of data to obtain a formula of the latest real situation. The preset parameters in the formula are set by a person skilled in the art according to the actual situation.

[0092] The above embodiments can be realized by software, hardware, firmware or any combination thereof. When realized by software, the above embodiments can be realized in the form of a computer program product in whole or in part. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the flow or function described in the embodiments of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network or other programmable devices. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another, for example, the computer instructions can be transmitted from one website, computer, server or data center to another website, computer, server or data center through wired or wireless (such as infrared, wireless, microwave, etc.) mode. The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server, data center and the like containing one or more available medium sets. The available medium can be a magnetic medium (such as a floppy disk, a hard disk, a magnetic tape), an optical medium (such as a DVD) or a semiconductor medium. The semiconductor medium can be a solid-state disk.

[0093] It should be understood that the size of the sequence number of each process described above in various embodiments of the present application does not mean the order of execution, and the execution order of each process should be determined according to its function and inherent logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.

[0094] The functions described above, if implemented in the form of software function units and sold or used as independent products, can be stored in a computer readable storage medium. Based on such understanding, the technical solutions of the present application essentially or the part that contributes to the prior art or the part of the technical solutions can be embodied in the form of a software product. The computer software product is stored in a storage medium and includes a number of instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk or an optical disk, and various media that can store program codes.

[0095] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art can easily think of changes or replacements within the technical scope disclosed in the present application, which should be covered within the protection scope of the present application. Therefore, the protection scope of the present application should be subject to the protection scope of the claims.

Claims

1. A virtual shopping mall navigation system based on VR technology, characterized in that: It includes a user initialization module, a navigation path planning module, a clipping hazard detection module, a path dynamic optimization module, and a navigation visualization module; The user initialization module is used to construct a virtual shopping mall space based on VR 3D modeling technology, and load the virtual avatar of the currently online user in real time, initializing the spatial position, posture data and motion data of the virtual avatar; The navigation path planning module is used to respond to the user's product selection instructions and call the path search algorithm to generate an initial navigation path from the current user location to the target product area; The clipping hazard perception module is used to obtain collision detection failure information of the user avatar and state synchronization delay information between the remote avatar and the local view during the execution of the navigation path. The collision detection failure information includes a collision detection failure coefficient, and the state synchronization delay information includes a state synchronization delay coefficient. The path dynamic optimization module is used to construct a heat map of the potential for clipping based on the collision detection failure coefficient and the state synchronization delay coefficient, and to iteratively correct the initial navigation path. The navigation visualization module is used to present the final revised navigation path to the user in a multimodal manner.

2. The virtual shopping mall navigation system based on VR technology according to claim 1, characterized in that: The logic for obtaining the collision detection failure coefficient is as follows: For each user virtual avatar i, at each timestamp t, record: Local observation state vector ,in Let i be the position vector of the user's virtual avatar at time t. The velocity vector of the user's virtual avatar i at time t. The pose vector of the user's virtual avatar i at time t; System predicted state vector ,in This refers to the lead time for system prediction; Calculate the local prediction bias vector ; Construct a globally perceived difference map ,in This is a collection of user virtual avatar nodes, where each node represents a user virtual avatar. Let it be the set of edges; For any pair of user virtual avatar nodes Calculate the state difference of i from the perspective of j: ,in For state difference degree, For user j, the system prediction state vector for i from their local perspective; Calculate the similarity weights of edges between user avatar nodes: ,in Similarity weights for user avatar node edges; Calculate the node inconsistency entropy: ,in The inconsistency entropy of nodes. The set of neighboring nodes of user virtual avatar node i; Calculate the collision detection failure coefficient: ,in The collision detection failure coefficient, The set of users who have spatial interactions with the current path. The average node inconsistency entropy for the user's virtual avatar node g within the current path range. ,in The node inconsistency entropy for the user's virtual avatar node g. , The duration of the user's virtual avatar node g within the current path range. The minimum spatial distance from user virtual avatar node i to the midpoint of the current path. This is a very small constant used to prevent division by zero.

3. The virtual shopping mall navigation system based on VR technology according to claim 1, characterized in that: The logic for obtaining the state synchronization delay coefficient is as follows: For any user virtual avatar u, at system timestamp t, calculate the cross-view residual tensor based on the received synchronization state packets from the perspectives of M remote users: ,in The cross-perspective residual tensor for the user's virtual avatar u Let v be the state synchronization error vector between the remote user virtual avatar v and the local user virtual avatar u. The local observation state vector for the user's virtual avatar u. To delay in the user's virtual avatar's perspective The system then predicts the state vector for the user's virtual avatar u. Obtain the time decay weight of the user's synchronization status within a preset time window T: ,in For time decay weight, ; Time-weighted calculation of the state within the window: ,in These are the time-weighted cross-view residual tensor coefficients; Environmental factors are introduced to model the interference in the synchronization state. These environmental factors include the number of other users within a unit spatial range in the current time window. An environmental coupling factor is defined as follows: ,in For the user's virtual avatar u, the environmental coupling factor, This represents the number of other users within a unit of space during the current time window. Calculate the state synchronization delay coefficient: ,in This is the state synchronization delay coefficient.

4. The virtual shopping mall navigation system based on VR technology according to claim 1, characterized in that: A heat map of the potential for clipping is constructed based on the collision detection failure coefficient and the state synchronization delay coefficient, as follows: Initial navigation path Divide into K equal path segments: ,in Let k be the kth path segment, where k = {1, 2, ..., K}, and K is the total number of path segments. For each path segment Get the path segment A collection of all online user virtual avatars within a meter range : ,in Virtual avatars for users With path segment The shortest distance; For each path segment Path segments are constructed based on the collision detection failure coefficient and state synchronization delay coefficient of the user's virtual avatar within the interaction area. Wear-through risk value: ,in Path segment The risk value of wearing out. Virtual avatars for users The collision detection failure coefficient, Virtual avatars for users State synchronization delay coefficient, These represent the preset proportional coefficients for the collision detection failure coefficient and the state synchronization delay coefficient, respectively. All are greater than 0. This is a very small constant to prevent division by zero; A heat map of potential clipping hazards is constructed based on the clipping risk value of each path segment. : .

5. The virtual shopping mall navigation system based on VR technology according to claim 4, characterized in that: The initial navigation path is iteratively corrected by comparing the clipping risk value in the clipping risk heat map with the preset clipping risk threshold, as follows: If the risk value of the clipping pattern exceeds the clipping pattern risk threshold, an iterative correction signal is generated. The path search algorithm is then invoked to regenerate the navigation path from the current user location to the target product area. A minimum cost function is defined, and this minimum cost function is used as the iterative condition to iteratively correct the initial navigation path. The minimum cost function is as follows: ,in The path length of the navigation path. , These represent the preset proportional coefficients for path length and the risk of penetration, respectively. , All are greater than 0; If the penetration risk value is less than or equal to the penetration risk value threshold, there is no need to generate an iterative correction signal.

Images