Building hybrid map model construction method and device for indoor positioning and navigation

Abstract

The invention provides a building hybrid map model construction method and device for indoor positioning and navigation, and the method comprises the steps: obtaining building information model data, and constructing a solid model based on the building information model data; constructing a network model based on passable elements in the entity model; constructing a voxel model based on space elements in the entity model; and constructing a hybrid map model based on the entity model, the network model and the voxel model, obtaining original coordinate data output by the positioning system as a to-be-matched positioning point set, and mapping the to-be-matched positioning point set to the hybrid map model through map matching. By integrating the advantages of an entity model, a network model and a voxel model, a high-precision hybrid map model is constructed, and a positioning result is accurately mapped to the model through a map matching algorithm, so that the precision problem in indoor position service application is systematically solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of indoor positioning and navigation, and in particular to a building hybrid map model construction method and device for indoor positioning and navigation. BACKGROUND

[0002] With the increasing complexity and scale of indoor environment, the application demand of Location-Based Services (LBS) in indoor environment is increasingly prominent. As the data basis and information carrier of indoor services, constructing high-availability indoor map has become a key scientific problem urgently needed to be solved in current indoor LBS application. In recent years, hybrid model has been widely applied in the field of indoor map research. However, the current research on indoor map mainly focuses on the model construction method, and the precision problem in indoor location service application is insufficiently concerned. SUMMARY

[0003] The present application provides a building hybrid map model construction method and device for indoor positioning and navigation, to solve the defects of insufficient attention to the precision problem in indoor location service application in the prior art. The technical solution of the present application is as follows:

[0004] In a first aspect, the present application provides a building hybrid map model construction method for indoor positioning and navigation, comprising:

[0005] Obtaining building information model data, and constructing an entity model based on the building information model data;

[0006] Constructing a network model based on the passable elements in the entity model;

[0007] Constructing a voxel model based on the spatial elements in the entity model;

[0008] Constructing a hybrid map model based on the entity model, the network model and the voxel model, obtaining original coordinate data output by a positioning system as a set of matching positioning points, and mapping the set of matching positioning points to the hybrid map model through map matching.

[0009] Optionally, constructing an entity model based on the building information model data comprises:

[0010] Importing building geometric information and semantic information based on the building information model data, and outputting original modeling data;

[0011] Pretreating the original modeling data to obtain intermediate model data;

[0012] According to indoor location service requirements, extracting basic elements and identification elements from the intermediate model data to obtain a basic model.

[0013] The semantic information is input into the basic model to obtain an entity model; the semantic information at least includes a room name, a functional area, a device facility, and multimedia data.

[0014] Optionally, the passable element includes a corridor, a floor entrance, and a staircase;

[0015] A network model is constructed based on the passable element in the entity model, including:

[0016] A center point of each floor corridor is extracted and connected to generate a single-layer skeleton network;

[0017] Based on the single-layer skeleton network, each target node is connected in sequence, and the target node and the corridor center point are connected to form a single-layer road network containing a horizontal path topology; the target node is a key functional node that needs to be connected in the network model, including a space connection point such as a door, a staircase entrance, and an elevator hall, and a user accessible target point such as a room entrance, a service facility, and a navigation sign;

[0018] The single-layer road network is used for vertical road network integration to obtain a network topology integrating horizontal paths and vertical paths; wherein the vertical road network integration at least includes connecting each floor to a connected node representing the entrance of the floor;

[0019] The semantic information anchor point and the map positioning anchor point are implanted in the network topology to obtain a network model;

[0020] The semantic information anchor point is implanted by the following way:

[0021] The indoor facility is abstracted and processed into a facility node, each facility node is configured with multi-element semantic information to obtain a semantic information anchor point, and the semantic information anchor point is added to the network topology; the multi-element semantic information at least includes text, picture, voice, and video data;

[0022] The map positioning anchor point is a walking anchor node, and the map positioning anchor point is implanted by the following way:

[0023] A preset pedestrian movement characteristic parameter is obtained;

[0024] Based on the network topology, the preset pedestrian movement characteristic parameter is selected as an interval standard of the walking anchor node in a horizontal direction; and in a vertical direction, the walking anchor node is arranged on each staircase step for a staircase, and the preset pedestrian movement characteristic parameter is also selected as an interval standard of the walking anchor node for a staircase platform.

[0025] Optionally, a voxel model is constructed based on the space element in the entity model, including:

[0026] Perform spatial classification processing on the spatial elements of the entity model according to the space type and navigability, and output the classified model;

[0027] Perform voxelization preprocessing on the classified model to obtain a voxelization-preprocessed model;

[0028] Perform scan filling processing on each spatial surface of the voxelization-preprocessed model using a scan filling method to obtain a surface voxel label result, and perform internal filling on the voxelization-preprocessed model according to the surface voxel label result, and output a voxel model.

[0029] Optionally, a hybrid map model is constructed based on the entity model, the network model and the voxel model, including:

[0030] Establish a spatial connection relationship between the entity model, the network model and the voxel model, and a mapping relationship between the network model nodes and the voxel model units, and output an integrated hybrid map model; the spatial connection relationship includes direct connection and indirect connection;

[0031] Establishing direct connection includes: establishing a one-to-one correspondence between a component node in the network model and a corresponding building component in the entity model, and establishing an association between a space set in the voxel model and all building components constituting the space in the entity model;

[0032] Establishing indirect connection includes: associating a plurality of components in the entity model through the voxel model by a room node in the network model, and associating a network model node through an entity model component by a voxel model unit.

[0033] Optionally, the original coordinate data output by the positioning system is obtained as a set of matching positioning points, and the set of matching positioning points is mapped to the hybrid map model through map matching, including:

[0034] Obtain the original coordinate data output by the positioning system as a set of matching positioning points;

[0035] Convert the set of matching positioning points to a model coordinate system based on the coordinate system conversion parameters of the hybrid map model, and output standardized positioning data;

[0036] Process the standardized positioning data using a buffer matching method to output a map matching result data set; the buffer matching method processing includes: establishing a buffer zone centered on the matching positioning point, and extracting all network nodes and voxel nodes in the buffer zone; traversing all network nodes and voxel nodes in the buffer zone to determine the network node and the voxel node closest to the matching positioning point as the corresponding map matching result.

[0037] In a second aspect, the present application also provides a building hybrid map model construction device for indoor positioning and navigation, including the following modules:

[0038] an entity model construction module configured to acquire building information model data and construct an entity model based on the building information model data;

[0039] a network model construction module configured to construct a network model based on passable elements in the entity model;

[0040] a voxel model construction module configured to construct a voxel model based on spatial elements in the entity model;

[0041] a hybrid model construction module configured to construct a hybrid map model based on the entity model, the network model and the voxel model, acquire raw coordinate data output by a positioning system as a set of to-be-matched positioning points, and map the set of to-be-matched positioning points to the hybrid map model through map matching.

[0042] In a third aspect, the present application also provides an electronic device, which comprises a memory, a processor, and a computer program stored in the memory and running on the processor, and the processor implements the method for constructing a hybrid map model of a building for indoor positioning and navigation according to the first aspect when executing the computer program.

[0043] In a fourth aspect, the present application also provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the method for constructing a hybrid map model of a building for indoor positioning and navigation according to the first aspect.

[0044] In a fifth aspect, the present application also provides a computer program product, which comprises a computer program, and the computer program is executed by a processor to implement the method for constructing a hybrid map model of a building for indoor positioning and navigation according to the first aspect.

[0045] Based on the above technical solutions, the present application has the following beneficial effects compared with the prior art:

[0046] The application provides a building hybrid map model construction method and device for indoor positioning and navigation. The method comprises the following steps: acquiring building information model (BIM) data, and constructing an entity model based on the data. The entity model not only captures the precise geometric shape of the building, but also integrates rich semantic information. This integration provides a solid data foundation for subsequent positioning and navigation, ensuring the accuracy and interpretability of the positioning results. The entity model is further refined to improve the description of indoor space and enhance the accuracy of positioning. A network model is constructed based on the passable elements in the entity model (such as corridors, staircases, elevators, etc.), which clearly defines the passable paths and connectivity within the building. This modeling approach helps optimize path planning algorithms and ensures that the positioning results accurately map to the actual passable paths, reducing positioning errors caused by unclear paths. A voxel model is constructed based on the spatial elements in the entity model (such as halls, corridors, rooms, etc.). The spatial elements are subdivided into small voxel blocks, each of which has unique spatial characteristics and location identifiers. This detailed description significantly improves the accuracy of indoor position matching and effectively reduces the offset error in traditional matching methods. The voxel model serves as an auxiliary matching tool for positioning results, providing more accurate and reliable indoor navigation services for users. By integrating the advantages of entity models, network models, and voxel models, a hybrid map model is constructed. This integration not only improves the richness and accuracy of the map model but also enhances the system's adaptability to complex indoor environments. After obtaining the raw coordinate data output by the positioning system as the set of matching points, the map matching algorithm is used to map these points to the hybrid map model. Due to the integration of multi-level data information in the hybrid map model, the map matching process can more accurately reflect the user's actual location, significantly improving the accuracy of indoor positioning.

[0047] Other features and advantages of the present application will be set forth in the descriptions below, and in part will be apparent from the description, or can be learned by practice of the present application. The purposes and other advantages of the present application will be realized and attained by the structures particularly pointed out in the description and the appended drawings.

[0048] In order to make the above-mentioned objectives, characteristics and advantages of the present application more apparent and easy to understand, the following preferred embodiments are specifically described below, and the accompanying drawings are used for detailed description. BRIEF DESCRIPTION OF DRAWINGS

[0049] In order to more clearly illustrate the technical solutions of the present application or prior art, the following will briefly introduce the drawings needed to be used in the embodiments or prior art description. Obviously, the drawings in the following description are some embodiments of the present application, and for those skilled in the art, other drawings can also be obtained without creative labor.

[0050] Figure 1 is a flowchart of a building hybrid map model construction method for indoor positioning and navigation provided by the present application.

[0051] Figure 2 is a network model abstraction table diagram of a single corridor space provided by the present application.

[0052] Figure 3 is an MLA anchor point diagram in a network model provided by the present application.

[0053] Figure 4 is a scanning filling method diagram provided by the present application.

[0054] Figure 5 is a triangular boundary voxel filling diagram provided by the present application.

[0055] Figure 6 is a direct connection diagram in a model provided by the present application.

[0056] Figure 7 is an indirect connection diagram in a model provided by the present application.

[0057] Figure 8 is a file storage structure diagram provided by the present application.

[0058] Figure 9 is a buffer matching method diagram provided by the present application.

[0059] Figure 10 is a building map overall scheduling loading scheme diagram provided by the present application.

[0060] Figure 11 is a surveying and mapping college F building modeling diagram provided by the present application.

[0061] Figure 12 is a lightweight processing effect diagram provided by the present application.

[0062] Figure 13 is a voxel model diagram of the surveying and mapping college F building provided by the present application.

[0063] Figure 14 is an experimental building part space voxelization comparison diagram provided by the present application.

[0064] Figure 15 is a building entity-network-voxel hybrid map construction result diagram of the surveying and mapping college F building provided by the present application.

[0065] Figure 16a 、 Figure 16b 、 Figure 16c 、 Figure 16dRespectively, the floor selection, indoor positioning, path planning, scene roaming interface schematic diagram of the mobile terminal positioning and navigation system provided by the application.

[0066] Figure 17 The structural schematic diagram of the building hybrid map model construction device for indoor positioning and navigation provided by the application.

[0067] Figure 18 The structural schematic diagram of the electronic device provided by the application. DETAILED DESCRIPTION

[0068] To make the objects, technical solutions and advantages of the present application clearer, the technical solutions in the present application will be described clearly and completely below in combination with the drawings in the present application. Obviously, the described embodiments are 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.

[0069] In modern society, with the rapid advancement of urbanization and the rapid development of economy, the lifestyle of human beings has changed significantly, and more and more daily activities have gradually shifted from outdoor to indoor. According to relevant survey data, people spend most of their time in indoor environments every day, and this proportion is particularly prominent among urban populations. However, with the accelerated advancement of urbanization and the continuous expansion of building scale, the complexity and diversity of indoor space are increasingly highlighted. Large shopping centers, airports, hospitals, museums, underground parking lots and other indoor places not only have complex structures, numerous regions and different functions, but also make the positioning and navigation needs of users in these environments become particularly urgent. In the face of complex indoor layout, traditional navigation methods are not up to the task, bringing many inconveniences to people's daily life and work. Therefore, it is crucial to develop an efficient and accurate building map. In the current research field of building map construction, the existing data models and data specifications still exhibit obvious limitations when applied to building map construction:

[0070] 1. The relative positioning of indoor road network elements has position deviation when used for map matching. In real-time positioning and navigation, there is often a problem that the actual trajectory of pedestrians does not completely coincide with the road network, at which time the actual position of pedestrians needs to be deviated to the road network, but this method will have a large matching error in the horizontal and vertical directions. Therefore, how to refine the indoor independent space, assist positioning and improve the accuracy of relative positioning is an important research direction of indoor location-based services at present.

[0071] 2. In the process of building map scheduling and visualization, the contradiction between quality and efficiency is inevitable due to the massive data of building models. Specifically, due to the limitation of hardware resources, high-precision map data is difficult to load and render efficiently, resulting in a decline in user experience. If the data is simplified to improve processing speed, the details and accuracy of the map will be sacrificed, which will affect the user's navigation and use effect. Therefore, it is urgent to seek breakthroughs through technological innovation and optimization algorithms to realize the visualization and use of buildings for multi-terminal (network terminal, mobile terminal).

[0072] The present application is based on BIM model, combined with the construction idea of BIMPN, extracts the semantic and topological information, and divides the indoor space elements based on this, and forms a voxel model by dividing the space elements. Then the connection relationship between the entity model, network model and space voxel is established by combining direct connection and indirect connection, realizing the construction of building entity-network-voxel mixed map model for indoor location service.

[0073] The building mixed map model construction method for indoor positioning and navigation provided by the present application aims to solve the precision problem in indoor positioning and navigation by constructing a building entity-network-voxel (ENV) mixed map model. This method integrates the advantages of entity model, network model and voxel model, and improves the precision and reliability of indoor positioning through multi-level data fusion and algorithm optimization. This method realizes the fine description of indoor space by voxelizing indoor independent space. Then, based on BIMPN, the connection between entity model, network model and voxel model is constructed to form an interactive building map mixed model. This method not only meets the demand for high-dimensional information expression of building indoor environment, but also significantly improves the position matching precision of building map in indoor positioning service application.

[0074] Reference Figure 1 The building mixed map model construction method for indoor positioning and navigation includes the following steps:

[0075] S110, obtain building information model data, and construct an entity model based on the building information model data.

[0076] In the process of constructing the entity model, first, the building information model (BIM) is taken as the core data basis, and the geometric and semantic information of the building is comprehensively extracted therefrom to provide accurate and rich data support for subsequent modeling. The specific construction steps include: importing the BIM data to obtain the original geometric shape and semantic attributes of the building, laying a foundation for model construction; data preprocessing is performed, and through denoising, alignment and simplification operations, the geometric consistency and quality of the data are ensured, thereby improving the accuracy and reliability of the model; element division and modeling are implemented, the entity model is subdivided into basic elements (such as wall, column, door and other structural components), identification elements (such as house number, display board and other positioning marks), special elements (such as bookshelf, chair and other specific functional facilities) and space elements (such as corridor space, hall space, house space, elevator space and the like), and fine modeling and optimization are performed for these elements to meet the needs of different application scenarios; semantic information is input, and key information such as room name, functional area division and equipment and facility configuration is recorded in detail to ensure that the semantic content contained in the model is comprehensive and accurate, and to provide users with an intuitive and meaningful navigation experience. Through this series of rigorous and systematic construction process, the entity model not only accurately reproduces the physical structure of the building, but also deeply integrates rich semantic information, thereby providing a solid data foundation and model support for indoor positioning and navigation services.

[0077] S120, constructing a network model based on the passable elements in the entity model.

[0078] In the network model construction process, the network construction method of BIMPN model is adopted, and the network model is constructed in detail based on the relationship structure between nodes. The specific construction process is as follows: basic road network drawing, strictly following the systematic idea of 'first main body, then detail, from horizontal to vertical', the center points of each floor corridor are connected first, and the basic skeleton structure on a single floor is constructed based on this; then, each target node in the floor is connected in detail one by one to ensure the integrity and accuracy of the road network; finally, in the vertical direction, the connected nodes on each floor are connected to complete the overall construction of the road network model on the whole building. In addition, in order to further improve the practicability and application breadth of the model, MLA anchors are introduced based on the road network model, including semantic information anchors (MLA(C)) and map positioning anchors (MLA(S)). The semantic information anchors are obtained by abstracting common facilities in indoor environment into nodes with specific semantic information, which provides necessary language information support for the application of indoor instance-based augmented reality (AR) technology; while the map positioning anchors are obtained by adding nodes at the edge of the map, which improves the accuracy of map matching and serves as the edge weight reference value for auxiliary navigation, thereby effectively improving the accuracy and practicability of the navigation system. Through this series of carefully designed construction steps, the network model not only realizes the accurate description of the complex road network inside the building, but also significantly enhances the practicability and application breadth of the model through the introduction of MLA anchors, providing strong support for indoor positioning and navigation services.

[0079] S130, constructing a voxel model based on the spatial elements in the entity model.

[0080] In the voxel model construction process, in view of the problem that the traditional matching method is prone to deviation error in horizontal and vertical directions, the concept of voxel model is innovatively proposed to improve the positioning matching accuracy by refining the indoor space description. The specific construction process is as follows:

[0081] ① Space element extraction:

[0082] Division basis: indoor space is divided in detail according to space type and navigability. Space type includes room, corridor, staircase, elevator, etc., and navigability is divided into non-navigable, navigable and semi-navigable.

[0083] Extraction target: focus on extracting navigable and semi-navigable spaces, which are the main consideration for positioning and navigation services, and ignore non-navigable spaces to reduce data processing amount.

[0084] ② Voxelization processing:

[0085] Preprocessing: select appropriate voxel size (such as 0.5x0.5x0.5m 3), to ensure the balance between the accuracy of the voxelization result and the computational efficiency. At the same time, a unified coordinate system is set to ensure the consistency of the voxelization process and reduce the computational load.

[0086] Surface voxelization: using the scan filling method, scanning along an axis (such as the y-axis), calculating the projection of each triangular facet of the model surface in the voxel space, and filling the voxels in the projected area. This method can efficiently convert the continuous three-dimensional model surface into a discrete voxel set.

[0087] Solid voxelization: after completing the surface voxelization, scanning the surface voxels in 6 directions (up, down, left, right, front, back) and filling the internal voxels to ensure that the entire solid model is completely voxelized. This step ensures that the voxel model not only contains surface information, but also contains internal structure information of the solid.

[0088] Through the above construction process, the voxel model can accurately describe the structural characteristics of the indoor space, providing high-precision spatial reference for indoor positioning and navigation. Compared with traditional matching methods, the voxel model effectively reduces the horizontal and vertical direction offset error, improves the positioning matching accuracy and reliability.

[0089] S140, based on the solid model, network model and voxel model, a hybrid map model is constructed, and the original coordinate data output by the positioning system is obtained as a set of matching positioning points, and the set of matching positioning points is mapped to the hybrid map model through map matching.

[0090] The application provides a building hybrid map model construction method for indoor positioning and navigation. By integrating the advantages of entity models, network models, and voxel models, a high-precision hybrid map model is constructed, and the positioning results are accurately mapped onto this model through a map matching algorithm, thereby systematically solving the precision problem in indoor location service applications. Specifically, by obtaining building information model (BIM) data and constructing an entity model based on these data, not only the precise geometric shape of the building is captured, but also rich semantic information (such as room names, functional areas, etc.) is integrated. This fusion provides a solid data foundation for subsequent positioning and navigation, ensuring the accuracy and interpretability of the positioning results. The entity model is finely modeled, including the division and modeling of basic elements (such as walls, columns, doors), identification elements (such as house numbers, display boards), thematic elements (such as bookshelves, chairs), and spatial elements (such as corridor spaces, hall spaces, house spaces, elevator spaces, etc.), further refining the description of indoor space and helping to improve positioning accuracy. Based on the passable elements (such as corridors, stairs, elevators, etc.) in the entity model, a network model is constructed, which clearly defines the passable paths and connectivity within the building. This modeling method helps to optimize the path planning algorithm, ensuring that the positioning results can be accurately mapped to the actual passable paths, reducing positioning errors caused by unclear paths. Based on the spatial elements (such as halls, corridors, rooms, etc.) in the entity model, a voxel model is constructed, which subdivides these spatial elements into small voxel blocks, each voxel block having unique spatial characteristics and location identification. This refined description significantly improves the accuracy of indoor position matching and effectively reduces the offset error in traditional matching methods. The voxel model serves as an auxiliary matching tool for positioning results, providing more accurate and reliable indoor navigation services for users by accurately matching the obtained location information with the finely divided voxel blocks. By integrating the advantages of entity models, network models, and voxel models, a hybrid map model is constructed. This fusion not only improves the richness and accuracy of the map model, but also enhances the system's adaptability to complex indoor environments. After obtaining the original coordinate data output by the positioning system as the set of matching points, these positioning points are mapped onto the hybrid map model through a map matching algorithm. Since the hybrid map model integrates multi-level data information, the map matching process can more accurately reflect the user's actual location, thereby significantly improving the accuracy of indoor positioning.

[0091] In some embodiments, the current entity model representing indoor building elements is mainly divided into two categories: geometric boundary model and building entity model. Geometric boundary model is constructed by capturing the boundaries of indoor space, using volume shapes to represent indoor space units. While building entity model is usually generated based on two-dimensional floor plans, or constructed through three-dimensional modeling software. Compared with geometric boundary model, building entity model can more accurately present the indoor building structure and provide users with a more realistic visual experience. Therefore, the present application chooses to use building entity model. The construction of the entity model based on the building information model data in S110 includes:

[0092] S1101, import building geometric information and semantic information based on building information model data, and output original modeling data.

[0093] First, the geometric information and semantic information of the building are imported from the BIM model. The geometric information includes the three-dimensional shape, size, spatial layout of the building, etc., while the semantic information covers the descriptive information such as room name, functional area, equipment and facilities. These information together constitute the original modeling data.

[0094] By importing BIM data, the accuracy and completeness of the entity model in geometric shape and semantic description are ensured, providing reliable data support for subsequent modeling.

[0095] S1102, pre-process the original modeling data to obtain intermediate model data.

[0096] The imported original modeling data is pre-processed, including denoising, alignment, simplification and other operations. Denoising aims to eliminate noise and redundant information in the data, alignment operation ensures the spatial consistency between different data sources, and simplification operation reduces data complexity and improves processing efficiency. The pre-processing process effectively improves the data quality, reduces the computational burden in the subsequent modeling process, and is conducive to the subsequent generation of high-quality entity model.

[0097] S1103, according to the demand of indoor location service, extract basic elements and identification elements from the intermediate model data to obtain the basic model.

[0098] According to the demand of indoor location service, extract basic elements (such as walls, columns, doors, windows, etc.) and identification elements (such as doorplates, display boards, display screens, etc.) from the pre-processed intermediate model data. These elements are the basic units of the entity model, and through fine modeling, the basic model is formed. By extracting key elements for modeling, the entity model can accurately reflect the actual structure and characteristics of the building, meeting the needs of indoor location service for space description.

[0099] Table 1 shows the classification of various elements. The basic elements form the core part of the building map, including walls, columns, doors, windows, floors, stairs, etc. The texture information of these elements is relatively simple, and they are converted from two-dimensional data to three-dimensional models to generate. The identification elements have unique features, which can be used as the basis for locating the features of the building, such as house numbers, display boards, display screens, floor signs, etc. The texture information of these elements needs to be matched with their features, and needs to be combined with the pictures and geometric information of the components to build. The thematic elements are not essential parts of the building map, and their addition can be determined according to specific needs, such as bookshelves and seats in a library. The texture information of these elements can be selected according to actual needs, and also needs to be combined with the pictures and geometric information of the components to build. The spatial elements refer to specific spatial areas surrounded by physical components such as walls and columns in the basic elements. In indoor environments, they include corridor spaces, hall spaces, house spaces, elevator spaces, etc. These spaces are not isolated, but are closely connected with the overall structure of the building, and are indispensable parts of the building interior. They together constitute the spatial layout and functional division of the building interior. Unlike basic elements and identification elements, spatial elements do not emphasize specific texture information. It focuses more on the geometric properties of space, such as range, shape and connectivity, and expresses the existence and form of space in a voxelized manner. For example, a room space can be defined by its boundary coordinates in three-dimensional space. Spatial elements can be regarded as voxelized target objects, i.e. they are discretized into a series of small three-dimensional units (voxels) for processing and analysis in the computer. This voxelized expression helps to more accurately describe the geometric characteristics and spatial relationships of space, such as calculating the size and volume of space, analyzing the connectivity between spaces, etc.

[0100] Table 1

[0101]

[0102] The core target and application direction of the entity model of the present application is mainly designed for the specific needs of the indoor location service field. Therefore, in the construction process of the entity model, the data content covered and integrated mainly focuses on two core categories: one is the basic elements, which constitute the basic framework and data support of the entity model, including walls, columns, doors, windows, floors, etc., which have low texture complexity (simple geometry), and are constructed by using the three-dimensional modeling method of two-dimensional data, mainly used for building structure expression. The second is the identification elements, which are mainly used to realize the accurate identification and positioning of indoor location, such as door plates, display boards, display screens, etc., which have high texture complexity (picture matching is required), and are constructed by using the method of combining geometry + picture, applied to AR enhancement, positioning identification and other scenes. The third is the space elements, such as corridor space, hall space, house space, elevator space, etc. This kind of element has no texture, which refers to the space surrounded by the entity components, that is, the voxelized target object, mainly applied to various applications of voxel expression space, such as positioning. As for other related thematic elements, considering the diversity and complexity of actual application scenarios, the present application does not include them all at once in the initial stage, but leaves a flexible expansion interface, so that subsequent targeted addition and optimization can be carried out according to the actual use and specific needs.

[0103] The construction of space elements is mainly based on the geometric information of entity components. By analyzing and processing the geometric model of basic elements (walls, columns, etc.), the boundary and range of the space are determined. For example, using the position and thickness information of the wall, different room spaces and corridor spaces can be divided. In this process, it is not necessary to combine picture information to construct like identification elements, mainly relying on the calculation and reasoning of geometric data.

[0104] In the indoor positioning system, space elements can be used as a reference framework for positioning. For example, by matching the position information of the positioning target (such as personnel or equipment) with the voxel model of the space element, the specific space area where the target is located can be more accurately determined, such as whether the personnel is in a certain room or on a corridor. Space elements also help to analyze the spatial relationship between different positions, such as distance, connectivity, etc., for path planning, navigation. For example, when planning a path from a hall to a certain room, the connectivity between space elements needs to be considered to avoid obstacles such as walls and choose the optimal passage space. Combined with other element information, space elements can also be used to identify the function of the space. For example, according to the size, shape of the space and the relationship with the surrounding identification elements (such as floor signs, room door plates), it can be judged that a space is a conference room, an office or a storage room, etc., thereby providing more rich semantic information for indoor location services.

[0105] S1104, semantic information is entered into the base model to obtain an entity model; the semantic information at least includes a room name, a functional area, a device facility, and multimedia data.

[0106] On the basis of the base model, detailed semantic information is entered, including a room name, a functional area, a device facility, and multimedia data (such as pictures, audio, video, etc.). These information enriches the connotation of the entity model, making it not only have geometric shapes, but also have rich semantic descriptions. The entry of semantic information makes the entity model more intelligent and humanized, and can provide more comprehensive and accurate indoor location services for users. For example, users can quickly locate the target position by querying the room name or functional area, and at the same time obtain relevant multimedia information to assist decision-making.

[0107] The entity model constructed based on BIM data in the application reaches a high precision level in geometric shape and semantic description, and provides a reliable data basis for indoor location services. Through preprocessing and element extraction, the data volume and computational burden are effectively reduced, and the modeling efficiency and processing speed are improved. The integration of rich semantic information and multimedia data makes the entity model more intelligent and humanized, and can meet the diversified needs of users. The entity model adopts modular design, which is convenient for extension and modification according to actual needs, to adapt to the needs of indoor location services in different scenarios.

[0108] In some embodiments, in constructing a building hybrid map model for indoor positioning and navigation, the construction of the network model is based on the passable elements (such as corridors, floor entrances and exits, and stairs) in the entity model. The above S120 constructs a network model based on the passable elements in the entity model, including:

[0109] S1201, extracting the center points of each floor corridor and connecting to generate a single-layer skeleton network.

[0110] The center points of each floor corridor are extracted from the entity model, and these center points are used as the basic nodes for path planning. Then, these center points are connected to form a single-layer skeleton network, which represents the main passable path in the floor. The single-layer skeleton network provides a basic framework for subsequent path planning and navigation, ensuring the connectivity and accessibility of the path.

[0111] S1202, based on the single-layer skeleton network, sequentially connecting each target node and connecting the target node and the corridor center point to form a single-layer road network containing horizontal path topology; the target node is a key functional node that needs to be connected in the network model, including space connection points such as doors, stair entrances and exits, and elevator halls, and user accessible target points such as room entrances, service facilities, and navigation signs;

[0112] On the basis of the single-layer skeleton network, the target nodes (such as space connection points such as doors, stair entrances, elevator halls, and user accessible target points such as room entrances, service facilities, and navigation signs) and the corridor center points are connected to form a single-layer road network containing a horizontal path topology. These target nodes are key positions that users need to reach or pay attention to in an indoor environment. The single-layer road network not only contains the main traffic paths within the floor, but also clearly indicates the positions of the key target nodes, providing detailed path information for indoor navigation.

[0113] S1203, integrating the vertical road network by using the single-layer road network to obtain a network topology integrating horizontal paths and vertical paths.

[0114] The single-layer road networks of each floor are vertically integrated to form a network topology containing horizontal paths and vertical paths. The integration of the vertical road network at least includes connecting the communication nodes representing the entrances of each floor to ensure the connectivity between floors. The integrated network topology can comprehensively reflect the traffic structure inside the building and support path planning and navigation across floors.

[0115] Figure 2 A network model abstract expression of a single corridor space is shown, which accurately depicts the traffic structure and key positioning elements of the corridor through nodes and connection relationships. The model includes corridor center nodes (distributed equidistantly along the corridor center line, used to construct a horizontal path skeleton), target nodes (such as stair entrances, elevator halls, and room entrances, marking space communication positions and user accessible targets), semantic information anchors (facilities such as fire hydrants and cameras are abstracted as nodes, and multiple information such as text, pictures, voice, and video are associated to enhance navigation interactivity), and map positioning anchors (pedestrian anchor nodes are set along the corridor and stairs according to preset pedestrian movement characteristic parameters, used to improve map matching accuracy and support cross-floor navigation). Each node forms a complete network through horizontal path topology (corridor center nodes are connected in turn) and vertical path integration (stair entrances are connected with floor communication nodes), improving path planning accuracy, enriching semantic information presentation, enhancing positioning matching robustness, and supporting seamless navigation across floors. For example, users can quickly query room positions and facility information through the model, or achieve efficient navigation between floors, fully demonstrating the practicability and reliability of the model in complex indoor environments.

[0116] As shown in (a) of Figure 2 , the single-layer road network includes door nodes, pedestrian anchor points, window nodes, room nodes, component nodes, communication nodes, main road lines, elevator nodes, virtual edges, and branch road lines. Figure 2 b and c of (a) of Figure 2 are top views of (b) and (c), respectively. Refer to Figure 2(b) of the figure shows that for the elevator, the elevator virtual nodes and the elevator door nodes are connected together. Figure 2 (c) of the figure shows that for the stairs, the stair platform nodes, the stair steps and the stair door nodes are connected together in the vertical direction.

[0117] S1204, implanting semantic information anchors and map positioning anchors in the network topology to obtain a network model.

[0118] To enhance the practicability and application range of the model, the present application introduces feature nodes: semantic information anchors MLA(C) and map positioning anchors MLA(S) on the basis of the existing BIMPN network model shown in (a) of the figure. Figure 3 The b, c and d parts of (a) are respectively the plan views of (b), (c) and (d) of the figure. The semantic information anchors are implanted by abstracting indoor facilities (doors, door numbers, fire hydrants, fire alarms, safety exit signs, wireless local area networks (WLAN), cameras, elevators, display boards, lighting lamps, electrical boxes, etc.) into facility nodes with specific semantic information, as shown in (c) of the figure. The main role of these nodes is to assist the application of indoor instantiation augmented reality (AR) technology and provide necessary language information support. In order to meet the needs of AR technology in content enhancement, AR enhancement data will adopt diversified storage forms. Multiple semantic information is configured for each facility node to obtain semantic information anchors, which are added to the network topology; the multiple semantic information at least includes text, pictures, voice and video data. The implantation of the semantic information anchors enriches the information content of the network model, so that users can obtain more information about the facilities during navigation and improve the practicability and interactivity of navigation. Figure 3 Figure 3 The map positioning anchors are walking anchor nodes, which are implanted by the following way: Figure 3 The preset walking movement characteristic parameters are obtained; based on the network topology, the preset walking movement characteristic parameters are selected as the interval standard of the walking anchor nodes in the horizontal direction; and in the vertical direction, the walking anchor nodes are arranged on each stair step for the stairs, and the preset walking movement characteristic parameters are also selected as the interval standard of the walking anchor nodes for the stair platforms. The above-mentioned preset walking movement characteristic parameters refer to the average stride of the user, which can be set according to actual needs, such as 0.5 m. The implantation of the map positioning anchors improves the accuracy of map matching, especially in the actual movement of the pedestrians, which can more accurately reflect the position change of the pedestrians and improve the accuracy and practicability of navigation.

[0119]

[0120] The preset walking movement characteristic parameters are obtained; based on the network topology, the preset walking movement characteristic parameters are selected as the interval standard of the walking anchor nodes in the horizontal direction; and in the vertical direction, the walking anchor nodes are arranged on each stair step for the stairs, and the preset walking movement characteristic parameters are also selected as the interval standard of the walking anchor nodes for the stair platforms. The above-mentioned preset walking movement characteristic parameters refer to the average stride of the user, which can be set according to actual needs, such as 0.5 m. The implantation of the map positioning anchors improves the accuracy of map matching, especially in the actual movement of the pedestrians, which can more accurately reflect the position change of the pedestrians and improve the accuracy and practicability of navigation. ​​

[0121] The map positioning anchor point MLA(S) is a node added to the edge of the map based on the existing end node, with the user's step as the basic unit. These nodes do not contain any semantic information, and their main function is to improve the accuracy of map matching and serve as an edge weight reference value during auxiliary navigation. When determining the interval unit of the walking anchor node, the invention has made careful considerations. Given that walking is the main way of changing the position of the human body in an indoor environment, after comprehensive analysis and experimental verification, the average user step of 0.5 meters is selected as the interval standard of the walking anchor node in the horizontal direction, as shown in (b) of FIG. 1. In the vertical direction, for stairs, the walking anchor node is set on each stair step, and for stair platforms, the average user step of 0.5 meters is also selected as the interval standard of the walking anchor node, as shown in (d) of FIG. 1. The setting of this standard aims to ensure that the walking anchor node can effectively reflect the actual movement of the user, thereby improving the accuracy and practicality of the navigation system. Figure 3 Figure 3 The network model of the invention integrates horizontal paths and vertical paths, comprehensively reflects the traffic structure inside buildings, and supports path planning and navigation across floors. By implanting semantic information anchor points and map positioning anchor points, the network model improves the accuracy of map matching, especially during the actual movement of pedestrians, it can more accurately reflect the change of the position of the pedestrian. The network model contains rich semantic information and target node information, providing detailed navigation guidance and facility information for users, and improving the practicality and interactivity of navigation. The network model adopts modular design, which is convenient for expansion and modification according to actual needs, to adapt to indoor positioning and navigation requirements in different scenarios.

[0122] In some embodiments, when the BIMPN model is applied to an indoor positioning and navigation system, in order to accurately obtain position information, it is usually necessary to forcibly match the indoor coordinates with the pre-set road network nodes. However, this traditional matching method often has many shortcomings in actual application, especially in the horizontal and vertical directions, which is prone to significant offset errors, which not only reduces the accuracy of positioning, but also negatively affects the user's navigation experience. In order to effectively solve this problem, the invention innovatively proposes the concept of voxel model.

[0123]

[0124] ​​The core of the method is to carefully divide and describe the indoor space where the pedestrian is located. Specifically, various spaces in the indoor environment, such as halls, corridors, rooms, staircases, and elevators, are voxelized. On this basis, these space elements are further divided into a plurality of small voxel blocks, each of which has unique spatial characteristics and location identification. By accurately matching the obtained location information with these finely divided voxel blocks, the accuracy of indoor location matching can be significantly improved. This method not only effectively reduces the offset error in traditional matching methods, but also provides more accurate and reliable indoor navigation services for users, greatly improving the user's indoor positioning experience. In the construction of a building hybrid map model for indoor positioning and navigation, the voxel model is constructed based on the space elements in the entity model. The voxel model constructed based on the space elements in the entity model described in S130 includes:

[0125] S1301, performing spatial classification processing on the space elements of the entity model according to the space type and navigability, and outputting the classified model.

[0126] First, the space elements in the entity model are classified according to the space type (such as room, corridor, staircase, elevator, and other spaces) and navigability (navigable, semi-navigable, and non-navigable). Among them, other spaces refer to the spaces inside building components (such as bookcases, wardrobes, walls, etc.). These spaces are usually not suitable for indoor positioning services, so the invention ignores this part of the space when extracting the space. For example, as shown in Table 2, the navigable spaces (such as corridors, staircases) and semi-navigable spaces (such as elevators that can be navigated at certain times) are distinguished, and non-navigable spaces (such as the interior of building components) are ignored. Semi-navigable spaces refer to spaces that cannot be navigated after a certain time, such as elevators that stop running at certain times, or spaces that are occupied by objects, which can be moved in necessary circumstances (such as indoor firefighting). Through spatial classification processing, the spatial structure and navigation characteristics of the building interior can be more clearly understood, providing a basis for subsequent voxelization processing.

[0127] Table 2

[0128]

[0129] S1302, voxelizing the classified model to obtain a voxelized preprocessed model.

[0130] The voxelization process involves converting a large number of mature face-based graphics three-dimensional models in a computer into a discrete voxel set representation. The voxelization preprocessing of the classified model includes selecting an appropriate voxel size (such as 0.5x0.5x0.5m 3) and setting coordinate system. The purpose of this step is to discretize the continuous three-dimensional space into a regular voxel set, which is convenient for subsequent processing. By establishing a unified coordinate system, the consistency of the voxelization process can be ensured. At the same time, the explicit coordinate space has a restraining effect, which can minimize the amount of calculation. The voxelization preprocessing simplifies the complexity of the three-dimensional model, making the subsequent voxelization processing more efficient and feasible.

[0131] S1303, using a scan filling method to perform scan filling processing on each spatial surface of the voxelization preprocessed model to obtain a surface voxel marking result, and performing internal filling on the voxelization preprocessed model according to the surface voxel marking result, and outputting a voxel model.

[0132] For voxelization of the model surface, the present application selects a scan filling method (Scanline Voxelization). The principle of this method is to scan along a certain axis for each triangular patch of the model surface, calculate the projection of the patch in the voxel space, and fill the voxels in the projection area, as shown in Figure 4 The scan filling method is used to scan each spatial surface of the voxelization preprocessed model. Specifically, scanning is performed along a certain axis (such as the y-axis), the projection of the patch in the voxel space is calculated, and the voxels in the projection area are filled. By assigning a value of 1 to the voxel blocks in the plane and a value of 0 to the external voxel blocks, a surface voxel marking result is obtained. According to the surface voxel marking result, internal filling is performed on the voxelization preprocessed model. Specifically, scanning is performed in 6 directions (up, down, left, right, front, and back) for the surface voxels, the scanning results are assigned values, the voxel values inside the matrix are set to 2, and the voxel values outside the matrix are set to 3, thereby realizing the filling of the model interior. Figure 4 H in the formula is the edge length of the voxel, which is 0.5m.

[0133] The surface of the solid model is composed of numerous irregular triangular meshes, and there are shared edges and shared vertices between these meshes, which leads to frequent occurrence of repeated voxelization phenomenon, thereby affecting the accuracy and efficiency of the solid model in the subsequent processing process. In view of the above problems, the present application adopts the following solutions.

[0134] The voxelization problem caused by the common edge can be properly solved by a specific processing method. In order to prevent repetition in the voxelization process, the integer coordinate points sharing the same boundary must be assigned to one of the triangular shapes. Therefore, when filling the pixels of the triangular boundary, the present application consistently follows the rule of "filling the left side but not the right side, and filling the lower side but not the upper side", which is described in detail as follows:

[0135] Each scan line intersects with a triangle at two points, and if the intersecting pixel point is on the left side, as shown in Figure 5As shown in (a), fill the gap if the intersecting feature points are located on the right, such as... Figure 5 As shown in (b), no filling is required. If one side of the triangle is parallel to the x-axis, and the intersecting pixels are located at the top, Figure 5 As shown in (c), fill is applied if the intersecting feature points are located at the bottom, such as... Figure 5 As shown in (d), no filling is required.

[0136] However, the situation of shared vertices is particularly complex because these vertices are shared by a variable number of triangles and lack a clear pattern. To address this, this invention introduces a priority determination mechanism during the voxelization process. This mechanism determines the priority of voxel assignment based on the size of the triangle's area; triangles with larger areas have higher priority, and their shared vertices are assigned priority during voxelization. According to this method, each shared vertex is assigned a unique voxel assignment.

[0137] After scanning and voxelizing each surface sequentially, the entire model is seamless and continuous. Since the voxel mapping space information is stored in a three-dimensional matrix V, after surface voxelization, the matrix space contains only voxel blocks with a value of 0 (no surface mapping) and voxel blocks with a value of 1 (complete surface mapping). Then, the surface voxels are scanned along six directions, and the scan results are assigned values: voxels inside the matrix are set to 2, and voxels outside the matrix are set to 3. This captures the internal portion of the surface voxels, thus filling the interior.

[0138] This invention employs a scan-fill method to ensure the accuracy of voxelization, precisely converting 3D models into discrete sets of voxels. This scan-fill method enables efficient processing of large-scale 3D model data, improving the efficiency of voxelization. Voxel models are easy to store and transmit, facilitating visualization and analysis on different platforms and devices. Furthermore, voxel models support various spatial analysis and processing algorithms, providing greater possibilities for indoor positioning and navigation.

[0139] This invention uses voxel models to subdivide indoor spaces into regular cubic units, enabling a more refined description of spatial and building features and improving the accuracy of indoor location matching. Compared to complex 3D models, voxel models simplify data processing, improve computational efficiency, and make real-time positioning and navigation possible. Voxel models are easy to visualize and analyze on different platforms and devices, exhibiting good multi-platform compatibility and facilitating user operation in various environments. Voxel models support various spatial analysis and processing algorithms, such as path planning and space occupancy analysis, providing greater functional support for indoor positioning and navigation.

[0140] In some embodiments, the hybrid map model integrates the entity model, the network model and the voxel model, and constitutes a comprehensive model. The construction of the model relies on the spatial position and semantic information between different element models, thereby establishing the interactive connection mechanism between the models. The hybrid map model is constructed based on the entity model, the network model and the voxel model in S140, including:

[0141] The spatial connection relationship between the entity model, the network model and the voxel model, and the mapping relationship between the network model node and the voxel model unit are established, and an integrated hybrid map model is output; the spatial connection relationship includes direct connection and indirect connection.

[0142] The direct connection refers to the existence of a corresponding relationship between two models, without the need for other models as an intermediate bridge. Figure 6 The main application scenario of the direct connection is represented by the node (i.e., the component node) in the network model representing a component, and the building component corresponding to the node in the entity model. It is represented by the voxel model expressing an indoor space element, and all building components constituting the space. It is represented by the voxel model expressing an indoor space element, and the network node (such as a door node, a component node, and a positioning anchor point) in the network model representing the room. The connection mode is that the relationship between the node element expressing a component in the network model and the building component in the entity model is one-to-one. The relationship between the voxel model and the building entity, the component node and the room node expressing the space is not one-to-one. The voxel is determined by multiple components, the room node expressing the space is composed of multiple voxel blocks, and the space expressed by the voxel may also contain multiple components and nodes.

[0143] Establishing the direct connection includes establishing a one-to-one correspondence between the component node in the network model and the corresponding building component in the entity model, and establishing an association between the space set in the voxel model and all building components constituting the space in the entity model.

[0144] The process of establishing the correspondence between the network model component node and the entity model building component is that the component node (such as a wall, a door, a staircase, etc.) in the network model is in one-to-one correspondence with the corresponding building component in the entity model. For example, the "door node" in the network model is directly associated with the geometric and semantic information representing the door in the entity model. This ensures that the path planning in the network model is consistent with the actual building structure in the entity model, and improves the accuracy of navigation.

[0145] The process of establishing the association between the voxel model space set and the entity model building component is that the space set in the voxel model (such as a room composed of multiple voxel blocks) is associated with all building components in the entity model that constitute the space. For example, the voxel set of a room is associated with all the components such as walls, floors, and ceilings in the entity model that represent the room. Through the space division of the voxel model, a fine-grained description of the entity model space is achieved, supporting more accurate spatial analysis and positioning.

[0146] Indirect connection refers to the absence of a corresponding relationship between two models, which requires other models as intermediaries, Figure 7 Indirect connection is represented in the room nodes in the network model and all elements in the entity model that constitute the room, as well as the voxel blocks and the entity model that constitutes them. The connection method is as follows: in the network model, the room is abstracted as a target node, while in the entity model, the room is composed of multiple building entities with boundary relationships, and the two are connected through the voxel model that represents the room.

[0147] Establishing indirect connection includes: the room node in the network model is associated with multiple components in the entity model through the voxel model, and the voxel model unit is associated with the network model node through the entity model component.

[0148] The process of associating the network model room node with the entity model component through the voxel model: the room node in the network model is not directly associated with a single component in the entity model, but is associated through the voxel model. For example, the room node is associated with the voxel set representing the room, and these voxel sets are associated with all components in the entity model that constitute the room. Simplify the structure of the network model, and at the same time realize indirect association with the entity model through the voxel model, supporting complex spatial queries and analysis.

[0149] The process of associating the voxel model unit with the network model node through the entity model component is that the unit in the voxel model (such as a single voxel block) is associated with the network model node through the component in the entity model. For example, a voxel block is associated with a wall in the entity model, and the wall is associated with a wall node in the network model. Realize the bidirectional association between the voxel model and the network model, support spatial analysis and navigation from the voxel level to the network level.

[0150] The process of establishing the mapping relationship between the network model nodes and the voxel model units is: for each node in the network model, a mapping relationship with the related units in the voxel model is established. For example, the room entrance node in the network model is mapped to the voxel unit representing the entrance in the voxel model, and the surrounding related voxel units. Through the mapping relationship, the positioning system can accurately map the original coordinate data to the voxel model units, and then determine the user's position in the network model through the association of the voxel model and the network model. When planning a path, the spatial information of the voxel model can be used to generate a path that is more consistent with the actual spatial structure.

[0151] The above spatial connection relationship and mapping relationship are integrated to output an integrated hybrid map model. The model includes the geometric and semantic information of the entity model, the topological structure of the network model, and the spatial division of the voxel model. The hybrid map model of the present application integrates multi-level information and supports more comprehensive indoor positioning and navigation services. It can be used for path planning, spatial query, facility positioning, emergency evacuation and other application scenarios.

[0152] The hybrid map model can more accurately describe the indoor space and reduce positioning errors through the spatial division of the voxel model and the topological structure of the network model. The hybrid map model integrates the geometric information of the entity model and the path information of the network model, supporting more reliable path planning and navigation. The fine-grained spatial description of the voxel model and the topological relationship of the network model enable the hybrid map model to support complex spatial analysis, such as spatial occupancy analysis and people flow simulation. The hybrid map model provides rich semantic information (such as room names, facility locations), enhances the interactivity and practicality of navigation, and improves the user experience. The hybrid map model can adapt to large and complex indoor environments such as shopping malls, hospitals, airports, etc., and meet the positioning and navigation needs in different scenarios.

[0153] Taking a shopping mall as an example, the application of the hybrid map model is described: the entity model includes the geometric and semantic information of the building structure, shop location, facility distribution, etc. of the shopping mall. The network model constructs the passable path network of the shopping mall and marks key nodes (such as elevators, stairs, restrooms, main shop entrances). The voxel model subdivides the shopping mall space into voxel blocks, and each voxel block is associated with a building component in the entity model. When a user queries the location of a certain brand shop, the hybrid map model can quickly locate the voxel unit of the shop and plan the optimal path. During navigation, the hybrid map model can be used to correct the positioning in real time to ensure the accuracy of the path. In an emergency, the hybrid map model can be used to simulate people flow evacuation and plan a safe evacuation path.

[0154] In the mixed map data storage system, due to the diversity and complexity of data types, it is difficult to meet the needs of efficient management and rapid retrieval by using a single storage method. Therefore, the present application proposes a hybrid storage scheme combining file storage and database storage to cope with the storage and management challenges of different types of data.

[0155] In terms of database storage, the present application uses a relational database management system to store structured data. Table 3 summarizes the table structure in the database, where each table corresponds to a feature category in the map. In order to facilitate management and query, these features are abstracted as objects and organized according to their scale and hierarchical relationship. Features are stored in a hierarchical manner according to the "scene-semantic-instance-connection" method, and users can easily query, update and manage data in the database through SQL statements.

[0156] Table 3

[0157]

[0158] The design of the file storage structure first needs to divide the folder hierarchy, as shown in the specific architecture Figure 8 , in order to facilitate the establishment of associated indexes with the data stored in the database. First, according to the scale of the large scene, divide it into a one-level directory of a certain building floor, then according to the hierarchical division method in the scene scale, establish a two-level file directory structure of the building folder nested elevator folder, staircase folder and each floor folder. Finally, according to the classification standard of entity models, establish a three-level directory of basic elements, identification elements, thematic elements, and spatial elements, and store the corresponding files in the directory into the file.

[0159] In some embodiments, map matching is a process of mapping the positioning result to the mixed map based on the initial positioning result using a matching algorithm. In indoor positioning scenarios, the positioning system usually returns a series of coordinate data, however, there is no special feature in the mixed map to represent the coordinates. Therefore, it is necessary to establish a connection between the coordinate information and the mixed map model through map matching, so as to realize the accurate alignment of the positioning result and the map. Map matching of the mixed map mainly includes two types: network node matching and voxel node matching. These two matching methods are suitable for different application scenarios and can meet the needs of path planning and location awareness. The above S140 obtains the original coordinate data output by the positioning system as a set of matching positioning points, and maps the set of matching positioning points to the mixed map model through map matching, including:

[0160] S210, obtaining the original coordinate data output by the positioning system as a set of matching positioning points.

[0161] The positioning system (such as Wi-Fi positioning, Bluetooth positioning, UWB positioning, etc.) outputs the original coordinate data of the user in real time, which constitutes the set of matching positioning points. Each positioning point contains coordinate information (such as x, y, z) and a timestamp. The original coordinate data provides the basis for subsequent map matching, but there are usually errors that need to be corrected through map matching.

[0162] S220, based on the coordinate system conversion parameters of the hybrid map model, converting the set of matching positioning points to the model coordinate system, and outputting the standardized positioning data.

[0163] The original coordinate data output by the positioning system is usually based on its own coordinate system, while the hybrid map model uses a unified model coordinate system. Through coordinate system conversion parameters (such as translation vectors, rotation matrices, scaling factors), the set of matching positioning points is converted from the positioning system coordinate system to the model coordinate system. The converted coordinate data may still have noise or outliers, which need to be standardized (such as filtering, denoising) to improve data quality. By using a unified coordinate reference, all positioning data is processed in the same coordinate system, avoiding errors caused by inconsistent coordinate systems. Standardization reduces the impact of noise and outliers, providing more reliable data for subsequent map matching.

[0164] S230, using the buffer matching method to process the standardized positioning data, and outputting the set of map matching result data; the buffer matching method includes: establishing a buffer zone centered on the matching positioning point, and extracting all network nodes and voxel nodes within the buffer zone; traversing all network nodes and voxel nodes within the buffer zone, and determining the network node and voxel node closest to the matching positioning point as the corresponding map matching result.

[0165] The matching technology adopted by the present application is based on the buffer matching method, and the principle is as shown in Figure 9 This method is based on spatial distance matching, and the purpose is to accurately correspond the positioning result to the network node or voxel center node of the hybrid map. The core idea is to calculate the spatial distance between the positioning point and each node in the hybrid map model to determine the closest node as the matching result. As shown in Figure 9 The buffer matching method process is: taking each standardized positioning point (i.e. the target node in Figure 9 ) as the center, establishing a buffer zone with a radius of R. The radius R of the buffer zone is determined according to the accuracy of the positioning system and the resolution of the hybrid map model. From the hybrid map model, all network nodes (i.e. Figure 9network nodes (such as corridor center nodes, room entrance nodes) and voxel nodes (such as voxel center points) in the buffer zone. The Euclidean distance between each node and the positioning point to be matched is calculated. The network node and voxel node closest to the positioning point to be matched are determined as the corresponding map matching result. The buffer zone matching method selects the closest node as the matching result by considering multiple candidate nodes around the positioning point, effectively reducing the positioning error. In a complex indoor environment (such as signal shielding, multipath effect), the buffer zone matching method can better handle the uncertainty of positioning data and improve the robustness of matching. By considering the nodes of the network model and the voxel model, the buffer zone matching method realizes the fusion of multiple models and fully utilizes the advantages of the hybrid map model.

[0166] The present application eliminates the coordinate system difference between the positioning system and the hybrid map model through coordinate system conversion and standardization processing, reducing the positioning error. Standardization processing and coordinate system conversion can be completed in the preprocessing stage, reducing the computational burden of real-time processing. The buffer zone matching method further improves the positioning accuracy by selecting the closest node as the matching result. The buffer zone matching method can handle the uncertainty of positioning data and adapt to complex indoor environments (such as signal shielding, multipath effect). Multi-model fusion (network model + voxel model) provides more comprehensive spatial information and enhances the robustness of matching. The buffer zone matching method has high computational efficiency and can meet the needs of real-time positioning and navigation. High-precision positioning and navigation results improve the user's navigation experience. The hybrid map model provides rich semantic information (such as room name, facility location), enhancing the interactivity and practicality of navigation.

[0167] Taking a shopping mall as an example, the application of the map matching process is explained: the positioning system outputs the original coordinate data of user A in the shopping mall as (x1, y1, z1). Convert (x1, y1, z1) to the coordinate system of the hybrid map model to obtain standardized positioning data (x1', y1', z1'). Establish a buffer zone with a radius of R centered on (x1', y1', z1'). Extract the network nodes (such as elevator entrance node E1, store entrance node S1) and voxel nodes (such as voxel center point V1) in the buffer zone. Calculate the distance between each node and (x1', y1', z1') to determine the closest node as the matching result. Assuming E1 is the closest node, the map matching result of user A is the elevator entrance. According to the map matching result, the optimal path from the current position to the elevator entrance is planned for user A.

[0168] Figure 10The application introduces a scheduling loading scheme of a building entity-network-voxel hybrid map model. The core principle of the application is "stratified classification and on-demand scheduling". In view of the complex structure of the building environment in the three-dimensional space, the indoor environment faced by the user at different height positions may be significantly different. Based on this actual condition, when formulating the scheduling and loading strategy, the structural characteristics of the building are fully considered, and the building is naturally divided into several different floors and vertical transportation modes connecting the floors. Specifically, first, in the vertical direction, the stratified scheduling loading mode is adopted with the floor as the basic unit, so as to ensure that the data of each floor can be loaded in an orderly and efficient manner. Then, when processing each specific floor, the classification method is further adopted in the horizontal direction, and the elements required in the floor are loaded in detail. Finally, in terms of on-demand scheduling, the dynamic loading strategy is designed, the priority and range of data loading are adjusted in real time according to the actual demand and interactive behavior of the user. The system dynamically evaluates the data demand in the current field of view, preferentially loads the floors and elements visible to the user, and delays the loading of the data that is invisible or temporarily unnecessary, so as to significantly reduce unnecessary data transmission and computing burden.

[0169] Referring to Figure 10 As shown in the figure, in terms of model construction, the building entity-network-voxel hybrid map model is generated from data sources such as geometric information (coordinates, features, etc.), basic attributes (name, function, etc.), and thematic attributes (audio, texture, etc.). The model divides the floors (1F, 2F, 3F,...) in the vertical direction, and in each floor, three kinds of sub-models are fused: the network model contains edges (central axis, horizontal connecting line, vertical connecting line, etc.), nodes (target node, connecting node, walking error node, etc.); the voxel model classifies rooms, corridors, stairs, elevators, etc. from the horizontal and vertical space angles; and the entity model covers basic elements (walls, elevators, doors, floors, etc.), thematic elements (office facilities, lighting facilities, fire facilities, etc.), and space elements (corridor space, hall space, house space, elevator space, etc.). In the data transmission layer, with the help of edge computing, 5G / 4G, WIFI and other networks, data transmission is realized through the cloud server, and data interaction with the application layer is realized by using function interfaces.

[0170] The application layer mainly implements multi-source sensor fusion positioning and indoor scene recognition positioning functions. The multi-source sensor fusion positioning determines the position by combining various sensor data, and the indoor scene recognition positioning judges the scene where the user is located. At the same time, by using visual technologies such as OpenGL ES, Cesium, Three.js, etc., the map model is displayed on terminal devices such as mobile phones, tablets, computers, etc., to facilitate users to view and use the building map information. The whole scheme provides a comprehensive and accurate solution for building indoor positioning and navigation and scene recognition through the fine construction of hybrid map models, efficient data transmission and diversified application functions.

[0171] The application also proposes a set of building hybrid map model visualization technology process system. The system aims to solve the high time consumption and high energy consumption problems of mass data processing in actual application services by constructing a building map data loading mechanism. Then, according to the needs of environmental development technology, the corresponding function interface is designed to realize the multi-end visualization of the building map. Through the development of a mobile terminal-oriented positioning and navigation system and a network terminal-oriented three-dimensional visualization and analysis system, the feasibility of the technology process is verified. The visualization scene constitutes the key of the whole architecture, and provides a container for carrying and integrating many elements in the three-dimensional world for three-dimensional applications. The main components of the visualization scene can be divided into:

[0172] (1) Model in the scene:

[0173] The model of the application mainly includes external model file formats such as GlTF and OBJ, and the frame adopts hierarchical classification loading technology to reduce the model loading power and time. At the same time, the mechanism of asynchronous loading and model data analysis is considered in the loading process, and the possible loading errors or exceptions are properly handled.

[0174] (2) Camera in the scene:

[0175] In the scene, the position and orientation of the camera directly determine the picture content and the breadth of the view angle displayed in the scene. Among them, the positioning of the camera determines the user's view angle, and the orientation of the camera determines the user's line of sight. By adjusting the positioning and orientation of the camera, the focus on different areas and objects in the scene can be achieved, thereby meeting the display of models in the scene.

[0176] (3) Light source in the scene:

[0177] When light is projected on the surface of an object, the object forms a light field by reflecting the light, and then the camera can capture the image of the object. In the framework of the application, to present the color of the object, in addition to the color attribute of the object itself, the assistance of light is also indispensable, otherwise the rendering result of the object will be black. The setting steps of the light source mainly include the design and implementation of:

[0178] Light source properties: A light source has multiple characteristics, including but not limited to color, position, and direction, which collectively affect the appearance of the light source in the scene.

[0179] Light source types: Different types of light sources can achieve distinctive effects, effectively meeting the specific needs of various application scenarios.

[0180] Multi-light support: In some complex scenes, a single light source cannot fully reveal the details of the model, so multiple light sources are needed to achieve a more realistic effect.

[0181] The present invention also completes the implementation of an interactive three-dimensional scene. WebGL, as a powerful web graphics rendering technology, provides rich and universal API interfaces that enable developers to efficiently convert complex three-dimensional model data into visible graphics elements in the browser. To further improve the overall development efficiency and user experience of the framework, the present invention related to WebGL mainly includes API interface encapsulation and shader.

[0182] API interface encapsulation: In the WebGL framework, numerous API interfaces are provided. The present invention will filter and encapsulate these interfaces based on functional requirements. The main types are as follows:

[0183] Operation data: Extract data from corresponding files and databases and transmit them.

[0184] Buffer: Establish an index range and bind the data within it.

[0185] Texture: Correctly use the mapping technology to give the model real texture information.

[0186] Management layer: Establish gradient layers in WebGL and associate each layer.

[0187] Shader: Create and compile shaders to create visual feedback in the scene.

[0188] About shader: Shaders mainly serve the purpose of graphics rendering, written in GLSL and run through GPU. The design of shaders in the present invention mainly includes:

[0189] Vertex shader, used to process vertex information of graphics such as position, normal, texture coordinates, etc., which determines the transformation and projection of graphics in three-dimensional space

[0190] Fragment shader, responsible for processing the color and texture of each pixel, generating the final image effect through calculation, such as lighting, shadow, reflection, etc.

[0191] In the process of building the entire framework, the renderer is located at the end of the flow, responsible for converting the complex information such as geometric data, material properties and lighting effects in the three-dimensional scene into two-dimensional images visible to users through a series of fine conversion and processing. The design and implementation of the renderer of the application mainly include:

[0192] (1) Initialization: the preparation work before rendering, which is the initialization creation of the project, including configuring the environment required by the project, installing the necessary dependent library, setting the basic structure and parameters of the project, etc.

[0193] (2) Scene configuration: the initial state of the environment is created, including the setting of the camera and the light source, the mapping of the three-dimensional coordinate system, the clipping and selection of the perspective window, etc.

[0194] (3) Coloring: the renderer works with the shader to pass the model data to the vertex shader and the fragment shader, process the complex logic such as lighting and texture mapping, and finally generate pixel color on the screen.

[0195] (4) Click interaction event: this part is often used as a direct display of system functions, by executing a click operation on the interface element, to trigger the related events of other interface elements.

[0196] (5) Compatibility and updateability: it should be ensured that it can work normally in various device and browser environments, and at the same time, the corresponding open interface should be designed to enhance the flexibility and updateability of the system.

[0197] The application selects a university surveying and mapping and urban spatial information college building F as the experimental area, which is a typical office scene with six floors, the corridor layout is relatively narrow, and the distribution of internal structural members presents a certain regularity. In terms of data source, the application selects CAD as the basic data, and then uses Revit software to perform three-dimensional modeling on the CAD drawing to obtain a BIM model, as shown in Figure 11 .

[0198] In terms of texture, the application collects the texture characteristics of the building in all aspects by fusing multi-source data collection technologies such as unmanned aerial vehicle aerial photography, smart phone photography and professional camera photography. Then, the collected photos are corrected to eliminate the image distortion caused by the variables such as shooting angle and light, to ensure the authenticity and practicality of the images.

[0199] Although the information in BIM is rich, the description of the building components is too fine, resulting in a large amount of redundant errors, so in order to reduce the unnecessary data amount in the model, the application simplifies and reconstructs the unnecessary lines and surfaces, redundant structural information, etc. Figure 12The lightweight processing result shown indicates that the simplification strategies (such as redundant structure elimination and line-surface optimization) adopted by the present application significantly reduce the data volume of the BIM model while retaining the geometric and semantic information of the key building components. This process reduces the complexity of subsequent calculations and improves the running efficiency of the ENV model. However, it should be noted that: (1) Precision trade-off: lightweight processing may result in the loss of some details, and the acceptable error range should be evaluated in combination with the application scenario. (2) Automation potential: machine learning algorithms can be introduced in the future to achieve adaptive lightweight processing and reduce manual intervention.

[0200] In the present application, the extraction of the entity model uses the Dynamo tool integrated by Revit software. Dynamo can be regarded as a graphical programming environment attached to Revit, which enables users to construct complex geometric shapes and logical processes through intuitive nodes and connection lines. Figure 13 The results of the extraction of the independent space of the surveying and mapping F building are obtained. The present application selects Python programming language to realize the voxelization of the spatial elements. The voxel block set after voxelization will be stored in the form of an entity model, and the program will calculate the center point coordinate information of each voxel block and export it as a text format for coordinate matching.

[0201] Through comparative analysis, it is confirmed that the spatial element description of the surveying and mapping F building is consistent with the actual situation. However, it is worth noting that the extraction result of the independent space voxel set is directly related to the quality of the model, so corresponding manual correction is needed for phenomena such as poor data source accuracy. Figure 14 The voxelization effect of part of the spatial elements of the experimental building is presented, and compared with the original elements constituting the space. By comparing the spatial elements before and after voxelization, it can be observed that the voxelization process not only retains the core features of the original spatial elements, but also simplifies the geometric complexity to a certain extent, thereby verifying the accuracy and effectiveness of the voxelization method.

[0202] Figure 15 The construction result of the building entity-network-voxel hybrid map model constructed by the present application for the surveying and mapping F building is shown. Figure 15 (a) in the present application presents the construction process and result of the network model in the hybrid map model. The process first involves the construction of a single-layer road network, followed by the construction of a vertical direction road network, and finally realizes the connection of the road networks of each layer. Figure 15 (b) in the present application discloses the construction method of the entity model in the hybrid map model. The method starts from three-dimensional modeling based on two-dimensional drawings, then constructs other related elements, and finally completes texture rendering. As for Figure 15(c) in the description of the construction method of the voxel model in the hybrid map model, the method is first divided into a pedestrian-oriented independent space, extracts spatial elements, and finally performs voxelization processing on these elements.

[0203] Figure 15 (d) in the description of the construction of the hybrid map model. According to the functional difference, the model can be divided into three categories: one is the road network expression for path planning, the second is the center point set expression for indoor positioning, and the third is the voxel expression for indoor spatial element expression.

[0204] Table 4 respectively counts the main elements of the hybrid map model of the Surveying and Mapping College F building. After comparing with the real building, the results show that the entity element part of the hybrid model shows good effect, can accurately restore the indoor environment of the building, and the node part has no obvious loss, which can fully meet the demand of indoor location service.

[0205] Table 4

[0206]

[0207] The application also tests and analyzes the building entity-network-voxel (ENV) hybrid map model, which includes: the local loading mechanism of the mobile positioning and navigation system, the loading efficiency test of the mobile positioning and navigation system, the design and function introduction of the mobile positioning and navigation system.

[0208] Regarding the local loading mechanism of the mobile positioning and navigation system: when performing three-dimensional model visualization on the mobile terminal, two major challenges are often faced: one is that network transmission is unstable or bandwidth is limited, resulting in model loading delay or even failure; the second is that the computing power of mobile devices is limited, which is difficult to smoothly render high-precision three-dimensional models, affecting user experience. In view of these problems, the application proposes a local storage + offline optimization loading solution. First, the original three-dimensional model that has been processed is output, and the model is converted to a lightweight GLTF compression format. Then, the optimized model resources (including geometric data, materials and text, etc.) are packaged and embedded in the application installation package, or stored in the local file system through silent download during the first start, ensuring offline availability. At runtime, the model is directly loaded from the local by using the mobile terminal graphics API, avoiding network requests. In addition, a progressive loading strategy is adopted, and the key parts within the visible range are preferentially loaded, further improving the smoothness of interaction. This scheme not only solves the problems of network dependence and performance bottleneck, but also is suitable for AR applications, offline BIM viewing and other scenarios, which optimizes the energy efficiency ratio while ensuring the visual effect.

[0209] Loading efficiency test of mobile positioning navigation system: To verify the optimization effect of local loading on mobile phones, the following comparative experiments are designed for mobile positioning navigation system.

[0210] Experimental object: F building of Surveying and Mapping Department mixed map model, initial data file is 100MB FBX format with texture image, after optimization processing, reduced to 56MB GLTF plus OBJ format with texture image.

[0211] Comparison group: scheme A, traditional network loading (original file, unoptimized model). Scheme B, local storage + progressive loading (optimized model embedded in installation package).

[0212] Test equipment: terminal Android system device, HAWEI Mate 40Pro (Mali-G78 MP24), VIVO X21 (Qualcomm Snapdragon 660AIE).

[0213] Experimental method: After clearing the cache of the mobile phone, 5G network, 4G network and Wi-Fi environment are used in turn, and scheme A and scheme B are implemented respectively. Each combination of network and scheme needs to be executed five times to record the loading time, frame rate performance and first rendering completion time of the model and other key parameters. Finally, the measurement results of each parameter are averaged for analysis. The test results are shown in Table 5.

[0214] Table 5

[0215]

[0216] Through careful analysis of the foregoing data, the local loading mechanism has shown its significant advantages in the field of three-dimensional applications on mobile platforms. First of all, it significantly shortens the startup time of the application, thereby improving the user experience. Secondly, the local loading mechanism effectively reduces the dependence on the network, ensuring that the application can run smoothly under various network conditions; finally, by reducing the data transmission volume, the local loading mechanism also helps to reduce the energy consumption of the application, thereby prolonging the battery life of the mobile device.

[0217] Design and function introduction of mobile positioning navigation system: The indoor positioning navigation system of mobile phone is based on the building entity-network-mixed map model, aiming to provide a mobile solution for indoor location services. The system takes F building of Beijing University of Civil Engineering and Architecture as the actual application scene, and its core functions include four main modules: floor display, indoor positioning, path planning and scene roaming.

[0218] The system main interface displays a three-dimensional model of the building, and the user can realize zooming in, zooming out, rotating, floor selection and other operations on the building through intuitive screen interaction. Figure 16a When the user is in the model, the built-in sensor of the mobile phone is combined with the built-in positioning algorithm of the system to obtain the current position information of the user, and is connected with the building map in the system to be displayed in the form of a positioning point, so as to realize the indoor real-time positioning function. Figure 16b In the map interface, the user can select a destination, and the system will automatically plan multiple paths for the user to select. Figure 16c After selecting a specific route, the system will display the corresponding path, and the user can enable the roaming mode to obtain an immersive experience. Figure 16d .

[0219] The building hybrid map model construction device for indoor positioning and navigation provided by the present application is described below, and the building hybrid map model construction device for indoor positioning and navigation described below can be correspondingly referred to the building hybrid map model construction method for indoor positioning and navigation described above.

[0220] The building hybrid map model construction device for indoor positioning and navigation provided by the present application, as shown in Figure 17 , comprises:

[0221] The entity model construction module 310 is used for acquiring building information model data and constructing an entity model based on the building information model data.

[0222] The network model construction module 320 is used for constructing a network model based on the passable elements in the entity model.

[0223] The voxel model construction module 330 is used for constructing a voxel model based on the spatial elements in the entity model.

[0224] The hybrid model construction module 340 is used for constructing a hybrid map model based on the entity model, the network model and the voxel model, acquiring original coordinate data output by a positioning system as a set of matching positioning points, and mapping the set of matching positioning points to the hybrid map model through map matching.

[0225] Figure 18 An example of an entity structure schematic diagram of an electronic device is shown in Figure 18As shown, the electronic device can include a processor 410, a communications interface 420, a memory 430, and a communications bus 440, wherein the processor 410, the communications interface 420, and the memory 430 complete mutual communication through the communications bus 440. The processor 410 can invoke a logical instruction in the memory 430 to execute the building hybrid map model construction method for indoor positioning and navigation.

[0226] In addition, the logical instruction in the memory 430 described above can be implemented in the form of a software function unit and sold or used as an independent product, and 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 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 several instructions for making a computer device (which can be a personal computer, a server, or a network device, etc.) execute all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage medium includes: a U disk, a mobile hard disk, a read-only memory (ROM, Read-Only Memory), a random access memory (RAM, Random Access Memory), a magnetic disk or an optical disk, and various media that can store program codes.

[0227] On the other hand, the present application also provides a computer program product, which includes a computer program, the computer program can be stored on a non-transitory computer readable storage medium, and the computer program is executed by a processor, and the computer can execute the building hybrid map model construction method for indoor positioning and navigation provided by the above-mentioned methods.

[0228] In another aspect, the present application also provides a non-transitory computer readable storage medium, which stores a computer program, and the computer program is executed by a processor to implement the building hybrid map model construction method for indoor positioning and navigation provided by the above-mentioned methods.

[0229] The device embodiments described above are only schematic, wherein the units shown as separate components can or can not be physically separated, and the components shown as units can or can not be physical units, i.e., they can be located in one place, or distributed on multiple network units. Part or all of the modules can be selected according to actual needs to achieve the purpose of the present embodiment. Those skilled in the art can understand and implement without creative labor.

[0230] Those skilled in the art can clearly understand the technical solutions of the various embodiments from the above description of the embodiments, and the various embodiments can be implemented by means of software with the necessary general hardware platforms, and of course, can also be implemented by hardware. Based on such understanding, the above technical solutions, essentially or in other words, the part of the prior art that makes a contribution, can be embodied in the form of a software product, which can be stored in a computer readable storage medium, such as a ROM / RAM, a magnetic disk, an optical disk, and the like, 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 the methods described in the various embodiments or some parts of the embodiments.

[0231] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present application, and not to limit them; although the present application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that: it can still modify the technical solutions recorded in the foregoing embodiments, or make equivalent replacement for some technical features therein; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present application.

Claims

1. A method for constructing a hybrid building map model for indoor positioning and navigation, characterized in that, The method comprises the following steps: obtaining building information model data, and constructing an entity model based on the building information model data; constructing a network model based on passable elements in the entity model; constructing a voxel model based on space elements in the entity model; constructing a hybrid map model based on the entity model, the network model and the voxel model, and obtaining original coordinate data output by a positioning system as a set of to-be-matched positioning points, and mapping the set of to-be-matched positioning points to the hybrid map model through map matching.

2. The building hybrid map model construction method for indoor positioning navigation according to claim 1, characterized in that, The method for constructing an entity model based on building information model data comprises the following steps: importing building geometry information and semantic information based on building information model data, and outputting original modeling data; preprocessing the original modeling data to obtain intermediate model data; extracting basic elements and identification elements from the intermediate model data according to indoor location service requirements to obtain a basic model through modeling; entering semantic information into the basic model to obtain an entity model; the semantic information at least includes room names, functional areas, equipment and facilities and multimedia data. 3.The building hybrid map model construction method for indoor positioning and navigation according to claim 1, wherein, The passable elements include corridors, floor entrances and staircases. The method for constructing a network model based on passable elements in the entity model comprises the following steps: extracting center points of each floor corridor and connecting them to generate a single-layer skeleton network; connecting target nodes one by one and connecting the target nodes and the corridor center points based on the single-layer skeleton network to form a single-layer road network containing horizontal path topology; the target nodes are key functional nodes in the network model, including space connection points such as doors, staircase entrances and elevator halls, and user accessible target points such as room entrances, service facilities and navigation signs; integrating vertical road networks by using the single-layer road network to obtain a network topology integrating horizontal paths and vertical paths; wherein the vertical road network integration at least includes connecting each floor to the connected node representing the floor entrance of the floor; implanting semantic information anchors and map positioning anchors in the network topology to obtain a network model; The semantic information anchors are implanted by the following method: abstracting indoor facilities into facility nodes, configuring multiple semantic information for each facility node to obtain semantic information anchors, and adding the semantic information anchors to the network topology; the multiple semantic information at least includes text, picture, voice and video data; The map positioning anchors are walking anchor nodes, and the map positioning anchors are implanted by the following method: obtaining preset pedestrian movement characteristic parameters; selecting the preset pedestrian movement characteristic parameters as the interval standard of the walking anchor nodes in the horizontal direction based on the network topology; and in the vertical direction, the walking anchor nodes are arranged on each stair step for staircases, and the preset pedestrian movement characteristic parameters are also selected as the interval standard of the walking anchor nodes for staircase platforms. 4.The building hybrid map model construction method for indoor positioning and navigation according to claim 1, wherein, The method for constructing a voxel model based on space elements in the entity model comprises the following steps: performing spatial classification processing on the space elements of the entity model according to space types and passability, and outputting the classified model; performing voxelization preprocessing on the classified model to obtain a voxelization-preprocessed model; The surface voxel labeling result is obtained by performing a scan filling process on each spatial surface of the voxelized preprocessed model using a scan filling method, and the voxelized preprocessed model is internally filled according to the surface voxel labeling result, and a voxel model is output. 5.The building hybrid map model construction method for indoor positioning and navigation according to claim 1, wherein, Based on the entity model, the network model and the voxel model, a hybrid map model is constructed, including: The spatial connection relationship between the entity model, the network model and the voxel model is established, and the mapping relationship between the network model node and the voxel model unit is established, and an integrated hybrid map model is output; the spatial connection relationship includes direct connection and indirect connection; The establishment of the direct connection includes: a one-to-one correspondence is established between the component node in the network model and the corresponding building component in the entity model, and a correlation is established between the spatial set in the voxel model and all building components constituting the space in the entity model; The establishment of the indirect connection includes: the room node in the network model is associated with multiple components in the entity model through the voxel model, and the voxel model unit is associated with the network model node through the entity model component. 6.The building hybrid map model construction method for indoor positioning and navigation according to claim 1, wherein, The raw coordinate data output by the positioning system is obtained as a set of matching positioning points, and the set of matching positioning points is mapped to the hybrid map model through map matching, including: The raw coordinate data output by the positioning system is obtained as a set of matching positioning points; Based on the coordinate system conversion parameters of the hybrid map model, the set of matching positioning points is converted to the model coordinate system, and standardized positioning data is output; The standardized positioning data is processed by using a buffer matching method, and a map matching result data set is output; the buffer matching method includes: a buffer is established with the matching positioning point as the center, and all network nodes and voxel nodes in the buffer are extracted; all network nodes and voxel nodes in the buffer are traversed, and the network node and the voxel node closest to the matching positioning point are determined as the corresponding map matching result. 7.A building hybrid map model construction device for indoor positioning and navigation, characterized by It includes: An entity model construction module is configured to obtain building information model data and construct an entity model based on the building information model data; A network model construction module is configured to construct a network model based on passable elements in the entity model; A voxel model construction module is configured to construct a voxel model based on spatial elements in the entity model; A hybrid model construction module is configured to construct a hybrid map model based on the entity model, the network model and the voxel model, and obtain raw coordinate data output by a positioning system as a set of matching positioning points, and map the set of matching positioning points to the hybrid map model through map matching.

8. An electronic device comprising a memory, a processor, and a computer program stored on the memory and running on the processor, characterized in that, The processor executes the computer program to implement the building hybrid map model construction method for indoor positioning and navigation according to any one of claims 1 to 6. 9.A non-transitory computer-readable storage medium having stored thereon a computer program, characterized in that, The computer program is executed by the processor to implement the building hybrid map model construction method for indoor positioning and navigation according to any one of claims 1 to 6.

10. A computer program product comprising a computer program, characterized in that, The computer program is executed by the processor to implement the building hybrid map model construction method for indoor positioning and navigation according to any one of claims 1 to 6.

Images