Route planning algorithm based on AR indoor navigation

Abstract

The invention discloses a path planning algorithm based on AR indoor navigation, and relates to the field of path planning, and the algorithm comprises the following steps: obtaining building information, constructing a three-dimensional building model, and setting corresponding local area nodes and area description types; acquiring a corresponding real-time environment image in the three-dimensional building model; semantic analysis is carried out on the real-time environment image, and marking is carried out in the three-dimensional building model according to a semantic analysis result and a region description type; obtaining user target information and corresponding user historical path information, and obtaining user preference information according to the user historical path information; setting corresponding path planning information in the three-dimensional building model according to the user target information and the user preference information; obtaining corresponding AR path information according to a semantic marking result of the corresponding real-time environment image in the path planning information; according to the invention, the individuation degree of the user demand in the path planning process is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of path planning, and particularly to a path planning algorithm based on AR indoor navigation. Background Art

[0002] With the continuous development of technology, AR technology has gradually become an indispensable part of people's lives. AR technology can integrate virtual information with the real world, providing users with a more convenient and intuitive usage experience. Traditional path planning methods mainly rely on pre-set map data and rules, and have poor adaptability to indoor spaces with rapidly changing environments and complex structural layouts. In addition, they cannot obtain user needs and environmental information in real time, resulting in the planning results may deviate from the actual usage scenarios. In the field of indoor navigation, the application of AR technology can effectively solve problems such as inaccurate positioning and difficult information acquisition in traditional navigation methods in complex environments, greatly improving the user experience.

[0003] After retrieval, the invention patent with the Chinese patent number CN108759840A discloses an indoor-outdoor integrated three-dimensional navigation path planning method. This method is based on indoor and outdoor three-dimensional models constructed by modeling software such as CityEngine and SketchUp. The three-dimensional geometric center points and lines of the navigable areas in the indoor and outdoor models of the building are respectively extracted, connected in sequence and integrated to form a complete indoor-outdoor integrated three-dimensional navigation path network. Then, the nearest facility point analysis tool in ArcGIS is used for shortest path analysis, and geographical modeling is completed through the ModelBuilder tool. Batch path analysis processing is realized through this model. The three-dimensional navigation path extraction and planning involved in this method provide a useful reference for the indoor-outdoor integrated intelligent navigation of robots or unmanned aerial vehicles.

[0004] Compared with the prior art, the invention patent with the Chinese patent number CN108759840A sets corresponding three-dimensional models according to the models in the building, and obtains the shortest path information according to the three-dimensional geometric center points and lines.

[0005] However, in the actual use process of the above method, only the corresponding building models and the planning of shortest path extraction are considered, without considering the complexity of indoor buildings and the needs of path users. Therefore, the accuracy and user demand individuality in the process of indoor navigation path planning are ignored. Summary of the Invention

[0006] The purpose of the present invention is to solve the disadvantages of low accuracy and low user demand individuality in the prior art, and to propose a path planning algorithm based on AR indoor navigation.

[0007] To achieve the above object, the present invention adopts the following technical solutions:

[0008] A path planning algorithm based on AR indoor navigation, comprising the following steps:

[0009] Step 1: Obtain corresponding building information, set up a three-dimensional building model according to the building information, and set corresponding local nodes and area description types according to the three-dimensional building model;

[0010] Step 2: Set up visual acquisition terminals according to the distribution of each local node in the three-dimensional building model, and obtain corresponding real-time environment images in the three-dimensional building model through the visual acquisition terminals;

[0011] Step 3: Perform semantic analysis on the real-time environment images obtained in the three-dimensional building model, and perform semantic marking on the three-dimensional building model according to the semantic analysis results and corresponding area description types;

[0012] Step 4: Obtain user target information submitted by the user account, collect user historical path information of the corresponding user account, and obtain corresponding user preference information according to the user historical path information;

[0013] Step 5: Analyze and process the corresponding semantic marking results in the three-dimensional building model according to the user target information and user preference information to obtain corresponding path planning information;

[0014] Step 6: Obtain the real-time environment images corresponding to the corresponding positions according to the path planning information, perform virtualization processing according to the semantic marking results corresponding to the real-time environment images, and obtain the AR path information corresponding to the path planning information.

[0015] The above technical solution further includes: The process of setting up the three-dimensional building model includes:

[0016] Obtain corresponding building information, where the building information includes basic geometric information and functional area information;

[0017] Build corresponding plane structures and three-dimensional structures according to the basic geometric information, analyze and process each plane structure in the vertical direction according to the three-dimensional structure to obtain corresponding space structures, set up a three-dimensional building model, describe and mark the constructed space structures according to the functional area information, perform classification processing according to the description and marking results, set the RGB data corresponding to the corresponding description and marking results according to the classification processing results, and map it into the three-dimensional building model to obtain the corresponding area description types in the three-dimensional building model;

[0018] Analyze and process the RGB data corresponding to different region description types within the three-dimensional building model, obtain the connection positions of different RGB data and the covered area of the same RGB data, set corresponding local nodes respectively according to the connection positions of different RGB data and the covered area of the same RGB data, and associate the positions where the local nodes are located with the three-dimensional building model.

[0019] Further, the process of obtaining the real-time environment image includes:

[0020] Obtain the distribution of each local node within the three-dimensional building model, set corresponding visual acquisition terminals according to the distribution of each local node, collect the real-time environment image at the corresponding position within the three-dimensional building model through the visual acquisition terminals, and mark the corresponding acquisition time and visual acquisition terminal.

[0021] Further, the process of semantic analysis of the real-time environment image includes:

[0022] Obtain the real-time environment image and the corresponding acquisition time and visual acquisition terminal, perform grayscale processing on the obtained real-time environment image, and obtain the corresponding real-time grayscale image;

[0023] Obtain the corresponding initial grayscale image according to the visual acquisition terminal corresponding to the real-time grayscale image, compare and analyze the real-time grayscale image corresponding to the corresponding acquisition time with the initial grayscale image, and obtain the corresponding real-time obstacle image according to the comparison and analysis result;

[0024] Preset a real-time monitoring period, map the real-time obstacle images corresponding to the real-time environment images obtained by the corresponding visual acquisition terminals to the corresponding unit time within the real-time monitoring period respectively according to the corresponding acquisition time, obtain the obstacle transformation data of the real-time obstacle images within the real-time monitoring period, and perform image segmentation on each obstacle transformation data;

[0025] Obtain the corresponding image detail data according to the image segmentation result, perform semantic analysis on the image detail data, and obtain the semantic type corresponding to the image detail data.

[0026] Further, the process of semantic marking of the three-dimensional building model includes:

[0027] Obtain the semantic types corresponding to each region description type within the three-dimensional building model, virtualize the real-time obstacle images corresponding to the corresponding semantic types, map the real-time obstacle images corresponding to the virtualization processing result to the corresponding positions within the three-dimensional building model, and perform semantic marking at the corresponding positions.

[0028] Further, the process of obtaining the user target information and user preference information corresponding to the user account includes:

[0029] Set up a user input window, obtain the corresponding user account through the user input window, and the user enters the corresponding user target information according to the needs through the user account;

[0030] Obtain the user's historical path information according to the user account, and respectively obtain the corresponding path feature information and area stay feature information by performing feature extraction on the user's historical path information;

[0031] Construct a user historical path information set according to the user's historical path information, and obtain the normalized evaluation level features corresponding to different feature information by analyzing and processing the feature information corresponding to each user historical path information in the user historical path information set;

[0032] Perform clustering analysis on the evaluation level features corresponding to each path feature information and area stay feature information in the user historical path information set to obtain the user preference data corresponding to the user account.

[0033] Further, the process of obtaining the corresponding path planning information includes:

[0034] Obtain the user target information, and obtain the corresponding user starting point information according to the connection relationship between the user terminal corresponding to the user account and the local node;

[0035] Set the path traversal range according to the positions of the user starting point information and the user target information in the three-dimensional building model, and perform traversal analysis on the local nodes within the path traversal range based on the big data analysis algorithm to obtain all the initial path information;

[0036] Set up an initial path node information table according to all the initial path information, sequentially set each local node planned in the initial path information as the cell corresponding to the corresponding row, and map it to the corresponding cell according to the marking result at the corresponding position in the three-dimensional building model;

[0037] Analyze and process the cells in the rows corresponding to all the initial path information in the initial path node information table and the user preference data, obtain the feature information corresponding to each initial path information, compare and analyze the obtained feature information with the user preference information, and obtain the corresponding similarity data;

[0038] Obtain the weight coefficients corresponding to different feature data in the user preference information, and obtain the initial evaluation data corresponding to each initial path information according to the similarity data and the weight coefficients;

[0039] Perform sorting analysis on the initial evaluation data, and set the corresponding initial path information as the path planning information according to the analysis result.

[0040] Further, the process of obtaining the AR path information corresponding to the path planning information includes:

[0041] Obtain the distribution of the path planning information in the three-dimensional building model, obtain the semantic types corresponding to the corresponding real-time environment images in the three-dimensional building model, perform virtualization settings according to the corresponding semantic types, and obtain the corresponding AR virtual images;

[0042] Integrate and update the AR virtual images at the corresponding positions in the path planning information, obtain the corresponding AR path information according to the update result, send the obtained AR path information to the user account, and let the user provide path planning feedback on the AR path information according to the requirements, and perform real-time adjustment according to the path planning feedback result.

[0043] The present invention has the following beneficial effects:

[0044] 1. In the present invention, by obtaining building information and setting a corresponding three-dimensional building model according to the building information, local nodes are set in the three-dimensional building model, and the position information of the corresponding user is obtained through the connection relationship between the local nodes and the user terminal, which maximally improves the accuracy in the indoor navigation process and avoids the type of event where the navigation is not smooth during the path planning process due to indoor signal reasons or distance reasons.

[0045] 2. In the present invention, by dividing different regions in the three-dimensional building model, setting corresponding RGB data to divide the three-dimensional building model into regions, and obtaining the corresponding region description types, the corresponding region division in the three-dimensional building model becomes visual, and then corresponding path planning information is set according to the visual result, which improves the path planning efficiency to a certain extent.

[0046] 3. In the present invention, by analyzing and processing the data information collected by the visual acquisition terminal, obtaining the semantic analysis results at the corresponding positions in the three-dimensional building model, setting a corresponding real-time monitoring period, and analyzing and processing different obstacle data, it provides a basis for planning in the subsequent path planning process. In addition, by semantically marking the real-time obstacle images and performing virtualization processing according to the semantic marking results, it also protects the privacy of various things in the corresponding building to a certain extent.

[0047] 4. In the present invention, by analyzing and processing the user's historical path information corresponding to the user account, obtaining the user preference data corresponding to the corresponding user account, setting the corresponding path planning information by combining the user preference data with the marking results in the three-dimensional building model, and in addition, performing real-time adjustment according to the user feedback, it improves the personalization in the path planning process corresponding to the user account to a certain extent. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 This is a schematic structural diagram of a path planning algorithm based on AR indoor navigation proposed by the present invention. Detailed implementation manners

[0049] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present invention without creative efforts shall fall within the protection scope of the present invention.

[0050] Embodiment 1

[0051] As Figure 1 shown, a path planning algorithm based on AR indoor navigation proposed by the present invention includes the following steps:

[0052] Step 1: Obtain corresponding building information, set a three-dimensional building model according to the building information, and set corresponding local nodes and area description types according to the three-dimensional building model;

[0053] Step 2: Set a visual acquisition terminal according to the distribution of each local node in the three-dimensional building model, and obtain corresponding real-time environment images in the three-dimensional building model through the visual acquisition terminal;

[0054] Step 3: Perform semantic analysis on the real-time environment images obtained in the three-dimensional building model, and perform semantic marking on the three-dimensional building model according to the semantic analysis results and corresponding area description types;

[0055] Step 4: Obtain user target information submitted by the user account, collect user historical path information of the corresponding user account, and obtain corresponding user preference information according to the user historical path information;

[0056] Step 5: Analyze and process the corresponding semantic marking results in the three-dimensional building model according to the user target information and user preference information to obtain corresponding path planning information;

[0057] Step 6: Obtain the real-time environment images corresponding to the corresponding positions according to the path planning information, perform virtualization processing according to the semantic marking results corresponding to the real-time environment images, and obtain the AR path information corresponding to the path planning information.

[0058] It should be further noted that, in the specific implementation process, the process of obtaining corresponding building information, setting a three-dimensional building model according to the building information, and setting corresponding local nodes and area description types includes:

[0059] Set up an indoor navigation path planning platform, which is used to plan the path information inside the corresponding building according to the user's needs, and display the set path information in AR, so as to facilitate the user to complete the corresponding guidance according to the corresponding path information;

[0060] In addition, there are three ports in the indoor navigation path planning platform, namely the indoor building analysis terminal, the user demand analysis terminal, and the indoor navigation analysis terminal. The path planning process based on AR indoor navigation is completed through the indoor building analysis terminal, the user demand analysis terminal, and the indoor navigation analysis terminal respectively;

[0061] An information entry window and a building management window are set in the indoor building analysis terminal;

[0062] Enter the corresponding building information through the information entry window. The building information includes basic geometric information, functional area division information, operation management information, and entrance and exit traffic information, where:

[0063] The basic geometric information is the plane layout information and three-dimensional structure information inside the corresponding building. The plane local information is the room distribution, corridor direction, building and elevator positions on the corresponding floor, etc. The three-dimensional structure information is the corresponding number of floors, floor height, and vertical drop between floors;

[0064] The functional area division information is the functional distribution information inside the corresponding building, and the obtained functional distribution information is marked with functions;

[0065] The operation management information includes the opening hours information, historical pedestrian flow density data, etc. inside the building corresponding to the corresponding functional area division information;

[0066] The entrance and exit traffic information includes the corresponding main entrance information, secondary entrance information, and traffic connection information;

[0067] The building management window is used to set up a three-dimensional building model according to the obtained building information. The specific implementation process includes:

[0068] Obtain the basic geometric information, and analyze and process it according to the corresponding plane layout information and three-dimensional structure information in the basic geometric information:

[0069] Obtain the plane layout information, build the plane structure according to the plane layout information, and mark and describe the corresponding positions in the built plane structure;

[0070] Obtain the three-dimensional structure information, connect each plane structure according to the three-dimensional structure information, analyze and process each plane structure in the vertical direction, obtain the corresponding space structure, and mark and describe the corresponding positions in the built space structure;

[0071] Set a 3D building model by obtaining the marker description results of the spatial structure and the planar structure;

[0072] Analyze and process the constructed 3D building model to obtain the corresponding marker descriptions within the planar structure and the spatial structure in the 3D building model, and analyze and process each marker description. The specific implementation process includes:

[0073] Classify the marker descriptions in the 3D building model, and set classification criteria in combination with function distinction information, entrance and exit traffic information, and operation management information. The classification criteria are the marker description types corresponding to public description information and non-public description information. In addition, during the process of the marker description types corresponding to public description information and non-public description information, set the corresponding color system mapping table, obtain the publicity degree and non-publicity degree of different marker description types in the public description information and non-public description information and correspond them with the corresponding color system mapping table, obtain the RGB data corresponding to the position of the corresponding marker description, and map the corresponding RGB data to the corresponding position in the 3D building model to obtain the corresponding area description type in the 3D building model;

[0074] Obtain the corresponding RGB data in the 3D building model, and set a corresponding local area network for the corresponding RGB data. The local area network is connected to the corresponding satellite navigation system, obtain the accurate location of the user in the corresponding 3D building model through the local area network, and upload the accurate location to the corresponding satellite navigation system through the local area network;

[0075] Obtain the RGB data corresponding to the public description information, and set the corresponding connection local area nodes according to the connection positions of different RGB data.

[0076] Obtain the area coverage corresponding to the same RGB data, and set area local area nodes according to the area coverage.

[0077] Set a corresponding local area network for the location information of the connection local area nodes and the area local area nodes in the 3D building model obtained, and mark and store the connection local area nodes and the area local area nodes in the local area network.

[0078] It should be further noted that in the specific implementation process, the process of setting a visual acquisition terminal according to the distribution of each local area node in the 3D building model and obtaining the corresponding real-time environment image in the 3D building model through the visual acquisition terminal includes:

[0079] Obtain the corresponding local area network in the corresponding 3D building model, and obtain the distribution of the corresponding area local area nodes and connection local area nodes in the local area network;

[0080] Set visual acquisition terminals according to the distribution of local nodes in the local area network, obtain the acquisition ranges corresponding to the respective visual acquisition terminals, obtain the coverage quantities of the local nodes within the respective acquisition ranges, and mark the visual acquisition terminals according to the coverage quantities of the respective local nodes;

[0081] Obtain the coverage quantities of the respective visual acquisition terminals, and set the coverage data sets corresponding to the respective visual acquisition terminals according to the local nodes corresponding to the coverage quantities;

[0082] Centering on the respective visual acquisition terminals, set a data acquisition radius, obtain the respective visual acquisition terminals within the data acquisition radius, sequentially combine the visual acquisition terminal corresponding to the center with the visual acquisition terminals within the data acquisition radius, obtain the corresponding coverage data sets according to the combination results, perform an intersection process on the pairwise corresponding coverage data sets, and obtain the intersection process results corresponding to the respective combinations;

[0083] Perform planning and management on the visual acquisition terminals within the data acquisition radius for the intersection process results, obtain the visually acquisition terminals with redundant planning, and mark the local nodes corresponding to the visually acquisition terminals with redundant planning;

[0084] Install the respective visual acquisition terminals according to the marking results of the local area network, and obtain the distribution of the respective visual acquisition terminals within the three-dimensional building model;

[0085] According to the distribution of the respective visual acquisition terminals within the three-dimensional building model, collect the respective real-time environment images by the respective visual acquisition terminals.

[0086] It should be further noted that in the specific implementation process, the process of performing semantic analysis on the real-time environment images obtained within the three-dimensional building model and performing semantic marking on the three-dimensional building model according to the semantic analysis results and the corresponding regional description types includes:

[0087] Obtain the real-time environment images collected by the respective visual acquisition terminals within the three-dimensional building model through the indoor navigation analysis terminal, perform quantization processing on the obtained real-time environment images, and perform semantic marking on the three-dimensional building model according to the quantization processing results. The specific implementation process includes:

[0088] Perform grayscale processing on the real-time environment images collected by the visual acquisition terminals to obtain the corresponding real-time grayscale images;

[0089] The initial environment images corresponding to the respective visual acquisition terminals are stored in the indoor navigation analysis terminal. The initial environment images are those collected within the corresponding three-dimensional building model without being used. Perform grayscale processing on the initial environment images to obtain the corresponding initial grayscale images;

[0090] Compare and analyze the obtained real-time grayscale image with the initial grayscale image, and obtain the corresponding real-time obstacle image according to the comparison and analysis results;

[0091] Preset a real-time monitoring period, and map the acquisition time corresponding to the real-time environment image obtained by the visual acquisition terminal to the unit time corresponding to the real-time monitoring period;

[0092] Analyze and process the real-time obstacle images corresponding to each unit time within the real-time monitoring period in sequence;

[0093] Based on the real-time obstacle image corresponding to the first unit time within the real-time monitoring period, sequentially obtain the real-time obstacle images corresponding to the remaining unit times, and obtain the obstacle transformation data corresponding to the real-time obstacle images within adjacent unit times, and perform image segmentation according to the obstacle transformation data;

[0094] If the corresponding positions of the real-time obstacle images corresponding to each unit time within the real-time monitoring period have not changed, obtain the image area where there is no change, perform segmentation processing on this area, and set a fixed obstacle mark;

[0095] If there are different changes in the corresponding positions of the real-time obstacle images corresponding to each unit time within the real-time monitoring period, obtain the unit time corresponding to the real-time obstacle image during the change process, perform regional division according to the change process of the real-time obstacle images within different unit times, perform segmentation processing on the corresponding area, and set it as a flexible obstacle mark;

[0096] Obtain the marking results of the corresponding areas within the real-time obstacle image, and perform image semantic analysis according to the fixed obstacle marking results and the flexible obstacle marking results;

[0097] Construct a corresponding semantic analysis model based on the deep learning algorithm; input the marking results of the corresponding areas within the obtained real-time obstacle image into the semantic analysis model respectively, and the semantic analysis model traverses and analyzes each pixel point within the real-time obstacle image to obtain image detail data, perform semantic analysis on the image detail data, and obtain the semantic type corresponding to the image detail data;

[0098] Obtain the semantic types corresponding to the description types of each area within the three-dimensional building model, virtualize the real-time obstacle images corresponding to the corresponding semantic types, map the real-time obstacle images corresponding to the virtualization processing results to the corresponding positions within the three-dimensional building model, and perform semantic marking at the corresponding positions.

[0099] It should be further noted that in the specific implementation process, the process of obtaining the user target information submitted by the user account, collecting the user historical path information of the corresponding user account, and obtaining the corresponding user preference information includes:

[0100] The user requirement analysis terminal is used to manage the corresponding user accounts, obtain the corresponding user target information and user preference information. The specific implementation process includes:

[0101] A user input window and a user management window are set in the user requirement analysis terminal;

[0102] Obtain the corresponding user account through the user input window, judge whether the corresponding user is a new user or an old user according to the user account, and complete the corresponding login through the user account;

[0103] If the user corresponding to the user account is a new user, obtain the user's subjective preference information, where the user's subjective preference information is the user's subjective preference information entered by the corresponding user, including route feature preference, transportation mode preference, path complexity preference, time tolerance preference, and route environment preference, etc.;

[0104] If the user corresponding to the user account is a new user, obtain the user's historical path information, where the user's historical path information includes historical path information, path passing location information, itinerary additional information, etc.;

[0105] Enter the user target information of the logged-in user account through the corresponding user terminal, obtain the connection result between the user terminal and the corresponding local node in the three-dimensional building model, obtain the corresponding user starting point information, and associate the user starting point information with the user target information;

[0106] Analyze and process the user's subjective preference information or user's historical path information obtained in the user account to obtain the corresponding user preference information. The specific implementation process includes:

[0107] Obtain the user's subjective preference information, extract keywords from the user's subjective preference information, obtain the corresponding feature information according to the keyword extraction result, and mark the obtained feature information as the user preference information corresponding to the user account;

[0108] Obtain the user's historical path information, set a user historical path information set according to the user's historical path information, respectively extract features from the corresponding user historical path information in the user historical path information set, and obtain path feature information and regional stay feature information, where:

[0109] The path feature information is the evaluation level feature corresponding to the path length and path complexity data between the user's starting point information and the user's target information. All the path lengths and path complexities within all the known passable paths are obtained based on the user's starting point information and the user's target information. The corresponding path length and path complexity data in the user's historical path information are evaluated respectively according to all the path lengths and path complexities to obtain the evaluation level features of the path length and the path complexity;

[0110] The area stay feature information is the evaluation level feature corresponding to different functional areas and obstacle avoidance areas obtained by analyzing and processing the stay time and frequency of the user in different indoor functional areas and the avoidance range and frequency of the obstacle avoidance areas;

[0111] In addition, it should be further noted that in the specific implementation process, the evaluation level features corresponding to the path feature information and the area stay feature information are obtained respectively according to the situations corresponding to each user's historical path information, and the evaluation level features include the corresponding maximum and minimum values;

[0112] The evaluation level features corresponding to the path feature information and the area stay feature information are respectively normalized;

[0113] The evaluation level features corresponding to the path feature information and the area stay feature information are obtained and marked as x, and the evaluation level features corresponding to the normalized ones are marked as X new ;

[0114]

[0115] The normalized analysis results corresponding to each user's historical path information in the user's historical path information set are clustered and analyzed for each information in the user's historical path information set through the K-Means clustering analysis algorithm to obtain the key evaluation level features corresponding to the path feature information and the area stay feature information;

[0116] The key evaluation level features corresponding to the path feature information and the area stay feature information are set as the user preference data, and the obtained user preference data is associated with the corresponding user account.

[0117] It should be further noted that in the specific implementation process, the process of analyzing and processing the corresponding semantic marking results in the three-dimensional building model according to the user's target information and the user's preference information to obtain the corresponding path planning information includes:

[0118] The indoor navigation analysis terminal is used to obtain the user's starting point information, user target information, and user preference information corresponding to the user account, analyze and process the corresponding data information in the user account based on the three-dimensional building model, and obtain the corresponding path planning information. The specific implementation process includes:

[0119] Map the user's starting point information and user target information into the three-dimensional building model, and sequentially analyze and process them according to the position information of the three-dimensional building model to which the user's starting point information and user target information belong, and set the path traversal range;

[0120] Obtain the connection local nodes and regional local nodes corresponding to the corresponding path traversal range in the three-dimensional building model;

[0121] Based on the big data analysis algorithm, traverse and analyze the connection local nodes and regional local nodes within the path traversal range to obtain all the initial path information. The initial path information includes the RGB data and corresponding semantic types in the three-dimensional building model corresponding to all local nodes in each initial path information;

[0122] Set the initial path node information table according to each local node included in each initial path information;

[0123] Map the RGB data and semantic types corresponding to each local node in the initial path information into the corresponding cells in sequence to obtain the initial path node information table corresponding to each initial path information;

[0124] Obtain the user preference information, and analyze and process each initial path information in the initial path node information table based on the user preference information;

[0125] Obtain the evaluation level characteristics corresponding to the path feature information and regional stay feature information in the user preference information. Analyze and process the evaluation level characteristics of different functional areas in the regional stay feature information and the evaluation level characteristics corresponding to the obstacle avoidance areas with the data information obtained in each cell in the initial path node information table to obtain the similarity data between each cell and the evaluation level characteristics corresponding to the user preference information, and perform marking processing;

[0126] In addition, it should be further noted that in the specific implementation process, during the process of analyzing and processing the data in the corresponding cells, obtain the corresponding real-time obstacle data, compare and analyze the real-time obstacle data with the avoidance range of the obstacle avoidance area in the user preference data, and judge whether the corresponding path meets the user preference data;

[0127] Obtain the rows corresponding to each cell of the initial path information, add up the path lengths corresponding to each cell in the corresponding row, compare and analyze the semantic marking results corresponding to each cell in the corresponding row with the semantic marking results corresponding to its adjacent cells. If the RGB data types in adjacent cells are the same, the path complexity is not increased; if the RGB data types in adjacent cells are different, the path complexity is increased;

[0128] Obtain the path length and path complexity corresponding to the corresponding initial path information;

[0129] According to the user preference data, normalize the path lengths and path complexities corresponding to the initial path information of all rows in the initial path node information table, and obtain the path feature information corresponding to each initial path information;

[0130] Obtain the path feature information and area stay feature information corresponding to the corresponding initial path information, and compare and analyze the obtained path feature information and area stay feature information with the user preference information;

[0131] Obtain the similarity data β between the path feature information and area stay feature information corresponding to the corresponding initial path information and the user preference information i ;

[0132] Respectively set the weight coefficients corresponding to the path feature information and area stay feature information. The specific implementation process includes:

[0133] Set matrix X according to the user preference data corresponding to each historical path information in the user historical path information set; perform standardization processing on matrix X, mark the standardized matrix as matrix Z, and calculate its corresponding covariance matrix Cov(Z), where:

[0134]

[0135] Respectively calculate the eigenvalues γ1, γ2,..., γ of the covariance matrix n and eigenvectors α1, α2,..., α n ; Sort the eigenvalues from large to small, select the corresponding cumulative contribution rate, obtain the corresponding weight coefficients according to the cumulative contribution rate, and mark the corresponding function types as i, i = 1, 2, k;

[0136] Mark the weight coefficients as ω i , where,

[0137] Analyze and process the initial path information according to the weight coefficients corresponding to the user preference data corresponding to different functional types, and obtain the initial evaluation data corresponding to the initial path information, CP, where:

[0138]

[0139] Sort according to the initial evaluation data corresponding to each initial path information, obtain the initial path information with the largest initial evaluation data according to the sorting result, and mark it as the path planning information.

[0140] It should be further noted that in the specific implementation process, the process of obtaining the real-time environmental image corresponding to the corresponding position according to the path planning information, and performing virtualization processing according to the semantic marking result corresponding to the real-time environmental image to obtain the AR path information corresponding to the path planning information includes:

[0141] The indoor navigation analysis end is used to obtain the path planning information corresponding to the corresponding user terminal, perform AR analysis according to the obtained path planning information, and obtain the corresponding AR path information;

[0142] Obtain the classification marking result in the corresponding real-time acquisition image in the three-dimensional building model to which the corresponding user terminal belongs in the path planning information for virtualization setting, and map it to the corresponding position according to the virtualization setting result;

[0143] When the user terminal reaches the corresponding local node, perform virtualization processing on the real-time acquisition image corresponding to the periphery of the corresponding local node to obtain the AR virtual image corresponding to the local node;

[0144] Integrate and update the AR virtual images at the corresponding positions in the path planning information in real time, obtain the corresponding AR path information according to the update result, and send the obtained AR path information to the user account. The user makes a path planning feedback on the AR path information according to the needs, and makes a real-time adjustment according to the path planning feedback result.

[0145] Although the embodiments of the present invention have been shown and described, for those of ordinary skill in the art, it can be understood that various changes, modifications, substitutions, and deformations can be made to these embodiments without departing from the principles and spirits of the present invention. The scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A path planning algorithm based on AR indoor navigation, characterized in that: The following steps are involved: Step 1: Obtain corresponding building information, set a three-dimensional building model according to the building information, and set corresponding local nodes and regional description types according to the three-dimensional building model; Step 2: Setting a visual acquisition terminal according to the distribution of each local node in the three-dimensional building model, and obtaining the corresponding real-time environment image in the three-dimensional building model through the visual acquisition terminal; Step 3: Perform semantic analysis on the real-time environment image obtained in the three-dimensional building model, and semantically mark the three-dimensional building model according to the semantic analysis results and the corresponding area description type; Step 4: Obtain the user target information submitted by the user account, collect the user history path information of the corresponding user account, and obtain the corresponding user preference information according to the user history path information; Step 5: Analyze and process the corresponding semantic labeling results in the 3D building model according to the user target information and the user preference information to obtain the corresponding path planning information; Step 6: Obtain a real-time environment image corresponding to the corresponding position according to the path planning information, perform virtualization processing according to the semantic labeling result corresponding to the real-time environment image, and obtain AR path information corresponding to the path planning information.

2. The path planning algorithm based on AR indoor navigation according to claim 1, characterized in that: The process of setting up a 3D building model includes: Acquire corresponding building information, wherein the building information includes basic geometric information and functional differentiation information; Build corresponding plane structures and three-dimensional structures according to basic geometric information, analyze and process each plane structure in the vertical direction according to the three-dimensional structure, obtain the corresponding spatial structure, and set the three-dimensional building model, describe and mark the constructed spatial structure according to the functional distinction information, classify and process the description and marking results, set the RGB data corresponding to the corresponding description and marking results according to the classification processing results, and map them to the three-dimensional building model to obtain the corresponding area description type in the three-dimensional building model; The RGB data corresponding to different area description types in the three-dimensional building model are analyzed and processed to obtain the connection positions of different RGB data and the coverage areas of the same RGB data. Corresponding local nodes are set according to the connection positions of different RGB data and the coverage areas of the same RGB data, and the positions of the local nodes are correlated with the three-dimensional building model.

3. The path planning algorithm based on AR indoor navigation according to claim 2, characterized in that: The process of acquiring real-time environmental images includes: The distribution of each local node in the three-dimensional building model is obtained, and a corresponding visual acquisition terminal is set according to the distribution of each local node. The real-time environmental image at the corresponding position in the three-dimensional building model is acquired through the visual acquisition terminal, and the corresponding acquisition time and visual acquisition terminal are marked.

4. The path planning algorithm based on AR indoor navigation according to claim 3 is characterized in that: The process of semantic analysis of real-time environment images includes: Acquire a real-time environment image and the corresponding acquisition time and visual acquisition terminal, and grayscale the acquired real-time environment image to acquire a corresponding real-time grayscale image; Obtain a corresponding initial grayscale image according to the visual acquisition terminal corresponding to the real-time grayscale image, compare and analyze the real-time grayscale image corresponding to the corresponding acquisition time with the initial grayscale image, and obtain a corresponding real-time obstacle image according to the comparison and analysis result; Preset a real-time monitoring cycle, map the real-time obstacle images corresponding to the real-time environment images obtained by the corresponding visual acquisition terminal to the corresponding unit time in the real-time monitoring cycle according to the corresponding acquisition time, obtain the obstacle transformation data of the real-time obstacle image in the real-time monitoring cycle, and perform image segmentation on each obstacle transformation data; The corresponding image detail data is obtained according to the image segmentation result, and the image detail data is semantically analyzed to obtain the semantic type corresponding to the image detail data.

5. The path planning algorithm based on AR indoor navigation according to claim 4, characterized in that: The process of semantically labeling a 3D building model includes: The semantic types corresponding to the description types of each area in the three-dimensional building model are obtained, and the real-time obstacle images corresponding to the corresponding semantic types are virtualized. The real-time obstacle images corresponding to the virtualization processing results are mapped to the corresponding positions in the three-dimensional building model, and semantic marking is performed at the corresponding positions.

6. The path planning algorithm based on AR indoor navigation according to claim 5, characterized in that: The process of obtaining the user target information and user preference information corresponding to the user account includes: Set up a user entry window, obtain the corresponding user account through the user entry window, and the user enters the corresponding user target information through the user account according to needs; Obtain the user's historical path information according to the user account, and extract features from the user's historical path information to obtain corresponding path feature information and area stay feature information; Construct a user historical path information set according to the user historical path information, and obtain normalized evaluation level features corresponding to different feature information by analyzing and processing the feature information corresponding to each user historical path information in the user historical path information set; Cluster analysis is performed on the evaluation level features corresponding to each path feature information and regional stay feature information in the user's historical path information set to obtain user preference data corresponding to the user account.

7. The path planning algorithm based on AR indoor navigation according to claim 6, characterized in that: The process of obtaining the corresponding path planning information includes: Obtain user target information, and obtain corresponding user starting point information according to the connection relationship between the user terminal corresponding to the user account and the local node; The path traversal range is set according to the location of the three-dimensional building model to which the user's starting point information and the user's target information belong, and the local nodes within the path traversal range are traversed and analyzed based on the big data analysis algorithm to obtain all the initial path information; An initial path node information table is set according to all the initial path information, each local node planned in the initial path information is set to a cell corresponding to a corresponding row in turn, and is mapped to a corresponding cell according to the marking result at the corresponding position in the three-dimensional building model; Analyze and process the cells in the rows corresponding to all the initial path information in the initial path node information table and the user preference data, obtain the feature information corresponding to each initial path information, compare and analyze the obtained feature information with the user preference information, and obtain the corresponding similarity data; Obtain weight coefficients corresponding to different feature data in the user preference information, and obtain initial evaluation data corresponding to each initial path information based on the similarity data and the weight coefficients; The initial evaluation data is sorted and analyzed, and the corresponding initial path information is set as the path planning information according to the analysis results.

8. The path planning algorithm based on AR indoor navigation according to claim 7, characterized in that: The process of obtaining the AR path information corresponding to the path planning information includes: Obtain the distribution of path planning information in the three-dimensional building model, obtain the semantic type corresponding to the corresponding real-time environment image in the three-dimensional building model, perform virtualization settings according to the corresponding semantic type, and obtain the corresponding AR virtual image; The AR virtual images at the corresponding positions in the path planning information are integrated and updated in real time. The corresponding AR path information is obtained according to the update results, and the obtained AR path information is sent to the user account. The user provides path planning feedback to the AR path information according to needs, and real-time adjustments are made based on the path planning feedback results.

Images