A method for positioning and labeling floor locations

Abstract

The application provides a floor position positioning and labeling method, floor geomagnetic data is extracted from floor sensor data; trajectory constraint is performed on the floor geomagnetic data to obtain a geomagnetic vector sequence, multi-scale extraction is performed on the geomagnetic vector sequence to obtain geomagnetic space-time dependent features; a positioning feature matching pool is determined based on the geomagnetic space-time dependent features and a prior global map, matching is performed according to the positioning feature matching pool and an environment image set to obtain a positioning matching entropy of the environment; motion data of a target object in a target building is obtained, and a positioning matching trajectory of the target object is determined according to the motion data and a position fingerprint database of the target building; the floor position where the target object is located is determined through the positioning matching entropy and the positioning matching trajectory, and then position labeling is performed in the prior global map of the target building. The above scheme can improve the representation and recognition ability of geomagnetic features in a wide or narrow floor environment, thereby increasing the floor position recognition accuracy and robustness.

Description

Technical Field

[0001] This application relates to the field of floor positioning technology, and more specifically, to a method for locating and marking floor positions. Background Technology

[0002] Floor location and marking refer to the precise determination and recording of the exact location of each floor within a multi-story building to facilitate navigation, management, and emergency response. This process is crucial in modern buildings, especially in complex structures such as large commercial complexes, hospitals, and high-rise residential buildings.

[0003] When performing floor location positioning, GPS signals inside buildings are often very weak or completely unavailable. Different buildings have different structures, materials, and layouts, leading to severe signal reflection and attenuation issues, significantly reducing the accuracy of wireless signal-based positioning systems (such as Wi-Fi and Bluetooth beacons). Secondly, multi-story buildings have a large number of floors and complex layouts, adding additional dimensions and complexity to the positioning system. In existing technologies, visual positioning in large indoor scenes requires feature searching in a large prior 3D map database, resulting in high complexity for positioning initialization. Furthermore, visual discrepancies in similar features within large indoor scenes can cause positioning initialization failures. In mixed topology environments, floors suffer from insufficient distinguishability of geomagnetic features, resulting in poor representation and identification capabilities of geomagnetic features in open or narrow environments. Therefore, improving the representation and identification capabilities of geomagnetic features in open or narrow environments to increase the accuracy and robustness of floor location identification has become a challenge for the industry. Summary of the Invention

[0004] This application provides a method for locating and marking floor positions, which can improve the ability to characterize and identify geomagnetic features in open or narrow environments, thereby increasing the accuracy and robustness of floor position identification.

[0005] Firstly, this application provides a method for locating and marking floor positions, comprising the following steps:

[0006] Collect floor sensor data for the target building;

[0007] Floor geomagnetic data is extracted from the floor sensor data. The floor geomagnetic data is then subjected to trajectory constraints based on the scene constraints of the target building to obtain a geomagnetic vector sequence. The geomagnetic vector sequence is then extracted at multiple scales to obtain geomagnetic spatiotemporal dependence features.

[0008] Based on the geomagnetic spatiotemporal dependence features and the prior global map of the target building, a location feature matching pool that integrates the geomagnetic features of each floor is determined. The location feature matching pool is then matched with the image set of the environment in which the target object is located in the target building to obtain the location matching entropy of the environment.

[0009] Acquire motion data of a target object within a target building, and determine the positioning and matching trajectory of the target object based on the motion data and the target building's location fingerprint database;

[0010] The floor location of the target object is determined by the location matching entropy and the location matching trajectory, and then the location is marked in the prior global map of the target building.

[0011] In some embodiments, extracting floor geomagnetic data from the floor sensor data specifically includes:

[0012] Extract all floor geomagnetic records from the floor sensor data;

[0013] Drift correction was performed on all floor geomagnetic records to obtain corrected floor geomagnetic records.

[0014] The floor geomagnetic data is determined based on the corrected geomagnetic records of all floors.

[0015] In some embodiments, floor sensor data of the target building are acquired using the built-in magnetic sensor and built-in gyroscope of a smartphone.

[0016] In some embodiments, the process of applying trajectory constraints to the floor geomagnetic data based on scene constraints of the target building to obtain a geomagnetic vector sequence specifically includes:

[0017] The floor geomagnetic data is fitted into a floor geomagnetic trajectory;

[0018] The floor geomagnetic trajectory is constrained according to the scene constraints of the target building to obtain the floor geomagnetic constrained trajectory;

[0019] Extract the geomagnetic vector sequence from the geomagnetic constraint trajectory of the floor.

[0020] In some embodiments, multi-scale extraction of the geomagnetic vector sequence to obtain geomagnetic spatiotemporal dependent features specifically includes:

[0021] The geomagnetic vector sequence is divided into multiple geomagnetic vector segments according to a preset sliding window.

[0022] All geomagnetic vector segments are input into a recurrent neural network model for multi-scale extraction, thereby obtaining geomagnetic spatiotemporal dependent features.

[0023] In some embodiments, determining the location feature matching pool that fuses the geomagnetic features of each floor based on the geomagnetic spatiotemporal dependence features and the prior global map of the target building specifically includes:

[0024] Based on the aforementioned geomagnetic spatiotemporal dependence features, the prior global map of the target building is cropped to obtain a local prior 3D map that integrates the geomagnetic features of each floor.

[0025] The local prior 3D map is used for geomagnetic positioning to obtain geomagnetic positioning results;

[0026] The cost of geomagnetic positioning is determined based on the geomagnetic positioning results.

[0027] Based on the geomagnetic positioning results, the geomagnetic positioning cost, and the prior global map, a positioning feature matching pool is determined that integrates the geomagnetic features of each floor.

[0028] In some embodiments, determining the location matching trajectory of a target object based on the motion data and the target building's location fingerprint database specifically includes:

[0029] The motion data is fitted to obtain the motion trajectory of the target object;

[0030] Extract the received signal strength sequence from the location fingerprint database of the target building;

[0031] The positioning and matching trajectory of the target object is determined by the motion trajectory of the target object and the received signal strength sequence.

[0032] In some embodiments, location marking in the prior global map of the target building means displaying the floor location of the target object in the prior global map of the target building in real time.

[0033] Secondly, this application provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, so that the computer device performs the above-described method for locating and marking floor positions.

[0034] Thirdly, this application provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the aforementioned method for locating and marking floor positions.

[0035] The technical solutions provided by the embodiments disclosed in this application have the following beneficial effects:

[0036] This application provides a method for locating and labeling floor positions, which involves collecting floor sensor data of a target building; extracting floor geomagnetic data from the floor sensor data; subjecting the floor geomagnetic data to trajectory constraints based on scene constraints of the target building to obtain a geomagnetic vector sequence; further extracting the geomagnetic vector sequence at multiple scales to obtain geomagnetic spatiotemporal dependence features; determining a location feature matching pool that fuses the floor geomagnetic features based on the geomagnetic spatiotemporal dependence features and a prior global map of the target building; matching the location feature matching pool with an image set of the environment in which the target object is located in the target building to obtain the location matching entropy of the environment; acquiring motion data of the target object in the target building; determining the location matching trajectory of the target object based on the motion data and the location fingerprint database of the target building; determining the floor position of the target object using the location matching entropy and the location matching trajectory; and then labeling the position in the prior global map of the target building.

[0037] Therefore, this application demonstrates an improvement in the representation and identification capabilities of geomagnetic features in open or narrow environments. Specifically, scene and trajectory constraints eliminate unreasonable trajectories and geomagnetic data. Through sliding window and multi-scale extraction, the variation characteristics of geomagnetic data at different temporal and spatial scales can be captured, making the extracted geomagnetic features more spatiotemporally dependent. The introduction of spatiotemporally dependent geomagnetic features allows the floor positioning model to capture longer-term and more stable feature information, improving its robustness to environmental changes. Furthermore, the positioning feature matching pool, which integrates floor geomagnetic features, combines precise geomagnetic features with a prior global map of the building, providing accurate matching criteria. The environmental image set provides additional contextual information, helping to better match environmental features and reducing errors that may occur due to relying solely on geomagnetic data. Secondly, combining motion data with signal strength data from a location fingerprint database effectively fuses information from different sources. Furthermore, by combining motion data and the location fingerprint database to determine the positioning matching trajectory, real-time trajectory tracking of the target object can be achieved. Finally, the floor location of the target object is determined and labeled using the positioning matching entropy and the positioning matching trajectory. In summary, the technical solution adopted in this application can improve the characterization and identification ability of geomagnetic features in open or narrow environments, thereby increasing the accuracy and robustness of floor location identification. Attached Figure Description

[0038] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this application. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0039] Figure 1 This is an exemplary flowchart of a method for locating and marking floor positions according to some embodiments of this application;

[0040] Figure 2 This is an exemplary flowchart illustrating the determination of geomagnetic vector sequences according to some embodiments of this application;

[0041] Figure 3 This is an exemplary flowchart illustrating the determination of a location matching trajectory according to some embodiments of this application;

[0042] Figure 4 This is a schematic diagram of the structure of a computer device for implementing a method for locating and marking floor positions according to some embodiments of this application. Detailed Implementation

[0043] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0044] This application provides a method for locating and labeling floor positions. The core of this method is to collect floor sensor data of a target building; extract floor geomagnetic data from the floor sensor data; apply trajectory constraints to the floor geomagnetic data based on scene constraints of the target building to obtain a geomagnetic vector sequence; then extract the geomagnetic vector sequence at multiple scales to obtain geomagnetic spatiotemporal dependency features; determine a location feature matching pool that fuses the floor geomagnetic features based on the geomagnetic spatiotemporal dependency features and a prior global map of the target building; match the location feature matching pool with an image set of the environment in which the target object is located within the target building to obtain the location matching entropy of the environment; acquire motion data of the target object within the target building; determine the location matching trajectory of the target object based on the motion data and the location fingerprint database of the target building; determine the floor position of the target object using the location matching entropy and the location matching trajectory; and then label the position on the prior global map of the target building. This method can improve the representation and identification capabilities of geomagnetic features in open or narrow floor environments, thereby increasing the accuracy and robustness of floor position recognition.

[0045] To better understand the above technical solutions, a detailed description of the solutions will be provided below in conjunction with the accompanying drawings and specific implementation methods. (Reference) Figure 1The figure is an exemplary flowchart of a method for locating and marking floor positions according to some embodiments of this application. The method 100 for locating and marking floor positions mainly includes the following steps:

[0046] In step 101, floor sensor data of the target building are collected.

[0047] In practice, the built-in magnetic sensor and gyroscope of a smartphone can be used to collect floor sensor data of the target building. This floor sensor data includes floor geomagnetic records and motion data of the target object. The floor geomagnetic records include the geomagnetic components of the floor, the recording time, and the recording number. The motion data includes changes in the target object's speed and orientation. It should be noted that the built-in magnetic sensor of the smartphone can detect minute changes in the magnetic field inside the target building. These changes are mainly caused by building materials (such as reinforced concrete) and electrical equipment (such as elevators and air conditioners). The built-in gyroscope of the smartphone can provide the rotation and motion information of the user device. That is, when the target object moves inside the target building, the gyroscope can record the target object's orientation changes and walking path.

[0048] In step 102, floor geomagnetic data is extracted from the floor sensor data. The floor geomagnetic data is then subjected to trajectory constraints based on the scene constraints of the target building to obtain a geomagnetic vector sequence. The geomagnetic vector sequence is then extracted at multiple scales to obtain geomagnetic spatiotemporal dependence features.

[0049] Optionally, in some embodiments, extracting floor geomagnetic data from the floor sensor data can be done in the following manner:

[0050] Extract all floor geomagnetic records from the floor sensor data;

[0051] Drift correction was performed on all floor geomagnetic records to obtain corrected floor geomagnetic records.

[0052] The floor geomagnetic data is determined based on the corrected geomagnetic records of all floors.

[0053] In practice, firstly, all floor geomagnetic records can be extracted from the floor sensor data, i.e., all floor geomagnetic records can be retrieved by traversing the record numbers; then, since the built-in magnetic sensor may drift over long-term use, i.e., the magnetic sensor reading gradually deviates from the actual value over time, it is necessary to perform drift correction on all extracted floor geomagnetic records. In this application, Kalman filtering technology is used for drift correction to eliminate noise and drift in the floor geomagnetic records; finally, the set of all corrected floor geomagnetic records can be used as the floor geomagnetic data.

[0054] Preferably, in some embodiments, reference is made to Figure 2 As shown, this figure is an exemplary flowchart of determining the geomagnetic vector sequence in some embodiments of this application. In this embodiment, the geomagnetic data of the floors is subjected to trajectory constraints based on the scene constraints of the target building to obtain the geomagnetic vector sequence, which can be achieved by the following steps:

[0055] First, in step 1021, the floor geomagnetic data is fitted into a floor geomagnetic trajectory;

[0056] Then, in step 1022, the floor geomagnetic trajectory is constrained according to the scene constraints of the target building to obtain the floor geomagnetic constrained trajectory;

[0057] Finally, in step 1023, the geomagnetic vector sequence is extracted from the geomagnetic constraint trajectory of the floor.

[0058] In practice, the process begins with arranging the floor geomagnetic data chronologically to form preliminary geomagnetic trajectories. Interpolation methods (such as linear interpolation and spline interpolation) are then used to smooth the data, fitting a continuous curve—the floor geomagnetic trajectory. Next, the floor geomagnetic trajectory is constrained by the target building's scene constraints. This involves creating a detailed scene model of the target building, including floor plans, layout, room locations, corridors, staircases, and elevators, generating a detailed scene constraint model. This model defines the accessibility and limitations of each area. The fitted floor geomagnetic trajectory is compared with the scene constraint model, identifying and correcting any unreasonable trajectory parts. The adjusted trajectory is then re-fitted and smoothed to obtain a floor geomagnetic constraint trajectory that conforms to the scene constraints. Finally, geomagnetic vectors are extracted from the floor geomagnetic constraint trajectory in chronological order. Each geomagnetic vector contains a magnetic field component, a timestamp, and a trajectory direction, resulting in a sequence of all geomagnetic vectors forming a geomagnetic vector sequence.

[0059] Preferably, in some embodiments, multi-scale extraction of the geomagnetic vector sequence to obtain geomagnetic spatiotemporal dependence features can be performed in the following manner:

[0060] The geomagnetic vector sequence is divided into multiple geomagnetic vector segments according to a preset sliding window.

[0061] All geomagnetic vector segments are input into a recurrent neural network model for multi-scale extraction, thereby obtaining geomagnetic spatiotemporal dependent features.

[0062] It should be noted that in this application, the geomagnetic spatiotemporal dependency feature represents the temporal and spatial dependency information of the geomagnetic vector sequence. This feature captures the spatiotemporal dependency relationship of geomagnetic data, providing more reliable and accurate feature information for precise positioning and navigation within the target building. Specifically, firstly, the size of the sliding window (e.g., 5 seconds, 10 seconds, etc.) can be preset based on historical experience, and the size can also be adjusted according to the specific application scenario. Then, the sliding window is used to slide across the geomagnetic vector sequence, thereby dividing the geomagnetic vector sequence into multiple geomagnetic vector segments. Each geomagnetic vector segment contains all the geomagnetic vector data within a sliding window. Finally, an appropriate recurrent neural network can be selected. Network models (such as LSTM or GRU) can effectively capture the temporal dependencies in geomagnetic vector segments through recurrent neural network models. For each geomagnetic vector segment, it is input as a time series into the recurrent neural network model. The recurrent neural network model processes the geomagnetic vector at each time step and passes contextual information through hidden states. After processing by the recurrent neural network model, each geomagnetic vector segment will generate a spatiotemporal dependency feature vector. This spatiotemporal dependency feature vector contains the temporal and spatial dependency information of the geomagnetic vector segment. The time scale and spatial scale of each spatiotemporal dependency feature vector are different. Thus, the spatiotemporal dependency feature vectors of all geomagnetic vector segments are summarized to form the final geomagnetic spatiotemporal dependency feature.

[0063] It should be noted that scene constraints and trajectory constraints can eliminate unreasonable trajectories and geomagnetic data. Thus, through sliding window and multi-scale extraction, the variation characteristics of geomagnetic data at different time and spatial scales can be captured, making the extracted geomagnetic features more spatiotemporally dependent. This allows for the extraction of effective geomagnetic features even in open floor environments, enhancing the representation capability. The introduction of geomagnetic spatiotemporally dependent features enables the floor positioning model to capture longer-term and more stable feature information, improving the robustness of the floor positioning model to environmental changes.

[0064] In step 103, a location feature matching pool that fuses the geomagnetic spatiotemporal dependence features and the prior global map of the target building is determined. The location feature matching pool is matched with the set of environmental images of the target object in the target building to obtain the location matching entropy of the environment.

[0065] Optionally, in some embodiments, determining the location feature matching pool that fuses the geomagnetic features of each floor based on the geomagnetic spatiotemporal dependence features and the prior global map of the target building can be done in the following way:

[0066] Based on the aforementioned geomagnetic spatiotemporal dependence features, the prior global map of the target building is cropped to obtain a local prior 3D map that integrates the geomagnetic features of each floor.

[0067] The local prior 3D map is used for geomagnetic positioning to obtain geomagnetic positioning results;

[0068] The cost of geomagnetic positioning is determined based on the geomagnetic positioning results.

[0069] Based on the geomagnetic positioning results, the geomagnetic positioning cost, and the prior global map, a positioning feature matching pool is determined that integrates the geomagnetic features of each floor.

[0070] It should be noted that in this application, the positioning feature matching pool is a feature pool composed of features in the prior global map used for positioning matching; the geomagnetic positioning cost represents the difference cost between the geomagnetic positioning result and the benchmark result, which can be used to measure the accuracy of the geomagnetic positioning result.

[0071] In practice, firstly, based on the geomagnetic spatiotemporal dependence characteristics, a local segment is extracted from the prior global map of the target building. This prior global map contains detailed information about the building's structure, layout, and each floor. The portion of the prior global map that has a high degree of matching with the geomagnetic spatiotemporal dependence characteristics is extracted to form a local prior 3D map. Then, geomagnetic positioning is performed within this local prior 3D map. Real-time geomagnetic data collected by the smartphone's built-in geomagnetic sensor is matched with the geomagnetic data in the local prior 3D map. A positioning algorithm (such as particle filtering or Kalman filtering) is used to determine the geomagnetic location based on the real-time geomagnetic data and the geomagnetic data in the local prior 3D map. First, the geomagnetic features are matched to obtain real-time geomagnetic positioning results. Second, the geomagnetic positioning results can be matched with the set benchmark results. The matching error between the geomagnetic positioning results and the set benchmark results can be calculated by using cosine similarity, and this matching error is used as the geomagnetic positioning cost. Finally, the positioning feature matching pool for fusing floor geomagnetic features can be determined based on the geomagnetic positioning results, the geomagnetic positioning cost, and the prior global map. That is, the geomagnetic positioning results are used as the center, and the geomagnetic positioning cost is used as the radius to truncate the prior global map. The positioning features in the truncated local 3D map are used as the positioning feature matching pool for fusing floor geomagnetic features.

[0072] Preferably, in some embodiments, the location matching entropy of the target object in the target building is obtained by matching the location feature matching pool and the image set of the environment in which the target object is located. This can be achieved in the following way:

[0073] The environmental image set of the target object in the target building is preprocessed, and then the environmental feature vector is extracted from the preprocessed environmental image set.

[0074] Each positioning feature in the positioning feature matching pool is matched with the environmental feature vector to obtain the matching confidence between the environmental feature vector and each positioning feature.

[0075] The location matching entropy of the environment is determined by all matching confidence scores.

[0076] It should be noted that, in this application, the location matching entropy represents the uncertainty of the matching between the environment in which the target object is located and the location feature matching pool; the matching confidence represents the degree of similarity between the environmental feature vector and the location features in the location feature matching pool.

[0077] In specific implementation, firstly, a set of images of the target object's environment can be acquired in real time using a smartphone camera. The set of environmental images should cover the field of view around the target object, including features such as walls, ground, and ceiling. Then, the acquired environmental image set can be preprocessed, including denoising, enhancement, and distortion correction, to ensure the quality of the environmental images. Computer vision algorithms (such as SIFT, SURF, ORB, etc.) can then be used to extract key feature points and descriptors from the environmental image set to form an environmental feature vector. Secondly, the environmental feature vector extracted from the environmental image set can be matched with localization features in a localization feature matching pool. The feature matching method used in this application is based on deep learning (such as Siamese Network), which will not be elaborated here. This allows for the calculation of the cosine similarity between the environmental feature vector and the localization features, which is used as the matching confidence between the environmental feature vector and the localization features. The matching confidence between the environmental feature vector and each localization feature can be obtained through the above method. Finally, the probability distribution corresponding to each matching confidence can be calculated, and the localization matching entropy of the environment can be calculated using all probability distributions.

[0078] It should be noted that the location feature matching pool, which integrates the geomagnetic features of each floor, combines precise geomagnetic features with a prior global map of the building, providing accurate matching basis. The environmental image set can provide additional contextual information, helping to better match environmental features and reduce errors that may occur due to relying solely on geomagnetic data. This can improve the robustness of the location method in complex environments. Furthermore, calculating the location matching entropy can quantify the uncertainty of the matching results and provide early warning of possible errors.

[0079] In step 104, motion data of the target object in the target building is obtained, and the positioning matching trajectory of the target object is determined based on the motion data and the location fingerprint database of the target building.

[0080] In practice, motion data of target objects in the target building can be obtained from floor sensor data of the target building.

[0081] Preferably, in some embodiments, reference is made to Figure 3As shown, this figure is an exemplary flowchart of determining the location matching trajectory in some embodiments of this application. In this embodiment, determining the location matching trajectory of the target object based on the motion data and the location fingerprint database of the target building can be achieved by the following steps:

[0082] First, in step 1041, the motion data is fitted to obtain the motion trajectory of the target object;

[0083] Then, in step 1042, the received signal strength sequence is extracted from the location fingerprint database of the target building;

[0084] Finally, in step 1043, the positioning matching trajectory of the target object is determined by the motion trajectory of the target object and the received signal strength sequence.

[0085] In practice, firstly, the motion data obtained from the floor sensor data of the target building is denoised, filtered, and corrected to ensure the quality of the motion data. Then, an appropriate motion model (such as a gait model or velocity model) is used to model the motion data, fitting it to the motion trajectory of the target object. This trajectory represents the target object's movement path within the target building. Next, a location fingerprint database contains signal strength data (such as Wi-Fi, Bluetooth, and geomagnetic signals) for various locations within the target building. This signal strength data provides a unique fingerprint for each location. Based on the target building's location fingerprint database, the received signal strength at the target location is extracted. Each location's signal strength is recorded at different times or locations. Each location point in the target object's motion trajectory is matched with the signal strength data in the database, thus obtaining a sequence of received signal strengths. Finally, based on the direction of each signal strength in the received signal strength sequence, a trajectory direction is assigned to each trajectory in the target object's motion trajectory, resulting in the target object's location matching trajectory.

[0086] It should be noted that combining motion data with signal strength data in the location fingerprint database can effectively fuse information from different sources and improve positioning accuracy. Motion data provides dynamic motion trajectory information, while signal strength data provides static features in the environment. The combination of the two can more accurately describe the location of the target object. Furthermore, by combining motion data and the location fingerprint database to determine the positioning matching trajectory, real-time trajectory tracking of the target object can be achieved.

[0087] In step 105, the floor position of the target object is determined by the positioning matching entropy and the positioning matching trajectory, and then the position is marked in the prior global map of the target building.

[0088] Preferably, in some embodiments, determining the floor location of the target object using the positioning matching entropy and the positioning matching trajectory can be achieved in the following manner:

[0089] The location matching entropy is used as a weight parameter;

[0090] Use the location matching trajectory as input data;

[0091] The weight parameters and the input data are input into a pre-trained floor positioning model for positioning, thereby obtaining the floor position of the target object.

[0092] In practice, firstly, the location matching entropy is input as a weight parameter into the floor location model. A lower location matching entropy corresponds to a higher weight, indicating that the location result is more reliable; a higher location matching entropy corresponds to a lower weight, indicating that the result has greater uncertainty. Then, the trajectory data is formatted into the input format required by the floor location model, including location coordinates, timestamps, motion status, etc. Finally, a pre-trained floor location model is selected. The floor location model selected in this application is a machine learning model. In actual implementation, other models can also be used as the floor location model. The weight parameter (location matching entropy) and the input data (location matching trajectory) are input into the model together. The floor location model makes predictions based on the input weight parameter and trajectory data to determine the floor location of the target object.

[0093] Optionally, in some embodiments, location marking in the prior global map of the target building means displaying the floor position of the target object in the prior global map of the target building in real time; in specific implementation, the location mark of the target object is updated in real time on the prior global map of the target building, that is, according to the determined floor position, a mark (such as an icon or coordinate point) is added or updated on the corresponding floor map.

[0094] Therefore, this application demonstrates an improvement in the representation and identification capabilities of geomagnetic features in open or narrow environments. Specifically, scene and trajectory constraints eliminate unreasonable trajectories and geomagnetic data. Through sliding window and multi-scale extraction, the variation characteristics of geomagnetic data at different temporal and spatial scales can be captured, making the extracted geomagnetic features more spatiotemporally dependent. The introduction of spatiotemporally dependent geomagnetic features allows the floor positioning model to capture longer-term and more stable feature information, improving its robustness to environmental changes. Furthermore, the positioning feature matching pool, which integrates floor geomagnetic features, combines precise geomagnetic features with a prior global map of the building, providing accurate matching criteria. The environmental image set provides additional contextual information, helping to better match environmental features and reducing errors that may occur due to relying solely on geomagnetic data. Secondly, combining motion data with signal strength data from a location fingerprint database effectively fuses information from different sources. Furthermore, by combining motion data and the location fingerprint database to determine the positioning matching trajectory, real-time trajectory tracking of the target object can be achieved. Finally, the floor location of the target object is determined and labeled using the positioning matching entropy and the positioning matching trajectory. In summary, the technical solution adopted in this application can improve the characterization and identification ability of geomagnetic features in open or narrow environments, thereby increasing the accuracy and robustness of floor location identification.

[0095] The foregoing has detailed examples of floor location positioning and marking methods provided in the embodiments of this application. It is understood that the corresponding apparatus, in order to achieve the above functions, includes hardware structures and / or software modules corresponding to the execution of each function. Those skilled in the art should readily recognize that, in conjunction with the units and algorithm steps of the various examples described in the embodiments disclosed herein, this application can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0096] In some embodiments, this application also provides a computer device, the computer device including a memory and a processor, the memory for storing a computer program, and the processor for calling and running the computer program from the memory, causing the computer device to perform the above-described method for locating and marking floor positions.

[0097] In some embodiments, reference Figure 4The dashed lines in the figure indicate that the unit or module is optional. This figure is a structural schematic diagram of a computer device for a floor location positioning and marking method according to an embodiment of this application. The floor location positioning and marking method described in the above embodiments can be achieved through... Figure 4 The computer device 400 shown is used to implement this, and the computer device 400 includes at least one processor 401, a memory 402 and at least one communication unit 405. The computer device 400 may be a terminal device, a server or a chip.

[0098] Processor 401 can be a general-purpose processor or a special-purpose processor. For example, processor 401 can be a central processing unit (CPU). The CPU can be used to control computer device 400, execute software programs, and process data of software programs. Computer device 400 may also include a communication unit 405 for signal input (receiving) and output (transmitting).

[0099] For example, computer device 400 may be a chip, communication unit 405 may be the input and / or output circuit of the chip, or communication unit 405 may be the communication interface of the chip, and the chip may be a component of terminal device, network device or other device.

[0100] For example, computer device 400 may be a terminal device or a server, and communication unit 405 may be a transceiver of the terminal device or the server, or communication unit 405 may be a transceiver circuit of the terminal device or the server.

[0101] The computer device 400 may include one or more memories 402 storing a program 404. The program 404 can be executed by a processor 401 to generate instructions 403, causing the processor 401 to execute the method described in the above method embodiments according to the instructions 403. Optionally, the memory 402 may also store data (such as a target audit model). Optionally, the processor 401 may also read data stored in the memory 402, which may be stored at the same storage address as the program 404, or it may be stored at a different storage address than the program 404.

[0102] The processor 401 and memory 402 can be configured separately or integrated together, for example, integrated on the system-on-chip (SOC) of the terminal device.

[0103] It should be understood that each step of the above method embodiment can be completed by hardware logic circuits or software instructions in processor 401. Processor 401 can be a central processing unit, digital signal processor (DSP), application specific integrated circuit (ASIC), field programmable gate array (FPGA), or other programmable logic device, such as discrete gate, transistor logic device, or discrete hardware component.

[0104] Those skilled in the art will understand that embodiments of this application can be provided as methods, systems, or computer program products. Therefore, this application can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, this application can take the form of a computer program product embodied on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0105] For example, in some embodiments, this application also provides a computer-readable storage medium storing instructions or code that, when executed on a computer, cause the computer to implement the above-described method for locating and marking floor positions.

[0106] Although preferred embodiments of this application have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including the preferred embodiments as well as all changes and modifications falling within the scope of this application.

[0107] Obviously, those skilled in the art can make various modifications and variations to this application without departing from the spirit and scope of this application. Therefore, if such modifications and variations fall within the scope of the claims of this application and their equivalents, this application also intends to include such modifications and variations.

Claims

1. A method for locating and marking floor positions, characterized in that, Includes the following steps: Collect floor sensor data for the target building; Floor geomagnetic data is extracted from the floor sensor data. The floor geomagnetic data is then subjected to trajectory constraints based on the scene constraints of the target building to obtain a geomagnetic vector sequence. The geomagnetic vector sequence is then extracted at multiple scales to obtain geomagnetic spatiotemporal dependence features. Based on the geomagnetic spatiotemporal dependence features and the prior global map of the target building, a location feature matching pool that integrates the geomagnetic features of each floor is determined. The location feature matching pool is then matched with the image set of the environment in which the target object is located in the target building to obtain the location matching entropy of the environment. Acquire motion data of a target object within a target building, and determine the positioning and matching trajectory of the target object based on the motion data and the target building's location fingerprint database; The floor location of the target object is determined by the location matching entropy and the location matching trajectory, and then the location is marked in the prior global map of the target building. Specifically, the location feature matching pool, which integrates the geomagnetic spatiotemporal dependence features and the prior global map of the target building, is determined by: Based on the geomagnetic spatiotemporal dependence features, the prior global map of the target building is cropped to obtain a local prior 3D map that integrates the geomagnetic features of each floor. The geomagnetic spatiotemporal dependence features represent the time and space dependence information of the geomagnetic vector sequence. The local prior 3D map is used for geomagnetic positioning to obtain geomagnetic positioning results; The geomagnetic positioning cost is determined by the geomagnetic positioning results, wherein the geomagnetic positioning results are matched with the set benchmark results, that is, the matching error between the geomagnetic positioning results and the set benchmark results is calculated by using cosine similarity, and the matching error is used as the geomagnetic positioning cost. Based on the geomagnetic positioning results, the geomagnetic positioning cost, and the prior global map, a positioning feature matching pool that integrates the geomagnetic features of each floor is determined. Specifically, the matching process based on the location feature matching pool and the image set of the environment in which the target object is located in the target building, to obtain the location matching entropy of the environment, includes: The environmental image set of the target object in the target building is preprocessed, and then the environmental feature vector is extracted from the preprocessed environmental image set. Each positioning feature in the positioning feature matching pool is matched with the environmental feature vector to obtain the matching confidence between the environmental feature vector and each positioning feature. The location matching entropy of the environment is determined by all the matching confidence scores. The location matching entropy represents the uncertainty of the matching between the environment of the target object and the location feature matching pool.

2. The method as described in claim 1, characterized in that, Extracting floor geomagnetic data from the floor sensor data specifically includes: Extract all floor geomagnetic records from the floor sensor data; Drift correction was performed on all floor geomagnetic records to obtain corrected floor geomagnetic records. The floor geomagnetic data is determined based on the corrected geomagnetic records of all floors.

3. The method as described in claim 1, characterized in that, The system uses the built-in magnetic sensor and gyroscope of a smartphone to collect floor sensor data from the target building.

4. The method as described in claim 1, characterized in that, Based on the scene constraints of the target building, trajectory constraints are applied to the floor geomagnetic data to obtain the geomagnetic vector sequence, which specifically includes: The floor geomagnetic data is fitted into a floor geomagnetic trajectory; The floor geomagnetic trajectory is constrained according to the scene constraints of the target building to obtain the floor geomagnetic constrained trajectory; Extract the geomagnetic vector sequence from the geomagnetic constraint trajectory of the floor.

5. The method as described in claim 1, characterized in that, Multi-scale extraction of the geomagnetic vector sequence yields geomagnetic spatiotemporal dependence features, specifically including: The geomagnetic vector sequence is divided into multiple geomagnetic vector segments according to a preset sliding window. All geomagnetic vector segments are input into a recurrent neural network model for multi-scale extraction, thereby obtaining geomagnetic spatiotemporal dependent features.

6. The method as described in claim 1, characterized in that, Determining the location matching trajectory of the target object based on the motion data and the target building's location fingerprint database specifically includes: The motion data is fitted to obtain the motion trajectory of the target object; Extract the received signal strength sequence from the location fingerprint database of the target building; The positioning and matching trajectory of the target object is determined by the motion trajectory of the target object and the received signal strength sequence.

7. The method as described in claim 1, characterized in that, Location marking in the prior global map of the target building means displaying the floor position of the target object in the prior global map of the target building in real time.

8. A computer device, characterized in that, The computer device includes a memory and a processor. The memory is used to store computer programs, and the processor is used to call and run the computer programs from the memory, causing the computer device to perform the method for locating and marking floor positions according to any one of claims 1 to 7.

9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores instructions or code that, when executed on a computer, cause the computer to perform the method for locating and marking floor positions as described in any one of claims 1 to 7.

Images