Indoor and outdoor seamless navigation method and system based on dynamic distribution of multi-source positioning weights

Abstract

This application provides a method and system for seamless indoor and outdoor navigation based on dynamic allocation of multi-source positioning weights. The seamless indoor and outdoor navigation system includes: a multi-source positioning data acquisition module for real-time acquisition and preprocessing of raw data from different positioning sources; a building structure big data engine for constructing a three-dimensional spatial constraint model; a positioning source reliability assessment module for quantifying the credibility of each positioning source in a specific scenario; a dynamic weight allocation decision module for real-time adjustment of positioning source weights based on a multi-factor decision model; a multi-source data fusion filtering module for fusing heterogeneous positioning data; and an indoor and outdoor scene switching control module for triggering and smoothing seamless switching between indoor and outdoor scenes. The above technical solution achieves efficient collaboration among multiple positioning sources, improving navigation accuracy and adaptability.

Description

Technical Field

[0001] This application relates to the field of multi-source fusion positioning technology, and in particular to an indoor and outdoor seamless navigation method and system based on dynamic allocation of multi-source positioning weights. Background Technology

[0002] With the booming development of smart cities, industrial IoT, emergency rescue, and many other fields, the demand for seamless, high-precision indoor and outdoor navigation is increasing daily. However, current single positioning technologies all have significant limitations and are unable to meet the navigation requirements in complex environments on their own.

[0003] In terms of satellite positioning, GPS and BeiDou systems can provide high-precision positioning in open outdoor areas and are widely used in scenarios such as vehicle navigation and outdoor exploration. However, when entering indoor spaces, underground parking lots, or urban areas with tall buildings, satellite signals are easily blocked or reflected, leading to a significant decrease in positioning accuracy or even failure.

[0004] Among short-range wireless positioning technologies, UWB offers high positioning accuracy and strong anti-interference capabilities, making it outstanding in industrial equipment positioning. However, it has high deployment costs and limited coverage. Bluetooth positioning is low-cost and widely used, often employed in indoor shopping mall navigation, but its accuracy is only at the meter level and its stability is poor.

[0005] While existing multi-source positioning fusion technologies attempt to combine the advantages of multiple positioning sources, they mostly employ fixed weight allocation methods. In different environments, the performance and reliability of each positioning source vary significantly, and fixed weights cannot be dynamically adjusted according to actual conditions, resulting in insufficient fusion positioning accuracy and reliability in specific scenarios. For example, when indoor UWB signals are interfered with, fixed weights still assign a large proportion, which will seriously affect the final positioning result. Summary of the Invention

[0006] This application provides an indoor and outdoor seamless navigation method and system based on dynamic allocation of multi-source positioning weights, which can realize efficient collaboration of multiple positioning sources and improve navigation accuracy and adaptability.

[0007] Firstly, a seamless indoor and outdoor navigation system based on dynamic allocation of multi-source positioning weights is provided, including:

[0008] The multi-source positioning data acquisition module is used to acquire raw data from different positioning sources in real time and perform preprocessing, and to align the timestamps of multi-sensor data through a hardware-level synchronization clock.

[0009] A building structure big data engine for constructing three-dimensional spatial constraint models;

[0010] The location source reliability assessment module is used to quantify the reliability of each location source in a specific scenario;

[0011] The dynamic weight allocation decision module is used to adjust the weight of the location source in real time based on a multi-factor decision model;

[0012] A multi-source data fusion and filtering module is used to fuse heterogeneous positioning data;

[0013] The indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes.

[0014] In the above technical solution, a multi-source positioning data acquisition module is set up to acquire raw data from different positioning sources in real time and perform preprocessing, and the timestamps of multi-sensor data are aligned through a hardware-level synchronization clock; a building structure big data engine is used to construct a three-dimensional spatial constraint model; a positioning source reliability assessment module is used to quantify the credibility of each positioning source in a specific scenario; a dynamic weight allocation decision module is used to adjust the weight of positioning sources in real time based on a multi-factor decision model; a multi-source data fusion filtering module is used to fuse heterogeneous positioning data; and an indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes. This achieves efficient collaboration of multiple positioning sources, improving navigation accuracy and adaptability.

[0015] In one specific implementation scheme, it also includes:

[0016] The positioning results output and visualization module is used to transform positioning data into understandable navigation information.

[0017] In one specific implementation scheme, the multi-source positioning data acquisition module includes:

[0018] The GPS / BeiDou receiver unit is used for dual-mode satellite signal reception.

[0019] UWB / Bluetooth beacon unit for integrating ultra-wideband pulse transmission and Bluetooth Low Energy 5.1 protocol;

[0020] A geomagnetic sensor array, including a triaxial magnetometer and a temperature compensation circuit, is used to eliminate environmental magnetic interference;

[0021] The visual positioning unit includes a binocular camera and a deep learning acceleration chip, used to support SLAM and feature point matching;

[0022] An inertial measurement unit, including a six-axis gyroscope and an accelerometer, is used for motion state prediction.

[0023] In one specific implementation scheme, the building structure big data engine includes:

[0024] BIM model parsing unit, used to parse geometric parameters in building information model;

[0025] A library of electromagnetic properties of materials, used to store attenuation coefficients for UWB / Bluetooth signals;

[0026] A dynamic obstacle detection unit is used to identify temporary obstacles;

[0027] The geomagnetic reference map generation unit is used to construct an indoor geomagnetic field intensity distribution map;

[0028] The path topology optimization unit is used to generate the optimal barrier-free path.

[0029] In one specific implementation scheme, the location source reliability assessment module includes:

[0030] Signal quality analysis unit, used to calculate the signal-to-noise ratio and multipath effect index of UWB / Bluetooth;

[0031] The geometric accuracy factor calculation unit is used to evaluate the impact of the spatial distribution of satellites / beacons on positioning errors;

[0032] The visual feature matching degree unit is used to count the number of feature point matches and the reprojection error.

[0033] The geomagnetic consistency verification unit is used to compare the real-time measurement value with the Mahalanobis distance of the reference map;

[0034] The motion state association unit is used to verify the physical validity of the positioning results using IMU data.

[0035] In one specific implementation scheme, the dynamic weight allocation decision module includes:

[0036] Fuzzy logic reasoning unit, used to construct three-dimensional membership functions;

[0037] The reinforcement learning optimization unit is used to train the weight allocation strategy using the DDPG algorithm.

[0038] The conflict resolution unit is used to enable DS evidence theory fusion decision-making when multiple location source results are contradictory.

[0039] The historical trajectory backtracking unit is used to correct short-term abnormal weight fluctuations using Kalman filtering;

[0040] User preference learning units are used to analyze user behavior patterns using hidden Markov models.

[0041] In one specific implementation scheme, the multi-source data fusion filtering module includes:

[0042] Loosely coupled filtering unit is used for weighted fusion after independent calculation of each positioning source;

[0043] Tightly coupled filtering unit, used to directly input the raw observations into the extended Kalman filter;

[0044] An adaptive covariance adjustment unit is used to dynamically adjust the EKF process noise matrix based on the reliability of the positioning source;

[0045] The outlier removal unit is used to detect and remove outliers using an improved RANSAC algorithm.

[0046] The smoothing filter unit is used to suppress high-frequency noise using a Savitzky-Golay filter.

[0047] In one specific implementation scheme, the indoor / outdoor scene switching control module includes:

[0048] A switching threshold learning unit is used to determine the switching boundary conditions based on a support vector machine classifier;

[0049] A pre-switching detection unit is used to predict the type of scene to be entered based on the WiFi signal strength gradient;

[0050] A gradient blending unit is used to smooth the positioning results using an exponentially weighted average in the switching transition zone;

[0051] The map matching correction unit is used to correct trajectory jumps when crossing scenes using a hidden map matching algorithm;

[0052] The user feedback interface unit is used to switch states via voice / vibration prompts and receive user confirmation.

[0053] In one specific implementation scheme, the positioning result output and visualization module includes:

[0054] Coordinate transformation unit, used to support real-time transformation between WGS84, CGCS2000 and local coordinate systems;

[0055] The path planning unit is used to combine the A* algorithm with dynamic obstacle avoidance to generate real-time navigation instructions.

[0056] AR overlay display unit, used to mark navigation arrows and distances in the real scene using the mobile phone camera;

[0057] A voice interaction unit is used to support bilingual (Chinese and English) broadcasting and dialect recognition;

[0058] The privacy protection unit is used to perform end-to-end encryption of location data using homomorphic encryption technology.

[0059] Secondly, a seamless indoor and outdoor navigation method based on dynamic allocation of multi-source positioning weights is provided, including the following steps:

[0060] The multi-source positioning data acquisition module is used to acquire raw data from different positioning sources in real time and perform preprocessing. The timestamps of the multi-sensor data are aligned by a hardware-level synchronization clock.

[0061] Utilize a building structure big data engine to construct a three-dimensional spatial constraint model;

[0062] The reliability of each positioning source in a specific scenario is quantified using the positioning source reliability assessment module;

[0063] The dynamic weight allocation decision module adjusts the weights of the location sources in real time based on a multi-factor decision model.

[0064] Heterogeneous positioning data are fused using a multi-source data fusion and filtering module;

[0065] The indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes.

[0066] In the above technical solution, a multi-source positioning data acquisition module is set up to acquire raw data from different positioning sources in real time and perform preprocessing, and the timestamps of multi-sensor data are aligned through a hardware-level synchronization clock; a building structure big data engine is used to construct a three-dimensional spatial constraint model; a positioning source reliability assessment module is used to quantify the credibility of each positioning source in a specific scenario; a dynamic weight allocation decision module is used to adjust the weight of positioning sources in real time based on a multi-factor decision model; a multi-source data fusion filtering module is used to fuse heterogeneous positioning data; and an indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes. This achieves efficient collaboration of multiple positioning sources, improving navigation accuracy and adaptability. Attached Figure Description

[0067] Figure 1 A structural block diagram of an indoor-outdoor seamless navigation system based on dynamic allocation of multi-source positioning weights, provided in an embodiment of this application;

[0068] Figure 2 This is a flowchart illustrating the indoor and outdoor seamless navigation method based on dynamic allocation of multi-source positioning weights provided in an embodiment of this application. Detailed Implementation

[0069] The present application will now be described in further detail with reference to the accompanying drawings and embodiments. Through these descriptions, the features and advantages of the present application will become clearer and more apparent.

[0070] The term “exemplary” as used herein means “serving as an example, embodiment, or illustration.” Any embodiment illustrated herein as “exemplary” is not necessarily to be construed as superior to or better than other embodiments. Although various aspects of embodiments are shown in the accompanying drawings, the drawings are not necessarily drawn to scale unless specifically indicated otherwise.

[0071] Furthermore, the technical features involved in the different embodiments of this application described below can be combined with each other as long as they do not conflict with each other.

[0072] To facilitate understanding of the indoor / outdoor seamless navigation method and system based on dynamic allocation of multi-source positioning weights provided in this application embodiment, its application scenario is first explained. The indoor / outdoor seamless navigation method and system based on dynamic allocation of multi-source positioning weights provided in this application embodiment are used to achieve efficient collaboration among multiple positioning sources, improving navigation accuracy and adaptability. With the booming development of smart cities, industrial IoT, emergency rescue, and many other fields, the demand for seamless, high-precision indoor / outdoor navigation is increasing daily. However, current single positioning technologies all have significant limitations and cannot independently meet the navigation requirements in complex environments. Regarding satellite positioning, GPS and BeiDou systems can provide high-precision positioning in open outdoor areas and are widely used in vehicle navigation, outdoor exploration, and other scenarios. However, when entering indoor spaces, underground parking lots, or urban areas with tall buildings, satellite signals are easily blocked or reflected, leading to a significant decrease in positioning accuracy or even failure. Among short-range wireless positioning technologies, UWB positioning has high accuracy and strong anti-interference capabilities, performing exceptionally well in industrial equipment positioning, but its deployment cost is high and its coverage is limited; Bluetooth positioning has low cost and widespread availability, often used for indoor shopping mall navigation, but its accuracy is only at the meter level and its stability is poor. While existing multi-source positioning fusion technologies attempt to combine the advantages of multiple positioning sources, they mostly employ fixed weight allocation methods. In different environments, the performance and reliability of each positioning source vary significantly, and fixed weights cannot be dynamically adjusted according to actual conditions, resulting in insufficient fusion positioning accuracy and reliability in specific scenarios. For example, when indoor UWB signals are interfered with, fixed weights still assign a large proportion, which severely affects the final positioning result. Therefore, this application provides an indoor-outdoor seamless navigation method and system based on dynamic allocation of multi-source positioning weights to achieve efficient collaboration among multiple positioning sources and improve navigation accuracy and adaptability. The following detailed description, with reference to specific accompanying drawings, illustrates the embodiments.

[0073] refer to Figure 1 and Figure 2 , Figure 1 A structural block diagram of an indoor-outdoor seamless navigation system based on dynamic allocation of multi-source positioning weights, provided in an embodiment of this application; Figure 2 This is a flowchart illustrating the indoor and outdoor seamless navigation method based on dynamic allocation of multi-source positioning weights provided in an embodiment of this application.

[0074] exist Figure 1 In this application, an embodiment provides a seamless indoor and outdoor navigation system based on dynamic allocation of multi-source positioning weights, comprising:

[0075] The multi-source positioning data acquisition module is used to acquire raw data from different positioning sources in real time and perform preprocessing, and to align the timestamps of multi-sensor data through a hardware-level synchronization clock.

[0076] A building structure big data engine for constructing three-dimensional spatial constraint models;

[0077] The location source reliability assessment module is used to quantify the reliability of each location source in a specific scenario;

[0078] The dynamic weight allocation decision module is used to adjust the weight of the location source in real time based on a multi-factor decision model;

[0079] A multi-source data fusion and filtering module is used to fuse heterogeneous positioning data;

[0080] The indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes.

[0081] In the above technical solution, a multi-source positioning data acquisition module is set up to acquire raw data from different positioning sources in real time and perform preprocessing, and the timestamps of multi-sensor data are aligned through a hardware-level synchronization clock; a building structure big data engine is used to construct a three-dimensional spatial constraint model; a positioning source reliability assessment module is used to quantify the credibility of each positioning source in a specific scenario; a dynamic weight allocation decision module is used to adjust the weight of positioning sources in real time based on a multi-factor decision model; a multi-source data fusion filtering module is used to fuse heterogeneous positioning data; and an indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes. This achieves efficient collaboration of multiple positioning sources, improving navigation accuracy and adaptability.

[0082] Specifically, the beneficial effects include:

[0083] Improving navigation accuracy: The multi-source positioning data acquisition module can acquire raw data in real time from various positioning sources such as GPS / BeiDou, UWB / Bluetooth beacons, geomagnetic sensors, visual positioning units, and inertial measurement units. It also ensures data synchronization and accuracy by precisely aligning the timestamps of multi-sensor data through a hardware-level synchronization clock. The 3D spatial constraint model constructed by the building structure big data engine provides prior knowledge for positioning source reliability assessment, making the assessment results more accurate. The positioning source reliability assessment module can quantify the credibility of each positioning source in a specific scenario. The dynamic weight allocation decision module adjusts the positioning source weights in real time based on this, allowing more reliable positioning sources to play a greater role in the positioning process. The multi-source data fusion and filtering module performs high-precision fusion of heterogeneous positioning data, ultimately achieving a continuous indoor and outdoor positioning error of less than 0.5 meters, a 60% improvement over traditional methods, providing users with extremely accurate navigation information.

[0084] Enhanced Scene Adaptability: The system boasts strong scene adaptability. Its building structure big data engine analyzes building information models, stores material electromagnetic properties, detects dynamic obstacles, generates geomagnetic reference maps, and optimizes path topology, enabling it to fully adapt to various complex indoor and outdoor environments. Whether in city streets lined with high-rise buildings or intricately laid-out indoor shopping malls and warehouses, the system accurately perceives surrounding environmental characteristics, providing reliable data for positioning and navigation. The dynamic weight allocation decision module flexibly adjusts the weights of positioning sources according to the characteristics of different scenarios, ensuring optimal positioning performance in various environments. For example, in indoor environments, it prioritizes data from UWB / Bluetooth beacons, geomagnetic sensors, and visual positioning units; in open outdoor areas, it fully leverages GPS / BeiDou.

[0085] Seamless scene switching is achieved: The indoor / outdoor scene switching control module triggers and smoothly transitions between indoor and outdoor scenes. The switching threshold learning unit accurately determines the switching boundary conditions based on a support vector machine classifier, and the pre-switching detection unit predicts the upcoming scene type in advance by using the WiFi signal strength gradient. The gradual fusion unit smooths the positioning results in the switching transition zone using an exponential weighted average method, the map matching correction unit corrects trajectory jumps when crossing scenes using a hidden map matching algorithm, and the user feedback interface unit provides voice / vibration prompts for the switching status and receives user confirmation. Simultaneously, the innovative "virtual beacon" technology deploys temporary UWB beacons at the indoor / outdoor boundary to simulate GPS signals, making the switching process virtually imperceptible to the user. The scene switching time is less than 0.3 seconds, truly achieving seamless navigation.

[0086] Enhancing System Robustness: Even in complex environments such as non-line-of-sight (NLOS) conditions, the system maintains over 95% availability. The location source reliability assessment module comprehensively evaluates the trustworthiness of location sources through multiple methods. The dynamic weight allocation decision module adjusts weights in real time based on the assessment results. The multi-source data fusion and filtering module effectively fuses heterogeneous data and removes outliers. This allows the system to provide stable positioning services even when some location sources are blocked or the data is inaccurate, relying on other reliable location sources and fusion algorithms, significantly improving system robustness.

[0087] Reduced system power consumption: The dynamic weight allocation decision module reduces sensor power consumption by 28% by rationally allocating the weights of positioning sources. The system dynamically adjusts the usage frequency and weight of each positioning source according to actual needs, avoiding unnecessary sensor operation. While ensuring positioning accuracy and navigation performance, it effectively extends the device's battery life, making it particularly suitable for energy-sensitive mobile terminal devices.

[0088] In one specific implementation scheme, it also includes:

[0089] The positioning results output and visualization module is used to transform positioning data into understandable navigation information.

[0090] The above technical solution enhances information comprehension and interactive experience: the positioning result output and visualization module transforms complex positioning data into intuitive and easy-to-understand navigation information, such as real-time location markers on the map, route guidance arrows, and distance prompts. Users can easily obtain their location and navigation direction without needing to understand professional positioning data, greatly reducing the barrier to entry and improving the interactive experience. Whether for ordinary users or special groups such as the elderly, everyone can quickly get started using the navigation system. It also enhances decision-making efficiency and accuracy: clear visualization information allows users to make quick decisions. In complex indoor and outdoor environments, users can quickly plan the optimal route based on intuitive navigation displays, avoiding decision-making errors caused by difficulties in understanding information, improving travel efficiency, and reducing time spent wandering in unfamiliar environments. Furthermore, it adapts to diverse application scenarios: this module can flexibly adjust the visualization form and content according to different scenarios, such as indoor shopping malls and outdoor streets, to meet diverse navigation needs, enabling the system to provide users with efficient and accurate navigation services in various scenarios.

[0091] In one specific implementation scheme, the multi-source positioning data acquisition module includes:

[0092] The GPS / BeiDou receiver unit is used for dual-mode satellite signal reception.

[0093] UWB / Bluetooth beacon unit for integrating ultra-wideband pulse transmission and Bluetooth Low Energy 5.1 protocol;

[0094] A geomagnetic sensor array, including a triaxial magnetometer and a temperature compensation circuit, is used to eliminate environmental magnetic interference;

[0095] The visual positioning unit includes a binocular camera and a deep learning acceleration chip, used to support SLAM and feature point matching;

[0096] An inertial measurement unit, including a six-axis gyroscope and an accelerometer, is used for motion state prediction.

[0097] In the above technical solution, the GPS / BeiDou receiver unit provides dual-mode reception, improving the stability and accuracy of outdoor positioning; the UWB / Bluetooth beacon unit integrates ultra-wideband and low-power Bluetooth, balancing high-precision positioning with low power consumption; the geomagnetic sensor array eliminates environmental magnetic interference, enhancing indoor positioning reliability; the visual positioning unit supports SLAM and feature point matching, adapting to complex environments; and the inertial measurement unit predicts motion states, compensating for the shortcomings of other positioning methods. Multi-source data complementarity enables high-precision positioning in all indoor and outdoor scenarios, providing users with more accurate and reliable navigation services.

[0098] In one specific implementation scheme, the building structure big data engine includes:

[0099] BIM model parsing unit, used to parse geometric parameters in building information model;

[0100] A library of electromagnetic properties of materials, used to store attenuation coefficients for UWB / Bluetooth signals;

[0101] A dynamic obstacle detection unit is used to identify temporary obstacles;

[0102] The geomagnetic reference map generation unit is used to construct an indoor geomagnetic field intensity distribution map;

[0103] The path topology optimization unit is used to generate the optimal barrier-free path.

[0104] In the above technical solution, the BIM model analysis unit acquires building geometric parameters, providing a spatial basis for positioning; the material electromagnetic property library stores signal attenuation coefficients, facilitating accurate UWB / Bluetooth positioning; the dynamic obstacle detection unit identifies temporary obstacles, enhancing environmental adaptability; the geomagnetic reference map generation unit constructs a geomagnetic field distribution map, improving geomagnetic positioning accuracy; and the path topology optimization unit generates an optimal barrier-free path, optimizing navigation routes. The collaboration of these units allows the system to better understand the built environment, achieving efficient and accurate navigation and enhancing the user's indoor and outdoor travel experience.

[0105] In one specific implementation scheme, the location source reliability assessment module includes:

[0106] Signal quality analysis unit, used to calculate the signal-to-noise ratio and multipath effect index of UWB / Bluetooth;

[0107] The geometric accuracy factor calculation unit is used to evaluate the impact of the spatial distribution of satellites / beacons on positioning errors;

[0108] The visual feature matching degree unit is used to count the number of feature point matches and the reprojection error.

[0109] The geomagnetic consistency verification unit is used to compare the real-time measurement value with the Mahalanobis distance of the reference map;

[0110] The motion state association unit is used to verify the physical validity of the positioning results using IMU data.

[0111] In the above technical solution, the signal quality analysis unit accurately judges the quality of UWB / Bluetooth signals by calculating the signal-to-noise ratio and multipath effect index; the geometric accuracy factor calculation unit assesses the impact of satellite / beacon spatial distribution, providing a basis for predicting positioning errors; the visual feature matching degree unit statistically analyzes feature points to measure the reliability of visual positioning; the geomagnetic consistency verification unit compares Mahalanobis distances to ensure accurate geomagnetic positioning; and the motion state association unit verifies the rationality of the results using IMU data. These units work together to comprehensively quantify the credibility of the positioning source, providing reliable support for dynamic weight allocation and improving navigation accuracy and stability.

[0112] In one specific implementation scheme, the dynamic weight allocation decision module includes:

[0113] Fuzzy logic reasoning unit, used to construct three-dimensional membership functions;

[0114] The reinforcement learning optimization unit is used to train the weight allocation strategy using the DDPG algorithm.

[0115] The conflict resolution unit is used to enable DS evidence theory fusion decision-making when multiple location source results are contradictory.

[0116] The historical trajectory backtracking unit is used to correct short-term abnormal weight fluctuations using Kalman filtering;

[0117] User preference learning units are used to analyze user behavior patterns using hidden Markov models.

[0118] In the above technical solution, the fuzzy logic reasoning unit constructs a three-dimensional membership function, which can flexibly handle the fuzziness of positioning source evaluation; the reinforcement learning optimization unit adopts the DDPG algorithm training strategy, making weight allocation more intelligent and efficient; the conflict resolution unit uses DS evidence theory to effectively resolve contradictions in results from multiple positioning sources; the historical trajectory backtracking unit uses Kalman filtering to correct short-term abnormal weight fluctuations and enhance stability; and the user preference learning unit analyzes behavior through a hidden Markov model to meet personalized needs. The collaboration of these units enables real-time, accurate, and personalized allocation of positioning source weights, improving navigation performance.

[0119] In one specific implementation scheme, the multi-source data fusion filtering module includes:

[0120] Loosely coupled filtering unit is used for weighted fusion after independent calculation of each positioning source;

[0121] Tightly coupled filtering unit, used to directly input the raw observations into the extended Kalman filter;

[0122] An adaptive covariance adjustment unit is used to dynamically adjust the EKF process noise matrix based on the reliability of the positioning source;

[0123] The outlier removal unit is used to detect and remove outliers using an improved RANSAC algorithm.

[0124] The smoothing filter unit is used to suppress high-frequency noise using a Savitzky-Golay filter.

[0125] In the above technical solution, the loosely coupled filtering unit performs independent calculations followed by weighted fusion, taking into account the characteristics of each positioning source; the tightly coupled filtering unit directly inputs the original observations into the extended Kalman filter, improving data utilization efficiency. The adaptive covariance adjustment unit dynamically adjusts the noise matrix according to the reliability of the positioning source, enhancing the filter's adaptability. The outlier removal unit uses an improved RANSAC algorithm to accurately detect and remove outliers, ensuring data quality. The smoothing filtering unit uses a Savitzky-Golay filter to suppress high-frequency noise, making the output more stable. These units work together to achieve efficient fusion of multi-source data, significantly improving the accuracy and reliability of navigation and positioning.

[0126] In one specific implementation scheme, the indoor / outdoor scene switching control module includes:

[0127] A switching threshold learning unit is used to determine the switching boundary conditions based on a support vector machine classifier;

[0128] A pre-switching detection unit is used to predict the type of scene to be entered based on the WiFi signal strength gradient;

[0129] A gradient blending unit is used to smooth the positioning results using an exponentially weighted average in the switching transition zone;

[0130] The map matching correction unit is used to correct trajectory jumps when crossing scenes using a hidden map matching algorithm;

[0131] The user feedback interface unit is used to switch states via voice / vibration prompts and receive user confirmation.

[0132] In the above technical solution, the switching threshold learning unit uses support vector machines to accurately determine the switching boundary, providing a reliable basis for scene switching; the pre-switching detection unit predicts the scene type in advance through WiFi signal strength gradients, enhancing the initiative of switching; the gradual fusion unit uses exponential weighted averaging to make the positioning results in the switching transition zone smoother; the map matching correction unit uses a hidden map matching algorithm to effectively correct trajectory jumps; and the user feedback interface unit provides voice / vibration prompts and receives confirmation, improving the user experience. The collaboration of these units enables seamless and stable switching between indoor and outdoor scenes, ensuring navigation continuity.

[0133] In one specific implementation scheme, the positioning result output and visualization module includes:

[0134] Coordinate transformation unit, used to support real-time transformation between WGS84, CGCS2000 and local coordinate systems;

[0135] The path planning unit is used to combine the A* algorithm with dynamic obstacle avoidance to generate real-time navigation instructions.

[0136] AR overlay display unit, used to mark navigation arrows and distances in the real scene using the mobile phone camera;

[0137] A voice interaction unit is used to support bilingual (Chinese and English) broadcasting and dialect recognition;

[0138] The privacy protection unit is used to perform end-to-end encryption of location data using homomorphic encryption technology.

[0139] In the above technical solution, the coordinate transformation unit enables real-time conversion between multiple coordinate systems to meet the needs of different scenarios; the path planning unit combines the A* algorithm with dynamic obstacle avoidance to generate accurate real-time navigation commands; the AR overlay display unit uses the mobile phone camera to annotate information in the real scene, making it intuitive and easy to use; the voice interaction unit supports Chinese, English, and dialects, improving the convenience of interaction; and the privacy protection unit uses homomorphic encryption technology to ensure the security of positioning data. The collaboration of these units makes the navigation results more accurate, the display more intuitive, and the interaction more user-friendly, while comprehensively protecting user privacy and enhancing user confidence.

[0140] In one feasible implementation, an indoor-outdoor seamless navigation system based on dynamic allocation of multi-source positioning weights includes:

[0141] 1: Multi-source positioning data acquisition module

[0142] Function: Acquire raw data from different location sources in real time and preprocess it.

[0143] Unit composition:

[0144] 1. GPS / BeiDou receiver unit: Used to support dual-mode satellite signal reception, and adopts anti-multipath interference design;

[0145] 2. UWB / Bluetooth Beacon Unit: Includes integrated ultra-wideband pulse transmitter and Bluetooth Low Energy 5.1 protocol;

[0146] 3. Geomagnetic sensor array: including a triaxial magnetometer and temperature compensation circuit, used to eliminate environmental magnetic interference;

[0147] 4. Visual positioning unit: includes a binocular camera and a deep learning acceleration chip (such as an NPU), supporting SLAM and feature point matching;

[0148] 5. Inertial Measurement Unit (IMU): Includes a six-axis gyroscope and an accelerometer, used for motion state prediction.

[0149] Beneficial effects include: achieving timestamp alignment of multi-sensor data through hardware-level synchronization clock with an error of <10μs.

[0150] 2: Building Structure Big Data Engine

[0151] Function: Construct a three-dimensional spatial constraint model to provide prior knowledge for the reliability assessment of the location source.

[0152] Unit composition:

[0153] 1. BIM Model Parsing Unit: Used to parse geometric parameters of walls, elevators, doors and windows, etc. in Building Information Model (BIM);

[0154] 2. Material Electromagnetic Properties Library: Used to store the attenuation coefficients of concrete, steel, and other materials for UWB / Bluetooth signals;

[0155] 3. Dynamic obstacle detection unit: Identifies temporary obstacles (such as mobile shelves) using lidar or vision.

[0156] 4. Geomagnetic reference map generation unit: used to construct an indoor geomagnetic field intensity distribution map based on crowdsourced data;

[0157] 5. Path topology optimization unit: used to generate barrier-free optimal paths in conjunction with Dijkstra's algorithm.

[0158] Beneficial effects include: integrating the electromagnetic properties of building structures with geomagnetic reference maps, improving the accuracy of location source reliability assessment by 27%.

[0159] 3: Location Source Reliability Assessment Module

[0160] Function: Quantify the reliability of each location source in a specific scenario.

[0161] Unit composition:

[0162] 1. Signal Quality Analysis Unit: Calculates the signal-to-noise ratio (SNR) and multipath effect index for UWB / Bluetooth;

[0163] 2. Geometric Factor of Precision (GDOP) Calculation Unit: Evaluates the impact of the spatial distribution of satellites / beacons on positioning errors;

[0164] 3. Visual Feature Matching Degree Unit: Statistically calculates the number of matched feature points and the reprojection error;

[0165] 4. Geomagnetic consistency verification unit: Compares the Mahalanobis distance between real-time measurements and the baseline map;

[0166] 5. Motion state association unit: Verifies the physical rationality of the positioning results through IMU data (such as acceleration change detection).

[0167] Beneficial effects include: employing a dynamic GDOP threshold adjustment algorithm to adaptively optimize satellite selection strategies based on user speed.

[0168] 4: Dynamic weight allocation decision module

[0169] Function: Adjusts the weight of the location source in real time based on a multi-factor decision model.

[0170] Unit composition:

[0171] 1. Fuzzy logic reasoning unit: Constructing a three-dimensional membership function of "speed-scenario-reliability";

[0172] 2. Reinforcement learning optimization unit: The weight allocation strategy is trained using the DDPG algorithm, and the reward function is to minimize the localization error;

[0173] 3. Conflict Resolution Unit: When results from multiple location sources contradict each other, the DS evidence theory fusion decision is activated;

[0174] 4. Historical trajectory backtracking unit: Kalman filtering is used to correct short-term abnormal weight fluctuations;

[0175] 5. User Preference Learning Unit: Analyze user behavior patterns using Hidden Markov Models (HMMs).

[0176] Beneficial effects include: adopting a "two-layer weight allocation mechanism", first coarsely allocating according to the scenario (such as indoor / outdoor), and then finely adjusting according to reliability.

[0177] 5: Multi-source data fusion and filtering module

[0178] Function: Enables high-precision fusion of heterogeneous positioning data.

[0179] Unit composition:

[0180] 1. Loosely coupled filtering unit: Each localization source is solved independently and then weighted and fused;

[0181] 2. Tightly Coupled Filtering Unit: Directly inputs the raw observations (such as pseudorange and phase) into the Extended Kalman Filter (EKF);

[0182] 3. Adaptive covariance adjustment unit: Dynamically adjusts the EKF process noise matrix based on the reliability of the positioning source;

[0183] 4. Outlier Removal Unit: An improved RANSAC algorithm is used to detect and remove outliers;

[0184] 5. Smoothing filter unit: A Savitzky-Golay filter is used to suppress high-frequency noise.

[0185] Beneficial effects include: the use of a "tight-loose coupling hybrid switching algorithm" to employ tight coupling when satellite signals are good and automatically switch to loose coupling when signals are blocked.

[0186] 6: Indoor / Outdoor Scene Switching Control Module

[0187] Function: Enables seamless switching triggering and smooth transition.

[0188] Unit composition:

[0189] 1. Switching threshold learning unit: The switching boundary conditions are determined based on a support vector machine (SVM) classifier;

[0190] 2. Pre-switching detection unit: Predicts the type of scene to be entered based on the WiFi signal strength gradient;

[0191] 3. Gradual Blending Unit: The positioning result is smoothed using an exponentially weighted average (EWA) in the switching transition zone;

[0192] 4. Map Matching Correction Unit: Utilizes Hidden Map Matching (HMM) algorithm to correct trajectory jumps when crossing scenes;

[0193] 5. User feedback interface unit: Switches states via voice / vibration prompts and receives user confirmation.

[0194] Beneficial effects include: using "virtual beacon" technology to deploy temporary UWB beacons at the indoor and outdoor boundaries to simulate GPS signals, achieving seamless switching.

[0195] 7: Location Result Output and Visualization Module

[0196] Function: Transforms location data into navigation information that users can understand.

[0197] Unit composition:

[0198] 1. Coordinate transformation unit: Supports real-time transformation between WGS84, CGCS2000, and local coordinate systems;

[0199] 2. Path planning unit: Combines the A* algorithm with dynamic obstacle avoidance to generate real-time navigation instructions;

[0200] 3. AR overlay display unit: Marks navigation arrows and distances in the real scene using the phone's camera;

[0201] 4. Voice interaction unit: Supports bilingual (Chinese and English) broadcasting and dialect recognition;

[0202] 5. Privacy Protection Unit: Homomorphic encryption technology is used to encrypt location data end-to-end.

[0203] Beneficial effects include: using the "semantic positioning" function to map coordinate points to natural language descriptions such as "elevator entrance on the 3rd floor" and "next to the shelf in area A".

[0204] The technical effects of the above technical solution include:

[0205] 1. Positioning accuracy: Continuous indoor and outdoor positioning error <0.5 meters, a 60% improvement over traditional methods;

[0206] 2. Switching latency: Scene switching time <0.3 seconds, achieving truly seamless navigation;

[0207] 3. Robustness: It can maintain more than 95% availability in NLOS (non-line-of-sight) environments;

[0208] 4. Energy consumption optimization: Sensor power consumption is reduced by 28% through dynamic weight allocation.

[0209] In one specific implementation scheme, a central control system is also included, comprising:

[0210] 1. Data Reception and Preprocessing Module

[0211] Multi-source data access unit: Adopting a high-efficiency data interface protocol, compatible with multiple communication methods, it can simultaneously receive raw data from the GPS / BeiDou receiving unit, UWB / Bluetooth beacon unit, geomagnetic sensor array, visual positioning unit, and inertial measurement unit in the multi-source positioning data acquisition module, ensuring data integrity and real-time performance.

[0212] Data verification unit: Uses a hash algorithm to verify the integrity of received data. By comparing it with a preset verification value, it quickly identifies possible data loss or errors during data transmission and triggers a retransmission mechanism in a timely manner to ensure data accuracy.

[0213] Time synchronization correction unit: Based on the feedback information of the hardware-level synchronization clock, the timestamps of each positioning source data are corrected a second time. A high-precision interpolation algorithm is used to further eliminate the time deviation caused by the difference in sensor sampling frequency, and ensure strict alignment of multi-source data in the time dimension.

[0214] The preliminary screening unit for abnormal data uses simple statistical rules, such as data range threshold judgment, to perform preliminary screening on the received raw data, quickly removing abnormal data that obviously exceeds the reasonable range, reducing the amount of data to be processed later, and improving the overall efficiency of the system.

[0215] Data format standardization unit: Converts and stores raw data from different positioning sources in a unified format, defines a standard data structure, including data type, precision, unit and other information, to facilitate unified processing and analysis of the data by subsequent modules.

[0216] Beneficial effects: The multi-source data access unit enables centralized reception of data from multiple positioning sources, the data verification unit ensures data integrity, the time synchronization correction unit ensures data time alignment, the abnormal data preliminary screening unit improves processing efficiency, and the data format standardization unit facilitates subsequent processing, providing a reliable data foundation for the system's accurate positioning.

[0217] 2. Positioning Strategy Decision Module

[0218] Scene Recognition Unit: Combining building information provided by the building structure big data engine with current location data, and using decision tree algorithms in machine learning, it quickly and accurately identifies the indoor and outdoor scenes where the user is located, and determines whether the user is in an indoor shopping mall, office building, or outdoor street, park, etc.

[0219] Location Demand Analysis Unit: Based on users' historical behavior data, current movement status, and application scenarios, the unit analyzes users' location needs using a Hidden Markov Model. For example, when shopping, users may pay more attention to the precise location of indoor stores, while while walking, they may focus more on the overall navigation path.

[0220] Location source priority evaluation unit: Combining the credibility information of each location source provided by the location source reliability evaluation module, the Analytic Hierarchy Process (AHP) is used to dynamically evaluate the priority of each location source under different scenarios, and determine which location sources contribute more to the location results in the current scenario.

[0221] Dynamic weight allocation strategy generation unit: Based on the location source priority evaluation results, it uses the Q-learning algorithm in reinforcement learning to generate a real-time location source weight allocation strategy. It continuously optimizes the weight allocation based on system feedback to make the location results more accurate and stable.

[0222] Emergency Positioning Strategy Unit: When some positioning sources are detected to be faulty or have abnormal data, the system automatically switches to the emergency positioning strategy. Based on the data from the remaining available positioning sources, a simplified positioning algorithm is used, such as dead reckoning based on the inertial measurement unit, to ensure that basic positioning services can still be provided under special circumstances.

[0223] Beneficial effects: The scene recognition unit accurately determines the user's scene, the positioning demand analysis unit understands the user's needs, the positioning source priority evaluation unit determines the importance of the positioning source, the dynamic weight allocation strategy generation unit achieves accurate weight allocation, and the emergency positioning strategy unit ensures positioning in special situations, thereby improving the system's positioning adaptability and reliability in different scenarios.

[0224] 3. Data Fusion and Optimization Module

[0225] Loosely coupled fusion execution unit: According to the weights generated by the dynamic weight allocation decision module, the location data independently calculated by each positioning source is weighted and fused. The weighted average algorithm is used to comprehensively consider the accuracy and reliability of different positioning sources to obtain the preliminary fusion positioning result.

[0226] Tightly Coupled Fusion Processing Unit: The raw observations from each positioning source are directly input into the Extended Kalman Filter (EKF) framework. Based on the measurement model of the positioning source and the system state equation, the unit performs more in-depth data fusion processing, making full use of the information in the raw data to improve positioning accuracy.

[0227] Adaptive noise adjustment unit: Based on the real-time reliability information of each positioning source provided by the positioning source reliability assessment module, the noise matrix in the EKF process is dynamically adjusted. Using a fuzzy control algorithm, the noise parameters are adaptively adjusted according to the range of changes in reliability indicators, so that the filter can better adapt to the data characteristics under different scenarios.

[0228] Outlier depth detection and removal unit: It adopts an improved random sampling consensus (RANSAC) algorithm to perform deep outlier detection on the fused data. Through multiple random samplings and model fitting, it can more accurately identify and remove abnormal data caused by environmental interference or sensor failure, thus ensuring the accuracy of the positioning results.

[0229] Data Smoothing and Optimization Unit: The Savitzky-Golay filter is used to smooth the positioning data after outlier removal. The appropriate window size and polynomial order are selected according to the characteristics of the data to effectively suppress high-frequency noise while retaining the trend information of the data, making the positioning results more stable and reliable.

[0230] Beneficial effects: The loosely coupled fusion execution unit and the tightly coupled fusion processing unit realize data fusion at different levels; the adaptive noise adjustment unit enhances the adaptability of the filter; the outlier depth detection and removal unit ensures data quality; and the data smoothing and optimization unit makes the positioning results more stable, thereby improving the accuracy and reliability of navigation and positioning.

[0231] 4. Scene switching control and coordination module

[0232] The switching condition comprehensive judgment unit integrates the switching boundary conditions determined by the switching threshold learning unit in the indoor and outdoor scene switching control module, the scene type information predicted by the pre-switching detection unit, and the current positioning data, and uses a multi-factor decision model to comprehensively judge whether to trigger scene switching.

[0233] The handover process planning unit: Based on the judgment results of the handover conditions, it formulates a detailed scene handover process plan, including the handover time nodes, the order of enabling and disabling each positioning source, and the adjustment of data fusion strategies, to ensure the orderly conduct of the handover process.

[0234] Gradual fusion coordination unit: In the handover transition zone, the coordination gradual fusion unit uses an exponential weighted average method to smooth the positioning results. At the same time, according to the handover process plan, the parameters of the exponential weighted average are dynamically adjusted to enable the positioning results to achieve a natural and smooth transition during the handover process.

[0235] Map Matching Correction Coordination Unit: Working closely with the Map Matching Correction Unit, it provides timely scene information and positioning data before and after the switch when switching between scenes, assisting the hidden map matching algorithm in accurately correcting trajectory jumps and ensuring the continuity and accuracy of the navigation trajectory.

[0236] User feedback processing unit: Receives user confirmation information from the user feedback interface unit and adjusts the scene switching process in real time based on user feedback. If the user is not satisfied with the switching prompt, the prompt method and timing can be optimized to improve the user experience.

[0237] Beneficial effects: The switching condition comprehensive judgment unit accurately determines the switching timing, the switching process planning unit formulates an orderly switching plan, the gradual fusion coordination unit achieves a smooth transition, the map matching correction coordination unit ensures trajectory continuity, and the user feedback processing unit improves the user experience, achieving seamless switching between indoor and outdoor scenes.

[0238] 5. Navigation command generation and output module

[0239] Path planning optimization unit: Combining the barrier-free optimal path information generated by the path topology optimization unit with the current positioning data and user destination, the improved A* algorithm is used for real-time path planning, taking into account the influence of dynamic obstacles, to generate a navigation path that is more in line with the actual environment.

[0240] Navigation instruction generation unit: Based on the navigation path generated by the path planning and optimization unit, it converts the path into specific navigation instructions, including information such as direction of travel, turning angle, and distance prompts. The instructions are presented in a concise and clear format to facilitate user understanding and execution.

[0241] Multimodal output coordination unit: coordinates the output of positioning results with the AR overlay display unit and voice interaction unit in the visualization module. It selects the appropriate output mode according to the user's usage habits and the current scenario. For example, it prioritizes voice prompts when walking and combines AR display when searching for a specific location.

[0242] Output content customization unit: Based on user behavior patterns and preference information obtained from the user preference learning unit, navigation instructions and visual content are customized. For example, if users prefer a simple interface, unnecessary display information is reduced; if users have a need for a specific language, corresponding voice broadcast is provided.

[0243] Privacy Protection Coordination Unit: Working closely with the Privacy Protection Unit, this unit ensures that users' location data and personal information are effectively encrypted and protected during the generation and output of navigation instructions. Homomorphic encryption technology is used to encrypt data involving user privacy end-to-end to prevent data leakage.

[0244] Beneficial effects: The route planning optimization unit generates better routes, the navigation instruction generation unit provides clear instructions, the multimodal output coordination unit selects appropriate output methods, the output content customization unit meets users' personalized needs, and the privacy protection coordination unit protects user privacy, thereby improving the quality of navigation services and user satisfaction.

[0245] 6. System Monitoring and Maintenance Module

[0246] Performance monitoring unit: Monitors the operating status of each module of the system in real time, including the real-time performance of data reception, the accuracy of positioning calculation, the effect of data fusion, and the smoothness of scene switching. It uses key performance indicators (KPIs) for quantitative evaluation and promptly identifies problems that occur during system operation.

[0247] Fault Diagnosis Unit: When the performance monitoring unit detects an anomaly in the system, it uses the Fault Tree Analysis (FTA) method to diagnose the fault, locate the module and cause of the fault, such as sensor failure, algorithm error, communication interruption, etc., and provide a basis for troubleshooting.

[0248] Automatic Repair Unit: For some common faults, such as temporary abnormal sensor data or temporary communication interruption, the repair mechanism is automatically triggered, and automatic repair is performed by methods such as data retransmission and algorithm parameter adjustment to restore the normal operation of the system as soon as possible.

[0249] Log recording and analysis unit: Records various data and events during system operation, including location data, operation instructions, fault information, etc. Data mining technology is used to analyze the logs to discover the patterns and potential problems in system operation, providing a reference for system optimization and upgrade.

[0250] System Upgrade Management Unit: Responsible for system software upgrades and configuration management. When new algorithms, functions, or data need to be updated, the updated content is deployed to the system through a secure upgrade mechanism to ensure continuous improvement in system performance and functionality, while ensuring the stability and reliability of the upgrade process.

[0251] Beneficial effects: The performance monitoring unit monitors the system status in real time, the fault diagnosis unit quickly locates the cause of the fault, the automatic repair unit restores the system operation in a timely manner, the log recording and analysis unit provides a basis for system optimization, and the system upgrade management unit ensures continuous system improvement, thereby enhancing the system's stability and maintainability.

[0252] exist Figure 2 This application provides a method for seamless indoor and outdoor navigation based on dynamic allocation of multi-source positioning weights, comprising the following steps:

[0253] The multi-source positioning data acquisition module is used to acquire raw data from different positioning sources in real time and perform preprocessing. The timestamps of the multi-sensor data are aligned by a hardware-level synchronization clock.

[0254] Utilize a building structure big data engine to construct a three-dimensional spatial constraint model;

[0255] The reliability of each positioning source in a specific scenario is quantified using the positioning source reliability assessment module;

[0256] The dynamic weight allocation decision module adjusts the weights of the location sources in real time based on a multi-factor decision model.

[0257] Heterogeneous positioning data are fused using a multi-source data fusion and filtering module;

[0258] The indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes.

[0259] In the above technical solution, a multi-source positioning data acquisition module is set up to acquire raw data from different positioning sources in real time and perform preprocessing, and the timestamps of multi-sensor data are aligned through a hardware-level synchronization clock; a building structure big data engine is used to construct a three-dimensional spatial constraint model; a positioning source reliability assessment module is used to quantify the credibility of each positioning source in a specific scenario; a dynamic weight allocation decision module is used to adjust the weight of positioning sources in real time based on a multi-factor decision model; a multi-source data fusion filtering module is used to fuse heterogeneous positioning data; and an indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes. This achieves efficient collaboration of multiple positioning sources, improving navigation accuracy and adaptability.

[0260] Those skilled in the art will know that this application can be implemented as a system, method, or computer program product.

[0261] Therefore, this disclosure can be implemented in the following forms: it can be entirely hardware, entirely software (including firmware, resident software, microcode, etc.), or a combination of hardware and software, generally referred to herein as a "circuit," "module," or "system." Furthermore, in some embodiments, this application can also be implemented as a computer program product in one or more computer-readable media, the computer-readable media containing computer-readable program code.

[0262] Any combination of one or more computer-readable media may be used. A computer-readable medium can be a computer-readable signal medium or a computer-readable storage medium. A computer-readable storage medium can be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples (a non-exhaustive list) of computer-readable storage media include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage device, magnetic storage device, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in connection with an instruction execution system, apparatus, or device.

[0263] Although embodiments of this application have been shown and described above, it is understood that the above embodiments are exemplary and should not be construed as limiting this application. Those skilled in the art can make changes, modifications, substitutions, and variations to the above embodiments within the scope of this application. Based on this, various substitutions and improvements can be made to this application, all of which fall within the protection scope of this application.

Claims

1. An indoor-outdoor seamless navigation system based on dynamic allocation of multi-source positioning weights, characterized in that, include: The multi-source positioning data acquisition module is used to acquire raw data from different positioning sources in real time and perform preprocessing, and to align the timestamps of multi-sensor data through a hardware-level synchronization clock. A building structure big data engine for constructing three-dimensional spatial constraint models; The location source reliability assessment module is used to quantify the reliability of each location source in a specific scenario; The dynamic weight allocation decision module is used to adjust the weight of the location source in real time based on a multi-factor decision model; A multi-source data fusion and filtering module is used to fuse heterogeneous positioning data; The indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes. The dynamic weight allocation decision module includes: Fuzzy logic reasoning unit, used to construct three-dimensional membership functions; The reinforcement learning optimization unit is used to train the weight allocation strategy using the DDPG algorithm. The conflict resolution unit is used to enable DS evidence theory fusion decision-making when multiple location source results are contradictory. The historical trajectory backtracking unit is used to correct short-term abnormal weight fluctuations using Kalman filtering; User preference learning unit, used to analyze user behavior patterns through hidden Markov models; The multi-source data fusion filtering module includes: Loosely coupled filtering unit is used for weighted fusion after independent calculation of each positioning source; Tightly coupled filtering unit, used to directly input the raw observations into the extended Kalman filter; An adaptive covariance adjustment unit is used to dynamically adjust the EKF process noise matrix based on the reliability of the positioning source; The outlier removal unit is used to detect and remove outliers using an improved RANSAC algorithm. A smoothing filter unit is used to suppress high-frequency noise using a Savitzky-Golay filter; The indoor / outdoor scene switching control module includes: A switching threshold learning unit is used to determine the switching boundary conditions based on a support vector machine classifier; A pre-switching detection unit is used to predict the type of scene to be entered based on the WiFi signal strength gradient; A gradient blending unit is used to smooth the positioning results using an exponentially weighted average in the switching transition zone; The map matching correction unit is used to correct trajectory jumps when crossing scenes using a hidden map matching algorithm; The user feedback interface unit is used to switch states via voice / vibration prompts and receive user confirmation.

2. The seamless indoor and outdoor navigation system based on dynamic allocation of multi-source positioning weight according to claim 1, characterized in that, Also includes: The positioning results output and visualization module is used to transform positioning data into understandable navigation information.

3. The seamless indoor and outdoor navigation system based on dynamic allocation of multi-source positioning weight according to claim 2, characterized in that, The multi-source positioning data acquisition module includes: The GPS / BeiDou receiver unit is used for dual-mode satellite signal reception. UWB / Bluetooth beacon unit for integrating ultra-wideband pulse transmission and Bluetooth Low Energy 5.1 protocol; A geomagnetic sensor array, including a triaxial magnetometer and a temperature compensation circuit, is used to eliminate environmental magnetic interference; The visual positioning unit includes a binocular camera and a deep learning acceleration chip, used to support SLAM and feature point matching; An inertial measurement unit, including a six-axis gyroscope and an accelerometer, is used for motion state prediction.

4. The indoor and outdoor seamless navigation system based on dynamic allocation of multi-source positioning weight according to claim 3, characterized in that, The building structure big data engine includes: BIM model parsing unit, used to parse geometric parameters in building information model; A library of electromagnetic properties of materials, used to store attenuation coefficients for UWB / Bluetooth signals; A dynamic obstacle detection unit is used to identify temporary obstacles; The geomagnetic reference map generation unit is used to construct an indoor geomagnetic field intensity distribution map; The path topology optimization unit is used to generate the optimal barrier-free path.

5. The seamless indoor and outdoor navigation system based on dynamic allocation of multi-source positioning weight according to claim 4, characterized in that, The location source reliability assessment module includes: Signal quality analysis unit, used to calculate the signal-to-noise ratio and multipath effect index of UWB / Bluetooth; The geometric accuracy factor calculation unit is used to evaluate the impact of the spatial distribution of satellites / beacons on positioning errors; The visual feature matching degree unit is used to count the number of feature point matches and the reprojection error. The geomagnetic consistency verification unit is used to compare the real-time measurement value with the Mahalanobis distance of the reference map; The motion state association unit is used to verify the physical validity of the positioning results using IMU data.

6. The indoor and outdoor seamless navigation system based on dynamic allocation of multi-source positioning weight according to claim 5, characterized in that, The positioning result output and visualization module includes: Coordinate transformation unit, used to support real-time transformation between WGS84, CGCS2000 and local coordinate systems; The path planning unit is used to combine the A* algorithm with dynamic obstacle avoidance to generate real-time navigation instructions. AR overlay display unit, used to mark navigation arrows and distances in the real scene using the mobile phone camera; A voice interaction unit is used to support bilingual (Chinese and English) broadcasting and dialect recognition; The privacy protection unit is used to perform end-to-end encryption of location data using homomorphic encryption technology. 7.A seamless indoor and outdoor navigation method based on dynamic distribution of multi-source positioning weights, characterized in that, Includes the following steps: The multi-source positioning data acquisition module is used to acquire raw data from different positioning sources in real time and perform preprocessing. The timestamps of the multi-sensor data are aligned by a hardware-level synchronization clock. Utilize a building structure big data engine to construct a three-dimensional spatial constraint model; The reliability of each positioning source in a specific scenario is quantified using the positioning source reliability assessment module; The dynamic weight allocation decision module adjusts the weights of the location sources in real time based on a multi-factor decision model. Heterogeneous positioning data are fused using a multi-source data fusion and filtering module; The indoor / outdoor scene switching control module is used to trigger and smoothly transition between indoor and outdoor scenes. The dynamic weight allocation decision module includes: Fuzzy logic reasoning unit, used to construct three-dimensional membership functions; The reinforcement learning optimization unit is used to train the weight allocation strategy using the DDPG algorithm. The conflict resolution unit is used to enable DS evidence theory fusion decision-making when multiple location source results are contradictory. The historical trajectory backtracking unit is used to correct short-term abnormal weight fluctuations using Kalman filtering; User preference learning unit, used to analyze user behavior patterns through hidden Markov models; The multi-source data fusion filtering module includes: Loosely coupled filtering unit is used for weighted fusion after independent calculation of each positioning source; Tightly coupled filtering unit, used to directly input the raw observations into the extended Kalman filter; An adaptive covariance adjustment unit is used to dynamically adjust the EKF process noise matrix based on the reliability of the positioning source; The outlier removal unit is used to detect and remove outliers using an improved RANSAC algorithm. A smoothing filter unit is used to suppress high-frequency noise using a Savitzky-Golay filter; The indoor / outdoor scene switching control module includes: A switching threshold learning unit is used to determine the switching boundary conditions based on a support vector machine classifier; A pre-switching detection unit is used to predict the type of scene to be entered based on the WiFi signal strength gradient; A gradient blending unit is used to smooth the positioning results using an exponentially weighted average in the switching transition zone; The map matching correction unit is used to correct trajectory jumps when crossing scenes using a hidden map matching algorithm; The user feedback interface unit is used to switch states via voice / vibration prompts and receive user confirmation.

Images