Method and system for realizing civil aviation industry airport AR navigation based on large model
Through large-scale architecture and AR technology, the problem of low efficiency of traditional navigation systems during layout changes and peaks is solved, high-precision navigation and personalized services are achieved, and user experience and system efficiency are improved.
Patent Information
- Application Number
- CN202411118501.X
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-08-15
- Publication Date
- 2024-11-29
AI Technical Summary
Traditional airport navigation systems are inefficient in the face of changes in layout and peak traffic, and cannot provide real-time navigation, and cannot provide users with customized navigation experience, which affects the travel experience of passengers.
Using a large model architecture, through designing data preprocessing, feature extraction, deep learning and real-time data fusion, we build high-precision three-dimensional scenes to achieve user precise positioning and personalized navigation path recommendations, and superimpose information into user's field of view through the AR navigation interface. , providing real-time interaction and intelligent decision-making support.
It improves the accuracy and stability of navigation, provides a tailor-made navigation experience, simplifies the information acquisition process, improves user satisfaction, and ensures the long-term effectiveness of the system through continuous learning and optimization.
Smart Images

Figure CN119048716A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of large models, and in particular to a method and system for realizing AR navigation of an airport in the civil aviation industry based on a large model. Background Art
[0002] With the popularity of air travel, airports are important nodes for passengers, and the convenience and accuracy of their navigation systems are crucial to passengers' travel experience. Traditional airport navigation systems mainly have the following methods:
[0003] 1. Signage system: Through clear icons and text instructions, it can guide passengers to find key areas such as boarding gates, security checks, and baggage claim. It is suitable for all passengers, regardless of whether they have or are familiar with electronic devices. It is almost everywhere in the airport, forming a continuous guidance path. However, the shortcomings of the signage system are also obvious. Once the layout or flight information changes, the update of the sign may lag behind, leading to misleading passengers. A large number of signs may occupy public space, affecting the beauty and traffic efficiency of the airport. Signage cannot give passengers an estimate of the time it will take to arrive at their destination.
[0004] 2. Manual service: Staff can provide personalized guidance and assistance based on the specific needs of passengers. For non-standard issues, such as special passenger services and emergency response, manual service is more flexible and effective. However, manual service has high labor costs and requires a large number of employees to maintain services, especially during peak hours. Manual service efficiency is low, and when there is a large flow of people, passengers may have to wait for a long time to get help. The quality of service varies, depending on the professional level and attitude of the staff.
[0005] Static maps: provide passengers with a bird's-eye view of the entire airport layout, which helps to plan the route of action at a macro level. Passengers can check it at any time, regardless of time and place. The disadvantages of static maps are: the information is fixed and cannot reflect real-time changes, such as temporarily closed channels or changed boarding gates. For passengers with poor sense of direction or first-time arrivals, it may be difficult to quickly understand the map and find the destination. Maps of large airports may be large and inconvenient to hold or carry with you. When looking at a large map for the first time, it is difficult to quickly find the target location. Summary of the invention
[0006] The purpose of the present invention is to provide a method and system for realizing AR navigation of airports in the civil aviation industry based on a large model, so as to solve the problems raised in the above-mentioned background technology.
[0007] To achieve the above object, the present invention provides the following technical solution: a method for realizing airport AR navigation in the civil aviation industry based on a large model, the method comprising the following steps:
[0008] Design large model architecture for processing and analyzing airport environmental data;
[0009] Environmental data collection and processing, used to collect images and sensor data inside the airport in real time;
[0010] 3D scene reconstruction, used to build a high-precision 3D model of the airport interior;
[0011] Visual positioning and sensor data fusion are used to achieve accurate user positioning;
[0012] User behavior analysis, used to generate personalized navigation paths and service recommendations;
[0013] AR navigation interface rendering, used to overlay navigation information onto the user’s actual field of view;
[0014] Real-time interaction and intelligent decision support, used to provide real-time interaction and decision support;
[0015] Feedback mechanism, used to enable large models to continuously learn and optimize based on user feedback.
[0016] Preferably, the specific operations of designing the large model architecture include:
[0017] Data preprocessing, which is responsible for cleaning, normalizing, and enhancing input data to improve the accuracy and robustness of the model, including noise removal, missing value filling, standardization, and normalization operations;
[0018] Feature extraction, using computer vision techniques to extract key features from 3D models and sensor data to achieve edge detection, corner detection, texture analysis, and object recognition;
[0019] Deep learning, which uses multi-layer neural networks to learn features and recognize patterns to understand complex spatial relationships and user behaviors;
[0020] Decision making, based on the learned knowledge, generates navigation decisions and path planning in real time, involving path search algorithms, Dijkstra algorithm, and machine learning models to predict traffic conditions and crowd density;
[0021] Optimize learning, continuously optimize model performance through continuous online learning and offline training, including using reinforcement learning algorithms to adjust the model based on user feedback and system performance.
[0022] Preferably, the specific operations of environmental data collection and processing include:
[0023] Deploy high-definition cameras at designated locations to capture high-resolution real-time images. The cameras have wide dynamic range and good low-light performance to adapt to various lighting conditions within the airport;
[0024] Use sensor networks to obtain information permissions for the phone's gyroscope, camera, Wi-Fi, LiDAR, and Wi-Fi / Bluetooth beacons;
[0025] Data synchronization ensures that timestamps of all data sources are synchronized to achieve accurate data fusion.
[0026] Preferably, the specific operations of three-dimensional scene reconstruction include:
[0027] Preset model: Preset the 3D model of the airport to provide a basis and reference for real-time modeling and optimize the blind spots of the camera;
[0028] Initial mapping: Using SLAM (Simultaneous Localization and Mapping) technology, a 3D model of the airport is constructed in real time from camera data. The SLAM algorithm handles changes in the dynamic environment, such as moving pedestrians and vehicles.
[0029] Feature matching: Deep learning models are used to identify and match significant features within the airport, such as gates, shops, and signage, which serve as important reference points during navigation.
[0030] Fine Modeling: In combination with LiDAR data, fine modeling of airport structures is performed to provide centimeter-level accuracy. LiDAR data provides high-precision 3D point clouds for generating detailed airport maps.
[0031] Dynamic updating: Updating the 3D model in real time to reflect changes in the airport environment, such as crowd movements and temporary roadblocks, involves using an event-driven update mechanism to update the model only when significant changes are detected.
[0032] Preferably, the specific operations of visual positioning and sensor data fusion include: visual positioning, sensor fusion, multi-source data calibration and real-time trajectory prediction;
[0033] The specific operations of user behavior analysis include: behavior analysis, historical behavior analysis, context awareness, and personalized content generation;
[0034] The specific operations of AR navigation interface rendering include: real-time rendering, visual guidance, information overlay, and user interaction;
[0035] Specific operations of real-time interaction and intelligent decision support include: natural language processing, intelligent recommendation, context-aware decision-making, and multi-round dialogue management.
[0036] An AR navigation system for an airport in the civil aviation industry based on a large model, the system comprising:
[0037] Large model architecture for processing and analyzing airport environmental data;
[0038] Data acquisition module, used to collect images and sensor data inside the airport in real time;
[0039] 3D scene reconstruction module, used to build a high-precision 3D model of the airport interior;
[0040] Visual positioning and sensor data fusion module, used to achieve accurate user positioning;
[0041] User behavior analysis module, used to generate personalized navigation paths and service recommendations;
[0042] AR navigation interface rendering module, used to overlay navigation information into the user's actual field of view;
[0043] Real-time interaction and intelligent decision support module, used to provide real-time interaction and decision support;
[0044] Feedback mechanism, used to enable large models to continuously learn and optimize based on user feedback.
[0045] Preferably, the specific operations of the large model architecture design include:
[0046] Data preprocessing, which is responsible for cleaning, normalizing, and enhancing input data to improve the accuracy and robustness of the model, including noise removal, missing value filling, standardization, and normalization operations;
[0047] Feature extraction, using computer vision techniques to extract key features from 3D models and sensor data to achieve edge detection, corner detection, texture analysis, and object recognition;
[0048] Deep learning, which uses multi-layer neural networks to learn features and recognize patterns to understand complex spatial relationships and user behaviors;
[0049] Decision making, based on the learned knowledge, generates navigation decisions and path planning in real time, involving path search algorithms, Dijkstra algorithm, and machine learning models to predict traffic conditions and crowd density;
[0050] Optimize learning, continuously optimize model performance through continuous online learning and offline training, including using reinforcement learning algorithms to adjust the model based on user feedback and system performance.
[0051] Preferably, the data acquisition module deploys a high-definition camera at a designated location to capture high-resolution real-time images. The camera has a wide dynamic range and good low-light performance to adapt to various lighting conditions in the airport;
[0052] Use sensor networks to obtain information permissions for the phone's gyroscope, camera, Wi-Fi, LiDAR, and Wi-Fi / Bluetooth beacons;
[0053] Data synchronization ensures that timestamps of all data sources are synchronized to achieve accurate data fusion.
[0054] Preferably, the three-dimensional scene reconstruction module includes:
[0055] Preset model: Preset the 3D model of the airport to provide a basis and reference for real-time modeling and optimize the blind spots of the camera;
[0056] Initial mapping: Using SLAM (Simultaneous Localization and Mapping) technology, a 3D model of the airport is constructed in real time from camera data. The SLAM algorithm handles changes in the dynamic environment, such as moving pedestrians and vehicles.
[0057] Feature matching: Deep learning models are used to identify and match significant features within the airport, such as gates, shops, and signage, which serve as important reference points during navigation.
[0058] Fine Modeling: In combination with LiDAR data, fine modeling of airport structures is performed to provide centimeter-level accuracy. LiDAR data provides high-precision 3D point clouds for generating detailed airport maps.
[0059] Dynamic updating: Updating the 3D model in real time to reflect changes in the airport environment, such as crowd movements and temporary roadblocks, involves using an event-driven update mechanism to update the model only when significant changes are detected.
[0060] Preferably, the visual positioning and sensor data fusion module includes: visual positioning, sensor fusion, multi-source data calibration and real-time trajectory prediction;
[0061] The user behavior analysis module includes: behavior analysis, historical behavior analysis, context awareness, and personalized content generation;
[0062] The AR navigation interface rendering module includes: real-time rendering, visual guidance, information overlay and user interaction;
[0063] The real-time interaction and intelligent decision support modules include: natural language processing, intelligent recommendation, context-aware decision-making, and multi-round dialogue management.
[0064] Compared with the prior art, the present invention has the following beneficial effects:
[0065] The invention proposes a method and system for realizing AR navigation of airports in the civil aviation industry based on a large model. The introduction of the large model greatly improves the data processing capability and learning efficiency, so that the system can quickly respond to user needs; the combination of three-dimensional reconstruction and visual positioning technology provides higher positioning accuracy and stability than traditional methods; the generation of personalized services enables users to obtain a tailor-made navigation experience and improves user satisfaction; the intuitive display of the AR navigation interface simplifies the process of users obtaining information and improves the convenience and intuitiveness of navigation; real-time interaction and intelligent decision support enable users to obtain effective help and guidance in complex environments; and the continuous learning and optimization of the large model ensure the long-term effectiveness and continuous improvement of system services. BRIEF DESCRIPTION OF THE DRAWINGS
[0066] Figure 1 is a flow chart of the method of the present invention;
[0067] Figure 2 The data collection flow chart of the present invention. DETAILED DESCRIPTION
[0068] In order to make the purpose and technical solution of the present invention clearly and completely described, and the advantages more clearly understood, the embodiments of the present invention are further described in detail with reference to the accompanying drawings. It should be understood that the specific embodiments described herein are part of the embodiments of the present invention, rather than all of the embodiments, and are only used to explain the embodiments of the present invention, and are not used to limit the embodiments of the present invention. All other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.
[0069] For example, see Figure 1 to Figure 2 The present invention provides a technical solution: a method for realizing airport AR navigation in the civil aviation industry based on a large model, the method comprising the following steps:
[0070] Design a large model architecture for processing and analyzing airport environmental data; specific operations include:
[0071] Data preprocessing, which is responsible for cleaning, normalizing, and enhancing input data to improve the accuracy and robustness of the model, including noise removal, missing value filling, standardization, and normalization operations;
[0072] Feature extraction, using computer vision techniques to extract key features from 3D models and sensor data to achieve edge detection, corner detection, texture analysis, and object recognition;
[0073] Deep learning, which uses multi-layer neural networks to learn features and recognize patterns to understand complex spatial relationships and user behaviors;
[0074] Decision making, based on the learned knowledge, generates navigation decisions and path planning in real time, involving path search algorithms, Dijkstra algorithm, and machine learning models to predict traffic conditions and crowd density;
[0075] Optimize learning, continuously optimize model performance through continuous online learning and offline training, including using reinforcement learning algorithms to adjust the model based on user feedback and system performance.
[0076] Environmental data collection and processing, used to collect images and sensor data inside the airport in real time; specific operations include:
[0077] Deploy high-definition cameras at designated locations to capture high-resolution real-time images. The cameras have wide dynamic range and good low-light performance to adapt to various lighting conditions within the airport;
[0078] Use sensor networks to obtain information permissions for the phone's gyroscope, camera, Wi-Fi, LiDAR, and Wi-Fi / Bluetooth beacons;
[0079] Data synchronization ensures that timestamps of all data sources are synchronized to achieve accurate data fusion.
[0080] 3D scene reconstruction is used to build a high-precision 3D model of the airport interior; specific operations include:
[0081] Preset model: Preset the 3D model of the airport to provide a basis and reference for real-time modeling and optimize the blind spots of the camera;
[0082] Initial mapping: Using SLAM (Simultaneous Localization and Mapping) technology, a 3D model of the airport is constructed in real time from camera data. The SLAM algorithm handles changes in the dynamic environment, such as moving pedestrians and vehicles.
[0083] Feature matching: Deep learning models are used to identify and match significant features within the airport, such as gates, shops, and signage, which serve as important reference points during navigation.
[0084] Fine Modeling: In combination with LiDAR data, fine modeling of airport structures is performed to provide centimeter-level accuracy. LiDAR data provides high-precision 3D point clouds for generating detailed airport maps.
[0085] Dynamic updating: Updating the 3D model in real time to reflect changes in the airport environment, such as crowd movements and temporary roadblocks, involves using an event-driven update mechanism to update the model only when significant changes are detected.
[0086] Visual positioning and sensor data fusion are used to achieve precise user positioning; specifically, they include: 1. Visual positioning: By analyzing the images captured by the camera, a deep learning model is used to identify the user's position.
[0087] 2. Sensor fusion: Combine the data information collected from the user side to improve positioning accuracy and robustness. The sensor fusion algorithm needs to be able to handle the uncertainty and errors of various sensor data.
[0088] 3. Multi-source data calibration: Ensure the consistency and accuracy of data from different sensors through calibration algorithms. Involves using calibration objects or known reference points to adjust sensor outputs.
[0089] 4. Real-time trajectory prediction: predict the user's movement trajectory to adjust the navigation strategy in advance. This involves using the time series prediction model, the long short-term memory network (LSTM).
[0090] User behavior analysis is used to generate personalized navigation paths and service recommendations; specifically, it includes:
[0091] 1. Behavior analysis: Analyze the user's arrival time, ticket processing progress, etc., and analyze the user's next possible behavior
[0092] 2. Historical behavior analysis: Analyze the user's historical behavior records, such as how they check in, whether they need charging services, etc.
[0093] 3. Context-aware: Considers the user’s current context, such as time, location, and itinerary, to generate the most appropriate navigation path.
[0094] 4. Personalized content generation: Use big models to generate personalized navigation instructions and related information, such as store discounts and gate change notifications.
[0095] AR navigation interface rendering is used to overlay navigation information into the user's actual field of view; specifically includes:
[0096] 1. Real-time rendering: Render the 3D model and navigation information into the user's field of view in real time. Use a high-performance graphics rendering engine and optimized rendering algorithms to ensure a smooth user experience.
[0097] 2. Visual guidance: Use arrows, highlights, and animation effects to guide users to their destination. Visual guidance design needs to consider the user’s attention and cognitive load.
[0098] 3. Information overlay: Useful information is overlaid in the user’s field of vision, such as boarding time, route distance, and direction instructions. Information overlay needs to be designed to be clear and not interfere with the user’s field of vision.
[0099] 4. User interaction: Allow users to interact with the AR interface through gestures, voice, or touch to get more information or change settings. User interaction design needs to be simple and intuitive to reduce the user's learning curve.
[0100] Real-time interaction and intelligent decision support, used to provide real-time interaction and decision support; specifically including:
[0101] 1. Natural language processing: Use NLP technology to parse the user's voice or text instructions and provide corresponding feedback. This involves the use of speech recognition and natural language understanding (NLU) technology.
[0102] 2. Intelligent recommendation: Intelligently recommend the best navigation path and airport services based on the user's current location and behavior. The recommendation system uses collaborative filtering or content recommendation algorithms.
[0103] 3. Context-aware decision making: Considering the user’s context and environmental factors to provide the most appropriate decision support. This involves using context-aware reasoning algorithms to understand the user’s needs and constraints, such as the user’s gender.
[0104] 4. Multi-turn dialogue management: Through multi-turn dialogue, the system can understand complex user needs and provide detailed navigation assistance. The dialogue management system needs to be able to process contextual information and user intent.
[0105] Feedback mechanism, used to enable the large model to continuously learn and optimize based on user feedback, including:
[0106] 1. Online learning: When users interact with the system, the model is updated in real time to adapt to changes in user behavior.
[0107] 2. Offline training: Regularly train the model using large amounts of collected data to improve accuracy and efficiency.
[0108] 3. Performance monitoring: Monitor the performance of the model to detect and solve problems in a timely manner, such as overfitting and underfitting.
[0109] 4. Feedback loop: Establish a user feedback mechanism and use user satisfaction and suggestions as an important reference for optimizing the model.
[0110] Also includes: Data security and privacy protection
[0111] When processing user data, the system strictly abides by the regulations on data security and privacy protection:
[0112] 1. Data encryption: Encrypt all sensitive data to prevent unauthorized access.
[0113] 2. Anonymization: When analyzing user data, anonymization technology is used to protect user identity.
[0114] 3. Access control: Implement strict access control policies to ensure that only authorized personnel can access data.
[0115] 4. Compliance Checks: Regular compliance checks are conducted to ensure that the system complies with all relevant data protection regulations.
[0116] Embodiment 2, based on embodiment 1, proposes an airport AR navigation system for the civil aviation industry based on a large model, the system comprising:
[0117] Large model architecture for processing and analyzing airport environmental data;
[0118] Data acquisition module, used to collect images and sensor data inside the airport in real time;
[0119] 3D scene reconstruction module, used to build a high-precision 3D model of the airport interior;
[0120] Visual positioning and sensor data fusion module, used to achieve accurate user positioning;
[0121] User behavior analysis module, used to generate personalized navigation paths and service recommendations;
[0122] AR navigation interface rendering module, used to overlay navigation information into the user's actual field of view;
[0123] Real-time interaction and intelligent decision support module, used to provide real-time interaction and decision support;
[0124] Feedback mechanism, used to enable large models to continuously learn and optimize based on user feedback.
[0125] The specific operations are as follows:
[0126] Large model architecture design
[0127] The big model architecture is the core of this system, responsible for processing, analyzing and learning all the collected data. The architecture consists of multiple sub-modules, including data preprocessing module, feature extraction module, deep learning module, decision making module and optimization learning module.
[0128] 1. Data preprocessing module: responsible for cleaning, normalizing and enhancing input data to improve the accuracy and robustness of the model, including operations such as removing noise, filling missing values, standardization and normalization.
[0129] 2. Feature extraction module: Use advanced computer vision technology to extract key features from 3D models and sensor data. Possible edge detection, corner detection, texture analysis and object recognition, etc.
[0130] 3. Deep learning module: Uses multi-layer neural networks to learn features and recognize patterns to understand complex spatial relationships and user behaviors.
[0131] 4. Decision-making module: Generates navigation decisions and path planning in real time based on learned knowledge. It involves path search algorithms, Dijkstra algorithms, and machine learning models to predict traffic conditions and crowd density.
[0132] 5. Optimize learning modules: Continuously optimize model performance through continuous online learning and offline training. This may include using reinforcement learning algorithms to adjust the model based on user feedback and system performance.
[0133] Environmental data collection and processing
[0134] In order to achieve accurate 3D reconstruction and navigation, the system needs to obtain detailed data of the airport environment in real time. This includes but is not limited to:
[0135] 1. High-definition cameras: deployed at key locations to capture high-resolution real-time images. These cameras need to have a wide dynamic range and good low-light performance to adapt to various lighting conditions within the airport.
[0136] 2. Sensor network: Get permission to access the phone's gyroscope information, camera information, Wi-Fi, LiDAR, and Wi-Fi / Bluetooth beacons. LiDAR can provide accurate distance measurement.
[0137] 3. Data synchronization: Ensure that timestamps of all data sources are synchronized to achieve accurate data fusion.
[0138] 3D scene reconstruction
[0139] 3D scene reconstruction is the basis for AR navigation. The system uses the following steps to reconstruct:
[0140] 1. Preset model: Preset the 3D model of the airport to provide a basis and reference for real-time modeling and optimize camera blind spots.
[0141] 2. Initial mapping: Using SLAM (Simultaneous Localization and Mapping) technology, a 3D model of the airport is constructed in real time from camera data. The SLAM algorithm needs to be able to handle changes in a dynamic environment, such as moving pedestrians and vehicles.
[0142] 3. Feature matching: Deep learning models are used to identify and match significant features inside the airport, such as boarding gates, shops, and signboards. These features will serve as important reference points during the navigation process.
[0143] 4. Fine Modeling: Combined with LiDAR data, fine modeling of airport structures is performed to provide centimeter-level accuracy. LiDAR data can provide high-precision 3D point clouds for generating detailed airport maps.
[0144] 5. Dynamic Update: Update the 3D model in real time to reflect changes in the airport environment, such as crowd movement and temporary roadblocks. This involves using an event-driven update mechanism to update the model only when significant changes are detected.
[0145] Visual positioning and sensor data fusion
[0146] Accurate user positioning is crucial for AR navigation. The system uses the following technologies to achieve this goal:
[0147] 1. Visual positioning: Identify the user’s location using a deep learning model by analyzing images captured by the camera.
[0148] 2. Sensor fusion: Combine the data information collected from the user side to improve positioning accuracy and robustness. The sensor fusion algorithm needs to be able to handle the uncertainty and errors of various sensor data.
[0149] 3. Multi-source data calibration: Ensure the consistency and accuracy of data from different sensors through calibration algorithms. Involves using calibration objects or known reference points to adjust sensor outputs.
[0150] 4. Real-time trajectory prediction: predict the user's movement trajectory to adjust the navigation strategy in advance. This involves using the time series prediction model, the long short-term memory network (LSTM).
[0151] User behavior analysis and personalized service generation
[0152] To provide a personalized navigation experience, the system needs to understand each user’s behavior and preferences:
[0153] 1. Behavior analysis: Analyze the user's arrival time, ticket processing progress, etc., and analyze the user's next possible behavior
[0154] 2. Historical behavior analysis: Analyze the user's historical behavior records, such as how they check in, whether they need charging services, etc.
[0155] 3. Context-aware: Considers the user’s current context, such as time, location, and itinerary, to generate the most appropriate navigation path.
[0156] 4. Personalized content generation: Use big models to generate personalized navigation instructions and related information, such as store discounts and gate change notifications.
[0157] AR navigation interface rendering
[0158] The AR navigation interface is a window for users to interact with the system:
[0159] 1. Real-time rendering: Render the 3D model and navigation information into the user's field of view in real time. Use a high-performance graphics rendering engine and optimized rendering algorithms to ensure a smooth user experience.
[0160] 2. Visual guidance: Use arrows, highlights, and animation effects to guide users to their destination. Visual guidance design needs to consider the user’s attention and cognitive load.
[0161] 3. Information overlay: Overlay useful information in the user’s field of view, such as boarding time, route distance, and direction instructions. Information overlay needs to be designed to be clear and not interfere with the user’s field of view.
[0162] 4. User interaction: Allow users to interact with the AR interface through gestures, voice, or touch to get more information or change settings. User interaction design needs to be simple and intuitive to reduce the user's learning curve
[0163] Real-time interaction and intelligent decision support
[0164] The system's interactive and decision support modules provide real-time assistance to users:
[0165] 1. Natural language processing: Use NLP technology to parse the user's voice or text instructions and provide corresponding feedback. This involves the use of speech recognition and natural language understanding (NLU) technology.
[0166] 2. Intelligent recommendation: Intelligently recommend the best navigation path and airport services based on the user's current location and behavior. The recommendation system uses collaborative filtering or content recommendation algorithms.
[0167] 3. Context-aware decision making: Considering the user’s context and environmental factors to provide the most appropriate decision support. This involves using context-aware reasoning algorithms to understand the user’s needs and constraints, such as the user’s gender.
[0168] 4. Multi-turn dialogue management: Through multi-turn dialogue, the system can understand complex user needs and provide detailed navigation assistance. The dialogue management system needs to be able to process contextual information and user intent.
[0169] Continuous learning and optimization of large models
[0170] Continuous learning and optimization of the system ensures long-term performance improvement:
[0171] 1. Online learning: When users interact with the system, the model is updated in real time to adapt to changes in user behavior.
[0172] 2. Offline training: Regularly train the model using large amounts of collected data to improve accuracy and efficiency.
[0173] 3. Performance monitoring: Monitor the performance of the model to detect and solve problems in a timely manner, such as overfitting and underfitting.
[0174] 4. Feedback loop: Establish a user feedback mechanism and use user satisfaction and suggestions as an important reference for optimizing the model.
[0175] Data security and privacy protection
[0176] When processing user data, the system strictly abides by the regulations on data security and privacy protection:
[0177] 1. Data encryption: Encrypt all sensitive data to prevent unauthorized access.
[0178] 2. Anonymization: When analyzing user data, anonymization technology is used to protect user identity.
[0179] 3. Access control: Implement strict access control policies to ensure that only authorized personnel can access data.
[0180] 4. Compliance Checks: Regular compliance checks are conducted to ensure that the system complies with all relevant data protection regulations.
[0181] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.
Claims
1. A method for realizing airport AR navigation in the civil aviation industry based on a large model, characterized by: The method comprises the following steps: Design large model architecture for processing and analyzing airport environmental data; Environmental data collection and processing, used to collect images and sensor data inside the airport in real time; 3D scene reconstruction, used to build a high-precision 3D model of the airport interior; Visual positioning and sensor data fusion are used to achieve accurate user positioning; User behavior analysis, used to generate personalized navigation paths and service recommendations; AR navigation interface rendering, used to overlay navigation information onto the user’s actual field of view; Real-time interaction and intelligent decision support, used to provide real-time interaction and decision support; Feedback mechanism, used to enable large models to continuously learn and optimize based on user feedback.
2. According to claim 1, a method for realizing airport AR navigation in the civil aviation industry based on a large model is characterized by: The specific operations for designing a large model architecture include: Data preprocessing, which is responsible for cleaning, normalizing, and enhancing input data to improve the accuracy and robustness of the model, including noise removal, missing value filling, standardization, and normalization operations; Feature extraction, using computer vision techniques to extract key features from 3D models and sensor data to achieve edge detection, corner detection, texture analysis, and object recognition; Deep learning, which uses multi-layer neural networks to learn features and recognize patterns to understand complex spatial relationships and user behaviors; Decision making, based on the learned knowledge, generates navigation decisions and path planning in real time, involving path search algorithms, Dijkstra algorithm, and machine learning models to predict traffic conditions and crowd density; Optimize learning, continuously optimize model performance through continuous online learning and offline training, including using reinforcement learning algorithms to adjust the model based on user feedback and system performance.
3. According to claim 1, a method for realizing airport AR navigation in the civil aviation industry based on a large model is characterized by: The specific operations of environmental data collection and processing include: Deploy high-definition cameras at designated locations to capture high-resolution real-time images. The cameras have wide dynamic range and good low-light performance to adapt to various lighting conditions within the airport; Use sensor networks to obtain information permissions for the phone's gyroscope, camera, Wi-Fi, LiDAR, and Wi-Fi / Bluetooth beacons; Data synchronization ensures that timestamps of all data sources are synchronized to achieve accurate data fusion.
4. According to claim 1, a method for realizing airport AR navigation in the civil aviation industry based on a large model is characterized by: The specific operations of 3D scene reconstruction include: Preset model: Preset the 3D model of the airport to provide a basis and reference for real-time modeling and optimize the blind spots of the camera; Initial mapping: Using SLAM (Simultaneous Localization and Mapping) technology, a 3D model of the airport is constructed in real time from camera data. The SLAM algorithm handles changes in the dynamic environment, such as moving pedestrians and vehicles. Feature matching: Deep learning models are used to identify and match significant features within the airport, such as gates, shops, and signage, which serve as important reference points during navigation. Fine Modeling: In combination with LiDAR data, fine modeling of airport structures is performed to provide centimeter-level accuracy. LiDAR data provides high-precision 3D point clouds for generating detailed airport maps. Dynamic updating: Updating the 3D model in real time to reflect changes in the airport environment, such as crowd movements and temporary roadblocks, involves using an event-driven update mechanism to update the model only when significant changes are detected.
5. According to claim 1, a method for realizing airport AR navigation in the civil aviation industry based on a large model is characterized by: The specific operations of visual positioning and sensor data fusion include: visual positioning, sensor fusion, multi-source data calibration and real-time trajectory prediction; The specific operations of user behavior analysis include: behavior analysis, historical behavior analysis, context awareness, and personalized content generation; The specific operations of AR navigation interface rendering include: real-time rendering, visual guidance, information overlay, and user interaction; Specific operations of real-time interaction and intelligent decision support include: natural language processing, intelligent recommendation, context-aware decision-making, and multi-round dialogue management.
6. An AR navigation system for an airport in the civil aviation industry based on a large model according to the method for realizing AR navigation for an airport in the civil aviation industry based on a large model according to any one of claims 1 to 5, characterized in that: The system comprises: Large model architecture for processing and analyzing airport environmental data; Data acquisition module, used to collect images and sensor data inside the airport in real time; 3D scene reconstruction module, used to build a high-precision 3D model of the airport interior; Visual positioning and sensor data fusion module, used to achieve accurate user positioning; User behavior analysis module, used to generate personalized navigation paths and service recommendations; AR navigation interface rendering module, used to overlay navigation information into the user's actual field of view; Real-time interaction and intelligent decision support module, used to provide real-time interaction and decision support; Feedback mechanism, used to enable large models to continuously learn and optimize based on user feedback.
7. According to claim 6, a large model-based AR navigation system for civil aviation airports is characterized by: The specific operations of large model architecture design include: Data preprocessing, which is responsible for cleaning, normalizing, and enhancing input data to improve the accuracy and robustness of the model, including noise removal, missing value filling, standardization, and normalization operations; Feature extraction, using computer vision techniques to extract key features from 3D models and sensor data to achieve edge detection, corner detection, texture analysis, and object recognition; Deep learning, which uses multi-layer neural networks to learn features and recognize patterns to understand complex spatial relationships and user behaviors; Decision making, based on the learned knowledge, generates navigation decisions and path planning in real time, involving path search algorithms, Dijkstra algorithm, and machine learning models to predict traffic conditions and crowd density; Optimize learning, continuously optimize model performance through continuous online learning and offline training, including using reinforcement learning algorithms to adjust the model based on user feedback and system performance.
8. According to claim 6, a large model-based AR navigation system for civil aviation airports is characterized by: Data acquisition module, which deploys high-definition cameras at designated locations to capture high-resolution real-time images. The cameras have wide dynamic range and good low-light performance to adapt to various lighting conditions in the airport; Use sensor networks to obtain information permissions for the phone's gyroscope, camera, Wi-Fi, LiDAR, and Wi-Fi / Bluetooth beacons; Data synchronization ensures that timestamps of all data sources are synchronized to achieve accurate data fusion.
9. The large-scale model-based AR navigation system for civil aviation airports according to claim 6 is characterized by: The 3D scene reconstruction module includes: Preset model: Preset the 3D model of the airport to provide a basis and reference for real-time modeling and optimize the blind spots of the camera; Initial mapping: Using SLAM (Simultaneous Localization and Mapping) technology, a 3D model of the airport is constructed in real time from camera data. The SLAM algorithm handles changes in the dynamic environment, such as moving pedestrians and vehicles. Feature matching: Deep learning models are used to identify and match significant features within the airport, such as gates, shops, and signage, which serve as important reference points during navigation. Fine Modeling: In combination with LiDAR data, fine modeling of airport structures is performed to provide centimeter-level accuracy. LiDAR data provides high-precision 3D point clouds for generating detailed airport maps. Dynamic updating: Updating the 3D model in real time to reflect changes in the airport environment, such as crowd movements and temporary roadblocks, involves using an event-driven update mechanism to update the model only when significant changes are detected.
10. The large model-based AR navigation system for civil aviation airports according to claim 6, characterized in that: The visual positioning and sensor data fusion module includes: visual positioning, sensor fusion, multi-source data calibration and real-time trajectory prediction; The user behavior analysis module includes: behavior analysis, historical behavior analysis, context awareness, and personalized content generation; The AR navigation interface rendering module includes: real-time rendering, visual guidance, information overlay and user interaction; The real-time interaction and intelligent decision support modules include: natural language processing, intelligent recommendation, context-aware decision-making, and multi-round dialogue management.
Citation Information
Cited By
Outdoor immersive text and travel hybrid positioning and dynamic rendering optimization method
CN120576758A
