Intelligent environment response type park scenic spot touring system and method and edge computing architecture

Abstract

The invention relates to the technical field of computer technology and garden management, and discloses an intelligent environment response type park scenic spot sightseeing system and method and an edge computing architecture.The system comprises a distributed environment monitoring network, an edge computing server and an AR visual interaction interface, the edge computing server integrates multi-modal data through a data fusion module, and the AR visual interaction interface is connected with the edge computing server; and dynamically planning and generating an optimal touring route which optimizes the environmental comfort, the path distance and the tourist congestion degree at the same time by using a reinforcement learning algorithm based on a multi-dimensional reward function, and carrying out real-time touring on the optimized route information and real-time environmental data by using an AR visual interaction interface. And the function of adaptively adjusting the display information according to the sight focus of the user is realized. According to the invention, through deep integration of environment, position and congestion degree data, dynamic adaptive adjustment of a touring route is realized, scientificity and real-time performance of navigation are effectively improved, and more intelligent, comfortable and efficient garden touring service is provided for tourists.

Description

Technical Field

[0001] This invention relates to the fields of computer technology and garden management technology, and in particular to intelligent environmental responsive park and scenic area tour systems, methods and edge computing architectures. Background Technology

[0002] With the advancement of technology, traditional garden tour guide methods are gradually evolving towards digitalization and intelligence. However, most current intelligent tour guide systems still have significant limitations. Existing technologies typically rely on preset fixed tour routes or static maps, and their core flaw lies in their inability to dynamically adjust to complex, real-time environments. When environmental parameters within the garden, such as air quality, temperature, and humidity, change, or when overcrowding occurs in specific areas, these systems cannot respond promptly and provide visitors with adaptive optimization suggestions.

[0003] Most existing tour guide systems employ traditional path planning algorithms, focusing solely on finding the shortest path and neglecting the actual comfort and experience needs of tourists. This one-dimensional approach results in impersonal tour routes, potentially leading tourists to areas with poor air quality, uncomfortable temperatures, or high crowds, significantly diminishing the overall experience. Furthermore, current tour guide systems also suffer from deficiencies in information presentation and interaction. They often present information in two-dimensional graphics or simple audio, lacking immersion and intuitiveness. Users must frequently switch their gaze between the real environment and their handheld devices, which not only distracts tourists from the scenery but also reduces information acquisition efficiency, failing to achieve truly natural, integrated human-computer interaction.

[0004] Therefore, there is an urgent need for an intelligent tour guide technology that can deeply integrate various real-time data, dynamically adapt to environmental changes, plan a comprehensive and optimal tour route, and present information in an immersive and adaptive manner. Summary of the Invention

[0005] The purpose of this invention is to provide an intelligent environment-responsive park and scenic area tour system, method, and edge computing architecture, which solves the problems of single path, information lag, and poor interactive experience in existing garden tours.

[0006] To achieve the above objectives, the present invention provides the following technical solution:

[0007] The first aspect of this invention provides an intelligent environmentally responsive park and scenic area tour system.

[0008] The system comprises a distributed environmental monitoring network, an edge computing server, and an AR visual interactive interface. The distributed environmental monitoring network includes multiple fixed monitoring base stations and multiple portable detection terminals. The fixed monitoring base stations are deployed at preset physical intervals within the park area to collect at least one environmental parameter for path planning in real time. The portable detection terminals collect visitor location information and communicate with the fixed monitoring base stations or the edge computing server. The edge computing server, as the core component of the edge computing architecture, is deployed within the park area and includes a data fusion module and a path planning engine. The data fusion module receives and processes heterogeneous data collected from the distributed environmental monitoring network. The path planning engine plans paths based on the processed data. The AR visual interactive interface runs on the portable detection terminals, overlaying environmental information and path guidance onto a real-world scene and allowing interaction with visitors.

[0009] Preferably, the environmental parameters used for path planning that the fixed monitoring base station collects in real time include negative oxygen ion concentration. ,temperature ,humidity and / or light intensity The portable detection terminal synchronizes data with the nearest fixed monitoring base station or the edge computing server via Bluetooth communication protocol and transmits tourist location information. To estimate regional tourist density This density can be calculated using the kernel density estimation method:

[0010] ;

[0011] in, Indicates the first In class data, for data points The probability density at that location is estimated. Normalization factor, yes The total number of tourists in this category of data. It's bandwidth. It's a kernel function. The point to be estimated was calculated. With the Data points The distance between them.

[0012] In one specific embodiment, the data fusion module of the edge computing server employs time series analysis algorithms to process the heterogeneous data in real time, including data cleaning, anomaly detection, missing value imputation, and data fusion, to generate a comprehensive environmental dataset for path planning. The time series analysis algorithms include an Autoregressive Moving Average (ARIMA) model and / or a Seasonally Decomposed Moving Average (STL) prediction algorithm. For missing sensor data... The general form of the ARIMA model is:

[0013] ;

[0014] in, It is in time Observed time series data, Fractional difference operator, It is a shift operator. It is the difference order. and These are the polynomials for the autoregressive and moving average components, respectively. .

[0015] Preferably, the edge computing server further includes a robust management module, which is used to initiate a degradation strategy of data interpolation, neighboring data reference, or historical data backtracking when some sensor data is abnormal or network communication is interrupted, so as to ensure the continuity of system data input.

[0016] In one specific embodiment, the path planning engine is implemented based on a reinforcement learning algorithm. The reinforcement learning algorithm determines the tourist's current location. The environmental parameters of the location and surrounding area and tourist density As status input It will move to an adjacent position. As an action And through a multi-dimensional reward function To conduct learning and decision-making.

[0017] Preferably, the reinforcement learning algorithm is the Q-learning algorithm. The multi-dimensional reward function... Taking into account environmental characteristics, tourist congestion, and route length, when the new state When the environmental parameters at a given location meet a preset optimal threshold, the reward is significantly increased. Specifically, when transitioning from state... Take action Reaching a new state Instant rewards received at that time Defined as:

[0018] ;

[0019] in, It's an environmental reward item, when the new state... Environmental parameter vector When the preset preferred threshold is met (for example, the concentration of negative oxygen ions is higher than a certain threshold, and the temperature and humidity are within a specific range), this item is positive; otherwise, it is zero or negative. It is a congestion-related reward item; It is the path cost item; These are the weighting coefficients for each reward, used to adjust the priority of different optimization objectives.

[0020] The Q-learninq algorithm updates the Q value according to the following rules:

[0021] ;

[0022] in, Indicates time At that time, the agent is in a state And take action The obtained Q value, on the left side This is the updated new value, while the one on the right... It is the old value before the update. This indicates an assignment operation. This is a learning rate parameter. Indicates time Take action Later, in time Instant rewards received This is a discount factor. Indicates the next state Choose the action that yields the maximum Q value. .

[0023] In one specific embodiment, the path planning engine periodically recalculates the scores of each potential path using the multi-dimensional reward function based on the latest environmental data and tourist density data, and dynamically adjusts the recommended route. Preferably, the AR visualization interactive interface combines simultaneous localization and mapping (SLAM) technology and integrates an eye-tracking module. The AR visualization interactive interface uses the tourist's gaze focus area analyzed by the eye-tracking module. Adaptively adjust information elements on the screen Display density (visibility) To avoid information obscuring the main landscape view. For example, visibility. It can be calculated in the following ways:

[0024] ;

[0025] in, Represent a Output values ​​or weights of class data, It is a maximum value or amplitude. Represented by natural constant An exponential function with base 0. It is a vector or coordinate. It is another vector or coordinate. This indicates that the person in charge is responsible for converting this distance into a decay factor. Representing vectors and The square of the Euclidean distance between them The standard deviation of the Gaussian function is represented by... It is variance.

[0026] A second aspect of the present invention provides an edge computing architecture.

[0027] This edge computing architecture includes one or more edge computing servers deployed within a specific physical area. Each edge computing server includes a data fusion module and a path planning engine. The data fusion module is configured to receive and process heterogeneous data in real time from a distributed environmental monitoring network deployed within the specific physical area. This heterogeneous data includes at least one environmental parameter and user location information. The path planning engine is connected to the data fusion module and is configured to dynamically generate one or more path planning decisions based on the environmental parameters and user location information processed by the data fusion module, using a reinforcement learning algorithm.

[0028] A third aspect of this invention provides a smart environment-responsive park / scenic area tour method. The method includes the following steps:

[0029] S1. Collect at least one environmental parameter and tourist location information for route planning in real time through a distributed environmental monitoring network;

[0030] S2. Utilize the data fusion module of the edge computing server to process the collected heterogeneous data in real time, including data cleaning, anomaly detection, missing value imputation, and data fusion;

[0031] S3. Utilize the path planning engine of the edge computing server, based on reinforcement learning algorithms, to dynamically plan and optimize the tour route according to the processed environmental parameters and tourist density;

[0032] S4. Through an AR visual interactive interface, environmental information and the optimized tour route are overlaid on the real-world scene, and the information display density is adaptively adjusted according to the tourist's focus area.

[0033] In summary, the present invention has at least one of the following beneficial technical effects:

[0034] 1. This invention collects environmental parameters and visitor density in real time by setting up a distributed environmental monitoring network, and uses a path planning engine in an edge computing server deployed in the park. Based on reinforcement learning algorithm, it uses real-time data as state input and makes decisions through a multi-dimensional reward function. This technical solution enables the system to dynamically generate tour routes that take into account environmental optimization, low congestion and path efficiency according to instantaneous changes in the environment and the distribution of people, breaking through the limitation of traditional tour guide systems that can only provide fixed or shortest paths.

[0035] 2. This invention integrates the data fusion module and the path planning engine on an edge computing server deployed within the park. This architecture enables heterogeneous data collected from the distributed environmental monitoring network to be processed and analyzed in real time locally, without the need for long-distance transmission to the cloud data center. As a result, the system avoids the problem of data obsolescence caused by network bandwidth limitations and transmission delays, ensuring that path planning decisions are based on the latest on-site conditions and improving the real-time response capability of the entire tour guide system.

[0036] 3. This invention integrates an eye-tracking module into the AR visual interactive interface. This module can analyze the area where tourists' eyes are focused and adaptively adjust the display density of information elements such as navigation guidance and environmental data based on the analysis results. When tourists focus their eyes on distant landscapes, the visibility of non-core information on the screen will be automatically reduced, thereby avoiding excessive obstruction of real landscapes by virtual information and ensuring that tourists can have a complete and undisturbed visual tour process while obtaining necessary guidance information. Attached Figure Description

[0037] Figure 1 This is a schematic diagram of the overall system architecture of the present invention;

[0038] Figure 2 This is a flowchart of the intelligent environmental responsive park and scenic area tour method of the present invention;

[0039] Figure 3 This is a schematic diagram of the distributed environmental monitoring network deployment of the present invention;

[0040] Figure 4 This is a functional block diagram of the edge computing server of the present invention;

[0041] Figure 5 This is a hardware structure block diagram of the fixed monitoring base station of the present invention. Detailed Implementation

[0042] See attached document Figure 1 , Figure 1This is a schematic diagram of the overall system architecture of the present invention. The present invention provides an intelligent environment-responsive park and scenic area tour system, which may include:

[0043] A distributed environmental monitoring network. The distributed environmental monitoring network includes multiple fixed monitoring base stations and multiple portable detection terminals. The fixed monitoring base stations are deployed at preset physical intervals within the park / tourist area to collect at least one environmental parameter in real time for path planning. The portable detection terminals are used to collect visitor location information and communicate with the fixed monitoring base stations or the edge computing server.

[0044] An edge computing server. This edge computing server, as a core component of the edge computing architecture, is deployed within the garden / tourist area. The edge computing server includes a data fusion module and a path planning engine. The data fusion module receives and processes heterogeneous data collected from the distributed environmental monitoring network. The path planning engine performs path planning based on the processed data.

[0045] An AR (Augmented Reality) visual interactive interface. This interface runs on the portable detection terminal and is used to overlay environmental information and route guidance onto the real-world scene, allowing for interaction with visitors.

[0046] In this system architecture, the distributed environmental monitoring network is responsible for front-end data acquisition. Fixed monitoring base stations are deployed within the park to continuously monitor environmental parameters in designated areas. Portable detection terminals move with visitors, acquiring their real-time location information. This data is transmitted to the edge computing server via wired or wireless connections.

[0047] See attached document Figure 4 , Figure 4 This is a functional block diagram of the edge computing server of the present invention. The edge computing server receives raw data streams from the distributed environmental monitoring network, and its data fusion module preprocesses and integrates them. The processed integrated environmental dataset is then invoked by its path planning engine to generate a dynamically optimized tour route. The route planning result is transmitted to the portable detection terminal via the network.

[0048] The AR visualization interface runs on the portable detection terminal, receiving path planning data and environmental information transmitted from the edge computing server. This virtual information is overlaid in real-time onto the real-world scene captured by the terminal device's camera, providing tourists with intuitive navigation guidance and environmental information. The interface also accepts input from tourists, enabling human-computer interaction.

[0049] The distributed environmental monitoring network described in this invention is designed to achieve real-time, distributed acquisition of multi-dimensional data in a garden environment. The network includes multiple fixed monitoring base stations and multiple portable detection terminals.

[0050] See attached document Figure 5 , Figure 5 This is a hardware structure block diagram of the fixed monitoring base station of the present invention. The fixed monitoring base station is an independent physical device deployed in a garden or tourist area. Its deployment follows a preset physical spacing to ensure sufficient coverage of environmental parameters in the target area. The preset physical spacing can be determined based on the terrain complexity, vegetation density, and spatial variation gradient of environmental parameters (such as negative oxygen ion concentration) of the garden. For example, in areas with flat terrain and uniform vegetation, the base station spacing can be set to 50 meters; while at key nodes with large terrain undulations or drastic changes in environmental parameters (such as near waterfalls or dense forests), the spacing can be shortened to 20 meters to achieve higher resolution data acquisition. In addition to collecting the aforementioned environmental parameters, in one specific embodiment, the fixed monitoring base station can also integrate sensors for measuring the concentration of PM2.5 and PM10 particulate matter in the air, as well as sensors for measuring environmental noise in decibels, to provide more comprehensive environmental quality data.

[0051] Each fixed monitoring base station typically integrates a main control unit, a power module (e.g., a lithium battery pack with solar charging capability), a communication module, and multiple sensors. The integrated sensors are used to collect environmental parameters in real time for path planning; for example, a negative oxygen ion concentration sensor measures the concentration of negative oxygen ions. Temperature sensors are used to measure ambient temperature. A humidity sensor is used to measure relative humidity. and a light intensity sensor for measuring light intensity. Each base station collects data at a preset sampling frequency (e.g., every 30 seconds) and packages the collected environmental parameter data.

[0052] The portable detection terminals are typically smart mobile devices (such as smartphones) carried by tourists or dedicated augmented reality glasses. Each portable detection terminal has a built-in Global Navigation Satellite System (GNSS) module (e.g., GPS, BeiDou) to obtain its own geographical coordinates. In addition, the portable detection terminal is equipped with a Bluetooth Low Energy (BLE) communication module for short-range data synchronization with the nearest fixed monitoring base station or for direct data transmission with an edge computing server. The portable detection terminal will collect its own location information. The data is periodically transmitted to an edge computing server. By aggregating location information reported by multiple portable detection terminals, the edge computing server can estimate the visitor density within a specified area in real time. For example, using kernel density estimation methods, the area density can be calculated. Tourist density within

[0053] ;

[0054] in, Indicates the first In class data, for data points The probability density at that location is estimated. Normalization factor, yes The total number of tourists in this category of data. It's bandwidth. It's a kernel function. The point to be estimated was calculated. With the Data points The distance between them.

[0055] The data fusion module included in the edge computing server is designed to receive and process heterogeneous data collected from a distributed environmental monitoring network in real time. This heterogeneous data includes environmental parameters from fixed monitoring base stations (such as negative oxygen ion concentration, temperature, humidity, and light intensity) and visitor location information from portable detection terminals. The data fusion module transforms the raw, scattered data into a structured, continuous, and reliable comprehensive environmental dataset.

[0056] The data fusion module first preprocesses the received raw heterogeneous data. This preprocessing includes data cleaning, anomaly detection, and missing value imputation. The data cleaning stage identifies and removes redundant data, non-standard format data, or values ​​exceeding physically reasonable ranges caused by sensor malfunctions or transmission errors. The anomaly detection stage identifies anomalous data points that significantly deviate from the normal pattern by analyzing the time-series characteristics and statistical distribution of the data. For example, judgments can be made based on statistical thresholds or the degree of deviation from the time-series trend.

[0057] Specifically, the anomaly detection can be achieved by using... The criterion is that a data point is identified as an anomaly when it deviates from the mean of the data within its time window by more than three standard deviations. For spatial anomaly detection, the density-based spatial clustering noise application (DBSCAN) algorithm can also be used to identify spatially isolated monitoring data points.

[0058] For missing data caused by temporary sensor outages or unstable communication, the data fusion module uses time series analysis algorithms for imputation. These time series analysis algorithms include an Autoregressive Moving Average (ARIMA) model and / or a Seasonally Decomposed Moving Average (STL) forecasting algorithm. Taking the ARIMA model as an example, its general form is used for forecasting and imputing missing values ​​in time series data:

[0059] ;

[0060] in, It is in time Observed time series data, Fractional difference operator, It is a shift operator. It is the difference order. and These are the polynomials for the autoregressive and moving average components, respectively. .

[0061] After preprocessing, the data fusion module will clean and interpolate various environmental parameters (such as negative oxygen ion concentration). ,temperature ,humidity Light intensity ) and the calculated regional tourist density The data is then integrated. This process associates these heterogeneous data sets with specific geographic grids and timestamps, constructing a unified, multi-dimensional, comprehensive environmental dataset. This dataset is stored in a structured format, such as a matrix or database record containing geographic coordinates, timestamps, and various environmental parameter values, for use and analysis by the route planning engine.

[0062] The path planning engine included in the edge computing server functions by dynamically generating optimized tour routes for tourists based on a comprehensive environmental dataset provided by the data fusion module and utilizing reinforcement learning algorithms. This path planning engine first models the garden tour guidance problem as a Markov Decision Process (MDP). This process is defined by a state space, an action space, and a reward function. State space... This describes the environmental state of the tourists, a specific state. Represented by a tuple ,in This represents the current geographical location of the visitor (e.g., a node in a garden map model). It is a vector that contains various environmental parameters of the location and its surrounding area. This refers to the tourist density in the area. (Action space) It is the set of movement actions that tourists can take, such as moving from the current location node. Move to any adjacent node Reward function Defined in state Take action below The system then receives an immediate reward. This function is a multi-dimensional function that comprehensively considers environmental characteristics, tourist congestion, and path length. When transitioning from state... Take action Reaching a new state Instant rewards received at that time Defined as:

[0063] ;

[0064] in, It's an environmental reward item, when the new state... Environmental parameter vector When the preset preferred threshold is met (for example, the concentration of negative oxygen ions is higher than a certain threshold, and the temperature and humidity are within a specific range), this item is positive; otherwise, it is zero or negative. It is a congestion-related reward item; It is the path cost item; These are the weighting coefficients for each reward, used to adjust the priority of different optimization objectives.

[0065] In this embodiment, the path planning engine uses the Q-learning algorithm to solve the Markov decision process. It should be understood that the Q-learning algorithm is only a preferred embodiment of the present invention. In other embodiments, other reinforcement learning algorithms can also be used, such as the State-Action-Reward-State-Action (SARSA) algorithm, or when the state space and action space are very large, deep reinforcement learning algorithms, such as Deep Q-Network (DQN), can be used, where the action value function Q(s,a) is approximated by a deep neural network to handle high-dimensional input states. The Q-learning algorithm learns an action value function... To assess the state Take action below The function learns through iterative updates, with the following update rules:

[0066] ;

[0067] in, Indicates time At that time, the agent is in a state And take action The obtained Q value, on the left side This is the updated new value, while the one on the right... It is the old value before the update. This indicates an assignment operation. This is a learning rate parameter. Indicates time Take action Later, in time Instant rewards received This is a discount factor. Indicates the next state Choose the action that yields the maximum Q value. .

[0068] To enable dynamic route adjustments, the route planning engine is configured to execute periodically. For example, the system can set a time period (e.g., 60 seconds) or a data change threshold. When a specified time is reached, or when changes in environmental data or visitor density reported by the data fusion module exceed this threshold, the route planning engine uses the latest comprehensive environmental dataset to recalculate the entire route from the visitor's current location. The calculation results are updated and pushed to the visitor's AR visualization interface, thus enabling real-time adjustments to the recommended route to adapt to dynamic changes in the environment and visitor flow.

[0069] See attached document Figure 3 , Figure 3 This diagram illustrates the deployment of the distributed environmental monitoring network of the present invention. The edge computing server further includes a robust management module, which is responsible for continuously monitoring the data input status and network communication link stability of the distributed environmental monitoring network during system operation. Its main function is to initiate corresponding degradation strategies when an anomaly is detected to ensure the continuity of system data input, thereby maintaining the normal operation of the path planning service.

[0070] The robustness management module detects anomalies by setting data validity rules and communication timeout mechanisms. When environmental parameter data or location information data received from a fixed monitoring base station or portable monitoring terminal experiences drastic fluctuations exceeding preset thresholds within a short period, remains stagnant for an extended period, or experiences a complete interruption of the data stream, the module classifies it as data anomaly. Simultaneously, when the network communication link with a specific base station or terminal fails to receive a heartbeat packet or data response within a specified time, the module classifies it as a network communication interruption.

[0071] To address communication interruptions caused by tourist movement or environmental interference, the robustness management module specifically performs the following functions:

[0072] Connection status monitoring and interruption determination: The module monitors the communication link quality between the portable testing terminal and the base station in real time through a heartbeat mechanism (Keep-alive). For example, if N heartbeat data packets (N can be set to 3-5 according to actual needs) are lost consecutively or the signal strength (RSSI) is lower than a preset threshold (such as -90dBm), the current connection is determined to be interrupted.

[0073] Intelligent handover and reconnection mechanism: Once a connection interruption is detected, the robust management module on the portable detection terminal will immediately trigger a channel scanning procedure to actively search for and connect to other nearby fixed monitoring base stations with the best signal strength, thereby achieving seamless network handover and ensuring the continuity of data flow.

[0074] Data caching and breakpoint resume: In extreme cases such as handover or temporary complete loss of connection due to signal blind spots, the module on the portable detection terminal will activate a data caching mechanism to temporarily store the collected location information and sensor data in local storage. Once a stable connection is re-established with any base station, the module will immediately package and upload the cached data to the edge computing server in chronological order, ensuring data integrity and preventing information loss.

[0075] When a short-term loss of some sensor data or an anomaly in a single data point is detected, the robustness management module can activate a data interpolation strategy. This strategy can utilize the time-series analysis capabilities within the data fusion module, such as using an ARIMA model to predictively fill in missing data points, or employing mathematical methods like linear interpolation or spline interpolation to estimate the missing value based on adjacent valid data points. This method is suitable for situations involving transient and localized incomplete sensor data.

[0076] If the data from a fixed monitoring base station is continuously abnormal or the duration of communication interruption with that base station exceeds a preset threshold, the robustness management module will activate a neighboring data reference strategy. Under this strategy, the system no longer directly relies on the data from the abnormal base station, but instead collects and analyzes environmental parameters collected by other geographically adjacent and functioning fixed monitoring base stations. The environmental parameters of the abnormal area can be estimated by spatially weighted averaging the data from these neighboring base stations, where the weights can be inversely proportional to the distance from the center of the abnormal area.

[0077] In extreme cases, such as when sensor data for a certain area is unavailable for an extended period and there is insufficient or unreliable data from neighboring base stations for reference, the robustness management module will initiate a historical data backtracking strategy. This strategy retrieves historical environmental datasets stored in the edge computing server, identifies and extracts average environmental parameter patterns for the anomalous area over the same time period in the past (e.g., the same hour on a day of the week, or similar seasonal periods), and uses these patterns as temporary substitute data for the current moment. This strategy ensures that even in cases of severe data loss, the path planning engine can obtain a relatively reasonable data input based on historical experience, preventing complete service interruption.

[0078] The AR visualization interactive interface is executed on the portable detection terminal. Its function is to integrate the path guidance and environmental information generated by the edge computing server with the real-world scene and present it to tourists in an adaptive manner.

[0079] To achieve accurate overlay and stable attachment of virtual information in real-world scenarios, the AR visualization interface incorporates Simultaneous Localization and Mapping (SLAM) technology. In some open scenarios where positioning accuracy requirements are relatively low, the AR visualization interface can also employ positioning technology based on the fusion of Global Navigation Satellite System (GNSS) and Inertial Measurement Unit (IMU) data as an alternative or supplement to SLAM. By fusing high-frequency IMU attitude data with low-frequency but absolutely accurate GNSS data using algorithms such as Kalman filtering, the pose of the terminal device can also be obtained, enabling the overlay of virtual information. This technology utilizes the camera and inertial measurement unit (IMU) sensors of a portable detection terminal to construct a real-time 3D map of the surrounding physical environment (e.g., in the form of point clouds or grids), and simultaneously calculates the terminal device's six-degree-of-freedom (6-DoF) pose (i.e., 3D spatial position and orientation) within this map. Through SLAM technology, the system can obtain the precise anchor points of virtual information (such as navigation arrows and data panels) in the real-world coordinate system, ensuring that this information can be correctly rendered on the corresponding physical objects (e.g., navigation arrows are close to the ground path).

[0080] The AR visualization interface also integrates an eye-tracking module. This module typically consists of an infrared light source and an infrared camera, deployed at specific locations on the AR glasses or portable device. It calculates the user's gaze point in real time by capturing the reflected light spots from the user's eyes and the position of their pupils.

[0081] Based on data acquired by the eye-tracking module, the AR visualization interface implements an adaptive information display density adjustment mechanism. This mechanism dynamically adjusts the display attributes of various information elements on the screen according to the user's focal area of ​​gaze, to avoid information obstructing the main landscape view. Specifically, for each information element on the screen (e.g., navigation arrows, environmental parameter display boxes, etc.), the system calculates its visibility based on its position and distance from the user's gaze point. For example, information elements... visibility It can be calculated using the following function:

[0082] ;

[0083] in, Represent a Output values ​​or weights of class data, It is a maximum value or amplitude. Represented by natural constant An exponential function with base 0. It is a vector or coordinate. It is another vector or coordinate. Representing vectors and The square of the Euclidean distance between them The standard deviation of the Gaussian function is represented by... It is variance.

[0084] See attached document Figure 2 , Figure 2 The flowchart below illustrates the intelligent environment-responsive park and scenic area tour method of the present invention. The present invention also provides an intelligent environment-responsive park and scenic area tour method, which details the overall workflow of the system described in the present invention, including the following steps:

[0085] S1: Data Acquisition

[0086] After the system is started, the distributed environmental monitoring network deployed within the park begins continuous data collection. Multiple fixed monitoring base stations in the network collect environmental parameters at their locations in real time at a preset sampling frequency (e.g., once every 30 seconds), including negative oxygen ion concentration, temperature, humidity, and light intensity. Simultaneously, portable monitoring terminals carried by visitors use their built-in Global Navigation Satellite System (GNSS) modules to acquire and record their real-time geographical location information. The collected raw environmental parameters and visitor location information are transmitted in real time to edge computing servers deployed within the park via wireless or wired communication links.

[0087] S2: Data Processing

[0088] The data fusion module within the edge computing server receives the heterogeneous data stream from step S1. This module first preprocesses the data, including data cleaning to remove invalid data, anomaly detection to identify data points deviating from the normal range, and imputation using time-series analysis algorithms such as ARIMA when data is missing. Subsequently, the module fuses the preprocessed environmental parameters with visitor location information and calculates the real-time visitor density for each area within the park based on the aggregated location information. The final output of this step is a structured, comprehensive environmental dataset with geographic location labels and timestamps, which provides input for subsequent path planning.

[0089] S3: Path Planning

[0090] When a tourist initiates a guided tour request or needs a route update during the tour, the path planning engine within the edge computing server is triggered. The engine obtains the tourist's current location and accesses the latest comprehensive environmental dataset generated in step S2. Based on a reinforcement learning algorithm (e.g., Q-learning), the engine takes the tourist's current state (including location, environmental parameters, and tourist density) as input, and selects the action that maximizes the multi-dimensional reward function (i.e., moving to the next location node) by querying a pre-trained Q-table or performing real-time calculations. Through continuous decision-making, the engine generates an optimized tour route from the current location to the target point, comprehensively considering environmental optima, low congestion, and path efficiency. This planning process is periodically repeated to respond to real-time changes in environmental data.

[0091] S4: AR Presentation and Interaction

[0092] Optimized tour route data generated by the path planning engine is sent to the tourist's portable detection terminal. The terminal's AR visualization interface receives this data and uses SLAM technology to perceive and locate the surrounding environment in real time. Subsequently, the interface renders path guidance (e.g., virtual arrows or highlighted paths) and relevant environmental information (e.g., negative oxygen ion concentration values ​​in the area ahead) as a virtual overlay on the real-world image captured by the camera. During the tourist's movement, an integrated eye-tracking module continuously monitors the area where the tourist's gaze is focused. Based on the gaze data, the AR visualization interface adaptively adjusts the visibility of information elements in non-focus areas to reduce landscape obstruction while ensuring clear readability when the tourist looks at the information elements. The entire process forms a closed loop, with the system continuously providing dynamically adjusted intelligent tour guidance services based on real-time changes in the environment and the tourist's behavior.

[0093] To avoid interfering with user experience due to frequent changes in the area of ​​focus caused by tourists moving or rapidly scanning their vision, the AR visualization interface 103 of this invention further integrates a dynamic smooth adjustment mechanism. This mechanism ensures the stability and comfort of adaptive display, and its specific logic is as follows:

[0094] A gaze stability determination mechanism is introduced: the system does not immediately trigger display switching the instant the user's gaze changes focus. Instead, a gaze dwell time threshold is introduced. For example, only when the user's gaze lingers on a new area (such as a UI element or a distant landscape) for more than a preset duration (e.g., 300 milliseconds) does the system determine that the user has made a genuine intention to shift focus and prepare to switch the display density. This effectively filters out invalid signals generated by involuntary eye movements or rapid scanning.

[0095] An adjustment cooling-off time threshold has been added: To prevent high-frequency jitter between two critical states, the system sets a minimum adjustment interval. That is, after a display density switch (e.g., from weakened to normal display), the system enters a brief cooling-off or lockout period. During this period, even if a change in gaze focus that meets the conditions is detected, the system will not perform another switch. This ensures sufficient time interval between two consecutive adaptive adjustments, providing the user with a stable visual interface.

[0096] Achieving smooth transition animations: All information density changes are not instantaneous, but achieved through smooth, gradual animations (such as fade-in / fade-out, smooth changes in transparency or size). For example, when an information element needs to return to normal display from a weakened state, it gradually becomes clear and prominent over 200-400 milliseconds, rather than jumping out instantly. This smooth transition greatly improves visual comfort, making state transitions natural and uninterrupted for the user.

Claims

1. An intelligent environmentally responsive park and scenic area tour system, characterized in that: include: A distributed environmental monitoring network includes multiple fixed monitoring base stations and multiple portable detection terminals. The fixed monitoring base stations are deployed at preset physical intervals within the park and are used to collect at least one environmental parameter for path planning in real time. The portable detection terminals are used to collect visitor location information and communicate with the fixed monitoring base stations or edge computing servers. An edge computing server, which serves as the core component of the edge computing architecture, is deployed within the garden visit area. The edge computing server includes a data fusion module and a path planning engine. The data fusion module is used to receive and process heterogeneous data collected from the distributed environmental monitoring network, and the path planning engine is used to plan a path based on the processed data. An AR visualization interactive interface runs on the portable detection terminal to overlay environmental information and route guidance onto the real-world scene and interact with tourists.

2. The intelligent environmental responsive park and scenic area tour system according to claim 1, characterized in that, The fixed monitoring base station is used to collect environmental parameters in real time, including negative oxygen ion concentration, temperature, humidity, and light intensity. The portable detection terminal synchronizes data with the nearest fixed monitoring base station or the edge computing server via Bluetooth communication protocol and transmits tourist location information to estimate the tourist density in the area.

3. The intelligent environmental responsive park and scenic area tour system according to claim 1, characterized in that, The data fusion module uses a time series analysis algorithm to process the heterogeneous data in real time. The real-time processing includes data cleaning, anomaly detection, missing value imputation, and data fusion to generate a comprehensive environmental dataset for path planning. The time series analysis algorithm includes an autoregressive moving average model and a seasonal decomposition moving average prediction algorithm.

4. The intelligent environmental responsive park and scenic area tour system according to claim 1, characterized in that, The edge computing server also includes a robust management module, which is used to initiate degradation strategies such as data interpolation, neighboring data reference, or historical data backtracking when some sensor data is abnormal or network communication is interrupted, in order to ensure the continuity of system data input.

5. The intelligent environmental responsive park and scenic area tour system according to claim 1, characterized in that, The path planning engine is based on a reinforcement learning algorithm. The reinforcement learning algorithm takes the tourist's current location, environmental parameters of the current location and surrounding area, and tourist density as state inputs, takes moving to an adjacent location as an action, and learns and makes decisions through a multi-dimensional reward function.

6. The intelligent environmental responsive park and scenic area tour system according to claim 5, characterized in that, The reinforcement learning algorithm is the Q-learning algorithm, and the multi-dimensional reward function comprehensively considers environmental comfort, tourist congestion, and path length.

7. The intelligent environmental responsive park and scenic area tour system according to claim 6, characterized in that, The path planning engine periodically recalculates the scores of each potential path using the multi-dimensional reward function based on the latest environmental data and tourist density data, and dynamically adjusts the recommended routes.

8. The intelligent environmental responsive park and scenic area tour system according to claim 6, characterized in that, The AR visualization interactive interface combines synchronous positioning and mapping technology and integrates an eye-tracking module. The AR visualization interactive interface adaptively adjusts the information display density based on the tourist's gaze focus area analyzed by the eye-tracking module, in order to avoid information obscuring the main landscape view.

9. An edge computing architecture, characterized in that, This includes one or more edge computing servers deployed within a specific physical area, the edge computing servers comprising: The data fusion module is configured to receive and process heterogeneous data from a distributed environmental monitoring network deployed within the specific physical area in real time, wherein the heterogeneous data includes at least one environmental parameter and user location information. The path planning engine, connected to the data fusion module, is configured to dynamically generate one or more path planning decisions using a reinforcement learning algorithm based on the environmental parameters and user location information processed by the data fusion module.

10. A smart environment-responsive park and scenic area tour method, employing the smart environment-responsive park and scenic area tour system as described in claims 1-9, characterized in that, Includes the following steps: S1. Collect at least one environmental parameter and tourist location information for route planning in real time through a distributed environmental monitoring network; S2. Utilize the data fusion module of the edge computing server to process the collected heterogeneous data in real time, including data cleaning, anomaly detection, missing value imputation, and data fusion; S3. Utilize the path planning engine of the edge computing server, based on reinforcement learning algorithms, to dynamically plan and optimize the tour route according to the processed environmental parameters and tourist density; S4. Through an AR visual interactive interface, environmental information and the optimized tour route are overlaid on the real-world scene, and the information display density is adaptively adjusted according to the tourist's focus area.

Images