An exhibition intelligent interaction control method, medium and device

Abstract

This invention discloses a method, medium, and device for intelligent interactive control of exhibitions, belonging to the field of intelligent interactive control technology for exhibitions. The method is applied to a distributed intelligent interactive node network, collecting visitor density data and spatial coordinate data of each interactive terminal in real time within the venue's exhibition areas; establishing a dynamic potential energy model of visitor flow to predict the interactive request load of each interactive terminal's location within a future time window; when the predicted load exceeds the service capacity threshold, triggering the terminal to send a load negotiation broadcast to neighboring interactive terminals; receiving available computing resource values ​​and relative azimuth angles returned by neighboring terminals; calculating the interactive task migration adaptability of each neighboring terminal, and sequentially distributing interactive requests exceeding the threshold to neighboring terminals according to adaptability from high to low. This invention achieves early load migration based on the prediction of visitor flow potential energy and integrates resource and azimuth angles for adaptability decisions, avoiding overload and response delays of exhibition interactive terminals.

Description

Technical Field

[0001] This invention relates to the field of intelligent interactive control technology for exhibitions, specifically to an intelligent interactive control method, medium, and device for exhibitions. Background Technology

[0002] Exhibition venues typically deploy numerous interactive terminals to provide visitors with services such as exhibit information inquiries and interactive guides. Each terminal needs to handle real-time interactive requests from surrounding visitors. Visitor flow within the venue is highly mobile, with pedestrian density changing rapidly over time in different exhibition areas, causing drastic fluctuations in the request load on each interactive terminal. When a terminal experiences a momentary overload, visitor interactive requests cannot be responded to in a timely manner, directly impacting the exhibition experience. Existing load balancing methods often rely on the current load state of the interactive terminal itself, setting static thresholds to trigger request transfers. However, transfer decisions are usually based solely on the idle computing resources of nearby terminals.

[0003] Existing technical solutions have significant drawbacks. First, load status perception relies entirely on the arrival of currently occurring requests, resulting in a passive response. By the time overload is detected, a large number of requests have already accumulated, leading to a delay in migration and making it impossible to avoid the initial impact of service congestion. Second, task migration is based solely on the availability of computing resources to select the target terminal, ignoring the terminal's physical location. In exhibition scenarios, the relative position between visitors and interactive terminals determines the quality of wireless communication and the sensory experience of interactive response. If a request is migrated to a terminal with an unsuitable location, the visitor's mobile terminal may need to switch antenna directions or reconnect, creating interaction interruptions and additional latency, weakening the continuity of the intelligent interactive system.

[0004] Therefore, two key issues need to be addressed: how to predict the load pressure of each interactive terminal in the future based on the visitor flow pattern, so that the load migration occurs before the overload actually occurs; and how to consider the available computing resources of the terminal and its spatial orientation relative to the triggering terminal when selecting the migration target terminal, so as to ensure that the interactive tasks after migration can obtain sufficient processing capacity, and that the service latency will not increase due to the mismatch of terminal location. Summary of the Invention

[0005] This invention provides a method, medium, and device for intelligent interactive control of exhibitions, aiming to solve the problems of passive response overload and migration decision-making that do not consider spatial orientation in existing exhibition interactive terminals, resulting in service congestion and increased interaction latency.

[0006] To achieve the above objectives, the present invention provides the following technical solution: The present invention provides an intelligent interactive control method for exhibitions, applied to a distributed intelligent interactive node network deployed in an exhibition center. This method establishes a dynamic potential energy model of the visitor flow field by real-time collection of visitor density data and spatial coordinate data of each interactive terminal in each exhibition area within the exhibition venue, and predicts the interactive request load of each interactive terminal's location within a future time window based on this dynamic potential energy model. When the predicted interactive request load of any interactive terminal exceeds its service capacity threshold, the interactive terminal is triggered to send a load negotiation broadcast to neighboring interactive terminals. In response to the load negotiation broadcast, negotiation response data returned by neighboring interactive terminals is received, including the available computing resources of the neighboring interactive terminals and their relative azimuth angle with the triggering terminal. Based on the available computing resources and relative azimuth angle returned by each neighboring interactive terminal, the interactive task migration adaptability of each neighboring interactive terminal is calculated, and according to the adaptability from high to low, the pending interactive requests exceeding the service capacity threshold are sequentially allocated to each neighboring interactive terminal. This solution dynamically links the macro-level flow of visitors with the micro-level load capacity of interactive terminal nodes, making load migration decisions in advance before the peak of interactive requests arrives. This effectively avoids overload and crash of individual interactive terminals due to sudden access, ensuring the continuity and real-time nature of exhibition interactive services.

[0007] As a preferred embodiment of the present invention, the specific implementation method for real-time acquisition of visitor density data and spatial coordinate data of each interactive terminal in each exhibition booth area within the exhibition venue is as follows: A millimeter-wave radar array deployed on the ceiling of the exhibition venue scans each booth area to obtain raw point cloud data. The raw point cloud data is then clustered, and after removing static background point clouds, the number of dynamic point clouds in each booth area is counted. This number of dynamic point clouds is used as the visitor density data. Simultaneously, the ultra-wideband positioning module built into each interactive terminal communicates with a fixed positioning base station within the venue to calculate the spatial coordinate data of each interactive terminal in the three-dimensional spatial coordinate system of the exhibition venue. Preferably, the scanning frequency of the millimeter-wave radar array is dynamically adjusted according to the real-time visitor density of each booth area; the higher the density, the higher the scanning frequency, thereby obtaining higher temporal accuracy density sampling in areas where visitors congregate, ensuring the timeliness and accuracy of the input data for the flow field model.

[0008] As a preferred embodiment of the present invention, the process of establishing a dynamic potential energy model of visitor flow and predicting the interactive request load within a future time window specifically involves: mapping the spatial coordinate data of each interactive terminal onto a two-dimensional booth plan to form an interactive terminal distribution grid; for each booth area, using its visitor density data as the flow field mass value of that area at the current moment, and calculating the flow velocity vector of visitors spreading from high-density areas to low-density areas based on the flow field mass value gradient between adjacent booth areas; substituting the spatial coordinate data of each interactive terminal into the dynamic potential energy function jointly constructed by the flow field mass values ​​and flow velocity vectors of all booth areas, and solving for the potential energy gradient change rate at the location of the interactive terminal; using this potential energy gradient change rate as an interactive request excitation factor, and combining it with the historical interactive request arrival rate of the interactive terminal, using an autoregressive moving average model to predict the interactive request load within a future time window. Preferably, a time decay factor is introduced into the dynamic potential energy function to reduce the influence weight of historical visitor density data on the current potential energy gradient change rate, making the prediction results more reflective of the latest trend of visitor flow.

[0009] As a preferred embodiment of the present invention, the specific conditions and methods for triggering load negotiation broadcast are as follows: A service capacity threshold for the interactive terminal is set as the product of the number of its processor cores and the maximum request processing rate per core; when the predicted interactive request load exceeds this service capacity threshold, the interactive terminal is marked as a triggering terminal; the triggering terminal, through a wireless ad hoc network, sends a load negotiation broadcast message containing the number and type of interactive requests to be migrated to all interactive terminals within a circular area centered on its own spatial coordinates and with a preset communication distance as its radius. By introducing the number of processor cores as a hardware constraint factor, the service capacity threshold objectively reflects the upper limit of the terminal's actual processing capability, avoiding load misjudgment caused by subjectively setting the threshold.

[0010] As a preferred embodiment of the present invention, the specific process of the neighboring interactive terminal generating and returning negotiation response data is as follows: After receiving the load negotiation broadcast message, the neighboring interactive terminal reads its current CPU occupancy rate and remaining memory capacity, performs a weighted fusion calculation on the two to obtain the available computing resource value; the neighboring interactive terminal obtains its own orientation angle through its built-in electronic compass, and simultaneously receives the orientation angle of the triggering terminal broadcast by the triggering terminal. Based on the difference between its own orientation angle and the triggering terminal's orientation angle, and combined with the direction of the line formed by its own spatial coordinate data and the triggering terminal's spatial coordinate data, the neighboring interactive terminal calculates the relative azimuth angle; the neighboring interactive terminal encapsulates the available computing resource value and the relative azimuth angle into a negotiation response data message and returns it to the triggering terminal along the original path. The introduction of the relative azimuth angle allows the subsequent migration adaptation calculation to not only consider the remaining computing power of the neighboring terminal, but also comprehensively measure the geographical matching degree between the terminal orientation and the relative position of the visitor, which is beneficial for migrating the interaction request to a terminal node that is more convenient for serving the target visitor.

[0011] As a preferred embodiment of the present invention, the process of calculating the interaction task migration adaptability and allocating the interaction requests to be processed is as follows: For each neighboring interaction terminal that returns negotiation response data, a two-dimensional migration evaluation vector is constructed, where the first element is the available computing resource value of the neighboring interaction terminal, and the second element is the absolute value of the difference between the relative azimuth angle of the neighboring interaction terminal and the preset ideal azimuth angle; the magnitude of the two-dimensional migration evaluation vector is calculated, and the reciprocal of the magnitude is used as the initial value of the interaction task migration adaptability of the neighboring interaction terminal; the interaction requests to be processed are sorted from high to low according to the request complexity to form a queue to be allocated; neighboring interaction terminals are selected in descending order of interaction task migration adaptability, and the interaction request with the highest request complexity in the queue to be allocated is allocated to the currently selected neighboring interaction terminal, and the interaction request is removed from the queue to be allocated. This allocation step is repeated until the queue to be allocated is empty or the available computing resource value of all neighboring interaction terminals has been exhausted. This allocation strategy prioritizes migrating high-complexity requests to the terminal with the best overall evaluation, reduces the risk of low-computing-power terminals being assigned heavy tasks, and improves the overall load balancing efficiency.

[0012] As a preferred embodiment of the present invention, if the queue to be allocated is still not empty after the task migration adaptation is allocated from high to low, the terminal is triggered to re-initiate load negotiation broadcast to the interactive terminal at a greater distance. By gradually expanding the negotiation range layer by layer, available resources are gradually found, avoiding network congestion caused by an excessively large single broadcast range.

[0013] In a preferred embodiment of the present invention, after distributing pending interaction requests exceeding the service capacity threshold to neighboring interactive terminals in sequence, the following operations are further performed: the triggering terminal sends an interaction request forwarding instruction to the neighboring interactive terminals to which the interaction requests were assigned. This forwarding instruction includes the original data packet of the pending interaction request and a session state snapshot generated by the triggering terminal for the interaction request. Upon receiving the forwarding instruction, the neighboring interactive terminal reconstructs the interaction session context based on the session state snapshot and establishes a direct interactive communication link with the visitor's mobile terminal based on the visitor identification information in the original data packet. By transmitting the session state snapshot instead of just the request data itself, the migrated interaction request can be seamlessly continued on the new terminal, allowing the visitor to switch service terminals without being aware of it.

[0014] As a preferred embodiment of the present invention, the specific process of session reconstruction and communication link establishment is as follows: The neighboring interactive terminal parses the session state snapshot, extracting the original session identifier, historical question-and-answer records, and incomplete interaction queue indexes; the neighboring interactive terminal registers the original session identifier in its local session management table and loads the historical question-and-answer records and incomplete interaction queue indexes into a memory buffer, completing the reconstruction of the interactive session context; based on the visitor's mobile terminal's media access control address and wireless signal strength in the original data packet, the neighboring interactive terminal uses directional antenna beamforming technology to establish a point-to-point wireless communication link with the visitor's mobile terminal and binds this communication link to the reconstructed session context. Directional beamforming technology improves the signal quality and anti-interference capability of the migrated link, ensuring the stable operation of the interaction process.

[0015] In a preferred embodiment of the present invention, after distributing pending interaction requests exceeding the service capacity threshold to neighboring interaction terminals in sequence, a load recovery step is further performed: the triggering terminal monitors its own real-time interaction request load. When the real-time interaction request load drops below the service capacity threshold and remains stable for a preset duration, the triggering terminal sends a load recovery broadcast to all neighboring interaction terminals that have been assigned interaction requests. In response to the load recovery broadcast, each neighboring interaction terminal that receives the broadcast encapsulates the running status data and intermediate calculation results of its currently processed interaction requests, originally belonging to the triggering terminal, into a session migration data packet and sends it back to the triggering terminal. After receiving the session migration data packet, the triggering terminal restores interaction control with the corresponding visitor's mobile terminal and notifies neighboring interaction terminals to release the corresponding session context and communication link resources. This recovery mechanism makes load migration a closed-loop process with a complete lifecycle, promptly recovering tasks after the triggering terminal's load returns to normal, releasing resources of neighboring terminals, and maintaining the long-term efficient operation of the entire interaction node network.

[0016] As a preferred embodiment of the present invention, before real-time data acquisition, an initialization configuration step is performed: interactive terminals are deployed in the physical space of the exhibition venue according to a regular hexagonal grid layout. Each interactive terminal is configured with a unique node identifier and an initial service capacity threshold. A global clock signal and a unified coordinate system reference point are synchronized to all interactive terminals through the venue's central server. After each interactive terminal starts up, it reports its node identifier and spatial coordinate data to the central server. The central server constructs a venue interactive terminal topology map based on the spatial coordinate data of all interactive terminals and broadcasts the topology map to all interactive terminals. The regular hexagonal grid layout ensures full coverage with equal distances between adjacent interactive terminals, eliminates the topological defects of unequal diagonal distances in rectangular layouts, and provides a uniform spatial basis for setting the communication radius for load negotiation broadcasting and discovering neighboring terminals.

[0017] The present invention also provides a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the aforementioned intelligent interactive control method for exhibitions. This storage medium enables the method to be deployed as software on existing exhibition interactive terminal hardware platforms, reducing system upgrade costs.

[0018] This invention also provides an intelligent interactive control device for exhibitions, including a processor and a memory. The memory stores a computer program, which, when executed by the processor, implements the aforementioned intelligent interactive control method for exhibitions. This device can be integrated into the central server or edge computing node of an exhibition venue to uniformly schedule and manage the entire interactive terminal network, thereby improving the level of intelligent management of exhibitions.

[0019] The technical effects and advantages provided by the present invention in the above technical solution are as follows: By constructing a dynamic potential energy model of visitor flow and predicting the load of interactive requests within future time windows, a shift from passive overload detection to proactive load pre-migration is achieved. This solution utilizes a millimeter-wave radar array deployed on the venue's roof to acquire real-time visitor density in each booth area and maps this density to the spatial coordinates of each interactive terminal, forming a prediction mechanism based on the potential energy of crowd diffusion. By introducing a flow velocity vector driven by the density gradient of adjacent areas and using the rate of change of the potential energy gradient as an excitation factor for interactive requests, combined with a time-series model of historical arrival rates, it can detect in advance localized load peaks that may arise due to visitor influx or changes in flow direction. This early-acquired predicted load triggers negotiation before the actual service capacity of the terminals is exhausted, allowing requests exceeding capacity to be pre-migrated, preventing interactive terminals from reaching saturation and ensuring real-time response to visitor interactive requests.

[0020] During the terminal selection process for load migration, the assignment is based on the task migration adaptability determined by the available computing resources and the relative azimuth angle. The migration decision not only collects the available computing resources value derived from the fusion of the CPU utilization and remaining memory capacity of neighboring terminals, but also incorporates the difference in orientation angle between each neighboring terminal and the triggering terminal. A migration evaluation vector is constructed using the available computing resources value as one dimension and the absolute value of the difference between the relative azimuth angle and the preset ideal azimuth angle as another dimension, with the adaptability measured by the reciprocal of the vector's magnitude. This design prioritizes the terminal whose azimuth angle is closest to the ideal value when computing resources are similar, allowing it to undertake the migration task. Because the receiving terminal is in a more reasonable relative physical orientation, it can directly establish an efficient beamforming communication link with the visitor's mobile terminal, eliminating the need for frequent switching of service terminals and the overhead of re-association and angle search, maintaining the continuity of the interactive session, and reducing the perception of interaction breakpoints caused by changes in spatial location. Attached Figure Description

[0021] To more clearly illustrate the technical solutions in the embodiments of this application or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in this invention. For those skilled in the art, other drawings can be obtained based on these drawings.

[0022] Figure 1 This is a flowchart of a smart interactive control method for exhibitions; Figure 2 This is a flowchart of the load negotiation process for interactive terminals; Figure 3 It is a curve showing the change in visitor density over time in different exhibition areas of the exhibition venue; Figure 4 It is a distribution map of available computing resources and relative azimuth angles of nearby interactive terminals. Detailed Implementation

[0023] To make the objectives, technical solutions, and advantages of the embodiments of the present invention clearer, the technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0024] See Figure 1This invention provides an intelligent interactive control method for exhibitions, applied to a distributed intelligent interactive node network deployed in an exhibition center. The method includes: real-time collection of visitor density data and spatial coordinate data of each interactive terminal in each booth area of ​​the exhibition venue; establishing a dynamic potential energy model of visitor flow field based on the visitor density data and spatial coordinate data, and predicting the interactive request load of each interactive terminal's location within a future time window based on the dynamic potential energy model; when the predicted interactive request load of any interactive terminal exceeds the service capacity threshold of that interactive terminal, triggering the interactive terminal to send a load negotiation broadcast to neighboring interactive terminals; responding to the load negotiation broadcast, receiving negotiation response data returned by neighboring interactive terminals, the negotiation response data including the available computing resources of neighboring interactive terminals and their relative azimuth angle with the triggering terminal; calculating the interactive task migration adaptability of each neighboring interactive terminal based on the available computing resources and relative azimuth angle returned by each neighboring interactive terminal, and sequentially allocating the pending interactive requests exceeding the service capacity threshold to each neighboring interactive terminal according to the order of the interactive task migration adaptability from high to low.

[0025] Example 1: In specific implementation, multiple millimeter-wave radars are installed on the ceiling of the exhibition hall, forming a millimeter-wave radar array that covers the entire exhibition area. Each millimeter-wave radar transmits a linear frequency modulated continuous wave signal to the exhibition area below at a fixed frame period and receives the reflected echo signal. The reflected echo signal is mixed, filtered, and sampled to obtain the intermediate frequency signal of the corresponding channel. The millimeter-wave radar performs range-dimensional fast Fourier transform and Doppler-dimensional fast Fourier transform on the intermediate frequency signal to generate a range-Doppler map. Points exceeding a preset constant false alarm rate (CFAR) threshold are extracted from the range-Doppler map as target points. Combining the millimeter-wave radar's own position coordinates and beam pointing angle, the target points are mapped onto the three-dimensional spatial coordinate system of the exhibition hall to form raw point cloud data. Each frame of raw point cloud data contains the three-dimensional coordinates, radial velocity, and echo intensity information of multiple points.

[0026] The steps for clustering the raw point cloud data are as follows: All points in a frame of raw point cloud data are used as the input point set, and a density-based spatial clustering algorithm is employed. In this algorithm, the neighborhood radius is set to 0.5 meters, and the minimum number of points within the neighborhood is set to 5. For each point in the input point set, if the number of points within its neighborhood radius is greater than or equal to the minimum number of points, that point is marked as a core point, and all points within that radius are grouped into the same cluster. After performing this operation on all points, at least one point cluster is obtained. Clusters with fewer than a preset threshold number of points are removed; this threshold is set to 10.

[0027] The removal of static background point clouds is achieved by comparing multiple consecutive frames of point clouds. K consecutive frames of point clouds are taken in time, with K set to 5. The cluster center position is calculated for each cluster of points in these 5 frames. If a cluster of points appears in 5 consecutive frames at the same spatial location, and the spatial movement distance of the cluster center position between adjacent frames is less than 0.1 meters, then the cluster is determined to be a static background point cloud, and it is removed from the original point cloud data of each frame. After removing the static background point cloud, the remaining point cloud in each frame is the dynamic point cloud. The number of dynamic point clouds in each booth area is counted. The counting method is as follows: Closed polygon boundaries corresponding to each booth area are pre-defined on the 2D floor plan of the exhibition hall. The 3D coordinates of each point in the dynamic point cloud are projected onto the 2D plane. It is determined whether the projected point falls within the closed polygon. Points falling within the closed polygon are counted, and the count results are used as visitor density data.

[0028] Meanwhile, each interactive terminal has a built-in ultra-wideband (UWB) positioning module, and at least four positioning base stations are fixedly installed within the venue. The coordinates of these base stations in the three-dimensional spatial coordinate system of the exhibition venue are precisely measured beforehand and stored on a central server. The UWB positioning module performs bidirectional ranging communication with each positioning base station. During ranging communication, the UWB positioning module of the interactive terminal sends a ranging request frame to a positioning base station. After receiving the ranging request frame, the positioning base station returns a ranging response frame after a fixed processing delay. The UWB positioning module of the interactive terminal records the sending time of the ranging request frame and the receiving time of the ranging response frame, and calculates the distance from the interactive terminal to that positioning base station using the time-of-flight method. The interactive terminal sequentially completes ranging with all positioning base stations, obtaining a set of distances from the interactive terminal to each positioning base station. Using the distance set and the spatial coordinates of all positioning base stations as input, the least squares method is used to solve the overdetermined equations, obtaining the spatial coordinate data of the interactive terminal in the three-dimensional spatial coordinate system of the exhibition venue. In the three-dimensional spatial coordinate system, the X-axis and Y-axis form a horizontal plane, and the Z-axis points vertically upwards.

[0029] Optionally, the specific implementation method for dynamically adjusting the scanning frequency of the millimeter-wave radar array according to the real-time visitor density of each booth area is as follows: Every preset adjustment period, the latest visitor density data for each booth area is read; the adjustment period is set to 5 seconds. For each millimeter-wave radar, one or more booth areas covered by the radar are determined, and the maximum value of the visitor density data within the covered area is taken as the reference density of the millimeter-wave radar. The preset density interval segmentation thresholds are: the first density interval is a visitor density less than or equal to 0.5 people per square meter; the second density interval is a visitor density greater than 0.5 people per square meter but less than or equal to 1.5 people per square meter; and the third density interval is a visitor density greater than 1.5 people per square meter. When the reference density falls within the first density interval, the scanning frequency of the millimeter-wave radar is set to 10 Hz; when the reference density falls within the second density interval, the scanning frequency is set to 20 Hz; and when the reference density falls within the third density interval, the scanning frequency is set to 30 Hz. The millimeter-wave radar collects data according to the updated scanning frequency. This adjustment method, which increases the scanning frequency as density increases, enables the acquisition of richer temporal resolution point clouds in high-density scenes, thereby improving the real-time accuracy of visitor density data collection.

[0030] See Figure 3 The horizontal axis of the graph represents time in seconds, covering an observation period of approximately 0 to 2400 seconds; the vertical axis represents visitor density in each booth area, in people per square meter (sqm). Solid lines, dashed lines, and dotted lines represent the visitor density trends in booth areas A, B, and C, respectively. The visitor density in booth area A fluctuates between 0.0 and 1.3 people / sqm, exhibiting periodic fluctuations with a period of approximately 400 seconds, indicating relatively active visitor movement and moderate density in this area. The visitor density in booth area B is generally lower, remaining within the range of 0.1 to 0.5 people / sqm with smaller fluctuations, indicating relatively sparse and stable visitor movement in this area. The visitor density in booth area C is significantly higher than in areas A and B, ranging from approximately 0.5 to 2.5 people / sqm, and exhibits strong periodic fluctuations, with peaks exceeding 2 people / sqm multiple times, indicating concentrated visitor activity and high visitor flow in this area. The visitor density curves for all three exhibition areas exhibited clear temporal dynamic variations, indicating that the dynamic point cloud data collected by the millimeter-wave radar array at different scanning frequencies effectively reflected the different visitor density distributions and trends in each exhibition area. The time-series visitor density data shown in this figure was collected in real-time by the millimeter-wave radar array and statistically analyzed after clustering and dynamic point cloud identification. It was used for the subsequent establishment of the dynamic potential energy model of the flow field and the prediction of interactive request load.

[0031] Example 2: In specific implementation, the method for mapping the spatial coordinate data of each interactive terminal onto the two-dimensional booth plan is as follows: The spatial coordinate data of the interactive terminal in the three-dimensional spatial coordinate system of the exhibition venue is obtained. The spatial coordinate data includes X-axis coordinate values, Y-axis coordinate values, and Z-axis coordinate values. The X-axis and Y-axis coordinate values ​​are extracted, while the Z-axis coordinate value is discarded. The interactive terminal is then projected onto the two-dimensional booth plan defined by the X-axis and Y-axis. The two-dimensional booth plan is pre-stored in vector form, containing the boundary outlines of each booth area, aisle locations, and fixed facility markings. After projection, each interactive terminal corresponds to a unique planar coordinate point on the two-dimensional booth plan. The planar coordinate points of all interactive terminals constitute an interactive terminal distribution grid. In the interactive terminal distribution grid, the spatial distance between adjacent interactive terminals is determined by calculating the Euclidean distance between the two planar coordinate points.

[0032] For each booth area, the steps to use its visitor density data as the flow field quality value are as follows: At each data collection time, acquire the visitor density data for each booth area. The visitor density data is the number of dynamic point clouds statistically obtained within that booth area. The number of dynamic point clouds is directly assigned to the flow field quality value for that booth area. The flow field quality value is then... It indicates that, for the first The quality value of the flow area in each booth area is recorded as follows: The unit is the number of people. The gradient of the flow field mass value between adjacent booth areas is calculated as follows: For any two adjacent booth areas sharing a boundary segment, take the difference in the flow field mass values ​​of the two adjacent booth areas, divide it by the distance between the geometric center points of the two adjacent booth areas, and obtain the flow field mass value gradient. Taking the first... Individual booth areas and the first Each booth area is an adjacent booth area, and the gradient of the flow field mass value is as follows: The calculation formula is as follows: ,in, Indicates the first The quality value of the flow area in each booth area, in units of people; Indicates the first The quality value of the flow area in each booth area, in units of people; Indicates the first From the geometric center point of each booth area to the... The straight-line distance between the geometric center points of each booth area, in meters. Mass gradient of the flow field. The direction is from the booth area with high flow field quality value to the booth area with low flow field quality value.

[0033] The process of calculating the flow velocity vector of visitors diffusing from high-density areas to low-density areas based on the flow field mass value gradient between adjacent booth areas is as follows: (Set a scaling factor...) proportionality coefficient The value ranges from 0.1 to 0.5, and is calibrated based on the width of the passageways within the exhibition venue and the average walking speed of visitors. The proportionality coefficient in this implementation... Set the value to 0.2. Adjust the scaling factor. Gradient of mass value of flow field The product is multiplied, and the direction of the product remains consistent with the direction of the mass gradient of the flow field. The magnitude of the product is a proportionality constant. Gradient of mass value of flow field The product of the absolute values ​​of the vectors forms the flow velocity vector. express, . The unit is meters per second.

[0034] The operation of substituting the spatial coordinate data of each interactive terminal into the dynamic potential energy function is as follows: When constructing the dynamic potential energy function, the flow field mass values ​​of all booth areas and the flow velocity vectors between all adjacent booth areas are used as known quantities to construct a scalar field covering the entire two-dimensional booth plan. For the coordinates on the two-dimensional booth plan... At any point, the dynamic potential energy function The dynamic potential energy function is determined by the distance from this point to the geometric center of each booth area and the mass value of the flow field in each booth area. The calculation method is a weighted summation with the mass value of the flow field in each booth area as the weight and the distance as the variable. The expression is:

[0035] In the above expression, Indicates the total number of exhibition booth areas; Indicates the first Each booth area is available at any time. The mass value of the flow field, in units of people; Indicates coordinates as The point to the first The Euclidean distance between the geometric center points of each booth area, in meters; Represents the spatial attenuation coefficient. The value of is related to the average coverage radius of the booth area, and is taken as the reciprocal of the average coverage radius of the booth area. The average coverage radius of the booth area is determined based on the radius of the equivalent circle of the booth area. If the average coverage radius of the booth area is 5 meters, then the space attenuation coefficient is... Value 0.2 per meter; Represents the natural constant.

[0036] The implementation method for introducing a time decay factor into the dynamic potential energy function is as follows: set the time decay factor. Time decay factor The value of is determined by the exponential decay model. ,in This represents the decay rate constant, with a value ranging from 0.01 to 0.1 per second. Based on the average visitor dwell time and flow inertia settings, the decay rate constant in this implementation is... The value is 0.05 per second; This represents the time difference between a historical moment and the current moment, in seconds. It also represents the flow field mass value at a historical moment. Multiply it by the time decay factor Then it is used in the calculation of the dynamic potential energy function, at which point the dynamic potential energy function is... The Middle The contribution items for each booth area are transformed into ,in Indicates the current moment. Representing a historical moment. The time decay factor varies with time difference. As the potential energy gradient changes, the influence of historical visitor density data on the current rate of change of potential energy gradually decreases.

[0037] The method for obtaining the rate of change of the potential energy gradient at the location of the interactive terminal by solving the dynamic potential energy function is as follows: Take the coordinates of the interactive terminal on the two-dimensional booth plan as... , coordinate Substitute into the dynamic potential energy function Calculate the first-order partial derivatives of the dynamic potential energy function along the X-axis. and the first-order partial derivative in the Y-axis direction The rate of change of the potential energy gradient is... This represents the rate of change of the potential energy gradient. The value is the square root of the sum of the squares of the two first-order partial derivatives: ,in, Represents the dynamic potential energy function At coordinates The first partial derivative of the interactive terminal position with respect to the X-axis, in person per meter; Represents the dynamic potential energy function At coordinates The first-order partial derivative with respect to the Y-axis at the location of the interactive terminal, in units of person per meter.

[0038] The specific method of using the potential energy gradient change rate as the interaction request incentive factor is as follows: Set the interaction request incentive factor to... This indicates the interaction request incentive factor. The value is taken as the rate of change of the potential energy gradient. A linear function, ,in Indicates the incentive ratio coefficient. The value is determined based on the historical interaction request trigger statistics of the interactive terminal, and the incentive ratio coefficient is used for calibration. The value ranges from 0.5 to 2.0. In this implementation, the excitation ratio coefficient... The value is 1.0.

[0039] In some embodiments, the structure of the autoregressive moving average model is defined as follows: The autoregressive moving average model is defined as follows: Identifier, autoregressive order The value is 3, which represents the order of the moving average. The value is 2. The autoregressive moving average model consists of an autoregressive part and a moving average part. The autoregressive part describes the linear relationship between the current interaction request arrival rate and the historical interaction request arrival rate, while the moving average part describes the linear relationship between the current interaction request arrival rate and the historical random disturbance term.

[0040] The steps for predicting the interaction request load within a future time window using an autoregressive moving average model are: obtaining the interaction terminal's past... The historical interaction request arrival rate sequence at each sampling point The value is 50, and the sampling interval is 10 seconds. The historical interaction request arrival rate sequence is compared with the interaction request activation factor corresponding to each sampling point. Multiplying these results yields the incentive-adjusted historical interaction request arrival rate sequence. A stationarity test is performed on this sequence; if non-stationary, first-order differencing is performed until stationarity is achieved. The parameters of the autoregressive moving average model are then estimated using the least squares method on the stationary sequence. The parameters of the autoregressive moving average model include the autoregressive coefficients. and moving average coefficient Based on the estimated parameters, the arrival rate of interaction requests within the future time window is recursively predicted. The future time window is set to the next sampling period, i.e., the next 10 seconds. The predicted arrival rate of interaction requests is multiplied by the duration of the future time window to obtain the interaction request load within the future time window.

[0041] Interaction request load represents the total number of interaction requests expected to arrive at the interactive terminal within a predicted future time window. Interaction request load is a dimensionless integer value, used for subsequent comparison with service capacity thresholds.

[0042] Example 3: See Figure 2 In specific implementation, the method for setting the service capacity threshold of the interactive terminal is as follows: obtain the number of processor cores of the interactive terminal, and the number of processor cores is... This indicates that the unit is "number"; it retrieves the maximum request processing rate of a single core on the interactive terminal, with the maximum request processing rate of a single core expressed as... This indicates the maximum request processing rate per second (RPS). The results were obtained through offline benchmark testing, which was performed before deployment on the interactive terminal. The load of standard interactive requests was gradually increased, and the maximum number of requests processed by a single core per unit time was recorded. The average of multiple tests was taken as the maximum request processing rate of a single core. Service capacity thresholds are based on This indicates the service capacity threshold. The calculation formula is: ,in, Indicates the number of processor cores, expressed in units. This indicates the maximum request processing rate per single core, measured in requests per second. This indicates the service capacity threshold, measured in times per second.

[0043] The operation method for comparing the predicted interaction request load with the service capacity threshold is as follows: obtain the interaction request load within a future time window, and the interaction request load is... This indicates that the unit is times; it retrieves the duration of future time windows, with the future time window duration expressed in terms of times. This indicates the unit is seconds; it represents the interactive request load. Divide by the duration of the future time window The predicted interaction request arrival rate is obtained, and the predicted interaction request arrival rate is... express, The unit is times per second. This will predict the arrival rate of interactive requests. With service capacity threshold If a comparison is made, If the predicted interaction request load exceeds the service capacity threshold, the corresponding interaction terminal is marked as a trigger terminal. The marking of a trigger terminal is achieved by modifying a flag bit in the interaction terminal's local status register; a high-level flag indicates the trigger terminal status.

[0044] The operation of the triggering terminal sending load negotiation broadcast messages through the wireless ad hoc network is as follows: The wireless transceiver module of the triggering terminal operates on a pre-allocated wireless channel, and the wireless ad hoc network adopts the independent basic service set mode in the IEEE 802.11 standard. The communication distance is preset to the spatial coordinate data of the triggering terminal. Preset communication distance The value is set at 30 meters, based on the typical deployment spacing of interactive terminals in exhibition halls and the effective transmission range of wireless signals in indoor environments. The center of the circle is the planar coordinate point on the two-dimensional booth floor plan where the trigger terminal's spatial coordinate data is located, with the preset communication distance as the reference. A circular area is defined with radius . The triggering terminal constructs a load negotiation broadcast message. The frame format of the load negotiation broadcast message includes a message type identifier field, a triggering terminal node identifier field, a triggering terminal spatial coordinate data field, a number of interaction requests to be migrated field, a request type list field, and a triggering terminal orientation angle field. The number of interaction requests to be migrated is the rounded difference between the predicted interaction request load and the product of the service capacity threshold and the future time window duration. The request type list records the identifier code for each request type. The triggering terminal broadcasts the load negotiation broadcast message to the wireless ad hoc network, filling the destination address field with the broadcast address.

[0045] In some embodiments, after receiving a load negotiation broadcast message, the neighboring interactive terminal reads its current CPU utilization rate by: the operating system of the neighboring interactive terminal providing a system call interface, and obtaining the proportion of time the CPU is in a non-idle state by calling system functions; the CPU utilization rate is then expressed as follows: express, The value ranges from 0% to 100%. The remaining memory capacity is read through the operating system's memory management interface, which displays the total amount of currently unused physical memory. This indicates that the unit is megabytes. It represents the CPU utilization rate. With remaining memory capacity The formula for weighted fusion calculation is: ,in, This represents the available computing resource value, which is a dimensionless numerical value ranging from 0 to 1. This indicates the CPU utilization rate, expressed as a percentage. This indicates the remaining memory capacity, in megabytes (MB). This indicates the total amount of physical memory installed in the interactive terminal, in megabytes. This represents the weighting coefficient of the central processing unit. The value is 0.6; This represents the memory weighting coefficient. The value is 0.4. This represents the CPU weighting coefficient. and memory weight coefficient The value is determined based on the fact that interactive request processing tasks are computationally intensive tasks, and the impact of CPU resources on processing power is greater than that of memory resources. Therefore, the CPU weight coefficient is set to a higher value than the memory weight coefficient.

[0046] The nearby interactive terminal obtains its orientation angle through a built-in electronic compass as follows: The electronic compass includes a three-axis magnetometer and a three-axis accelerometer. The three-axis magnetometer measures the components of the Earth's magnetic field along the three axes of the nearby interactive terminal's coordinate system, and the three-axis accelerometer measures the components of gravitational acceleration along the three axes. By fusing the data from the three-axis magnetometer and the three-axis accelerometer, the yaw angle of the nearby interactive terminal relative to the Earth's magnetic north pole is calculated. This yaw angle is then converted into an orientation angle based on the unified coordinate system reference direction of the venue. The terminal's own orientation angle is then calculated using... The unit is degrees. The method by which a neighboring interactive terminal receives the triggering terminal's orientation angle broadcast by the triggering terminal is as follows: the triggering terminal appends its own orientation angle field to the load negotiation broadcast message. The triggering terminal's orientation angle is expressed as... It indicates that the unit is degrees, and the method for measuring the orientation angle of the trigger terminal is the same as the method for measuring the orientation angle of the adjacent interactive terminal itself.

[0047] The process by which a neighboring interactive terminal calculates its relative azimuth angle based on the difference between its own orientation angle and the triggering terminal's orientation angle, combined with the direction of the line connecting its own spatial coordinate data and the triggering terminal's spatial coordinate data, is as follows: The triggering terminal's spatial coordinate data is obtained and contained in the load negotiation broadcast message. The planar coordinates of the triggering terminal's spatial coordinates on the two-dimensional booth plan are... Obtain the spatial coordinate data of the adjacent interactive terminal itself. The planar coordinates of its own spatial coordinates on the two-dimensional booth floor plan are: Calculate the direction angle of the line connecting the nearest interactive terminal to the triggering terminal, where the direction angle is... Indicates the direction angle of the connecting line. The angle between the direction of the line and the reference direction of the venue's unified coordinate system, expressed in degrees. Calculated using the arctangent function: ,in Represents the arctangent function in the four quadrants. Calculates its own orientation angle. Angle of orientation of the trigger terminal The difference is denoted as the orientation difference. , The unit is degrees. The relative azimuth angle is... Indicates the relative azimuth angle. The calculation method is as follows: The calculation results are then subjected to a modulo 360-degree operation to determine the relative azimuth angle. The value falls within the range of 0 degrees to 360 degrees. When the relative azimuth angle... When the angle is greater than 180 degrees, take As a corrected relative azimuth angle, the final value of the relative azimuth angle ranges from 0 degrees to 180 degrees.

[0048] Optionally, nearby interactive terminals will have access to computing resource values. Angle with relative azimuth The operation of encapsulating a negotiation response data message is as follows: A negotiation response data message is constructed, which includes a message type identifier field, a neighboring interactive terminal node identifier field, an available computational resource value field, a relative azimuth angle field, and a neighboring interactive terminal spatial coordinate data field. The available computational resource value field is filled with the calculated values. The value, relative azimuth angle field, is filled with the calculated value. The neighboring interactive terminal reads the source address of the load negotiation broadcast message, sets the destination address of the negotiation response data message to the source address of the load negotiation broadcast message, and returns the negotiation response data message to the triggering terminal via unicast along the original path through the wireless ad hoc network. The original path refers to the reverse route traversed by the load negotiation broadcast message in the wireless ad hoc network. In the independent basic service set mode, if the neighboring interactive terminal and the triggering terminal are within single-hop communication range, the negotiation response data message is sent directly to the triggering terminal.

[0049] See Figure 4 In the graph, the horizontal axis represents the relative azimuth angle between neighboring interactive terminals and the triggering terminal, in degrees, ranging from 0 to 180 degrees; the vertical axis represents the available computing resources of neighboring interactive terminals, ranging from 0 to 1. The scatter plot distribution in the graph specifically reflects the correspondence between the available computing resources of multiple neighboring interactive terminals and their relative azimuth angles. The vertical axis value represents the resource availability index calculated by weighted fusion of CPU utilization and remaining memory capacity; a larger value indicates more abundant available computing resources. The horizontal axis value represents the orientation relationship between neighboring interactive terminals and the triggering terminal; a smaller angle indicates that the neighboring interactive terminal is closer to the ideal orientation towards the triggering terminal.

[0050] The ideal azimuth angle, marked by a dashed line in the diagram, is 0 degrees, representing the location where the communication link gain is greatest when the nearest interactive terminal faces the triggering terminal. Most of the scattered points are concentrated within the relative azimuth angle range of 0 to 90 degrees, and the available computing resources within this range are relatively evenly distributed, mainly concentrated in the 0.3 to 0.8 range, indicating that most nearby interactive terminals have a certain availability of computing resources under relatively favorable orientations. In areas with larger azimuth angles (90 to 180 degrees), the available computing resources of nearby interactive terminals are more dispersed, with some terminals having available resources close to zero, indicating that terminals far from the ideal orientation have lower overall computing resource utilization or are under high load.

[0051] Overall, the graph shows no obvious linear or monotonic trend, but there is a certain non-uniform distribution of the relative azimuth angles and available computing resources of neighboring interactive terminals. This distribution characteristic is consistent with the processing mechanism for load negotiation response data of neighboring interactive terminals. That is, by combining the two-dimensional information of available computing resources and relative azimuth angles to construct a two-dimensional migration evaluation vector, the migration adaptability of interactive tasks is further calculated to achieve the goal of optimally allocating interactive requests to be migrated to neighboring terminals. The data in the graph serves as the input basis for the migration evaluation vector, reflecting the diversity of neighboring interactive terminals in terms of spatial orientation and resource status, and providing data support for subsequent adaptation ranking based on the reciprocal of the magnitude of the two-dimensional migration evaluation vector.

[0052] Example 4: In specific implementation, for each neighboring interactive terminal that returns negotiation response data, the operation method for constructing a two-dimensional migration evaluation vector is as follows: the trigger terminal parses the available computing resource value and relative azimuth angle of each neighboring interactive terminal from the received negotiation response data packet. The available computing resource value is expressed as follows: This indicates a dimensionless numerical value, ranging from 0 to 1. The relative azimuth angle is expressed as... The unit is degrees, and the value ranges from 0 to 180 degrees. The preset ideal azimuth angle is set to 0 degrees. The preset ideal azimuth angle represents the azimuth angle when the nearby interactive terminal is facing the trigger terminal. At this time, the directional antenna beam of the nearby interactive terminal and the communication link gain of the trigger terminal are maximized. The relative azimuth angle of the nearby interactive terminals is... The difference between the angle of the two positions is calculated at 0 degrees and the preset ideal orientation. The absolute value of the difference is then calculated. express, The unit is degrees. The constructed two-dimensional migration evaluation vector is based on... express, The first element is the value of available computing resources. The second dimension element is the absolute value of the difference. ,Right now .

[0053] The process of calculating the magnitude of the two-dimensional transfer evaluation vector is as follows: [The process involves dividing the two-dimensional transfer evaluation vector into...] Square each element separately, sum them, and then take the square root of the sum. The magnitude of the two-dimensional transfer evaluation vector is... The calculation formula is as follows: ,in, This represents the available computing resources of nearby interactive terminals and is a dimensionless value. This represents the absolute value of the difference between the relative azimuth angle of a nearby interactive terminal and the preset ideal azimuth angle, in degrees. The magnitude of the two-dimensional migration evaluation vector. The dimensions include the result of a mixture of dimensionless numerical values ​​and degrees.

[0054] Optionally, the operation method of using the reciprocal of the magnitude of the two-dimensional transfer evaluation vector as the initial value of the interactive task transfer fitness is as follows: Calculate The reciprocal of the interactive task migration adaptability. express, .when When approaching zero, the adaptability of interactive tasks during migration Approaching infinity, to avoid numerical overflow, an interactive task migration adaptation degree is set. The upper limit is 1000. Interactive task migration adaptability. The larger the value, the more suitable the nearby interactive terminal is for undertaking the migration task, because the available computing resources are higher. The larger the absolute value of the difference The smaller the value, the greater the magnitude of the two-dimensional transfer evaluation vector. The smaller the value, the better the adaptability to interactive task migration. The larger.

[0055] In some embodiments, the operation of sorting pending interaction requests according to their complexity from high to low is as follows: The terminal is triggered to read a list of pending interaction requests exceeding the service capacity threshold. Each pending interaction request carries a request type identifier and a request complexity parameter. The request complexity parameter is pre-set with different complexity scores according to the request type. The complexity score setting rules are as follows: text query requests have a complexity score of 1, image recognition requests have a complexity score of 3, video stream analysis requests have a complexity score of 5, and 3D model rendering requests have a complexity score of 8. The request complexity parameter of each pending interaction request is obtained, and the list of pending interaction requests is sorted from high to low using a quicksort algorithm to form a queue for allocation. The pending interaction request at the front of the queue has the highest request complexity, and the pending interaction request at the back has the lowest request complexity.

[0056] The operation method for selecting neighboring interactive terminals in descending order of interactive task migration adaptability is as follows: All neighboring interactive terminals that return negotiation response data are selected according to their interactive task migration adaptability. Sort the terminals from highest to lowest to form a candidate terminal list. Select the interactive task migration adaptability from the candidate terminal list. The nearest neighboring interactive terminal is selected as the current allocation target. The interactive request with the highest complexity is retrieved from the head of the allocation queue, marked as allocated, and its corresponding request identifier and content are written into the migration task list assigned to the current allocation target. Simultaneously, the available computing resources of the nearest interactive terminal are... The estimated resource consumption for this allocation is deducted. This estimated resource consumption is proportionally calculated based on the request complexity parameter. A pending interaction request with a complexity parameter of 8 consumes 0.2 available computing resources; a pending interaction request with a complexity parameter of 5 consumes 0.125 available computing resources; a pending interaction request with a complexity parameter of 3 consumes 0.075 available computing resources; and a pending interaction request with a complexity parameter of 1 consumes 0.025 available computing resources. The available computing resources of neighboring interactive terminals are then updated after the deduction. The remaining value. Remove the assigned pending interaction requests from the assignment queue.

[0057] Repeat the above allocation steps. Each time, take the interaction request with the highest complexity from the current head of the queue to be allocated, and select the current interaction task migration adaptation from the candidate terminal list. The highest and remaining available computing resource value Allocation is performed on neighboring interactive terminals with a value greater than zero. Allocation stops when the queue of terminals to be allocated becomes empty, or when the remaining available computing resources of all neighboring interactive terminals in the candidate terminal list reach a certain threshold. All resources have been exhausted, and the expected allocation queue is not empty. This is due to the remaining available computing resources of all neighboring interactive terminals. When allocation is exhausted and stops, and the queue to be allocated is still not empty, the terminal is triggered to re-initiate load negotiation broadcast to interactive terminals at a greater distance.

[0058] In some embodiments, the specific method for triggering the terminal to re-initiate load negotiation broadcast to a more distant interactive terminal is as follows: [The text abruptly shifts to a different topic] ... preset communication distance... An additional step value is added to the existing distance, set to 15 meters, to create a new preset communication distance. , The unit is meters. The triggering terminal uses its own spatial coordinates as the center and a new preset communication distance. The expanded circular area is defined by radius. The node identifiers of previously responding neighboring interactive terminals are added to an exclusion list. An exclusion list field is appended to the load negotiation broadcast message. Interactive terminals receiving the load negotiation broadcast message that find their own node identifier in the exclusion list will not respond. This triggers the terminal to send a new load negotiation broadcast message within the expanded circular area, repeating the complete process of receiving negotiation response data and calculating and allocating interactive task migration adaptation.

[0059] In practice, after distributing pending interaction requests exceeding the service capacity threshold to nearby interactive terminals, the triggering terminal sends an interaction request forwarding instruction to each assigned neighboring interactive terminal as follows: For each assigned neighboring interactive terminal, the triggering terminal generates an interaction request forwarding instruction. This instruction includes the original data packet of the pending interaction request and a session state snapshot generated by the triggering terminal for that request. The original data packet contains the original request message sent by the visitor's mobile terminal, carrying visitor identification information, request type, request parameters, and the visitor's mobile terminal's media access control address. The session state snapshot is generated by the triggering terminal for each pending interaction request during allocation confirmation, and includes the original session identifier, historical question-and-answer records, and an index of incomplete interaction queues. The triggering terminal then sends the interaction request forwarding instruction via unicast through the wireless ad hoc network to the corresponding neighboring interactive terminal.

[0060] In some embodiments, the steps for a nearby interactive terminal that receives an interaction request forwarding instruction to reconstruct the interaction session context based on a session state snapshot are as follows: The nearby interactive terminal parses the received session state snapshot and extracts the original session identifier, the historical question-and-answer records of interactions, and the index of incomplete interaction queues from the session state snapshot. The original session identifier is a globally unique identifier assigned when the triggering terminal establishes an interaction session with the visitor's mobile terminal, and the original session identifier is stored in string form. The historical question-and-answer records of interactions are the completed dialogue content between the triggering terminal and the visitor, and are stored in text sequence form. The index of incomplete interaction queues is the position number of the interaction request that the triggering terminal has not yet processed in the original processing queue. The nearby interactive terminal checks whether the original session identifier already exists in the local session management table. If it does not exist, it creates a new record in the local session management table, registers the original session identifier in the local session management table, and marks the registration status as active. The nearby interactive terminal allocates a contiguous storage area in the memory buffer, copies the historical question-and-answer records of interactions to the memory buffer in chronological order, and loads the task items corresponding to the index of incomplete interaction queues into the pending queue area in the memory buffer, thus completing the reconstruction of the interaction session context.

[0061] In practical implementation, the steps for the proximity interactive terminal to establish a direct interactive communication link with the visitor's mobile terminal based on the visitor identification information in the original data packet are as follows: The proximity interactive terminal extracts the Media Access Control (MAC) address and wireless signal strength of the visitor's mobile terminal from the original data packet. The MAC address is the unique hardware address of the visitor's mobile terminal's wireless network card, with a length of 48 bits. The proximity interactive terminal scans the surrounding wireless signals through its own wireless transceiver module to obtain the MAC address and received signal strength indication value corresponding to each wireless signal source. By traversing the scan results, the target wireless signal source that matches the MAC address in the original data packet is found, and the received signal strength indication value of the target wireless signal source is recorded. The proximity interactive terminal uses directional antenna beamforming technology. Directional antenna beamforming technology adjusts the phase and amplitude weights of each element in the antenna array to align the main lobe of the antenna radiation pattern with the direction of the target wireless signal source. The calculation basis for the antenna array phase weights is: determining the target direction angle based on the received signal angle of arrival estimation result, and calculating the optimal weight vector of each element using the minimum variance distortionless response algorithm. After the antenna array is adjusted according to the optimal weight vector, it forms a high-gain beam in the target direction, establishing a point-to-point wireless communication link with the visitor's mobile terminal. The nearby interactive terminal binds the link identifier of the point-to-point wireless communication link with the original session identifier in the reconstructed session context. The binding is achieved by writing the link identifier of the wireless communication link into the record corresponding to the original session identifier in the local session management table, thus completing the association between the interactive session and the communication channel.

[0062] Example 5: In specific implementation, after distributing pending interaction requests exceeding the service capacity threshold to neighboring interaction terminals, each terminal is triggered to start a load monitoring thread. This thread collects the real-time interaction request arrival rate of the triggering terminal at fixed intervals. The collection interval is set to 2 seconds. Within each collection interval, the triggering terminal counts the number of newly arriving interaction requests and divides this number by the collection interval duration of 2 seconds to obtain the real-time interaction request arrival rate. The real-time interaction request arrival rate is then... This indicates the rate at which real-time interactive requests arrive at the triggering terminal, measured in times per second. With service capacity threshold Comparison, service capacity threshold Number of processor cores Maximum request processing rate per single core The product of, in seconds.

[0063] When real-time interactive requests reach the rate Drop to service capacity threshold The following triggers the terminal to start a stable duration timer, with the timer starting at [time value]. The moment of initial establishment. The preset stability period is... express, The timer is set to 30 seconds, based on the assumption that the rate of interactive requests arriving remains below the service capacity threshold for 30 consecutive seconds, indicating that the load peak caused by visitor movement has definitively subsided and is not a temporary fluctuation. During the stable duration timer's counting process, if the timer reaches... It appeared again before In the event of a stable duration, the timer resets to zero and pauses, then waits again. Once the conditions are met, the stable duration timer will restart. The timer will restart when the accumulated value of the stable duration timer reaches the preset stable duration. When the terminal determines that the load of real-time interactive requests has dropped below the service capacity threshold and has remained stable for a preset duration, the load recovery conditions are met.

[0064] The triggering terminal sends a load recovery broadcast to all neighboring interactive terminals that have been assigned interaction requests as follows: The triggering terminal queries its locally maintained task migration record table, which records the node identifier and corresponding migration task list for each neighboring interactive terminal that has been assigned an interaction request. The triggering terminal constructs a load recovery broadcast message, which includes a message type identifier field, a triggering terminal node identifier field, a recovery command code field, and a migration task identifier list field. The migration task identifier list field contains globally unique interaction request identifiers corresponding to all interaction requests that need to be recovered. The interaction request identifier is composed of the original session identifier and the request sequence number. The triggering terminal broadcasts the load recovery broadcast message via the wireless ad hoc network, with a transmission radius covering the spatial range of all neighboring interactive terminals that have been assigned interaction requests.

[0065] In some embodiments, in response to a load recovery broadcast, each nearby interactive terminal that receives the load recovery broadcast performs the encapsulation and return operation of a session migration data packet. After receiving the load recovery broadcast message, the nearby interactive terminal parses the migration task identifier list field, iterates through each interaction request identifier in the migration task identifier list, and searches the local session management table for a session record that matches the interaction request identifier. For each matched session record, the nearby interactive terminal extracts the currently processed runtime status data, which includes the current state identifier of the dialogue state machine, the latest completed interaction step number, the prompt message content awaiting visitor response, and the remaining time value of the session timeout timer. The nearby interactive terminal extracts intermediate calculation results, which include a list of keywords extracted from visitor input, cached exhibit information data obtained from queries, and intermediate frame cache pointers for completed partial rendering tasks. The nearby interactive terminal serializes the runtime status data and intermediate calculation results according to a predefined serialization format and encapsulates them into a session migration data packet. The session migration data packet contains an interaction request identifier field, a runtime status data field, an intermediate calculation result field, and a data integrity verification field. The data integrity verification field uses a cyclic redundancy check code, calculated from the binary content of the running status data field and the intermediate calculation result field. The neighboring interactive terminal sends the encapsulated session migration data packet back to the triggering terminal via unicast, with the destination address being the network address corresponding to the triggering terminal's node identifier.

[0066] After receiving the session migration data packet, the triggering terminal resumes interaction control with the corresponding visitor's mobile terminal through the following steps: The triggering terminal performs a cyclic redundancy check on the received session migration data packet. If the check passes, it extracts the interaction request identifier from the packet and uses this identifier as an index to search for the corresponding original session information in the triggering terminal's local task recovery record. The triggering terminal parses the runtime status data field and sets the corresponding local session state machine to the same state based on the state identifier of the dialogue state machine. It then confirms the starting node for interaction recovery based on the latest completed interaction step number. The triggering terminal writes intermediate calculation results to its local buffer, allowing subsequent interaction processing to directly reuse intermediate results calculated by neighboring interaction terminals. After completing session context recovery, the triggering terminal notifies neighboring interaction terminals to release the corresponding session context and communication link resources.

[0067] In some embodiments, the triggering terminal notifies the neighboring interactive terminal to release the corresponding session context and communication link resources as follows: The triggering terminal sends a resource release instruction message to the neighboring interactive terminal. The resource release instruction message includes a message type identifier field, a triggering terminal node identifier field, a release confirmation code field, and a list of interaction request identifiers to be released field. After receiving the resource release instruction message, the neighboring interactive terminal searches for the corresponding record in the local session management table according to each interaction request identifier in the interaction request identifier list, and changes the status of the record from active to released. The neighboring interactive terminal clears the historical question-and-answer records and incomplete interaction queue indexes from the memory buffer, releasing the storage area occupied by the session in the memory buffer. The neighboring interactive terminal finds the link identifier of the wireless communication link bound to the interaction request identifier, actively disconnects the point-to-point wireless communication link with the visitor's mobile terminal through the wireless transceiver module, releases the phase weight configuration resources allocated by the directional antenna beamforming, and deletes the link identifier from the local active link table. After the resource release is completed, the neighboring interactive terminal returns a resource release confirmation message to the triggering terminal, and the triggering terminal updates the status of the corresponding migration task record in the local task migration record table to recycled.

[0068] In practical implementation, before real-time collection of visitor density data for each booth area and spatial coordinate data for each interactive terminal within the exhibition venue, the initialization configuration steps include the physical deployment of the interactive terminals. The method for deploying the interactive terminals in a hexagonal grid layout within the physical space of the exhibition venue is as follows: A rectangular outer frame covering the entire booth area is determined on the floor plan of the exhibition venue, and a corner point of the rectangular outer frame is selected as the layout origin. Starting from the layout origin, a grid with a side length of... A regular hexagonal grid, the side length of the regular hexagonal grid. The effective service radius of the interactive terminal is determined by measuring the signal coverage radius of the terminal's built-in wireless transceiver module in a typical indoor environment. The side length of the regular hexagonal grid is also considered. Take the effective service radius The interactive terminal is positioned at the geometric center of each regular hexagon. It is fixed to the ceiling of the venue via a slingshot, with its antenna panel facing the ground. Each interactive terminal is programmed with a unique node identifier at the factory; this identifier is a 48-bit binary number represented as a hexadecimal string. The initial service capacity threshold for each interactive terminal is written into its firmware based on its hardware configuration; this threshold is the product of the number of processor cores and the maximum request processing rate per core.

[0069] In some embodiments, the operation of synchronizing the global clock signal and a unified coordinate system reference point to all interactive terminals via the venue's central server is as follows: The venue's central server is equipped with a Global Navigation Satellite System timing module or a Network Time Protocol (NAT) client to obtain Coordinated Universal Time (UTC) as the global clock reference. During the initialization phase, the venue's central server broadcasts a clock synchronization message to all interactive terminals via wired Ethernet. The clock synchronization message contains the current timestamp value of the central server. After receiving the clock synchronization message, the interactive terminal calibrates its local clock counter against the received timestamp value to eliminate the offset between the local clock and the global clock. The unified coordinate system reference point is established as follows: A fixed reference point is selected within the exhibition venue, located at the center of the main entrance. The absolute coordinates of the fixed reference point in the venue's architectural coordinate system are measured. The absolute coordinates of the fixed reference point are used as the origin of the exhibition venue's three-dimensional spatial coordinate system. The X-axis is parallel to the main passageway of the venue, the Y-axis is perpendicular to the X-axis and points to the side of the venue, and the Z-axis is perpendicular to the ground and upwards. The coordinates of all positioning base stations are calibrated based on the unified coordinate system reference point.

[0070] Optionally, after each interactive terminal starts up, it reports its node identifier and spatial coordinate data to the central server as follows: After powering on, the interactive terminal first loads the node identifier from its firmware, initializes the ultra-wideband positioning module, establishes ranging communication with the fixed positioning base stations within the venue, and calculates its own spatial coordinate data. The interactive terminal sends a registration message to the central server via wired Ethernet or wireless network. The registration message includes a node identifier field, a spatial coordinate data field, and an initial service capacity threshold field. After receiving the registration messages reported by all interactive terminals, the central server collects the node identifiers and spatial coordinate data of all interactive terminals and constructs a topology map of the venue's interactive terminals.

[0071] The method for constructing the venue's interactive terminal topology is as follows: The central server creates a graph data structure. Each node in the node set of the graph data structure corresponds to an interactive terminal. The node attributes include the interactive terminal's node identifier, spatial coordinate data, and initial service capacity threshold. For any two interactive terminal nodes, the Euclidean distance between their spatial coordinate data is calculated. If the Euclidean distance is less than a preset adjacency distance threshold, an edge is established between the two nodes, with the edge weight set to the Euclidean distance. The preset adjacency distance threshold is set to the side length of a regular hexagonal grid. The graph data structure is twice the size of the target graph. After the graph data structure is established, the central server uses Dijkstra's algorithm to calculate the shortest path between any two interactive terminal nodes and appends the shortest path information to the graph data structure. The central server broadcasts the completed venue interactive terminal topology map to all interactive terminals. Each interactive terminal receives the map and stores it in its local memory for subsequent self-organizing network routing decisions between interactive terminals.

[0072] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application.

Claims

1. A method for intelligent interactive control of exhibitions, characterized in that, The method, applied to a distributed intelligent interactive node network deployed in a convention and exhibition center, includes: Real-time collection of visitor density data for each exhibition area within the convention and exhibition venue, as well as spatial coordinate data for each interactive terminal; Based on the visitor density data and the spatial coordinate data, a dynamic potential energy model of the visitor flow field is established, and the interaction request load of each interactive terminal in the future time window is predicted based on the dynamic potential energy model. When the predicted interaction request load of any interactive terminal exceeds the service capacity threshold of that interactive terminal, the interactive terminal is triggered to send a load negotiation broadcast to neighboring interactive terminals. In response to the load negotiation broadcast, the system receives negotiation response data returned by a nearby interactive terminal. The negotiation response data includes the available computing resources of the nearby interactive terminal and the relative azimuth angle between the terminal and the triggering terminal. Based on the available computing resources and relative azimuth angles returned by each neighboring interactive terminal, the interaction task migration adaptation degree of each neighboring interactive terminal is calculated. Then, according to the interaction task migration adaptation degree from high to low, the pending interaction requests that exceed the service capacity threshold are sequentially allocated to each neighboring interactive terminal.

2. The intelligent interactive control method for exhibitions according to claim 1, characterized in that, Real-time collection of visitor density data for each exhibition booth area within the exhibition venue, as well as spatial coordinate data for each interactive terminal, specifically: The millimeter-wave radar array deployed on the ceiling of the exhibition hall scans the area of ​​each booth to obtain raw point cloud data; The original point cloud data is clustered, and after removing the static background point cloud, the number of dynamic point clouds in each booth area is counted. This number of dynamic point clouds is used as the visitor density data. Meanwhile, by using the ultra-wideband positioning module built into each interactive terminal to communicate with the fixed positioning base station in the venue, the spatial coordinate data of each interactive terminal in the three-dimensional spatial coordinate system of the exhibition venue can be calculated.

3. The intelligent interactive control method for exhibitions according to claim 2, characterized in that, The scanning frequency of the millimeter-wave radar array is dynamically adjusted according to the real-time visitor density of each booth area; the higher the density, the higher the scanning frequency.

4. The intelligent interactive control method for exhibitions according to claim 2, characterized in that, Based on the visitor density data and the spatial coordinate data, a dynamic potential energy model of the visitor flow field is established. Based on this dynamic potential energy model, the interaction request load at the location of each interactive terminal within a future time window is predicted, specifically: The spatial coordinate data of each interactive terminal is mapped onto a two-dimensional booth plan to form an interactive terminal distribution grid. For each booth area, its visitor density data is used as the flow field mass value of that area at the current moment, and the flow velocity vector of visitors spreading from high-density areas to low-density areas is calculated based on the flow field mass value gradient between adjacent booth areas. Substitute the spatial coordinate data of each interactive terminal into the dynamic potential energy function constructed by the flow field mass value and flow velocity vector of all booth areas, and solve for the potential energy gradient change rate at the location of the interactive terminal. Using the potential energy gradient change rate as the interaction request excitation factor, and combining it with the historical interaction request arrival rate of the interactive terminal, an autoregressive moving average model is used to predict the interaction request load within the future time window.

5. The intelligent interactive control method for exhibitions according to claim 4, characterized in that, The dynamic potential energy function introduces a time decay factor to reduce the weight of the influence of historical visitor density data on the current potential energy gradient change rate.

6. The intelligent interactive control method for exhibitions according to claim 4, characterized in that, When the predicted interaction request load of any interactive terminal exceeds the service capacity threshold of that interactive terminal, the interactive terminal is triggered to send a load negotiation broadcast to neighboring interactive terminals, specifically as follows: The service capacity threshold of this interactive terminal is set to be the product of the number of its processor cores and the maximum request processing rate of a single core; When the predicted interactive request load exceeds the service capacity threshold, the interactive terminal is marked as a triggering terminal. The triggering terminal sends a load negotiation broadcast message containing the number and type of interaction requests to be migrated to all interactive terminals within a circular area centered on its own spatial coordinates and with a preset communication distance as the radius, through a wireless ad hoc network.

7. The intelligent interactive control method for exhibitions according to claim 6, characterized in that, In response to the load negotiation broadcast, negotiation response data returned by a nearby interactive terminal is received. This negotiation response data includes the available computing resources of the nearby interactive terminal and the relative azimuth angle between the terminal and the triggering terminal. Specifically: After receiving the load negotiation broadcast message, the nearby interactive terminal reads its current CPU utilization rate and remaining memory capacity, and performs a weighted fusion calculation on the CPU utilization rate and remaining memory capacity to obtain the available computing resource value. The neighboring interactive terminal obtains its own orientation angle through its built-in electronic compass, and at the same time receives the orientation angle of the trigger terminal broadcast by the trigger terminal. Based on the difference between its own orientation angle and the orientation angle of the trigger terminal, and combined with the direction of the line formed by its own spatial coordinate data and the spatial coordinate data of the trigger terminal, the relative azimuth angle is calculated. The neighboring interactive terminal encapsulates the available computing resource value and the relative azimuth angle into a negotiation response data message and returns it to the triggering terminal along the original path.

8. The intelligent interactive control method for exhibitions according to claim 7, characterized in that, Based on the available computing resources returned by each neighboring interactive terminal and the relative azimuth angle, the interaction task migration adaptation degree of each neighboring interactive terminal is calculated. Then, according to the interaction task migration adaptation degree from high to low, the pending interaction requests exceeding the service capacity threshold are sequentially allocated to each neighboring interactive terminal. Specifically: For each neighboring interactive terminal that returns negotiation response data, a two-dimensional migration evaluation vector is constructed, where the first element is the available computing resource value of the neighboring interactive terminal, and the second element is the absolute value of the difference between the relative azimuth angle of the neighboring interactive terminal and the preset ideal azimuth angle. Calculate the magnitude of the two-dimensional migration evaluation vector, and use the reciprocal of the magnitude as the initial value of the interaction task migration adaptation degree of the neighboring interactive terminal; The pending interaction requests are sorted from highest to lowest complexity to form a queue for assignment. Select neighboring interactive terminals in descending order of their interactive task migration adaptability. Assign the interactive request with the highest request complexity in the queue to the currently selected neighboring interactive terminal and remove the request from the queue. Repeat this assignment process until the queue is empty or all available computing resources of neighboring interactive terminals are exhausted.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by the processor, it implements a smart interactive control method for exhibitions as described in any one of claims 1 to 8.

10. A smart interactive control device for exhibitions, comprising a processor and a memory, wherein the memory stores a computer program, characterized in that, When the computer program is executed by the processor, it implements a smart interactive control method for exhibitions as described in any one of claims 1 to 8.

Images