An AI-assisted exhibition route design method and system
By using AI to assist in the design of exhibition hall circulation and utilizing multi-source data and graph neural networks to simulate the dynamics of pedestrian flow, the problem of difficulty in quantifying the flow patterns of people in traditional exhibition hall design has been solved, and efficient and dynamic pedestrian flow management and path optimization have been achieved within the exhibition hall.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- SHENZHEN KASTONE INTELLIGENT TECHNOLOGY DEVELOPMENT CO LTD
- Filing Date
- 2026-02-09
- Publication Date
- 2026-06-02
AI Technical Summary
Traditional exhibition hall flow design relies on personal experience and cannot quantify and analyze crowd flow patterns, resulting in congestion in popular exhibition areas and sparse crowds in less popular areas. Furthermore, it cannot be dynamically adjusted, and the design scheme lacks data support.
An AI-assisted exhibition hall circulation design method is adopted, which uses multi-source data fusion, spatiotemporal modeling and graph neural network to simulate the dynamic flow of people, and combines reinforcement learning to optimize the exhibition hall layout and generate the optimal visitor path.
It enables accurate prediction and dynamic optimization of visitor flow in the exhibition hall, effectively avoiding congestion, improving booth coverage and venue utilization, and meeting the needs of visitors.
Smart Images
Figure CN122133228A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of exhibition hall circulation design technology, and in particular to an AI-assisted exhibition hall circulation design method and system. Background Technology
[0002] The core objective of exhibition hall circulation design is to guide visitors to tour the exhibition hall efficiently and in an orderly manner by rationally planning visitor paths and optimizing the layout of booths and facilities, while ensuring the smooth operation and safety of the exhibition hall. In all types of exhibition halls, circulation design directly affects the performance of the core functions of the exhibition hall and is a key indicator for measuring the level of exhibition hall design.
[0003] Traditional exhibition hall circulation design mainly relies on the designer's personal experience and static planning. Designers usually lay out the exhibits based on their importance and spatial aesthetics, making it difficult to quantify and predict the actual flow of large crowds. This results in congestion in popular exhibition areas, sparse crowds in less popular areas, and low overall browsing efficiency. Furthermore, traditional methods cannot utilize historical and real-time visitor flow data for dynamic adjustments, which has limitations.
[0004] The reason for this problem is that traditional design patterns cannot quantify and analyze crowd flow patterns through historical and real-time pedestrian flow data, resulting in a lack of data support for design solutions. Furthermore, existing design tools do not have the ability to simulate and predict large-scale crowd flow, making it difficult to anticipate congestion and other problems in advance, thus presenting limitations. Summary of the Invention
[0005] The purpose of this invention is to provide an AI-assisted exhibition hall circulation design method and system to solve the problems mentioned in the background art.
[0006] To achieve the above objectives, the present invention provides the following technical solution: an AI-assisted exhibition hall circulation design method and system, comprising the following steps; S1. Multi-source data is fused and spatially discretized to obtain basic spatial information of the target exhibition hall. Historical pedestrian flow distribution data is fused to discretize the continuous space of the exhibition hall into a grid model composed of multiple unit grids. S2. Construct a spatiotemporal state graph. Based on the grid model, construct an initial graph structure for the exhibition hall with unit grids as nodes and inter-grid connectivity as edges. Node features include static and dynamic attributes. S3. Simulate the dynamics of pedestrian flow by inputting the spatiotemporal state map into the graph neural network. The graph neural network simulates the movement, gathering and diffusion of people in the exhibition hall through the message passing mechanism, and outputs the predicted pedestrian density and path prediction flow. S4. Define the reinforcement learning optimization objective and optimization, transform the layout optimization problem into a reinforcement learning task, and define the state, action and comprehensive optimization function. S5, AI reinforcement learning: AI obtains the optimal layout scheme by performing actions in a simulation environment and updating the strategy according to the optimization function; S6. Generate the optimal visitor route. Based on the optimized layout and improved route planning, generate the visitor route and output the visualization.
[0007] Preferably, in step S1, the basic spatial information of the target exhibition hall includes the exhibition hall floor plan, size parameters, entrance and exit locations, and distribution of fixed facilities; historical pedestrian flow distribution data comes from sensor monitoring, ticketing system statistics, and historical image analysis; the grid model is divided into uniform grids, each grid has a unique identifier, and its coordinate position on the exhibition hall floor plan is recorded.
[0008] Preferably, in step S2, the static attributes of the nodes include whether the grid is a booth, entrance / exit, or obstacle area; the dynamic attributes include the real-time pedestrian density of the corresponding grid and the historical pedestrian density value of the grid in the same period. The edges between nodes represent the direct passage relationship between adjacent grids, and the weight of the edges is preset according to the passage convenience between the grids.
[0009] Preferably, in step S3, the graph neural network adopts a multi-layer graph convolutional network architecture, which aggregates the features of adjacent nodes layer by layer through a multi-layer message passing mechanism, and outputs the predicted pedestrian density and the predicted traffic flow of the main path for each grid.
[0010] Preferably, in step S4, the state includes the current exhibition hall layout and the pedestrian flow prediction results; the actions include adjusting the booth locations, entrance and exit locations and key passage widths, and the comprehensive optimization function consists of a congestion penalty term, a coverage optimization term and a popular booth dispersion optimization term.
[0011] Preferably, the congestion penalty item is determined based on the distribution and density value of high pedestrian density areas, the coverage optimization item is determined based on the proportion of booths accessible to visitors, and the popular booth dispersion optimization item is determined based on the uniformity of the spatial distribution of popular booths.
[0012] Preferably, in step S5, AI is used to repeatedly perform actions selection, pedestrian flow state prediction, optimization calculation and strategy update operations in a simulation environment, and a layout scheme that maximizes long-term cumulative optimization is obtained through multiple iterations.
[0013] Preferably, in step S6, multiple alternative routes are generated with the goal of minimizing visitor congestion by introducing a predicted crowd density factor, and the optimized exhibition hall layout map, predicted crowd cloud map, and recommended visit routes are displayed in a graphical interface.
[0014] Preferably, the system includes; The data acquisition module is used to acquire historical heat maps of pedestrian flow, real-time location data and basic information of the exhibition hall space, and to complete data cleaning and normalization processing. The spatial discretization construction module is used to divide the exhibition space into a uniform grid and record the adjacency relationship, while generating a spatiotemporal state diagram structure containing static and dynamic attributes based on the grid. The graph neural network crowd flow simulation module is based on a spatiotemporal state graph input graph neural network. It simulates crowd flow dynamics through a message passing mechanism and outputs predicted crowd flow density and path flow. The reinforcement learning optimization module configures the state, actions and comprehensive optimization functions of reinforcement learning, trains the AI to iteratively optimize in a simulation environment, and finally outputs the optimal exhibition hall layout scheme. The path planning module generates the optimal visitor path based on the optimized layout and predicted visitor flow through an improved path planning algorithm. The visualization output module presents the optimized layout, pedestrian flow prediction cloud map, and recommended routes in a graphical interface.
[0015] The technical effects and advantages of this invention are as follows: 1. This invention, through multi-source data fusion and spatiotemporal modeling technology, combined with AI intelligent optimization, can predict the dynamic flow of people in the exhibition hall, adaptively optimize the layout and visitor path, effectively avoid congestion, improve booth coverage and venue utilization, and meet user needs.
[0016] 2. Based on the construction of a spatiotemporal state map using a grid model and combined with the message passing mechanism of a graph neural network, it is possible to accurately simulate the movement, gathering and diffusion patterns of crowds within the exhibition hall, thereby achieving accurate prediction of crowd density and path flow, and providing a reliable decision-making basis for dynamic optimization. Attached Figure Description
[0017] Figure 1 This is an overall flowchart of the present invention; Figure 2 This is a flowchart of the spatial discretization and map construction process of the present invention; Figure 3 This is a flowchart of the graph neural network-based crowd flow simulation of the present invention; Figure 4 This is a flowchart illustrating the reinforcement learning optimization process of this invention. Figure 5 This is a flowchart illustrating the optimal path generation and visualization process of this invention. Figure 6 This is a diagram showing the system composition and data flow of the present invention. Detailed Implementation
[0018] 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, and 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.
[0019] This invention provides, for example Figure 1 The present invention relates to an AI-assisted exhibition hall circulation design method and system, comprising the following steps; S1. Multi-source data is fused and spatially discretized to obtain basic spatial information of the target exhibition hall. Historical pedestrian flow distribution data is fused to discretize the continuous space of the exhibition hall into a grid model composed of multiple unit grids. S2. Construct a spatiotemporal state graph. Based on the grid model, construct an initial graph structure for the exhibition hall with unit grids as nodes and inter-grid connectivity as edges. Node features include static and dynamic attributes. S3. Simulate the dynamics of pedestrian flow by inputting the spatiotemporal state map into the graph neural network. The graph neural network simulates the movement, gathering and diffusion of people in the exhibition hall through the message passing mechanism, and outputs the predicted pedestrian density and path prediction flow. S4. Define the reinforcement learning optimization objective and optimization, transform the layout optimization problem into a reinforcement learning task, and define the state, action and comprehensive optimization function. S5, AI reinforcement learning: AI obtains the optimal layout scheme by performing actions in a simulation environment and updating the strategy according to the optimization function; S6. Generate the optimal visitor route. Based on the optimized layout and improved route planning, generate the visitor route and output the visualization.
[0020] Example 1: In this example, relevant data about the exhibition hall is first collected, including the hall's floor plan, size, entrances and exits, and the locations of fixed facilities such as restrooms and service counters. Then, historical pedestrian flow data is collected, such as the number of people monitored by sensors, the number of visitors entering the park through the ticketing system, and the pedestrian distribution analyzed from historical surveillance videos. The entire exhibition hall space is divided into several small grids with unique numbers and location information, creating a grid model. Based on this grid model, a map reflecting changes in space and pedestrian flow is built. Each grid is a node, recording whether it is an exhibition booth, entrance / exit, or an inaccessible obstacle zone, as well as the real-time and historical pedestrian flow density. Whether grids are traversable is used as a connecting edge; the easier the passage, the higher the edge weight. This map is then input into an AI model, allowing the model to simulate where people will be most numerous and which routes will be crowded in the future, outputting pedestrian flow prediction results. Subsequently, the problem of optimizing the exhibition hall layout and routes is transformed into a task where the AI continuously tries to adjust and find the optimal solution, clearly defining the state the AI needs to focus on, i.e., the current state of the exhibition hall. The AI's reward and punishment system is comprised of three components: layout and visitor flow prediction, possible adjustments (such as moving booth locations, opening or adjusting entrances and exits, widening or narrowing key passageways), optimization rules for judging the effectiveness of adjustments (deducting points for reduced congestion, adding points for higher visitor coverage, and evenly distributing popular booths to avoid overcrowding), and a virtual simulation environment similar to the real exhibition hall. During the AI learning phase, the AI selects adjustment actions according to certain rules, applies these actions to the virtual exhibition hall, and then re-predicts visitor flow using the previous model. The score for each adjustment is calculated based on the reward and punishment criteria, and the AI updates its adjustment strategy. After hundreds to thousands of trials, the AI finds the optimal exhibition hall layout that maximizes the total score. Finally, based on this optimal layout and visitor flow prediction, several suitable visiting routes are planned to minimize the probability of congestion. The optimized exhibition hall layout, visitor flow heatmap, and recommended routes are then displayed visually on screens within the exhibition area or on a public WeChat account. Visitors can access these recommended routes by adding the WeChat account, meeting their needs.
[0021] This invention provides, for example Figures 2 to 6 The present invention relates to an AI-assisted exhibition hall circulation design method and system, comprising the following steps; S1. Multi-source data is fused and spatially discretized to obtain basic spatial information of the target exhibition hall. Historical pedestrian flow distribution data is fused to discretize the continuous space of the exhibition hall into a grid model composed of multiple unit grids. S2. Construct a spatiotemporal state graph. Based on the grid model, construct an initial graph structure for the exhibition hall with unit grids as nodes and inter-grid connectivity as edges. Node features include static and dynamic attributes. S3. Simulate the dynamics of pedestrian flow by inputting the spatiotemporal state map into the graph neural network. The graph neural network simulates the movement, gathering and diffusion of people in the exhibition hall through the message passing mechanism, and outputs the predicted pedestrian density and path prediction flow. S4. Define the reinforcement learning optimization objective and optimization, transform the layout optimization problem into a reinforcement learning task, and define the state, action and comprehensive optimization function. S5, AI reinforcement learning: AI obtains the optimal layout scheme by performing actions in a simulation environment and updating the strategy according to the optimization function; S6. Generate the optimal visitor route. Based on the optimized layout and improved route planning, generate the visitor route and output the visualization.
[0022] In step S1, the basic spatial information of the target exhibition hall includes the exhibition hall floor plan, size parameters, entrance and exit locations, and distribution of fixed facilities; historical pedestrian flow distribution data comes from sensor monitoring, ticketing system statistics, and historical image analysis; the grid model is divided into uniform grids, each grid has a unique identifier, and its coordinate position on the exhibition hall floor plan is recorded.
[0023] In step S2, the static attributes of the nodes include whether the grid is a booth, entrance / exit, or obstacle area; the dynamic attributes include the real-time pedestrian density of the corresponding grid and the historical pedestrian density value of the grid in the same period. The edges between nodes represent the direct passage relationship between adjacent grids, and the weight of the edges is preset according to the passage convenience between the grids.
[0024] In step S3, the graph neural network adopts a multi-layer graph convolutional network architecture, which aggregates the features of adjacent nodes layer by layer through a multi-layer message passing mechanism, and outputs the predicted pedestrian density and the predicted traffic flow of the main path for each grid.
[0025] In step S4, the state includes the current exhibition hall layout and the pedestrian flow prediction results; the actions include adjusting the booth locations, entrance and exit locations and key passage widths. The comprehensive optimization function consists of a congestion penalty term, a coverage optimization term, and a popular booth dispersion optimization term.
[0026] The congestion penalty item is determined based on the distribution and density value of high-traffic-density areas, the coverage optimization item is determined based on the proportion of booths accessible to visitors, and the popular booth dispersion optimization item is determined based on the uniformity of the spatial distribution of popular booths.
[0027] In step S5, AI performs actions selection, pedestrian flow prediction, optimization calculation, and strategy update operations in a simulation environment in a loop, and obtains a layout scheme that maximizes long-term cumulative optimization through multiple iterations.
[0028] In step S6, by introducing the predicted crowd density factor, multiple alternative routes are generated with the goal of minimizing the congestion experience of the audience, and the optimized exhibition hall layout map, predicted crowd cloud map and recommended visiting routes are displayed in a graphical interface.
[0029] The system includes; The data acquisition module is used to acquire historical heat maps of pedestrian flow, real-time location data and basic information of the exhibition hall space, and to complete data cleaning and normalization processing. The spatial discretization construction module is used to divide the exhibition space into a uniform grid and record the adjacency relationship, while generating a spatiotemporal state diagram structure containing static and dynamic attributes based on the grid. The graph neural network crowd flow simulation module is based on a spatiotemporal state graph input graph neural network. It simulates crowd flow dynamics through a message passing mechanism and outputs predicted crowd flow density and path flow. The reinforcement learning optimization module configures the state, actions and comprehensive optimization functions of reinforcement learning, trains the AI to iteratively optimize in a simulation environment, and finally outputs the optimal exhibition hall layout scheme. The path planning module generates the optimal visitor path based on the optimized layout and predicted visitor flow through an improved path planning algorithm. The visualization output module presents the optimized layout, pedestrian flow prediction cloud map, and recommended routes in a graphical interface.
[0030] Example 2: In this example, the data acquisition module installs infrared sensors and high-definition cameras at key locations within the exhibition hall, then connects to the ticketing system to collect past visitor heatmaps, real-time personnel location information, and spatial information of the exhibition hall itself. This data, including visitor density and exhibition hall dimensions, is then formatted and stored in a database for later use. The spatial discretization module divides the exhibition hall space into several small grids, mapping the attributes of each grid to visitor data to create a spatial and visitor flow change map. The graph neural network visitor flow simulation module uses a professional AI framework to build an analysis model. The generated map is input, allowing the model to learn the relationships between grids and visitor flow patterns. After training, it can output predicted data on future visitor flow and route congestion. The reinforcement learning optimization module uses an AI framework... The learning model is built by setting the states the AI should focus on, the adjustment actions it can take, and the reward and punishment standards. A virtual exhibition hall simulation environment is then created, allowing the AI model to interact with this virtual environment, continuously trying adjustments, calculating scores, and updating strategies. After training, it outputs the optimal exhibition hall layout. The path planning module takes into account the crowd prediction level. After inputting the optimal layout and crowd prediction data, it outputs several visitor routes that minimize congestion, providing the route's location, length, and approximate visit time. The visualization output module then visually displays the optimized exhibition hall layout, crowd heatmap, and recommended routes, either on displays within the exhibition area or on a public WeChat account. Visitors can obtain recommended routes by adding the WeChat account, meeting their needs.
[0031] Finally, it should be noted that the above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. An AI-assisted exhibition hall circulation design method, characterized in that, Includes the following steps: S1. Multi-source data is fused and spatially discretized to obtain basic spatial information of the target exhibition hall. Historical pedestrian flow distribution data is fused to discretize the continuous space of the exhibition hall into a grid model composed of multiple unit grids. S2. Construct a spatiotemporal state graph. Based on the grid model, construct an initial graph structure for the exhibition hall with unit grids as nodes and inter-grid connectivity as edges. Node features include static and dynamic attributes. S3. Simulate the dynamics of pedestrian flow by inputting the spatiotemporal state map into the graph neural network. The graph neural network simulates the movement, gathering and diffusion of people in the exhibition hall through the message passing mechanism, and outputs the predicted pedestrian density and path prediction flow. S4. Define the reinforcement learning optimization objective and optimization, transform the layout optimization problem into a reinforcement learning task, and define the state, action and comprehensive optimization function. S5, AI reinforcement learning: AI obtains the optimal layout scheme by performing actions in a simulation environment and updating the strategy according to the optimization function; S6. Generate the optimal visitor route. Based on the optimized layout and improved route planning, generate the visitor route and output the visualization.
2. The AI-assisted exhibition hall circulation design method according to claim 1, characterized in that, In step S1, the basic spatial information of the target exhibition hall includes the exhibition hall floor plan, size parameters, entrance and exit locations, and distribution of fixed facilities; historical pedestrian flow distribution data comes from sensor monitoring, ticketing system statistics, and historical image analysis; the grid model is divided into uniform grids, each grid has a unique identifier, and its coordinate position on the exhibition hall floor plan is recorded.
3. The AI-assisted exhibition hall circulation design method according to claim 1, characterized in that, In step S2, the static attributes of the nodes include whether the grid is a booth, entrance / exit, or obstacle area; the dynamic attributes include the real-time pedestrian density of the corresponding grid and the historical pedestrian density value of the grid in the same period. The edges between nodes represent the direct passage relationship between adjacent grids, and the weight of the edges is preset according to the passage convenience between the grids.
4. The AI-assisted exhibition hall circulation design method according to claim 1, characterized in that, In step S3, the graph neural network adopts a multi-layer graph convolutional network architecture, which aggregates the features of adjacent nodes layer by layer through a multi-layer message passing mechanism, and outputs the predicted pedestrian density and the predicted traffic flow of the main path for each grid.
5. The AI-assisted exhibition hall circulation design method according to claim 1, characterized in that, In step S4, the state includes the current exhibition hall layout and the pedestrian flow prediction results; the actions include adjusting the booth locations, entrance and exit locations and key passage widths. The comprehensive optimization function consists of a congestion penalty term, a coverage optimization term, and a popular booth dispersion optimization term.
6. The AI-assisted exhibition hall circulation design method according to claim 5, characterized in that, The congestion penalty item is determined based on the distribution and density value of high-traffic-density areas, the coverage optimization item is determined based on the proportion of booths accessible to visitors, and the popular booth dispersion optimization item is determined based on the uniformity of the spatial distribution of popular booths.
7. The AI-assisted exhibition hall circulation design method according to claim 1, characterized in that, In step S5, AI performs actions selection, pedestrian flow prediction, optimization calculation, and strategy update operations in a simulation environment in a loop, and obtains a layout scheme that maximizes long-term cumulative optimization through multiple iterations.
8. The AI-assisted exhibition hall circulation design method according to claim 1, characterized in that, In step S6, by introducing the predicted crowd density factor, multiple alternative routes are generated with the goal of minimizing the congestion experience of the audience, and the optimized exhibition hall layout map, predicted crowd cloud map and recommended visiting routes are displayed in a graphical interface.
9. An AI-assisted exhibition hall circulation design system, based on the AI-assisted exhibition hall circulation design method according to any one of claims 1-8, characterized in that, The system includes: The data acquisition module is used to acquire historical heat maps of pedestrian flow, real-time location data and basic information of the exhibition hall space, and to complete data cleaning and normalization processing. The spatial discretization construction module is used to divide the exhibition space into a uniform grid and record the adjacency relationship, while generating a spatiotemporal state diagram structure containing static and dynamic attributes based on the grid. The graph neural network crowd flow simulation module is based on a spatiotemporal state graph input graph neural network. It simulates crowd flow dynamics through a message passing mechanism and outputs predicted crowd flow density and path flow. The reinforcement learning optimization module configures the state, actions and comprehensive optimization functions of reinforcement learning, trains the AI to iteratively optimize in a simulation environment, and finally outputs the optimal exhibition hall layout scheme. The path planning module generates the optimal visitor path based on the optimized layout and predicted visitor flow through an improved path planning algorithm. The visualization output module presents the optimized layout, pedestrian flow prediction cloud map, and recommended routes in a graphical interface.