Intelligent seat management system and method
By combining multimodal perception with pressure and infrared sensors and an LSTM prediction model with a collaborative filtering algorithm, the problems of low seat-finding efficiency, prominent seat-hogging, and blind management in university library seat management have been solved, achieving fair allocation and efficient management of seat resources.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- ANHUI POLYTECHNIC UNIV MECHANICAL & ELECTRICAL COLLEGE
- Filing Date
- 2026-02-12
- Publication Date
- 2026-06-02
AI Technical Summary
The existing seat management systems in public learning spaces such as university libraries cannot accurately distinguish between real people sitting in seats and objects occupying seats, resulting in low efficiency in finding seats, prominent seat-hogging, lack of data support for management decisions, unfair resource allocation, and a lack of predictive guidance.
Employing a multimodal sensing mechanism combining pressure and infrared sensors, along with collaborative filtering algorithms and LSTM time series prediction models, the system achieves accurate identification and personalized recommendations of seat availability, provides real-time seat vacancy probability prediction, and allows interaction via WeChat mini-program and data dashboard.
It effectively reduces invalid seat reservations, improves seat-finding efficiency and the fairness of resource allocation, provides dynamic decision-making basis, and enhances management efficiency and user experience.
Smart Images
Figure CN122133986A_ABST
Abstract
Description
Technical Field
[0001] This invention belongs to the fields of Internet of Things and artificial intelligence technology, specifically relating to a smart seat management system and method based on multimodal perception and intelligent prediction. Background Technology
[0002] Currently, seating management in public learning spaces such as university libraries mainly adopts traditional manual management or simple reservation systems. Traditional manual management relies on administrators to patrol, remove items reserving seats, and mediate disputes, requiring a significant investment of manpower. Simple reservation systems can only achieve basic reservation and release functions, lacking real-time status awareness capabilities, unable to effectively distinguish between "real people sitting" and "items reserving seats," and unable to cope with issues such as users not showing up and differences in seat usage between peak and off-peak periods.
[0003] Existing technologies have significant drawbacks: First, students find seats inefficiently, with peak periods often resulting in a shortage of seats and a significant waste of time. Second, seat hogging is widespread, undermining the fairness of resource allocation and causing disputes among students. Third, management decisions lack data support, relying on experience-based judgments for spatial layout optimization and opening hours adjustments, leading to low management efficiency. Fourth, existing visualization systems are lagging in data updates, only displaying the current status and failing to provide predictive guidance, thus failing to fundamentally balance the supply and demand imbalance of seats.
[0004] Therefore, there is an urgent need for a seat management system and method to avoid the problems of low seat-finding efficiency, prominent seat-hogging, blind management decisions, and lack of predictive guidance in the existing public space seat management. Summary of the Invention
[0005] The purpose of this invention is to provide a smart seat management system and method based on multimodal perception and intelligent prediction to solve the problems of low seat search efficiency, prominent seat hogging, blind management decisions, and lack of predictive guidance.
[0006] Based on the above concept, the technical solution adopted by this invention is as follows: According to a first aspect of the present invention, a smart seat management system is provided, comprising: The sensing module is used to collect seat occupancy status through pressure sensors deployed on the seat and infrared array sensors, and to collect user identity information through NFC tags. The communication module is used to transmit the data collected by the sensing module; The data processing and algorithm module is used to receive and process the data, generate personalized seat recommendation results based on the collaborative filtering algorithm, and generate seat vacancy probability prediction results based on the LSTM time series prediction model. The application interaction module is used to display the recommendation results and prediction results to the user and receive the user's operation instructions.
[0007] In some embodiments, the system uses a dual verification mechanism of pressure sensors and infrared sensors to distinguish between people sitting in the seats and objects occupying the seats.
[0008] In some embodiments, the data processing and algorithm module is further configured to construct a user-item rating matrix based on the user's historical behavior data, as input to the collaborative filtering algorithm.
[0009] In some embodiments, the data processing and algorithm module is further configured to train the LSTM model with continuous historical seat usage data of not less than 3 months, wherein the historical data includes at least the occupancy period, duration, and regional information.
[0010] In some embodiments, the application interaction module includes a user-side WeChat mini-program and an administrator-side data dashboard, which is used to display regional utilization rate, turnover rate, and peak-hour analysis reports generated based on seat usage data.
[0011] According to a second aspect of the present invention, a smart seat management method is provided, comprising the following steps: S1. Real-time acquisition of multimodal sensing data of the seat, the data including at least pressure data, infrared thermal data, and user authentication data; S2. Clean and merge the collected data to determine the actual occupancy status of the seats; S3. Based on the processed historical and real-time data, personalized seat recommendation information and future seat availability probability information are generated through machine learning models. S4. Push the recommendation information and idle probability information to the user terminal; S5. Update the seat status and record user behavior data based on the operation instructions fed back by the user terminal.
[0012] In some embodiments, determining the actual occupancy status of a seat includes: when the pressure data exceeds a first threshold and the infrared thermal sensing data is valid, determining that a person is seated; when the pressure data exceeds the first threshold but the infrared thermal sensing data is invalid, determining that an object is occupying the seat.
[0013] In some embodiments, generating personalized seat recommendation information specifically includes: calculating and ranking the recommendation scores for each seat based on the user's historical preference data and the behavioral data of similar users using a collaborative filtering algorithm.
[0014] In some embodiments, generating seat vacancy probability information for future time periods specifically includes: organizing historical occupancy status data in a time series manner, inputting it into an LSTM prediction model, and outputting the vacancy probability level of each seat within the next 30-60 minutes.
[0015] In some embodiments, a user's credit score is updated based on the user's historical performance record, and usage restrictions are imposed on users whose credit scores are below a threshold.
[0016] The beneficial effects of this invention are as follows: 1. This invention uses a dual verification mechanism of pressure sensor and infrared sensor to accurately distinguish between "real people sitting" and "objects occupying seats". Combined with a collaborative filtering recommendation algorithm based on user behavior, it effectively reduces invalid seat occupation, realizes fair allocation and rapid matching of seat resources, and improves users' seat search efficiency and user experience.
[0017] 2. This invention uses an LSTM time series model to predict and analyze seat usage data, and combines it with a continuously collected data closed-loop optimization model to provide managers with dynamic and quantitative decision-making basis, realizing advance prediction and dynamic scheduling of seat resource supply and demand, and improving space management efficiency and long-term resource utilization. Attached Figure Description
[0018] Figure 1 This is a technical architecture diagram of the intelligent seat management system of the present invention; Figure 2 This is a flowchart of the core user operation of the intelligent seat management system of the present invention; Figure 3 This is a schematic diagram illustrating the working principle of the AI model of this invention. Detailed Implementation
[0019] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numerals in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Rather, they are merely examples of apparatuses and methods consistent with some aspects of the invention as detailed in the appended claims.
[0020] The terminology used in this application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. Unless otherwise defined, the technical or scientific terms used in this application should be understood in their ordinary sense by one of ordinary skill in the art to which this invention pertains. The words “a” or “one” and similar terms used in this application specification and claims do not indicate a limitation of quantity, but rather indicate the presence of at least one. “A plurality” means two or more. The words “comprising” or “including” and similar terms mean that the element or object preceding “comprising” or “including” covers the element or object listed following “comprising” or “including” and its equivalents, and does not exclude other elements or objects. The words “connected” or “linked” and similar terms are not limited to physical or mechanical connections and can include electrical connections, whether direct or indirect. The words “above” and / or “below” and similar terms are for ease of description only and are not limited to a location or spatial orientation. The singular forms “a,” “the,” and “the” used in this application specification and appended claims are also intended to include the plural forms, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used herein refers to and includes any or all possible combinations of one or more associated listed items.
[0021] The technical concept of this invention includes: First, in response to the problems in the existing public space seating management, such as unfair resource allocation and low efficiency of users finding seats due to the inability to accurately distinguish between actual occupancy and seats occupied by objects, as well as the imbalance of peak and off-peak seat usage and insufficient basis for management decisions due to the lack of data-driven prediction and scheduling mechanisms, a systematic solution is proposed.
[0022] Secondly, this solution achieves accurate judgment of seat status by integrating a multimodal perception mechanism of pressure and infrared sensors, uses a collaborative filtering algorithm to build a personalized recommendation model to improve seat matching efficiency, and introduces an LSTM time series prediction model to achieve probabilistic prediction of future idle resources. Thus, a closed-loop optimization is formed at the three levels of perception, recommendation and scheduling, systematically addressing the above problems.
[0023] Therefore, this invention aims to construct a complete technical path from real-time perception and intelligent analysis to dynamic scheduling and continuous optimization, and to achieve intelligent, fair and efficient management of public seating resources through the deep integration of data and algorithms.
[0024] This application provides a smart seat management system, including: The sensing module is used to collect seat occupancy status through pressure sensors deployed on the seat and infrared array sensors, and to collect user identity information through NFC tags. The communication module is used to transmit the data collected by the sensing module; The data processing and algorithm module is used to receive and process the data, generate personalized seat recommendation results based on the collaborative filtering algorithm, and generate seat vacancy probability prediction results based on the LSTM time series prediction model. The application interaction module is used to display the recommendation results and prediction results to the user and receive the user's operation instructions.
[0025] This invention uses a dual verification mechanism of pressure sensors and infrared sensors to accurately distinguish between "real people sitting" and "objects occupying seats." Combined with a collaborative filtering recommendation algorithm based on user behavior, it effectively reduces invalid seat occupation, achieves fair allocation and rapid matching of seat resources, and improves users' seat-finding efficiency and user experience.
[0026] The intelligent seat management method described in this application includes the following steps: S1. Real-time acquisition of multimodal sensing data of the seat, the data including at least pressure data, infrared thermal data, and user authentication data; S2. Clean and merge the collected data to determine the actual occupancy status of the seats; S3. Based on the processed historical and real-time data, personalized seat recommendation information and future seat availability probability information are generated through machine learning models. S4. Push the recommendation information and idle probability information to the user terminal; S5. Update the seat status and record user behavior data based on the operation instructions fed back by the user terminal.
[0027] The following is in conjunction with the appendix Figures 1 to 3 This application provides a detailed description of a smart seat management system and method.
[0028] In a preferred embodiment of the present invention, a smart seating management system deployed in a university library is provided. For example... Figure 1 As shown, the system adopts a four-layer architecture, including a sensing module, a communication module, a data processing and algorithm module, and an application interaction module. The sensing module hardware mainly includes a thin-film pressure sensor, an infrared thermal imaging array sensor, and an NFC tag. A pressure sensor is embedded under each learning seat to detect whether an object is placed on it; an infrared sensor is installed on the back or side of the seat, with its detection window facing the seat area, to identify specific wavelengths of infrared radiation emitted by the human body; an NFC tag is affixed to the seat desktop, with an internal code uniquely bound to the seat ID. All sensors are connected to the coordinator node in their respective seat areas via wires. This node is responsible for collecting raw signals and performing preliminary analog-to-digital conversion and encapsulation.
[0029] The communication module utilizes the library's comprehensive Wi-Fi network. Each coordinator node uploads encapsulated data packets to the designated cloud server gateway in real time via the Wi-Fi module. To ensure transmission reliability, the network is configured with a Quality of Service (QoS) policy, giving higher priority to sensor data streams, ensuring that status change information reaches the cloud within 2 seconds, and optimizing the response time for user authentication requests to within 1 second.
[0030] The data processing and algorithm modules are deployed on elastic computing services. The data processing service cleans and formats the incoming real-time data. Cleaning rules include removing abnormal data with pressure values that are consistently 0 or exceed the range, filtering transient noise in infrared data caused by sudden changes in ambient temperature, and integrating pressure, infrared, NFC check-in events, and timestamps into a unified format "seat status record." The processed data is stored in a time-series database for real-time querying and simultaneously synchronized to a big data platform, where it is used by an AI algorithm engine for online inference.
[0031] The AI algorithm engine includes two core model services, such as Figure 3 As shown, the model provides recommendation services and prediction services. The recommendation model employs a user-based collaborative filtering algorithm, where the "rating" in the user-item rating matrix is implicitly calculated based on user behavior such as historical seat usage time, number of active selections, and whether a seat is marked as a "preferred seat." The prediction model uses a neural network structure consisting of three LSTM layers and one fully connected layer. The input is a time series of seat occupancy rates for each region, divided by hour over the past 24 hours, and the output is the probability of each seat being available in the next hour.
[0032] For the user side, the application interaction module developed a WeChat mini-program called "Smart Seating". The mini-program's homepage displays the real-time seat status of each floor and area of the library in the form of a visual map (using green, yellow, and red to represent vacant, occupied, and temporarily unoccupied, respectively), and provides two core function entry points: "Intelligent Recommendation" and "Vacancy Prediction". For the administrator side, a web-based data visualization backend was developed, which displays multi-dimensional data analysis reports in chart form.
[0033] In this embodiment, the system's workflow is as follows: Figure 2 As shown, the specific steps include: When a student enters the library and needs to find a seat, they open a WeChat mini-program. The system first obtains the student's WeChat-linked student ID information through the mini-program to complete identity pre-verification. The student clicks the "Smart Recommendation" button, requesting that the request, along with their anonymized user ID and current timestamp, be sent to the data processing and algorithm module. Upon receiving the request, the recommendation model service first queries the Redis cache for a list of all currently available seats, then retrieves the user's historical seat selection records and those of other users with similar behavior from the user behavior database, calculating the user's estimated preference score for each available seat. Calculation factors include the quietness of the seat area (derived from historical data), distance from a window, availability of power outlets, and other static attributes, as well as the frequency with which the user selected seats in the same time period and area over the past week. Finally, the model outputs a seat list sorted in descending order of recommendation score, along with the recommendation reasons (e.g., "Your preferred quiet area," "Power outlets available"), and returns them to the mini-program frontend. The entire process takes less than 3 seconds.
[0034] Students can also view the "Availability Prediction" function. The prediction model service starts from the current time, traces back the past 24 hours of seat occupancy data across the entire venue, and combines this with contextual features such as whether it's a weekday or weekend, or whether it's exam week, to run an LSTM model for inference. The model outputs the probability of each occupied seat becoming available within the next 60 minutes. The mini-program frontend maps this probability to three levels: "High (>70%)", "Medium (30%-70%)", and "Low (<30%)", and displays them as intuitive icons overlaid on the seat map. For example, a currently occupied seat displaying a "High" availability probability suggests to the user that the seat is likely to become available soon, helping them plan their wait or find another seat.
[0035] After selecting a recommended available seat, the student walks to it. Before sitting down, they use their mobile phone (with NFC enabled) to tap an NFC tag on the table. Once the tag is read, its unique ID and read timestamp are sent to the platform via a mini-program. The platform verifies that the seat's current status is "available" and matches the user's reservation or selection record, completing the check-in and updating the seat status to "occupied (verified)" in real time. When leaving the seat, if it's a short absence (e.g., to the restroom), the user can click the "Temporarily Leave" button in the mini-program, changing the seat status to "Temporarily Away" and starting a countdown (e.g., 30 minutes). If the user returns before the countdown ends and confirms again via NFC or the mini-program, the status returns to "Occupied." If the user does not return within the time limit, the system will automatically release the seat and may record a "Temporary Away Timeout" in the user's credit record.
[0036] Throughout the process, the pressure sensor and infrared sensor of the sensing module operate continuously. The data fusion service of the data processing and algorithm module periodically (e.g., every 10 seconds) compares the pressure data, infrared data, and system status data from the same seat. In this embodiment, a specific data fusion and status judgment rule is defined: when the pressure sensor reading exceeds 5 kg for three consecutive cycles (i.e., the first threshold), and the infrared sensor simultaneously detects a valid human heat source signal (surface temperature within the range of 30-42 degrees Celsius), the system determines that the seat is occupied. If the system's logical status is "idle" or "temporarily vacant," but the sensor determines that the seat is occupied, the system will automatically update the status to "occupied (unauthenticated)" and send a reminder to users who may have reserved the seat via a mini-program. Conversely, if the pressure data persists but there is no valid infrared signal, it is determined that the seat is occupied by an item. The system will first send a reminder to the user marked as occupying the seat via a mini-program, "Please do not use items to occupy the seat"; if the situation does not change after a period of time (e.g., 15 minutes), the system will generate an "abnormal seat occupation" event in the management backend for the administrator to handle on-site. This dual verification mechanism is one of the core technical means to solve the seat-hogging problem in this embodiment.
[0037] In this embodiment, the model training process is conducted offline and periodically. Every Sunday morning, the system starts a new round of model training. The training task extracts all historical data from the past 12 weeks, including over 500,000 seat status records and 100,000 user check-in and seat selection behavior records. First, feature engineering is performed to construct a "user-seat feature" interaction matrix for the recommendation model and hourly "region-time-occupancy" sequence samples for the prediction model. The collaborative filtering model is trained using alternating least squares, aiming to minimize the root mean square error between the predicted score and the implicit feedback. Training continues until the validation set no longer significantly decreases or reaches the preset iteration limit of 100 rounds.
[0038] In this embodiment, the system also implements a complete credit and scheduling management mechanism. Each user initially has 100 credit points. When behaviors such as "failure to check in after reservation," "failure to release after temporary departure time," or "being identified by sensors as having an item occupying a seat and failing to correct the situation after being reminded" occur, the system will automatically deduct the corresponding credit points. When the credit score is below 80 points, the user will be restricted from using the "reservation" function and can only use "on-site seat selection"; when it is below 60 points, the priority of using "intelligent recommendation" will be reduced. This set of credit data, together with seat usage data (such as the occupancy rate of different areas at different times, average occupancy time, seat turnover rate, etc.), constitutes the foundation of the management backend data analysis module. Administrators can customize and generate daily, weekly, or monthly reports. The reports, in the form of heat maps, line graphs, bar charts, etc., intuitively reveal the spatiotemporal patterns of resource usage. For example, the report may show that the occupancy rate of the quiet area on the east side of the third floor consistently reaches 98% during the 14:00-16:00 time period during exam week, while the utilization rate of the discussion area on the fourth floor is less than 40% on weekday mornings. These data-driven insights can support administrators in making informed decisions, such as adding temporary seating in quiet areas during peak hours, adjusting seminar area opening policies, or optimizing library lighting and air conditioning zoning strategies, thereby fundamentally improving the management efficiency of space resources. Other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the disclosure herein. The invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of the invention are indicated by the following claims.
[0039] It should be understood that the present invention is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.
Claims
1. A smart seat management system, characterized in that, include: The sensing module is used to collect seat occupancy status through pressure sensors deployed on the seat and infrared array sensors, and to collect user identity information through NFC tags. The communication module is used to transmit the data collected by the sensing module; The data processing and algorithm module is used to receive and process the data, generate personalized seat recommendation results based on the collaborative filtering algorithm, and generate seat vacancy probability prediction results based on the LSTM time series prediction model. The application interaction module is used to display the recommendation results and prediction results to the user and receive the user's operation instructions.
2. The system according to claim 1, characterized in that, The system uses a dual verification mechanism of pressure sensors and infrared sensors to distinguish between real people sitting in the seats and objects occupying the seats.
3. The system according to claim 1, characterized in that, The data processing and algorithm module is also used to construct a user-item rating matrix based on the user's historical behavior data, which serves as the input to the collaborative filtering algorithm.
4. The system according to claim 1, characterized in that, The data processing and algorithm module is also used to train the LSTM model with continuous historical seat usage data of no less than 3 months, wherein the historical data includes at least the occupancy period, duration and regional information.
5. The system according to claim 1, characterized in that, The application interaction module includes a user-side WeChat mini-program and an administrator-side data dashboard. The data dashboard is used to display regional utilization rate, turnover rate, and peak-hour analysis reports generated based on seat usage data.
6. A smart seat management method, characterized in that, Includes the following steps: S1. Real-time acquisition of multimodal sensing data of the seat, the data including at least pressure data, infrared thermal data and user authentication data; S2. Clean and merge the collected data to determine the actual occupancy status of the seats; S3. Based on the processed historical and real-time data, personalized seat recommendation information and future seat availability probability information are generated through machine learning models. S4. Push the recommendation information and idle probability information to the user terminal; S5. Update the seat status and record user behavior data based on the operation instructions fed back by the user terminal.
7. The method according to claim 6, characterized in that, The method for determining the actual occupancy status of a seat includes: when the pressure data exceeds a first threshold and the infrared thermal sensing data is valid, it is determined that a person is seated; when the pressure data exceeds the first threshold but the infrared thermal sensing data is invalid, it is determined that an object is occupying the seat.
8. The method according to claim 6, characterized in that, The generation of personalized seat recommendation information specifically includes: calculating and ranking the recommendation scores of each seat based on the user's historical preference data and the behavioral data of similar users through a collaborative filtering algorithm.
9. The method according to claim 6, characterized in that, The process of generating seat vacancy probability information for future time periods specifically includes: organizing historical occupancy status data in a time series manner, inputting it into an LSTM prediction model, and outputting the vacancy probability level of each seat within the next 30-60 minutes.
10. The method according to claim 6, characterized in that, The user's credit score is updated based on their historical performance record, and usage restrictions are imposed on users whose credit scores are below the threshold.