An intelligent home voice interaction and remote monitoring system suitable for the elderly

Abstract

The application discloses an old-age adaptation intelligent home voice interaction and remote monitoring system, belongs to the field of intelligent home and old-age care monitoring, and comprises a perception interaction module, an intelligent judgment module and a remote monitoring module. The perception interaction module is provided with an old-age adaptation voice interaction terminal and distributed multi-modal perception equipment, and completes contactless multi-signal acquisition and old-age adaptation voice interaction; the intelligent judgment module quickly judges an emergency state through an edge computing terminal, a cloud service terminal performs deep fusion analysis and self-learning optimization, a three-level grading early warning mechanism is arranged, corresponding early warning is triggered after voice verification; and the remote monitoring module realizes early warning receiving, state checking and remote linkage. The system only transmits collected signal characteristic values, data transmission is encrypted, and storage is provided with authority management, contactless all-scene old-age adaptation monitoring is realized, judgment accuracy and data security are taken into account, life service and health monitoring integration is achieved, and the system is adapted to home-based old-age care needs.

Description

Technical Field

[0001] This invention relates to the field of smart home and elderly care monitoring technology, specifically to an age-friendly smart home voice interaction and remote monitoring system. Background Technology

[0002] Existing remote monitoring solutions for the elderly mostly adopt single visual monitoring or portable wearable device monitoring modes, which have problems such as privacy leaks, incomplete scene coverage, and high operation thresholds. Some monitoring solutions can only realize post-fall alarms and cannot provide early warnings of potential health risks. Moreover, the interaction methods are not optimized for the physiological characteristics of the elderly, resulting in low acceptance among the elderly. At the same time, the judgment mechanism of traditional monitoring systems is simple, with a high error rate, and the security of data transmission and storage is also difficult to guarantee, failing to meet the needs of full-scene, contactless, and intelligent monitoring in home-based elderly care scenarios. Summary of the Invention

[0003] The purpose of this invention is to provide an age-friendly smart home voice interaction and remote monitoring system to solve the technical problems of existing monitoring solutions, such as limited scenario coverage, unsuitability for the elderly, only post-event alarm capability, and low data security.

[0004] The technical solution of the present invention to solve the above-mentioned technical problems is as follows: An age-friendly smart home voice interaction and remote monitoring system includes a perception and interaction module, an intelligent judgment module, and a remote monitoring module. The perception and interaction module completes contactless collection of various types of signals in the home environment, and simultaneously realizes age-friendly local voice interaction and command transmission and reception. The intelligent judgment module performs hierarchical intelligent analysis and status judgment on the collected signals and implements system self-learning optimization. The remote monitoring module receives the judgment results from the intelligent judgment module and completes early warning reception, two-way linkage, and monitoring and handling. The three modules work together to realize the entire process of age-friendly voice interaction and remote monitoring.

[0005] In a further embodiment, the perception and interaction module includes an age-friendly voice interaction terminal and a distributed multimodal perception device. The age-friendly voice interaction terminal is the core interaction entry point of the system, supporting far-field voice wake-up, dialect semantic recognition, high-volume slow-speed voice broadcasting, and two-way voice intercom. It receives voice commands from the elderly and triggers corresponding operations. The distributed multimodal perception device includes millimeter-wave radar, low-power visual cameras, and environmental sensors, which are distributed and deployed in various activity areas of the home to complete contactless acquisition of human motion and static, physiological signs, spatial positioning signals, human behavior contour signals, and home environment parameter signals, thereby achieving full-scene signal acquisition.

[0006] In a further embodiment, the intelligent judgment module is divided into an edge computing terminal and a cloud server terminal. The edge computing terminal receives real-time acquisition signals from the perception and interaction module to quickly identify and judge emergency states such as falls. The cloud server terminal receives the feature values ​​of the data collected by the perception and interaction module, completes deep data fusion analysis, establishes a health behavior model specifically for the elderly, and realizes trend analysis and potential health risk judgment in non-emergency states.

[0007] In a further embodiment, the edge computing terminal uses dual verification logic to determine the fall status. It combines human posture signals collected by millimeter-wave radar and human behavior contour signals collected by low-power vision cameras to comprehensively analyze the characteristics of human posture changes and contour morphology to obtain the judgment result of the fall status.

[0008] In a further embodiment, the cloud server performs deep fusion analysis through the correlation analysis of physiological signs and behavioral states. It quantitatively correlates the collected physiological sign parameters and behavioral state parameters, compares the deviation of the parameters from the normal benchmark values, and obtains the health risk assessment results for the elderly. The system self-learning optimization is implemented by the cloud server, which continuously collects the elderly's physiological signs, behavioral states, and various assessment results data. Based on the data changes, it dynamically adjusts the threshold parameters for various state assessments and the analysis parameters of the health behavior model to optimize the system's assessment accuracy.

[0009] In a further embodiment, the intelligent judgment module is equipped with a three-level early warning mechanism, with the early warning level being associated with the health risk judgment result and the emergency state judgment result. The first-level early warning only triggers the information push operation of the remote monitoring module, the second-level early warning triggers the information push operation of the remote monitoring module and the local voice broadcast and inquiry operation of the perception and interaction module, and the third-level early warning triggers the multi-terminal information push operation of the remote monitoring module, the local sound and light alarm of the perception and interaction module, and the linkage operation of home smart home devices.

[0010] In a further embodiment, after the intelligent judgment module obtains the emergency state judgment result, it first initiates a local voice verification operation through the perception and interaction module, and executes subsequent warning triggering operations based on the elderly person's voice response; if there is no voice response from the elderly person, the three-level warning mechanism is directly triggered.

[0011] In a further embodiment, the remote monitoring module includes a family member terminal, a community elderly care platform terminal, and a medical institution terminal. Each terminal can receive early warning information, view the elderly person's monitoring status information, and have a two-way voice intercom function with the sensing and interaction module. When a level 3 early warning is triggered, each terminal simultaneously receives the early warning information, and each terminal can send a rescue linkage command to the sensing and interaction module to realize remote control of smart home devices.

[0012] In a further embodiment, when the perception and interaction module transmits data to the intelligent judgment module, it only uploads the feature values ​​of the collected signals and does not transmit the original collected data; all data transmission processes in the system are implemented through encryption protocols, and the cloud server sets up an access control mechanism for all stored monitoring data, allowing only authorized users to access the monitoring data of the corresponding elderly person.

[0013] In a further embodiment, the sensing and interaction module receives voice commands initiated by the elderly, which can trigger operations such as controlling smart home devices, monitoring physiological signs, and remotely calling family members, thereby integrating the functions of daily life services and health monitoring. The present invention has the following beneficial effects: This invention achieves a deep integration of home-based elderly care monitoring and age-friendly voice interaction through a three-layer collaborative architecture consisting of a perception and interaction module, an intelligent judgment module, and a remote monitoring module. The contactless data collection method of the distributed multimodal perception device not only covers the monitoring needs of all home scenarios but also avoids privacy leaks from the source. The dedicated optimization of the age-friendly voice interaction terminal lowers the operating threshold for the elderly and realizes the integration of life services and health monitoring functions. The intelligent judgment module's hierarchical analysis mechanism balances rapid response in emergency situations with early warning of potential health risks. Combined with dual verification of fall detection using mathematical algorithms, quantitative analysis of health risks, and a self-learning optimization mechanism, it effectively improves the accuracy and adaptability of status judgments. The tiered early warning mechanism and the multi-terminal linked remote monitoring module enable differentiated handling of different risk levels, significantly improving the efficiency of monitoring and rescue. Simultaneously, the data transmission feature value uploading method, encryption protocol, and access control mechanism comprehensively ensure the security of monitoring data transmission and storage. The overall solution aligns with the actual needs of home-based elderly care, possessing high adaptability and practicality. Attached Figure Description

[0014] Figure 1 This is a block diagram of the overall system architecture of the present invention; Figure 2 This is a block diagram of the perception and interaction layer structure of the present invention; Figure 3 This is a block diagram of the intelligent judgment layer structure of the present invention; Figure 4 This is a block diagram of the remote monitoring layer structure of the present invention.

[0015] In the diagram: 1-Perception and interaction module, 11-Aging-friendly voice interaction terminal, 12-Distributed multimodal sensing device, 121-Millimeter-wave radar, 122-Low-power vision camera, 123-Environmental sensor; 2-Intelligent judgment module, 21-Edge computing terminal, 22-Cloud server; 3-Remote monitoring module, 31-Family terminal, 32-Community elderly care platform terminal, 33-Medical institution terminal. Detailed Implementation

[0016] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.

[0017] refer to Figures 1-4 As shown, an age-friendly smart home voice interaction and remote monitoring system consists of a perception interaction module 1, an intelligent judgment module 2, and a remote monitoring module 3. These three modules communicate via a TCP / IP encrypted communication protocol to achieve data interaction and command transmission, forming a complete monitoring loop encompassing signal acquisition, intelligent analysis, and early warning response. This collaboratively completes the entire process of age-friendly voice interaction and remote monitoring in a home setting. The perception interaction module 1, serving as the core entry point for data acquisition and human-computer interaction, is distributed across key activity areas in the elderly person's home environment, enabling contactless acquisition of various physical signals and the transmission and reception of local voice interaction commands. The intelligent judgment module 2 adopts a layered architecture combining local and cloud components, consisting of a local edge computing terminal 21 and a remote cloud server 22. These components respectively enable millisecond-level rapid judgment of emergency situations such as falls and in-depth analysis and trend warnings of the elderly person's long-term health status. The remote monitoring module 3 is deployed on the terminals of family members, community elderly care platforms, and medical institutions, enabling the reception of multi-level early warning information, viewing of the elderly person's monitoring status, and remote emergency response. This three-layer architecture, with its distinct functions and deep collaboration, ensures the integrity and efficiency of the monitoring process.

[0018] In one embodiment, the sensing and interaction module 1 includes an age-friendly voice interaction terminal 11 and a distributed multimodal sensing device 12. The overall system is deployed in a distributed manner in the main activity areas of the home, such as the living room, bedroom, bathroom, kitchen, and balcony, in accordance with the principle of "full coverage, no dead angles, and low interference". This enables full-dimensional signal collection and convenient voice interaction in the home environment. All signal collection processes are implemented in a contactless manner, without requiring the elderly to wear any devices.

[0019] The age-friendly voice interaction terminal 11 serves as the core human-computer interaction interface of the system. At least one unit is deployed in each of the elderly's frequently used areas, such as the living room and bedroom. The terminal features a built-in 6-microphone ring array, an offline voice recognition module, a high-power voice broadcast module, and a two-way voice intercom module. It supports far-field voice wake-up from 5-8 meters away and is resistant to noise interference in the home environment. The terminal integrates semantic recognition models for Mandarin and major northern and southern dialects, accurately recognizing the elderly's everyday spoken commands. The volume and speed of the voice broadcast module can be preset, defaulting to a high-volume, slow-speed broadcast mode to accommodate the physiological characteristics of hearing loss in the elderly. This terminal can receive various proactive voice commands initiated by the elderly, triggering corresponding operations based on the command type. These include life service operations such as turning smart home devices on and off and adjusting modes, activating physiological sign monitoring such as heart rate and respiratory rate, and monitoring operations such as calling family members or requesting help from the remote monitoring module 3. This integrates life service and health monitoring functions, allowing the elderly to complete various needs simply through voice commands without needing to learn complex device operations.

[0020] Distributed multimodal sensing device 12: Composed of millimeter-wave radar 121, low-power vision camera 122, and environmental sensor 123, these devices are deployed as needed according to the functional characteristics of the home area, working together to accurately collect multiple types of signals. Millimeter-wave radar 121: As the core sensing unit, one unit is deployed in each room. Utilizing a 77GHz frequency-modulated continuous wave radar, it can penetrate clothing, bedding, thin furniture, and other obstructions without collecting any visual information, thus avoiding privacy leaks at the source. This radar collects human motion and static signals, physiological signs, and spatial positioning signals by transmitting and receiving radar echoes. The collection of physiological signs is achieved through calculation of the phase changes of the radar echoes; the core calculation formula is as follows: In the formula, Heart rate / respiratory rate, in Hz; A complete cycle of heartbeat / respiration is measured in seconds (s). The phase change of the radar echo within one period is expressed in rad; t represents the phase change. The corresponding time is expressed in seconds (s). Spatial positioning signals are acquired using the time difference of arrival (TDOA) method with multiple radars. The core calculation formula is: In the formula, (x, y, z) represent the three-dimensional coordinates of the elderly person in their home space; , , ), ( , , ), ( , , () represents the three-dimensional coordinates of three distributed millimeter-wave radars; The speed of electromagnetic wave propagation is 3×10⁸ m / s; The time difference between the arrival of the sensing signal at the first and second radars is measured in seconds. The time difference between the arrival of the sensing signal at the second and third radars is expressed in seconds.

[0021] The low-power visual camera 122 is deployed only in open areas without privacy, such as living rooms and bedrooms. It employs a human contour recognition algorithm, collecting only the behavioral contour signals of the human body, without capturing any detailed information such as facial features, to assist millimeter-wave radar in accurately judging behavioral states. The camera operates in a low-power mode, only activating high-definition acquisition when the millimeter-wave radar detects changes in human movement or stillness; otherwise, it remains in sleep mode, reducing device power consumption and data acquisition volume.

[0022] Environmental Sensor 123: Deployed as needed in areas such as the kitchen, bathroom, and living room, it collects signals of home environmental parameters such as temperature and humidity, gas concentration, smoke concentration, door and window opening status, and light intensity, enabling dual monitoring of the elderly's physical condition and the home environment. When environmental parameters are abnormal, they can be linked with the elderly's physical condition signals for analysis, improving the comprehensiveness of monitoring.

[0023] In one embodiment, the intelligent judgment module 2 is divided into an edge computing terminal 21 and a cloud server terminal 22. The edge computing terminal 21 is a local smart home gateway equipped with a lightweight neural network inference model. The cloud server terminal 22 is a remote cloud platform deployed on a public cloud, equipped with a deep fusion analysis model and a self-learning optimization model. The two interact with each other through an encrypted network to collaboratively complete the hierarchical intelligent analysis, status judgment and system self-learning optimization of the collected signals, taking into account both the response speed in emergency situations and the foresight of long-term health monitoring.

[0024] Edge computing terminal 21: Interacts with the perception and interaction module 1 in real-time via the home local area network, receiving real-time acquisition signals from the perception and interaction module 1. It prioritizes rapid identification and judgment of emergency states such as falls, with a response latency controlled within 1 second. During the fall judgment process, a dual verification logic of millimeter-wave posture and visual contour is employed. It simultaneously receives human posture signals collected by millimeter-wave radar 121 and human behavioral contour signals collected by low-power visual camera 122. The fall judgment result is obtained through feature extraction and comprehensive matching. The core judgment formula is: In the formula, This represents the change in vertical height of the human body, in meters (m). This is the baseline value of the elderly person's vertical height in a normal upright position, in meters; The actual vertical height of the human body at a certain moment, in meters; The height threshold for determining a human fall is pre-determined based on the elderly person's height and limb characteristics, and the unit is meters (m). The contact area between the human visual outline and the ground at a certain moment is measured in m2. The threshold for the outline contact area used to determine a fall is pre-calibrated based on the elderly person's body shape characteristics, and the unit is m2. When the conditions of both formulas are met simultaneously, the system determines that the person has fallen. This dual verification logic effectively avoids the misjudgment problem caused by a single device. For example, when the elderly person sits or lies down normally, although the height change condition is met, the outline contact area condition is not met, and the system will not determine that the person has fallen.

[0025] Cloud server 22: Transmits the feature values ​​of the collected data to the sensing interaction module 1 via an encrypted public network, without receiving any raw collected data. This reduces network transmission pressure and further ensures data security. The cloud server 22 performs deep data fusion analysis on the received feature values. The specific implementation process is as follows: First, based on the elderly person's basic health data (such as age, height, weight, and basic medical history) and the initially collected daily behavior data, a health behavior model specific to the elderly person is established. Then, through correlation analysis between physiological signs and behavioral states, physiological sign parameters such as heart rate and respiratory rate collected by millimeter-wave radar are quantitatively correlated with behavioral state parameters such as spatial positioning, activity level, daily routine, and range of motion. The core quantification formula is: In the formula, This is a quantified value for health risk, without units. The weighting coefficient for physiological signs ranges from (0,1) and is dynamically adjusted based on the elderly person's underlying medical history. The coefficient increases for elderly people with underlying diseases. Let be the weight coefficient of the behavioral state, with a value range of (0,1), and satisfy . + =1; The actual values ​​of physiological parameters collected at a certain moment; The normal baseline values ​​for the elderly person's physiological parameters were determined based on basic health data and long-term collected data. The actual value of the behavioral state parameters collected at a certain moment; The parameters representing the elderly person's behavioral status are set as normal baseline values ​​based on their daily habits. The cloud server 22 calculates a quantitative value for health risk. By comparing the results with preset risk thresholds, the health risk assessment results of the elderly can be obtained, realizing trend analysis and potential health risk assessment in non-emergency situations.

[0026] System self-learning optimization: Implemented entirely by cloud server 22, which continuously collects the elderly's physiological signs, behavioral status data, judgment results for various states, and early warning and response results to establish a long-term monitoring database for the elderly. Based on the long-term trends of the data in the database, the threshold parameters for various state judgments, such as fall detection and health risk assessment, are dynamically adjusted using the gradient descent method. Simultaneously, the analysis parameters of the health behavior model are optimized. The core update formula is: In the formula, This is the health risk warning threshold after the (n+1)th update; This is the threshold for the nth health risk warning. The learning rate is defined as (0,1), and its value is adjusted based on the frequency and stability of data updates; L is the loss function used for threshold determination. ,in The number of samples in a single update. For the first The system quantifies the health risk of each sample. The self-learning optimization mechanism enables the system's judgment model to adapt to the elderly person's physical condition and behavioral habits in real time. For example, if the elderly person's activity level decreases after surgery or their limb movements become smaller, the system will automatically adjust the corresponding judgment threshold, continuously improving the accuracy of the system's status judgment.

[0027] In one embodiment: the intelligent judgment module 2 triggers a corresponding three-level early warning mechanism based on the health risk judgment result obtained by the cloud server 22 and the emergency state judgment result obtained by the edge computing terminal 21. Different early warning levels are strongly correlated with the degree of risk, and differentiated early warning triggering measures are matched to achieve accurate and efficient handling of risks, avoid excessive early warnings from interfering with the elderly, and ensure timely rescue in high-risk situations.

[0028] Level 1 Warning: When the health risk quantification value calculated by the cloud server reaches level 22. satisfy When the risk level is determined to be low, a Level 1 warning is triggered. At this time, only the corresponding risk warning information is pushed to the family member's terminal of the remote monitoring module 3, without any local warning operation. The warning content includes the elderly person's current status and risk type, so that family members can keep abreast of the elderly person's health trend and avoid interfering with the elderly person's normal life.

[0029] Level 2 warning: When the health risk quantification value F calculated by the cloud server 22 meets the requirements... When the situation is assessed as a medium health risk, a level-two warning is triggered. At this time, two operations are performed simultaneously: first, a risk warning message is pushed to the family terminal of the remote monitoring module 3; second, a local voice broadcast and inquiry are initiated through the age-friendly voice interaction terminal 11 of the perception and interaction module 1 to confirm the elderly person's physical condition and whether they need help. The voice inquiry content uses conversational and gentle expressions to suit the elderly person's comprehension ability.

[0030] Level 3 warning: When the edge computing terminal 21 determines that an emergency situation such as a fall has occurred, or when the health risk quantification value F calculated by the cloud server terminal 22 meets the requirements... ≥ When the situation is assessed as high health risk, a Level 3 warning is triggered. Simultaneously, three operations are executed: first, the warning information is pushed to the family terminal, community elderly care platform terminal, and medical institution terminal of the remote monitoring module 3, including the elderly person's real-time location, current status, and risk level; second, a local sound and light alarm is initiated through the age-friendly voice interaction terminal 11, providing dual sound and light alerts to facilitate timely discovery and assistance by those nearby; and third, smart home devices are automatically activated, such as turning on lights in the elderly person's area, unlocking smart door locks, and shutting off kitchen gas valves, facilitating both on-site and remote rescue efforts.

[0031] in, As a basic threshold for health risk early warning, This is the threshold for escalating from a Level 1 warning to a Level 2 warning. The threshold for upgrading from a Level II warning to a Level III warning, and meeting the following conditions. < < Each threshold can be dynamically adjusted through the self-learning optimization mechanism of the cloud server 22.

[0032] After the edge computing terminal 21 obtains the emergency status judgment result such as a fall, the system does not directly trigger the level three warning. Instead, it first initiates a local voice verification operation through the age-friendly voice interaction terminal 11 of the perception and interaction module 1 to confirm with the elderly whether they need rescue. If the elderly's voice response is received within the preset time range, the corresponding warning trigger operation is executed according to the response content. If there is no voice response from the elderly within the preset time range, it is determined that the elderly have lost the ability to move or express themselves, and the level three warning mechanism is directly triggered to ensure the timeliness of rescue in emergency situations to the greatest extent.

[0033] In one embodiment: the remote monitoring module 3 includes a family member terminal 31, a community elderly care platform terminal 32, and a medical institution terminal 33. Each terminal is a mobile terminal (phone, tablet, or computer terminal) with a dedicated monitoring program installed in this system. It interacts with the cloud server 22 of the intelligent judgment module 2 through an encrypted public network. Each terminal is configured with different functional permissions according to the needs of the user, so as to realize remote monitoring with multiple subjects working together.

[0034] Basic Function Implementation: Each terminal has three core basic functions, and the implementation of these functions is deeply adapted to the needs of the elderly: First, the early warning information receiving function can receive early warning information of various levels pushed by the intelligent judgment module 2 in real time and accurately. The early warning information is presented in the form of pop-up window + sound prompts to ensure that the recipient can discover it in time. Second, the elderly monitoring status information viewing function can view the elderly's real-time physiological parameters, behavioral status, home space location, environmental parameters and other monitoring information. All information is displayed in the form of visual charts and text, which is simple and easy to understand. Third, the two-way voice intercom function can realize real-time, zero-delay two-way voice intercom with the elderly-friendly voice interaction terminal 11 of the perception interaction module 1, directly communicate with the elderly, understand the elderly's actual status and needs, and the voice can automatically amplify the volume to adapt to the hearing characteristics of the elderly.

[0035] The three-level early warning linkage mechanism is implemented as follows: When a level three early warning is triggered, the family member terminal 31, the community elderly care platform terminal 32, and the medical institution terminal 33 will simultaneously receive the early warning information. The operators of each terminal can take corresponding actions according to their responsibilities: family members can communicate with the elderly through two-way voice intercom or remotely trigger smart home devices for rescue assistance; the community elderly care platform can dispatch nearby elderly care service personnel to the scene for inspection; and the medical institution can provide remote rescue guidance or arrange emergency personnel to go to the scene based on the elderly's basic health data and current status.

[0036] Remote linkage command implementation: Each terminal has the function of sending remote rescue linkage commands. Operators can send rescue linkage commands to the sensing and interaction module 1 through the terminal. The command is transmitted to the edge computing terminal 21 at home through an encrypted network. The edge computing terminal 21 triggers the corresponding smart home device to perform actions, realizing remote control of smart home devices, such as remotely opening smart door locks, remotely closing gas valves, and remotely adjusting indoor lights, which can cooperate with the on-site rescue work and improve rescue efficiency.

[0037] In one embodiment, this system establishes a comprehensive, end-to-end data security mechanism from two core dimensions: data transmission and data storage. This mechanism maximizes the protection of the elderly's monitoring data and privacy information, meeting the privacy protection needs of home-based elderly care.

[0038] Data transmission security: When the perception and interaction module 1 transmits data to the intelligent judgment module 2, it only extracts features from the raw collected signals and uploads the signal feature values, without transmitting any raw collected data, which greatly reduces the risk of data leakage during transmission. At the same time, all data transmission processes between all modules of the system are implemented through national cryptographic-level encrypted communication protocols. Data is encrypted before transmission and decrypted and verified after transmission to ensure the security and integrity of data transmission and prevent data from being stolen or tampered with.

[0039] Data storage security: The cloud server 22 of the intelligent judgment module 2 sets up a strict access control mechanism for all elderly monitoring data, behavior data, and health data stored. Different access permissions are assigned according to user type (family members, community elderly care workers, medical staff), and multi-level account password verification is set up. Only authorized users can access the monitoring data of the corresponding elderly person, and unauthorized users cannot view or obtain any data. At the same time, the cloud data adopts a distributed storage method and performs multi-node backup to ensure the storage security and reliability of the data and prevent data loss.

[0040] In one embodiment, the overall workflow of the system is as follows: The distributed multimodal sensing device 12 of the sensing and interaction module 1 continuously and non-contactly collects the elderly person's dynamic and static body movements, physiological signs, spatial positioning signals, and home environment parameter signals; the age-friendly voice interaction terminal 11 receives the elderly person's active voice commands and triggers corresponding operations; after feature extraction, the collected signals are transmitted in real time to the intelligent judgment module 2; the edge computing terminal 21 completes the rapid judgment of emergency states such as falls; the cloud server terminal 22 completes deep fusion analysis and health risk judgment, and implements self-learning optimization; the intelligent judgment module 2 triggers the corresponding three-level graded early warning mechanism based on the judgment result and pushes early warning information to the remote monitoring module 3; after receiving the early warning information, the remote monitoring module 3 completes status viewing, two-way voice intercom, and remote rescue linkage; at the same time, the system implements a secure data transmission and storage mechanism throughout the entire process to ensure the data and privacy security of the elderly person.

[0041] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. 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 age-friendly smart home voice interaction and remote monitoring system, characterized in that, It includes a perception and interaction module, an intelligent judgment module, and a remote monitoring module; The perception and interaction module completes contactless collection of multiple types of signals in the home environment, and at the same time realizes age-friendly local voice interaction and command transmission and reception. The intelligent judgment module performs hierarchical intelligent analysis and status judgment on the collected signals and implements system self-learning optimization. The remote monitoring module receives the judgment results from the intelligent judgment module, completes early warning reception, two-way linkage and monitoring and handling, and the three-layer module works together to realize the entire process of age-friendly voice interaction and remote monitoring.

2. The age-friendly smart home voice interaction and remote monitoring system according to claim 1, characterized in that, The perception and interaction module includes an age-friendly voice interaction terminal and distributed multimodal perception devices. The age-friendly voice interaction terminal is the core interaction entry point of the system, supporting far-field voice wake-up, dialect semantic recognition, loud and slow speech broadcasting, and two-way voice intercom. It receives voice commands from the elderly and triggers corresponding operations. The distributed multimodal perception devices include millimeter-wave radar, low-power visual cameras, and environmental sensors, which are distributed in various activity areas of the home to complete contactless acquisition of human motion and static, physiological signs, spatial positioning signals, human behavior contour signals, and home environment parameter signals, realizing full-scene signal acquisition.

3. The age-friendly smart home voice interaction and remote monitoring system according to claim 1, characterized in that, The intelligent judgment module is divided into an edge computing terminal and a cloud server terminal. The edge computing terminal receives real-time acquisition signals from the perception and interaction module to quickly identify and judge emergency states such as falls. The cloud server terminal receives the feature values ​​of the data collected by the perception and interaction module, completes in-depth data fusion analysis, establishes a health behavior model exclusive to the elderly, and realizes trend analysis and potential health risk judgment in non-emergency states.

4. The age-friendly smart home voice interaction and remote monitoring system according to claim 3, characterized in that, The edge computing terminal uses dual verification logic to determine the fall status. It combines human posture signals collected by millimeter-wave radar and human behavior contour signals collected by low-power vision cameras to comprehensively analyze the characteristics of human posture changes and contour morphology to obtain the fall status judgment result.

5. The age-friendly smart home voice interaction and remote monitoring system according to claim 3, characterized in that, The cloud server performs deep fusion analysis through the correlation analysis of physiological signs and behavioral status, quantitatively correlates the collected physiological sign parameters and behavioral status parameters, compares the degree of deviation of the parameters with the normal benchmark value, and obtains the health risk assessment result of the elderly. The system's self-learning optimization is implemented by the cloud server, which continuously collects data on the elderly's physiological signs, behavioral status, and various judgment results. Based on data changes, it dynamically adjusts the threshold parameters for various status judgments and the analysis parameters of the health behavior model to optimize the system's judgment accuracy.

6. The age-friendly smart home voice interaction and remote monitoring system according to claim 1, characterized in that, The intelligent judgment module is equipped with a three-level early warning mechanism. The early warning level is related to the health risk judgment result and the emergency state judgment result. The first-level early warning only triggers the information push operation of the remote monitoring module. The second-level early warning triggers the information push operation of the remote monitoring module and the local voice broadcast and inquiry operation of the perception and interaction module. The third-level early warning triggers the multi-terminal information push of the remote monitoring module, the local sound and light alarm of the perception and interaction module, and the linkage operation of home smart home devices.

7. The age-friendly smart home voice interaction and remote monitoring system according to claim 6, characterized in that, After the intelligent judgment module obtains the emergency state judgment result, it first initiates a local voice verification operation through the perception and interaction module, and then executes the subsequent warning trigger operation based on the elderly's voice response result. If there is no response from the elderly person via voice, the level three warning mechanism will be triggered directly.

8. The age-friendly smart home voice interaction and remote monitoring system according to claim 7, characterized in that, The remote monitoring module includes a family member terminal, a community elderly care platform terminal, and a medical institution terminal. Each terminal can receive early warning information, view the elderly's monitoring status information, and have a two-way voice intercom function with the sensing and interaction module. When a level 3 early warning is triggered, each terminal receives the early warning information simultaneously, and each terminal can send a rescue linkage command to the sensing and interaction module to realize remote control of smart home devices.

9. The age-friendly smart home voice interaction and remote monitoring system according to claim 1, characterized in that, When the perception and interaction module transmits data to the intelligent judgment module, it only uploads the feature values ​​of the collected signals and does not transmit the original collected data. All data transmission processes in the system are implemented through encryption protocols. The cloud server sets up an access control mechanism for all stored monitoring data, and only authorized users can access the monitoring data of the corresponding elderly person.

10. The age-friendly smart home voice interaction and remote monitoring system according to any one of claims 1-9, characterized in that, The sensory interaction module receives active voice commands from the elderly, which can trigger operations such as controlling smart home devices, monitoring physiological signs, and remotely calling family members, thus integrating life services and health monitoring functions.

Images