Aging-appropriate furniture safety risk ai assessment method and system

Abstract

This invention discloses an AI-based method and system for assessing the safety risks of age-friendly furniture, relating to the field of artificial intelligence assessment technology. The method includes: placing a virtual body into a virtual three-dimensional environment; locating the mismatch risk points between the user's body and the furniture to be assessed by applying simulated perturbations; extracting biomechanical parameters at these risk points and determining physical risk values; determining psychological stress values ​​based on the biomechanical parameters through dynamic extrapolation via a physiological-psychological state mapping relationship; fusing physical risk values ​​and psychological stress values ​​to generate current safety risk assessment results; and predicting long-term safety risks of furniture in the same virtual three-dimensional environment based on the simulated physiological trajectory of aging and frailty using the virtual body. This invention improves the safety assurance level and forward-looking assessment capability of age-friendly furniture under real-world usage conditions by introducing mechanical perturbations and simulating typical instability conditions in a virtual three-dimensional environment.

Description

Technical Field

[0001] This invention relates to the field of artificial intelligence assessment technology, and in particular to an AI assessment method and system for the safety risks of age-friendly furniture. Background Technology

[0002] With the rapid development of intelligent and age-friendly product design for the elderly, important research directions in the field of home safety are gradually deepening. In existing technologies, the safety assessment of age-friendly furniture mainly relies on sensor monitoring and empirical models. This involves collecting physiological data such as the elderly user's posture, contact pressure, and movement trajectory during furniture use to conduct static or semi-dynamic analysis of the furniture's stability, support, and fall prevention performance. Some studies have introduced computer vision and human posture recognition technologies to achieve risk identification based on motion capture, such as detecting abnormal loads and postural deviations during elderly people sitting, lying down, or getting up, thereby assisting in furniture design optimization.

[0003] Existing safety assessment technologies for age-friendly furniture suffer from a fundamental limitation: their risk detection mechanisms are passive and lack foresight. Current methods generally rely on post-hoc analysis of predetermined actions, typically identifying risks only after users have completed actions such as sitting, standing, and walking, based on observed postural changes and force distribution. They lack mechanisms for active perturbation and critical condition simulation, failing to provide early warnings of potential instability before actions occur. When the user or furniture is in a stable surface state, they cannot actively apply micro-perturbations to verify the structural or postural safety margins, making it difficult to identify hidden instability factors in a timely manner. This passive assessment approach results in delayed risk identification, a lack of predictability and exploratory capabilities, and an inability to support proactive safety assurance in real-world interaction scenarios for elderly users. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides an AI-based method for assessing the safety risks of age-friendly furniture, which solves the problem that existing assessment methods lack active perturbation and forward-looking simulation mechanisms, resulting in delayed risk identification and the inability to detect potential instability hazards before elderly users actually interact with the furniture.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides an AI-based method for assessing the safety risks of age-friendly furniture, comprising:

[0008] Construct a virtual body for the user; and form a physiological-psychological mapping relationship by integrating the intensity of the user's facial micro-expressions with biomechanical parameters;

[0009] The virtual body is placed in a virtual 3D environment, and the mismatch risk points between the user's body and the furniture to be evaluated are located by applying simulated perturbations; the biomechanical parameters under the risk points are extracted and the physical risk value is determined.

[0010] Based on biomechanical parameters, the psychological stress value is determined by dynamic extrapolation through the physiological-psychological state mapping relationship; the physical risk value and the psychological stress value are integrated to generate the current safety risk assessment result.

[0011] Based on virtual body simulation of the physiological trajectory of aging and frailty, long-term safety risk prediction is performed on furniture in the same virtual 3D environment.

[0012] Based on the current and long-term safety risk assessment results, we will output a concrete, personalized assessment report on user-furniture interaction and quantitative modification suggestions for furniture structure.

[0013] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method for forming a physiological-psychological mapping relationship includes:

[0014] The system guides users to complete a standard sequence of movements, obtains facial video streams and motion information of joints throughout the body, calculates the mechanical relationships between body segments, and obtains biomechanical parameters. These biomechanical parameters then drive a standard human body structure to form a virtual body that reflects the user's physiological characteristics.

[0015] Micro-expression intensity values ​​are obtained from facial video streams and aligned with biomechanical parameters by timestamps; and a mapping function from biomechanical parameters to micro-expression intensity values ​​is constructed through nonlinear regression analysis to form a physiological-psychological mapping relationship.

[0016] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method for locating the mismatch risk points between the user's body and the furniture to be assessed by applying simulated perturbations includes:

[0017] The virtual body is placed into a virtual 3D environment containing the furniture to be evaluated, according to the spatial relationships of the actual use scenario.

[0018] Drive the virtual body to execute a standardized daily behavior sequence. During the interaction between the virtual body and the furniture to be evaluated, simulate typical instability conditions and apply mechanical disturbances to the core joints of the virtual body and the load-bearing structure of the furniture to be evaluated to obtain dynamic disturbance response data under the disturbance.

[0019] Based on dynamic disturbance response data, the ability of the virtual body to maintain static balance and dynamic motion continuity under mechanical disturbance is monitored; when any stability assessment index based on dynamic disturbance response data exceeds the instability judgment threshold set for elderly users, the interaction state is judged as a mismatch risk point between the user's body and the furniture to be evaluated.

[0020] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method of applying mechanical disturbances to the core joints of the virtual body and the load-bearing structure of the furniture to be assessed through typical instability simulation includes:

[0021] A multi-type mechanical perturbation library is established based on the physics engine and dynamic constraints of the virtual 3D environment.

[0022] At key posture nodes where the virtual body executes a standardized daily behavior sequence, perturbation modes are called from a multi-type mechanical perturbation library based on preset frailty parameters of the elderly, and the direction and amplitude parameters of the mechanical perturbation vector are set to simulate the posture deviations generated by elderly users during the interaction process.

[0023] Displacement and stiffness disturbances are applied to the load-bearing and support structures of the furniture to be evaluated to simulate the changes in support capacity caused by insufficient structural strength and loose connections of the furniture to be evaluated.

[0024] Under various types of mechanical disturbances, the mechanical transmission path and energy distribution between the virtual body and furniture are obtained, and the corresponding dynamic disturbance response data are output.

[0025] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method for determining the physical risk value includes:

[0026] During the process of the virtual body performing standardized daily behavior sequences, biomechanical parameters corresponding to the mismatch risk points between the user's body and the furniture to be evaluated are acquired in real time.

[0027] The deviation of real-time acquired biomechanical parameters from preset safety thresholds is compared; based on the degree of deviation, the deviation magnitude and duration of each biomechanical parameter are comprehensively evaluated to obtain a physical risk value.

[0028] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method for determining the psychological stress value includes:

[0029] The biomechanical parameters corresponding to the mismatch risk points are input into the physiological-psychological mapping relationship; the biomechanical parameters are converted into a multidimensional psychological stress feature vector through the mapping function contained in the physiological-psychological mapping relationship.

[0030] Based on user physiological characteristics, the multidimensional psychological stress feature vector is dynamically weighted, fused, and normalized to output a quantitative value of the potential psychological stress level of users with mismatched risk points, which is used as the psychological stress value.

[0031] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method for generating the current safety risk assessment result includes:

[0032] By analyzing the interaction between physical risk value and psychological stress value, weights are dynamically assigned to physical risk value and psychological stress value based on the interaction; and a comprehensive risk score is obtained by weighting physical risk value and psychological stress value.

[0033] The comprehensive risk score is mapped to a risk level range to determine the safety risk assessment level; based on the safety risk assessment level and the comprehensive risk score, the current safety risk assessment result is generated.

[0034] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method for predicting long-term safety risks includes:

[0035] Based on the virtual body and according to the typical pathological characteristics of frailty syndrome, a degenerative virtual body is constructed.

[0036] The degenerate virtual body is placed in a virtual 3D environment and performs a standardized sequence of daily behaviors; new mismatch risk points between the degenerate virtual body and the furniture design to be evaluated are located by applying simulated perturbations.

[0037] Based on the degenerative biomechanical parameters corresponding to the new mismatch risk points, the physical risk value is determined; and the degenerative biomechanical parameters are input into the physiological-psychological mapping relationship to predict the user's psychological stress level and determine the psychological stress value.

[0038] The physical risk value and psychological stress value corresponding to the new mismatch risk point are integrated to generate a long-term safety risk assessment result.

[0039] As a preferred embodiment of the AI-based safety risk assessment method for age-friendly furniture described in this invention, the method for outputting a concrete user-furniture interaction personalized assessment report and quantitative modification suggestions for the furniture structure includes:

[0040] Based on current and long-term safety risk assessment results, a personalized assessment report on user-furniture interaction is generated; and through 3D animation, high-risk interaction scenarios between the virtual body and the furniture to be assessed in a virtual 3D environment are vividly reproduced.

[0041] Based on the key biomechanical parameters and environmental spatial factors that lead to instability in high-risk interaction scenarios, inverse simulation is used to derive furniture structure optimization parameters applicable to the current state and the long-term degradation state, and to generate quantitative modification suggestions for furniture structures.

[0042] Secondly, this invention provides an AI-based safety risk assessment system for age-friendly furniture, including:

[0043] The relationship modeling module is used to construct the user's virtual body; and establishes a physiological-psychological mapping relationship by integrating the intensity of the user's facial micro-expressions with biomechanical parameters;

[0044] The active perturbation module is used to place the virtual body into a virtual 3D environment, apply simulated perturbations to locate the mismatch risk points between the user's body and the furniture to be evaluated, and extract the biomechanical parameters at the risk points to determine the physical risk value.

[0045] The assessment module is used to dynamically extrapolate based on biomechanical parameters and through the physiological-psychological state mapping relationship to determine the psychological stress value; it integrates the physical risk value and the psychological stress value to generate the current safety risk assessment result.

[0046] The prediction module is used to predict the long-term safety risks of furniture in the same virtual 3D environment based on the physiological trajectory of aging and frailty simulated by virtual bodies.

[0047] The modification suggestion module is used to output a concrete personalized assessment report of user-furniture interaction and quantitative modification suggestions for furniture structure based on the current and long-term safety risk assessment results.

[0048] The beneficial effects of this invention are as follows: By introducing mechanical disturbances and simulating typical instability conditions in a virtual 3D environment, this invention actively triggers the dynamic interactive response between the user's virtual body and the furniture to be evaluated, breaking through the limitations of traditional passive risk identification methods. By applying multiple types of mechanical disturbances to the core joints of the virtual body and the load-bearing structure of the furniture, it can reveal hidden instability risk points under seemingly stable postures, achieving early perception of potential dangers. It can not only accurately capture the virtual body's posture deviation, changes in support force, and balance recovery ability under disturbances, but also adaptively adjust the intensity and direction of disturbances according to different physiological characteristics of the elderly, quantitatively assessing the stability margin during the interaction process. Therefore, this invention achieves a shift from "post-event identification" to "pre-event warning," significantly improving the safety level and forward-looking assessment capability of age-friendly furniture under real-world usage conditions. Attached Figure Description

[0049] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0050] Figure 1This is a flowchart of the AI ​​assessment method for safety risks of age-friendly furniture in this invention;

[0051] Figure 2 This is a schematic diagram of the AI ​​assessment system for safety risks of age-friendly furniture in this invention;

[0052] Figure 3 This is a flowchart illustrating the generation of the physiological-psychological mapping relationship in this invention;

[0053] Figure 4 This is a flowchart illustrating the process of locating the mismatch risk points between the user and the furniture to be evaluated in this invention. Detailed Implementation

[0054] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0055] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0056] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0057] Reference Figure 1 , Figure 2 , Figure 3 and Figure 4 As one embodiment of the present invention, this embodiment provides an AI-based method for assessing the safety risks of age-friendly furniture, including the following steps:

[0058] Methods for establishing a physiological-psychological mapping relationship include:

[0059] The system guides users to complete a standard sequence of movements, obtains facial video streams and full-body joint movement information, calculates the mechanical relationships between body segments, and obtains biomechanical parameters. These biomechanical parameters then drive a standard human body structure to form a virtual body that reflects the user's physiological characteristics.

[0060] Specifically, in a standardized experimental environment, users are guided to perform a pre-designed sequence of standard movements covering typical daily activities such as sitting, standing, walking, bending over, and getting up through voice and animation demonstrations. During this process, a high-resolution camera is used to synchronously capture the user's facial video stream at a specific frame rate, while a depth-sensing camera is used to synchronously capture the user's full-body joint motion information at the same time reference.

[0061] Furthermore, based on the user's full-body joint motion information, the angular velocity and angular acceleration of the joint motion trajectory are calculated through numerical differentiation; combined with the inertial parameters of the standard human body model, the interaction force vector and torque vector between adjacent body segments are recursively calculated according to the Newton-Euler equations. The calculation expressions are as follows:

[0062] ;

[0063] ;

[0064] in, It acts on the first Force vectors of proximal joints of each body segment It is the first The quality of each body segment It acts on the first The first body segment (i.e., the first) Force vector of the proximal joint of the distal segment of each body segment. It is the first The linear acceleration vector of the center of mass of each body segment. It is an index variable for body segments. It acts on the first Torque vector of proximal joints of each body segment It acts on the first Torque vector of proximal joints of each body segment From the first The vector of the proximal joint of each body segment pointing towards its center of mass. It is the first The inertial tensor of each body segment It is the first The angular acceleration vector of each body segment It is the first The angular velocity vector of each body segment, From the first Each body segment's proximal joint points towards its distal joint (i.e., the first...) Vectors of the proximal joints of each body segment;

[0065] After completing the recursive calculations for all body segments, the difference in the interaction torque vectors between body segments at the joints yields the joint net torque vector. Based on the joint net torque, muscle activation and force values ​​are calculated, and the expression is:

[0066] ;

[0067] ;

[0068] in, It is the first The activation level of a mass muscle. It is the total number of muscles that cross the target joint. Indicates muscle. It is the first The lever arm vector of a muscle relative to the center of the joint. It is the first The maximum isometric contractile force of a muscle mass. It is the first The unit direction vector of a muscle mass. It is the net torque vector of the joint. It is the first The force value of a muscle group;

[0069] The user's biomechanical parameters, such as joint torque vectors, muscle activation and force values, are input into a standard human body model as core driving parameters; the bone size, muscle mechanical properties and joint dynamic parameters are recalibrated to complete personalized initialization and generate a virtual body that matches the user's physiological characteristics.

[0070] Micro-expression intensity values ​​are obtained from facial video streams and aligned with biomechanical parameters by timestamps; and a mapping function from biomechanical parameters to micro-expression intensity values ​​is constructed through nonlinear regression analysis to form a physiological-psychological mapping relationship.

[0071] Specifically, key feature points such as the eyebrows, corners of the eyes, and corners of the mouth are extracted from facial video streams and their movement trajectories are tracked. The temporal changes of geometric features such as the distance between the eyebrows and the tilt of the corners of the mouth are calculated and matched with a pre-built facial action library. The facial action library is built based on the Facial Action Coding System (FACS). By collecting facial videos of subjects under known exertion and tension, the geometric configurations of key feature points such as the eyebrows, corners of the eyes, and corners of the mouth corresponding to action units such as AU4 (brow muscle contraction) and AU7 (eyelid tightening) are labeled. By comparing the geometric features extracted in real time with the configurations in the facial action library, and based on preset quantitative standards (for example, a decrease in the distance between the eyebrows exceeding the baseline value by more than 15% corresponds to intensity level 3, and a drooping angle of the corners of the mouth exceeding the baseline value by more than 10 degrees corresponds to intensity level 4), standardized facial micro-expression intensity values ​​are obtained.

[0072] The time series of micro-expression intensity values ​​and biomechanical parameter time series were precisely aligned using a unified time reference. A neural network model was employed for nonlinear regression analysis: a multilayer perceptron was constructed, comprising an input layer, two hidden layers (64 and 32 nodes respectively, using ReLU activation function), and an output layer (using Sigmoid activation function). The aligned biomechanical parameters were used as input features, with the corresponding micro-expression intensity values ​​as the target labels. The model was trained using the Adam optimizer (exemplary learning rate 0.001) with the objective of minimizing the mean squared error (exemplary batch size 32, training epochs ≥ 100).

[0073] Biomechanical parameters are Z-score normalized and then input into a trained neural network model. The input layer of the neural network model receives the normalized biomechanical parameters, which undergo nonlinear feature transformation through two hidden layers. Finally, the sigmoid activation function of the output layer outputs a scalar value between 0 and 1. This scalar value is a quantitative representation of the user's psychological stress state corresponding to the current biomechanical parameters, where 0 represents no psychological stress and 1 represents the maximum psychological stress state. Through this end-to-end mapping process, a quantitative conversion from physical biomechanical parameters to psychological stress state is achieved, thus forming a physiological-psychological mapping relationship between biomechanical parameters and the user's psychological stress state.

[0074] Methods for locating the risk points of mismatch between the user's body and the furniture to be evaluated by applying simulated perturbations include:

[0075] Existing safety assessment technologies for age-friendly furniture suffer from significant limitations due to their passivity and static nature. Their assessments rely entirely on post-hoc analysis of predetermined actions, lacking proactive risk detection mechanisms and failing to identify potential compatibility defects before actual user use. This passive response model struggles to simulate sudden instability situations in real-world scenarios, resulting in incomplete risk assessments, insufficient early warning capabilities, and an inability to provide proactive safety guarantees for the elderly.

[0076] The virtual body is placed into a virtual 3D environment containing the furniture to be evaluated, according to the spatial relationships of the actual use scenario.

[0077] Specifically, a 3D laser scanning device is used to acquire precise geometric point cloud data of the user's living area, while a multi-view color image of the scene is acquired through a multi-view camera array; the point cloud data and the multi-view images are then registered and fused to generate a virtual 3D environment.

[0078] The 3D model of the furniture to be evaluated is registered to the virtual 3D environment according to its position and orientation in the real environment; the virtual body is placed into the corresponding position in the virtual 3D environment according to the user's initial position and posture in the real scene through coordinate transformation.

[0079] The system drives the virtual body to execute standardized daily behavior sequences. During the interaction phase between the virtual body and the furniture to be evaluated, it simulates typical instability conditions and applies mechanical disturbances to the core joints of the virtual body and the load-bearing structure of the furniture to be evaluated, thereby obtaining dynamic disturbance response data under the disturbance.

[0080] Specifically, the virtual body is driven by a physics engine to execute standardized daily behavior sequences. During the interaction phase where the virtual body comes into contact with the furniture to be evaluated, mechanical disturbances simulating sudden muscle strength decline are applied to the core joints of the virtual body's hip, knee, and ankle joints through typical instability simulation. At the same time, displacement and stiffness disturbances simulating sudden changes in structural support capacity are applied to the load-bearing structures of the seat, armrests, and backrest of the furniture to be evaluated. The physics engine calculates the dynamic response of the virtual body and the furniture to be evaluated under mechanical disturbances in real time, and obtains dynamic disturbance response data including joint torque, contact pressure, and changes in body posture.

[0081] Based on dynamic disturbance response data, the ability of the virtual body to maintain static balance and dynamic motion continuity under mechanical disturbance is monitored; when any stability assessment index based on dynamic disturbance response data exceeds the instability judgment threshold set for elderly users, the interaction state is judged as a mismatch risk point between the user's body and the furniture to be evaluated.

[0082] It should be noted that the process of setting the instability judgment threshold includes: based on the biomechanical characteristics and safety standards of the elderly population, setting corresponding safety boundary values ​​for key stability assessment indicators in the dynamic disturbance response data; for example, setting an instability judgment threshold for joint torque that does not exceed 85% of the peak torque of healthy elderly people under this action, setting an instability judgment threshold for contact pressure that does not exceed 70% of the tissue tolerance pressure limit, and setting an instability judgment threshold for the center of mass shift in body posture changes that does not exceed 60% of the distance to the support surface boundary.

[0083] It should be noted that the following are examples of risk points where the user's body is mismatched with the furniture to be evaluated: risk points of postural instability and imbalance, where the knee joint torque exceeds 85% of the peak torque of a healthy elderly person in this action when standing up; risk points of action execution obstruction, where the shoulder joint abduction torque needs to increase by more than 30% to complete the support action; risk points of insufficient structural support, where an instantaneous displacement of more than 5mm occurs when bearing weight and causes compensatory lateral flexion of the lumbar spine; and risk points of contact interface mismatch, where the contact pressure in the ischial tuberosity area continuously exceeds 70% of the tissue's tolerance pressure limit.

[0084] By constructing a virtual environment and introducing an active disturbance mechanism, in-depth analysis and proactive early warning of potential risks are achieved. By simulating typical instability conditions and quantitatively analyzing dynamic responses, mismatch risk points in the human-furniture interaction process can be accurately located. This elevates risk assessment from passive observation to proactive detection, improving the accuracy and predictability of safety assessments for age-friendly furniture and providing a scientific basis for personalized safety modifications.

[0085] Methods for applying mechanical perturbations to the core joints of a virtual body and the load-bearing structure of the furniture being evaluated through simulation of typical instability conditions include:

[0086] Based on the physics engine and dynamic constraints of the virtual 3D environment, a multi-type mechanical perturbation library is established.

[0087] Specifically, based on a physics engine-based dynamics simulation framework, a multi-type mechanical perturbation library is constructed, including a sudden muscle strength decay mode, a support surface sliding mode, and a structural instantaneous deformation mode. The sudden muscle strength decay mode is achieved by setting the activation level of specific muscle groups in the virtual body to decrease from a baseline value by a specified percentage within a preset time window. For example, the preset time window can be set to a range of 200 to 500 milliseconds, simulating a sudden muscle strength decay condition by linearly decreasing the quadriceps activation level from 100% to 60% within a 300-millisecond time window. The support surface sliding mode is achieved by... The support surface is simulated by applying a sudden tangential velocity vector. For example, an instantaneous horizontal tangential velocity of 0.5 m / s is applied to the seat support surface. The instantaneous structural deformation mode is achieved by applying boundary conditions of transient stiffness coefficient decay and displacement change to the load-bearing structure of the furniture to be evaluated. For example, the stiffness coefficient of the chair leg support structure is decayed to 50% of the initial value within 200 milliseconds and an instantaneous displacement of 10 mm is generated. Each perturbation mode in the multi-type mechanical perturbation library contains preset mechanical perturbation vector direction parameters, amplitude parameters, and duration parameters, which can be adaptively adjusted according to the physiological characteristics of elderly users.

[0088] At key posture nodes where the virtual body executes a standardized daily behavior sequence, perturbation modes are called from a multi-type mechanical perturbation library based on preset frailty parameters of the elderly, and the direction and amplitude parameters of the mechanical perturbation vector are set to simulate the posture deviations generated by elderly users during interaction.

[0089] Specifically, the preset frailty parameters for the elderly include muscle strength decline level, balance ability score and joint stability index, which are obtained through a comprehensive geriatric assessment scale and biomechanical testing; based on the muscle strength decline level, the corresponding sudden muscle strength decline mode is automatically matched from a multi-type mechanical disturbance library; based on the balance ability score, the support surface sliding mode is called; and based on the joint stability index, the instantaneous structural deformation mode is selected.

[0090] During the key posture node of the sitting-up phase of the virtual body, when the hip flexion angle is detected to reach an exemplary 60 degrees, based on the moderate muscle strength decay level, the quadriceps activation is invoked, and a sudden muscle strength decay mode is implemented, which linearly decreases from 100% to 50% within an exemplary 300 milliseconds; the direction of the mechanical disturbance vector is set to be consistent with the direction of the human body's instability trend.

[0091] The perturbation amplitude parameter is dynamically adjusted according to the real-time joint load level. For example, when the physics engine detects that the knee joint torque value reaches 40 N·m, the perturbation amplitude is adjusted to 60% of the baseline value; when it is in the range of 25-40 N·m, 80% of the baseline value is used as the perturbation amplitude; when it is below 25 N·m, the perturbation amplitude is increased to 120% of the baseline value. This negative adjustment mechanism avoids evaluation distortion caused by excessive interference under high load conditions, and ensures that potential instability risks can be fully stimulated under low load conditions. The physics engine simulates in real time the sudden increase in knee flexion torque and the posture deviation of the body's center of gravity shifting backward, accurately simulating the difficulty of getting up due to insufficient muscle strength in elderly users. At the same time, based on the balance ability score, a compensatory perturbation opposite to the direction of instability is applied to the support surface of the assistive device, completely reproducing the dynamic imbalance process caused by control disorder during the interaction of elderly users.

[0092] Displacement and stiffness disturbances are applied to the load-bearing and support structures of the furniture to be evaluated to simulate the changes in support capacity caused by insufficient structural strength and loose connections.

[0093] It should be noted that displacement disturbance is achieved by applying abrupt displacement vectors at key connection points of the load-bearing and support structures of the furniture under evaluation; stiffness disturbance is achieved by dynamically adjusting the material stiffness parameters of the load-bearing and support structures of the furniture under evaluation, for example, reducing the bending stiffness of the armrest support tube by 40% from its initial value within 200 milliseconds.

[0094] The application of displacement and stiffness disturbances is synchronized with the interaction of the virtual body. When the contact force between the virtual body and the furniture to be evaluated is detected to be more than 1.5 times the normal expected contact force and lasts for more than 200 milliseconds, a disturbance sequence is triggered. The dynamic response under the combined action of displacement and stiffness disturbances is calculated, resulting in non-uniform settlement and structural deformation of the support surface of the furniture to be evaluated, which in turn causes the virtual body's center of gravity to shift and attitude instability, accurately reproducing the balance instability process caused by structural defects of the furniture to be evaluated.

[0095] Under various types of mechanical disturbances, the mechanical transmission path and energy distribution between the virtual body and furniture are obtained, and the corresponding dynamic disturbance response data are output.

[0096] Specifically, based on the physics engine of the virtual 3D environment, the dynamics of the interaction between the virtual body and the furniture to be evaluated are solved in real time. Under the condition of multiple types of mechanical disturbances acting simultaneously, the force state of the virtual body and the furniture to be evaluated and their response changes in the time domain are calculated respectively. According to Newton's second law and rigid body dynamics equations, the physics engine solves the linear acceleration, angular acceleration, force vector and energy transfer of each virtual body and furniture structural node in each simulation time step, and outputs the corresponding force data, attitude angle data and energy change data of the virtual body and furniture structural nodes in the form of time series.

[0097] On the virtual body side, the reaction forces, joint torques, and angular accelerations of each core joint (including hip, knee, ankle, shoulder, and elbow) are extracted to calculate the mechanical transmission path within the body; the joint torques within each simulation time step are recorded and stored to generate joint torque time series data; at the same time, the changes in normal force and friction force at the contact surface between the body and the furniture are continuously sampled to form a contact reaction force curve.

[0098] It should be noted that the expression for calculating the contact reaction force is:

[0099] ;

[0100] in, time The scalar value of the contact reaction force. It is a moment The square of the normal reaction force component, It is a moment The square of the tangential frictional force component;

[0101] It should be noted that the mechanical transmission path characterizes the dynamic distribution relationship of force and energy from the external disturbance point through the body segments to the body's center of mass. For example, by establishing a directed graph structure with joint nodes as vertices, using the joint torque vector magnitude as the edge weight, and employing the shortest energy transmission path algorithm to determine the main energy flow direction, an energy transmission path diagram of the virtual body under disturbance is obtained.

[0102] During the solution process, the physics engine updates the attitude angles of each body segment in real time based on the integral results of the joint node angular acceleration, and outputs the corresponding angle change sequence according to the simulation step size, thereby obtaining the attitude angle change sequence, which is used to describe the attitude stability changes of the virtual body during the disturbance process.

[0103] On the furniture side, by monitoring the stress tensor distribution and displacement response at the load-bearing structural nodes (such as seat support points, armrest connection points, and backrest force-bearing nodes), the energy absorption rate and deformation energy inside the furniture are calculated. The physics engine updates the strain energy density field of each load-bearing structural node in real time using the finite element approximation method, and constructs an energy distribution matrix with the load-bearing structural nodes and time step as indexes to characterize the energy dissipation distribution of the furniture during the disturbance process.

[0104] It should be noted that the expression for calculating the strain energy density at the nodes of the furniture load-bearing structure is as follows:

[0105] ;

[0106] in, It is the first Each furniture load-bearing structural node at time strain energy density, It is the first At the moment of each furniture load-bearing structure node The stress tensor, It is the first At the moment of each furniture load-bearing structure node The strain tensor;

[0107] The energy changes of the virtual body and the furniture are registered on a unified time axis to keep the energy change data of both sides synchronized in the time dimension. Based on this, the interaction power between the virtual body and the furniture is calculated according to the power integral relationship, and a power curve is plotted with time as the horizontal axis. The peak value of the power curve corresponds to the instantaneous energy impact intensity, and the integral area of ​​the curve represents the total energy transfer during the entire disturbance process.

[0108] It should be noted that the expression for calculating the interaction power between the virtual body and the furniture is:

[0109] ;

[0110] in, It is a moment The power of interaction between virtual bodies and furniture It is a moment The interaction force vector between the virtual body and the furniture. It is a moment The relative velocity vector at the contact interface between the virtual body and the furniture;

[0111] Based on this, multi-dimensional simulation output data, including joint torque time series data, contact reaction force curves, attitude angle change sequences, energy distribution matrices and power curves, are synchronously sampled and formatted to form a standardized dynamic disturbance response dataset.

[0112] Methods for determining physical risk values ​​include:

[0113] During the process of the virtual body performing standardized daily behavior sequences, biomechanical parameters corresponding to the mismatch risk points between the user's body and the furniture to be evaluated are acquired in real time.

[0114] Specifically, based on the dynamic disturbance response data of the physics engine in the virtual 3D environment, the joint torque, angular velocity, angular acceleration, and contact reaction force of the core joints of the virtual body are extracted at each simulation time step; simultaneously, the normal reaction force and tangential friction force components of the contact interface between the virtual body and the furniture to be evaluated are acquired to obtain the scalar value of the contact reaction force; combined with the sequence of virtual body posture angle changes and the change in center of mass position, characteristic parameters reflecting the body's balance state and dynamic stability are extracted. This yields a set of biomechanical parameters at the mismatch risk points.

[0115] The deviation of real-time acquired biomechanical parameters from preset safety thresholds is compared; based on the degree of deviation, the deviation magnitude and duration of each biomechanical parameter are comprehensively evaluated to obtain a physical risk value.

[0116] Specifically, the process of setting the preset safety threshold includes: establishing a biomechanical database of healthy elderly people by collecting biomechanical data of healthy elderly volunteers performing standardized daily behavior sequences, analyzing the safe distribution range of each biomechanical parameter, and taking the exemplary 95th percentile as the preset safety threshold.

[0117] Furthermore, the relative percentage deviation between the real-time values ​​of each biomechanical parameter and the corresponding preset safety threshold is calculated. When a biomechanical parameter exceeds the preset safety threshold, the duration and magnitude of the exceedance are recorded. The deviation of each biomechanical parameter is weighted and fused, with the weighting coefficient determined according to the contribution of each parameter to the fall risk. For example, the weight of hip joint torque is 0.3, the weight of knee joint torque is 0.4, and the weight of contact reaction force is 0.3. The cumulative deviation of each biomechanical parameter within the assessment time window is calculated by integration, and finally, a physical risk value in the range of 0-1 is obtained, where 0 represents no risk and 1 represents the maximum risk.

[0118] Methods for determining psychological stress levels include:

[0119] The biomechanical parameters corresponding to the mismatch risk points are input into the physiological-psychological mapping relationship; through the mapping function contained in the physiological-psychological mapping relationship, the biomechanical parameters are converted into a multidimensional psychological stress feature vector.

[0120] Specifically, the biomechanical parameters corresponding to the mismatch risk points between the user's body and the furniture to be evaluated are Z-score normalized, and the normalized biomechanical parameters are input into a neural network model. The input layer of the neural network model receives the normalized biomechanical parameters, which are then subjected to nonlinear feature transformation through two hidden layers. The first hidden layer maps the input features to an exemplary 64-dimensional feature space, and the second hidden layer further compresses them to an exemplary 32-dimensional feature space. Finally, the activation values ​​of the exemplary 32 neurons in the second hidden layer together constitute a multidimensional psychological stress feature vector reflecting the user's level of tension, perceived effort, and sense of unease.

[0121] Based on user physiological characteristics, the multidimensional psychological stress feature vector is dynamically weighted, fused, and normalized to output a quantitative value of the potential psychological stress level of users with mismatched risk points, which is used as the psychological stress value.

[0122] It should be noted that a dynamic weighting coefficient adjustment rule is established based on the user's physiological characteristics. The weight coefficients of each dimension of the multidimensional psychological stress feature vector are dynamically adjusted according to the user's muscle strength decline level, balance ability score, and joint stability index. For example, for users with severe muscle strength decline, the weight coefficient of the effort perception dimension is increased to 0.5; for users with low balance ability scores, the weight coefficient of the instability dimension is increased to 0.4; and for users with poor joint stability index, the weight coefficient of the tension dimension is increased to 0.6. The multidimensional psychological stress feature vector is fused into a single psychological stress intermediate value through weighted summation. Then, the Sigmoid function is used to map the psychological stress intermediate value to the range of 0-1, and the quantitative value of the psychological stress level is output, where 0 represents no psychological stress and 1 represents the maximum psychological stress.

[0123] Methods for generating current security risk assessment results include:

[0124] By analyzing the interaction between physical risk value and psychological stress value, weights are dynamically assigned to physical risk value and psychological stress value based on the interaction; and a comprehensive risk score is obtained by weighting physical risk value and psychological stress value.

[0125] It should be noted that by comparing and analyzing the correspondence between physical risk values ​​and psychological stress values ​​in historical case data, the statistical correlation between the two types of risk values ​​in specific interaction scenarios was identified. For example, in the case of posture instability risk points, statistics show that when the physical risk value exceeds 0.7, 87% of the cases are accompanied by an increase in psychological stress value of more than 0.2; in the case of operational difficulty risk points, 83% of the cases show that a psychological stress value of 0.6 leads to an increase in physical risk value of more than 10%. Through this statistical analysis based on a large number of case data, the mutual influence relationship between physical risk values ​​and psychological stress values ​​was determined.

[0126] Based on the quantitative impact of physical risk and psychological stress, a dynamic weighting mechanism is established: For example, when the Pearson correlation coefficient between physical risk and psychological stress is greater than 0.7, both are assigned equal weights (0.5 each); when the correlation coefficient is between 0.3 and 0.7, the weight ratio is adjusted according to the type of risk point; when the correlation coefficient is less than or equal to 0.3, physical risk is assigned a weight of 0.9, and psychological stress is assigned a weight of 0.1. A comprehensive risk score within the range of 0-1 is obtained through weighted calculation.

[0127] The comprehensive risk score is mapped to a risk level range to determine the safety risk assessment level; based on the safety risk assessment level and the comprehensive risk score, the current safety risk assessment result is generated.

[0128] It should be noted that the risk level ranges are set based on statistical data on safety accidents among the elderly and clinical medical research. The comprehensive risk score is divided into four levels: a comprehensive risk score of less than 0.3 is low risk and indicates safety; 0.3 to 0.6 is medium risk and indicates caution; 0.6 to 0.8 is high risk and indicates warning; and greater than 0.8 is extremely high risk and indicates danger. The current safety risk assessment results include both the comprehensive risk score and the corresponding risk level label.

[0129] Methods for predicting long-term security risks include:

[0130] Based on the virtual body, and according to the typical pathological characteristics of frailty syndrome, a degenerative virtual body is constructed.

[0131] It should be noted that the typical pathological features of frailty syndrome include: muscle loss due to sarcopenia and limited range of motion due to joint degeneration, as well as decreased neuromuscular control. The key physiological parameters of the virtual body are degraded to construct a degraded virtual body that reflects changes in the user's physiological state.

[0132] The degenerate virtual body is placed in a virtual 3D environment and performs a standardized sequence of daily behaviors; new mismatch risk points between the degenerate virtual body and the furniture design to be evaluated are located by applying simulated perturbations.

[0133] It should be noted that the constructed degenerate virtual body is placed into the corresponding position in the virtual 3D environment containing the furniture to be evaluated through coordinate transformation; the degenerate virtual body is driven to execute the same standardized daily behavior sequence as the current evaluation stage; during the interaction stage where the degenerate virtual body comes into contact with the furniture to be evaluated, mechanical disturbances simulating sudden decline in muscle strength are applied to the degenerate virtual body through typical instability simulation, while displacement disturbances and stiffness disturbances simulating sudden changes in structural support capacity are applied to the furniture to be evaluated;

[0134] Based on the dynamic disturbance response data generated by the degraded virtual body under mechanical disturbance, the ability of the degraded virtual body to maintain static balance and dynamic motion continuity is monitored. When any stability assessment index based on the dynamic disturbance response data exceeds the instability judgment threshold set for elderly users, the interaction state is judged as a new mismatch risk point between the degraded virtual body and the furniture design to be evaluated.

[0135] Based on the degenerative biomechanical parameters corresponding to the new mismatch risk points, the physical risk value is determined; and the degenerative biomechanical parameters are input into the physiological-psychological mapping relationship to predict the user's psychological stress level and determine the psychological stress value.

[0136] It should be noted that during the interaction between the degraded virtual body and the furniture to be evaluated, the degraded biomechanical parameters corresponding to new mismatch risk points are extracted in real time. The degraded biomechanical parameters are compared with the preset safety threshold, and the relative deviation percentage, exceedance range, and duration of the degraded biomechanical parameters are calculated. According to the relative contribution of each degraded biomechanical parameter to the fall risk, corresponding weights are assigned. Through weighted fusion and integral calculation, the physical risk value in the range of 0-1 is finally obtained.

[0137] The degenerative biomechanical parameters corresponding to the new mismatch risk points are normalized using Z-score; then, the normalized degenerative biomechanical parameters are input into the neural network model; the psychological stress value in the range of 0-1 is output to quantify and predict the psychological stress level of users when facing potential risks in future physiological states.

[0138] The physical risk value and psychological stress value corresponding to the new mismatch risk point are integrated to generate a long-term safety risk assessment result.

[0139] Specifically, based on the interaction between physical risk value and psychological stress value, weights are dynamically assigned to the physical risk value and psychological stress value corresponding to new mismatch risk points; a weighted calculation is performed to obtain a long-term comprehensive risk score; the long-term comprehensive risk score is mapped to a risk level range to determine the safety risk assessment level; and a long-term safety risk assessment result containing risk level identifier and comprehensive risk score is generated based on the safety risk assessment level.

[0140] Methods for generating concrete, personalized user-furniture interaction assessment reports and quantitative recommendations for furniture structural modifications include:

[0141] Based on current and long-term safety risk assessment results, a personalized assessment report on user-furniture interaction is generated; and through 3D animation, high-risk interaction scenarios between the virtual body and the furniture to be assessed in a virtual 3D environment are vividly reproduced.

[0142] It should be noted that by integrating the risk level labels and comprehensive risk scores from the current and long-term safety risk assessment results, and combining the dynamic disturbance response data corresponding to the mismatch risk points between the user's body and the furniture to be assessed, as well as the new mismatch risk points between the degraded virtual body and the design of the furniture to be assessed, a multi-dimensional personalized assessment report of user-furniture interaction is generated, which includes a risk assessment summary, risk point distribution map, biomechanical parameter anomaly analysis, and safety warning suggestions.

[0143] Based on a virtual 3D environment, dynamic disturbance response data corresponding to the mismatch risk points between the user's body and the furniture to be evaluated, as well as the new mismatch risk points between the degraded virtual body and the design of the furniture to be evaluated, are extracted. The complete interaction process between the virtual body and the furniture to be evaluated at the moment when the instability judgment threshold is exceeded is reproduced frame by frame through 3D animation. The 3D animation synchronously displays the real-time change curves of key biomechanical parameters such as joint torque and contact reaction force of the core joints of the virtual body, and highlights the key body segments and furniture structural parts that lead to instability with bright colors. At the same time, the dynamic evolution process of posture instability is highlighted through slow motion playback and multi-view switching.

[0144] Based on the key biomechanical parameters and environmental spatial factors that lead to instability in high-risk interaction scenarios, inverse simulation is used to derive furniture structure optimization parameters applicable to the current state and the long-term degradation state, and to generate quantitative modification suggestions for furniture structures.

[0145] It should be noted that, based on key biomechanical parameters and environmental spatial factors in high-risk interaction scenarios, an optimization objective function is established with the goal of improving key stability assessment indicators in dynamic disturbance response data. Key structural parameters such as seat height, armrest position, and backrest angle of the furniture under evaluation are adjusted iteratively through inverse simulation. The dynamic disturbance response data of the virtual body and the optimized furniture under evaluation are recalculated under the same mechanical disturbance conditions until all stability assessment indicators meet the instability judgment threshold requirements. For both the current virtual body and the degraded virtual body in the future, sets of furniture structural optimization parameters for the current state and the future independent degradation state are derived. Combining the existing structural characteristics and processing constraints of the furniture under evaluation, specific structural modification schemes are formulated, generating quantitative modification suggestions that clearly specify the adjustment range of each structural component's dimensions, material replacement specifications, and connection strength requirements.

[0146] This embodiment also provides an AI-based safety risk assessment system for age-friendly furniture, including:

[0147] The relationship modeling module is used to construct the user's virtual body; and establishes a physiological-psychological mapping relationship by integrating the intensity of the user's facial micro-expressions with biomechanical parameters;

[0148] The active perturbation module is used to place the virtual body into a virtual 3D environment, apply simulated perturbations to locate the mismatch risk points between the user's body and the furniture to be evaluated, and extract the biomechanical parameters at the risk points to determine the physical risk value.

[0149] The assessment module is used to dynamically extrapolate based on biomechanical parameters and through the physiological-psychological state mapping relationship to determine the psychological stress value; it integrates the physical risk value and the psychological stress value to generate the current safety risk assessment result.

[0150] The prediction module is used to predict the long-term safety risks of furniture in the same virtual 3D environment based on the physiological trajectory of aging and frailty simulated by virtual bodies.

[0151] The modification suggestion module is used to output a concrete personalized assessment report of user-furniture interaction and quantitative modification suggestions for furniture structure based on the current and long-term safety risk assessment results.

[0152] This embodiment also provides a computer device applicable to the AI ​​assessment method for safety risks of age-friendly furniture, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the AI ​​assessment method for safety risks of age-friendly furniture as proposed in the above embodiment.

[0153] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0154] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the AI ​​assessment method for safety risks of age-friendly furniture as proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0155] In summary, this invention overcomes the limitations of traditional passive risk identification methods by introducing mechanical disturbances and simulating typical instability conditions in a virtual 3D environment to proactively trigger dynamic interaction between the user's virtual body and the furniture being evaluated. By applying various types of mechanical disturbances to the core joints of the virtual body and the load-bearing structure of the furniture, it can reveal hidden instability risk points beneath seemingly stable postures, enabling early detection of potential dangers. Furthermore, it can accurately capture the virtual body's posture shift, changes in support forces, and balance recovery ability under disturbances, and can adaptively adjust the intensity and direction of disturbances according to different physiological characteristics of the elderly, quantifying the stability margin during the interaction process. Therefore, this invention achieves a shift from "post-event identification" to "pre-event warning," significantly improving the safety level and forward-looking assessment capabilities of age-friendly furniture under real-world usage conditions.

[0156] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. An AI-based method for assessing the safety risks of age-friendly furniture, characterized in that: include: Construct a virtual body for the user; and form a physiological-psychological mapping relationship by integrating the intensity of the user's facial micro-expressions with biomechanical parameters; By placing a virtual body into a virtual 3D environment and applying simulated perturbations, the potential mismatch risk points between the user's body and the furniture to be evaluated can be located. The biomechanical parameters at this risk point are extracted and the physical risk value is determined. Specific steps include: The virtual body is placed into a virtual 3D environment containing the furniture to be evaluated, according to the spatial relationships of the actual use scenario. Drive the virtual body to execute a standardized daily behavior sequence. During the interaction between the virtual body and the furniture to be evaluated, simulate typical instability conditions and apply mechanical disturbances to the core joints of the virtual body and the load-bearing structure of the furniture to be evaluated to obtain dynamic disturbance response data under the disturbance. Based on dynamic disturbance response data, the ability of the virtual body to maintain static balance and dynamic motion continuity under mechanical disturbance is monitored; when any stability assessment index based on dynamic disturbance response data exceeds the instability judgment threshold set for elderly users, the interaction state is judged as a mismatch risk point between the user's body and the furniture to be evaluated. A multi-type mechanical perturbation library is established based on the physics engine and dynamic constraints of the virtual 3D environment. At key posture nodes where the virtual body executes a standardized daily behavior sequence, perturbation modes are called from a multi-type mechanical perturbation library based on preset frailty parameters of the elderly, and the direction and amplitude parameters of the mechanical perturbation vector are set to simulate the posture deviations generated by elderly users during the interaction process. Displacement and stiffness disturbances are applied to the load-bearing and support structures of the furniture to be evaluated to simulate the changes in support capacity caused by insufficient structural strength and loose connections of the furniture to be evaluated. Under various types of mechanical disturbances, the mechanical transmission path and energy distribution between the virtual body and furniture are obtained, and the corresponding dynamic disturbance response data are output. Based on biomechanical parameters, the psychological stress value is determined by dynamic extrapolation through the physiological-psychological state mapping relationship; the physical risk value and the psychological stress value are integrated to generate the current safety risk assessment result. Based on virtual body simulation of the physiological trajectory of aging and frailty, long-term safety risk prediction is performed on furniture in the same virtual 3D environment. The specific steps include: Based on the virtual body and according to the typical pathological characteristics of frailty syndrome, a degenerative virtual body is constructed. The degenerate virtual body is placed in a virtual 3D environment and performs a standardized sequence of daily behaviors; new mismatch risk points between the degenerate virtual body and the furniture design to be evaluated are located by applying simulated perturbations. Based on the degenerative biomechanical parameters corresponding to the new mismatch risk points, the physical risk value is determined; and the degenerative biomechanical parameters are input into the physiological-psychological mapping relationship to predict the user's psychological stress level and determine the psychological stress value. The physical risk value and psychological stress value corresponding to the new mismatch risk point are integrated to generate a long-term safety risk assessment result; Based on the current and long-term safety risk assessment results, we will output a concrete, personalized assessment report on user-furniture interaction and quantitative modification suggestions for furniture structure.

2. The AI-based safety risk assessment method for age-friendly furniture as described in claim 1, characterized in that, The methods for forming a physiological-psychological mapping relationship include: The system guides users to complete a standard sequence of movements, obtains facial video streams and motion information of joints throughout the body, calculates the mechanical relationships between body segments, and obtains biomechanical parameters. These biomechanical parameters then drive a standard human body structure to form a virtual body that reflects the user's physiological characteristics. Micro-expression intensity values ​​are obtained from facial video streams and aligned with biomechanical parameters by timestamps; and a mapping function from biomechanical parameters to micro-expression intensity values ​​is constructed through nonlinear regression analysis to form a physiological-psychological mapping relationship.

3. The AI-based safety risk assessment method for age-friendly furniture as described in claim 1, characterized in that, The method for determining the physical risk value includes: During the process of the virtual body performing standardized daily behavior sequences, biomechanical parameters corresponding to the mismatch risk points between the user's body and the furniture to be evaluated are acquired in real time. The deviation of real-time acquired biomechanical parameters from preset safety thresholds is compared; based on the degree of deviation, the deviation magnitude and duration of each biomechanical parameter are comprehensively evaluated to obtain a physical risk value.

4. The AI-based safety risk assessment method for age-friendly furniture as described in claim 1, characterized in that, The methods for determining psychological stress levels include: The biomechanical parameters corresponding to the mismatch risk points are input into the physiological-psychological mapping relationship; the biomechanical parameters are converted into a multidimensional psychological stress feature vector through the mapping function contained in the physiological-psychological mapping relationship. Based on user physiological characteristics, the multidimensional psychological stress feature vector is dynamically weighted, fused, and normalized to output a quantitative value of the potential psychological stress level of users with mismatched risk points, which is used as the psychological stress value.

5. The AI-based safety risk assessment method for age-friendly furniture as described in claim 1, characterized in that, The method for generating the current security risk assessment result includes: By analyzing the interaction between physical risk value and psychological stress value, weights are dynamically assigned to physical risk value and psychological stress value based on the interaction; and a comprehensive risk score is obtained by weighting physical risk value and psychological stress value. The comprehensive risk score is mapped to a risk level range to determine the safety risk assessment level; based on the safety risk assessment level and the comprehensive risk score, the current safety risk assessment result is generated.

6. The AI-based safety risk assessment method for age-friendly furniture as described in claim 1, characterized in that, The methods for outputting a concrete, personalized evaluation report on user-furniture interaction and quantitative recommendations for furniture structural modifications include: Based on current and long-term safety risk assessment results, a personalized assessment report on user-furniture interaction is generated; and through 3D animation, high-risk interaction scenarios between the virtual body and the furniture to be assessed in a virtual 3D environment are vividly reproduced. Based on the key biomechanical parameters and environmental spatial factors that lead to instability in high-risk interaction scenarios, inverse simulation is used to derive furniture structure optimization parameters applicable to the current state and the long-term degradation state, and to generate quantitative modification suggestions for furniture structures.

7. An AI-based safety risk assessment system for age-friendly furniture, based on the AI-based safety risk assessment method for age-friendly furniture as described in any one of claims 1 to 6, characterized in that, include: The relationship modeling module is used to construct the user's virtual body; and establishes a physiological-psychological mapping relationship by integrating the intensity of the user's facial micro-expressions with biomechanical parameters; The active perturbation module is used to place the virtual body into the virtual 3D environment and locate the mismatch risk points between the user's body and the furniture to be evaluated by applying simulated perturbations; The biomechanical parameters at this risk point are extracted and the physical risk value is determined. Specific steps include: The virtual body is placed into a virtual 3D environment containing the furniture to be evaluated, according to the spatial relationships of the actual use scenario. Drive the virtual body to execute a standardized daily behavior sequence. During the interaction between the virtual body and the furniture to be evaluated, simulate typical instability conditions and apply mechanical disturbances to the core joints of the virtual body and the load-bearing structure of the furniture to be evaluated to obtain dynamic disturbance response data under the disturbance. Based on dynamic disturbance response data, the ability of the virtual body to maintain static balance and dynamic motion continuity under mechanical disturbance is monitored; when any stability assessment index based on dynamic disturbance response data exceeds the instability judgment threshold set for elderly users, the interaction state is judged as a mismatch risk point between the user's body and the furniture to be evaluated. A multi-type mechanical perturbation library is established based on the physics engine and dynamic constraints of the virtual 3D environment. At key posture nodes where the virtual body executes a standardized daily behavior sequence, perturbation modes are called from a multi-type mechanical perturbation library based on preset frailty parameters of the elderly, and the direction and amplitude parameters of the mechanical perturbation vector are set to simulate the posture deviations generated by elderly users during the interaction process. Displacement and stiffness disturbances are applied to the load-bearing and support structures of the furniture to be evaluated to simulate the changes in support capacity caused by insufficient structural strength and loose connections of the furniture to be evaluated. Under various types of mechanical disturbances, the mechanical transmission path and energy distribution between the virtual body and furniture are obtained, and the corresponding dynamic disturbance response data are output. The assessment module is used to dynamically extrapolate based on biomechanical parameters and through the physiological-psychological state mapping relationship to determine the psychological stress value; it integrates the physical risk value and the psychological stress value to generate the current safety risk assessment result. The prediction module is used to predict the long-term safety risks of furniture in the same virtual 3D environment based on the physiological trajectory of aging and frailty simulated by a virtual body. Specific steps include: Based on the virtual body and according to the typical pathological characteristics of frailty syndrome, a degenerative virtual body is constructed. The degenerate virtual body is placed in a virtual 3D environment and performs a standardized sequence of daily behaviors; new mismatch risk points between the degenerate virtual body and the furniture design to be evaluated are located by applying simulated perturbations. Based on the degenerative biomechanical parameters corresponding to the new mismatch risk points, the physical risk value is determined; and the degenerative biomechanical parameters are input into the physiological-psychological mapping relationship to predict the user's psychological stress level and determine the psychological stress value. The physical risk value and psychological stress value corresponding to the new mismatch risk point are integrated to generate a long-term safety risk assessment result; The modification suggestion module is used to output a concrete personalized assessment report of user-furniture interaction and quantitative modification suggestions for furniture structure based on the current and long-term safety risk assessment results.

Images