Embodied cognition-based empathy ability evaluation and intervention system and method

Abstract

The application discloses a system and method for evaluating and intervening in empathy ability based on embodied cognition, and the system comprises a sand table body, a plurality of task models, a multi-modal perception module, an environment simulation module, a wearable force feedback device and a central processing module; the task model is provided with an RFID tag, and reciprocity behavior data is generated by identification of the RFID reader; the multi-modal perception module collects emotional signals of a user; the environment simulation module constructs an emotional interaction scene; the wearable force feedback device collects pressure signals when the user moves the task model, and applies programmable physical resistance to the user's hands; and the central processing module calculates an embodied empathy coupling index, and dynamically adjusts the physical resistance of the wearable force feedback device according to the embodied empathy coupling index. The application mathematically maps physical parameters and psychological parameters for the first time, and realizes objective and quantitative evaluation of empathy ability through an EECI index.

Description

Technical Field

[0001] This invention relates to the field of interdisciplinary technology of educational psychology and artificial intelligence, specifically to a method for assessing embodied cognition and empathy using an intelligent sand table system. Background Technology

[0002] Traditional cognitive views consider the brain like a standalone computer, with thinking occurring solely within the brain. Embodied cognition, on the other hand, posits that cognition, thought, emotion, and understanding are not solely shaped by the brain, but rather by the body, senses, and environment. Physical actions and sensory experiences directly determine how one thinks and feels. Embodied cognition is the underlying physiological mechanism of empathy, giving rise to emotions and attitudes of gratitude.

[0003] Current methods of gratitude education for teenagers often employ one-way lectures, questionnaires, or role-playing. However, traditional questionnaires are susceptible to subjective masking effects and lack genuine physical experience, making their effectiveness highly dependent on subjective judgment and failing to form a closed loop of "perception-feedback-evaluation." Existing educational hardware largely remains at the visual and auditory level of interaction, failing to transform abstract "psychological pressure" and "empathy" into concrete physical perceptions. Current gratitude education experiences, such as household chores, are often mechanical repetitive actions, making it difficult to trigger deep psychological empathy and lacking intervention mechanisms that dynamically adjust based on the user's real-time state. Therefore, there is an urgent need for a system that can deeply couple psychological theories (embodied cognition) with hardware control to address the problems of traditional educational methods lacking immersion, objective quantitative indicators, and dynamic intervention capabilities.

[0004] Sandplay therapy, also known as sand tray therapy, is a popular international psychotherapeutic method. It is widely used in schools and kindergartens for children's psychological education and therapy; it is also highly popular in universities and adult psychological clinics. By evoking a childlike spirit, it provides a pathway to inner peace, thereby resolving issues such as psychosomatic disorders, maladaptive social behavior, and personality development disorders, ultimately achieving the goal of gratitude education.

[0005] Therefore, how to obtain the empathy coupling index in sandplay games, thereby solving the technical problems of traditional gratitude education lacking immersion and objective quantitative indicators; at the same time, how to improve the guidance of gratitude education while playing games are technical problems that need to be solved. Summary of the Invention

[0006] 1. The technical problem to be solved:

[0007] To address the aforementioned technical problems, this invention provides a method for assessing embodied cognition and empathy using an intelligent sand table system.

[0008] 2. Technical Solution:

[0009] An empathy assessment and intervention system based on embodied cognition is characterized by comprising a sand table ontology, multiple task models, a multimodal perception module, an environmental simulation module, a wearable force feedback device, and a central processing module.

[0010] The task model is placed inside the sand table body and is equipped with an RFID tag. An RFID reader is installed at a preset position inside the sand table body. When the task model is moved to the preset position, the RFID reader identifies it and generates a task completion signal, which serves as the user's reciprocal behavior data.

[0011] The multimodal perception module is used to collect images and sounds of the user while completing a task and generate the user's emotional signals;

[0012] The environment simulation module uses language, music, and lighting to construct emotional interaction scenarios for users during the task completion process;

[0013] The wearable force feedback device is fitted onto the user's hand to collect pressure signals when the user moves the task model and can apply programmable physical resistance to the user's hand.

[0014] The central processing module is pre-set with an embodied empathy coupling index algorithm, which is used to receive the reciprocal behavior data, emotional signals and stress signals, calculate the embodied empathy coupling index, and dynamically adjust the physical resistance of the wearable force feedback device according to the embodied empathy coupling index.

[0015] The multimodal sensing module, RFID reader, and wearable force feedback device are all communicatively connected to the central processing module to transmit the collected data to the central processing module.

[0016] Furthermore, the multimodal perception module includes a camera, a microphone array, an image processing unit, and a speech signal extraction unit; the camera collects the user's facial information and transmits the collected information to the image processing unit; the image processing unit presets a facial key point extraction algorithm, extracting the intensity values ​​of multiple preset facial action units as facial features; the microphone array collects the speech signals emitted by the user when performing a task; the speech signal extraction unit extracts the variance between the Mel-frequency cepstral coefficients (MFCC) and the preset fundamental frequency (F0) in the speech signal as speech features;

[0017] The wearable force feedback device has a built-in torque sensor that collects the real-time pressure value received by the device and outputs it to the central processing module.

[0018] Furthermore, the wearable force feedback device is a lightweight force feedback glove based on a knuckle damper, comprising a wrist support unit, a force feedback joint module, and a transmission unit. The wrist support unit integrates a main control circuit board, a battery, and a wireless communication module. The force feedback joint module is installed at the base of the finger joint and includes a micro motor, a reduction gear set, and a torque sensor. The transmission unit includes a finger ring and a flexible steel wire embedded in the glove lining. One end of the flexible steel wire is wound around the output shaft of the micro motor, and the other end is fixed to the finger ring. In use, the finger ring is fitted onto the end of the corresponding finger. When the system needs to apply resistance, the main control circuit board controls the micro motor to generate a torque opposite to the direction of the finger's grip, which pulls the finger ring through the flexible steel wire, generating a damping sensation when the user bends their finger. The wireless communication module in the wrist support unit is communicatively connected to the central processing module to receive resistance control commands and transmit the pressure signal collected by the torque sensor.

[0019] An empathy assessment and intervention method based on embodied cognition includes the following steps:

[0020] Step 1: Obtain user information and set the basic resistance and safety threshold of the wearable force feedback device based on the user information; the user information includes the user's age, weight, and historical embodied empathy coupling index data;

[0021] Step 2: Select an emotional interaction scenario that includes task information, and construct the corresponding emotional interaction scenario through the environment simulation module;

[0022] Step 3: The user moves the task model according to the task information. During the movement, the user's emotional signal data, reciprocity behavior data, and stress signals are collected in real time; the central processing module calculates the embodied empathy coupling index in real time based on the collected data.

[0023] Step 4: Dynamically adjust the task difficulty based on the real-time empathy coupling index; when the empathy coupling index is lower than a preset threshold, increase the physical resistance of the wearable force feedback device to enhance the user's sense of physical load; when the user completes a preset reciprocal behavior or the empathy coupling index reaches a preset target threshold, reduce the physical resistance and / or emit encouraging voice or music as positive feedback; when the pressure signal applied by the user is continuously lower than a preset moving threshold and the task model is not moved, determine that the user's current strength is insufficient, and dynamically reduce the task difficulty of the current task, including replacing it with a lighter task model or shortening the carrying distance.

[0024] Furthermore, the Embodied Empathy Coupling Index (EECI) is calculated using the following formula;

[0025] In the above formula, α, β, and γ are the preset corresponding weight coefficients, and satisfy 0 < α < 1, 0 < β < 1, 0 < γ < 1, and α + β + γ = 1; the values ​​of α, β, and γ are dynamically adjusted according to the type of emotional interaction scenario. The physical load ratio is defined as follows: P represents the real-time pressure value of the torque sensor built into the wearable force feedback device; Pmax is the maximum torque threshold set according to the user's weight, age, and safety regulations; ΔE / Δt is the rate of change of emotion within the time interval Δt, where Δt is the time interval between two consecutive acquisitions of emotional signals by the multimodal perception module or the length of a preset fixed time window; ΔE represents the emotional arousal level; the emotional arousal level ΔE is obtained by jointly calculating the camera and microphone array of the multimodal perception module. The reciprocal behavior conversion rate is represented by Rs, where Rs is the number of times the task is completed as identified by the sandbox body through RFID, and Rt is the number of times the target assistance is set for the current task. When Rs≥Rt, the reciprocal behavior conversion rate is counted as 1.

[0026] The emotional arousal value ΔE is output through a pre-trained lightweight neural network. The input layer of this neural network receives the facial features and voice features, and the output layer outputs an emotional arousal value between 0 and 1.

[0027] Furthermore, the lightweight neural network adopts an architecture combining a convolutional neural network and a long short-term memory network; the convolutional neural network is used to perform sliding window convolution calculation on the input facial action unit intensity sequence to extract local spatial correlation features of facial muscle groups; the long short-term memory network is used to receive the local spatial correlation features and perform temporal analysis in combination with speech features to calculate the changing trend of emotions within a continuous time window, thereby outputting the current emotional arousal level ΔE.

[0028] Furthermore, it also includes a multi-layered user security protection strategy, including:

[0029] Hardware-level physical protection: The force feedback glove has a mechanical limiting structure at the finger joints to limit the maximum bending angle of the fingers; an elastic buffer component is connected in series in the transmission unit to absorb impact energy;

[0030] Software-level intelligent control: When adjusting the physical resistance P, the system ensures that the rate of change of resistance is lower than the preset rate of change, thus ensuring a smooth increase or decrease in resistance. Simultaneously, it features a multimodal emergency stop function based on voice keywords and abnormal movement detection. When a preset help-seeking keyword is recognized or a violent struggle by the user's hands is detected, the resistance is reduced to the lowest safe value within 100 milliseconds. User-level guidance: The system loads different safety parameter baselines based on the user's age and suggests the duration of each use.

[0031] Furthermore, the method for obtaining the maximum torque threshold Pmax includes: Default baseline loading: When the system starts, based on the user's entered age and weight information, a preset group experience default value is loaded as the initial maximum torque threshold Pmax; Dynamic optimization: In the early stages of the task, the system gradually increases the resistance in steps, captures the user's actual action response peak in the current state in real time, and smoothly updates it to the current maximum torque threshold Pmax to ensure that the physical load ratio is always based on the user's current actual ability.

[0032] 3. Beneficial effects:

[0033] (1) The present invention provides a method for assessing embodied cognition and empathy using an intelligent sand table system. Based on sand table games, an empathy assessment and intervention system based on embodied cognition is constructed. The system collects facial features and voice features of users to generate emotional data, collects reciprocal behavior data and pressure information collected by wearable force feedback devices, analyzes the embodied empathy coupling index of the tested subjects, and dynamically adjusts the task difficulty and physical resistance according to the real-time calculated embodied empathy coupling index to form a closed loop of "assessment-intervention-reassessment", thereby improving the pertinence and effectiveness of educational intervention.

[0034] (2) The present invention provides a method for assessing embodied cognition and empathy using an intelligent sand table system. For the first time, it mathematically maps "physical parameters (resistance)" and "psychological parameters (empathy level)" and achieves an objective and quantitative assessment of empathy ability through the embodied empathy coupling index.

[0035] (3) The present invention provides a method for assessing embodied cognition and empathy using an intelligent sand table system. By using a wearable force feedback device, abstract psychological stress is transformed into concrete physical load, and combined with the sand table scene, a deep embodied cognitive experience is created.

[0036] In summary, this method achieves mutual constraints and collaboration among hardware modules through the embodied empathy coupling index algorithm, effectively avoiding simple hardware assembly. Simultaneously, the system supports multi-dimensional scene switching, providing an objective and immersive intervention tool for adolescent mental health. Attached Figure Description

[0037] Figure 1 This is a block diagram illustrating the hardware architecture and data flow of the present invention as implemented in this embodiment;

[0038] Figure 2 This is a flowchart illustrating the calculation of the embodied empathy coupling index in this invention;

[0039] Figure 3 This is a schematic diagram of the empathy assessment and intervention system involved in Specific Implementation Example 3.

[0040] Explanation of reference numerals in the attached diagram: 1. Wearable force feedback device; 2. Sand table body; 3. Environment simulation module; 4. Camera and microphone array; 5. User A; 6. User B; 7. Task model. Detailed Implementation

[0041] The present invention will now be described in detail with reference to the accompanying drawings.

[0042] An empathy assessment and intervention system based on embodied cognition is characterized by comprising a sand table ontology, multiple task models, a multimodal perception module, an environmental simulation module, a wearable force feedback device, and a central processing module.

[0043] The task model is placed inside the sand table body and is equipped with an RFID tag. An RFID reader is installed at a preset position inside the sand table body. When the task model is moved to the preset position, the RFID reader identifies it and generates a task completion signal, which serves as the user's reciprocal behavior data.

[0044] In one embodiment of the invention, the task model employs a physical model structure internally encapsulated with a passive ultra-high frequency (UHF) RFID tag. This tag includes an RFID chip and antenna, and stores a unique model identification code (ID). The outer shell material of the task model is selected from non-metallic materials (such as ABS plastic or wood) that do not affect the transmission of RFID signals, ensuring the normal radiation performance of the tag antenna. An RFID reader antenna is embedded below or to the side of each preset target location on the sand table body. This reader antenna is communicatively connected to the RFID read / write module and the central processing module. To reduce the false reading rate, the effective reading distance of the reader antenna is set to 3cm~8cm, ensuring that only task model tags precisely placed directly above the target location can be activated for reading, while other models in the surrounding area or models passed by by the hand will not be misread. When the user moves the task model to the preset target location, the RFID tag enters the effective identification area of ​​the reader antenna. The chip inside the tag is activated after receiving working power through the antenna, and then sends out a response signal carrying its own model ID. After receiving the response signal, the reader antenna demodulates and decodes the signal by the RFID read / write module, and uploads the decoded model ID and the current reading timestamp to the central processing module through the communication interface. After receiving the above information, the central processing module determines that the model has been successfully delivered to the target location as the event "a tag with model ID Y was read at target location X at a certain time", and records the event as a valid assistance action (the Rs count is automatically incremented by 1).

[0045] When multiple task models are placed simultaneously or sequentially at different target locations, to prevent signal collisions caused by multiple tags responding simultaneously, the RFID read / write module incorporates a time-slot anti-collision algorithm. This algorithm executes as follows: When initiating a round of inventory checks, the read / write module broadcasts a control command containing the total number of time slots N to all tags within the identification area; upon receiving the command, each tag randomly generates an integer between 1 and N as its own response delay time slot; each tag only sends a response signal when its assigned time slot arrives. The read / write module sequentially receives response signals from different tags within each time slot. If a tag ID is successfully decoded within a time slot, the tag is marked as read; if a collision is detected (i.e., multiple tags responding simultaneously within the same time slot causing decoding failure), the tag is reread in the next round of inventory checks. Through this mechanism, the central processing module can sequentially and accurately identify the identity of each task model placed at each target location.

[0046] The multimodal perception module is used to collect images and sounds of the user while completing a task and generate the user's emotional signals.

[0047] The environment simulation module uses language, music, and lighting to construct corresponding emotional interaction scenarios during the task completion process.

[0048] The wearable force feedback device is fitted onto the user's hand to collect pressure signals when the user moves the task model and apply programmable physical resistance to the user's hand.

[0049] The central processing module is pre-programmed with an embodied empathy coupling index algorithm to receive the reciprocal behavior data, emotional signals, and stress signals, calculate the embodied empathy coupling index, and dynamically adjust the physical resistance of the wearable force feedback device based on the embodied empathy coupling index.

[0050] The multimodal sensing module, RFID reader, and wearable force feedback device are all communicatively connected to the central processing module to transmit the collected data to the central processing module.

[0051] Furthermore, the multimodal perception module includes a camera, a microphone array, an image processing unit, and a speech signal extraction unit; the camera collects the user's facial information and transmits the collected information to the image processing unit; the image processing unit presets a facial key point extraction algorithm, extracting the intensity values ​​of multiple preset facial action units as facial features; the microphone array collects the speech signals emitted by the user when performing a task; the speech signal extraction unit extracts the variance between the Mel-frequency cepstral coefficients (MFCC) and the preset fundamental frequency (F0) in the speech signal as speech features;

[0052] The wearable force feedback device has a built-in torque sensor for outputting real-time pressure values ​​to the central processing module.

[0053] Furthermore, the wearable force feedback device is a lightweight force feedback glove based on a knuckle damper, comprising a wrist support unit, a force feedback joint module, and a transmission unit. The wrist support unit integrates a main control circuit board, a battery, and a wireless communication module. The force feedback joint module is installed at the base of the finger joint and includes a micro motor, a reduction gear set, and a torque sensor. The transmission unit includes a finger ring and a flexible steel wire embedded in the glove lining. One end of the flexible steel wire is wound around the motor output shaft, and the other end is fixed to the finger ring. In use, the finger ring is fitted onto the end of the corresponding finger. When the system needs to apply resistance, the main control circuit board controls the micro motor to generate a torque opposite to the direction of the finger's grip, which pulls the finger ring through the flexible steel wire, generating a damping sensation when the user bends their finger. The wireless communication module in the wrist support unit is communicatively connected to the central processing module to receive resistance control commands and transmit the pressure signal collected by the torque sensor.

[0054] In one embodiment of the present invention, the wearable force feedback device collects pressure signals when a user moves a task model. The specific acquisition logic is as follows: The torque sensor, with a range of 0~10 N·m, is installed inside the force feedback joint module and is used to sense the reverse torque value on the output shaft of the micro-motor in real time. When the user bends their finger to grasp the task model, the finger bending action pulls a flexible steel wire through a finger ring. The steel wire transmits the tension to the motor output shaft, and the torque sensor thereby senses the resistance torque proportional to the user's gripping force. This resistance torque value is the real-time pressure value P.

[0055] To exclude invalid pressure signals when a user bends their fingers without grasping the model, the central processing module has built-in action validity determination logic: the pressure value reported by the torque sensor is only included in the EECI calculation within the time period after the RFID reader identifies that the task model has been moved from its initial position; if the user does not touch any task model, the pressure signal during this period will be marked as invalid data and filtered out.

[0056] The force feedback gloves can be equipped with a five-finger fully driven structure or a three-finger driven structure (thumb-index-middle finger only) to adapt to the gripping methods of different task models; the force feedback joint modules corresponding to each finger are independently controlled, and the central processing module can adjust the physical resistance value of each finger individually.

[0057] An empathy assessment and intervention method based on embodied cognition includes the following steps:

[0058] Step 1: Obtain user information and set the basic resistance and safety threshold of the wearable force feedback device based on the user information; the user information includes the user's age, weight, and historical embodied empathy coupling index data;

[0059] Step 2: Select an emotional interaction scenario that includes task information, and construct the corresponding emotional interaction scenario through the environment simulation module;

[0060] Step 3: The user moves the task model according to the task information. During the movement, the user's emotional signal data, reciprocity behavior data, and stress signals are collected in real time; the central processing module calculates the embodied empathy coupling index in real time based on the collected data.

[0061] Step 4: Dynamically adjust the task difficulty based on the real-time empathy coupling index; when the empathy coupling index is lower than a preset threshold, increase the physical resistance of the wearable force feedback device to enhance the user's sense of physical load; when the user completes a preset reciprocal behavior or the empathy coupling index reaches a preset target threshold, reduce the physical resistance [yz2.1] and / or emit encouraging voice music as positive feedback; when the pressure signal applied by the user is continuously lower than a preset moving threshold and the task model is not moved, determine that the user's current strength is insufficient, and dynamically reduce the task difficulty of the current task, including replacing it with a lighter task model or shortening the carrying distance.

[0062] Furthermore, the Embodied Empathy Coupling Index (EECI) is calculated using the following formula; ;

[0063] In the above formula, α, β, and γ are the preset corresponding weight coefficients, and satisfy 0 < α < 1, 0 < β < 1, 0 < γ < 1, and α + β + γ = 1; the values ​​of α, β, and γ are dynamically adjusted according to the type of emotional interaction scenario. The physical load ratio is given by P, where P represents the real-time pressure value from the torque sensor built into the wearable force feedback device. max The maximum torque threshold is set based on the user's weight, age, and safety regulations; ΔE / Δt is the rate of change of emotion within the time interval Δt, where Δt is the time interval between two consecutive acquisitions of emotion signals by the multimodal perception module or the length of a preset fixed time window; ΔE represents the emotional arousal level; the emotional arousal level ΔE is obtained by jointly calculating the camera and microphone array of the multimodal perception module. R represents the conversion rate of reciprocal behavior. s R represents the number of times the task was completed as identified by RFID on the sand table itself. t The target number of times to help the current task is set, when R s ≥R t At that time, the conversion rate of this reciprocal behavior is counted as 1;

[0064] The emotional arousal value ΔE is output through a pre-trained lightweight neural network. The input layer of this neural network receives the facial features and voice features, and the output layer outputs an emotional arousal value between 0 and 1.

[0065] Furthermore, the lightweight neural network adopts an architecture combining a convolutional neural network and a long short-term memory network; the convolutional neural network is used to perform sliding window convolution calculation on the input facial action unit intensity sequence to extract local spatial correlation features of facial muscle groups; the long short-term memory network is used to receive the local spatial correlation features and perform temporal analysis in combination with speech features to calculate the changing trend of emotions within a continuous time window, thereby outputting the current emotional arousal level ΔE.

[0066] In one embodiment of the present invention, the emotional arousal value ΔE is obtained through joint calculation by a pre-trained lightweight neural network. The specific multimodal fusion logic is as follows: the input layer of the neural network adopts a dual-branch parallel structure: the first branch receives the facial action unit intensity sequence, with an input dimension of 12×T (T is the time window length, ranging from 10 to 30 frames); the second branch receives the speech feature vector sequence, with an input dimension of 13×T (13 dimensions include 12-dimensional MFCC and 1-dimensional fundamental frequency F0 variance). After feature extraction by independent one-dimensional convolutional layers, the two branches are concatenated in the network's fusion layer to form a joint feature matrix with a dimension of 25×T. The fused feature matrix is ​​input to a long short-term memory network layer for temporal modeling, and finally outputs an emotional arousal value between 0 and 1 through a fully connected layer.

[0067] To simplify computation and improve real-time performance, when the signal-to-noise ratio of the speech signal is below a preset threshold (e.g., 20dB) or the user does not speak, the neural network automatically switches to a single-modal mode, relying solely on facial features to output emotional arousal. Conversely, when the camera is obstructed or facial feature extraction fails, it relies solely on speech features for output. This single-modal degradation strategy ensures that the system can stably output the ΔE value under any environmental conditions.

[0068] Furthermore, it also includes a multi-layered user security protection strategy, including:

[0069] Hardware-level physical protection: The force feedback glove has a mechanical limiting structure at the finger joints to limit the maximum bending angle of the fingers; an elastic buffer component is connected in series in the transmission unit to absorb impact energy;

[0070] Software-level intelligent control: When adjusting the physical resistance P, the system uses a resistance change rate limiting algorithm to ensure smooth increases and decreases in resistance; at the same time, it has a multimodal emergency stop function based on voice keywords and abnormal action detection. When a preset help keyword is recognized or a violent struggle of the user's hand is detected, the resistance is reduced to the lowest safe value within 100 milliseconds; User-level standardized guidance: The system loads different safety parameter baselines according to the user's age and suggests the duration of each use.

[0071] Furthermore, the method for obtaining the maximum torque threshold Pmax includes: Default baseline loading: When the system starts, based on the user's entered age and weight information, a preset group experience default value is loaded as the initial maximum torque threshold Pmax; Dynamic optimization: In the early stages of the task, the system gradually increases the resistance in steps, captures the user's actual action response peak in the current state in real time, and smoothly updates it to the current maximum torque threshold Pmax to ensure that the physical load ratio is always based on the user's current actual ability.

[0072] Example 1:

[0073] As attached Figure 1 As shown, this embodiment illustrates the hardware architecture and data flow diagram of this solution; the central processing module is the core of this solution. Data acquisition path (solid arrow): The multimodal perception module and the sandbox are responsible for real-time acquisition of the user's micro-expressions, voice tone, physiological data, and task completion signals, and transmit them to the central processing module. Data processing: After receiving the data, the central processing module calculates the empathy coupling index through the empathy coupling index calculation unit, and the threshold determination unit judges the user's current state, dynamically adjusting the task difficulty and physical resistance according to the user's current state.

[0074] Control execution path (dashed arrow): Based on the calculation results, the central processing module sends instructions to the wearable force feedback device to adjust the torque, and at the same time sends instructions to the intelligent sand table interaction module to control the environment simulation module. In this embodiment, the environment simulation module is in the form of an energy tower, which provides sound and light feedback to form a closed-loop control system.

[0075] Example 2:

[0076] As attached Figure 2The diagram shows the specific flowchart for calculating the Embodied Empathy Coupling Index (EECI) using this method. Steps S1 and S2: After system startup, resistance parameters are initialized, and user emotional arousal and reciprocity behavior data are collected in real time. Steps S3 and S4: The central processing module calculates the current index using the Embodied Empathy Coupling Index algorithm. If the EECI is lower than a preset threshold, the system executes step S5, automatically increasing the physical resistance P to forcibly awaken the user's embodied cognition by increasing physical load. Steps S6 and S7: If the EECI is higher than the threshold, it further determines whether the user's reciprocity behavior meets the standard. If it does, the system triggers the sound and light rewards in the sandbox and reduces resistance, completing one positive intervention cycle; if it does not meet the standard, the monitoring state continues.

[0077] Example 3:

[0078] This implementation serves as an example to illustrate the various emotional interaction scenarios to which this method is applicable and their specific implementation methods.

[0079] 1. Single-player mode; In this mode, the system generates a virtual guide character (such as "Thanksgiving Elf") through projection or built-in speakers. The character gives prompts to the user according to a preset script (such as "Now you need to move this heavy object. Imagine your mother is carrying your luggage in the wind and snow").

[0080] The virtual guide dynamically adjusts the intensity and frequency of its voice encouragement based on real-time changes in the EECI index. When a user completes a reciprocal action (such as placing a "thank you" module on the sandbox), the system automatically triggers an audio-visual reward, with the same effect as in the two-person mode. The single-person mode is mainly used in family scenarios or when pairing for training is not possible. Its core algorithm is exactly the same as the two-person mode, except that the source of the helping action in Rs / Rt is replaced from "another real user" to "a symbolic mutual assistance task actively completed by the user".

[0081] 2. Two-player mode:

[0082] As attached Figure 3 As shown, this diagram illustrates a two-person collaboration scenario. User A (labeled 5) wears a wearable force feedback device and is responsible for moving the task model 7 within the sandbox body 2. In this embodiment, the wearable force feedback device 1 uses a resistance glove, which outputs variable damping force based on the system algorithm to simulate the feeling of weight. User B (labeled 6) is responsible for observing User A's state and placing auxiliary modules or interacting with the sandbox. The camera / microphone array 4 is located directly in front of or above the user to capture User A's facial expressions and vocal emotions. The energy tower of the environmental simulation module 3 serves as a feedback device; when the system detects that both parties have completed effective cooperative behavior (EECI index met), the energy tower emits specific light effects or sounds to provide positive encouragement to the users.

[0083] (1) Asymmetrical load-bearing transport scenario for two people;

[0084] Scenario Setting: Simulate the embodied cognitive experience of "parents carrying heavy objects for their children in extreme weather". User A (playing the person carrying the load) wears resistance gloves, and User B (playing the facilitator) stands in front of the sand table.

[0085] Hardware linkage and parameter logic: System setting initial safety resistance If the multimodal perception module detects that user A has a calm expression while carrying the item ( The EECI index is low. The central processing module automatically increases its level. Weighting will increase resistance to User A physically experienced a real heaviness and exertion, which triggered changes in facial micro-expressions and rapid breathing. Significant increase. User B, observing User A's strain, proactively placed a "weight-sharing" module on the sand table. The system's RFID detected this action. The target was met, the EECI index exceeded the threshold, the energy tower on the sand table lit up with fragments of "understanding", and the resistance gloves automatically released the force. This completes the intervention loop.

[0086] If user A stops moving for more than 5 seconds due to excessive resistance, the system determines that the user has "given up". At this time, the resistance will not be increased. Instead, an encouraging voice will be played: "Do you need to take a break? Imagine that your parents are also very tired..." After the user starts moving again, the resistance will be slowly reduced from the current value to the initial value, and the closed loop will be adjusted again.

[0087] Throughout the process, the aforementioned security protection mechanism is active in real time, and if any abnormal physiological or behavioral changes of the user are detected at any time, the resistance will automatically drop to zero.

[0088] (2) Environmental resistance simulation scenario – “Guardianship in wind and rain”

[0089] Scenario Setting: Simulates the hardships parents face when picking up or dropping off their children or going to work in inclement weather, using environmental disturbances to evoke empathy. User A attempts to move the "umbrella" module to the "home" location in the sandbox.

[0090] Hardware linkage and parameter logic: The physics feedback module generates high-frequency random pulse vibrations (5Hz-10Hz, with random fluctuations in intensity) to simulate strong wind resistance. If user A fails the task due to hand tremors and becomes depressed, the system plays a "thunderstorm sound effect" through the sand table interaction module and further increases the viscous resistance of the gloves. When user B places the "windbreak" module on the sand table, the physics feedback module instantly eliminates the vibration and resistance, and the energy tower lights up with a warm yellow light, symbolizing a safe haven.

[0091] Throughout the process, the aforementioned security protection mechanism is active in real time, and if any abnormal physiological or behavioral changes of the user are detected at any time, the resistance will automatically drop to zero.

[0092] (3) Endurance and perseverance scenario – “The long wait”

[0093] Scenario setting: Simulates scenarios requiring parents to endure long periods of time, such as queuing for registration at a hospital or working under the scorching sun. User A needs to keep their arm suspended in the air or press the "start button" on the sand table for 3 minutes.

[0094] Hardware linkage and parameter logic: The physical force feedback module is set to a constant reverse torque (e.g.) ), over time According to linear equations The intensity increases slowly, simulating muscle fatigue. The system monitors the amplitude of user A's limb tremors and the frequency of sighs. When fatigue accumulates and causes the EECI index to drop, if user B clicks the "wipe sweat" or "pass water" button, the physical force feedback module provides 5 seconds of reverse torque support, allowing user A to feel a sense of ease and "being supported".

[0095] Throughout the process, the aforementioned security protection mechanism is active in real time, and if any abnormal physiological or behavioral changes of the user are detected at any time, the resistance will automatically drop to zero.

[0096] (4) Detailed operation scenario - "Fragile gift"

[0097] Scenario: Simulate parents caring for a young child or passing a precious item, experiencing the psychological pressure of "being extremely careful." User A needs to move the "egg" module from one end to the other.

[0098] Hardware linkage and parameter logic: The physical force feedback module activates a high-sensitivity damping mode, setting an extremely small acceleration threshold (e.g., 0.5 m / s²). If the movement speed is too fast, the glove immediately "locks in" (instantly increasing damping to its maximum value), and the sand table emits a "crackling" sound. If user A exhibits excessive tension leading to stiff operation, and user B places a "pad" module on the sand table, the system relaxes the acceleration threshold, and the glove damping becomes gentler.

[0099] Throughout the process, the aforementioned security protection mechanism is active in real time, and if any abnormal physiological or behavioral changes of the user are detected at any time, the resistance will automatically drop to zero.

[0100] Although the present invention has been disclosed above with reference to preferred embodiments, these are not intended to limit the invention. Any person skilled in the art can make various changes or modifications without departing from the spirit and scope of the invention. Therefore, the scope of protection of the present invention should be defined by the scope of the claims of this application.

[0101] Example 4: Another implementation of the task model

[0102] As an alternative implementation, the identification method of the task model is not limited to RFID passive tags; it can also be achieved by combining magnetic contact switches with RFID. Specifically, the bottom of the task model is provided with a first magnetic connector and a first contact electrode, and a second magnetic connector and a second contact electrode are correspondingly provided at the preset target position. The first and second magnetic connectors are attracted by each other, which is used to guide the task model to accurately place and assist in physical contact. When the task model is placed at the target position, the first and second contact electrodes make close contact with each other under magnetic assistance, forming an electrical conduction circuit, triggering an identification signal. At the same time, the RFID reader reads the model ID. The central processing module only determines that a valid assistance action has been completed when both the "physical contact conduction" and "RFID reading success" conditions are met simultaneously. This solution can further avoid false triggers caused by the tag being misread by nearby readers, improving the reliability of reciprocal behavior data.

Claims

1. A system for assessing and intervening in empathy based on embodied cognition, characterized in that: It includes a sand table, multiple task models, a multimodal perception module, an environment simulation module, a wearable force feedback device, and a central processing module; The task model is placed inside the sand table body and is equipped with an RFID tag. An RFID reader is installed at a preset position inside the sand table body. When the task model is moved to the preset position, the RFID reader identifies it and generates a task completion signal, which serves as the user's reciprocal behavior data. The multimodal perception module is used to collect images and sounds of the user while completing a task and generate the user's emotional signals; The environment simulation module uses language, music, and lighting to construct emotional interaction scenarios for users during the task completion process; The wearable force feedback device is fitted onto the user's hand to collect pressure signals when the user moves the task model and can apply programmable physical resistance to the user's hand. The central processing module is pre-set with an embodied empathy coupling index algorithm, which is used to receive the reciprocal behavior data, emotional signals and stress signals, calculate the embodied empathy coupling index, and dynamically adjust the physical resistance of the wearable force feedback device according to the embodied empathy coupling index. The multimodal sensing module, RFID reader, and wearable force feedback device are all communicatively connected to the central processing module to transmit the collected data to the central processing module.

2. The empathy assessment and intervention system based on embodied cognition according to claim 1, characterized in that: The multimodal perception module includes a camera, a microphone array, an image processing unit, and a speech signal extraction unit; the camera collects the user's facial information and transmits the collected information to the image processing unit; The image processing unit has a preset facial key point extraction algorithm, which extracts the intensity values ​​of multiple preset facial action units as facial features; the microphone array collects the voice signal emitted by the user when performing the task; the voice signal extraction unit extracts the variance between the Mel-frequency cepstral coefficients (MFCC) and the preset fundamental frequency (F0) in the voice signal as voice features. The wearable force feedback device has a built-in torque sensor that collects the real-time pressure value received by the device and outputs it to the central processing module.

3. The empathy assessment and intervention system based on embodied cognition according to claim 1, characterized in that: The wearable force feedback device is a lightweight force feedback glove based on a knuckle damper, comprising a wrist support unit, a force feedback joint module, and a transmission unit. The wrist support unit integrates a main control circuit board, a battery, and a wireless communication module. The force feedback joint module is installed at the base of the finger joint and includes a micro motor, a reduction gear set, and a torque sensor. The transmission unit includes a finger ring and a flexible steel wire embedded in the glove lining. One end of the flexible steel wire is wound around the output shaft of the micro motor, and the other end is fixed to the finger ring. In use, the finger ring is fitted onto the end of the corresponding finger. When the system needs to apply resistance, the main control circuit board controls the micro motor to generate a torque opposite to the direction of the finger's grip, which pulls the finger ring through the flexible steel wire, generating a damping sensation when the user bends their finger. The wireless communication module in the wrist support unit is connected to the central processing module to receive resistance control commands and transmit the pressure signal collected by the torque sensor.

4. A method for assessing and intervening in empathy based on embodied cognition, implemented using any one of the empathy assessment and intervention systems based on embodied cognition as described in claims 1-3, characterized in that: Includes the following steps: Step 1: Obtain user information and set the basic resistance and safety threshold of the wearable force feedback device based on the user information; the user information includes the user's age, weight, and historical embodied empathy coupling index data; Step 2: Select an emotional interaction scenario that includes task information, and construct the corresponding emotional interaction scenario through the environment simulation module; Step 3: The user moves the task model according to the task information. During the movement, the user's emotional signal data, reciprocity behavior data, and stress signals are collected in real time; the central processing module calculates the embodied empathy coupling index in real time based on the collected data. Step 4: Dynamically adjust the task difficulty based on the real-time empathy coupling index; when the empathy coupling index is lower than a preset threshold, increase the physical resistance of the wearable force feedback device to enhance the user's sense of physical load; when the user completes a preset reciprocal behavior or the empathy coupling index reaches a preset target threshold, reduce the physical resistance and / or emit encouraging voice or music as positive feedback; when the pressure signal applied by the user is continuously lower than a preset moving threshold and the task model is not moved, determine that the user's current strength is insufficient, and dynamically reduce the task difficulty of the current task, including replacing it with a lighter task model or shortening the carrying distance.

5. The method for assessing and intervening in empathy based on embodied cognition as described in claim 4, characterized in that: The Embodied Empathy Coupling Index (EECI) is calculated using the following formula; ； In the above formula, α, β, and γ are the preset corresponding weight coefficients, and satisfy 0 < α < 1, 0 < β < 1, 0 < γ < 1, and α + β + γ = 1; the values ​​of α, β, and γ are dynamically adjusted according to the type of emotional interaction scenario. The physical load ratio is given by P, where P represents the real-time pressure value from the torque sensor built into the wearable force feedback device. max The maximum torque threshold is set based on the user's weight, age, and safety regulations; ΔE / Δt is the rate of change of emotion within the time interval Δt, where ΔE is the time interval between two consecutive acquisitions of emotion signals by the multimodal perception module or the length of a preset fixed time window; ΔE represents the emotional arousal level; the emotional arousal level ΔE is obtained by jointly calculating the camera and microphone array of the multimodal perception module. R represents the conversion rate of reciprocal behavior. s R represents the number of times the task was completed as identified by RFID on the sand table itself. t The target number of times to help the current task is set, when R s ≥R t At that time, the conversion rate of this reciprocal behavior is counted as 1; The emotional arousal value ΔE is output through a pre-trained lightweight neural network. The input layer of this neural network receives the facial features and voice features, and the output layer outputs an emotional arousal value between 0 and 1.

6. The method for assessing and intervening in empathy based on embodied cognition as described in claim 5, characterized in that: The lightweight neural network adopts an architecture combining a convolutional neural network and a long short-term memory network. The convolutional neural network is used to perform sliding window convolution calculations on the input facial action unit intensity sequence to extract local spatial association features of facial muscle groups. The long short-term memory network is used to receive the local spatial association features and perform temporal analysis in combination with speech features to calculate the changing trend of emotions within a continuous time window, thereby outputting the current emotional arousal level ΔE.

7. The method for assessing and intervening in empathy based on embodied cognition as described in claim 4, characterized in that: It also includes a multi-layered user security protection strategy, including: Hardware-level physical protection: The force feedback glove has a mechanical limiting structure at the finger joints to limit the maximum bending angle of the fingers; an elastic buffer component is connected in series in the transmission unit to absorb impact energy; Software-level intelligent control: When adjusting the physical resistance P, the system ensures that the rate of change of resistance is lower than the preset rate of change, thus ensuring a smooth increase or decrease in resistance. Simultaneously, it features a multimodal emergency stop function based on voice keywords and abnormal movement detection. When a preset help-seeking keyword is recognized or a violent struggle by the user's hands is detected, the resistance is reduced to the lowest safe value within 100 milliseconds. User-level guidance: The system loads different safety parameter baselines based on the user's age and suggests the duration of each use.

8. The method for assessing and intervening in empathy based on embodied cognition according to claim 5, characterized in that: The method for obtaining the maximum torque threshold Pmax includes: Default baseline loading: When the system starts, based on the age and weight information entered by the user, a preset group experience default value is loaded as the initial maximum torque threshold Pmax; Dynamic optimization: In the early stage of the task, the system gradually increases the resistance in steps, captures the peak value of the user's actual action response in the current state in real time, and smoothly updates it to the current maximum torque threshold Pmax to ensure that the physical load ratio is always based on the user's current actual ability.

Images