Massage robot and its control method, control device and storage medium

Abstract

This disclosure provides a massage robot and its control method, control device, and storage medium. In one embodiment, the control method for the massage robot includes: acquiring multimodal signals from a user, including electroencephalogram (EEG) signals, electrodermal transfer signal (EDS) signals, skin temperature signals, body impedance signals, and pressure array signals; fusing the multimodal signals to obtain the user's fatigue index, comfort index, and risk index; determining the user's massage path, target area, and massage parameters based on the fatigue index, comfort index, and risk index; and generating massage operation instructions to control the robotic arm of the massage robot to move its actuator to the target area along the massage path and to control the actuator to perform massage operations according to the massage parameters. This embodiment significantly improves the comfort, accuracy, and safety of the massage robot.

Description

Technical Field

[0001] This disclosure relates to the field of robotics, and more particularly to a massage robot and its control method, control device and storage medium. Background Technology

[0002] Currently, smart massage devices mainly provide basic massage for areas such as the shoulders, neck, and back through preset programs and parameters. Although the physical implementation of smart massage devices is relatively mature, most rely on simple feedback from single sensors such as pressure sensors, position sensors, or heart rate sensors for adjustment, resulting in poor adaptability. Improvements are still needed in terms of safety, personalization, and comfort. Summary of the Invention

[0003] In view of this, the present disclosure provides a massage robot and its control method, control device and storage medium.

[0004] According to a first aspect of this disclosure, a method for controlling a massage robot is provided, the method comprising:

[0005] Acquire the user's multimodal signals, including electroencephalogram (EEG) signals, electrodermal signal, skin temperature, body impedance signal, and pressure array signal;

[0006] The multimodal signals are fused to obtain the user's fatigue index, comfort index, and risk index. The user's massage path, target area, and massage parameters are determined based on the fatigue index, comfort index, and risk index.

[0007] Generate massage operation instructions to control the robotic arm of the massage robot to move the execution end of the massage robot to the target area according to the massage path and control the execution end to perform massage operations according to the massage parameters.

[0008] According to some embodiments of the first aspect of this disclosure, the method further includes:

[0009] Adaptive threshold security detection is performed on the electroencephalogram (EEG) signal, electrodermal signal, skin temperature, body impedance signal, and pressure array signal in the multimodal signal respectively;

[0010] If the number of signals in the multimodal signals that pass the adaptive threshold security detection exceeds the first preset threshold, then the current processing flow or massage operation of the massage robot continues, and the massage intensity of the current massage operation is reduced.

[0011] If the number of signals in the multimodal signals that pass the adaptive threshold security detection does not exceed the first preset threshold, an alarm is issued to terminate the current processing flow or massage operation of the massage robot.

[0012] In some embodiments of the first aspect of this disclosure, the adaptive threshold security detection of the electroencephalogram (EEG) signal, electrodermal signal, skin temperature, body impedance signal, and pressure array signal in the multimodal signal includes:

[0013] The β / α power ratio of the EEG signal is calculated in real time. When the β / α power ratio is less than or equal to the β / α power ratio threshold, the adaptive security detection of the EEG signal is determined to be passed. The β / α ratio threshold is an adaptive threshold.

[0014] The skin conductivity slope is calculated based on the skin conductivity signal. When the skin conductivity slope is less than or equal to the skin conductivity change rate threshold, the adaptive safety detection of the skin conductivity signal is determined to be passed. The skin conductivity change rate threshold is the adaptive threshold.

[0015] The skin temperature in the massage contact area is detected, and the adaptive safety detection of the skin temperature signal is determined to be passed when the skin temperature in the massage contact area is less than or equal to a skin temperature threshold, wherein the skin temperature threshold is an adaptive threshold.

[0016] The body impedance change amplitude is calculated using the user's body impedance signal. When the body impedance change amplitude is less than or equal to a body impedance change amplitude threshold, the adaptive security detection of the body impedance signal is determined to pass. The body impedance change amplitude threshold is the adaptive threshold.

[0017] The pressure value at each point in the pressure array signal is detected to exceed a pressure threshold. If the number of points in the pressure array signal with a pressure value greater than the pressure threshold does not exceed a second preset threshold, the adaptive safety detection of the pressure array signal is determined to be passed. The pressure threshold is the adaptive threshold.

[0018] In some embodiments of the first aspect of this disclosure, fusing the multimodal signals to obtain the user's fatigue index, comfort index, and risk index includes: extracting features from each signal in the multimodal signals to obtain the features of each signal, and fusing the features of each signal to obtain the user's fatigue index, comfort index, and risk index.

[0019] In some embodiments of the first aspect of this disclosure, the fusion processing of the features of each signal to obtain the user's fatigue index, comfort index, and risk index includes:

[0020] The weights of each signal in the multimodal signal are determined through sensor quality assessment;

[0021] The user's current state vector and state estimation covariance matrix are determined based on the weights of each signal in the multimodal signal. The state vector includes relative α power, β / α ratio, reciprocal of skin conductivity slope, instantaneous body temperature, pressure distribution entropy, and mean body electrical impedance.

[0022] The comfort index, the fatigue index, and the risk index are calculated based on the user's current state vector and the state estimation covariance matrix.

[0023] In some embodiments of the first aspect of this disclosure, the comfort index is the difference between the result of the fundamental function operation of the state vector and a first penalty term, wherein the first penalty term is the product of the preset comfort adjustment coefficient and a first trace operation result, and the first trace operation result is the trace operation result of the product of the state estimation covariance matrix and a preset comfort weight; and / or, the fatigue index is the difference between the result of the fundamental function operation of the state vector and a second penalty term, wherein the second penalty term is the product of the preset fatigue adjustment coefficient and a second trace operation result, and the second trace operation result is the trace operation result of the product of the state estimation covariance matrix and a preset fatigue weight; and / or, the risk index is the difference between the result of the fundamental function operation of the state vector and a third penalty term, wherein the third penalty term is the product of the preset risk adjustment coefficient and a third trace operation result, and the third trace operation result is the trace operation result of the product of the state estimation covariance matrix and a preset risk weight.

[0024] In some embodiments of the first aspect of this disclosure, the method further includes: acquiring one or a combination of electroencephalogram (EEG) signals, electrodermal signals, skin temperature, body impedance signals, pressure array signals, facial expressions, and user commands during the massage operation; and dynamically adjusting the massage parameters based on one or a combination of the EEG signals, electrodermal signals, skin temperature, body impedance signals, pressure array signals, facial expressions, and user commands during the massage operation.

[0025] According to a second aspect of this disclosure, a control device for a massage robot is provided, the control device comprising:

[0026] The signal acquisition unit is used to acquire the user's multimodal signals, including electroencephalogram (EEG) signals, electrodermal (ED) signals, skin temperature, body impedance signals, and pressure array signals.

[0027] The fusion unit is used to fuse the multimodal signals to obtain the user's fatigue index, comfort index, and risk index, and to determine the user's massage path, target area, and massage parameters based on the fatigue index, comfort index, and risk index.

[0028] A massage operation unit is used to generate massage operation instructions to control the robotic arm of the massage robot to move the execution end of the massage robot to the target area according to the massage path and to control the execution end to perform massage operations according to the massage parameters.

[0029] According to a third aspect of this disclosure, a massage robot is provided, the massage robot comprising: a pressure sensor array, an electroencephalogram (EEG) sensor, a skin conductance sensor, a body temperature sensor, a bioimpedance sensor, a control unit, a robotic arm, and an actuator, wherein the pressure sensor array, EEG sensor, skin conductance sensor, body temperature sensor, bioimpedance sensor, robotic arm, and actuator are respectively connected to the control unit, the control unit comprising one or more processors and a memory storing a program, the program comprising instructions, which, when executed by the processor, cause the processor to perform the method described above.

[0030] According to a fourth aspect of this disclosure, a computer-readable storage medium storing a program, the program including instructions that, when executed by one or more processors of a computing device, cause the computing device to perform the method described above.

[0031] This embodiment monitors various indicators such as user comfort, fatigue, and risk level by real-time acquisition and fusion of multimodal signals, and plans massage paths and determines massage parameters based on these multiple indicators. Compared with traditional massage methods based on only a single sensor, it can not only more comprehensively and accurately reflect the user's real state, but also achieve highly personalized adaptive adjustment, realizing comprehensiveness, personalization, and safety that existing single-sensor control does not possess.

[0032] The embodiments disclosed herein possess both global capabilities, i.e., overall state perception, and real-time and adaptive learning capabilities, which can significantly improve the comfort, accuracy, and safety of massage robots. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of this disclosure or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of this disclosure. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 A flowchart illustrating the massage robot control method provided in this embodiment of the present disclosure;

[0035] Figure 2 This is a schematic diagram illustrating the specific implementation process of the massage robot control method provided in the embodiments of this disclosure;

[0036] Figure 3 This is a schematic diagram of the structure of the massage robot control device provided in an embodiment of the present disclosure;

[0037] Figure 4 A schematic structural block diagram of a massage robot provided in an embodiment of this disclosure. Detailed Implementation

[0038] The technical solutions of the embodiments of this disclosure will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this disclosure, and not all embodiments. Based on the embodiments of this disclosure, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this disclosure.

[0039] The terminology used in the embodiments of this disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of this disclosure. The singular forms “a,” “the,” and “the” as used in the embodiments of this disclosure and the appended claims are also intended to include the plural forms unless the context clearly indicates otherwise.

[0040] Depending on the context, words such as "if," "when," etc., used here can be interpreted as "when," "when," "in response to determination," or "in response to detection." Similarly, depending on the context, the phrases "if determination" or "if detection (of the stated condition or event)" can be interpreted as "when determination," "in response to determination," "when detection (of the stated condition or event)," or "in response to detection (of the stated condition or event)."

[0041] As described in the background section, the intelligent massage devices of this technology have poor adaptability and still need improvement in terms of safety, personalization, and comfort.

[0042] Currently, most intelligent massage devices focus on the following areas: 1. Roller pressure adjustment based on pressure sensors; 2. Automatic generation and positioning optimization of massage paths; 3. Safety control measures based on heart rate or temperature. It is evident that these technologies are typically limited to triggering by a single sensor signal, lacking joint analysis of multi-source physiological signals. This leads to the following problems: 1) Due to reliance on a single sensor signal, they cannot accurately reflect the complex state of the human body; 2) Insufficient personalization and adaptive capabilities; 3) Lack of multi-dimensional safety and comfort guarantees, posing a risk of overstimulation or poor efficacy.

[0043] In view of this, the present disclosure provides a massage robot and its control method, control device, and storage medium, which determines the user's fatigue index, comfort index, and risk index by real-time acquisition and fusion of multi-source physiological signals. Massage parameters are then determined in real-time based on these indices and directly applied to the massage execution level. Compared with traditional massage methods that rely on a single sensor (such as a pressure sensor or a timing program), the present disclosure not only more comprehensively and accurately reflects the user's true state but also enables highly personalized adaptive massage adjustment.

[0044] Figure 1 A flowchart illustrating the massage robot control method provided in an embodiment of this disclosure is shown. See also... Figure 1 The control method for a massage robot may include the following steps:

[0045] Step 101: Acquire the user's multimodal signals, which include electroencephalogram (EEG) signals, electrodermal conductance (EDA) signals, skin temperature, body impedance signals, and pressure array signals.

[0046] Step 102: Fuse multimodal signals to obtain the user's fatigue index, comfort index, and risk index; and determine the user's massage path, target area, and massage parameters based on the fatigue index, comfort index, and risk index.

[0047] Step 103: Generate massage operation instructions to control the robotic arm of the massage robot to move the execution end of the massage robot to the target area according to the massage path and control the execution end to perform massage operations according to the massage parameters.

[0048] Further, before step 102 and / or during the execution of step 103, see Figure 2 The method of this embodiment may further include: step 104, performing adaptive threshold safety detection on the EEG signal, skin conductance signal, skin temperature, body impedance signal, and pressure array signal in the multimodal signal respectively; if the number of signals passing the adaptive threshold safety detection in the multimodal signal exceeds a first preset threshold, then the current processing flow or massage operation of the massage robot continues, and the massage intensity of the current massage operation is reduced; if the number of signals passing the adaptive threshold safety detection in the multimodal signal does not exceed the first preset threshold, then an alarm is issued, and the current processing flow or massage operation of the massage robot is terminated.

[0049] In practical applications, the massage intensity can be reduced by a predetermined ratio or by a fixed value. This disclosure does not limit the specific method for reducing the massage intensity; it can be flexibly selected based on the actual application scenario.

[0050] Therefore, safety can be improved through adaptive threshold safety detection to avoid tissue overheating or other similar bodily injuries.

[0051] In some examples, adaptive threshold security detection of EEG signals may include: real-time calculation of the power ratio of beta waves (8–13 Hz) to alpha waves (13–30 Hz) of the EEG signal (i.e., the β / α power ratio), which reflects the user's level of anxiety; comparing the β / α power ratio with a β / α power ratio threshold; if the β / α power ratio is less than or equal to the threshold, the EEG signal security detection is considered passed; if the β / α power ratio is greater than the threshold, the EEG signal security detection is considered failed. Here, the β / α ratio threshold is an adaptive threshold.

[0052] In some examples, the β / α power ratio threshold can be set within an empirical range, referencing the β / α power ratio range observed in humans under stress in relevant studies. For instance, the β / α power ratio threshold could range from approximately 1.5 to 2.5. When a user first uses the massage robot, a personalized β / α power ratio threshold can be determined through benchmark testing. During subsequent use, the β / α power ratio threshold can be dynamically adjusted using a self-learning mechanism within the aforementioned range based on user data.

[0053] The adaptive threshold safety detection of electrodermal (EDS) signals can include: calculating the skin conductivity slope dG / dt (also known as the real-time rate of change of skin conductivity) based on the EDS signals. The skin conductivity slope dG / dt reflects the user's real-time sympathetic nerve excitation level. The skin conductivity slope dG / dt is compared with a skin conductivity rate of change threshold. If the skin conductivity slope dG / dt is less than or equal to the skin conductivity rate of change threshold, the EDS signal safety detection is considered passed. If the skin conductivity slope dG / dt is greater than the skin conductivity rate of change threshold, it can be determined that the level of tension or fatigue has increased, and the EDS signal safety detection is considered failed. The skin conductivity rate of change threshold is an adaptive threshold.

[0054] In some examples, when a user first uses the massage robot or during its initialization phase, a baseline value of the user's skin conductivity at rest can be pre-recorded through user testing or user settings, and a dynamic floating threshold for the user's skin conductivity can be set. The threshold for the rate of change of skin conductivity is the product of the baseline value of the user's skin conductivity at rest and the dynamic floating threshold. The range of the dynamic floating threshold can be pre-set, for example, set to "20%~30% above the baseline," i.e., +20% to +30%. When the user uses the massage robot for the first time, the dynamic floating threshold is set to an initial value within the range of the dynamic floating threshold. During subsequent use of the massage robot, the dynamic floating threshold can be dynamically adjusted based on the user's usage data within this range.

[0055] In some examples, the adaptive threshold safety detection of the skin temperature signal may include: detecting the skin temperature of the massage contact area using a body temperature sensor, comparing the skin temperature of the massage contact area with a skin temperature threshold; if the skin temperature of the massage contact area is less than or equal to the skin temperature threshold, the skin temperature signal safety detection is considered passed; if the skin temperature of the massage contact area is greater than the skin temperature threshold, the skin temperature signal safety detection is considered failed. Here, the skin temperature threshold is an adaptive threshold.

[0056] The skin temperature threshold can be set to the normal comfort zone for humans, such as 34℃~37℃. When a user uses the massage robot for the first time, the user's skin temperature baseline can be determined through user testing. The user's skin temperature threshold can then be set as the user's skin temperature baseline ± a preset amplitude x, where the preset amplitude x ranges from 1℃ to 1.5℃. During subsequent use of the massage robot, the user's skin temperature threshold can be dynamically adjusted based on user usage data, clinical databases, or individualized adjustment instructions, with the user's permission. This skin temperature threshold is not a fixed, single value.

[0057] In some examples, the adaptive threshold safety detection of the pressure array signal may include: detecting the pressure array signal of the massage contact area or the human body contact area through a pressure sensor array; detecting whether the pressure value at each point in the pressure array signal exceeds a pressure threshold; if the pressure value at any point in the pressure array signal exceeds a second preset threshold, the pressure array signal safety detection is determined to have failed; if the pressure value at any point in the pressure array signal is less than the second preset threshold, the pressure value is determined to have passed the pressure array signal safety detection. This pressure threshold is an adaptive threshold.

[0058] In some examples, the pressure threshold range can be set empirically. For instance, it can be set as a reference range under normal human conditions, such as 30-50 Newtons. When a user uses the massage robot for the first time, their personalized pressure threshold can be determined through a sensitivity test. During subsequent use, the user's pressure threshold can be dynamically adjusted within the aforementioned range based on feedback signals and user data using a self-learning mechanism to avoid excessive pressure at a single point, which could negatively impact the user's massage experience.

[0059] In some examples, adaptive threshold safety detection of body impedance signals may include: detecting the user's body impedance signal using a bioimpedance sensor, which reflects contact quality and tissue tolerance; calculating the amplitude of body impedance change using the user's body impedance signal; comparing the amplitude of body impedance change ΔZ with a threshold value; if the amplitude of body impedance change ΔZ is less than or equal to the threshold value, the body impedance safety detection is considered passed; if the amplitude of body impedance change ΔZ is greater than the threshold value, there may be contact abnormalities or displacement errors, and the body impedance safety detection is considered failed. Here, the threshold value for body impedance change is an adaptive threshold.

[0060] When a user uses the massage robot for the first time, the user's resting body impedance baseline can be determined through user testing. The threshold for the change in the user's body impedance can be set to 20% to 30% of the user's resting body impedance baseline. During subsequent use of the massage robot, the user's resting body impedance baseline or the threshold for the change in body impedance can be automatically updated based on the user's usage data to improve recognition accuracy.

[0061] It should be noted that, in order to ensure the reliability of the system and user safety, the various thresholds involved in the embodiments of this disclosure are not fixed values, but are dynamically determined based on individual user differences, adaptive learning and reference database.

[0062] The control strategy for adaptive threshold security detection in step 104 can be flexibly set.

[0063] In some examples, the control strategy for safety detection may include: if the number of signals exceeding its adaptive threshold does not exceed a first preset threshold, the massage intensity can be automatically reduced and a protection mode entered according to a preset ratio (e.g., 20–30%); if the number of signals exceeding its adaptive threshold simultaneously exceeds the first preset threshold, an emergency shutdown is triggered (e.g., triggering the emergency shutdown procedure of the EEU in the massage robot), and the massage operation parts of the massage robot, such as the actuator and robotic arm, are shut down to ensure user safety. Here, the upper limit of the number of signals can be flexibly set. Assuming the massage robot contains 7 sensors for detecting the user's physiological state, the upper limit of the number of signals can be set to 5, 6, or 7.

[0064] In some examples, the safety detection and control strategy may also include: triggering an alarm when a specific signal (e.g., skin temperature signal or body impedance signal) exceeds its adaptive threshold to remind the user to make manual adjustments, while automatically reducing the massage frequency and / or massage intensity to avoid overheating or overstimulation, thereby ensuring no tissue damage occurs and further improving safety and reliability.

[0065] In step 101, multimodal signals can be acquired in real time using the multimodal sensors of the massage robot. Specifically, electroencephalogram (EEG) signals can be acquired by an EEG sensor, electrodermal signal can be acquired by an electrodermal sensor, skin temperature can be acquired by a body temperature sensor, body impedance signals can be acquired by a bioimpedance sensor, and pressure array signals can be acquired by a pressure sensor array. In practical applications, all signals in the multimodal signal set are acquired synchronously.

[0066] Pressure sensor arrays can be deployed inside the backrest, seat, and actuator of the massage robot. The pressure sensor array monitors the pressure array signals in real time, indicating the pressure distribution between the user's body and the massage robot. It can reflect the force distribution and contact area of ​​various parts of the user's body, and detect the user's sitting posture, body contact points, and local force conditions.

[0067] The EEG sensor can be a standalone head-mounted device. The EEG sensor can collect EEG signals containing parameters such as alpha, beta, and theta waves. The EEG signals can reflect the degree of nerve excitation and relaxation, and can be used to assess relaxation level, attention level and emotional state, serving as an important basis for adjusting the massage rhythm and intensity.

[0068] Electrodermal sensors can be deployed via wristbands or hand-held electrodes. The electrodermal signals collected by the sensors reflect the activity of the sympathetic nervous system, that is, the level of emotional arousal and tension, and can be used to assess the user's fatigue level and stress level.

[0069] Body temperature sensors and bioimpedance sensors can be embedded in the massage contact points of the actuator (such as the fingertips or palm of a bionic hand). The body temperature sensor can monitor the skin temperature of the user's massage area, which reflects the local temperature rise of the massage area. The bioimpedance sensor collects body impedance signals, which reflect the tissue impedance of the massage area. The local temperature rise and tissue impedance can reflect the local metabolic level and contact status. By using skin temperature and body impedance signals, local overstimulation during the massage process can be effectively avoided.

[0070] Before step 102, the method of this embodiment may further include: preprocessing the multimodal signals. Specifically, preprocessing may include, but is not limited to, filtering, denoising, and normalization, to improve the quality of the multimodal signals. Different types of signals can use different filtering methods. For example, EEG signals can use a 0.5–40 Hz bandpass filter, and EEG signals can use a 0.05–1 Hz low-pass filter. Denoising methods may include, but are not limited to, wavelet denoising and moving average. For example, EEG signals can use wavelet denoising, while pressure array signals can use moving average for denoising. Normalization may use, but is not limited to, Z-score normalization, which ensures the uniformity of dimensions across different sensors.

[0071] In step 102, fuzzy estimation can be used to achieve the fusion processing of multimodal signals. Specifically, the process of fusing multimodal signals to obtain the user's fatigue index, comfort index, and risk index may include: extracting features from each signal in the multimodal signals to obtain the features of each signal, and fusing the features of each signal to obtain the user's fatigue index, comfort index, and risk index.

[0072] Specifically, the characteristics of pressure array signals include, but are not limited to, centroid coordinates (Xc, Yc) and pressure distribution entropy Hp; the characteristics of electroencephalogram (EEG) signals include, but are not limited to, relative α power and β / α power ratio; the characteristics of electrodermal (EDS) signals include, but are not limited to, skin conductivity slope dG / dt and average skin conductivity level Gmean; the characteristics of skin temperature signals include, but are not limited to, instantaneous skin temperature T and skin temperature change rate dT / dt; and the characteristics of body impedance signals include, but are not limited to, mean body impedance Zmean and body impedance change amplitude ΔZ. The body impedance change amplitude can also be referred to as the contact stability index.

[0073] To ensure the robustness and adaptability of the system, the fusion processing in this embodiment can employ a three-layer fusion computation and control method. Specifically, the fusion processing in step 102 may include: sensor quality assessment and weighted modeling, Bayesian adaptive Kalman / covariance intersection fusion, and uncertainty-aware control decision-making. This three-layer fusion computation and control method can explicitly introduce the uncertainty of multimodal signals into the control loop, thereby achieving safe and personalized closed-loop massage regulation.

[0074] The comfort index reflects the stability of a user's body position and contact, the fatigue index reflects sympathetic nerve activity and muscle tension, and the risk index reflects the user's subjective comfort level. All three indices can be determined through a three-layer fusion calculation. Specifically, an exemplary implementation of step 102, which fuses the characteristics of each signal to obtain the user's fatigue, comfort, and risk indices, may include: determining the weights of each signal in the multimodal signal through sensor signal quality assessment; determining the state vector and state estimation covariance matrix based on the characteristics and weights of each signal in the multimodal signal using extended Kalman filtering or covariance methods; finally, determining a penalty term based on the state vector, preset index weights, and the state estimation covariance matrix; and obtaining the index values ​​through the state vector and penalty term. Here, the preset index weights include preset comfort weights, preset fatigue weights, and preset risk weights.

[0075] In some examples, the process of determining the weights of various sensor signals through sensor signal quality assessment may include:

[0076] First, for each sensor i of the massage robot, the weight distribution of sensor i is obtained through Bayesian update.

[0077] For example, the weight distribution of sensor i can be obtained by the following equation (1).

[0078] (1)

[0079] in, Indicates the Dirichlet distribution. For the residual of sensor i, This is the mapping function between the residuals and the weight adjustments. This represents the weight distribution of sensor i.

[0080] In practical applications, the residual of sensor i The prediction model can be determined based on the difference between the measured signal of sensor i and the predicted signal estimated by its prediction model. Different types of sensors have different prediction models. For example, the prediction model for a pressure sensor can be, but is not limited to, a pre-fitted regression model, which indicates the mapping relationship between the pressure sensor's motor control signal and operating parameters such as motor current and its massage intensity. For example, the prediction model for a body temperature sensor can be, but is not limited to, a time-series model, which indicates the trend of temperature change over time. As another example, the prediction model for a conductance skin sensor can be, but is not limited to, a physiological model, which can be trained using historical data from the conductance skin sensor. The prediction model for an electroencephalogram (EEG) sensor can be, but is not limited to, a statistical model, etc. In practical applications, the prediction models for each sensor are not limited to the above situations, and this disclosure does not impose any limitations on this.

[0081] Secondly, based on weight distribution The weight of sensor i is calculated. For example, the weight of sensor i can be calculated using the following equation (2).

[0082] (2)

[0083] in, Represents the weight distribution The expected value, i.e., the weight distribution The weighted average of all possible values.

[0084] In practical applications, after determining the weights of each sensor signal through sensor signal quality assessment, further adjustments can be made based on the user's personalized characteristics. For example, for users sensitive to alpha waves in EEG signals, the weight of the EEG signal can be increased. This increases the weight of EEG signals in comfort assessments. For users who sweat easily, the weight of TENS signals can be increased. This is to increase the weight of skin electrical signals in fatigue assessment.

[0085] In some examples, the process of determining the state vector based on the features and weights of various sensor signals using extended Kalman filtering or covariance methods may include: under multi-sensor input, using extended Kalman filtering (EKF) or covariance intersection (CI) to perform posterior estimation to obtain the state vector based on multi-modal signals and their weights.

[0086] The covariance matrix and state vector after the fusion of all sensor signals can be obtained by the following equations (3) to (4).

[0087] (3)

[0088] (4)

[0089] in, Let i represent the covariance matrix of sensor i. This represents the local state estimate of sensor i. This represents the weight of sensor i. This represents the covariance matrix after fusing all sensor signals. This represents the state vector after the signals from all sensors have been fused. It represents the inverse of the covariance matrix after the signals from all sensors are fused.

[0090] Wherein, the state vector It can be expressed as the following formula (5):

[0091] (5)

[0092] in, Represents relative α power. Indicates the β / α power ratio. This represents the reciprocal of the slope of skin conductivity. Represents the entropy of pressure distribution. Indicates the instantaneous value of skin temperature. Indicates the mean body impedance. This indicates the transpose operation. This represents the state vector after the signals from all sensors have been fused.

[0093] In some examples, the comfort index, fatigue index and risk index can be calculated by the following formulas (6) to (8).

[0094] (6)

[0095] (7)

[0096] (8)

[0097] Where S represents the comfort index, F represents the fatigue index, and R represents the risk index. Representing the covariance matrix, it can be used to estimate uncertainty. Represents the state vector. Represents trace operation. This indicates the preset comfort weight. This indicates the preset fatigue weight. This indicates the preset risk level weight. , , These are the comfort adjustment coefficient, fatigue adjustment coefficient, and risk adjustment coefficient, respectively.

[0098] in, As a penalty item for the comfort index, This is a penalty item for the fatigue index. This is a penalty item for the risk level index. It is a state variable This is the base function for variables; the specific function form can be defined in advance. For example, It can be defined as a multi-objective optimization function. The values ​​represent the comfort index, fatigue index, and risk index under ideal massage effect.

[0099] By introducing an uncertainty penalty term into the calculation of the comfort index S, fatigue index F, and risk index R, the system can automatically adopt a more conservative strategy to perform massage operations under high uncertainty conditions, thereby improving the massage effect and user experience while enhancing the safety of the massage operation.

[0100] In equations (6) to (8), the comfort weights Fatigue weight and risk weight It can be pre-configured and flexibly adjusted. During the user's use of the massage robot, with the user's permission, the comfort weight can be automatically optimized through a self-learning mechanism based on different users' usage data. Fatigue weight and risk weight .

[0101] In step 102, a massage control scheme can be dynamically generated and iteratively modified based on the fatigue index, comfort index, and risk index to determine the user's massage path, target area, and massage parameters. Specifically, the process of determining the user's massage path, target area, and massage parameters based on the fatigue index, comfort index, and risk index in step 102 may include, but is not limited to: ① selection of target areas such as shoulders, waist, and hips; ② using digital acupoint maps to achieve millimeter-level precise positioning to complete acupoint combination planning; ③ optimizing the massage path and trajectory based on the principle of ensuring complete coverage and reducing ineffective repetition; ④ adaptive adjustment of various massage parameters such as the location of each acupoint, massage intensity, massage duration, massage technique, and / or massage rhythm.

[0102] Specifically, massage paths and trajectories can be optimized through path optimization and obstacle avoidance constraints. Path optimization algorithms can be, but are not limited to, gradient descent, conjugate gradient method, quasi-Newton method, etc. Obstacle avoidance constraints can prevent encountering obstacles such as wounds or bony protrusions on the body during the massage. Obstacle avoidance constraints can include, but are not limited to, the following: for each path point and each obstacle, the distance between each path point and each obstacle is less than or equal to a preset safety distance.

[0103] Furthermore, during the path optimization process, path smoothness can be optimized by minimizing the curvature or turning angle of the path points, ensuring a smooth massage path and avoiding sharp turns, thereby further improving comfort.

[0104] Furthermore, during the path optimization process, constraints such as boundary constraints can be added. Boundary constraints may include, but are not limited to, requiring all path points to be within the target area.

[0105] Furthermore, in path optimization and obstacle avoidance constraints, coverage optimization and re-optimization avoidance can be used. Coverage optimization ensures that the massage area is fully covered, avoiding omissions or repeated massages. Re-optimization avoidance can prevent repeated massages and improve system efficiency and response speed. For example, re-optimization avoidance can be achieved by setting one or more of the following strategies in path optimization: 1) If the target area only changes slightly (e.g., minor adjustment of obstacle position), optimization can start from the previously planned path; 2) Record the already covered areas and reduce the coverage requirements for these areas in subsequent planning, thereby avoiding repeated massages.

[0106] As a result, while ensuring that the massage covers the target area, it avoids wounds, bony protrusions, etc. on the user's body, thereby improving the massage effect and avoiding injury, and further enhancing the reliability and safety of the massage operation.

[0107] In step 102, after determining the massage path, a path node threshold verification can be further performed to correct the massage path. Path node threshold verification verifies the safety, feasibility, and effectiveness of each node or predetermined key node on the massage path after its generation. In some examples, path node threshold verification includes one or more of the following: 1) Whether the massage intensity of the path node is within the preset safety range of the corresponding acupoint (different areas have different intensity ranges); 2) Whether the massage intensity of the path node is within the effective intensity range of the corresponding acupoint, i.e., whether the pressure is effective; 3) Whether the position of the path node is within the preset tolerance range of the corresponding acupoint. In specific applications, the content of the path node threshold verification can be flexibly adjusted as needed, and different threshold verification content can be set for different path nodes or different acupoints. This disclosure does not impose any limitations on this.

[0108] Threshold verification of path nodes can ensure the safety, feasibility and effectiveness of massage paths before actual execution, further reducing the risks of massage operations.

[0109] The following effects can be achieved through the aforementioned step 102: 1) When the pressure distribution entropy shows that the shoulder is under concentrated pressure, the shoulder and neck area is automatically selected as the target area and acupoints are located to obtain the acupoint combination in the shoulder area; 2) When the fatigue index value is too high, the massage intensity can be automatically reduced and the massage duration can be extended to relieve fatigue; 3) When the risk index value is insufficient, the massage rhythm and heat assistance can be automatically increased to improve the pleasure level; 4) When the comfort index value is close to the critical value, the massage intensity or massage frequency can be adjusted in time to reduce the stimulation intensity or switch to low power mode.

[0110] Step 102 obtains various user indicators by fusing multimodal signals and determines massage parameters based on these indicators, which can achieve at least the following beneficial effects: ① Joint judgment by pressure sensor array and comfort index avoids the errors of traditional fixed paths and can effectively improve the accuracy of acupoint positioning; ② Combining EEG signals and skin electrical signals allows the massage rhythm to match the user's real feelings, resulting in better personalized adaptability; ③ Safety detection based on body impedance signals and skin temperature can ensure that no tissue damage occurs, further improving safety; ④ Comprehensive massage path planning and massage parameter determination based on the user's comfort index, fatigue index, and risk index can relieve muscle fatigue while taking into account psychological relaxation and comfort experience, resulting in better overall health benefits.

[0111] In step 103, after generating the massage operation command, the massage operation command is sent to the EEU via the CCU, driving the actuator to complete the corresponding massage operation. Specifically, the control process in step 103 may include: the massage robot's CCU generating a massage command based on the massage path, target area, and massage parameters and sending it to the massage robot's EEU; the EEU generating a position command for the robotic arm and an action command for the actuator based on the massage command and sending them to the robotic arm and actuator respectively; the action command for the actuator being generated based on the massage parameters; and the position command for the robotic arm being generated based on the massage path; the robotic arm responding to the position command moving the actuator to the target area; and the actuator responding to the action command performing a massage operation on the user's target area.

[0112] Further, see Figure 2 The massage robot control method of this disclosure embodiment may further include: step 105, adjusting massage parameters based on user feedback during the massage operation. Specifically, see... Figure 2Step 105 may include: acquiring one or a combination of the following signals collected by the user during the massage operation: electroencephalogram (EEG), electrodermal signal, skin temperature, body impedance signal, pressure array signal, facial expression image, and user commands; and adjusting massage parameters based on one or a combination of these signals dynamically generated during the massage operation. This allows for continuous correction of the massage operation within a real-time feedback loop.

[0113] In some examples, step 102 can be repeatedly executed based on multimodal signals such as electroencephalogram (EEG) signals, electrodermal osmosis (EDS) signals, skin temperature, body impedance signals, and pressure array signals to adjust massage parameters. The specific execution process is the same as the processing procedure of step 102 described above, and will not be repeated here.

[0114] In some examples, the steps of adjusting massage parameters based on EEG signals may include: calculating in real time the power ratio of alpha waves (8–13 Hz) to beta waves (13–30 Hz) in the EEG signal (i.e., the α / β power ratio); if the α / β power ratio increases by a first predetermined proportion (e.g., 10%) of the α / β baseline value, it indicates that the user has entered a relaxed state, and the current massage parameters and massage path can be maintained; if the α / β power ratio decreases by a second predetermined proportion (e.g., 90%) of the α / β baseline value, the massage intensity can be automatically reduced proportionally (e.g., reduced by about 15%), and the massage duration can be shortened, while also prompting the user so that the user can actively adjust their posture in a timely manner.

[0115] Alpha waves are brain waves with frequencies between 8 and 13 Hz. They typically appear when the eyes are closed, the brain is relaxed, and the patient is awake and at rest. Alpha waves are a sign that the brain is in a calm, introspective state, not focused on the external world. For example, alpha wave activity increases during meditation or quiet rest. Beta waves are brain waves with frequencies between 13 and 30 Hz. They typically dominate when the eyes are open, thinking, problem-solving, focusing, or experiencing anxiety or tension. Beta waves represent a highly active and excited state of the cerebral cortex. The alpha / beta power ratio can be used to measure the balance between the brain's "relaxed" and "excited" states.

[0116] The α / β baseline value can be preset. Specifically, the average α / β power ratio measured when the user is in a quiet, neutral state can be detected beforehand, and this average value can be used as the user's α / β baseline value. For each user, their personalized baseline value can be monitored and saved before the massage.

[0117] In some examples, adjusting massage parameters based on skin conductance signals may include: if the user's skin conductance increases by more than a preset first conductance threshold (e.g., 0.2 microseconds) within a first predetermined duration (e.g., within 10 seconds), the actuator can be controlled to immediately reduce the massage intensity and switch to a gentle mode; if the user's skin conductance changes by less than a preset second conductance threshold (e.g., 0.05 microseconds) within the first predetermined duration (e.g., within 10 seconds), it indicates that the user is stably relaxed, and the massage duration can be extended proportionally (e.g., extended by approximately 20%).

[0118] In some examples, adjusting massage parameters based on the user's facial expression images can include: capturing the user's facial expression images through a camera during the massage, performing facial expression recognition on the images, and determining that the user is currently uncomfortable if their expression is characterized by furrowed brows or downturned corners of the mouth, thus controlling the execution end to immediately reduce the massage intensity or massage path; if the user's expression is characterized by a wide smile, a gentle smile, or upturned corners of the mouth, the execution end can maintain the current massage intensity and duration.

[0119] The specific implementation process of adjusting massage parameters based on skin temperature is similar to that of the aforementioned electroencephalogram (EEG) signals, electrodermal signals, and facial expression images, and will not be described in detail here.

[0120] Specifically, user instructions can be, but are not limited to, voice instructions, touch instructions, or any other type of user-issued instructions.

[0121] In some examples, a user can issue a voice command like "lighter," and the massage robot will respond by decreasing the current massage intensity according to a pre-configured intensity adjustment range (e.g., 3 Newtons). Conversely, a user can issue a voice command like "upper," and the massage robot will respond by moving the coordinates of the acupoint location upwards by 5 millimeters according to a pre-configured position adjustment range (e.g., 5 millimeters). In practical applications, speech recognition algorithms, such as keyword matching algorithms, can be used to achieve real-time adjustment of massage parameters based on user voice commands. This ensures that the massage robot can accurately execute user voice commands even in noisy environments.

[0122] In some examples, users can also make fine adjustments via the armrest panel. For instance, the human-computer interaction module of the massage robot can provide physical buttons such as intensity sliders, four-way position buttons, and / or mode switching buttons. Users can input touch commands by operating these physical buttons, and the massage robot's control unit responds to these touch commands by adjusting massage parameters such as massage intensity, massage location, massage duration, and / or massage techniques within preset safety boundaries (e.g., a maximum neck massage intensity of 30 Newtons). Thus, users can instantly correct massage parameters through voice, panel input, etc., achieving a closed-loop user interaction and further enhancing the user experience.

[0123] In addition to voice commands and touch commands, user commands can also be input through various other methods such as eye tracking, gesture recognition, and remote control via a mobile app. This disclosure does not limit the specific form of user commands or their input methods.

[0124] Furthermore, the method in this embodiment may further include: a human-computer interaction module acquiring user instructions, and a control unit selecting a massage mode and / or adjusting massage parameters in response to the user instructions. The human-computer interaction module may include, but is not limited to, a microphone, a touchscreen, physical buttons, and communication components for communicating with external devices. User instructions may be, but are not limited to, voice instructions, touch instructions, button instructions, or instruction signals from external devices. For example, a human-computer interaction interface can be provided via a touchscreen, allowing the user to perform touch operations to input user instructions. The communication component receives user instructions from external devices such as mobile terminals, which have a massage application installed, allowing the user to operate within the massage application to input user instructions. It should be noted that the human-computer interaction module is not limited to the above-mentioned types and may also include, for example, gesture recognition components. This embodiment does not impose any limitations on this.

[0125] Massage modes can be pre-configured. For example, massage modes may include, but are not limited to, "deep relaxation," "rapid recovery," and "neck and shoulder focus." This disclosure does not limit the types of massage modes or their configuration methods.

[0126] As described above, users can easily adjust massage parameters proactively through various methods such as voice, touch, and apps. Furthermore, this embodiment of the invention utilizes a dual mapping of multiple physiological signals and the user's subjective experience to ultimately form an intelligent, adaptive massage closed-loop optimization framework: it can dynamically match individual differences while achieving a balance between massage effectiveness, safety, and comfort.

[0127] Furthermore, the method in this embodiment may also include: the control unit controlling a display component such as a touch screen to display the massage process and various indicators of the user in a visual manner in real time, so that the user can intuitively understand the massage process and further improve the user experience and safety.

[0128] In this embodiment, the synergistic effect of multimodal sensors significantly improves the robustness and personalization of the massage robot system: the pressure sensor array analyzes the force distribution and body posture information to achieve high-precision acupoint recognition; the alpha and beta wave characteristics in the electroencephalogram (EEG) signal are used to assess relaxation and pleasure, providing a basis for rhythm and circadian rhythm optimization; changes in skin conductivity sensitively reflect tension or fatigue, triggering force reduction or stimulus shift; monitoring skin temperature and body impedance not only safeguards safety thresholds but also assesses contact quality and tissue tolerance. In other words, this embodiment monitors various indicators such as user comfort, fatigue, and risk level through real-time acquisition and fusion of multi-source physiological signals, and plans massage paths and determines massage parameters based on these indicators. Compared to traditional massage methods based on only a single sensor, this not only more comprehensively and accurately reflects the user's true state but also enables highly personalized adaptive adjustment, achieving comprehensiveness, personalization, and safety that existing single-sensor control lacks.

[0129] Figure 3 A schematic diagram of the structure of the massage robot control device provided in an embodiment of this disclosure is shown. See also Figure 3 The massage robot control device 300 may include:

[0130] The signal acquisition unit 301 is used to acquire the user's multimodal signals, including electroencephalogram (EEG) signals, electrodermal conductance (EDA) signals, skin temperature, body impedance signals, and pressure array signals.

[0131] The fusion unit 302 is used to fuse multimodal signals to obtain the user's fatigue index, comfort index and risk index, and to determine the user's massage path, target area and massage parameters based on the fatigue index, comfort index and risk index;

[0132] The massage operation unit 303 is used to generate massage operation instructions to control the robotic arm of the massage robot to move the execution end of the massage robot to the target area according to the massage path and to control the execution end to perform massage operation according to the massage parameters.

[0133] Furthermore, the massage robot control device 300 may also include: an adaptive threshold safety detection unit 304, used to perform adaptive threshold safety detection on the electroencephalogram (EEG) signal, electrodermal signal, skin temperature, body impedance signal, and pressure array signal in the multimodal signals respectively; if the number of signals passing the adaptive threshold safety detection in the multimodal signals exceeds a first preset threshold, the current processing flow or massage operation of the massage robot continues, and the massage intensity of the current massage operation is reduced; if the number of signals passing the adaptive threshold safety detection in the multimodal signals does not exceed the first preset threshold, an alarm is issued, and the current processing flow or massage operation of the massage robot is terminated.

[0134] Furthermore, the adaptive threshold security detection unit 304 can be specifically used for:

[0135] The β / α power ratio of the EEG signal is calculated in real time. When the β / α power ratio is less than or equal to the β / α power ratio threshold, the adaptive safety detection of the EEG signal is determined to be passed. The β / α ratio threshold is the adaptive threshold.

[0136] The skin conductivity slope is calculated based on the skin conductivity signal. When the skin conductivity slope is less than or equal to the skin conductivity change rate threshold, the adaptive safety detection of the skin conductivity signal is determined to be passed. The skin conductivity change rate threshold is the adaptive threshold.

[0137] The system detects the skin temperature in the massage contact area. When the skin temperature in the massage contact area is less than or equal to a skin temperature threshold, the adaptive safety detection of the skin temperature signal is determined to be passed. The skin temperature threshold is the adaptive threshold.

[0138] The body impedance change amplitude is calculated using the user's body impedance signal. When the body impedance change amplitude is less than or equal to the body impedance change amplitude threshold, the adaptive safety detection of the body impedance signal is determined to pass. The body impedance change amplitude threshold is the adaptive threshold.

[0139] The pressure array signal is tested to determine whether the pressure value at each point exceeds the pressure threshold. If the number of points in the pressure array signal with a pressure value greater than the pressure threshold does not exceed the second preset threshold, the adaptive safety detection of the pressure array signal is determined to be passed. The pressure threshold is the adaptive threshold.

[0140] Furthermore, the fusion unit 302 can be specifically used to: extract features from each signal in the multimodal signal to obtain the features of each signal, and perform fusion processing on the features of each signal to obtain the user's fatigue index, comfort index and risk index.

[0141] Furthermore, the fusion unit 302 can be specifically used to: determine the weight of each signal in the multimodal signal through sensor quality assessment; determine the user's current state vector and state estimation covariance matrix based on the weight of each signal in the multimodal signal, wherein the state vector includes relative α power, β / α ratio, reciprocal of skin conductivity slope, instantaneous body temperature, pressure distribution entropy, and mean body electrical impedance; and calculate the comfort index, fatigue index, and risk index based on the user's current state vector and state estimation covariance matrix.

[0142] Furthermore, the comfort index is the difference between the result of the fundamental function operation of the state vector and the first penalty term, whereby the first penalty term is the product of a preset comfort adjustment coefficient and the result of the first trace operation, and the result of the first trace operation is the trace operation result of the product of the state estimation covariance matrix and the preset comfort weights; and / or,

[0143] The fatigue index is the difference between the result of the fundamental function operation of the state vector and the second penalty term. The second penalty term is the product of a preset fatigue adjustment coefficient and the result of the second trace operation. The result of the second trace operation is the trace of the product of the state estimation covariance matrix and the preset fatigue weight; and / or,

[0144] The risk index is the difference between the result of the basic function operation of the state vector and the third penalty term. The third penalty term is the product of the preset risk adjustment coefficient and the result of the third trace operation. The result of the third trace operation is the trace operation result of the product of the state estimation covariance matrix and the preset risk weight.

[0145] Furthermore, the massage robot control device 300 may also include: a feedback correction unit 305, used to acquire one or a combination of the following during the massage operation: electroencephalogram (EEG) signals, electrodermal signals, skin temperature, body impedance signals, pressure array signals, facial expressions, and user commands; and to dynamically adjust the massage parameters based on one or a combination of the following during the massage operation: EEG signals, electrodermal signals, skin temperature, body impedance signals, pressure array signals, facial expressions, and user commands.

[0146] Further technical details regarding the massage robot control device 300 can be found in the preceding section on massage robot control methods, and will not be repeated here. The massage robot control device 300 can be implemented as software, hardware, or a combination of both. For example, the massage robot control device 300 can be implemented as, but is not limited to, the control unit 401 of the massage robot 400 described below.

[0147] Figure 4 A schematic diagram of the structure of the massage robot provided in an embodiment of this disclosure is shown. See also... Figure 4 The massage robot may include: a pressure sensor array 402, an EEG sensor 403, a skin conductance sensor 404, a body temperature sensor 405, a bioimpedance sensor 406, a control unit 401, a robotic arm 407, and an actuator 408. The pressure sensor array 402, EEG sensor 403, skin conductance sensor 404, body temperature sensor 405, bioimpedance sensor 406, robotic arm 407, and actuator 408 are respectively connected to the control unit 401. The control unit 401 includes one or more processors and a memory storing programs. The programs include instructions, which, when executed by the processor, cause the processor to perform the control method of the massage robot described above.

[0148] The aforementioned multimodal sensors of the massage robot can form a multimodal perception layer. Further, this multimodal perception layer may also include a camera 409, which can be used to capture images of the user's facial expressions and provide them to the control unit. In specific applications, the multimodal perception layer of the massage robot may also include, for example, electromyography (EMG) sensors, heart rate variability sensors, blood oxygen sensors, or other types of sensors to indirectly monitor the degree of relaxation or fatigue. This disclosure does not limit the types of sensors used in the multimodal perception layer.

[0149] See Figure 4 The control unit 401 may include a Central Control Unit (CCU) 4011 and an Edge Execution Unit (EEU) 4012. In some examples, the CCU and EEU can be implemented as two independent control subunits, i.e., the CCU and EEU can each contain a processor and memory, respectively. In some examples, the CCU and EEU can be implemented as two independent processor architectures sharing memory. This disclosure does not limit the specific implementation of the CCU and EEU.

[0150] The CCU 4011 can be used to preprocess (denoise, normalize, and extract features) multimodal signals and fuse them to obtain the user's fatigue index, comfort index, and risk index. As the "brain" of the massage robot system, the CCU 4011 is responsible for receiving multimodal signals from the multimodal perception layer, performing time synchronization, feature extraction, and fusion calculations, and outputting the comfort index S, fatigue index F, and risk index R. These three indices reflect the user's subjective comfort level, objective fatigue accumulation, and safety risk level, respectively. For example, the CCU 4011 can incorporate various algorithm modules, including Kalman filters, fuzzy logic inference, Bayesian inference, and weighted multi-index evaluation, to achieve multimodal signal fusion.

[0151] The EEU 4012 can be used to determine and / or dynamically adjust the user's massage path, target area, and massage parameters based on the fatigue index, comfort index, and risk index provided by the CCU4011. This includes determining the massage area selection, meridian point location, path coverage sequence, massage intensity, massage frequency, and massage duration. It generates massage operation commands to directly control massage actuators, such as robotic arms or actuators, to perform the massage operation. Furthermore, the EEU 4012 can maintain high-speed communication with the multimodal sensing layer to receive feedback in real time and quickly adjust massage parameters, ensuring a stable closed-loop massage control.

[0152] The massage actuator consists of a robotic arm 407 and an actuator 408. The robotic arm 407 can deliver the actuator 408 to the target area or acupoint, and the actuator 408 can perform massage operations according to the massage parameters determined by the EEU 4012. In specific applications, the actuator 408 can be implemented in various forms, including but not limited to a robotic hand, a bionic hand, a wearable exoskeleton, a flexible pneumatic actuator, an external massage operation platform, an airbag, a vibration module, and a heat pack, to adapt to the needs of different scenarios. This disclosure does not limit the specific type and implementation of the actuator.

[0153] Furthermore, the massage robot 400 of this embodiment may also include a human-computer interaction module 410, which receives user commands and provides them to the control unit 401. Specifically, the human-computer interaction module can provide a user-initiated input and information output interface, including a touch screen, a voice command interface, or a mobile app. Users can select preset modes (such as "deep relaxation," "rapid recovery," or "neck and shoulder focus"), or make real-time adjustments during use. In addition, the human-computer interaction module can also be used to visually present the massage process and current status indicators, allowing users to intuitively understand their relaxation level and safety risks.

[0154] The human-computer interaction module 410 may include, but is not limited to, touch buttons, microphones, eye-tracking components, gesture recognition components, brain-computer interfaces, etc., for receiving user commands in order to adjust massage parameters through the user commands. This disclosure does not limit the types of components in the human-computer interaction module 410.

[0155] The massage robot 400 provided in this embodiment has an overall system architecture including a CCU 4011, an EEU 4012, a multimodal perception layer, and a human-computer interaction module 410. These four components work together to form a closed-loop control link for massage, including data acquisition, information processing, action execution, and user feedback.

[0156] In practical applications, massage robots can be implemented as, but are not limited to, smart massage chairs.

[0157] In addition, embodiments of this disclosure also provide a computer-readable storage medium having a computer program stored thereon, the program including instructions that, when executed by one or more processors of a computing device, perform the steps of the aforementioned massage robot control method.

[0158] The aforementioned programs (also known as software, software applications, or code) include the machine instructions of a programmable processor and can be implemented using object-oriented programming languages, assembly language, or machine language.

[0159] With the development of time and technology, the meaning of "medium" has become increasingly broad. The dissemination of computer programs is no longer limited to tangible media; they can also be downloaded directly from the network. Any combination of one or more computer-readable storage media can be used. Computer-readable storage media can be, but is not limited to, electrical, magnetic, optical, electromagnetic, infrared, or semiconductor systems, devices, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM or flash memory), optical fiber, portable compact disk read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this document, a computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, device, or apparatus.

[0160] The technical solutions provided in this disclosure have been described in detail above. Specific examples have been used to illustrate the principles and implementation methods of this disclosure. The descriptions of the embodiments above are only for the purpose of helping to understand the methods and core ideas of this disclosure. Furthermore, those skilled in the art will recognize that, based on the ideas of this disclosure, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of this disclosure.

[0161] The above description is merely a preferred embodiment of this disclosure and is not intended to limit this disclosure. Any modifications or equivalent substitutions made within the spirit and principles of this disclosure should be included within the scope of protection of this disclosure.

Claims

1. A control method for a massage robot, characterized in that, The method includes: Acquire the user's multimodal signals, including electroencephalogram (EEG) signals, electrodermal signal, skin temperature, body impedance signal, and pressure array signal; The multimodal signals are fused to obtain the user's fatigue index, comfort index, and risk index. The user's massage path, target area, and massage parameters are determined based on the fatigue index, comfort index, and risk index. Generate massage operation instructions to control the robotic arm of the massage robot to move the execution end of the massage robot to the target area according to the massage path and control the execution end to perform massage operation according to the massage parameters; Adaptive threshold security detection is performed on the electroencephalogram (EEG) signal, electrodermal signal, skin temperature, body impedance signal, and pressure array signal in the multimodal signal respectively; If the number of signals in the multimodal signals that pass the adaptive threshold security detection exceeds the first preset threshold, then the current processing flow or massage operation of the massage robot continues, and the massage intensity of the current massage operation is reduced. If the number of signals in the multimodal signals that pass the adaptive threshold security detection does not exceed the first preset threshold, an alarm is issued to terminate the current processing flow or massage operation of the massage robot.

2. The method according to claim 1, characterized in that, The adaptive threshold security detection of the EEG signal, SCART signal, skin temperature, body impedance signal, and pressure array signal in the multimodal signal includes: The β / α power ratio of the EEG signal is calculated in real time. When the β / α power ratio is less than or equal to the β / α power ratio threshold, the adaptive security detection of the EEG signal is determined to be passed. The β / α ratio threshold is an adaptive threshold. The skin conductivity slope is calculated based on the skin conductivity signal. When the skin conductivity slope is less than or equal to the skin conductivity change rate threshold, the adaptive safety detection of the skin conductivity signal is determined to be passed. The skin conductivity change rate threshold is the adaptive threshold. The skin temperature in the massage contact area is detected, and the adaptive safety detection of the skin temperature is determined to be passed when the skin temperature in the massage contact area is less than or equal to a skin temperature threshold, wherein the skin temperature threshold is an adaptive threshold. The body impedance change amplitude is calculated using the user's body impedance signal. When the body impedance change amplitude is less than or equal to a body impedance change amplitude threshold, the adaptive security detection of the body impedance signal is determined to pass. The body impedance change amplitude threshold is the adaptive threshold. The pressure value at each point in the pressure array signal is detected to exceed a pressure threshold. If the number of points in the pressure array signal with a pressure value greater than the pressure threshold does not exceed a second preset threshold, the adaptive safety detection of the pressure array signal is determined to be passed. The pressure threshold is the adaptive threshold.

3. The method according to claim 1, characterized in that, The process of fusing the multimodal signals to obtain the user's fatigue index, comfort index, and risk index includes: extracting features from each signal in the multimodal signals to obtain the features of each signal, and fusing the features of each signal to obtain the user's fatigue index, comfort index, and risk index.

4. The method according to claim 3, characterized in that, The process of fusing the features of each signal to obtain the user's fatigue index, comfort index, and risk index includes: The weights of each signal in the multimodal signal are determined through sensor quality assessment; The user's current state vector and state estimation covariance matrix are determined based on the weights of each signal in the multimodal signal. The state vector includes relative α power, β / α ratio, reciprocal of skin conductivity slope, instantaneous body temperature, pressure distribution entropy, and mean body electrical impedance. The comfort index, the fatigue index, and the risk index are calculated based on the user's current state vector and the state estimation covariance matrix.

5. The method according to claim 4, characterized in that, The comfort index is the difference between the result of the basic function operation of the state vector and the first penalty term. The first penalty term is the product of the preset comfort adjustment coefficient and the result of the first trace operation. The result of the first trace operation is the trace operation result of the product of the state estimation covariance matrix and the preset comfort weight. And / or, The fatigue index is the difference between the result of the basic function operation of the state vector and the second penalty term. The second penalty term is the product of the preset fatigue adjustment coefficient and the result of the second trace operation. The result of the second trace operation is the trace operation result of the product of the state estimation covariance matrix and the preset fatigue weight. And / or, The risk index is the difference between the result of the basic function operation of the state vector and the third penalty term. The third penalty term is the product of the preset risk adjustment coefficient and the result of the third trace operation. The result of the third trace operation is the trace operation result of the product of the state estimation covariance matrix and the preset risk weight.

6. The method according to claim 1, characterized in that, The method further includes: Acquire one or a combination of the following during the massage operation: electroencephalogram (EEG) signals, electrodermal signals, skin temperature, body impedance signals, pressure array signals, facial expression images, and user commands. The massage parameters are dynamically adjusted based on one or a combination of the following: electroencephalogram (EEG) signals, electrodermal signals, skin temperature, body impedance signals, pressure array signals, facial expressions, and user commands during the massage operation.

7. A control device for a massage robot, characterized in that, The control device for the massage robot includes: The signal acquisition unit is used to acquire the user's multimodal signals, including electroencephalogram (EEG) signals, electrodermal (ED) signals, skin temperature, body impedance signals, and pressure array signals. The fusion unit is used to fuse the multimodal signals to obtain the user's fatigue index, comfort index, and risk index, and to determine the user's massage path, target area, and massage parameters based on the fatigue index, comfort index, and risk index. A massage operation unit is used to generate massage operation instructions to control the robotic arm of the massage robot to move the execution end of the massage robot to the target area according to the massage path and to control the execution end to perform massage operation according to the massage parameters; The adaptive threshold safety detection unit is used to perform adaptive threshold safety detection on EEG signals, skin conductance signals, skin temperature, body impedance signals, and pressure array signals in the multimodal signals respectively. If the number of signals passing the adaptive threshold safety detection in the multimodal signals exceeds a first preset threshold, the current processing flow or massage operation of the massage robot continues, while reducing the massage intensity of the current massage operation. If the number of signals passing the adaptive threshold safety detection in the multimodal signals does not exceed the first preset threshold, an alarm is issued, and the current processing flow or massage operation of the massage robot is terminated.

8. A massage robot, characterized in that, The massage robot includes: a pressure sensor array, an electroencephalogram (EEG) sensor, a skin conductance sensor, a body temperature sensor, a bioimpedance sensor, a control unit, a robotic arm, and an actuator. The pressure sensor array, EEG sensor, skin conductance sensor, body temperature sensor, bioimpedance sensor, robotic arm, and actuator are respectively connected to the control unit. The control unit includes one or more processors and a memory storing programs. The programs include instructions that, when executed by the processor, cause the processor to perform the method as described in any one of claims 1 to 6.

9. A computer-readable storage medium storing a program, the program comprising instructions that, when executed by one or more processors of a computing device, cause the computing device to perform the method as claimed in any one of claims 1 to 6.

Images