A massage physiotherapy robot safety detection method and system based on heart rate monitoring

Abstract

This invention discloses a safety detection method and system for massage therapy robots based on heart rate monitoring, belonging to the field of safety detection technology for therapy robots. Before the first therapy service, a three-dimensional point cloud of the therapist's body surface is acquired through a visual perception module, an individualized three-dimensional mesh model is reconstructed, and the therapy area is labeled. At the start of therapy, the electronic tag of the end effector is read to identify the therapy method, and the corresponding anomaly judgment threshold and response strategy are loaded. During therapy, heart rate signals are collected in real time, and the heart rate change rate and heart rate variability parameters are calculated to determine if abnormal heart rate fluctuations occur. When an anomaly is detected, a multi-indicator fusion risk score is calculated and a graded safety response is triggered. When the response is triggered, the action point of the end effector is back-calculated to the individualized mesh model to determine the corresponding human body area identifier. This invention achieves a deep integration of heart rate physiological perception, individualized human body area mapping, and closed-loop control for therapy safety, significantly improving the safety and personalized experience of massage therapy.

Description

Technical Field

[0001] This invention belongs to the field of safety testing technology for physiotherapy robots, specifically a safety testing method and system for massage physiotherapy robots based on heart rate monitoring. Background Technology

[0002] With the widespread application of robotics technology in the field of health and wellness therapy, massage therapy robots are gradually entering medical institutions, community health and wellness centers, and home healthcare settings. Existing massage robots can perform various therapeutic functions such as electromagnetic stimulation, shockwaves, moxibustion, and vibration massage. However, the following safety hazards exist during the therapy process: excessive therapeutic force causing pain and discomfort to the user; electromagnetic stimulation intensity exceeding the user's tolerance range; excessively high moxibustion temperatures causing burns; and the inability to promptly communicate sudden physical discomfort to the user.

[0003] Existing safety solutions largely rely on mechanical force sensor feedback or visual distance detection, lacking real-time monitoring of the user's physiological state, especially the autonomic nervous system's response. When a user experiences stress during physiotherapy, their heart rate exhibits characteristic fluctuations, a physiological indicator that can serve as an effective basis for safety monitoring. However, traditional heart rate monitoring methods often use wristbands or finger clips, which suffer from inconvenience in wearing, significant motion artifact interference, and the potential for allergies caused by contact electrodes. Furthermore, different physiotherapy methods have fundamentally different physiological mechanisms affecting the human body, making it difficult to effectively adapt to the diverse needs of physiotherapy scenarios using a uniform, fixed threshold for safety assessment.

[0004] Another key shortcoming is that existing systems, when recording user discomfort events, typically only record the coordinates of the robotic arm's end effector in Cartesian or joint space, lacking the ability to reverse-map the robotic arm's position to the user's individualized body regions. Because the user's lying posture and body position vary during each treatment, relying solely on the robotic arm's coordinates cannot accurately reproduce treatment points such as "Jianjing" and "Jiaji" acupoints on the lower back. This makes it difficult to accurately utilize optimized parameters in subsequent treatments, limiting the continuous improvement of treatment effectiveness and safety.

[0005] Therefore, a safety testing method for massage therapy robots is needed that can flexibly adapt to various heart rate acquisition methods, be compatible with different physiotherapy techniques, and accurately correlate physiological feedback with individualized human body areas. Summary of the Invention

[0006] To solve the above problems, the technical solution provided by the present invention is as follows:

[0007] The present invention provides a safety detection method for a massage therapy robot based on heart rate monitoring, comprising the following steps:

[0008] S1. Before the first physiotherapy service, the three-dimensional point cloud of the physiotherapy patient's body surface is obtained through the visual perception module, an individualized three-dimensional mesh model is reconstructed, and at least one physiotherapy human body area is marked on the model.

[0009] S2. When physiotherapy begins, the current physiotherapy mode is identified by reading the electronic tag of the physiotherapy end effector, and the heart rate abnormality judgment threshold and response strategy associated with the physiotherapy mode are loaded from the preset physiotherapy mode-parameter configuration library;

[0010] S3. Equip physiotherapy patients with heart rate monitoring devices to collect heart rate data at rest and calculate baseline heart rate HR_baseline and baseline parameters of heart rate variability;

[0011] S4. During the physiotherapy process, heart rate signals are collected in real time, and the heart rate change rate ΔHR = HR_mean - HR_baseline and heart rate variability parameters are continuously calculated using a sliding time window. The abnormal heart rate fluctuations are determined based on the loaded threshold.

[0012] S5. When abnormal fluctuations are detected, calculate the multi-indicator fusion risk score S_danger, and trigger a graded safety response based on the score and the response strategy corresponding to the current physiotherapy method.

[0013] S6. When a safety response is triggered, obtain the contact position of the therapeutic end point on the human body surface, and calculate the position back onto the individualized three-dimensional mesh model through the calibrated spatial transformation relationship to determine the corresponding therapeutic human body region identifier r_j;

[0014] S7. Record the human body region identifier r_j and the current physiotherapy parameters, calculate the optimized physiotherapy parameters and store them in association with the region identifier;

[0015] S8. When the next physiotherapy is applied to the same human body area, the system will register the human body area identifier with the individualized model based on the surface point cloud under the current posture, map the position corresponding to the human body area identifier to the current body surface, and automatically call the optimized physiotherapy parameters.

[0016] Preferably, the determination in step S4 includes:

[0017] When ΔHR exceeds the first threshold Th_1 and the LF / HF ratio exceeds the threshold Th_LFHF, it is determined to be acute discomfort;

[0018] When ΔHR exceeds the second threshold Th_2 for N consecutive sliding windows, and the SDNN value decreases by more than 20% from the baseline, it is determined to be a continuous stress.

[0019] When the instantaneous heart rate exceeds 90% of the preset maximum safe heart rate, it is determined to be an abnormal emergency and the highest level of safety response is immediately triggered.

[0020] Preferably, the calculation formula for the multi-index fusion risk score in step S5 is as follows:

[0021] S_danger = w1·f(ΔHR / Th1) + w2·f(LF / HF / Th_LFHF) + w3·f(P_pain / 0.5) + w4·f((SDNN_baseline-SDNN) / (0.2·SDNN_baseline));

[0022] Where f(x) = min(max(x, 0), 1), w1 + w2 + w3 + w4 = 1, and P_pain is the pain probability estimate based on binary logistic regression.

[0023] Preferably, the formula for calculating the pain probability estimate P_pain is:

[0024] P_pain = 1 / (1 + e^{-z}), z = β0+ β1·HR_mean + β2·(LF / HF) + β3·SDNN+ β4·RMSSD;

[0025] Where β0, β1, β2, β3, and β4 are regression coefficients obtained by fitting the training samples.

[0026] Preferably, the calculation formula for the optimized physiotherapy parameters in step S7 is:

[0027] P_j^opt = P_j · (1-α · S_danger);

[0028] Where P_j is the physiotherapy parameter vector when the safety response is triggered, and α is the optimization coefficient.

[0029] Preferably, the registration in step S8 adopts a non-rigid point cloud registration algorithm to obtain the deformation field Φ of the current pose point cloud relative to the individualized three-dimensional mesh model. The set of positions mapped from the human body region identifier r_j to the current body surface is Φ(r_j). The robotic arm moves to the corresponding physiotherapy area according to this position.

[0030] Preferably, the physiotherapy methods include electromagnetic stimulation, shock waves, moxibustion, and vibration massage, with different physiotherapy methods corresponding to different Th1, Th2, Th_LFHF, and N values.

[0031] Preferably, the first threshold for shockwave therapy is set to +20 bpm, the first threshold for moxibustion therapy is set to +12 bpm, and the number of consecutive windows N for moxibustion therapy is set to 4.

[0032] Preferably, in the graded safety response, a mild response reduces the physiotherapy intensity to 80% of the current value, a moderate response reduces it to 50% and sends an early warning to the technician's terminal, and an emergency response immediately stops the physiotherapy and retracts the robotic arm to a safe position before sending an alarm to the technician's terminal. Furthermore, the intensity coefficients of each response are dynamically adjusted according to the physiotherapy method.

[0033] Preferably, the heart rate monitoring device in step S3 is at least one of the following: a wrist-worn photoplethysmography (PPG) sensor, a chest patch electrocardiogram (ECG) electrode, a piezoelectric or capacitive micro-vibration sensor built into a physiotherapy bed, or a non-contact sensor based on ultra-wideband / millimeter-wave radar.

[0034] A safety detection system for a massage therapy robot based on heart rate monitoring, comprising:

[0035] The visual perception module is used to acquire the three-dimensional point cloud of the body surface of the physiotherapist before the first service, reconstruct an individualized three-dimensional mesh model and label the physiotherapist body area;

[0036] Heart rate monitoring device, used to collect users' heart rate data in real time;

[0037] The massage therapy robot body includes a multi-axis robotic arm, a quick-change therapy end effector, a force / torque sensor, and a speaker. Each therapy end effector has a built-in electronic tag that stores the code of the therapy method type.

[0038] The central control unit includes an HRV analysis engine, a physiotherapy method recognition and threshold matching engine, a safety decision-maker, a human body modeling and position calculation module, and a parameter optimization and recording module, which are used for the above methods;

[0039] The cloud-based data platform is used to store user data, individualized 3D mesh models, multi-mode threshold configuration, and human body region optimization parameter records.

[0040] Technician terminal, used to receive security incident alarms and support remote intervention.

[0041] Preferably, when the human body modeling and position inverse calculation module triggers a safety response, it obtains the contact position of the therapeutic end point and uses the hand-eye calibration matrix and the individualized three-dimensional mesh model to map the contact position to the corresponding human body region identifier, in order to replace the simple record of the robotic arm joint angle or Cartesian coordinates.

[0042] Preferably, the safety determiner performs a graded response based on the danger score S_danger: a mild response is triggered when 0.6 ≤ S_danger < 0.8, a moderate response is triggered when 0.8 ≤ S_danger < 0.95, and an emergency response is triggered when S_danger ≥ 0.95.

[0043] Preferably, the central control unit has a communication timeout detection function, which triggers a protective stop when the heart rate monitoring device loses data continuously for more than a preset time.

[0044] Preferably, the parameter optimization recording module uses a recursive optimization formula for identifying the same human body region:

[0045] P_j^opt,(k+1) = P_j^opt,(k) · (1-α· S_danger^(k));

[0046] Perform iterative optimization until the hazard score is below 0.6 or the parameters reach the preset safety lower limit.

[0047] Beneficial effects

[0048] Compared with the prior art, the technical solution provided by this invention has the following advantages:

[0049] This invention discloses a safety detection method and system for a massage therapy robot based on heart rate monitoring. Before the first therapy service, a three-dimensional point cloud of the therapist's body surface is acquired through a visual perception module, an individualized three-dimensional mesh model is reconstructed, and the therapy area is labeled. At the start of therapy, the electronic tag of the end effector is read to identify the therapy mode, and the corresponding anomaly judgment threshold and response strategy are loaded. During therapy, heart rate signals are collected in real time, and the heart rate change rate and heart rate variability parameters are calculated to determine whether abnormal heart rate fluctuations occur. When an anomaly is detected, a multi-index fusion risk score is calculated and a graded safety response is triggered. When the response is triggered, the action point of the end effector is back-calculated to the individualized mesh model to determine the corresponding human body area identifier. The area identifier and the current therapy parameters are recorded, optimized parameters are calculated, and stored in association. When the same area is treated again, the area identifier is mapped to the current body surface through registration of the body surface point cloud and the model, and the optimized parameters are automatically called. This invention achieves a deep integration of heart rate physiological perception, individualized human body area mapping, and closed-loop control of therapy safety, significantly improving the safety and personalized experience of massage therapy. Attached Figure Description

[0050] Figure 1 This is a schematic diagram of the overall structure of a massage therapy robot safety detection system based on heart rate monitoring according to the present invention.

[0051] Figure 2 This is a flowchart illustrating step S1 in Example 1;

[0052] Figure 3 This is a diagram showing the mapping relationship between the adaptive threshold configuration of the physiotherapy method and the response action in step S2 of Example 1;

[0053] Figure 4 This is a detailed diagram of the heartbeat abnormality detection and safety response technology in step S4 of Example 1. Detailed Implementation

[0054] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0055] It should be noted that the terms "first," "second," etc., in the specification, claims, and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate for the embodiments of this application described herein. Furthermore, the terms "comprising" and "having," and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0056] In this application, the terms "upper," "lower," "left," "right," "front," "rear," "top," "bottom," "inner," "outer," "middle," "vertical," "horizontal," "lateral," and "longitudinal" indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. These terms are primarily for the purpose of better describing this application and its embodiments, and are not intended to limit the indicated device, element, or component to having a specific orientation, or to be constructed and operated in a specific orientation.

[0057] Furthermore, in addition to indicating location or positional relationship, some of the aforementioned terms may also have other meanings. For example, the term "above" may also be used in some cases to indicate a certain dependency or connection relationship. Those skilled in the art can understand the specific meaning of these terms in this application based on the specific circumstances.

[0058] Furthermore, the terms "installation," "setup," "equipped with," "connection," "linking," and "socketing" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral structure; they can refer to a mechanical connection or an electrical connection; they can refer to a direct connection or an indirect connection through an intermediate medium, or an internal connection between two devices, components, or parts. Those skilled in the art can understand the specific meaning of these terms in this application based on the specific circumstances.

[0059] It should be noted that, unless otherwise specified, the embodiments and features described in this application can be combined with each other. This application will now be described in detail with reference to the accompanying drawings and embodiments.

[0060] Example 1

[0061] See attached document Figure 2-4 This embodiment of the safety detection method for a massage therapy robot based on heart rate monitoring includes the following steps:

[0062] S1. Before the first physiotherapy service, the three-dimensional point cloud of the physiotherapy patient's body surface is obtained through the visual perception module, an individualized three-dimensional mesh model is reconstructed, and at least one physiotherapy body area is marked on the model.

[0063] Specifically, one or more depth cameras (such as Intel RealSense D435) are installed above the physiotherapy bed to acquire depth and color images of the user in a supine or prone position at a rate of 30 frames per second. After each frame is denoised and filtered, the multi-view point cloud data is fused into a complete body surface point cloud using a point cloud stitching algorithm. Subsequently, the Poisson surface reconstruction algorithm is used to reconstruct the body surface point cloud into a mesh, generating an individualized 3D mesh model M=(V,E,F), where V is the vertex set, E is the edge set, and F is the triangular facet set. The model contains approximately 100,000 to 300,000 triangular facets. After the model is established, the system extracts key joint points based on the human skeleton detection algorithm, derives the spatial coordinates of commonly used physiotherapy acupoints / regions according to the rules of traditional Chinese medicine meridians or anatomy, and labels the physiotherapy human body regions such as "left Jianjing acupoint region", "right Jianjing acupoint region", "Mingmen acupoint region", and "Yaoyangguan acupoint region" on the model, and obtains the set of physiotherapy regions R={r1,r2,…,r_K}. Each region r_k contains its triangular facet index and region name label on the individualized model M.

[0064] During physiotherapy, a calibrated depth camera is mounted on the end effector of the robotic arm, or a vision perception module is used to continuously acquire point clouds of the patient's body surface. When a safety calibrator triggers a heart rate abnormality event, the system records the current Cartesian pose of the robotic arm's end effector in the robot's base coordinate system: X_e = (p_e, R_e). Subsequently, the system uses the hand-eye calibration matrix and the robotic arm's kinematic model to transform the end effector pose to the camera coordinate system or directly calculates the contact position p_contact of the end effector's point of action on the human body surface.

[0065] The system projects p_contact onto the individualized human body mesh model M and finds the semantic label of its corresponding triangle or nearest vertex. If the contact point falls within a labeled region r_k (by checking if the spatial distance between the contact point and the region's envelope is less than a threshold δ, defaulting to 2cm), the current event is associated with region r_k; otherwise, a temporary region identifier is created for the contact point. Ultimately, the system stores the human body region identifier r_k, rather than simply the robotic arm joint angle or Cartesian pose.

[0066] When a user undergoes subsequent physical therapy, their lying posture may differ from that during modeling. In this case, the system acquires a point cloud of the current posture using a visual perception module and performs non-rigid registration with the individualized model M to obtain the deformed current mesh model M'. Based on this deformation field, the stored human body region identifier r_k is mapped to the corresponding body surface position in the current posture, thereby guiding the robotic arm to move to the correct physical therapy area and achieving robust reuse of "region memory".

[0067] S2. When physiotherapy begins, the current physiotherapy mode is identified by reading the electronic tag of the physiotherapy end effector, and the heart rate abnormality judgment threshold and response strategy associated with the physiotherapy mode are loaded from the preset physiotherapy mode-parameter configuration library.

[0068] Specifically, each end effector (including electromagnetic stimulation module, shockwave module, moxibustion module, and vibration massage module) has a built-in passive RFID electronic tag, which stores the physiotherapy mode type code. An RFID reader / writer is installed at the end of the robotic arm; once the end effector is in place, the RFID reader / writer automatically reads the code information from the tag. The central control unit loads the corresponding threshold configuration from a preset physiotherapy mode-parameter configuration library based on the read physiotherapy mode code, including the first threshold Th1, the second threshold Th2, the LF / HF threshold Th_LFHF, and the number of continuous windows N. For example, the first threshold for shockwave physiotherapy is set to +20 bpm, the first threshold for moxibustion physiotherapy is set to +12 bpm, and the number of continuous windows N for moxibustion physiotherapy is set to 4.

[0069] S3. Configure a heart rate monitoring device for physiotherapists to collect heart rate data at rest and calculate the baseline heart rate HR_baseline and the baseline parameters of heart rate variability.

[0070] Specifically, the heart rate monitoring device is at least one of the following: a wrist-worn photoplethysmography (PPG) sensor, a chest patch ECG electrode, a piezoelectric or capacitive micro-vibration sensor built into the physiotherapy bed, or a non-contact sensor based on ultra-wideband / millimeter-wave radar. Before the start of physiotherapy, heart rate data is collected for 2 minutes at rest. For PPG / ECG signals, bandpass filtering (cutoff frequency 0.4Hz~4.0Hz) is performed to remove respiratory drift and high-frequency noise, and the R-peak or pulse peak is detected to extract successive heartbeat intervals. For BCG or radar signals, an adaptive template matching algorithm is used to extract heartbeat intervals, and motion artifacts are eliminated using acceleration or body motion signals. Based on the extracted RR interval sequence, parameters such as baseline heart rate HR_baseline, baseline SDNN (SDNN_baseline), baseline RMSSD, and baseline LF / HF ratio are calculated.

[0071] S4. During the physiotherapy process, heart rate signals are collected in real time, and the heart rate change rate ΔHR=HR_mean-HR_baseline and heart rate variability parameters are continuously calculated using a sliding time window. The abnormal heart rate fluctuations are determined based on the loaded threshold.

[0072] Specifically, a sliding time window with a length of 60 seconds and a step size of 10 seconds is used. Within each window, the mean heart rate (HR_mean), heart rate variability (ΔHR), SDNN, RMSSD, LF power, HF power, and LF / HF ratio are calculated. The judgment includes the following three cases:

[0073] (1) When ΔHR exceeds the first threshold Th1 and the LF / HF ratio exceeds the threshold Th_LFHF, it is determined to be acute discomfort;

[0074] (2) When ΔHR exceeds the second threshold Th2 in N consecutive sliding windows, and the SDNN value decreases by more than 20% from the baseline, it is determined to be a continuous stress.

[0075] (3) When the instantaneous heart rate exceeds 90% of the preset maximum safe heart rate HR_max, it is determined to be an abnormal sudden event and the highest level of safety response is immediately triggered.

[0076] S5. When abnormal fluctuations are detected, calculate the multi-indicator fusion risk score S_danger, and trigger a graded safety response based on the score and the response strategy corresponding to the current physiotherapy method.

[0077] Specifically, the formula for calculating the multi-indicator fusion hazard score is as follows:

[0078] S_danger = w1·f(ΔHR / Th1) + w2·f(LF / HF / Th_LFHF) + w3·f(P_pain / 0.5) + w4·f((SDNN_baseline − SDNN) / (0.2·SDNN_baseline));

[0079] Where f(x) = min(max(x,0),1), w1 + w2 + w3 + w4 = 1, and the default weights are w1 = 0.30, w2 = 0.25, w3 = 0.25, and w4 = 0.20 respectively. P_pain is the pain probability estimate based on binary logistic regression, and the calculation formula is P_pain = 1 / (1 + e^(−z)), z = β0 + β1·HR_mean + β2·(LF / HF) + β3·SDNN + β4·RMSSD, where β0, β1, β2, β3, and β4 are the regression coefficients obtained by fitting the training samples. In the graded safety response, a mild response is triggered when 0.6≤S_danger<0.8 (reducing the physiotherapy intensity to 80% of the current value), a moderate response is triggered when 0.8≤S_danger<0.95 (reducing to 50% and sending an alert to the technician's terminal), and an emergency response is triggered when S_danger≥0.95 (immediately stopping the physiotherapy and retracting the robotic arm to a safe position before sending an alarm to the technician's terminal). The intensity coefficients of each response are dynamically adjusted according to the physiotherapy method.

[0080] S6. When a safety response is triggered, obtain the contact position of the therapeutic end point on the human body surface, and calculate the position back onto the individualized three-dimensional mesh model through the calibrated spatial transformation relationship to determine the corresponding therapeutic human body region identifier r_j.

[0081] Specifically, a calibrated depth camera is mounted at the end effector of the robotic arm. When a heart rate abnormality event is triggered by the safety detector, the system records the current Cartesian space pose of the robotic arm's end effector in the robot's base coordinate system, X_e=(p_e,R_e). Using the hand-eye calibration matrix and the robotic arm's kinematic model, the end effector pose is transformed to the camera coordinate system, and the contact position p_contact of the therapeutic end effector on the human body surface is calculated. The hand-eye calibration uses the Tsai algorithm, which transforms the robotic arm pose so that the depth camera can capture images of the checkerboard calibration board from multiple angles, and solves for the homogeneous transformation matrix from the camera coordinate system to the robotic arm's base coordinate system. p_contact is projected onto the individualized human body mesh model M, and the semantic label of the triangular facet to which it belongs is found, thus determining the corresponding human body region identifier r_j.

[0082] S7. Record the human body region identifier r_j and the current physiotherapy parameters, calculate the optimized physiotherapy parameters and store them in association with the region identifier.

[0083] Specifically, for the marked human body region r_j, an initial physiotherapy intensity parameter vector P_j=[F_j,I_j,T_j]^T is set, where F_j is the contact force, I_j is the electromagnetic stimulation current intensity, and T_j is the moxibustion temperature. The optimized parameter vector is P_j^opt=P_j·(1-α·S_danger), where α is the optimization coefficient (default is 0.6). The optimized parameters must satisfy P_j^opt≥P_min (preset safety lower limit) and not exceed the upper limit of the device's physical parameters. The optimized parameters P_j^opt are associated with the human body region identifier r_j and stored in the user's personal physiotherapy parameter database.

[0084] S8. When the next physiotherapy is applied to the same human body area, the system will register the human body area identifier with the individualized model based on the surface point cloud under the current posture, map the position corresponding to the human body area identifier to the current body surface, and automatically call the optimized physiotherapy parameters.

[0085] Specifically, the system acquires the point cloud in the current pose through a visual perception module and performs non-rigid point cloud registration with the individualized model M. A non-rigid registration algorithm based on embedded deformation maps is used to obtain the deformation field Φ of the current pose point cloud relative to the individualized 3D mesh model. The set of positions mapped from the human body region identifier r_j to the current body surface is Φ(r_j), which is the set of positions obtained after applying the deformation field Φ to the triangular facets or vertex set contained in region r_j. The robotic arm plans a path based on this position and moves to the corresponding physiotherapy area, automatically calling the optimized physiotherapy parameters P_j^opt as the initial physiotherapy parameters.

[0086] The central control unit has a communication timeout detection function. When the heart rate monitoring device loses data continuously for more than a preset time (default 5 seconds), it triggers a protective stop, immediately suspends the movement of the robotic arm, and prompts the user to check the heart rate monitoring device.

[0087] Example 2

[0088] See attached document Figure 1 This embodiment of a massage therapy robot safety detection system based on heart rate monitoring includes a visual perception module, a heart rate monitoring device, a massage therapy robot body, a central control unit, a cloud data platform, and a technician terminal.

[0089] The visual perception module is used to acquire 3D point clouds of the patient's body surface before the first service, reconstruct an individualized 3D mesh model, and label the treatment areas. This module includes one or more depth cameras (such as Intel RealSense D435), which are fixedly mounted above the treatment bed or near the robotic arm base, acquiring depth and color images at a rate of 30 frames per second and transmitting the data to the central control unit via a USB 3.0 interface.

[0090] The heart rate monitoring device is used to collect the user's heart rate data in real time. It is at least one of the following: a wrist-worn photoplethysmography (PPG) sensor, a chest patch ECG electrode, a piezoelectric or capacitive micro-vibration sensor built into the therapy bed, or a non-contact sensor based on ultra-wideband / millimeter-wave radar. Each heart rate monitoring device is equipped with a communication module to send the raw signal or calculated heart rate data to the central control unit via Bluetooth, WiFi, or ZigBee.

[0091] The massage therapy robot consists of a multi-axis robotic arm, a quick-change end effector, force / torque sensors, and a speaker. The multi-axis robotic arm employs a six-axis collaborative design (such as the UR10e) with a repeatability accuracy of ±0.1mm. The quick-change end effector supports rapid replacement of electromagnetic stimulation, shockwave, moxibustion, and vibration massage modules; each end effector has a built-in RFID tag storing the treatment method type code. A force / torque sensor, using a six-dimensional force / torque sensor (such as the ATI Mini45) and a 1kHz sampling rate, is installed between the robotic arm's end and the end effector to monitor contact force in real time. The speaker is mounted near the robotic arm base or the treatment bed to play soothing voice prompts.

[0092] The central control unit includes an HRV analysis engine, a physiotherapy method recognition and threshold matching engine, a safety decision-maker, a human body modeling and position inversion module, and a parameter optimization and recording module, used to execute the method described in Example 1. The HRV analysis engine performs bandpass filtering, R-peak detection, RR interval extraction, and HRV parameter calculation on the heart rate signal. The physiotherapy method recognition and threshold matching engine reads the end effector's electronic tag via an RFID reader and loads the corresponding threshold set from the configuration library. The safety decision-maker calculates a multi-index fusion hazard score S_danger; a mild response is triggered when 0.6 ≤ S_danger < 0.8, a moderate response when 0.8 ≤ S_danger < 0.95, and an emergency response when S_danger ≥ 0.95. When a safety response is triggered, the human body modeling and position inversion module obtains the contact position of the physiotherapy end effector's point of action, and uses a hand-eye calibration matrix and an individualized 3D mesh model to map the contact position to the corresponding human body region identifier, replacing simple robotic arm joint angle or Cartesian coordinate recording. The parameter optimization and recording module uses the recursive optimization formula P_j^opt,(k+1)=P_j^opt,(k)·(1−α·S_danger^(k)) to iteratively optimize the same human body region identifier until the danger score is lower than 0.6 or the parameter reaches the preset safety lower limit. The central control unit has a communication timeout detection function, which triggers a protective stop when the heart rate monitoring device loses data continuously for more than a preset time.

[0093] The cloud-based data platform stores user data, individualized 3D mesh models, multi-method threshold configurations, and human body region optimization parameter records. The platform is deployed on cloud servers, equipped with relational and non-relational databases, and supports cross-device data synchronization.

[0094] The technician terminal is used to receive security incident alarms and support remote intervention. The technician terminal is a tablet or smartphone with a dedicated physiotherapy monitoring app installed, communicating with the central control unit and cloud data platform via WiFi / 4G / 5G networks. When an emergency response is triggered, the system pushes an alarm data packet to the technician terminal via the MQTT protocol (QoS=2, encrypted transmission), containing basic user information, heart rate data curves, current physiotherapy parameters, body area markers, hazard scores, and suggested treatment measures. Technicians can remotely confirm alarms, view real-time physiotherapy data, and issue commands through the terminal.

[0095] Compared with the prior art, this application has the following advantages:

[0096] 1. This invention achieves dynamic monitoring of the user's autonomic nervous system response by collecting heart rate and heart rate variability parameters in real time during physiotherapy. It can capture characteristic heart rate fluctuations in the early stages when the user experiences stress response due to physiotherapy discomfort, making up for the shortcomings of traditional force sensor feedback or visual distance detection in that they cannot perceive the user's physiological state, and significantly improving the safety of physiotherapy.

[0097] 2. This invention provides a heart rate monitoring device for physiotherapy patients. This device can be at least one of the following: a wrist-worn photoplethysmography sensor, a chest patch electrocardiogram electrode, a piezoelectric or capacitive micro-vibration sensor built into the physiotherapy bed, or a non-contact sensor based on ultra-wideband / millimeter-wave radar. This enables flexible adaptation to various heart rate acquisition methods, adapts to different physiotherapy scenarios and user preferences, and reduces wearing discomfort and motion artifact interference.

[0098] 3. This invention identifies the current physiotherapy mode by reading the electronic tag of the physiotherapy end effector, and loads the heart rate abnormality judgment threshold and response strategy associated with the physiotherapy mode from the preset physiotherapy mode-parameter configuration library. Different physiotherapy modes (electromagnetic stimulation, shock wave, moxibustion, vibration massage) correspond to different Th1, Th2, Th_LFHF and N values, which effectively solves the problem that a unified fixed threshold cannot adapt to the needs of physiotherapy in multiple scenarios.

[0099] 4. This invention integrates four physiological indicators—heart rate change rate, LF / HF ratio, pain probability estimate, and SDNN decrease rate—into a multi-indicator risk score, constructing a comprehensive risk assessment system. Through a graded safety response mechanism (mild response reduced to 80%, moderate response reduced to 50% with warning, and emergency response immediately stopped with alarm), it achieves differentiated and refined safety intervention, balancing safety and the continuity of physical therapy.

[0100] 5. When a safety response is triggered, this invention back-calculates the point of action of the physiotherapy end to the individualized three-dimensional mesh model through the calibrated spatial transformation relationship, and determines the corresponding physiotherapy human body area identifier. This replaces the traditional method of recording the joint angle of the robotic arm or the Cartesian coordinate, overcomes the position deviation problem caused by changes in the user's lying posture, and achieves accurate position memory "centered on the human body area".

[0101] 6. This invention associates and stores the optimized physiotherapy parameters with human body region identifiers. When the same human body region is treated again, the region identifier is mapped to the current body surface through a non-rigid point cloud registration algorithm and the optimized parameters are automatically called. This realizes the continuous optimization and iteration of physiotherapy parameters and the reuse of experience, and improves the ability to continuously improve the physiotherapy effect and safety.

[0102] 7. The central control unit of this invention has a communication timeout detection function. When the data of the heart rate monitoring device is continuously lost for more than a preset time, a protective stop is triggered, avoiding blind spots in safety monitoring caused by signal loss and further ensuring the safety of the physiotherapy process.

[0103] 8. This invention uses a parameter optimization recording module to iteratively optimize the identification of the same human body area using a recursive optimization formula until the risk score is lower than the safety threshold or the parameter reaches the preset safety lower limit. This achieves personalized adaptive adjustment of physiotherapy parameters. As the number of uses increases, the system's physiotherapy adaptability to the user continues to improve.

[0104] 9. This invention stores user data, individualized three-dimensional mesh models, multi-mode threshold configuration, and human body area optimization parameter records through a cloud data platform, supports cross-device data synchronization, and allows users to access historical optimization parameters when receiving services in different physiotherapy institutions or on different physiotherapy robots, thus expanding the system's application scenarios and long-term service capabilities.

[0105] 10. This invention has a clear structure and a high degree of automation. The technician terminal supports remote alarm reception and intervention, realizing a human-machine collaborative safety management mode. It can be widely used in physiotherapy scenarios such as medical institutions, community health care, and home healthcare.

[0106] The above-described embodiments are merely illustrative of certain implementations of the present invention, and are described in a relatively specific and detailed manner. However, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements are all within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the appended claims.

Claims

1. A safety detection method for a massage therapy robot based on heart rate monitoring, characterized in that: Includes the following steps: S1. Before the first physiotherapy service, the three-dimensional point cloud of the physiotherapy patient's body surface is obtained through the visual perception module, an individualized three-dimensional mesh model is reconstructed, and at least one physiotherapy human body area is marked on the model. S2. When physiotherapy begins, the current physiotherapy mode is identified by reading the electronic tag of the physiotherapy end effector, and the heart rate abnormality judgment threshold and response strategy associated with the physiotherapy mode are loaded from the preset physiotherapy mode-parameter configuration library; S3. Equip physiotherapy patients with heart rate monitoring devices to collect heart rate data at rest and calculate baseline heart rate HR_baseline and baseline parameters of heart rate variability; S4. During the physiotherapy process, heart rate signals are collected in real time, and the heart rate change rate ΔHR = HR_mean - HR_baseline and heart rate variability parameters are continuously calculated using a sliding time window. The abnormal heart rate fluctuations are determined based on the loaded threshold. S5. When abnormal fluctuations are detected, calculate the multi-indicator fusion risk score S_danger, and trigger a graded safety response based on the score and the response strategy corresponding to the current physiotherapy method. S6. When a safety response is triggered, obtain the contact position of the therapeutic end point on the human body surface, and calculate the position back onto the individualized three-dimensional mesh model through the calibrated spatial transformation relationship to determine the corresponding therapeutic human body region identifier r_j; S7. Record the human body region identifier r_j and the current physiotherapy parameters, calculate the optimized physiotherapy parameters and store them in association with the region identifier; S8. When the next physiotherapy is applied to the same human body area, the system will register the human body area identifier with the individualized model based on the surface point cloud under the current posture, map the position corresponding to the human body area identifier to the current body surface, and automatically call the optimized physiotherapy parameters.

2. The safety detection method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that, The determination in step S4 includes: When ΔHR exceeds the first threshold Th_1 and the LF / HF ratio exceeds the threshold Th_LFHF, it is determined to be acute discomfort; When ΔHR exceeds the second threshold Th_2 for N consecutive sliding windows, and the SDNN value decreases by more than 20% from the baseline, it is determined to be a continuous stress. When the instantaneous heart rate exceeds 90% of the preset maximum safe heart rate, it is determined to be an abnormal emergency and the highest level of safety response is immediately triggered.

3. The safety detection method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that, The calculation formula for the multi-indicator fusion risk score mentioned in step S5 is as follows: S_danger = w1·f(ΔHR / Th1) + w2·f(LF / HF / Th_LFHF) + w3·f(P_pain / 0.5) +w4·f((SDNN_baseline − SDNN) / (0.2·SDNN_baseline)); Where f(x) = min(max(x, 0), 1), w1 + w2 + w3 + w4 = 1, and P_pain is the pain probability estimate based on binary logistic regression.

4. The method according to claim 3, characterized in that, The formula for calculating the pain probability estimate P_pain is as follows: P_pain = 1 / (1 + e^{-z}), z = β0 + β1·HR_mean + β2·(LF / HF) + β3·SDNN +β4·RMSSD; Where β0, β1, β2, β3, and β4 are regression coefficients obtained by fitting the training samples.

5. The safety detection method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that, The calculation formula for the optimized physiotherapy parameters mentioned in step S7 is as follows: P_j^opt = P_j · (1-α · S_danger); Where P_j is the physiotherapy parameter vector when the safety response is triggered, and α is the optimization coefficient.

6. The safety detection method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that, The registration in step S8 uses a non-rigid point cloud registration algorithm to obtain the deformation field Φ of the current pose point cloud relative to the individualized three-dimensional mesh model. The set of positions mapped from the human body region identifier r_j to the current body surface is Φ(r_j). The robotic arm moves to the corresponding physiotherapy area according to this position.

7. The safety detection method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that, The physiotherapy methods include electromagnetic stimulation, shock waves, moxibustion, and vibration massage. Different physiotherapy methods correspond to different Th1, Th2, Th_LFHF, and N values.

8. The safety detection method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that: The first threshold for shockwave therapy is set to +20 bpm, and the first threshold for moxibustion therapy is set to +12 bpm. The number of consecutive windows N for moxibustion therapy is set to 4.

9. A safety testing method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that: In the graded safety response, a mild response reduces the physiotherapy intensity to 80% of the current value, a moderate response reduces it to 50% and sends an early warning to the technician's terminal, and an emergency response immediately stops the physiotherapy and retracts the robotic arm to a safe position before sending an alarm to the technician's terminal. The intensity coefficients of each response are dynamically adjusted according to the physiotherapy method.

10. A safety detection method for a massage therapy robot based on heart rate monitoring according to claim 1, characterized in that: The heart rate monitoring device mentioned in step S3 is at least one of the following: a wrist-worn photoplethysmography (PPG) sensor, a chest patch electrocardiogram (ECG) electrode, a piezoelectric or capacitive micro-vibration sensor built into a physiotherapy bed, or a non-contact sensor based on ultra-wideband / millimeter-wave radar.

11. A safety detection system for a massage therapy robot based on heart rate monitoring, characterized in that, include: The visual perception module is used to acquire the three-dimensional point cloud of the body surface of the physiotherapist before the first service, reconstruct an individualized three-dimensional mesh model and label the physiotherapist body area; Heart rate monitoring device, used to collect users' heart rate data in real time; The massage therapy robot body includes a multi-axis robotic arm, a quick-change therapy end effector, a force / torque sensor, and a speaker. Each therapy end effector has a built-in electronic tag that stores the code of the therapy method type. The central control unit includes an HRV analysis engine, a physiotherapy method recognition and threshold matching engine, a safety decision-maker, a human body modeling and position calculation module, and a parameter optimization and recording module, for executing the method of any one of claims 1 to 10; The cloud-based data platform is used to store user data, individualized 3D mesh models, multi-mode threshold configuration, and human body region optimization parameter records. Technician terminal, used to receive security incident alarms and support remote intervention.

12. A safety detection system for a massage therapy robot based on heart rate monitoring according to claim 11, characterized in that, When the safety response is triggered, the human body modeling and position inverse calculation module obtains the contact position of the therapeutic end point and uses the hand-eye calibration matrix and the individualized three-dimensional mesh model to map the contact position to the corresponding human body area identifier, in order to replace the simple record of the robotic arm joint angle or Cartesian coordinates.

13. A safety detection system for a massage therapy robot based on heart rate monitoring according to claim 11, characterized in that, The safety decision-maker performs graded responses based on the hazard score S_danger: a mild response is triggered when 0.6 ≤ S_danger < 0.8, a moderate response is triggered when 0.8 ≤ S_danger < 0.95, and an emergency response is triggered when S_danger ≥ 0.

95.

14. A safety detection system for a massage therapy robot based on heart rate monitoring according to claim 11, characterized in that, The central control unit has a communication timeout detection function. When the heart rate monitoring device loses data continuously for more than a preset time, it triggers a protective stop.

15. A safety detection system for a massage therapy robot based on heart rate monitoring according to claim 11, characterized in that, The parameter optimization recording module uses a recursive optimization formula for the identification of the same human body region: P_j^opt,(k+1) = P_j^opt,(k) · (1-α · S_danger^(k)); Perform iterative optimization until the hazard score is below 0.6 or the parameters reach the preset safety lower limit.

Images