Abdominal acupoint intelligent massage system and method based on multi-mode perception

Abstract

The invention relates to the technical field of massage, in particular to an intelligent abdominal acupoint massage system and method based on multi-modal sensing, and the system comprises a multi-modal sensing module which comprises a 3D depth camera, an infrared thermal imager, a myoelectricity sensor and a millimeter wave radar; the intelligent decision-making module is used for updating a massage strategy in real time by applying a Poisson reconstruction algorithm, a multi-modal fusion positioning algorithm and a dynamic coordinate mapping method; the execution device comprises an end effector and is composed of a bionic massage head for executing massage, a six-dimensional force sensor for feeding back a pressure value in real time and an infrared temperature control module; the three-degree-of-freedom mechanical arm is composed of xyz translation compensation and a rotating joint and is used for controlling the end effector to move; the power module is used for supplying power to the system. The device has the advantages that accurate acupoint tracking and positioning can be provided for a user, safe, controllable, autonomously adjustable massage force is provided, and different massage schemes are set according to individuation of the physique of the user.

Description

[Technical Field]

[0001] This invention relates to the field of massage technology, and in particular to an intelligent abdominal acupoint massage system and method based on multimodal perception. [Background Technology]

[0002] With the rapid development of artificial intelligence technology and the continuous innovation and integration of traditional Chinese medicine concepts, traditional Chinese medicine is gradually moving towards intelligentization in disease prevention, health management, and precision treatment. In the practice of traditional Chinese medicine, the accurate location of acupoints is the key to the successful implementation of acupuncture, massage, and other therapies.

[0003] However, traditional acupoint location methods rely heavily on the physician's experience. Traditional manual massage suffers from strong subjectivity in acupoint location and inconsistent operational standards. Differences in patient body type, anatomical features, and physician skill levels often lead to insufficient accuracy and consistency in acupoint location. Furthermore, existing massage equipment lacks precise acupoint identification capabilities and dynamic adjustment of abdominal acupoint depth, thus affecting treatment efficacy and therapeutic evaluation.

[0004] The main methods currently used by mainstream massage robots in the industry are as follows: some use pressure sensors to locate acupoints, but this method cannot adapt to differences in body shape; some rely on pre-calibrated human models, which does not solve the problem of body position changes; and some have fixed mechanical trajectories, which cannot achieve dynamic massage along the meridians. [Summary of the Invention]

[0005] The technical problem to be solved by the present invention is to overcome the shortcomings of the prior art and provide an intelligent abdominal acupoint massage system and method based on multimodal perception, which can provide users with accurate acupoint tracking and positioning, provide safe, controllable and autonomously adjustable massage intensity, and set different massage programs based on the user's physical condition.

[0006] In order to solve the above technical problems, the present invention adopts the following technical solutions:

[0007] A multimodal perception-based intelligent abdominal acupoint massage system, including

[0008] The multimodal perception module includes a 3D depth camera for collecting abdominal point cloud data and constructing a three-dimensional topological map of the abdomen; an infrared thermal imager for locating acupoints by identifying temperature anomalies; an electromyography sensor for real-time monitoring of the tension of the rectus abdominis muscle and adaptive adjustment of massage intensity; and a millimeter-wave radar for real-time tracking of body surface positions caused by respiration and controlling the pace of massage.

[0009] The intelligent decision-making module uses the Poisson reconstruction algorithm, multimodal fusion positioning algorithm, and dynamic coordinate mapping method to update the massage strategy in real time;

[0010] The actuator includes an end effector consisting of a bionic massage head for performing massage, a six-dimensional force sensor for real-time pressure feedback, and an infrared temperature control module; and a three-degree-of-freedom robotic arm consisting of xyz translation compensation and rotary joints, used to control the movement of the end effector.

[0011] The power module supplies power to the lithium battery, providing stable power support for the entire system and ensuring its normal operation.

[0012] As a preferred option, the multimodal fusion localization algorithm includes abdominal surface point cloud data, acupoint temperature field data, and traditional meridian model data. The abdominal surface point cloud data is first processed using principal component analysis to calculate the acupoint normal vector. Then, a covariance matrix is ​​constructed within the neighborhood of each point p in the point cloud. The specific calculation formula is as follows: In the formula, k = 30; calculate the eigenvalues ​​and eigenvectors of the matrix, and the eigenvector corresponding to the smallest eigenvalue is the normal vector.

[0013] As a preferred approach, the curvature of the normal vector is estimated: the target point and its neighborhood are selected; a tangent plane is established; with the normal vector as the Z-axis, a vector that is not collinear with the normal vector is selected, and the local X-axis and Y-axis are determined by Schmitt orthogonalization; the neighborhood points are projected onto the tangent plane.

[0014] Preferably, for each neighborhood point P(x,y,z), after calculating its tangent plane coordinates, a quadratic surface is fitted to obtain the mean curvature, Gaussian curvature, and principal curvature. The specific calculation formula is as follows:

[0015] u i =Δp i ·x,υ i =Δp i ·y

[0016] ω i =au i 2 +bu i υ i +cυ i 2

[0017] H = a + c, K = 4ac - b 2 ,

[0018] In the formula, Δp is the relative coordinate of the point to the tangent plane, H is the mean curvature, and K is the Gaussian curvature.

[0019] As a preferred approach, the Poisson reconstruction algorithm proceeds sequentially by constructing an octree, calculating the normal vector divergence field, solving the Poisson equation, and extracting isosurfaces, ultimately outputting a triangular mesh.

[0020] As a preferred method, the dynamic coordinate mapping method first determines the key feature points. The center of the umbilicus is determined by the extreme value of the curvature of the 3D point cloud and the highest point of the thermal imaging temperature. The pubic symphysis is identified by the depth mutation point. Then, a dynamic coordinate system is established. The Z-axis runs along the Ren meridian from the center of the umbilicus to the pubic symphysis. The X-axis is perpendicular to the Z-axis and parallel to the treatment bed plane. The Y-axis is determined by pointing to the ventral side with the right hand.

[0021] As a preferred option, acupoint temperature field data and traditional meridian model data can be processed by setting dynamic weights and normalization. The specific calculation formula is Cacu=wT·F(T)+wG·F(G)+wM·F(M), where Cacu is the comprehensive confidence score of the acupoint, and the closer the confidence score is to 1, the more likely the point is to be an acupoint; wT is the weight coefficient value of the temperature field; wG is the weight coefficient value of the curvature of the abdominal geometric surface; wM is the weight coefficient value of the meridian data; F(T) is the normalization function of the temperature field; F(G) is the normalization function of the curvature of the abdominal geometric surface; and F(M) is the normalization function of the meridian data.

[0022] As a preferred option, the normalization formula is: In the formula, x represents the specific values ​​of temperature field, curvature of abdominal geometric surface, and matching degree of meridian data; and Substitute the minimum and maximum values ​​of the temperature field, the curvature of the abdominal geometric surface, and the matching degree of the meridian data into the normalization functions of the temperature field, the curvature of the abdominal geometric surface, and the matching degree of the meridian data.

[0023] As a preferred method, millimeter-wave radar monitors breathing fluctuations, and the robotic arm moves along the Z-axis to control the massage intensity and frequency. When the user turns more than 15°, repositioning is triggered.

[0024] When the electromyography (EMG) value of the EMG sensor is less than the threshold, the pressure is increased in a gradient of 1 N / s until it approaches the threshold. At the same time, the user's respiratory rate and user interaction feedback are observed to determine whether to continue increasing the pressure. When the EMG value is greater than the threshold, the pressure is reduced by 0.2 N / s.

[0025] Infrared thermal imaging monitors the temperature of acupoints. When the temperature difference rises above the preset value, it indicates overtreatment, and the massage should be suspended.

[0026] The intelligent abdominal acupoint massage method based on multimodal perception includes the following steps:

[0027] After the user lies down, the device is turned on, and the 3D depth camera scans the point cloud data of the abdomen to establish a three-dimensional coordinate system of the abdomen.

[0028] The pubic symphysis and the umbilicus are used as reference points;

[0029] The system performs relevant calculations based on the acupoints to be massaged and provides automatic dynamic compensation.

[0030] The massage begins with the set initial pressure and dynamically tracks the user's state in real time to adjust the massage intensity and pace.

[0031] The beneficial effects of using the present invention are as follows:

[0032] 1. Multimodal perception module, including a 3D depth camera for collecting abdominal point cloud data and constructing a 3D topological map of the abdomen; an infrared thermal imager for locating acupoints by detecting temperature anomalies, which can be customized according to the user's constitution, such as the infrared temperature control module for adding moxibustion to the Qihai acupoint for those with a cold constitution; an electromyography sensor for real-time monitoring of the tension of the rectus abdominis muscle, adaptively adjusting the massage intensity, and monitoring electromyography signals and respiratory rate to adjust the force curve in real time, thereby improving the massage and health care effect; and millimeter-wave radar for real-time tracking of body surface positions caused by respiration and controlling the speed of the massage.

[0033] The intelligent decision-making module uses the Poisson reconstruction algorithm, multimodal fusion positioning algorithm, and dynamic coordinate mapping method to update the massage strategy in real time. It can customize or select the appropriate massage method based on the user's selection or medical history.

[0034] The actuator includes an end effector consisting of a bionic massage head for performing massage, a six-dimensional force sensor for real-time pressure feedback, and an infrared temperature control module; and a three-degree-of-freedom robotic arm consisting of xyz translational compensation and rotary joints, used to control the movement of the end effector.

[0035] 2. The Poisson reconstruction algorithm, proposed by Kazhdan M in 2006, is a mesh-based 3D reconstruction algorithm based on octrees and the Poisson equation. It transforms discrete point clouds into a continuous, watertight mesh model by solving the Poisson equation, and is a classic computer graphics algorithm. Its core idea is to treat the point cloud normal field as a gradient field, solve the indicator function using the inverse Poisson equation, and then extract the isosurface to obtain the 3D model.

[0036] In this application, the Poisson reconstruction algorithm sequentially constructs an octree, calculates the normal vector divergence field, solves the Poisson equation, and extracts isosurfaces, ultimately outputting a triangular mesh.

[0037] 3. The multimodal fusion localization algorithm includes abdominal curved point cloud data, acupoint temperature field data, and traditional meridian model data. Multimodal fusion localization uses abdominal curved point cloud data + acupoint temperature field data + traditional meridian model data, which improves the accuracy of acupoint recognition. The acupoint localization error is less than 3mm, which is much better than the localization error of more than 10mm of traditional methods. It can be seen that its acupoint recognition accuracy is greatly improved.

[0038] These features and advantages of the present invention will be disclosed in detail in the following specific embodiments and accompanying drawings. [Attached Image Description]

[0039] The invention will be further described below with reference to the accompanying drawings:

[0040] Figure 1 This is a schematic diagram of the massage method according to Embodiment 1 of the present invention;

[0041] Figure 2 This is a system construction diagram.

Detailed Implementation Methods

[0042] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, and not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of this application.

[0043] The concepts involved in this application will first be explained with reference to the accompanying drawings. It should be noted that the following explanation of each concept is only to make the content of this application easier to understand and does not imply any limitation on the scope of protection of this application.

[0044] Example 1:

[0045] Intelligent abdominal acupoint massage system based on multimodal perception, such as Figures 1 to 2 As shown, including

[0046] The multimodal perception module includes a 3D depth camera for collecting abdominal point cloud data and constructing a 3D topological map of the abdomen; an infrared thermal imager that uses temperature anomalies to help locate acupoints and can be customized according to the user's constitution, such as an infrared temperature control module that can add moxibustion to the Qihai acupoint for those with a cold constitution; an electromyography (EMG) sensor for real-time monitoring of the tension of the rectus abdominis muscle, adaptively adjusting the massage intensity, and monitoring EMG signals and respiratory rate to adjust the force curve in real time, thereby improving the massage and health care effect; and a millimeter-wave radar for real-time tracking of body surface positions caused by respiration and controlling the speed of the massage.

[0047] The intelligent decision-making module uses the Poisson reconstruction algorithm, multimodal fusion positioning algorithm, and dynamic coordinate mapping method to update the massage strategy in real time. It can customize or select the appropriate massage method based on the user's selection or medical history.

[0048] The actuator includes an end effector consisting of a bionic massage head for performing massage, a six-dimensional force sensor for real-time pressure feedback, and an infrared temperature control module; and a three-degree-of-freedom robotic arm consisting of xyz translation compensation and rotary joints, used to control the movement of the end effector.

[0049] The power module supplies power to the lithium battery, providing stable power support for the entire system and ensuring its normal operation.

[0050] A 3D depth camera is a device that uses active optical measurement technology to acquire the three-dimensional coordinates (X / Y / Z) and contour information of the surface of an object in a scene. Its core output is a depth map, where each pixel value represents the actual distance from that point to the camera.

[0051] Abdominal point cloud data is a set of discrete points describing the three-dimensional morphology of the abdominal surface, obtained through scanning with a 3D depth camera. The discrete point information includes: geometric information, attribute information, and medical features. Geometric information includes three-dimensional coordinates (X / Y / Z, in millimeters); attribute information includes RGB color, infrared reflectivity, and timestamps; medical features include curvature and normal vectors, used to identify acupoint depressions.

[0052] It is worth mentioning that the dynamic coordinate mapping method requires the first determination of key feature points. The center of the umbilicus is determined by the extreme value of the curvature of the 3D point cloud and the highest point of the thermal imaging temperature. The pubic symphysis is identified by the depth mutation point. Then, a dynamic coordinate system is established. The Z-axis runs along the Ren meridian from the center of the umbilicus to the pubic symphysis. The X-axis is perpendicular to the Z-axis and parallel to the treatment bed plane. The Y-axis is determined by pointing to the ventral side with the right hand.

[0053] The abdominal surface reconstruction uses the Poisson reconstruction algorithm to transform sparse point clouds into watertight meshes, accurately representing the anatomical structure of the periumbilical depression.

[0054] The Poisson reconstruction algorithm, proposed by Kazhdan M in 2006, is a mesh-based 3D reconstruction algorithm based on octrees and the Poisson equation. It transforms discrete point clouds into a continuous, watertight mesh model by solving the Poisson equation, and is a classic computer graphics algorithm. Its core idea is to treat the point cloud normal field as a gradient field, solve the indicator function using the inverse Poisson equation, and then extract the isosurface to obtain the 3D model.

[0055] In this application, the Poisson reconstruction algorithm sequentially constructs an octree, calculates the normal vector divergence field, solves the Poisson equation, and extracts isosurfaces, ultimately outputting a triangular mesh.

[0056] The multimodal fusion localization algorithm includes abdominal curved point cloud data, acupoint temperature field data, and traditional meridian model data. Multimodal fusion localization uses abdominal curved point cloud data, acupoint temperature field data, and traditional meridian model data, which improves the accuracy of acupoint recognition. The acupoint positioning error is less than 3mm, which is significantly better than the positioning error of more than 10mm of traditional methods.

[0057] Calculate the point cloud data of the abdominal surface:

[0058] The first step is to calculate the acupoint normal vector using principal component analysis. For each point p in the point cloud, a covariance matrix is ​​constructed within its neighborhood. The specific calculation formula is as follows: In the formula, k = 30;

[0059] The second step is to calculate the eigenvalues ​​and eigenvectors of the matrix. The eigenvector corresponding to the smallest eigenvalue is the normal vector.

[0060] Among them, the curvature estimation of the normal vector is performed as follows: (1) Select the target point and its neighborhood.

[0061] (2) Establish the tangent plane

[0062] (3) Using the normal vector as the Z-axis, select a vector that is not collinear with the normal vector, and determine the local X-axis and Y-axis through Schmitt orthogonalization; project the neighborhood points onto the tangent plane.

[0063] For example, for each neighborhood point P(x,y,z), after calculating its tangent plane coordinates, a quadratic surface is fitted to obtain the mean curvature, Gaussian curvature, and principal curvature.

[0064] For each neighborhood point P(x,y,z), calculate its tangent plane coordinates, i.e., u i =Δp i ·x,υ i =Δp i ·y

[0065] In the formula, Δp is the relative coordinate of the point to the tangent plane.

[0066] (4) Fitting the quadratic surface

[0067] The local surface can be represented by the following formula: ω i =au i 2 +bu i υ i +cυ i 2 Substitute the neighborhood coordinates to solve for a, b, c.

[0068] (5) Calculate curvature

[0069] Calculate the mean curvature, Gaussian curvature, and principal curvature based on the coefficients of the quadratic surface:

[0070] H = a + c, K = 4ac - b 2 , H is the mean curvature, K is the Gaussian curvature, and κ is the principal curvature.

[0071] Acupoint temperature field data and traditional meridian model data can be processed by setting dynamic weights and normalization. Traditional meridian model data can refer to traditional Chinese medicine meridian data, such as historical data such as "GB / T 30232-2013 Meridian Positioning Specification". Acupoint temperature field refers to the specific temperature distribution characteristics of acupoints and their surrounding areas. High temperature points indicate that the central area of ​​the acupoint is 0.5-1.5℃ higher than the surrounding skin, which is related to enhanced local metabolism and capillary density.

[0072] Gradient change represents the temperature conduction trend along the meridian. For example, stimulation of the Zusanli acupoint can cause the temperature to rise along the Stomach Meridian. The specific calculation formula is Cacu=wT·F(T)+wG·F(G)+wM·F(M).

[0073] In the formula, Cacu is the overall confidence score of the acupoint, and the closer the confidence score is to 1, the more likely the point is to be an acupoint; wT is the weighting coefficient of the temperature field; wG is the weighting coefficient of the curvature of the abdominal geometric surface; wM is the weighting coefficient of the meridian data; F(T) is the normalization function of the temperature field; F(G) is the normalization function of the curvature of the abdominal geometric surface; and F(M) is the normalization function of the meridian data.

[0074] The normalization formula is as follows: In the formula, x represents the specific values ​​of temperature field, curvature of abdominal geometric surface, and matching degree of meridian data; and Substitute the minimum and maximum values ​​of the temperature field, the curvature of the abdominal geometric surface, and the matching degree of the meridian data into the normalization functions of the temperature field, the curvature of the abdominal geometric surface, and the matching degree of the meridian data.

[0075] The minimum and maximum values ​​of the temperature field, abdominal geometric surface curvature, and meridian data matching degree are normalized based on the function to improve the stability of data acquisition. The minimum and maximum values ​​of all temperature fields, abdominal geometric surface curvature, and meridian data matching degree are normalized. The smaller the change, the higher the stability of the data. Furthermore, the use of multiple data for monitoring improves the accuracy of acupoint massage and enhances the massage effect.

[0076] Normalization is a technique well-known to those skilled in the art. The normalization function can be linear normalization or standard normalization, etc. The specific normalization method is not limited here.

[0077] Example 2:

[0078] The intelligent abdominal acupoint massage method based on multimodal perception includes the following steps:

[0079] After the user lies down, the device is turned on, and the 3D depth camera scans the point cloud data of the abdomen to establish a three-dimensional coordinate system of the abdomen.

[0080] The pubic symphysis and the umbilicus are used as reference points;

[0081] The system performs relevant calculations based on the acupoints to be massaged and provides automatic dynamic compensation.

[0082] The massage begins with the set initial pressure and dynamically tracks the user's state in real time to adjust the massage intensity and pace.

[0083] Taking massaging the Guanyuan acupoint as an example:

[0084] 1. Initial positioning:

[0085] The user lies flat, and a 3D camera establishes a three-dimensional coordinate system for the abdomen.

[0086] 2. Identify the pubic symphysis and umbilicus as reference points.

[0087] 3. Calculate the location of Guanyuan acupoint: 3 cun below the navel (±0.5cm dynamic compensation).

[0088] 4. Used to select abdominal massage methods. It is worth mentioning that the massage methods are not limited to the following examples:

[0089] Examples of massage techniques:

[0090] (1) Massage acupoints, such as massaging the Zhongwan acupoint;

[0091] (2) Knead and push along the midline of the abdomen, such as from the Zhongwan acupoint to the navel, using a certain speed curve and force curve to push and knead back and forth;

[0092] (3) Abdominal massage: With the navel as the center and the abdomen to the Zhongwan acupoint as the radius, massage in a clockwise or counterclockwise direction.

[0093] For massage techniques, a 6-dimensional force sensor is installed at the end of the robotic arm. Based on the 6-dimensional force sensor, the force information and corresponding torque information of the real-time massage force in multiple directions are detected in real time. The multiple directions can be forces in the X, Y, and Z directions in the spatial coordinate system. The corresponding torque information are the torques around the X, Y, and Z axes, respectively. The massage strategy adjustment updated by the intelligent decision-making module is sent to the massage head to adjust the posture and direction of the massage head based on the adjustment data.

[0094] 5. Dynamic tracking:

[0095] Millimeter-wave radar monitors breathing fluctuations, and the robotic arm's Z-axis motion compensates for these fluctuations, adjusting the massage intensity and frequency. Repositioning is triggered when the user's turn exceeds 15°.

[0096] When the electromyography (EMG) value is below the threshold, the pressure is increased in increments of 1 N / s until the set threshold is reached. The user's respiratory rate and interactive feedback are observed to determine whether to continue increasing the pressure. When the EMG value is above the threshold, the pressure is reduced by 0.2 N / s.

[0097] Infrared thermal imaging monitors the temperature index of acupoints. When the temperature difference rises above the preset value (2.5℃), it indicates overtreatment, and massage should be suspended.

[0098] It should be noted that, for the sake of simplicity, the foregoing embodiments are all described as a series of actions. However, those skilled in the art should understand that this application is not limited to the described order of actions, as some steps may be performed in other orders or simultaneously according to this application. Furthermore, those skilled in the art should also understand that the embodiments described in the specification are preferred embodiments, and the actions involved are not necessarily essential to this application.

[0099] The embodiments of this application have been described in detail above. Specific examples have been used in the illustration. The description of the above embodiments is only for the purpose of helping to understand the method and core ideas of this application. At the same time, for those skilled in the art, 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 application.

[0100] This application is described with reference to flowchart illustrations and / or block diagrams of methods, hardware products, and computer program products according to embodiments of this application. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart... Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0101] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The functions specified in one or more boxes. Memory may include: flash drives, read-only memory (ROM), random access memory (RAM), hard disks or optical disks, etc.

[0102] Although this application has been described herein in conjunction with various embodiments, those skilled in the art, by reviewing the accompanying drawings, disclosure, and appended claims, will understand and implement other variations of the disclosed embodiments in carrying out the claimed application. In the claims, the word "comprising" does not exclude other components or steps, and "a" or "an" does not exclude multiple instances. While different dependent claims may recite certain measures, this does not mean that these measures cannot be combined to produce a good effect.

[0103] Those skilled in the art will understand that all or part of the steps in the various methods of any of the above method embodiments can be implemented by a program instructing related hardware. The program can be stored in a computer-readable storage medium, which may include: flash drive, read-only memory (ROM), random access memory (RAM), disk or optical disk, etc.

[0104] It is understood that any product that is controlled or configured to perform the processing method of the flowchart described in the embodiments of this application, such as the apparatus and computer program product of the above flowchart, falls within the scope of the related products described in this application.

[0105] Obviously, those skilled in the art can make various modifications and variations to the device provided in this application without departing from the spirit and scope of this application. Therefore, if these modifications and variations of this application fall within the scope of the claims of this application and their equivalents, this application also intends to include these modifications and variations.

Claims

1. An intelligent abdominal acupoint massage system based on multimodal perception, characterized in that, include The multimodal perception module includes a 3D depth camera for collecting abdominal point cloud data and constructing a three-dimensional topological map of the abdomen; an infrared thermal imager for locating acupoints by detecting abnormal temperature points; an electromyography sensor for real-time monitoring of the tension of the rectus abdominis muscle and adaptively adjusting the massage intensity; and a millimeter-wave radar for real-time tracking of body surface positions caused by respiration and controlling the speed of the massage. The intelligent decision-making module uses the Poisson reconstruction algorithm, multimodal fusion positioning algorithm, and dynamic coordinate mapping method to update the massage strategy in real time; The actuator includes an end effector consisting of a bionic massage head for performing massage, a six-dimensional force sensor for real-time pressure feedback, and an infrared temperature control module; and a three-degree-of-freedom robotic arm consisting of xyz translation compensation and rotary joints, used to control the movement of the end effector. The power module supplies power to the lithium battery, providing stable power support for the entire system and ensuring its normal operation.

2. The intelligent abdominal acupoint massage system based on multimodal perception as described in claim 1, characterized in that, The multimodal fusion localization algorithm includes abdominal surface point cloud data, acupoint temperature field data, and traditional meridian model data. The abdominal surface point cloud data is first processed using principal component analysis to calculate the acupoint normal vector. Then, a covariance matrix is ​​constructed within the neighborhood of each point p in the point cloud. The specific calculation formula is as follows: Calculate the eigenvalues ​​and eigenvectors of the matrix. The eigenvector corresponding to the smallest eigenvalue is the normal vector.

3. The intelligent abdominal acupoint massage system based on multimodal perception as described in claim 2, characterized in that, Curvature estimation of the normal vector: Select the target point and its neighborhood; establish the tangent plane; take the normal vector as the Z-axis, select vectors that are not collinear with the normal vector, and determine the local X-axis and Y-axis through Schmitt orthogonalization; project the neighborhood points onto the tangent plane.

4. The intelligent abdominal acupoint massage system based on multimodal perception as described in claim 3, characterized in that, For each neighborhood point P(x,y,z), after calculating its tangent plane coordinates, a quadratic surface is fitted to obtain the mean curvature, Gaussian curvature, and principal curvature. The specific calculation formula is as follows: u i =Δp i ·x,υ i =Δp i ·y oh i =au i 2 +bu i u i +cy i 2 H=a+c,K=4ac-b 2 , , In the formula, Δp is the relative coordinate of the point to the tangent plane, H is the mean curvature, K is the Gaussian curvature, and κ is the principal curvature.

5. The intelligent abdominal acupoint massage system based on multimodal perception as described in claim 1, characterized in that, The Poisson reconstruction algorithm proceeds sequentially by constructing an octree, calculating the normal vector divergence field, solving the Poisson equation, and extracting isosurfaces, ultimately outputting a triangular mesh.

6. The intelligent abdominal acupoint massage system based on multimodal perception as described in claim 1, characterized in that, The dynamic coordinate mapping method first determines key feature points. The center of the umbilicus is determined by the extreme value of the curvature of the 3D point cloud and the highest point of the thermal imaging temperature. The pubic symphysis is identified by the depth mutation point. Then, a dynamic coordinate system is established. The Z-axis runs along the Ren meridian from the center of the umbilicus to the pubic symphysis. The X-axis is perpendicular to the Z-axis and parallel to the treatment bed plane. The Y-axis is determined by pointing to the ventral side with the right hand.

7. The intelligent abdominal acupoint massage system based on multimodal perception as described in claim 2, characterized in that, The acupoint temperature field data and traditional meridian model data can be processed by setting dynamic weights and normalization. The specific calculation formula is Cacu=wT·F(T)+wG·F(G)+wM·F(M), where Cacu is the comprehensive confidence score of the acupoint, and the closer the confidence score is to 1, the more likely the point is to be an acupoint; wT is the weight coefficient value of the temperature field; wG is the weight coefficient value of the curvature of the abdominal geometric surface; wM is the weight coefficient value of the meridian data; F(T) is the normalization function of the temperature field; F(G) is the normalization function of the curvature of the abdominal geometric surface; and F(M) is the normalization function of the meridian data.

8. The intelligent abdominal acupoint massage system based on multimodal perception as described in claim 7, characterized in that, The normalization formula is: In the formula, x represents the specific values ​​of temperature field, curvature of abdominal geometric surface, and matching degree of meridian data; and Substitute the minimum and maximum values ​​of the temperature field, the curvature of the abdominal geometric surface, and the matching degree of the meridian data into the normalization functions of the temperature field, the curvature of the abdominal geometric surface, and the matching degree of the meridian data.

9. A method for intelligent abdominal acupoint massage based on multimodal perception, characterized in that: Includes the following steps: After the user lies down, the device is turned on, and the 3D depth camera scans the point cloud data of the abdomen to establish a three-dimensional coordinate system of the abdomen. The pubic symphysis and the umbilicus are used as reference points; The system performs relevant calculations based on the acupoints to be massaged and provides automatic dynamic compensation. The massage begins with the set initial pressure and dynamically tracks the user's state in real time to adjust the massage intensity and pace.

Images