Qi and collateral dredging robot regulation and control method based on 3D acupuncture point calibration

Abstract

The invention discloses a 3D acupoint calibration-based qi and collateral dredging robot regulation and control method, which comprises the following steps of: constructing a user massage manipulation demand prediction model, and predicting massage manipulation demands of different users; preset acupuncture points are calibrated on the body of the user through green labels, space coordinates of other acupuncture points on the body of the user are positioned and updated in real time, and a massage manipulation demand prediction result is adjusted; recognizing facial expressions of the user, recognizing body actions, voice content and voice emotion feedback, and generating a massage parameter adjustment strategy of the qi-smoothing collateral-dredging robot; and according to user feedback data, evaluating the effectiveness of the massage effect and the massage parameter adjustment strategy, adjusting the massage parameters of the qi-smoothing and collateral-dredging robot and optimizing the user massage manipulation demand prediction model. According to the invention, through integrating the body characteristics of the user and real-time feedback, the massage manipulation and the robot parameters are accurately regulated and controlled, personalized and intelligent massage experience is provided, and the problems of insufficient personalization and poor real-time performance in the prior art are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent health robots, and particularly to a regulation method for a qi-promoting and meridian-unblocking robot based on 3D acupoint calibration. Background Art

[0002] With the development of technology, intelligent massage robots have gradually entered the market, attempting to make up for the deficiencies in traditional massage methods. However, existing massage devices usually operate based on preset fixed parameters, which cannot be dynamically adjusted according to the user's body characteristics and real-time feedback. Especially when massaging specific acupoints, there is a lack of precise acupoint positioning technology, and the massage effect is often affected. For example, traditional massage devices are difficult to accurately identify and locate specific acupoints on the human body. Especially when the user's body shape and posture change, the accuracy of the massage position may be affected, resulting in a significant reduction in the massage effect. In addition, most existing intelligent massage robots work based on fixed programs and parameters. These devices usually rely on set modes for massage, such as setting fixed intensity, rhythm, etc., and cannot be adjusted according to the individual differences of users. Especially when faced with various factors such as body shape characteristics, health conditions, facial expressions, and body movements, these devices often cannot make effective responses. Due to differences in the body shapes, health conditions, muscle structures, etc. of different users, the fixed massage parameters often cannot meet the specific needs of each user. Therefore, there are still significant deficiencies in existing massage devices in providing a personalized massage experience. Additionally, the massage techniques of existing massage robots often rely on preset parameters, lacking instant perception and feedback adjustment of the user's body state. Therefore, the accuracy and real-time performance of the massage effect are poor, and the user experience is not always ideal. In addition to the deficiencies in massage parameters, massage robots in the prior art often ignore the importance of real-time user feedback. During the massage process, information such as the user's body movements, facial expressions, and voice feedback can provide important feedback data for the device, helping the system evaluate the current massage effect and then dynamically adjust the massage parameters. However, most existing massage robots lack such a feedback mechanism and therefore cannot make effective adjustments according to the user's immediate needs, which directly affects the personalization and comfort of the massage effect. To sum up, there are still many deficiencies in existing massage robot technologies in terms of personalized adjustment, real-time response to user needs, dynamic adjustment of massage parameters, and integration of user feedback, and they cannot provide a comprehensive and accurate user experience. To address these problems, a new technical solution is urgently needed that can comprehensively consider the user's body characteristics, real-time feedback, and personalized needs to provide a more customized and accurate massage service. Summary of the Invention

[0003] In view of the problems existing in the above prior art, the present invention provides a regulation method for a qi-promoting and meridian-unblocking robot based on 3D acupoint calibration, mainly including:

[0004] Obtain the user's gender, age, occupation information, health status, and massage technique requirements, construct a prediction model for the user's massage technique requirements, and predict the massage technique requirements of different users;

[0005] Calibrate the preset acupoints on the user's body with green label stickers, combine the user's body shape feature data, real-time locate and update the spatial coordinates of other acupoints on the user's body, and adjust the prediction result of the massage technique requirements based on the user's body shape feature data;

[0006] According to the user monitoring images and voice feedback obtained in real time by the 3D camera, identify the user's facial expressions, body movements, voice content, and voice emotion feedback, determine whether it is necessary to adjust the massage parameters of the qi-promoting and collaterals-unblocking robot, and generate a massage parameter adjustment strategy for the qi-promoting and collaterals-unblocking robot;

[0007] According to the user feedback data, evaluate the massage effect and the effectiveness of the massage parameter adjustment strategy, and adjust the massage parameters of the qi-promoting and collaterals-unblocking robot and optimize the user massage technique requirement prediction model based on the effectiveness evaluation result.

[0008] Furthermore, the obtaining the user's gender, age, occupation information, health status, and massage technique requirements, constructing a prediction model for the user's massage technique requirements, and predicting the massage technique requirements of different users includes:

[0009] Obtain the user's gender, age, occupation information, and health status through the user data center. The occupation information includes occupation name, working years, working intensity, working type, and job tasks; obtain the user's massage technique requirements through the massage technique collector and save them to the massage technique database. The massage technique requirements include acupoints, strength, rhythm, frequency, and massage heads; use a recurrent neural network for model training according to the user's gender, age, occupation information, and health status, and the user's massage technique requirements, to construct a prediction model for the user's massage technique requirements; according to the current user's gender, age, occupation information, and health status, use the prediction model for the user's massage technique requirements to predict the massage technique requirements of this user; if the occupation name of the current user does not exist in the user data center, obtain the massage technique requirements of a similar occupation name of the current user's occupation name.

[0010] It also includes that if the occupation name of the current user does not exist in the user data center, obtain the massage technique requirements of a similar occupation name of the current user's occupation name, specifically including:

[0011] Obtain the occupational information of different occupational groups through occupational health reports and relevant ergonomic research reports, and mark the common pain areas and high-risk areas of different occupational groups. The occupational information includes occupational name, working years, working intensity, working type, and job tasks. If the occupational name of the current user does not exist in the user data center, calculate the similarity between the common pain areas and high-risk areas of the current user's occupational group and those of the user F occupational group using the cosine similarity calculation method, obtain the similar occupational names with a similarity higher than the preset first similarity threshold to the common pain areas and high-risk areas of the current user's occupational group, and use the user massage technique demand prediction model to predict the massage technique demand of this user.

[0012] Furthermore, calibrating the preset acupoints on the user's body with green label stickers, combining with the user's body shape characteristic data, real-time positioning and updating the spatial coordinates of other acupoints on the user's body, and adjusting the massage technique demand prediction result based on the user's body shape characteristic data, includes:

[0013] Use green label stickers to calibrate the preset acupoints on the user's body. Combine with the user's body shape characteristic data and use 3D vision recognition technology to locate the spatial positions of other acupoints on the user's body and determine the corresponding spatial coordinates to the calibrated acupoints. The body shape characteristic data includes muscle structure, joint position, and angle. Continuously photograph and identify the user's body through a 3D camera to real-time update the acupoint spatial coordinate data. If the user's posture changes, automatically adjust the spatial coordinates of the acupoints. Obtain the user's body shape characteristic data, and adjust the massage technique demand prediction result according to the differences in the user's body shape characteristic data. The body shape characteristic data includes muscle structure, joint position, and angle. Based on the adjusted massage technique demand, set the massage parameters for the qi-promoting and collaterals-unblocking robot, and perform massage actions on the user based on the real-time updated acupoint spatial coordinate data.

[0014] It also includes obtaining the user's body shape characteristic data and adjusting the massage technique demand prediction result according to the differences in the user's body shape characteristic data. Specifically, it includes:

[0015] Continuously capture the user's body through a 3D camera to obtain user images, and use a convolutional neural network for model training to identify the user's body shape feature data, which includes muscle structure, joint positions, and angles; calculate the similarity between the real-time obtained user's body shape feature data and the preset reference body shape feature data through the cosine similarity calculation method. If the similarity is less than the preset second similarity threshold, adjust the prediction result of the massage technique requirements according to the user's body shape feature differences, including establishing a model of the body massage area based on the user's body shape feature data, and adjusting the prediction results of the massage technique requirements for each body massage area based on the muscle structure characteristics, joint positions, and angles of the muscle groups in different body massage areas, and determining the user's massage strategy. Each body massage area includes, but is not limited to, the back, shoulders, neck, and waist.

[0016] Furthermore, based on the user monitoring images and voice feedback obtained in real time by the 3D camera, identify the user's facial expressions, body movements, voice content, and voice emotion feedback, and determine whether it is necessary to adjust the massage parameters of the qi-promoting and collaterals-unblocking robot, and generate a massage parameter adjustment strategy for the qi-promoting and collaterals-unblocking robot, including:

[0017] In real time, obtain the user monitoring images and user voice during the massage process through the 3D camera; use a convolutional neural network for model training based on the user monitoring images to identify the user's facial expressions and body movements. Facial expressions include, but are not limited to, smiling, being comfortable, and being surprised. Body movements include, but are not limited to, raising the hand, bending down, and stretching; use the Transformer model to convert the obtained user voice into text and identify the user's voice content; frame the user voice with a preset window size and perform Fourier transform on each frame signal to obtain the frequency domain representation; compress the spectrum using the Mel frequency scale, perform logarithmic transformation on the Mel spectrum, and apply discrete cosine transform to obtain the Mel frequency cepstral coefficient features; use a recurrent neural network for model training based on the obtained Mel frequency cepstral coefficient features to construct a user voice emotion recognition model and identify the user's voice emotion feedback. Voice emotion feedback includes, but is not limited to, being happy, being painful, and being calm; combine the user's facial expressions, body movements, voice content, and voice emotion feedback to determine whether it is necessary to adjust the massage parameters of the qi-promoting and collaterals-unblocking robot; if it is necessary to adjust the massage parameters of the qi-promoting and collaterals-unblocking robot, generate a massage parameter adjustment strategy for the qi-promoting and collaterals-unblocking robot based on the user's facial expressions, body movements, voice content, and voice emotion feedback, and gradually adjust the massage parameters of the qi-promoting and collaterals-unblocking robot.

[0018] Furthermore, based on the user feedback data, evaluate the effectiveness of the massage effect and the massage parameter adjustment strategy, and adjust the massage parameters of the qi-promoting and collaterals-unblocking robot and optimize the user massage technique requirement prediction model based on the effectiveness evaluation result, including:

[0019] Obtain the feedback data during the user's massage process and the feedback data after the massage, evaluate the massage effect and the effectiveness of the massage parameter adjustment strategy. The feedback data during the user's massage process includes facial expressions, body movements, voice content, and voice emotion feedback. If the massage effect and the effectiveness of the massage parameter adjustment strategy are lower than the preset standard, adjust the massage parameters of the Qi-promoting and Meridian-clearing robot based on the user feedback data, and optimize the user's massage technique demand prediction model to obtain an optimized user massage strategy. Implement the optimized user massage strategy, obtain new user feedback data, and continuously optimize the user's technique demand prediction model and the parameter adjustment strategy of the massage robot.

[0020] The technical solutions provided by the embodiments of the present invention may include the following beneficial effects:

[0021] The present invention provides a method for regulating a Qi-promoting and Meridian-clearing robot based on 3D acupoint calibration. According to the user's gender, age, occupation, health status, and body shape characteristics, combined with real-time feedback information, the present invention accurately predicts and dynamically adjusts the user's massage needs. Through green label sticker calibration and 3D vision recognition technology, the spatial coordinates of acupoints are updated in real time. Combining with the user's body shape characteristic data, it ensures that the massage robot can accurately locate the target acupoints and adjust the massage parameters according to the user's posture changes, realizing a personalized massage experience. The present invention can dynamically adjust parameters such as massage intensity and rhythm according to the user's real-time facial expressions, body movements, and voice emotion feedback, solves the problems of lack of real-time response and personalized adjustment of traditional massage robots, and ensures that each massage can accurately meet the user's needs. By comprehensively considering the user's body characteristics and real-time feedback, the present invention accurately regulates the massage techniques and robot parameters, provides a personalized and intelligent massage experience, overcomes the problems of insufficient personalization and poor real-time performance in the prior art, significantly improves the personalization and comfort of the massage effect, and promotes the intelligent development of massage robot technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0022] Figure 1 is a flowchart of a method for regulating a Qi-promoting and Meridian-clearing robot based on 3D acupoint calibration of the present invention;

[0023] Figure 2 is a schematic diagram of a method for regulating a Qi-promoting and Meridian-clearing robot based on 3D acupoint calibration of the present invention;

[0024] Figure 3 is another schematic diagram of a method for regulating a Qi-promoting and Meridian-clearing robot based on 3D acupoint calibration of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0025] In order to make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0026] As Figures 1-3 , a regulation method of a qi-promoting and collaterals-unblocking robot based on 3D acupoint calibration in this embodiment may specifically include:

[0027] Step S101, obtain the user's gender, age, occupation information, health status, and massage technique requirements, construct a user massage technique requirement prediction model, and predict the massage technique requirements of different users.

[0028] Obtain the user's gender, age, occupation information, and health status through the user data center. The occupation information includes occupation name, working years, working intensity, working type, and post tasks. Obtain the user's massage technique requirements through the massage technique collector and save them to the massage technique database. The massage technique requirements include acupoints, strength, rhythm, frequency, and massage heads. Use a recurrent neural network for model training according to the user's gender, age, occupation information, health status, and the user's massage technique requirements, and construct a user massage technique requirement prediction model. According to the current user's gender, age, occupation information, and health status, use the user massage technique requirement prediction model to predict the massage technique requirements of this user. If the occupation name of the current user does not exist in the user data center, obtain the massage technique requirements of similar occupation names of the current user's occupation name.

[0029] Exemplarily, there is a user A, aged 32, female, with the occupation of a marketing manager and has been in this position for 8 years. User A has a relatively high work intensity. The work type is sedentary and mental labor that often requires attending meetings, planning, and promoting marketing activities. The work tasks involve long hours of mental work, communicating with customers, and team collaboration. User A is in good health. When using the massage device, User A recorded the following massage needs. User A likes massages on the back, shoulders, and neck, with a moderate massage intensity, a moderate rhythm, and a relatively high frequency. The massage head selected is a rolling massage head, which is suitable for soothing tense muscles, especially in the shoulders and neck. All these massage need information is saved into the massage technique database through the massage technique collector. Obtain the historical massage need data of user F from the massage technique database. By combining the gender, age, occupation information, health status, and massage need data of different users, use a recurrent neural network for model training to construct a user massage technique need prediction model. This model will learn the massage need patterns of people in different industries and with different health statuses under different circumstances. If user B, aged 34, female, with the occupation of a public relations manager and 10 years of work experience, has a relatively high work intensity. The work type is sedentary and mental labor that involves communicating and coordinating with customers, organizing, and planning various media activities. The work tasks are to stay in the office for a long time, handle a large number of emails, phone conferences, and prepare speech materials. At the same time, it is also necessary to frequently go out to participate in various industry activities and meetings. User B is in good health. Using the user massage technique need prediction model, it is predicted that user B needs a medium massage intensity, a moderate rhythm, a relatively high frequency, with key massage areas on the back, shoulders, and neck, and a rolling massage head that soothes muscles is used. User C, aged 30, female, with the occupation of a financial analyst and 6 years of work experience, has a relatively high work intensity. User C's work type is sedentary and mental labor for financial calculations in the office. The work tasks are to be responsible for analyzing the company's financial statements, formulating budgets, and predicting financial conditions, handling a large amount of numbers and data, and frequently participating in internal meetings and team discussions. User C is in good health. If there is no massage technique need data for the occupation name of financial analyst in the massage technique database, then obtain the massage technique needs of similar occupation names of financial analysts.

[0030] Among them, if the current user's occupation name does not exist in the user data center, then obtain the massage technique needs of similar occupation names of the current user's occupation name.

[0031] Obtain the occupational information of different occupational groups through occupational health reports and relevant ergonomic research reports, and mark the common pain areas and high-risk areas of different occupational groups. The occupational information includes occupational name, working years, work intensity, work type, and job tasks. If the occupational name of the current user does not exist in the user data center, calculate the similarity between the common pain areas and high-risk areas of the current user's occupational group and those of the user F occupational group through the cosine similarity calculation method, obtain the similar occupational names whose similarity to the common pain areas and high-risk areas of the current user's occupational group is higher than the preset first similarity threshold, and use the user massage technique demand prediction model to predict the massage technique demand of this user.

[0032] Exemplarily, User C, 30 years old, female, is a financial analyst with 6 years of working experience and a relatively high work intensity. The work type of User C is sedentary and mental work of financial calculation in the office. The job tasks are to be responsible for analyzing the company's financial statements, formulating budgets, and predicting financial situations, dealing with a large amount of numbers and data, and frequently participating in internal meetings and team discussions, with good health. To predict the massage technique demand of User C, refer to the data in the occupational health report and ergonomic research report, which provide the common pain areas and high-risk areas for different occupational groups. For example, financial analysts and their user F occupations, such as data analysts, software engineers, etc., often feel discomfort in the shoulders, neck, and waist. The work types of these occupational groups mostly involve sitting for a long time, staring at the computer screen for a long time, and concentrating on processing numbers and data mentally, resulting in fatigue and tension in the upper body muscles, especially the shoulders and neck. In the case of User C, the shoulders, neck, and waist are identified as common pain areas and marked as high-risk areas. Obtain the data of the common pain areas of the financial analyst group and compare it with the data of the user F occupations, such as the software engineer or market analyst group. If the occupational information of User C cannot be directly found in the data center, calculate the cosine similarity to measure the similarity of the common pain areas and high-risk areas of different occupational groups. If it is calculated that the similarity of the pain areas and high-risk areas between the financial analyst group and the data analyst group reaches the preset first similarity threshold of 0.8. Based on this result, the data analyst is taken as a similar occupation, and the occupational name of the user, financial analyst, is replaced with the data analyst to predict the massage technique demand of User C. The prediction result shows that User C may need massages for the shoulders, neck, and waist, with moderate massage strength, moderate rhythm, and high frequency, and the massage head is selected as the rolling type or the type suitable for relieving muscles to help User C relieve the muscle tension and fatigue caused by sedentary work.

[0033] Step S102: Use a green label sticker to mark the preset acupoints on the user's body. Combine the user's body shape characteristic data to locate and update the spatial coordinates of other acupoints on the user's body in real time, and adjust the prediction result of the massage technique requirements based on the user's body shape characteristic data.

[0034] Use a green label sticker to mark the preset acupoints on the user's body. Combine the user's body shape characteristic data and use 3D vision recognition technology to locate the spatial positions of other acupoints on the user's body and determine the corresponding spatial coordinates of the marked acupoints. The body shape characteristic data includes muscle structure, joint position and angle. Continuously capture and identify the user's body through a 3D camera to update the acupoint spatial coordinate data in real time. If the user's posture changes, automatically adjust the spatial coordinates of the acupoints. Obtain the user's body shape characteristic data and adjust the prediction result of the massage technique requirements according to the differences in the user's body shape characteristic data. The body shape characteristic data includes muscle structure, joint position and angle. Set the massage parameters for the qi-promoting and collaterals-unblocking robot based on the adjusted massage technique requirements, and perform massage actions on the user based on the real-time updated acupoint spatial coordinate data.

[0035] Exemplarily, there is a user, User D, who is 32 years old, female, with an occupation as a product manager and 8 years of work experience. Her work intensity is relatively high. The work type of User D mainly involves long-term sedentary and mental labor with long hours of computer use. Her work tasks include writing project documents, analyzing product data, and communicating and discussing with team members. Her health condition is good. Before starting to use the massage device, green label stickers are used to mark the preset acupoints on User D's body. These acupoints include commonly used massage acupoints such as Jianjing acupoint, Hegu acupoint, and Tianzong acupoint. Through 3D vision recognition technology, first, a 3D camera is used to capture and obtain the body shape feature data of User D, including muscle structure, joint positions and their angles, etc., and the spatial coordinates corresponding to the marked acupoints are determined through these data. If the shoulder muscles of User D are relatively tight, by accurately identifying the shoulder position and angle, the spatial coordinates of the Jianjing acupoint are determined as (x1, y1, z1), and corresponding adjustments are made in combination with the shoulder shape of User D. During the massage process, the 3D camera continuously captures and recognizes User D's body in real time, and the spatial coordinate data of the acupoints is updated in real time. If User D slightly adjusts her sitting posture during the massage, resulting in a change in the shoulder position, then the new coordinates of the Jianjing acupoint are automatically updated through 3D vision recognition technology, and the adjusted shoulder coordinates of User D become (x2, y2, z2). Then, the massage action is adjusted according to this new coordinate to ensure that the massage head accurately aligns with the Jianjing acupoint of User D. In addition, according to the differences in the body shape characteristics of User D, the prediction results of the massage technique requirements are adjusted. For example, if the shoulder muscles of User D are tighter than those of ordinary people, or the joint angles of User D are different, then according to these differences, the massage intensity and rhythm are adjusted accordingly. For example, if the muscle structure in the shoulder area of User D is more tense than normal, the massage force will be adjusted to be stronger to help effectively relieve muscle tightness. Based on these real-time updated body shape feature data and the adjusted massage technique requirements, massage parameters are set for the Qi-promoting and collaterals-unblocking robot. For example, a medium to strong massage force, moderate rhythm and frequency are set for the shoulder, and at the same time, it is ensured that the movement trajectory of the massage head is consistent with the spatial coordinates of acupoints such as the Jianjing acupoint and Tianzong acupoint. As User D's posture changes, the massage robot will also adjust its massage path and force to ensure maximum relief of her muscle discomfort.

[0036] Among them, obtain the body shape feature data of the user, and adjust the prediction result of the massage technique requirement according to the difference in the body shape feature of the user.

[0037] Continuously photograph the user's body through a 3D camera to obtain user images, and use a convolutional neural network for model training to identify the user's body shape feature data, where the body shape feature data includes muscle structure, joint positions, and angles. Calculate the similarity between the body shape feature data of the user obtained in real time and the preset reference body shape feature data through the cosine similarity calculation method. If the similarity is less than the preset second similarity threshold, adjust the prediction result of the massage technique requirements according to the differences in the user's body shape features, including establishing a model of the body massage area based on the user's body shape feature data, and adjusting the prediction results of the massage technique requirements for each body massage area based on the muscle structure characteristics, joint positions, and angles of the muscle groups in different body massage areas, and determining the user's massage strategy. Each body massage area includes but is not limited to the back, shoulders, neck, and waist.

[0038] Exemplarily, a user named User E, 35 years old, male, with an occupation of engineer and 10 years of work experience. The work type of User E is long-term sedentary and mental labor in designing and developing technical solutions. The work task is to sit in front of the computer for a long time and occasionally go out for equipment inspection and on-site debugging, and the health status is good. When User E uses the massage device, the 3D camera continuously takes pictures of User E's body to obtain the image data of User E's body. Using a convolutional neural network, the body shape feature data of User E is recognized, mainly including the muscle structure of User E, the position and angle of the joints. If it is recognized through these data that the shoulder muscle structure of User E is relatively tense, the shoulder joint angle is deviated outward, and there is also a certain degree of tension in the lumbar muscles, especially in the lumbar spine area, and the joint range of motion is small. Through the cosine similarity calculation method, the similarity between the real-time body shape feature data of User E and the preset reference body shape feature data is calculated. If the calculated similarity is 0.78, and the preset second similarity threshold is 0.85, this means that the similarity between User E's body shape features and the reference body shape data is lower than the threshold, so it is considered that there are certain differences in User E's body shape features and personalized adjustment is required. According to the differences in User E's body shape features, the prediction result of the massage technique requirements is adjusted. Specifically, a model of the body massage area is established according to the muscle structure characteristics, joint positions and angles of User E. Since the shoulder muscles of User E are relatively tight, the joint range of motion is small, and there is also a certain sense of compression in the back muscle group, these areas are analyzed key points to determine the personalized requirements for each body massage area. For example, according to the shoulder muscle structure and the shoulder joint angle of User E, it is concluded that User E needs moderate deep massage on the shoulders to relieve the discomfort caused by muscle tension and limited joint movement. Since the joint range of motion of the lumbar spine is small, the massage intensity is adjusted to be moderate, and the massage head selection may tend to be soft type so as not to exert too much pressure on the lumbar spine. For the back, since User E's back muscles are fatigued after long-term sitting work, the system will recommend using moderate strength and rhythm, especially focusing on relaxing the middle back area.

[0039] Step S103, according to the user monitoring image and voice feedback obtained in real time by the 3D camera, identify the user's facial expression, body movement recognition, voice content and voice emotion feedback, determine whether it is necessary to adjust the massage parameters of the smooth qi and dredge collaterals robot, and generate a massage parameter adjustment strategy for the smooth qi and dredge collaterals robot.

[0040] The user monitoring images and user voices during the massage process are obtained in real time through a 3D camera. Based on the user monitoring images, a convolutional neural network is used for model training to identify the user's facial expressions and body movements. The facial expressions include but are not limited to smiling, being comfortable, and being surprised. The body movements include but are not limited to raising the hand, bending down, and stretching. The Transformer model is used to convert the obtained user voice into text to identify the user's voice content. The user voice is framed with a preset window size, and the Fourier transform is performed on each frame signal to obtain the frequency domain representation. The mel frequency scale is used to compress the spectrum, the logarithmic transform is performed on the mel spectrum, and the discrete cosine transform is applied to obtain the mel-frequency cepstral coefficient features. Based on the obtained mel-frequency cepstral coefficient features, a recurrent neural network is used for model training to construct a user voice emotion recognition model to identify the user's voice emotion feedback. The voice emotion feedback includes but is not limited to being happy, being painful, and being calm. Combining the user's facial expressions, body movements, voice content, and voice emotion feedback, it is judged whether it is necessary to adjust the massage parameters of the smooth qi and dredge collaterals robot. If it is necessary to adjust the massage parameters of the smooth qi and dredge collaterals robot, a massage parameter adjustment strategy for the smooth qi and dredge collaterals robot is generated based on the user's facial expressions, body movements, voice content, and voice emotion feedback, and the massage parameters of the smooth qi and dredge collaterals robot are gradually adjusted.

[0041] Exemplarily, during the massage process, user E obtains real-time user monitoring images and voice feedback of user E through a 3D camera to monitor the emotional changes and comfort level of user E during the massage. The facial expressions and body movements of user E are obtained through the camera, and a convolutional neural network is used for model training to identify that user E smiles at the beginning of the massage, and user E slightly stretches user E's hands. Based on the data of these facial expressions and body movements, it is inferred that user E is currently in a relaxed state. At the same time, user E starts to interact with the system through voice. User E says that the massage intensity is good and the shoulders no longer feel so tight, but the neck still aches a bit. The system uses a Transformer model to convert the voice of user E into text, divides the voice into frames through a preset window size, then performs a Fourier transform on each frame signal to obtain the frequency-domain representation, compresses the spectrum using the Mel frequency scale, and performs a logarithmic transformation. Finally, a discrete cosine transform is applied to obtain the Mel frequency cepstral coefficient features. Based on the obtained Mel frequency cepstral coefficient features, a user voice emotion recognition model trained by a recurrent neural network is used to identify that the voice emotion feedback of user E is calm. This indicates that the overall emotional state of user E is relatively relaxed. Combining the facial expressions, body movements, voice content, and voice emotion feedback of user E, it is judged that user E generally feels comfortable during the massage process, but still has discomfort in the neck area. Based on this information, it is necessary to adjust the massage parameters of the smooth qi and dredge collaterals robot, especially the massage intensity and frequency for user E's neck. Based on the facial smile and stretching movements of user E indicating a relaxed state, and the voice feedback of user E indicating that the shoulders have been relieved but the neck still aches, the massage parameters are adjusted, including increasing the massage intensity in the neck area, adjusting from medium intensity to strong intensity to more effectively relieve the neck pain, adjusting the massage rhythm in the neck area, moderately accelerating the rhythm, and increasing the frequency of the massage head to better meet the relaxation needs of user E. For the back and shoulder areas, maintain the current massage intensity and rhythm because user E has expressed satisfaction with these parts. After these adjustments, the smooth qi and dredge collaterals robot starts to execute the new massage parameters and continues to monitor the facial expressions and voice feedback of user E, and further fine-tune the massage strategy according to the real-time reactions of user E to ensure that user E obtains the most suitable relaxation experience throughout the process.

[0042] Step S104, according to the user feedback data, evaluate the effectiveness of the massage effect and the massage parameter adjustment strategy, and adjust the massage parameters of the smooth qi and dredge collaterals robot and optimize the user massage technique requirement prediction model based on the effectiveness evaluation result.

[0043] Obtain the feedback data during the user's massage process and the feedback data after the massage, evaluate the massage effect and the effectiveness of the massage parameter adjustment strategy. The feedback data during the user's massage process includes facial expressions, body movements, voice content, and voice emotion feedback. If the massage effect and the effectiveness of the massage parameter adjustment strategy are lower than the preset standard, adjust the massage parameters of the qi-promoting and collaterals-unblocking robot based on the user feedback data, and optimize the user massage technique demand prediction model to obtain an optimized user massage strategy. Implement the optimized user massage strategy, obtain new user feedback data, and continuously optimize the user technique demand prediction model and the parameter adjustment strategy of the massage robot.

[0044] Exemplarily, there is a user, User F, who is 45 years old, male, and works as a senior lawyer with 20 years of work experience. Due to long-term high-intensity work, User F often needs to handle a large number of cases, sitting at the desk for a long time, and often feels tense and fatigued in the shoulders, neck, and back. During the process of using the Changqi Tongluo robot for massage, the 3D camera captures User F's facial expressions, body movements, and voice content in real time, and analyzes User F's voice emotion feedback through an emotion recognition model. Ten minutes after the massage starts, User F's facial expression shows a slight smile, there are slight stretching movements in the body, and the voice feedback is that the fatigue in the shoulders has been slightly relieved, but the neck still feels a bit tight. Based on these data, it is initially judged that the shoulders of User F have been relieved, but the massage intensity and rhythm of the neck may still need to be further adjusted. Based on this feedback, the parameters of the massage robot are adjusted, especially for the neck area of User F. Therefore, the intensity of neck massage is adjusted from medium to strong, and the rhythm and frequency of massage are appropriately increased. After the adjustment, the feedback data of User F is continuously monitored, and User F's facial expressions, body movements, and voice emotions are obtained in real time. Twenty minutes after the adjustment, User F's voice feedback is that the neck now feels much better and the sense of relaxation of the whole person is stronger. At the same time, User F's facial expression also shows obvious comfort, the body movements are more relaxed, and the tension in the shoulders and neck is significantly reduced. At this time, the feedback data is analyzed again, and it is considered that the massage effect has been significantly improved. However, in the feedback data after the massage ends, User F still mentions that there is some soreness in the back, especially in the spinal area. According to this new feedback, the system evaluates the effectiveness of the massage effect and the massage parameter adjustment strategy, and finds that although the massage effect of the neck has been improved, the relaxation effect of the back, especially the spinal area, has not reached the expected standard, and the effectiveness of the massage strategy and parameter adjustment is still lower than the preset standard. Therefore, the massage parameters are further adjusted according to User F's feedback data, especially in the back area. The intensity of back massage is adjusted from medium to strong, and the frequency of back massage is reduced to make it more suitable for User F's physical condition. At the same time, the type of massage head is optimized to better act on the spinal area. In addition, the system also optimizes the user's massage technique demand prediction model according to User F's overall feedback data. Through the analysis of multiple feedbacks, the model can better capture the massage demand characteristics of different body areas, such as User F's different demands for the neck, shoulders, and back, and User F's voice emotion feedback. The system applies these optimization results to the next massage process to ensure that the massage robot can adjust the massage parameters more accurately. After these optimizations, User F undergoes massage again. After the implementation of the new optimization strategy, User F indicates in the subsequent feedback that the effect of this massage is very good, the fatigue in the spine and shoulders has completely disappeared, and the overall sense of relaxation is very strong.At this time, the system adjusted and optimized the user's massage technique demand prediction model again according to the new feedback data, and at the same time fine-tuned the parameters of the qi-promoting and collaterals-unblocking robot to ensure that every subsequent massage experience can be further optimized. Through this continuous optimization process based on feedback data, the system can continuously improve the user's massage experience, not only improving the massage effect, but also dynamically adjusting the massage technique and the parameters of the robot according to the user's personalized needs to ensure that the best effect can be achieved in each massage.

[0045] The above description is only the preferred embodiment of the present application and the explanation of the applied technical principles. Those skilled in the art should understand that the scope of the invention involved in the present application is not limited to the technical solution formed by the specific combination of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or their equivalent features without departing from the concept of the present application. For example, the technical solution formed by mutually replacing the above features with the technical features (but not limited to) disclosed in the present application that have similar functions.

Claims

1. A method for regulating and controlling a robot for promoting qi flow and unblocking meridians based on 3D acupoint calibration, characterized in that: The method comprises: Obtain the user's gender, age, occupation information, health status and massage technique needs, build a user massage technique demand prediction model, and predict the massage technique needs of different users; The preset acupuncture points on the user's body are marked with green stickers, and the spatial coordinates of other acupuncture points on the user's body are located and updated in real time based on the user's body shape feature data, and the massage technique demand prediction results are adjusted based on the user's body shape feature data; According to the user monitoring images and voice feedback obtained in real time by the 3D camera, the user's facial expressions, body movements, voice content and voice emotional feedback are identified to determine whether the massage parameters of the Changqi Tongluo robot need to be adjusted, and a massage parameter adjustment strategy for the Changqi Tongluo robot is generated; According to user feedback data, the effectiveness of massage effect and massage parameter adjustment strategy is evaluated, and based on the effectiveness evaluation results, the massage parameters of the Changqi Tongluo robot are adjusted and the user massage technique demand prediction model is optimized.

2. The method according to claim 1, wherein: The method of obtaining the user's gender, age, occupation information, health status and massage technique requirements, building a user massage technique requirement prediction model, and predicting the massage technique requirements of different users includes: The user's gender, age, occupational information and health status are obtained through the user data center. The occupational information includes occupational name, years of work, work intensity, work type, and job tasks; the user's massage technique requirements are obtained through the massage technique collector and saved in the massage technique database. The massage technique requirements include acupoints, strength, rhythm, frequency and massage head; according to the user's gender, age, occupational information and health status, as well as the user's massage technique requirements, a recurrent neural network is used for model training to build a user massage technique requirement prediction model; according to the current user's gender, age, occupational information and health status, the user's massage technique requirements are predicted using the user massage technique requirement prediction model; if the current user's occupational name does not exist in the user data center, the massage technique requirements of occupational names similar to the current user's occupational name are obtained.

3. The method according to claim 2, wherein: If the occupation name of the current user does not exist in the user data center, the massage technique requirements of occupation names similar to the occupation name of the current user are obtained, including: Through occupational health reports and related ergonomic research reports, the occupational information of different occupational groups is obtained, and the common pain sites and high-risk areas of different occupational groups are marked. The occupational information includes occupational name, years of work, work intensity, work type, and job tasks; if the occupational name of the current user does not exist in the user data center, the similarity between the common pain sites and high-risk areas of the current user's occupational group and the common pain sites and high-risk areas of user F's occupational group is calculated through the cosine similarity calculation method, and similar occupational names whose similarity with the common pain sites and high-risk areas of the current user's occupational group is higher than the preset first similarity threshold are obtained, and the user massage technique demand prediction model is used to predict the user's massage technique demand.

4. The method according to claim 1, wherein: The method of marking preset acupuncture points on the user's body by using green label stickers, locating and updating the spatial coordinates of other acupuncture points on the user's body in real time in combination with the user's body shape feature data, and adjusting the massage technique demand prediction result based on the user's body shape feature data includes: Use green stickers to mark preset acupoints on the user's body, and use 3D visual recognition technology in combination with the user's body shape feature data to locate the spatial positions of other acupoints on the user's body, and determine the spatial coordinates corresponding to the marked acupoints. The body shape feature data include muscle structure, joint position and angle. The 3D camera continuously captures and identifies the user's body, and updates the acupoint spatial coordinate data in real time. If the user's posture changes, the acupoint spatial coordinates are automatically adjusted. The user's body shape feature data is obtained, and the massage technique demand prediction results are adjusted according to the differences in the user's body shape features. The body shape feature data include muscle structure, joint position and angle. Based on the adjusted massage technique requirements, the massage parameters of the Changqi Tongluo Robot are set, and massage actions are performed on the user based on the real-time updated acupoint spatial coordinate data.

5. The method according to claim 4, wherein: The step of obtaining the user's body shape feature data and adjusting the massage technique demand prediction result according to the user's body shape feature difference includes: The user's body is continuously photographed by a 3D camera to obtain the user's image, and a convolutional neural network is used to train the model to identify the user's body feature data, which includes muscle structure and joint position and angle; the cosine similarity calculation method is used to calculate the similarity between the user's body feature data obtained in real time and the preset benchmark body feature data. If the similarity is less than the preset second similarity threshold, the massage technique demand prediction result is adjusted according to the difference in the user's body features, including establishing a model of the body massage area according to the user's body feature data, and adjusting the massage technique demand prediction result of each body massage area based on the muscle structure characteristics and joint position and angle of muscle groups in different body massage areas, and determining the user's massage strategy. Each body massage area includes but is not limited to the back, shoulders, neck and waist.

6. The method according to claim 1, wherein: The method comprises: identifying the user's facial expression, body movement, voice content and voice emotion feedback based on the user monitoring image and voice feedback acquired in real time by the 3D camera, judging whether it is necessary to adjust the massage parameters of the robot for promoting qi and unblocking meridians, and generating a massage parameter adjustment strategy for the robot for promoting qi and unblocking meridians, including: The user monitoring images and user voice of the user during the massage are acquired in real time through a 3D camera; based on the user monitoring images, a convolutional neural network is used to train the model to recognize the user's facial expressions and body movements, including but not limited to smiling, comfort and surprise, and body movements including but not limited to raising hands, bending over and stretching; the acquired user voice is converted into text using a Transformer model to recognize the user's voice content; the user voice is framed with a preset window size, and each frame signal is Fourier transformed to obtain a frequency domain representation; the spectrum is compressed using the Mel frequency scale, the Mel spectrum is logarithmically transformed, and a discrete cosine transform is applied The invention relates to a novel method for massaging the Qi Tongluo robot and a novel method for massaging the Qi Tongluo robot. The method comprises the steps of: obtaining a Mel-frequency cepstral coefficient feature by using a recurrent neural network to train a model based on the obtained Mel-frequency cepstral coefficient feature, constructing a user voice emotion recognition model, and identifying the user's voice emotion feedback, which includes but is not limited to happiness, pain, and calmness; combining the user's facial expressions, body movements, voice content, and voice emotion feedback, determining whether it is necessary to adjust the massage parameters of the Qi Tongluo robot; if it is necessary to adjust the massage parameters of the Qi Tongluo robot, generating a massage parameter adjustment strategy for the Qi Tongluo robot based on the user's facial expressions, body movements, voice content, and voice emotion feedback, and gradually adjusting the massage parameters of the Qi Tongluo robot.

7. The method according to claim 1, wherein: The method of evaluating the effectiveness of massage effect and massage parameter adjustment strategy according to user feedback data, and adjusting the massage parameters of the Changqi Tongluo robot and optimizing the user massage technique demand prediction model based on the effectiveness evaluation results, includes: Obtain user feedback data during and after the massage, and evaluate the effectiveness of the massage effect and the massage parameter adjustment strategy. The user feedback data during the massage process includes facial expressions, body movements, voice content, and voice emotion feedback. If the massage effect and the effectiveness of the massage parameter adjustment strategy are lower than the preset standard, adjust the massage parameters of the Changqi Tongluo robot based on the user feedback data, and optimize the user massage technique demand prediction model to obtain the optimized user massage strategy. Implement the optimized user massage strategy, obtain new user feedback data, and continuously optimize the user technique demand prediction model and the massage robot's parameter adjustment strategy.

Images