Multi-part cooperative massage adjusting method of qi and collateral dredging robot

Abstract

The invention discloses a multi-part cooperative massage adjusting method of a qi and collateral dredging robot, which comprises the following steps: identifying an independent function area and a function dependent area of a user body by using a 3D visual identification technology; constructing a body strain area identification model, and identifying a body strain area of the user; determining a massage mode of the body strain area based on the independent function area where the body strain area of the user is located; converting the human body thermal imaging picture into a temperature matrix, and predicting massage parameters of an independent functional area where the body strain area of the user is located; judging whether the massage paths during cooperative massage have an overlapping risk or not, and locally adjusting the massage paths with the overlapping risk; the massage effects of single-part step-by-step massage and multi-part cooperative massage are evaluated, and the massage parameter prediction model is optimized. According to the technical scheme, the integrity, accuracy and safety of the intelligent massage equipment in multi-part cooperative conditioning are effectively improved, and the comfort and health management effect of a user are remarkably improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of intelligent health robots, and in particular to a multi-site collaborative massage adjustment method of a free-flowing meridian robot. BACKGROUND

[0002] With the improvement of people's health awareness, massage equipment has gradually shifted from traditional manual massage to intelligent massage systems centered on robots. These devices are widely used in home health care, sports rehabilitation, medical assistance and other scenarios, and to some extent, they improve user comfort and health management efficiency. However, the current technology still faces many key problems in actual application, making it difficult to meet the growing demand for personalization, intelligence and multi-site collaborative conditioning. First, existing massage robots generally lack precise perception of human body structure. Most devices rely on pre-set programs or simple contact sensing control, and cannot adapt to individual differences in user body shape, posture changes, muscle distribution and skeletal structure. This leads to insufficient accuracy in identifying specific massage target sites, especially deep fatigue areas, and can cause errors or omissions, affecting the massage effect. Second, in multi-region massage scenarios, existing devices are difficult to achieve collaborative control. Modern people often experience fatigue or pressure in multiple body parts, such as neck and shoulder tension, waist and back pain, and lower limb pain. However, most current massage robots can only process each site based on single-site, linear logic, and cannot identify and handle the functional dependency between body regions. They also lack the ability to condition multiple related sites simultaneously. This segmented approach not only reduces efficiency, but also can affect the overall conditioning. Third, in multi-robot concurrent operation, there is a lack of effective path planning and coordination mechanisms. With the development of multi-robot systems, some massage devices support multi-robot collaborative work. However, there is still a lack of stable algorithms for motion trajectory prediction, path conflict detection and avoidance. Once multiple massage heads operate simultaneously in the same area or adjacent areas, it is easy to cause physical collisions, path interference or excessive stimulation, affecting device reliability and user experience. Therefore, current intelligent massage devices have obvious shortcomings in multi-dimensional perception, regional collaboration and path planning. There is an urgent need for deeper technical breakthroughs in system recognition capabilities, control strategies and user adaptability to achieve highly intelligent, dynamically adaptive, safe and efficient personalized massage experiences. SUMMARY

[0003] The present application provides a multi-site collaborative massage adjustment method for a free-flowing meridian robot to address the problems of existing technology. The method mainly includes:

[0004] The free-flowing meridian robot acquires user images and uses 3D vision recognition technology to identify independent functional areas and functionally dependent areas of the user's body.

[0005] By using the Qi-regulating and meridian-clearing robot, thermal imaging images of the user's body are obtained, and a model for identifying areas of physical strain is constructed to identify these areas.

[0006] Using image registration technology, the independent functional area where the user's strained area is located is identified, and the massage method for the strained area is determined based on the independent functional area where the user's strained area is located.

[0007] Using image processing methods based on computer vision and numerical calculation, thermal imaging images of the human body in the independent functional area where the user's body strain area is located are converted into a temperature matrix. Combined with the user's basic information data, massage parameters of the independent functional area where the user's body strain area is located are predicted.

[0008] A motion trajectory simulation algorithm was used to simulate the collaborative massage process of multiple airflow and meridian-clearing robots. The algorithm was used to determine whether there was a risk of overlapping massage paths when multiple airflow and meridian-clearing robots performed collaborative massage, and to make local adjustments to the massage paths with overlapping risks.

[0009] Based on user physical condition data before and after massage, the massage effects of single-area step-by-step massage and multi-area synergistic massage are evaluated, and the massage parameter prediction model is optimized.

[0010] Furthermore, the process of acquiring user images through the airflow and circulation machine, and using 3D visual recognition technology to identify independent functional areas and function-dependent areas of the user's body, includes:

[0011] The robot continuously captures images of the user's body using a 3D camera, and then trains a model using a convolutional neural network to identify the user's body shape features, including muscle structure and joint positions and angles. Pre-defined acupoints are marked on the user's body using green stickers. Combined with the body shape data, 3D visual recognition technology identifies independent functional areas of the user's body. Based on human anatomical data, functionally dependent independent functional areas are determined and classified as functionally dependent regions.

[0012] Furthermore, the process of acquiring thermal imaging images of the user's body using the airflow and circulation robot, constructing a model for identifying areas of physical strain, and identifying these areas includes:

[0013] The human body thermal imaging picture of the user is obtained through the thermal imaging camera mounted on the robot, the human body thermal imaging picture of the user is preprocessed, and the preprocessed human body thermal imaging picture is stored in a human body thermal imaging picture monitoring database; the historical human body thermal imaging picture of the user is obtained through the human body thermal imaging picture monitoring database, and the body fatigue area is marked in the picture; a convolutional neural network is used for model training to construct a body fatigue area identification model; and the body fatigue area of the user is identified according to the real-time human body thermal imaging picture of the user.

[0014] Further, the body fatigue area of the user is identified through image registration technology, and the massage mode of the body fatigue area is determined based on the independent functional area where the body fatigue area is located.

[0015] The human body thermal imaging picture of the user is aligned to the same coordinate system through image registration technology, and the independent functional area where the body fatigue area is located is determined based on the identification result of the user's body independent functional area and the identification result of the functional dependent area; the massage mode of the body fatigue area is determined according to the independent functional area where the body fatigue area is located, and the massage mode includes multi-site cooperative massage and single-site step-by-step massage; if there is a functional dependence between the independent functional areas where the body fatigue area is located, the massage mode of multi-site cooperative massage is used for the independent functional areas with functional dependence, otherwise, the massage mode of single-site step-by-step massage is used.

[0016] Further, the human body thermal imaging picture of the user's body fatigue area is converted into a temperature matrix through an image processing method based on computer vision and numerical calculation, and the massage parameters of the independent functional area where the body fatigue area is located are predicted in combination with the basic information data of the user, including:

[0017] The human body thermal imaging picture of the user's body fatigue area is converted into a temperature matrix through an image processing method based on computer vision and numerical calculation, wherein the temperature value of each pixel point corresponds to an element in the matrix; the basic information data of the user is obtained through the user data center, including gender and age; the massage parameters of the user are labeled according to the basic information data of the user, the temperature matrix of the independent functional area where the body fatigue area is located, and the historical data of the body part where the independent functional area is located, a recurrent neural network is used for model training to construct a massage parameter prediction model, and the massage parameters of the independent functional area where the body fatigue area is located are determined, including the massage force, massage frequency, massage method and massage trajectory of the massage head.

[0018] Further, the motion trajectory simulation algorithm is used to simulate the collaborative massage process of the plurality of free-flowing collaterals massage robots, to determine whether there is an overlap risk in the massage path when the plurality of free-flowing collaterals massage robots perform collaborative massage, and to locally adjust the massage path with the overlap risk, comprising:

[0019] If the single-part step-by-step massage mode is used for the user's body fatigue area, the massage parameters of different independent functional areas of the free-flowing collaterals massage robot are set according to the predicted massage parameters of the independent functional area where the user's body fatigue area is located, and each independent functional area is massaged step by step; if the multi-part collaborative massage mode is used for the user's body fatigue area, the massage trajectory data of the massage head of the free-flowing collaterals massage robot responsible for each independent functional area in the function-dependent area is obtained, a motion trajectory simulation algorithm is used to simulate the collaborative massage process of the plurality of free-flowing collaterals massage robots, the position of the massage head at each time is obtained, and it is determined whether there is an overlap risk in the massage path when the plurality of free-flowing collaterals massage robots perform collaborative massage; if there is an overlap risk, the massage path with the conflict is locally adjusted to obtain adjusted massage head massage trajectory data; according to the predicted massage parameters of the independent functional area where the user's body fatigue area is located, the massage parameters of the free-flowing collaterals massage robot responsible for each independent functional area in the function-dependent area are set in combination with the adjusted massage head massage trajectory data, and the plurality of free-flowing collaterals massage robots are used to perform collaborative massage on each independent functional area in the function-dependent area.

[0020] Further, the massage effect of the single-part step-by-step massage and the multi-part collaborative massage is evaluated according to the user's body state data before and after massage, and the massage parameter prediction model is optimized, comprising:

[0021] The user's body state data before and after massage is obtained, the massage effect of the single-part step-by-step massage and the multi-part collaborative massage is evaluated, and the user's body state data includes muscle relaxation degree, relief condition of the fatigue area and comfort score; if the massage effect is lower than the preset standard, the massage parameters of the independent functional area are adjusted based on the user's body state data, and the massage parameter prediction model is optimized to obtain the optimized massage parameters of the independent functional area; the optimized massage parameters of the independent functional area are implemented, the user's body state data before and after massage is re-obtained, and the massage parameter prediction model is continuously optimized.

[0022] The technical scheme provided by the embodiment of the application can include the following beneficial effects:

[0023] This invention provides a multi-site collaborative massage adjustment method using a Qi-regulating and meridian-clearing robot. By integrating 3D visual recognition and thermal imaging analysis, this invention accurately identifies independent functional areas and functionally dependent areas of the user's body, pinpointing specific areas of strain and achieving intelligent identification of target areas and customized massage strategies. Furthermore, by constructing a temperature matrix and integrating basic user information, this invention predicts personalized massage parameters, significantly improving the targeting and effectiveness of the massage. Unlike traditional massage robots that use single-site, linear processing logic, this invention employs a collaborative control mechanism involving multiple Qi-regulating and meridian-clearing robots to identify functional dependencies between areas, supporting simultaneous adjustment of multiple related areas such as the neck, shoulders, lower back, and lower limbs, thus improving the overall effectiveness and efficiency of the treatment. This invention uses a motion trajectory simulation algorithm to collaboratively schedule multiple Qi-regulating and meridian-clearing robots, identifying and avoiding path conflicts to ensure safety and smoothness during simultaneous massage of multiple areas. In addition, based on user feedback before and after massage, this invention continuously optimizes the parameter prediction model, possessing adaptive learning capabilities, significantly enhancing personalized and dynamic adjustment capabilities, and comprehensively improving the professionalism and stability of massage services. This invention not only significantly improves the overall effect and precision of massage, but also provides a safer, more efficient, and intelligent personalized massage experience in scenarios involving synergistic treatment of multiple body parts, significantly enhancing user comfort and health management outcomes. Attached Figure Description

[0024] Fig. 1 This is a flowchart of a multi-site coordinated massage adjustment method using a Qi-regulating and meridian-clearing robot according to the present invention;

[0025] Fig. 2 This is a schematic diagram of a multi-site coordinated massage adjustment method using a Qi-regulating and meridian-clearing robot according to the present invention;

[0026] Fig. 3 This is another schematic diagram of a multi-site coordinated massage adjustment method of a Qi-regulating and meridian-clearing robot according to the present invention. Detailed Implementation

[0027] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and specific embodiments.

[0028] like Figs. 1-3 This embodiment of a method for multi-site coordinated massage adjustment using a Qi-regulating and meridian-clearing robot may specifically include:

[0029] Step S101: Obtain user images through the Qi and Blood Circulation Machine, and use 3D visual recognition technology to identify independent functional areas and functionally dependent areas of the user's body.

[0030] The 3D camera installed on the robot continuously photographs the user's body, obtains the user's image, and uses a convolutional neural network to train the model to identify the user's body shape feature data, including muscle structure and joint position and angle. A green label sticker is used to mark the preset acupoints on the user's body, combined with the user's body shape feature data, using 3D vision recognition technology, to identify the user's body's independent functional areas. According to the human anatomy data, determine the independent functional areas that exist in functional dependence, and divide the independent functional areas that exist in functional dependence into functional dependence areas.

[0031] For example, a 3D camera installed on the body of the robot continuously photographs a 170 cm tall, 65 kg adult male user A, obtaining a three-dimensional image of the user's body. A convolutional neural network is used to train the model on historical user images, and the trained convolutional neural network model is used to identify the user's body shape feature data, including the starting point and ending point coordinates of the trapezius muscle, the connection angle between the humerus and the scapula is 72°, and the muscle thickness in the lumbar region is 1.4 cm, etc. These data can fully reflect the user's muscle direction and joint range of motion. By attaching a green label sticker to the user's body, specific acupoints such as Fengchi, Jianjing, Weizhong, etc. are marked. Through 3D vision recognition technology, combined with sticker calibration and body shape feature data, the user's body's independent functional areas are identified. The independent functional area refers to a group of body parts that have relatively independent functions in physiological function, such as the neck and shoulder area, the waist and back area, and the lower limb area. The muscles and joints in each area work together to complete specific movements, but have weak connections with other areas, and have relatively independent functional performance. Based on the human anatomy database, the physiological relationship between these independent functional areas is further analyzed, and it is found that user A's Fengchi, Jianjing, and the upper part of the trapezius muscle in the neck and shoulder area have a clear synergistic relationship with the upper back area containing the medial border of the scapula, the rhomboid muscle, and the thoracic segment in terms of nerve conduction path and muscle traction direction. The trapezius muscle extends from the neck to the shoulder and upper back, playing a key role in maintaining the posture of the upper body. Therefore, these two areas are divided into a functional dependence area, i.e. the neck-shoulder-upper back functional dependence area. The functional dependence area refers to the close relationship between multiple independent functional areas in structure and function. These independent functional areas have significant linkage effects when adjusting movements or physiological responses, and need to be coordinated.

[0032] Step S102, through the robot, obtain the user's body thermal image, construct a body strain area recognition model, and identify the user's body strain area.

[0033] The thermal imaging camera installed on the robot is used to obtain a thermal image of the user, and the thermal image of the user is preprocessed and stored in a thermal image database. The preprocessing operation includes adjusting the brightness and contrast of the image. The historical thermal image of the user is obtained from the thermal image database, and the body fatigue area is marked in the image. The convolutional neural network is used for model training to build a body fatigue area recognition model. According to the real-time thermal image of the user, the body fatigue area recognition model is used to identify the body fatigue area of the user.

[0034] For example, a 32-year-old male user B weighing 70 kg is photographed by the thermal imaging camera of the robot, and a thermal image of the user B is obtained. The thermal image reflects the surface temperature distribution of different parts of the user B's body. The temperature of the right scapular region of the user B is significantly higher than that of the surrounding area, reaching 38.6℃, while the temperature of the other shoulder region is only 36.1℃. The image is preprocessed, including adjusting the brightness to enhance the image clarity by 20% and increasing the contrast by 15% to highlight the difference in heat distribution. The preprocessed image is saved in the thermal image database. By calling 27 thermal images stored in the database in the past three months, the temperature distribution in different time periods is compared, and typical body fatigue areas such as shoulder, waist and elbow areas with abnormal temperature rise are manually marked on these historical images. The temperature is usually above 37.8℃. These data are used to train the convolutional neural network to build a body fatigue area recognition model. The body fatigue area refers to the part of the soft tissue that is stressed for a long time due to overuse or improper posture, and its thermal imaging feature is usually characterized by temperature rise and concentrated distribution. Before using the robot for massage, the latest thermal image is obtained in real time, and the body fatigue area recognition model is used to identify the average temperature of the right shoulder muscle group of the user B as 38.4℃, which is about 2.1℃ higher than the surrounding area, and the area is highly matched with the typical fatigue heat distribution pattern identified by the model. Therefore, it is determined that the right shoulder area is the body fatigue area of the user B.

[0035] In step S103, the image registration technology is used to identify the independent functional area where the user's body fatigue area is located, and the massage method of the body fatigue area is determined based on the independent functional area where the user's body fatigue area is located.

[0036] The human body thermal imaging picture of the user and the user image are aligned to the same coordinate system through image registration technology, and based on the identification result of the independent functional area of the user's body and the identification result of the function-dependent area, the independent functional area where the user's body fatigue area is located is determined, and whether there is a function-dependent relationship between the independent functional areas is judged. According to the independent functional area where the user's body fatigue area is located, the massage mode of the body fatigue area is determined, and the massage mode includes multi-site cooperative massage and single-site step-by-step massage. If there is a function-dependent relationship between the independent functional areas where the user's body fatigue area is located, multi-site cooperative massage is used for the independent functional areas with function-dependent relationship, otherwise, single-site step-by-step massage is used.

[0037] For example, through image registration technology, the thermal imaging picture of the user C taken at the latest time is aligned with the conventional user image obtained previously into a unified three-dimensional coordinate system, and it is ensured that the temperature information displayed in the image can be accurately corresponded to the position of a specific anatomical part on the surface of the user's body. In the processing result after image fusion, it is identified by the body fatigue area identification model that there is a persistent high temperature anomaly in the left scapular area and the upper back area of the user C, and the average surface temperature reaches 38.2℃, which is 2.3℃ higher than that of the adjacent area, indicating that the left scapular area and the upper back area of the user C are body fatigue areas. The user's body area structure information previously divided based on the green label sticker and 3D visual recognition technology is called to confirm that the two fatigue areas of the user C are respectively located in the neck-shoulder independent functional area and the upper back independent functional area. According to the division result of the function-dependent area, it is identified that the neck-shoulder independent functional area and the upper back independent functional area are divided into one function-dependent area. Based on this identification result, it is judged that the fatigue area of the user C is located in an independent functional area combination with a function-dependent relationship, so it is decided to adopt the multi-site cooperative massage mode, and the Qiangqi Tongluo robot will process the left scapular and upper trapezius muscle areas simultaneously during the massage process, and the two mechanical arms work cooperatively to relax multiple interrelated muscle groups, so as to enhance the coverage range and overall soothing effect of the massage. In contrast, if the fatigue area appears in a single area such as the lower leg or forearm, which has weak dependence on other areas, the single-site step-by-step massage strategy is selected, and only local step-by-step massage is implemented to prevent excessive stimulation or resource waste caused by multi-site intervention.

[0038] In step S104, an image processing method based on computer vision and numerical calculation is used to convert the human body thermal imaging picture of the independent functional area where the user's body fatigue area is located into a temperature matrix, and combined with the basic information data of the user, the massage parameters of the independent functional area where the user's body fatigue area is located are predicted.

[0039] An image processing method based on computer vision and numerical calculation is used to convert the thermal imaging picture of the independent functional area where the user's body fatigue area is located into a temperature matrix, wherein the temperature value of each pixel point corresponds to an element in the matrix. Through the user data center, the basic information data of the user is obtained, including gender and age. According to the basic information data of the user, the temperature matrix of the independent functional area where the body fatigue area is located, and the historical data of the body part where the independent functional area is located, and the user's massage parameters are labeled, a recurrent neural network is used for model training, a massage parameter prediction model is constructed, and the massage parameters of the independent functional area where the user's body fatigue area is located are determined, including the massage intensity, massage frequency, massage method and massage trajectory of the massage head.

[0040] Exemplarily, in a service process, the free-flowing collaterals robot identifies that a 42-year-old female user D has obvious strain phenomenon in the left lower waist area, which belongs to the independent functional area of the waist. Through image registration and thermal imaging analysis, it is found that there is a concentrated temperature rise in this area, with a maximum temperature of 38.7°C and an average temperature of 36.5°C in the surrounding area. Using image processing methods based on computer vision and numerical calculation, relying on OpenCV and Numpy libraries, the pixel area of this region in the thermal imaging picture is intercepted and converted into a 10x10 temperature matrix. Each pixel represents an actual 0.5 square centimeter of skin surface area. A typical behavior in the matrix is [36.4, 36.6, 37.0, 37.2, 38.1, 38.5, 38.7, 38.3, 37.5, 37.1], reflecting a local heating trend of temperature rising from the edge to the center, indicating that the soft tissue in this area has obvious fatigue or inflammatory response. Through the API of the user data center, the basic information of the user is obtained, including female, 42 years old. Combined with the gender and age characteristics of the user, the basic information of the user is retrieved from the historical database, such as the temperature distribution of the strain in the independent functional area of the waist, the massage response feedback and the effect after treatment of female users aged 40-45 years old, and the massage parameters are manually labeled, including the intensity, frequency, type of massage, and massage trajectory. The type of massage includes but is not limited to kneading, pushing and tapping, and the massage trajectory includes but is not limited to S-shaped, spiral and straight reciprocating. According to the obtained user basic information data, the temperature matrix of the independent functional area where the body strain area is located, the historical data of the body part where the independent functional area is located, and the user massage parameters of the independent functional area, a recurrent neural network is used for model training to build a massage parameter prediction model. Using the massage parameter prediction model, the most suitable massage parameters for the left lower waist of user D are predicted to include intensity of 18N, frequency of 12 times / minute, type of spiral kneading, and massage trajectory of elliptical trajectory distributed symmetrically around the spine. This scheme aims to avoid single-point stimulation and improve deep muscle relaxation effect by using uniform transition trajectory. Finally, the free-flowing collaterals robot uploads the predicted massage parameters to the dispatch center to guide the massage robot arm to complete personalized conditioning with precise parameters during task execution, significantly improving the user's comfort experience and recovery effect.

[0041] In step S105, a motion trajectory simulation algorithm is used to simulate the cooperative massage process of multiple free-flowing collaterals robots, to determine whether there is an overlap risk in the massage path when multiple free-flowing collaterals robots cooperate, and to make local adjustments to the massage path with an overlap risk.

[0042] If the user's body fatigue area uses a single-part step-by-step massage method, the massage parameters of the independent functional areas where the user's body fatigue area is located are predicted, the massage parameters of the different independent functional areas of the Qixi Tongluo robot are set, and each independent functional area is massaged step by step. If the user's body fatigue area uses a multi-part cooperative massage method, the massage trajectory data of the Qixi Tongluo robot massage head responsible for each independent functional area in the functional dependent area is obtained, a motion trajectory simulation algorithm is used to simulate the cooperative massage process of multiple Qixi Tongluo robots, the position of the massage head at each time is obtained, and it is judged whether there is an overlap risk in the massage path when multiple Qixi Tongluo robots are used for cooperative massage. If there is an overlap risk, the massage path with the conflict is adjusted locally to obtain the adjusted massage head massage trajectory data. According to the predicted massage parameters of the independent functional areas where the user's body fatigue area is located, the adjusted massage head massage trajectory data is combined, and the massage parameters of the Qixi Tongluo robot responsible for each independent functional area in the functional dependent area are set respectively. Through multiple Qixi Tongluo robots, each independent functional area in the functional dependent area is cooperatively massaged.

[0043] Exemplarily, the Qiyongtongluo robot detects that a 172 cm tall, 46-year-old male user E has a right scapular area with strain, which belongs to the shoulder independent functional area and has a clear functional dependence relationship with the right upper arm independent functional area. The two areas frequently work together to complete upper limb abduction and lifting actions, so the multi-site coordinated massage method is used for conditioning. The massage parameters predicted by the recurrent neural network are called, including the massage intensity required by the shoulder and back area of 22 N, the massage frequency of 10 times per minute, the deep kneading massage method, and the S-shaped path. The recommended parameters for the right upper arm area are intensity 15 N, frequency 14 times per minute, mainly pushing, and the trajectory is a spiral around the radius rotating direction. These parameters are respectively configured to the corresponding two Qiyongtongluo robots, and the massage trajectory data of each robot is called. Through the motion trajectory simulation algorithm, the paths of the two massage heads in the process of coordinated execution are dynamically simulated, the resolution is set to one time frame every 0.2 seconds, and the spatial coordinate changes of each massage head within 5 seconds are tracked. The simulation results show that in the time period of 3.2 seconds to 3.8 seconds, the paths of the two massage heads at the junction of the scapula and the humerus have a space overlap risk of about 4 cm, which may cause the massage head to physically interfere or the user experience to be unsatisfactory. To solve this problem, a local path adjustment strategy is adopted, and the massage path of the area responsible for one of the robots is slightly offset. The trajectory is horizontally translated by about 3 cm and the frequency is increased by 2 times per minute, so that it avoids the running trajectory of the other massage head without affecting the overall rhythm. The new trajectory after path adjustment has no overlap risk through simulation verification, so the finally confirmed massage parameters and path instructions are respectively sent to the two Qiyongtongluo robots. In the coordinated execution stage, the two robots run synchronously in the right shoulder and right upper arm area according to their respective task settings. One massages along the S-shaped scapula for deep kneading, and the other uniformly pushes the biceps muscle in the spiral direction, achieving dynamic obstacle avoidance collaborative massage in multiple independent functional areas in the functionally dependent area, effectively improving the effect of deep relaxation and nerve traction coordination. If in another case, the user's body strain area is determined not to have a functional dependence relationship with other areas, for example, the strain areas are located in the right upper arm independent functional area and the right lower leg rear independent functional area, respectively, then the single-site step-by-step massage method is selected. The massage parameters of the two independent functional areas are set to the Qiyongtongluo robot, and the two independent functional areas are sequentially massaged in steps.

[0044] In step S106, the massage effect of single-site step-by-step massage and multi-site coordinated massage is evaluated according to the user's body state data before and after massage, and the massage parameter prediction model is optimized.

[0045] The user's body state data before and after massage is obtained to evaluate the massage effect of single-site step-by-step massage and multi-site coordinated massage, and the user's body state data includes muscle relaxation degree, relief status of strained area and comfort score. If the massage effect is lower than the preset standard, the massage parameters of the independent functional area are adjusted based on the user's body state data, and the massage parameter prediction model is optimized to obtain the optimized massage parameters of the independent functional area. The optimized massage parameters of the independent functional area are implemented, the user's body state data before and after massage is re-obtained, and the massage parameter prediction model is continuously optimized.

[0046] For example, in a massage service for a 38-year-old female user F, the robot performs single-site step-by-step massage on the left shoulder of the user F. The initial massage parameters are predicted by the massage parameter prediction model, including massage force 16N, massage frequency 12 strokes per minute, deep kneading technique, and S-shaped curve trajectory. Before massage, the integrated muscle tension sensor evaluates the muscle relaxation degree of the user in this area as 62%, indicating that the muscle is still moderately tense, and the thermal imaging detection result shows that the average temperature of the strained area is 38.4℃, which is significantly higher than the surrounding area. The user gives a comfort evaluation score of 6.5 and feedbacks that the stiffness is not completely relieved. After massage, the body state data is re-evaluated, and the muscle relaxation degree is only improved to 68%, and the average temperature of the strained area is reduced to 37.9℃, and the overall relief amplitude is insufficient. Since these evaluation results are lower than the system set preset effect standard, such as relaxation degree improvement ≥15% and comfort score ≥8, the massage parameter optimization process is started, including adjusting the massage parameters of the independent functional area based on the current user feedback and sensor data, increasing the massage force to 19N, the frequency to 14 strokes per minute, and changing the technique to superimposed spiral push, and the trajectory from S-shaped to more suitable for the direction of muscle fiber hook-shaped. The pre- and post-service data in this massage service are integrated with historical samples to optimize the massage parameter prediction model. The optimized massage parameter prediction model is used for the second round of massage, and the pre- and post-massage data is re-obtained. After massage, the evaluation shows that the muscle relaxation degree is improved to 83%, the average temperature of the strained area is reduced to 37.2℃, the user gives a comfort evaluation score of 9.2, and leaves a message that the shoulder is significantly relaxed and no longer feels hard. The difference between the pre- and post-optimization state and the parameter adjustment path are recorded in the model training set for continuous iteration and optimization of future prediction accuracy.

[0047] The above description is only the preferred embodiment of the present application and the explanation of the technical principles. It should be understood by those skilled in the art that the scope of the protection of the present application is not limited to the technical solutions formed by the specific combinations of the above technical features. It should also cover other technical solutions formed by the combinations of the above technical features or their equivalents without departing from the concept of the present application. For example, the technical solutions formed by replacing the above features with the technical features with similar functions disclosed in the present application (but not limited to) and the like.

Claims

1. A multi-site coordinated massage adjustment method for a free-air meridian-collateral vessel robot, characterized in that, The method comprises: Obtain user images through the free-flowing qi and collaterals machine, use 3D visual recognition technology to identify the independent functional areas and function-dependent areas of the user's body; Obtain the user's human thermal imaging pictures through the free-flowing qi and collaterals robot, construct a body fatigue area identification model, and identify the user's body fatigue area; Use image registration technology to identify the independent functional area where the user's body fatigue area is located, and determine the massage method of the body fatigue area based on the independent functional area where the user's body fatigue area is located; Use an image processing method based on computer vision and numerical calculation to convert the human thermal imaging pictures of the independent functional area where the user's body fatigue area is located into a temperature matrix, combine the user's basic information data, and predict the massage parameters of the independent functional area where the user's body fatigue area is located; Use a motion trajectory simulation algorithm to simulate the collaborative massage process of multiple free-flowing qi and collaterals robots, determine whether there is an overlap risk in the massage path of the multiple free-flowing qi and collaterals robots, and make local adjustments to the massage path with an overlap risk; According to the user's body state data before and after massage, evaluate the massage effect of single-site step-by-step massage and multi-site collaborative massage, and optimize the massage parameter prediction model.

2. The method of claim 1, wherein, The method comprises: The free-flowing qi and collaterals robot is installed with a 3D camera, which continuously photographs the user's body to obtain user images, and uses a convolutional neural network for model training to identify the user's body shape feature data, including muscle structure and joint position and angle; green label stickers are used to mark the preset acupoints on the user's body, and the 3D visual recognition technology is used to identify the independent functional areas of the user's body based on the user's body shape feature data; according to human anatomy data, the independent functional areas with function dependence are determined, and the independent functional areas with function dependence are divided into function-dependent areas.

3. The method of claim 1, wherein, The free-flowing qi and collaterals robot is installed with a thermal imaging camera, which obtains the user's human thermal imaging pictures, performs image preprocessing operations on the user's human thermal imaging pictures, and stores the preprocessed human thermal imaging pictures into a human thermal imaging picture monitoring database; the image preprocessing operations include adjusting the brightness and contrast of the pictures; the human thermal imaging picture monitoring database is used to obtain the user's historical human thermal imaging pictures, mark the body fatigue area in the pictures, use a convolutional neural network for model training, and construct a body fatigue area identification model; the body fatigue area identification model is used to identify the user's body fatigue area according to the real-time obtained user's human thermal imaging pictures. The method comprises:

4. The method of claim 1, wherein, ​ The human body thermal imaging picture of the user and the user image are aligned to the same coordinate system through image registration technology, the independent functional area where the user body fatigue area is located is determined based on the identification result of the independent functional area of the user body and the identification result of the functional dependent area, and whether there is functional dependence between the independent functional areas is judged.

5. The method of claim 1, wherein, The massage mode of the body fatigue area is determined according to the independent functional area where the body fatigue area is located, and the massage mode includes multi-site cooperative massage and single-site step-by-step massage. The human body thermal imaging picture of the user body fatigue area in the independent functional area is converted into a temperature matrix by using the image processing method based on computer vision and numerical calculation, and the massage parameters of the independent functional area where the user body fatigue area is located are predicted in combination with the basic information data of the user, including:

6. The method of claim 1, wherein, The human body thermal imaging picture of the user body fatigue area in the independent functional area is converted into a temperature matrix by using the image processing method based on computer vision and numerical calculation, and the massage parameters of the independent functional area where the user body fatigue area is located are predicted in combination with the basic information data of the user, including: The massage parameters of the independent functional area where the user body fatigue area is located are determined by using a recurrent neural network to train a model and construct a massage parameter prediction model according to the basic information data of the user, the temperature matrix of the independent functional area where the body fatigue area is located, and the historical data of the body part where the independent functional area is located, and labeling the user massage parameters, and the massage parameters include the massage force, massage frequency, massage method and massage trajectory of the massage head. The massage path of the multiple free-flowing collaterals robots is simulated by using a motion trajectory simulation algorithm, whether there is an overlap risk in the massage path of the multiple free-flowing collaterals robots is judged, and the massage path with the overlap risk is adjusted locally, including: If the user's body fatigue area uses single-part step-by-step massage, the massage parameters of the independent functional areas where the user's body fatigue area is located are predicted, and the massage parameters of the different independent functional areas of the Qiangqi Tongluo robot are set, and each independent functional area is massaged step by step; if the user's body fatigue area uses multi-part cooperative massage, the massage trajectory data of the Qiangqi Tongluo robot massage head responsible for each independent functional area in the functional dependent area is obtained, a motion trajectory simulation algorithm is used to simulate the cooperative massage process of multiple Qiangqi Tongluo robots, the position of the massage head at each time is obtained, and it is judged whether there is an overlap risk in the massage path when multiple Qiangqi Tongluo robots are used for cooperative massage; if there is an overlap risk, the massage path with conflict is locally adjusted to obtain the adjusted massage head massage trajectory data; according to the predicted massage parameters of the independent functional areas where the user's body fatigue area is located, the massage parameters of the Qiangqi Tongluo robot responsible for each independent functional area in the functional dependent area are set in combination with the adjusted massage head massage trajectory data, and each independent functional area in the functional dependent area is cooperatively massaged by multiple Qiangqi Tongluo robots.

7. The method of claim 1, wherein, The massage effect of single-part step-by-step massage and multi-part cooperative massage is evaluated according to the user's body state data before and after massage, and the massage parameter prediction model is optimized, including: The user's body state data before and after massage is obtained, the massage effect of single-part step-by-step massage and multi-part cooperative massage is evaluated, and the user's body state data includes muscle relaxation degree, relief condition of fatigue area and comfort score; if the massage effect is lower than the preset standard, the massage parameters of the independent functional areas are adjusted based on the user's body state data, and the massage parameter prediction model is optimized to obtain the optimized massage parameters of the independent functional areas; the optimized massage parameters of the independent functional areas are implemented, the user's body state data before and after massage is re-obtained, and the massage parameter prediction model is continuously optimized.

Images