Training auxiliary system for tiger play action of five birds

Abstract

The invention relates to the technical field of training assistance, and provides a training assistance system for five-poultry paddling and tiger paddling actions, which comprises a data acquisition module, an action recognition module, a whole body coordination evaluation module and a user interface module. The data acquisition module is used for acquiring motion data when a user executes a tiger play action through a plurality of sensor groups; the action recognition module performs preprocessing and feature extraction on the collected data, decomposes the tiger play action into a basic action sequence, and recognizes the matching condition of the current action and a standard action template through a deep learning network; the whole body coordination evaluation module establishes a biomechanical model based on a reverse dynamics method, calculates multi-dimensional coordination indexes including joint angle change, mechanical load distribution and dynamic stability, and generates an action coordination score through a support vector regression model; and the user interface module is used for displaying identification and evaluation results and providing improvement suggestions when the actions do not conform to specifications or the coordination is insufficient.

Description

Technical Field

[0001] The present invention relates to the technical field of training aids, and more specifically, to a training aid system for the tiger-play movement of the Five-Animal Exercises. Background Art

[0002] The Five-Animal Exercises is a traditional Chinese health care method created by the Eastern Han Dynasty medical scientist Hua Tuo. It imitates the movement characteristics of five animals: tiger, deer, bear, ape, and bird. By combining physical movement with breathing regulation, it can achieve the effects of strengthening the body and regulating the mind and body. Among them, the tiger-play, as an important part of the Five-Animal Exercises, mainly imitates the movements of the tiger such as pouncing, twisting, squatting, and grasping, emphasizing the explosiveness and fluency of the movements. These movements not only require participants to have high physical flexibility and strength but also need to maintain good overall coordination during the movement process. Therefore, the tiger-play has important application values in both traditional health care and modern sports training.

[0003] With the development of science and technology, many sports training aid systems have emerged, especially in the fields of action recognition and supervised training. Systems based on sensors and virtual reality (VR) technology have gradually become the mainstream. For example, training devices that collect human motion data through wearable devices and provide real-time feedback to users in combination with virtual environments. These systems are widely used in rehabilitation training, physical education, and health management, and can effectively improve the interactivity and efficiency of training. In addition, the progress of action decomposition and recognition technology enables training systems to accurately analyze complex action sequences and provide users with detailed evaluations of action standardization.

[0004] Document 1 (Smith JD, et al., Virtual Reality-Based Gait Training System for Stroke Rehabilitation. Rehabilitation Science, 2018) proposed a cyber-physical system that combines virtual reality (VR) and intelligent sensors for action recognition and training supervision. This system collects the motion data of users through low-cost sensors, uses the non-negative matrix factorization algorithm to recognize gait and gesture actions, and provides real-time feedback to users in combination with virtual scenarios. This technology has significant advantages in improving action recognition efficiency and enhancing user experience. However, this system mainly focuses on the recognition of local action features and has not deeply explored the comprehensive evaluation of overall coordination, especially its application in complex action sequences is relatively limited.

[0005] Reference 2 (Chang M, et al., Whole-Body Coordination in Complex Dance Sequences: Implications for Skill Acquisition. Human Movement Science, 2020) studied the changes in whole-body coordination of dancers with different skill levels in complex dance tasks, and verified the phenomenon of the movement freedom changing from "freezing" to "thawing" during the skill improvement process. The research results show that as the skill level increases, the movement of dancers becomes more fluent, and the coordination of each joint is significantly improved. It can be seen that whole-body coordination is crucial for the execution of complex movements, especially in evaluating skill differences and movement fluency.

[0006] Although the existing technologies have made certain progress in the fields of action recognition and training assistance, the support for complex traditional actions such as the Tiger Play is still insufficient. On the one hand, the action recognition of existing technologies mostly targets single actions or simple sequences, and it is difficult to accurately capture the complex action features involving multi-joint cooperation and dynamic changes in the Tiger Play, such as the sense of power in pouncing and the flexibility of trunk torsion; on the other hand, the existing systems lack a comprehensive evaluation of whole-body coordination and usually rely on single-point monitoring or simple evaluation methods, unable to comprehensively quantify multi-dimensional indicators. In addition, for the traditional action of the Tiger Play, most of the existing technologies lack a specific action template library and a biomechanical model matching its characteristics, resulting in obvious deficiencies in the adaptability and accuracy of training guidance. Summary of the Invention

[0007] To overcome the above defects of the existing technology, the present invention provides a training assistance system for the Tiger Play actions of the Five-Animal Exercises. By combining deep learning networks and biomechanical modeling techniques, the action recognition module decomposes and recognizes the Tiger Play action sequence, the whole-body coordination evaluation module calculates multi-dimensional indicators, the action coordination is evaluated based on the support vector regression model and improvement suggestions are generated, and the action standardization and coordination scores are fed back in real time, solving the problem that the coordination of the Tiger Play actions cannot be comprehensively quantified in the existing technology.

[0008] To achieve the above object, the present invention provides the following technical solutions:

[0009] A training assistance system for the tiger-play movement of the Five-Animal Exercises, comprising a data acquisition module and a user interface module, including an action recognition module and a whole-body coordination evaluation module; wherein, the data acquisition module is connected to the action recognition module, the action recognition module is connected to the whole-body coordination evaluation module, and the whole-body coordination evaluation module is connected to the user interface; the action recognition module is used to decompose the tiger-play movement into an ordered combination of basic action sequences and identify them based on a deep learning network; the whole-body coordination evaluation module is used to evaluate the limb coordination of the user when performing the tiger-play movement, including a biomechanical modeling unit, a coordination index calculation unit, and a coordination evaluation unit. The biomechanical modeling unit establishes a biomechanical model of the tiger-play movement based on the inverse dynamics method; the coordination index calculation unit is used to calculate the joint angle consistency index, the mechanical load distribution index, and the dynamic stability index; the coordination evaluation unit calculates the action coordination score based on a support vector regression model. Among them, the formula for calculating the dynamic stability index is:

[0010]

[0011] In the formula, A hull is the convex hull area of the projection of the center-of-gravity trajectory, A base is the support area, a j is the acceleration vector of the j-th joint, ω j is the angular velocity vector of the j-th joint, g is the acceleration due to gravity, and M is the total number of joints in the human body model.

[0012] As a further solution of the present invention, the biomechanical modeling unit includes the following steps:

[0013] Step S1: Model the human body as a linkage system connected by joints. Virtual spring dampers are set at the shoulder, hip, knee, and ankle joints of the linkage system to simulate the elasticity and damping of the joints. A three-dimensional torsional joint is established at the trunk, and a three-degree-of-freedom spherical hinge structure is established at the shoulder.

[0014] Step S2: Perform the first-stage inverse dynamics calculation. According to the kinematic parameters output by the action recognition module, the angular acceleration and linear acceleration of each joint are calculated using the numerical difference method.

[0015] Step S3: Perform the second-stage inverse dynamics calculation. Using the Newton-Euler recursive algorithm, calculate the forces and torques of each joint step by step in the order from the foot to the head.

[0016] Step S4: Extract the peak values, average values, and integral values of the forces and torques of each joint during the complete action cycle.

[0017] As a further solution of the present invention, the coordination index calculation unit calculates the joint angle consistency index based on the movement coordination of adjacent joints, and its formula is:

[0018]

[0019] In the formula, θ p,1 (t) is the angle of the first joint in the pth pair of adjacent joints at time t, and θ p,2 (t) is the angle of the second joint in the pth pair of adjacent joints at time t. T is the total length of the sampling time, ω p is the angular velocity vector of the pth pair of adjacent joints, P is the total number of adjacent joint pairs in the human body model, and p is the number of the adjacent joint pair.

[0020] As a further solution of the present invention, the coordination index calculation unit calculates the mechanical load distribution index based on the distribution of joint forces, and its formula is:

[0021]

[0022] In the formula, T j is the torque vector of the jth joint, ω j is the angular velocity vector of the jth joint, F j is the force vector of the jth joint, and M is the total number of joints in the human body model.

[0023] As a further solution of the present invention, the working steps of the coordination evaluation unit include:

[0024] Step A1: Construct a training set based on the action data of professional practitioners of the Tiger Play. Each training sample includes the joint angle consistency index, the mechanical load distribution index, the dynamic stability index, the peak joint torque, the peak joint pressure, and the peak joint shear force; and use the coordination score obtained by the professional coach's scoring as the training label;

[0025] Step A2: Train a support vector machine regression model, use the radial basis kernel function and select parameters using the cross-validation method;

[0026] Step A3: Input the feature vector output by the coordination index calculation unit into the trained support vector regression model; calculate the action coordination score based on the support vector regression model;

[0027] Step A4: Generate improvement suggestions when the action coordination score is less than the set threshold.

[0028] As a further solution of the present invention, the action recognition module includes a motion decomposition unit, wherein the motion decomposition unit is used to decompose the input tiger-play action into four basic actions: pouncing, torso twisting, squatting, and grasping; and establish a standard template library for basic actions, including the standard limb positions, motion trajectories, speed change characteristics, and action sequence standards of each basic action.

[0029] As a further solution of the present invention, the action recognition module further includes a feature extraction unit for preprocessing and feature extraction of the data of different sensor groups; wherein, the data of the first sensor group is used to obtain the acceleration characteristics of the shoulder pouncing action, the data of the second sensor group is used to obtain the angular velocity characteristics of the torso twisting action, the data of the third sensor group is used to obtain the acceleration characteristics of the double-knee squatting action, the data of the fourth sensor group is used to obtain the electromyography characteristics of the two-hand grasping action, and the data of the fifth sensor group is used to obtain the three-dimensional spatial position characteristics of the body joint points.

[0030] As a further solution of the present invention, the action recognition module further includes an action recognition unit, and the action recognition unit constructs an action evaluation model based on a deep learning network; compares the currently executed basic action with the action characteristics in the standard template library in real time, evaluates the action standardization and determines whether the action sequence meets the standard sequence requirements; when the action execution meets the standardization requirements and the sequence meets the standard sequence, transmits the action data to the whole-body coordination evaluation module for further analysis; otherwise, prompts the user to adjust the action through the user interface module.

[0031] As a further solution of the present invention, the data acquisition module includes a first sensor group arranged on the user's shoulders to collect the acceleration and angular velocity information of the shoulder pouncing action; a second sensor group arranged at a specific position of the user's spine to collect the angular velocity information of the torso twisting action; a third sensor group arranged on the user's knees to collect the acceleration information of the squatting action; a fourth sensor group arranged on the palms of the user's hands to collect the electromyography signals of the grasping action; and a fifth sensor group including infrared reflective marker points distributed at each joint point of the user to collect the three-dimensional spatial coordinate data of the body joint points.

[0032] As a further solution of the present invention, the user interface module is used to display the action standardization determination result of the action recognition module and the action coordination score of the whole-body coordination evaluation module, and output a prompt message when the action does not conform to the standard sequence or the coordination score is insufficient.

[0033] Compared with the prior art, the beneficial effects of a training assistance system for the tiger-play action of the Five-Animal Exercises of the present invention are as follows:

[0034] The present invention comprehensively obtains various motion data of the shoulders, spine, knees, hands, and body key points through a data acquisition module, and uses an action recognition module to decompose the movements of the Tiger Play into basic action sequences, and combines a deep learning network to achieve accurate recognition and real-time feedback. In this way, it can dynamically capture the multi-dimensional characteristics of the complex movements of the Tiger Play, such as the sense of power in pouncing and the flexibility of trunk torsion. Existing technologies mostly only support the recognition of single actions or simple sequences, and it is difficult to cover the characteristics of multi-joint cooperation and dynamic changes involved in the movements of the Tiger Play, resulting in insufficient adaptability of training guidance and being unable to comprehensively support the practice and evaluation of complex traditional movements.

[0035] The present invention innovatively introduces a whole-body coordination evaluation module, calculates multi-dimensional indicators such as joint angle consistency, mechanical load distribution, and dynamic stability through biomechanical modeling, and generates a coordination score based on a support vector regression model to provide users with accurate analysis of action standardization and coordination. Existing technologies usually adopt single-point monitoring or simple evaluation methods, lacking the comprehensive evaluation ability of whole-body coordination. Especially in complex action sequences, it is impossible to comprehensively quantify the user's movement coordination and it is difficult to provide targeted improvement suggestions for users. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 It is a schematic diagram of the hardware composition of the virtual reality action capture system in Document 1.

[0037] Figure 2 It is a skeletal sequence diagram of the 'Alternate Basic' action of the cha-cha dance.

[0038] Figure 3 It is a distribution diagram of kinematic marker points of the human body.

[0039] Figure 4 It is a schematic diagram of the structure of a training assistance system for the Tiger Play movements of the Five-Animal Exercises according to the present invention.

[0040] Figure 5 It is a schematic diagram of the user interface of a training assistance system for the Tiger Play movements of the Five-Animal Exercises according to the present invention.

[0041] Figure 6 It is a schematic diagram of the human body link system of a training assistance system for the Tiger Play movements of the Five-Animal Exercises according to the present invention.

[0042] In the figure, HMD: Head-Mounted Display; Digital Glove: Digital Glove; Glove Sensor Model: Glove Sensor Model; Musical count: Musical Count; Back Basic: Back Basic Step; Hip Twist Chasse: Hip Twist Chasse Step; Forward Check Basic: Forward Check Basic Step; Ronde Chasse: Ronde Chasse Step. Detailed Implementation Manner

[0043] Next, the technical solutions in this embodiment will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0044] Embodiment 1

[0045] A training assistance system for the tiger play action of Wuqinxi includes a data acquisition module, an action recognition module, a whole-body coordination evaluation module, and a user interface module.

[0046] In the embodiments of the present invention, the data acquisition module is connected to the action recognition module, the action recognition module is connected to the whole-body coordination evaluation module, and the whole-body coordination evaluation module is connected to the user interface.

[0047] The data acquisition module in the embodiments of the present invention is used to collect sensor data when the user performs the tiger play action, including a first sensor group arranged on the user's shoulders, including a three-axis acceleration sensor and a three-axis gyroscope, for collecting the acceleration and angular velocity data of the shoulder striking action; a second sensor group arranged at the T4, T8, and T12 positions of the user's spine, including a three-axis gyroscope, for collecting the angular velocity data of the trunk torsion action and the relative movement data between each segment; a third sensor group arranged on the user's knees, including a three-axis acceleration sensor, for collecting the acceleration data of the squatting action; a fourth sensor group including electromyogram sensors arranged on the palms of the user's hands, for collecting the electromyographic signals of the grasping action; a fifth sensor group including infrared reflective markers arranged at the joint positions of the user, for collecting the three-dimensional space coordinate data of the body joint points.

[0048] In the embodiments of the present invention, accelerometers / gyroscopes are additionally installed at key parts such as the shoulders, spine (T4 / T8 / T12), and knees, aiming to capture the key movement characteristics (such as pouncing, twisting, and squatting) of the Tiger Play with high precision and in real time. The kinematic data (position, angle, angular velocity) of other joints in the whole body (such as the hips, ankles, elbows, wrists, etc.) are mainly obtained by the fifth sensor group and the corresponding optical detection devices and are solved through numerical differentiation or Kalman filtering, so as to obtain the complete spatio-temporal information of the whole body. In this way, the force / moment of each joint can be calculated by inverse dynamics, meeting the requirements of biomechanical modeling and coordination evaluation of the present invention.

[0049] The action recognition module is used to decompose the Tiger Play actions into an ordered combination of basic action sequences and recognize them based on a deep learning network.

[0050] The action recognition module in the embodiments of the present invention includes a motion decomposition unit. Among them, the motion decomposition unit is used to decompose the input Tiger Play actions into four basic actions: pouncing, torso twisting, squatting, and grasping; the decomposition of the basic actions is based on the extraction of expert knowledge of the Tiger Play action schools and the analysis of action characteristics; a standard template library of basic actions is established, where the standard template library includes: the standard limb positions, motion trajectories, speed change characteristics of each basic action, and the action sequence standard of "squat first and then twist, twist first and then pounce, and grasp when pouncing"; the standard template library is input into the action recognition unit.

[0051] The feature extraction unit in the embodiments of the present invention is used to receive the sensor data collected by the data acquisition module; perform low-pass filtering on the acceleration and angular velocity data collected by the first sensor group to extract the acceleration peak value and acceleration direction vector of the shoulder pouncing action; perform wavelet transform denoising on the angular velocity data collected by the second sensor group to extract the angular velocity peak values at the positions of T4, T8, and T12 of the torso; perform mean filtering on the acceleration data collected by the third sensor group to extract the acceleration peak value of the double-knee squatting action; perform rectification and smoothing processing on the electromyography signals collected by the fourth sensor group to extract the root mean square value of the electromyography signals of the two-handed palm grasping action; perform Kalman filtering on the three-dimensional space coordinate data collected by the fifth sensor group to extract the spatial position data and center of gravity position data of the joint points.

[0052] The action recognition unit in the embodiments of the present invention constructs an action evaluation model based on a deep learning network; compares the currently executed basic actions with the action characteristics in the standard template library in real time to evaluate the action normality; judges whether the action sequence meets the standard order requirements; when the action execution meets the normality requirements and the sequence meets the standard order, the action data is passed to the whole-body coordination evaluation module for further analysis; otherwise, the user is prompted by the user interface module to adjust the action.

[0053] The action recognition unit in the embodiment of the present invention includes a cascaded CNN and LSTM network; wherein, the CNN is used to extract spatial features from the sensor features output by the feature extraction unit; the LSTM is used to extract temporal features from the spatial features.

[0054] The whole-body coordination evaluation module is used to evaluate the limb coordination of the user when performing the Tiger Play action, including: a biomechanical modeling unit, a coordination index calculation unit, and a coordination evaluation unit.

[0055] The biomechanical modeling unit in the embodiment of the present invention is used to establish a biomechanical model of the Tiger Play action based on the inverse dynamics method, including the following steps:

[0056] Step S1, model the human body as a linkage system connected by joints, set virtual spring dampers at the shoulder, hip, knee, and ankle joints of the linkage system to simulate the elasticity and damping of the joints, establish a three-dimensional torsion joint at the trunk, and establish a three-degree-of-freedom ball hinge structure at the shoulder;

[0057] Step S2, perform the first-stage inverse dynamics calculation, and calculate the angular acceleration and linear acceleration of each joint by using the numerical difference method according to the kinematic parameters output by the action recognition module;

[0058] Step S3, perform the second-stage inverse dynamics calculation, and use the Newton-Euler recursive algorithm to calculate the forces and torques of each joint step by step in the order from the foot to the head;

[0059] Step S4, extract the peak values, average values, and integral values of the forces and torques of each joint during the complete action cycle.

[0060] In the embodiment of the present invention, after performing Kalman filtering on the three-dimensional space coordinate data collected by the fifth sensor group, calculate the angles of each joint based on the spatial position relationship between adjacent joint points; calculate the angular velocity of each joint by using the central difference method; calculate the angular acceleration of each joint by using the five-point difference method.

[0061] The coordination index calculation unit is used to calculate the joint angle consistency index, the mechanical load distribution index, and the dynamic stability index.

[0062] The coordination index calculation unit in the embodiment of the present invention calculates the joint angle consistency index based on the motion coordination of adjacent joints, and the calculation formula of the joint angle consistency index is:

[0063]

[0064] where θ p,1θ(t) is the angle of the first joint in the p-th pair of adjacent joints at time t. p,2 θ(t) is the angle of the second joint in the p-th pair of adjacent joints at time t, T is the total sampling time length, ω p is the angular velocity vector of the p-th pair of adjacent joints, P is the total number of adjacent joint pairs in the human body model, and p is the number of the adjacent joint pair.

[0065] The adjacent joint pairs in the embodiments of the present invention include: head-neck, neck-upper torso, upper torso-middle torso, middle torso-lower torso, lower torso-left thigh, left thigh-left calf, left calf-left foot, lower torso-right thigh, right thigh-right calf, right calf-right foot, upper torso-left upper arm, left upper arm-left forearm, left forearm-left palm, upper torso-right upper arm, right upper arm-right forearm, right forearm-right palm.

[0066] The coordination index calculation unit in the embodiments of the present invention calculates a mechanical load distribution index based on the distribution of joint forces, and the calculation formula of the mechanical load distribution index is:

[0067]

[0068] In the formula, T j is the moment vector of the j-th joint, ω j is the angular velocity vector of the j-th joint, F j is the force vector of the j-th joint, and M is the total number of joints in the human body model.

[0069] The formula for calculating the dynamic stability index in the embodiments of the present invention is:

[0070]

[0071] In the formula, A hull is the convex hull area of the projection of the center of gravity trajectory, A base is the support area, a j is the acceleration vector of the j-th joint, ω j is the angular velocity vector of the j-th joint, g is the acceleration due to gravity, and M is the total number of joints in the human body model.

[0072] The coordination evaluation unit in the embodiments of the present invention is used to calculate an action coordination score based on a support vector regression model.

[0073] The working steps of the coordination evaluation unit in the embodiments of the present invention include:

[0074] Step A1: Construct a training set based on the motion data of professional practitioners of the Tiger Play. Each training sample includes indicators of joint angle consistency, mechanical load distribution, dynamic stability, peak joint torque, peak joint pressure, and peak joint shear force; and use the coordination score obtained by professional coaches' scoring as the training label.

[0075] Step A2: Train a support vector machine regression model, select the radial basis kernel function as the kernel function; use the cross-validation method to select the optimal penalty factor and kernel function parameters; use the sequential minimal optimization algorithm to train the support vector regression model so that the root mean square error between the score output by the model and the score given by professional coaches is less than the preset threshold.

[0076] Step A3: Input the feature vector output by the coordination index calculation unit into the trained support vector regression model; calculate the motion coordination score from 0 to 100 based on the support vector regression model; among them, 90 - 100 points indicate excellent motion coordination, 80 - 89 points indicate good, 70 - 79 points indicate medium, 60 - 69 points indicate passing, and below 60 points indicate failing.

[0077] Step A4: When the motion coordination score is less than the set threshold, generate specific improvement suggestions based on the following rules: when the joint angle consistency index is lower than the first preset threshold, prompt to optimize the motion rhythm of adjacent joints; when the mechanical load distribution index is lower than the second preset threshold, prompt to adjust the force application sequence and force transmission path; when the dynamic stability index is lower than the third preset threshold, prompt to adjust the center of gravity control and balance position; when the peak joint torque, peak pressure, or peak shear force exceeds the corresponding safety threshold, prompt to reduce the amplitude of the corresponding motion; display the scoring result and specific improvement suggestions on the user interface.

[0078] The user interface module is used to display the motion standard determination result of the motion recognition module and the motion coordination score of the whole body coordination evaluation module, and output a prompt message when the motion does not conform to the standard sequence or the coordination score is insufficient.

[0079] Example 2

[0080] The biomechanical modeling unit is used to establish a biomechanical model of the Tiger Play based on the inverse dynamics method, including the following steps:

[0081] Step S1: As Figure 6As shown, the human body is modeled as a linkage system of 17 rigid body segments connected by 16 joints. The rigid body segments include: head, neck, upper torso, middle torso, lower torso, left and right upper arms, left and right forearms, left and right palms, pelvis, left and right thighs, left and right calves, left and right feet. Virtual spring dampers are set at the shoulder, hip, knee and ankle joints of the linkage system to simulate the elasticity and damping of the joints, and a three-dimensional torsional joint is established at the torso part, and a three-degree-of-freedom ball hinge structure is established at the shoulder.

[0082] Step S2, perform the first-stage inverse dynamics calculation. According to the kinematic parameters output by the action recognition module, use the numerical difference method to calculate the angular acceleration and linear acceleration of each joint. The kinematic parameters include: the three-dimensional spatial position coordinates of the joint, joint angle, and angular velocity.

[0083] Step S3, perform the second-stage inverse dynamics calculation. Use the Newton-Euler recurrence algorithm to calculate the force and moment of each joint step by step in the order from the foot to the head. The force and moment of each joint satisfy:

[0084] F i =m i *l i +m i *g;

[0085] M i =I i *α i +ω i ×(I i *ω i );

[0086] Among them, F i is the force of the i-th joint, M i is the moment of the i-th joint, m i is the mass of the distal segment connected by the joint, l i is the linear acceleration of the joint, g is the acceleration due to gravity, I i is the moment of inertia of the distal segment connected by the joint, α i is the angular acceleration of the joint, ω i is the angular velocity of the joint;

[0087] Step S4, extract the peak value, average value and integral value of the force and moment of each joint during the complete action cycle. For the shoulder joint, set the moment threshold to; for the spine, set the pressure threshold to; for the knee joint, set the shear force threshold to.

[0088] The following is a Python code example for establishing a biomechanical model of the Tiger Play movement based on the inverse dynamics method. Please note that this example is only a starting point and may need to be adjusted according to the actual situation and device interface in practical applications;

[0089]

[0090]

[0091]

[0092]

[0093]

[0094]

[0095]

[0096] This code is only for example, and appropriate modifications and adjustments need to be made according to the specific situation in practical applications.

[0097] The above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art within the technical scope disclosed by this application can easily think of changes or substitutions, which should be covered within the protection scope of this application. Therefore, the protection scope of this application should be subject to the protection scope of the claims.

[0098] Finally: The above is only the preferred embodiment of the present invention and is not used to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A training auxiliary system for the tiger play of the Five Animal Play, comprising a data acquisition module and a user interface module, characterized in that: It includes an action recognition module and a whole-body coordination evaluation module; wherein the data acquisition module is connected to the action recognition module, the action recognition module is connected to the whole-body coordination evaluation module, and the whole-body coordination evaluation module is connected to the user interface; the action recognition module is used to decompose the tiger play action into an ordered combination of basic action sequences and identify them based on a deep learning network; the whole-body coordination evaluation module is used to evaluate the limb coordination of the user when performing the tiger play action, and includes a biomechanical modeling unit, a coordination index calculation unit and a coordination evaluation unit, wherein the biomechanical modeling unit establishes a biomechanical model of the tiger play action based on an inverse dynamics method; the coordination index calculation unit is used to calculate a joint angle consistency index, a mechanical load distribution index and a dynamic stability index; the coordination evaluation unit calculates the action coordination score based on a support vector regression model, wherein the formula for calculating the dynamic stability index is: In the formula, A hull is the convex hull area of ​​the projection of the centroid trajectory, A base is the support area, a j is the acceleration vector of the j-th joint, ωj is the angular velocity vector of the j-th joint, g is the gravitational acceleration, and M is the total number of joints in the human body model.

2. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The biomechanical modeling unit comprises the following steps: Step S1, modeling the human body as a linkage system connected by joints, setting virtual spring dampers at the shoulder, hip, knee and ankle joints of the linkage system to simulate the elasticity and damping of the joints, establishing a three-dimensional torsion joint at the torso, and establishing a three-degree-of-freedom ball hinge structure at the shoulder; Step S2, performing the first stage of inverse dynamics calculation, and calculating the angular acceleration and linear acceleration of each joint using a numerical difference method according to the kinematic parameters output by the action recognition module; Step S3, executing the second stage of inverse dynamics calculation, using the Newton-Euler recursive algorithm, and calculating the forces and moments of each joint step by step in the order from the foot to the head; Step S4, extracting the peak value, mean value and integral value of the force and torque of each joint in the complete motion cycle.

3. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The coordination index calculation unit calculates the joint angle consistency index based on the motion synergy of adjacent joints, and its formula is: In the formula, θ p，1 (t) is the first joint angle of the pth pair of adjacent joints at time t, θ p，2 (t) is the second joint angle of the pth pair of adjacent joints at time t, T is the total length of sampling time, ω p is the angular velocity vector of the pth pair of adjacent joints, P is the total number of adjacent joint pairs in the human body model, and p is the number of the adjacent joint pair.

4. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The coordination index calculation unit calculates the mechanical load distribution index based on the distribution of joint forces, and the formula is: Where, T j is the torque vector of the jth joint, ω j is the angular velocity vector of the jth joint, F j is the force vector of the jth joint, and M is the total number of joints in the human body model.

5. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The working steps of the coordination assessment unit include: Step A1, constructing a training set based on the action data of professional tiger play practitioners, each training sample includes joint angle consistency index, mechanical load distribution index, dynamic stability index, joint torque peak, joint pressure peak and joint shear force peak; and using the coordination score obtained by professional coaches as a training label; Step A2, training a support vector machine regression model, using a radial basis kernel function and selecting parameters using a cross-validation method; Step A3, inputting the feature vector output by the coordination index calculation unit into a trained support vector regression model; and calculating the movement coordination score based on the support vector regression model; Step A4: When the movement coordination score is less than a set threshold, an improvement suggestion is generated.

6. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The action recognition module includes a motion decomposition unit, wherein the motion decomposition unit is used to decompose the input tiger play action into four basic actions: pouncing, trunk twisting, squatting and grasping; and establish a basic action standard template library, including the standard limb position, motion trajectory, speed change characteristics and action sequence standards of each basic action.

7. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The action recognition module also includes a feature extraction unit, which is used to preprocess and extract features from data from different sensor groups; wherein the data from the first sensor group is used to obtain the acceleration characteristics of the shoulder lunging action, the data from the second sensor group is used to obtain the angular velocity characteristics of the trunk twisting action, the data from the third sensor group is used to obtain the acceleration characteristics of the double-knee squatting action, the data from the fourth sensor group is used to obtain the electromyographic characteristics of the two-handed grasping action, and the data from the fifth sensor group is used to obtain the three-dimensional spatial position characteristics of the body joints.

8. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The action recognition module also includes an action recognition unit, which builds an action evaluation model based on a deep learning network; compares the currently executed basic action with the action features in the standard template library in real time, evaluates the action standardization and determines whether the action sequence meets the standard sequence requirements; when the action execution meets the standard requirements and the sequence meets the standard sequence, the action data is transmitted to the whole body coordination evaluation module for further analysis; otherwise, the user is prompted to adjust the action through the user interface module.

9. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The data acquisition module includes a first sensor group, which is arranged on the user's shoulders to collect acceleration and angular velocity information of the shoulder punching action; a second sensor group, which is arranged at a specific position of the user's spine to collect angular velocity information of the torso twisting action; a third sensor group, which is arranged on the user's knees to collect acceleration information of the squatting action; and a fourth sensor group, which is arranged on the palms of the user's hands to collect electromyographic signals of the grasping action; The fifth sensor group includes infrared reflective marking points distributed at each joint of the user to collect three-dimensional spatial coordinate data of the body joints.

10. A training auxiliary system for the tiger play of the Five Animal Play according to claim 1, characterized in that: The user interface module is used to display the action normativeness determination result of the action recognition module and the action coordination score of the whole body coordination assessment module, and output prompt information when the action does not meet the standard sequence or the coordination score is insufficient.

Images