Motion planning generation method, system and equipment based on AI driving

Abstract

The invention discloses a motion plan generation method, system and device based on AI driving, and relates to the technical field of data processing. The method comprises the following steps: obtaining an initial motion scheme; performing height development prediction association optimization based on the initial exercise scheme to obtain a first exercise scheme; performing acute injury risk prediction on the target user to obtain an exercise acute injury risk coefficient; performing chronic strain risk prediction to obtain an exercise chronic strain risk coefficient; performing excessive fatigue risk prediction to obtain an excessive fatigue risk coefficient of exercise; performing exercise risk optimization adjustment on the first exercise scheme to obtain a second exercise scheme; and performing nutritional feature compensation on the second exercise scheme to obtain a third exercise scheme, and performing exercise execution rewarding on the target user based on the third exercise scheme. The technical problem of insufficient safety of the exercise scheme caused by lack of systematic risk assessment in exercise planning in the prior art is solved, and the technical effect of improving the safety of the exercise scheme is achieved.

Description

Technical Field

[0001] This invention relates to the field of data processing technology, and specifically to an AI-driven motion planning generation method, system, and device. Background Technology

[0002] AI-powered vertical jumpers, designed specifically for children, can guide them through continuous and regular vertical jump training at an adjustable target height. This can, to some extent, promote growth hormone secretion and enhance bone load-bearing capacity and lower limb muscle strength. However, because children's bones and joints are not yet fully developed, excessive training load or inappropriate rhythm can easily lead to acute sports injuries, chronic strain, or over-fatigue, negatively impacting their growth and development. Existing exercise planning methods typically set fixed parameters based on age or weight, lacking the ability to dynamically assess and adjust for individual differences and risk factors, making it difficult to achieve truly safe and effective personalized training guidance. Summary of the Invention

[0003] This application provides an AI-driven method, system, and device for generating motion plans, which solves the technical problem that the lack of systematic risk assessment in motion planning in the prior art leads to insufficient safety of motion plans.

[0004] A first aspect of this application provides an AI-driven motion planning generation method, the method comprising: Exercise planning is performed based on the individual characteristics and exercise goals of the target user to obtain an initial exercise plan. This initial plan is then optimized by predicting and associating with height development to obtain a first exercise plan. Based on this first exercise plan, the target user's acute injury risk is predicted to obtain an acute injury risk coefficient. Similarly, the target user's chronic strain risk is predicted based on the first exercise plan to obtain a chronic strain risk coefficient. Furthermore, the target user's over-fatigue risk is predicted based on the first exercise plan to obtain an over-fatigue risk coefficient. The first exercise plan is then optimized and adjusted based on the acute injury risk coefficient, the chronic strain risk coefficient, and the over-fatigue risk coefficient to obtain a second exercise plan. Finally, the second exercise plan is compensated for nutritional features using an AI-guided nutrition model to obtain a third exercise plan. Finally, the target user is rewarded for exercising based on the third exercise plan.

[0005] A second aspect of this application provides an AI-driven motion planning generation system, the system comprising: The exercise planning module performs exercise planning based on the target user's individual characteristics and exercise goals to obtain an initial exercise plan. It then optimizes the initial exercise plan by predicting and associating it with height development to obtain a first exercise plan. The first risk prediction module predicts the target user's acute injury risk based on the first exercise plan, obtaining an acute injury risk coefficient. The second risk prediction module predicts the target user's chronic strain risk based on the first exercise plan, obtaining a chronic strain risk coefficient. The third risk prediction module predicts the target user's over-fatigue risk based on the first exercise plan, obtaining an over-fatigue risk coefficient. The optimization module optimizes and adjusts the first exercise plan based on the acute injury risk coefficient, the chronic strain risk coefficient, and the over-fatigue risk coefficient, obtaining a second exercise plan. The compensation module compensates the second exercise plan for nutritional characteristics based on an AI-guided nutrition model, obtaining a third exercise plan, and rewards the target user for exercise performance based on the third exercise plan.

[0006] A third aspect of this application provides an electronic device, comprising: a memory for storing executable instructions; and a processor for implementing the AI-driven motion planning generation method provided in this application when executing the executable instructions stored in the memory.

[0007] One or more technical solutions provided in this application have at least the following technical effects or advantages: First, exercise planning is performed based on the individual characteristics and exercise goals of the target user to obtain an initial exercise plan. This initial plan is then optimized through height development prediction to obtain the first exercise plan. Next, based on the first exercise plan, the target user's acute injury risk is predicted, yielding an acute injury risk coefficient; the target user's chronic strain risk is predicted, yielding a chronic strain risk coefficient; and the target user's over-fatigue risk is predicted, yielding an over-fatigue risk coefficient. Then, based on the acute injury risk coefficient, chronic strain risk coefficient, and over-fatigue risk coefficient, the first exercise plan is optimized and adjusted to obtain the second exercise plan. Finally, the second exercise plan is supplemented with nutritional features using an AI-guided nutrition model to obtain the third exercise plan, and the target user is rewarded for exercising based on the third exercise plan. This approach solves the technical problem of insufficient safety in exercise plans due to the lack of systematic risk assessment in existing technologies, achieving the technical effect of improving the safety and individual suitability of exercise plans while ensuring the achievement of exercise goals. Attached Figure Description

[0008] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0009] Figure 1 This is a schematic diagram of the AI-driven motion planning generation method provided in an embodiment of this application; Figure 2 This is a schematic diagram of the structure of an AI-driven motion planning and generation system provided in an embodiment of this application; Figure 3 This is a schematic diagram of the structure of an exemplary electronic device of this application.

[0010] Explanation of reference numerals in the attached drawings: motion planning module 11, first risk prediction module 12, second risk prediction module 13, third risk prediction module 14, optimization module 15, compensation module 16, processor 21, memory 22, input device 23, output device 24. Detailed Implementation

[0011] This application addresses the technical problem of insufficient safety in existing motion planning due to the lack of systematic risk assessment in motion planning by providing an AI-driven motion planning generation method, system, and device.

[0012] The technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only a part of the embodiments of this application, and not all of them. All other embodiments obtained by those skilled in the art based on the embodiments of this application without creative effort are within the scope of protection of this application.

[0013] It should be noted that the terms "comprising" and "having" are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or server that includes a series of steps or units is not necessarily limited to those steps or units that are explicitly listed, but may include other steps or modules that are not explicitly listed or that are inherent to these processes, methods, products, or devices.

[0014] Example 1, as Figure 1 As shown in the embodiments of this application, an AI-driven motion planning generation method is provided, wherein the method includes: Exercise planning is performed based on the individual characteristics and exercise goals of the target user to obtain an initial exercise plan. Then, height development prediction and correlation optimization are performed based on the initial exercise plan to obtain the first exercise plan.

[0015] In this embodiment, the individual characteristics and exercise goals of the target user are input into the AI ​​height-boosting device to obtain an initial exercise plan. The individual characteristics include, but are not limited to, height, weight, gender, age, bone age, past exercise habits, physiological cycle status, growth and development level, exercise ability assessment indicators, past injury records and rehabilitation status, etc.; the exercise goals include, but are not limited to, the expected developmental indicators (such as target height growth), the target achievement period, the acceptable training frequency, the expected training duration and intensity restrictions, etc.; the initial exercise plan includes general training configurations such as phased training goal arrangements, daily training content (such as movement type, number of repetitions, number of sets, rhythm control, etc.), exercise intensity level, training time arrangement and precautions.

[0016] After generating the initial exercise plan, to further enhance the guiding effect of exercise intervention on children's height development, the initial exercise plan is optimized based on a height development prediction mechanism to obtain the final first exercise plan. Specifically, key training parameters, including exercise type, intensity, frequency, and duration, are extracted from the initial exercise plan and combined with multi-dimensional physical indicators such as the target user's bone age, height, weight, flexibility, muscle strength, and developmental rhythm to construct an individual physical characteristic vector. Using the individual physical characteristic vector as input, a multi-dimensional adaptation model built on a deep neural network is invoked to evaluate the suitability of multiple preset candidate exercise items. The model outputs multiple evaluation indicators for each candidate item, including height development response factors and training compliance prediction values. These indicators are fused using a weighted formula to calculate a personalized recommendation score for each candidate item. The candidate exercise items are ranked according to the recommendation scores, and several exercise items with higher recommendation scores are selected as the final recommended optimized training content, thereby generating the first exercise plan for the current user and realizing intelligent recommendation of exercise items and personalized optimization of training plans based on individual user differences. The height development response factor is used to measure the bone stimulation effect and longitudinal height growth potential of target users after performing the exercise. It is based on the historical exercise-height feedback sample set for predictive modeling. The higher the score, the greater the average efficacy of the exercise in inducing height growth in a group with similar physical characteristics. The training compliance prediction value is used to measure the user's interest fit and probability of continuous completion in a specific exercise. The system is based on data such as user age, psychological characteristic labels, and historical training completion rate, and uses compliance behavior model to fit and predict, avoiding the selection of exercises that users may resist or find difficult to persist in for a long time.

[0017] Furthermore, based on the individual characteristics and movement goals of the target user, movement planning is performed to obtain an initial movement plan, including: Based on the individual characteristic information and the movement target information, a big data retrieval of movement data is performed to obtain a matching movement scheme library; trigger degree evaluation is performed on each scheme in the matching movement scheme library to obtain a trigger degree evaluation distribution; the trigger degree evaluation is maximized and optimized in the matching movement scheme library based on the trigger degree evaluation distribution to generate the initial movement scheme.

[0018] First, based on the individual characteristics and exercise goals of the target user, a joint retrieval operation targeting big data in sports is performed. This big data includes structured historical exercise program samples, adaptation records for multiple user groups, training feedback data, and expert-annotated data. By constructing a feature vector index model, the target user's characteristics and exercise goals are embedded into the retrieval space. Based on similarity calculation rules, exercise programs with high adaptability across multiple feature dimensions are extracted from the big data to form a matching exercise program library. Each program in the matching exercise program library includes a program identifier, adaptable user feature tags, exercise combination, training cycle settings, and effect feedback records. Next, for each program in the matching exercise program library, its frequency of occurrence in historical exercise data is calculated. This frequency is defined as the program triggering degree, used to measure the relative probability that the program will be invoked or recommended among users with similar characteristics. Based on the triggering degree values ​​of all candidate programs, a triggering degree evaluation distribution is constructed to characterize the breadth of applicability and historical effectiveness of this batch of programs under the target feature conditions. Finally, the triggering degree evaluation distribution is optimized by maximizing the triggering degree, which means extracting the target scheme with the highest triggering degree from the matching motion scheme library to generate the initial motion scheme.

[0019] Based on the first exercise plan, the target user is subjected to acute injury risk prediction to obtain an acute injury risk coefficient.

[0020] Furthermore, based on the first exercise plan, the target user is subjected to acute injury risk prediction to obtain an acute exercise injury risk coefficient, including: Based on the individual characteristic information, a user profile of the target user is constructed; the user profile is subjected to multiple simulation training according to the first exercise plan to obtain multiple exercise simulation data; confidence fusion is performed on the multiple exercise simulation data to obtain first exercise simulation confidence data; acute injury risk assessment is performed based on the first exercise simulation confidence data to generate the acute injury risk coefficient of the exercise.

[0021] Specifically, based on the individual characteristics of the target users, a user profile is constructed. This user profile is a multi-dimensional structured dataset generated by combining physiological characteristic data and motor ability characteristics. It covers information such as height, weight, skeletal structure parameters, joint range of motion, muscle strength level, postural stability index, historical exercise records, and past injury status, to realistically reflect the user's exercise capacity and potential risk characteristics. The training content, movement type, rhythm control, and load parameters from the first exercise plan are applied to the user profile, and a simulation engine based on a sports biomechanics model is invoked to perform multiple simulation training sessions. Each simulation training session outputs key indicators such as joint force, muscle stretch, impact acceleration, and movement deviation angle during the exercise, thereby obtaining multiple time-labeled exercise simulation data. Multiple motion simulation data are subjected to confidence fusion processing. Specifically, fusion algorithms such as weighted averaging, Bayesian update, or Kalman filtering are used, assigning different weights according to the reliability of the data source and the historical verification error rate. This eliminates the influence of occasional biases in a single simulation, resulting in more stable first motion simulation confidence data. This first motion simulation confidence data serves as a stable and reliable basis for mechanical response, reflecting the user's average stress state and potential injury trend under the current motion scheme. Based on the first motion simulation confidence data, an acute injury risk assessment model is invoked to determine the risk, generating an acute motion injury risk coefficient. This coefficient measures the likelihood of short-term biomechanical failures such as strains, ankle sprains, and impact injuries when the user executes the current first motion scheme.

[0022] Furthermore, based on the first motion simulation confidence data, an acute injury risk assessment is performed to generate the acute motion injury risk coefficient, including: Based on the first motion simulation confidence data, acute injury-related features are identified to construct a first acute injury-related feature vector; multiple learners are trained under supervision based on the acute injury vector sample set and the acute injury risk sample set to obtain multiple acute injury risk assessment models that meet the predetermined accuracy of risk assessment; the first acute injury-related feature vector is input into the multiple acute injury risk assessment models to obtain multiple acute injury risk values; the mean of the multiple acute injury risk values ​​is calculated to obtain the motion acute injury risk coefficient.

[0023] Based on the confidence data from the first motion simulation, key mechanical and motion feature parameters associated with acute injuries are extracted, including but not limited to: peak joint impact force, muscle stretching rate, trajectory deviation, rate of change of impact acceleration, and posture control error index. Through feature selection and standardization, a structured first acute injury-associated feature vector is constructed. This feature vector characterizes the potential triggering factors for acute injuries in the target user under the current exercise load. Next, a pre-defined acute injury risk model training mechanism is invoked. This mechanism is based on two historically constructed datasets: one is an acute injury vector sample set, containing feature vectors extracted from different populations across various types of exercise; the other is an acute injury risk sample set, recording the actual injury event labels (e.g., whether sprains, strains, falls, etc.) corresponding to the above samples. These two types of data are used as supervised input to supervise the training of multiple machine learning machines (e.g., random forests, gradient boosting trees, support vector machines, neural networks, etc.), iteratively optimizing their parameter configurations to obtain multiple acute injury risk assessment models. Cross-validation and accuracy checks are then used to ensure that they meet the preset risk assessment accuracy requirements. The constructed first acute injury-related feature vector is input into multiple acute injury risk assessment models for inference and prediction, resulting in multiple acute injury risk values. Each risk value represents the probability or risk level of a user's short-term injury under a specific model. Finally, statistical normalization and mean fusion operations are performed on the multiple acute injury risk values ​​to obtain the acute injury risk coefficient.

[0024] Based on the first exercise plan, the target user is subjected to a chronic strain risk prediction to obtain a chronic strain risk coefficient.

[0025] Based on the first motion simulation confidence data, feature recognition operations related to chronic strain are performed. Chronic strain-related features include, but are not limited to: local repetitive stress accumulation, joint load duration, muscle fatigue index, cumulative postural deviation, and periodic changes in motion path deviation. Through feature filtering and normalization modeling, a first chronic strain-related feature vector is constructed to indicate the target user's potential chronic strain during exercise. This vector reflects the trend of indicators related to long-term repetitive motion injuries. Based on the established chronic strain vector sample set and the corresponding chronic strain risk sample set, multiple supervised learners are trained to form a set of chronic strain risk assessment models with different structures. The chronic strain vector sample set includes a large number of labeled historical individuals' chronic injury feature vectors under different training loads and frequencies; the chronic strain risk sample set corresponds to the injury labels (such as tendon strain, spinal compression injury, joint wear, etc.) of each training sample after long-term exercise. Multiple models (such as LGBM, SVM, RNN, etc.) are trained using supervised learning methods, and their output errors are cross-validated, retaining multiple models whose prediction accuracy meets the requirements. Then, the first chronic strain-related feature vector is input into the aforementioned multiple chronic strain risk assessment models to obtain multiple chronic strain risk values. Each risk value represents the probability or level of chronic strain in the target user as assessed by the corresponding model. Finally, the average of the multiple chronic strain risk values ​​is calculated to generate a sports-related chronic strain risk coefficient. The sports-related chronic strain risk coefficient is used to quantify the risk of long-term structural injury in a user performing the first exercise plan.

[0026] Based on the first exercise plan, the risk of excessive fatigue of the target user is predicted, and the risk coefficient of excessive exercise fatigue is obtained.

[0027] Based on the confidence data from the first motion simulation, we perform over-fatigue-related feature identification and extract key indicators related to accumulated exercise fatigue, including but not limited to: the amplitude of heart rate fluctuations per unit time, the density of abnormal points in respiratory exchange rate, the continuous tremor characteristics of electromyographic signals, the degree of imbalance in energy consumption-compensation ratio, and the amount of neural control fluctuations in the completion of movement. Through feature extraction, denoising, and structure mapping operations, we construct a first over-fatigue-related feature vector representing the fatigue load state of the target user. Subsequently, we train and optimize multiple supervised learning models using a pre-constructed fatigue vector sample set and an over-fatigue risk sample set. The fatigue vector sample set contains multimodal physiological parameter features related to fatigue from historical exercise data, while the over-fatigue risk sample set contains level labels for various fatigue events (such as mild fatigue, moderate fatigue, severe overwork, etc.). Through supervised training on various learners (such as Multilayer Perceptron (MLP), Random Forest, Long Short-Term Memory (LSTM), etc.), we obtain a set of high-precision over-fatigue risk assessment models. Next, the aforementioned first over-fatigue-related feature vector is input into multiple over-fatigue risk assessment models to obtain multiple over-fatigue risk values. Each risk value represents the degree or probability of fatigue risk corresponding to the target user's current exercise plan under a specific model. Finally, the average of the multiple over-fatigue risk values ​​is calculated to generate an exercise over-fatigue risk coefficient. The exercise over-fatigue risk coefficient is used to assess the degree of phased overexertion of the target user's physical fitness by the current exercise load.

[0028] Based on the risk coefficients for acute exercise injury, chronic exercise strain, and excessive exercise fatigue, the first exercise plan is optimized and adjusted to obtain the second exercise plan.

[0029] Based on the risk coefficients of acute sports injury, chronic strain, and over-fatigue, a comprehensive analysis of the first exercise plan is conducted to construct an exercise risk matrix. Based on the exercise risk matrix and preset safety constraints, multi-dimensional adjustment and optimization are performed to dynamically adjust training parameters and movement plans, ultimately generating a second exercise plan that meets safety and personalization requirements, so as to effectively reduce various risks during the user's exercise process.

[0030] Furthermore, based on the risk coefficients for acute sports injuries, chronic sports strain, and excessive exercise fatigue, the first exercise plan is optimized and adjusted to obtain a second exercise plan, including: Based on the risk coefficients for acute sports injuries, chronic sports strain, and excessive sports fatigue, a sports risk matrix is ​​constructed. If the sports risk matrix does not satisfy the sports risk constraint matrix, the first sports plan is adjusted with multiple degrees of freedom to generate a first sports adjustment space. Risk constraint optimization is performed on the first sports adjustment space based on the sports risk constraint matrix to obtain a second sports adjustment space. Weights are allocated based on multiple sports risk indicators to establish a comprehensive sports risk assessment model. The comprehensive sports risk assessment model is used to minimize the comprehensive sports risk in the second sports adjustment space to generate the second sports plan.

[0031] First, based on the risk coefficients for acute sports injury, chronic sports strain, and excessive exercise, a sports risk matrix is ​​constructed for the target user. This matrix comprehensively reflects the risk level of the current primary exercise plan across the three risk dimensions of acute injury, chronic strain, and excessive fatigue. A predefined sports risk constraint matrix is ​​set as a safety threshold standard for sports risk, used to limit the maximum acceptable range of each risk indicator in the sports risk matrix. When the sports risk matrix is ​​determined not to meet the requirements of the sports risk constraint matrix, the primary exercise plan is adjusted with multiple degrees of freedom. The adjustment includes multiple parameters such as exercise type, training intensity, training frequency, and recovery time, forming multiple adjustment combinations and constructing a first space for sports adjustment. Based on the sports risk constraint matrix, risk constraint optimization is performed on multiple plans in the first space for sports adjustment, filtering out all plans that meet the risk constraint conditions, forming a second space for sports adjustment after risk constraints. Based on the multiple sports risk indicators (risk of acute sports injury, risk of chronic sports strain, and risk of excessive exercise), weights are allocated according to the importance of different risk indicators to establish a comprehensive sports risk assessment model, comprehensively evaluating the risk performance of each plan. The weights can be determined through expert experience assignment, dynamic adjustment by machine learning, or adaptive determination based on user health status feedback. The formula for the comprehensive sports risk assessment model is: R = w1 × R1 + w2 × R2 + w3 × R3, where R is the comprehensive sports risk score, R1, R2, and R3 are the risk coefficients for acute sports injury, chronic sports strain, and excessive sports fatigue, respectively, and w1, w2, and w3 are the corresponding weights.

[0032] Furthermore, based on the motion risk constraint matrix, risk constraint optimization is performed on the first space of motion adjustment to obtain the second space of motion adjustment, including: The q-th motion adjustment scheme is extracted from the first space of motion adjustment, where q is a positive integer; multidimensional motion risk prediction is performed on the target user based on the q-th motion adjustment scheme to construct the motion risk matrix of the q-th scheme; it is determined whether the motion risk matrix of the q-th scheme satisfies the motion risk constraint matrix; if the motion risk matrix of the q-th scheme satisfies the motion risk constraint matrix, the q-th motion adjustment scheme is added to the second space of motion adjustment.

[0033] The q-th motion regulation scheme is extracted sequentially from the first space of motion regulation, where q is a positive integer representing the sequence number of the current scheme. Based on the q-th motion regulation scheme and combined with the individual characteristic information of the target user, multidimensional motion risk prediction is performed on the scheme, and a corresponding motion risk matrix for the q-th scheme is constructed. The motion risk matrix for the q-th scheme includes multidimensional risk indicators such as acute injury risk coefficient, chronic strain risk coefficient, and over-fatigue risk coefficient. It is determined whether the motion risk matrix for the q-th scheme meets the preset motion risk constraint matrix, that is, whether each risk indicator is within the acceptable safety threshold range. If the determination result is satisfactory, the q-th motion regulation scheme is added to the second space of motion regulation as a feasible scheme that meets the risk constraint requirements. This process is repeated for all schemes in the first space of motion regulation, and finally a second space of motion regulation containing all risk-compliant schemes is obtained.

[0034] The second exercise plan is compensated for nutritional features based on the AI ​​nutrition guidance model to obtain the third exercise plan, and the target user is rewarded for exercising based on the third exercise plan.

[0035] Based on the predicted energy consumption characteristics of the target user according to the second exercise plan, this consumption characteristic data, along with the exercise target information, is input into the AI ​​nutrition guidance model to generate the first nutrition plan. An allergy profile is constructed based on the target user's allergy records, and the first nutrition plan is desensitized and optimized to obtain the second nutrition plan. Nutritional feature compensation is then applied to the second exercise plan based on this second nutrition plan to obtain the third exercise plan. Based on the third exercise plan, the actual exercise execution process of the target user is monitored. When the completion meets preset quality indicators, an exercise completion quality coefficient is calculated, and exercise execution reward information is generated accordingly to incentivize the target user to continue scientific exercise.

[0036] Based on the third exercise plan, the system rewards target users for their exercise performance. This includes: based on the exercise tasks generated that day, target users can complete designated movements through video practice or by performing free vertical jumps for target detection. The system collects exercise completion data in real time, obtains the number of valid vertical jumps, and generates corresponding reward points (such as coins) based on the quality of movement completion and calorie consumption. These coins can be used in the children's virtual development system to exchange for virtual food, props, or to unlock new skills and appearances of their virtual pets, thereby enabling children to perceive and receive positive feedback on the results of their exercise behavior. Furthermore, target users can bring their trained virtual pets into the metaverse social space to engage in virtual interactions, competitive games, or scenario-based tasks with other users, enhancing the social relevance and sustained appeal of their exercise behavior. This mechanism, through a reward system that combines virtual and real elements, establishes a positive feedback loop of exercise-incentive-growth, improving children's long-term enthusiasm and self-motivation in participating in exercise training.

[0037] Furthermore, based on the AI ​​nutrition guidance model, nutritional feature compensation is applied to the second exercise plan to obtain a third exercise plan, including: Based on the second exercise plan, the exercise consumption of the target user is predicted to obtain exercise consumption feature data; the exercise consumption feature data and the exercise target information are input into the AI ​​nutrition guidance model to obtain a first nutrition plan; based on the target user's allergy record set, a user allergy map is constructed; based on the user allergy map, the first nutrition plan is desensitized and optimized to obtain a second nutrition plan, and based on the second nutrition plan, the second exercise plan is nutritional feature compensated to obtain the third exercise plan.

[0038] First, based on information such as the type of exercise, intensity, frequency, and duration included in the second exercise plan, the energy consumption of the target user is predicted, and an exercise consumption model is constructed to obtain the exercise consumption feature data. Then, the exercise consumption feature data and the target user's exercise goals (such as height increase, muscle gain, fat loss, and improved physical fitness) are input into a preset AI nutrition guidance model. This AI nutrition guidance model generates a matching first nutrition plan based on historical nutrition intake samples, exercise performance feedback samples, and a physiological parameter evolution model to guide the user in appropriate nutritional supplementation before, during, and after exercise. Further, the target user's allergy record set is obtained, including the user's known allergies to certain substances. This process involves identifying allergy information related to foods, nutrients, and additives, constructing a user allergy profile based on an allergy record set, and representing the limiting boundaries of nutrient intake. Based on this, a sensitivity analysis is performed on the first nutritional plan to identify nutrient components that conflict with the user's allergy profile. Desensitization optimization is then implemented to eliminate or replace potential allergens, thereby generating a second nutritional plan that meets individual safe intake requirements. Finally, based on the nutritional composition of the second nutritional plan, nutritional feature compensation is applied to the second exercise plan, including adjusting the timing of exercise plans and nutritional arrangements, and recommending supplement forms. This results in a third exercise plan that highly matches nutritional and exercise interventions, providing the target user with a more scientific, safe, and personalized comprehensive exercise execution plan.

[0039] Furthermore, rewarding the target user for exercise performance based on the third exercise scheme includes: When the target user exercises according to the third exercise plan, exercise monitoring data is obtained; the exercise monitoring data is evaluated for completion quality according to the third exercise plan to obtain an exercise completion quality coefficient, and exercise execution reward information is generated based on the exercise completion quality coefficient.

[0040] When a target user engages in exercise according to the third exercise plan, the system collects exercise monitoring data through wearable sensor devices (such as inertial measurement units, heart rate monitors, and posture detection devices). This data includes parameters such as exercise trajectory, movement amplitude, duration, heart rate zone, and frequency of completion. Based on the preset execution standards and evaluation indicators in the third exercise plan, the system evaluates the quality of the exercise monitoring data. A completion quality assessment model compares the deviation between the user's actual execution data and the standard reference values, outputting an exercise completion quality coefficient. This coefficient measures the accuracy, stability, and overall compliance of the exercise movements. The system then matches the exercise completion quality coefficient with a reward strategy rule base, considering the target user's historical performance trends and growth incentive mechanisms, to generate corresponding exercise execution reward information. This reward information may include virtual badge points, task level advancement, reward redemption permissions, and personalized positive feedback prompts, enhancing the user's motivation and sustainability in exercising.

[0041] In summary, the embodiments of this application have at least the following technical effects: First, exercise planning is performed based on the individual characteristics and exercise goals of the target user to obtain an initial exercise plan. This initial plan is then optimized through height development prediction to obtain the first exercise plan. Next, based on the first exercise plan, the target user's acute injury risk is predicted, yielding an acute injury risk coefficient; the target user's chronic strain risk is predicted, yielding a chronic strain risk coefficient; and the target user's over-fatigue risk is predicted, yielding an over-fatigue risk coefficient. Then, based on the acute injury risk coefficient, chronic strain risk coefficient, and over-fatigue risk coefficient, the first exercise plan is optimized and adjusted to obtain the second exercise plan. Finally, the second exercise plan is supplemented with nutritional features using an AI-guided nutrition model to obtain the third exercise plan, and the target user is rewarded for exercising based on the third exercise plan. This approach solves the technical problem of insufficient safety in exercise plans due to the lack of systematic risk assessment in existing technologies, achieving the technical effect of improving the safety and individual suitability of exercise plans while ensuring the achievement of exercise goals.

[0042] Example 2, based on the same inventive concept as the AI-driven motion planning generation method in the foregoing examples, such as... Figure 2 As shown, this application provides an AI-driven motion planning and generation system, wherein the system includes: Exercise planning module 11: Based on the individual characteristics and exercise goals of the target user, exercise planning is performed to obtain an initial exercise plan, and height development prediction and correlation optimization are performed based on the initial exercise plan to obtain a first exercise plan; First risk prediction module 12: Based on the first exercise plan, acute injury risk prediction is performed on the target user to obtain an acute exercise injury risk coefficient; Second risk prediction module 13: Based on the first exercise plan, chronic strain risk prediction is performed on the target user to obtain a chronic strain risk coefficient; Third risk prediction module 14: Based on the first exercise plan, over-fatigue risk prediction is performed on the target user to obtain an over-fatigue risk coefficient; Optimization module 15: Based on the acute exercise injury risk coefficient, the chronic strain risk coefficient, and the over-fatigue risk coefficient, the first exercise plan is optimized and adjusted to obtain a second exercise plan; Compensation module 16: Based on the AI ​​nutrition guidance model, nutritional characteristics are compensated for the second exercise plan to obtain a third exercise plan, and exercise execution rewards are given to the target user based on the third exercise plan.

[0043] Furthermore, the motion planning module 11 is used to perform the following methods: Based on the individual characteristic information and the movement target information, a big data retrieval of movement data is performed to obtain a matching movement scheme library; trigger degree evaluation is performed on each scheme in the matching movement scheme library to obtain a trigger degree evaluation distribution; the trigger degree evaluation is maximized and optimized in the matching movement scheme library based on the trigger degree evaluation distribution to generate the initial movement scheme.

[0044] Furthermore, the first risk prediction module 12 is used to perform the following method: Based on the individual characteristic information, a user profile of the target user is constructed; the user profile is subjected to multiple simulation training according to the first exercise plan to obtain multiple exercise simulation data; confidence fusion is performed on the multiple exercise simulation data to obtain first exercise simulation confidence data; acute injury risk assessment is performed based on the first exercise simulation confidence data to generate the acute injury risk coefficient of the exercise.

[0045] Furthermore, the first risk prediction module 12 is used to perform the following method: Based on the first motion simulation confidence data, acute injury-related features are identified to construct a first acute injury-related feature vector; multiple learners are trained under supervision based on the acute injury vector sample set and the acute injury risk sample set to obtain multiple acute injury risk assessment models that meet the predetermined accuracy of risk assessment; the first acute injury-related feature vector is input into the multiple acute injury risk assessment models to obtain multiple acute injury risk values; the mean of the multiple acute injury risk values ​​is calculated to obtain the motion acute injury risk coefficient.

[0046] Furthermore, the optimization module 15 is used to perform the following method: Based on the risk coefficients for acute sports injuries, chronic sports strain, and excessive sports fatigue, a sports risk matrix is ​​constructed. If the sports risk matrix does not satisfy the sports risk constraint matrix, the first sports plan is adjusted with multiple degrees of freedom to generate a first sports adjustment space. Risk constraint optimization is performed on the first sports adjustment space based on the sports risk constraint matrix to obtain a second sports adjustment space. Weights are allocated based on multiple sports risk indicators to establish a comprehensive sports risk assessment model. The comprehensive sports risk assessment model is used to minimize the comprehensive sports risk in the second sports adjustment space to generate the second sports plan.

[0047] Furthermore, the optimization module 15 is used to perform the following method: The q-th motion adjustment scheme is extracted from the first space of motion adjustment, where q is a positive integer; multidimensional motion risk prediction is performed on the target user based on the q-th motion adjustment scheme to construct the motion risk matrix of the q-th scheme; it is determined whether the motion risk matrix of the q-th scheme satisfies the motion risk constraint matrix; if the motion risk matrix of the q-th scheme satisfies the motion risk constraint matrix, the q-th motion adjustment scheme is added to the second space of motion adjustment.

[0048] Furthermore, the compensation module 16 is used to perform the following method: Based on the second exercise plan, the exercise consumption of the target user is predicted to obtain exercise consumption feature data; the exercise consumption feature data and the exercise target information are input into the AI ​​nutrition guidance model to obtain a first nutrition plan; based on the target user's allergy record set, a user allergy map is constructed; based on the user allergy map, the first nutrition plan is desensitized and optimized to obtain a second nutrition plan, and based on the second nutrition plan, the second exercise plan is nutritional feature compensated to obtain the third exercise plan.

[0049] Furthermore, the compensation module 16 is used to perform the following method: When the target user exercises according to the third exercise plan, exercise monitoring data is obtained; the exercise monitoring data is evaluated for completion quality according to the third exercise plan to obtain an exercise completion quality coefficient, and exercise execution reward information is generated based on the exercise completion quality coefficient.

[0050] Example 3, Figure 3 This is a schematic diagram of the structure of an electronic device provided in Embodiment 3 of the present invention, showing a block diagram of an exemplary electronic device suitable for implementing the embodiments of the present invention. Figure 3 The electronic device shown is merely an example and should not be construed as limiting the functionality or scope of the embodiments of the present invention. Figure 3 As shown, the electronic device includes a processor 21, a memory 22, an input device 23, and an output device 24; the number of processors 21 in the electronic device can be one or more. Figure 3 Taking a processor 21 as an example, the processor 21, memory 22, input device 23, and output device 24 in an electronic device can be connected via a bus or other means. Figure 3 Taking the example of a connection between China and Israel via a bus.

[0051] The memory 22, as a computer-readable storage medium, can be used to store software programs, computer-executable programs, and modules, such as the program instructions / modules corresponding to the AI-driven motion planning generation method in this embodiment of the invention. The processor 21 executes various functional applications and data processing of the electronic device by running the software programs, instructions, and modules stored in the memory 22, thereby realizing the aforementioned AI-driven motion planning generation method.

[0052] It should be noted that the order of the embodiments described above is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. Furthermore, the above description focuses on specific embodiments of this specification. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired results. In some implementations, multitasking and parallel processing are possible or may be advantageous.

[0053] The above description is only a preferred embodiment of this application and is not intended to limit this application. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of this application should be included within the protection scope of this application.

[0054] This specification and accompanying drawings are merely illustrative examples of this application and are intended to cover any and all modifications, variations, combinations, or equivalents within the scope of this application. Clearly, those skilled in the art can make various alterations and modifications to this application without departing from its scope. Therefore, if such modifications and modifications fall within the scope of this application and its equivalents, this application intends to include such modifications and modifications.

Claims

1. An AI-driven motion planning generation method, characterized in that, include: Exercise planning is performed based on the individual characteristics and exercise goals of the target user to obtain an initial exercise plan. Then, height development prediction and correlation optimization are performed based on the initial exercise plan to obtain a first exercise plan. Based on the first exercise plan, the target user is subjected to acute injury risk prediction to obtain an acute injury risk coefficient. Based on the first exercise plan, the target user is subjected to chronic strain risk prediction to obtain the exercise chronic strain risk coefficient. Based on the first exercise plan, the risk of excessive fatigue of the target user is predicted, and the risk coefficient of excessive exercise fatigue is obtained. The exercise risk of the first exercise plan is optimized and adjusted based on the exercise acute injury risk coefficient, the exercise chronic strain risk coefficient, and the exercise over-fatigue risk coefficient to obtain the exercise second plan; The second exercise plan is compensated for nutritional features based on the AI ​​nutrition guidance model to obtain the third exercise plan, and the target user is rewarded for exercising based on the third exercise plan.

2. The AI-driven motion planning generation method as described in claim 1, characterized in that, Motion planning is performed based on the individual characteristics and motion goals of the target user to obtain an initial motion plan, including: Based on the individual characteristic information and the movement target information, a big data retrieval of movement is performed to obtain a matching movement scheme library; Triggering degree evaluation is performed on each scheme in the matching motion scheme library to obtain the triggering degree evaluation distribution; The initial motion scheme is generated by optimizing the matching motion scheme library to maximize the triggering degree based on the triggering degree evaluation distribution.

3. The AI-driven motion planning generation method as described in claim 1, characterized in that, Based on the first exercise plan, the target user is subjected to acute injury risk prediction to obtain an acute exercise injury risk coefficient, including: Based on the individual characteristic information, a user profile of the target user is constructed; The user profile is subjected to multiple simulation training sessions based on the first exercise scheme to obtain multiple exercise simulation data. The first motion simulation confidence data is obtained by performing confidence fusion on the multiple motion simulation data. An acute injury risk assessment is performed based on the first motion simulation confidence data to generate the acute motion injury risk coefficient.

4. The AI-driven motion planning generation method as described in claim 3, characterized in that, Based on the first motion simulation confidence data, an acute injury risk assessment is performed to generate the acute motion injury risk coefficient, including: Based on the first motion simulation confidence data, acute injury-related features are identified, and a first acute injury-related feature vector is constructed. Multiple learners are trained under supervision based on acute injury vector sample sets and acute injury risk sample sets to obtain multiple acute injury risk assessment models that meet the predetermined accuracy of risk assessment. The first acute injury-associated feature vector is input into the multiple acute injury risk assessment models to obtain multiple acute injury risk values. The average of the multiple acute injury risk values ​​is calculated to obtain the acute injury risk coefficient of sports.

5. The AI-driven motion planning generation method as described in claim 1, characterized in that, Based on the risk coefficients for acute sports injuries, chronic sports strain, and excessive exercise fatigue, the first exercise plan is optimized and adjusted to obtain a second exercise plan, including: A sports risk matrix is ​​constructed based on the aforementioned acute sports injury risk coefficient, the aforementioned chronic sports strain risk coefficient, and the aforementioned excessive sports fatigue risk coefficient. If the motion risk matrix does not satisfy the motion risk constraint matrix, the first motion scheme is adjusted with multiple degrees of freedom to generate a first motion adjustment space; Based on the motion risk constraint matrix, the first space of motion adjustment is optimized for risk constraints to obtain the second space of motion adjustment. A comprehensive sports risk assessment model is established by weighting multiple sports risk indicators. Based on the motion comprehensive risk assessment model, the second space for motion regulation is optimized to minimize the comprehensive risk of motion, thereby generating the second motion scheme.

6. The AI-driven motion planning generation method as described in claim 5, characterized in that, Based on the motion risk constraint matrix, the first space of motion adjustment is optimized to obtain the second space of motion adjustment, including: The q-th motion adjustment scheme is extracted based on the first space of motion adjustment, where q is a positive integer; Based on the qth motion adjustment scheme, multidimensional motion risk prediction is performed on the target user, and a motion risk matrix of the qth scheme is constructed. Determine whether the motion risk matrix of the q-th scheme satisfies the motion risk constraint matrix; If the motion risk matrix of the qth scheme satisfies the motion risk constraint matrix, the qth motion adjustment scheme is added to the second space of motion adjustment.

7. The AI-driven motion planning generation method as described in claim 1, characterized in that, Based on the AI ​​nutrition guidance model, nutritional feature compensation is applied to the second exercise plan to obtain a third exercise plan, including: Based on the second exercise plan, the exercise consumption of the target user is predicted to obtain exercise consumption characteristic data. The exercise consumption characteristic data and the exercise target information are input into the AI ​​nutrition guidance model to obtain a first nutrition plan; Construct a user allergy map based on the target user's allergy record set; The first nutrition plan is desensitized and optimized based on the user allergy profile to obtain a second nutrition plan. Then, the second exercise plan is compensated for nutritional characteristics based on the second nutrition plan to obtain the third exercise plan.

8. The AI-driven motion planning generation method as described in claim 1, characterized in that, Based on the third exercise scheme, the target user is rewarded for their exercise performance, including: When the target user exercises according to the third exercise plan, exercise monitoring data is obtained; The third exercise scheme is used to evaluate the quality of the exercise monitoring data to obtain an exercise completion quality coefficient, and exercise execution reward information is generated based on the exercise completion quality coefficient.

9. An AI-driven motion planning and generation system, characterized in that, The system for implementing the AI-driven motion planning generation method according to any one of claims 1-8, the system comprising: Exercise planning module: Based on the individual characteristics and exercise goals of the target user, exercise planning is performed to obtain an initial exercise plan, and height development prediction and correlation optimization are performed based on the initial exercise plan to obtain a first exercise plan; First risk prediction module: Based on the first exercise plan, predicts the acute injury risk of the target user and obtains the acute injury risk coefficient of the exercise. Second risk prediction module: Based on the first exercise plan, predict the risk of chronic strain injury for the target user and obtain the exercise chronic strain injury risk coefficient; The third risk prediction module: predicts the risk of excessive fatigue of the target user based on the first exercise plan, and obtains the risk coefficient of excessive exercise fatigue. Optimization module: Based on the risk coefficients of acute exercise injury, chronic exercise strain, and excessive exercise fatigue, the first exercise plan is optimized and adjusted to obtain the second exercise plan; Compensation module: Based on the AI ​​nutrition guidance model, nutritional feature compensation is performed on the second exercise plan to obtain the third exercise plan, and exercise execution rewards are given to the target user based on the third exercise plan.

10. An electronic device, characterized in that, The electronic device includes: Memory, used to store executable instructions; The processor, when executing executable instructions stored in the memory, implements the AI-driven motion planning generation method according to any one of claims 1-8.

Images