An energy button and exoskeleton combined simulation motion posture error compensation method, device and storage medium
By deploying energy buttons on exoskeleton joints, collecting and standardizing motion data, performing posture calculation and error compensation, the problem of posture misalignment in exoskeleton co-simulation was solved, and high-precision motion control was achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- BEIJING ANZHEN HOSPITAL AFFILIATED TO CAPITAL MEDICAL UNIV
- Filing Date
- 2026-04-10
- Publication Date
- 2026-07-14
AI Technical Summary
In existing exoskeleton co-simulation technologies, multi-sensor data is susceptible to noise, has large format differences, insufficient processing accuracy, and lacks a real-time error compensation mechanism, leading to virtual-physical posture misalignment and affecting simulation reliability and control accuracy.
By deploying energy buttons on key joints of the exoskeleton, raw motion data is collected, standardized and optimized, and posture calculations are performed. A digital twin platform is used to calculate posture deviations, build an error compensation model, and correct the motion data of the exoskeleton control module in real time.
It achieves accurate acquisition and real-time compensation of exoskeleton motion posture errors, improves the posture consistency between the virtual model and the physical entity, and enhances the accuracy of exoskeleton control and simulation reliability.
Smart Images
Figure CN122389306A_ABST
Abstract
Description
Technical Field
[0001] This invention mainly relates to the field of digital twin simulation processing technology, specifically to a method, device, and storage medium for compensating for motion posture errors in a joint simulation of an energy button and an exoskeleton. Background Technology
[0002] Exoskeleton robots, as intelligent equipment integrating mechanical design, motion control, and ergonomics, have been widely used in rehabilitation medicine, industrial assistance, and military operations. Their core requirement is to achieve precise assistance or enhancement of human movements through coordinated movement with the human body. With the development of digital twin technology, exoskeleton co-simulation systems have become a key tool for optimizing motion control algorithms and reducing the debugging costs of physical prototypes. By constructing a 1:1 virtual twin model of the exoskeleton, motion trajectories can be pre-simulated and control parameters tested in a virtual environment, and then the verified strategies can be deployed to the physical entity, significantly improving R&D and application efficiency.
[0003] However, existing exoskeleton co-simulation technology still faces core bottlenecks: First, the raw motion data collected by multiple sensors is easily affected by noise and format differences, resulting in insufficient processing accuracy and difficulty in supporting accurate posture calculation; Second, due to simulation delay, model simplification, and entity loss, posture mapping deviations are prone to occur between the virtual exoskeleton model and the physical entity during dynamic motion; Third, the lack of a real-time adaptive error compensation mechanism means that offline calibration or simple linear compensation cannot cope with dynamic time-varying errors, leading to virtual-physical posture misalignment, which seriously affects the reliability of simulation and the control accuracy of the exoskeleton, thus restricting the practical application value of exoskeleton co-simulation technology. Summary of the Invention
[0004] The technical problem to be solved by the present invention is to provide a method, device and storage medium for compensating for motion posture errors by combining an energy button and an exoskeleton, in order to address the shortcomings of the prior art.
[0005] The technical solution of this invention to solve the above-mentioned technical problems is as follows: A method for compensating for motion posture errors in a combined simulation of an energy button and an exoskeleton, comprising: Raw motion data of the coordinated movement of the human body and exoskeleton is collected by energy buttons deployed in key joints of the exoskeleton. The data format of the raw motion data is standardized, and the standardized raw motion data is optimized to obtain the target motion data. The target motion data is processed by human body and exoskeleton coordinated posture calculation to obtain the actual posture data of the exoskeleton joints. Based on the digital twin platform, virtual model joint posture data is obtained, and posture deviation is calculated between the actual posture data of the exoskeleton joint and the virtual model joint posture data to obtain the posture deviation value. An error compensation model is constructed, and error compensation is performed on the error compensation model based on the attitude deviation value; The motion data of the exoskeleton simulation control module is corrected in real time by the compensated error compensation model, and control commands are generated to drive the movement of the exoskeleton physical entity.
[0006] Another technical solution of the present invention to solve the above-mentioned technical problems is as follows: A device for compensating for motion posture errors by combining an energy button and an exoskeleton, comprising: The acquisition module is used to collect raw motion data of the coordinated movement of the human body and the exoskeleton through energy buttons deployed on key joints of the exoskeleton; The standardization processing module is used to standardize the data format of the original motion data and optimize the standardized original motion data to obtain the target motion data. The error compensation module is used to: perform human and exoskeleton coordinated posture calculation on the target motion data to obtain the actual posture data of the exoskeleton joints; Based on the digital twin platform, virtual model joint posture data is obtained, and posture deviation is calculated between the actual posture data of the exoskeleton joint and the virtual model joint posture data to obtain the posture deviation value. An error compensation model is constructed, and error compensation is performed on the error compensation model based on the attitude deviation value; The control module is used to correct the motion data of the exoskeleton simulation control module in real time through the compensated error compensation model, generate control commands, and drive the movement of the exoskeleton physical entity through the control commands.
[0007] Another technical solution of the present invention to solve the above-mentioned technical problems is as follows: an energy button and exoskeleton joint simulation motion posture error compensation device, including a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the energy button and exoskeleton joint simulation motion posture error compensation method as described above.
[0008] Another technical solution of the present invention to solve the above-mentioned technical problems is as follows: a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the above-described method for compensating for motion posture errors of the energy button and exoskeleton in joint simulation.
[0009] The beneficial effects of this invention are as follows: It constructs a complete process for exoskeleton co-simulation motion posture error compensation, forming a closed loop from raw motion data acquisition to control command generation. By deploying and integrating energy buttons in key joints, it achieves accurate acquisition of raw data on human and exoskeleton coordinated motion, providing a comprehensive foundation for subsequent data processing. Subsequently, through standardized processing, posture calculation, deviation calculation, error compensation, and control command generation, it systematically covers the key links of error compensation, providing a framework support for solving the problem of posture misalignment between the virtual model and the physical entity of the exoskeleton. Attached Figure Description
[0010] Figure 1 A flowchart of the method for compensating for motion posture errors in the joint simulation of energy buttons and exoskeleton provided in an embodiment of the present invention; Figure 2 This is a block diagram of the energy button and exoskeleton combined simulation motion posture error compensation device provided in an embodiment of the present invention. Detailed Implementation
[0011] The principles and features of the present invention are described below with reference to the accompanying drawings. The examples given are only for explaining the present invention and are not intended to limit the scope of the present invention.
[0012] like Figure 1 As shown, this embodiment of the invention provides a method for compensating for motion posture errors in a joint simulation of an energy button and an exoskeleton, including: S1. Collect raw motion data of the coordinated movement of the human body and the exoskeleton through energy buttons deployed in key joints of the exoskeleton; S2. Standardize the data format of the original motion data and optimize the standardized original motion data to obtain the target motion data; S3. Perform human and exoskeleton coordinated posture calculation on the target motion data to obtain the actual posture data of the exoskeleton joints. S4. Based on the digital twin platform, obtain the joint posture data of the virtual model, calculate the posture deviation between the actual posture data of the exoskeleton joint and the posture data of the virtual model joint, and obtain the posture deviation value. S5. Construct an error compensation model and perform error compensation on the error compensation model based on the attitude deviation value; S6. The motion data of the exoskeleton simulation control module is corrected in real time by the compensated error compensation model, and control commands are generated to drive the movement of the exoskeleton physical entity through the control commands.
[0013] Specifically, the exoskeleton's key joints include deployment at the hip, knee, and ankle joints, and the power button includes an integrated multi-axis IMU and pressure sensors.
[0014] The above embodiments construct a complete process for exoskeleton co-simulation motion posture error compensation, forming a closed loop from raw motion data acquisition to control command generation. By deploying integrated energy buttons at key joints, accurate acquisition of raw data on human-exoskeleton coordinated motion is achieved, providing a comprehensive foundation for subsequent data processing. Subsequently, through standardization processing, posture calculation, deviation calculation, error compensation, and control command generation, the key links of error compensation are systematically covered, providing a framework support for solving the problem of posture misalignment between the virtual exoskeleton model and the physical entity.
[0015] Preferably, the data format of the original motion data is standardized, and the standardized original motion data is optimized to obtain the target motion data, including: The original motion data is standardized using a normalization method. The motion state is determined based on the joint contact pressure and the standardized original motion data. A corresponding window length is set according to the determined motion state. The standardized original motion data is then denoised using a sliding window with the set length. Outliers are then repaired using the Grubbs criterion to obtain the target motion data.
[0016] Specifically, the format of raw motion data is standardized through normalization methods. The core objective is to solve the "format difference problem" of data collected by energy buttons (integrating multi-axis IMU and pressure sensor) deployed in key joints of the exoskeleton (hip, knee, ankle). The multi-axis IMU collects data such as triaxial acceleration, triaxial angular velocity, and triaxial magnetic field strength (units are m / s², rad / s, and μT, respectively), while the pressure sensor collects joint contact pressure data (unit is N). The dimensions and numerical ranges of different types of data vary greatly.
[0017] Normalization, through mathematical transformations (such as linear normalization mapping data to the [0,1] or [-1,1] interval), eliminates the influence of different dimensions and differences in numerical ranges of different data, ensuring that all original motion data are in a unified format and numerical range. This avoids calculation deviations caused by inconsistent data formats during subsequent motion state judgment and filtering / denoising, thus ensuring the consistency and accuracy of data processing.
[0018] After data standardization, the current state of the exoskeleton (static or dynamic) needs to be determined based on joint contact pressure and standardized raw motion data. The core logic is to determine the state through a dual dimension of "pressure + angular velocity" to avoid misjudgment problems caused by a single dimension.
[0019] Based on the above motion state judgment results, sliding windows of different lengths are set, and the "sliding mean filtering algorithm" is used to denoise the standardized raw data. The core is to find a balance between "denoising effect" and "data timeliness" through the window length of "state adaptation".
[0020] After sliding window denoising, outliers may still exist in the data (such as jump data caused by momentary sensor failure or extreme values caused by sudden external interference). These need to be identified and corrected using the Grubbs criterion to ensure data integrity and authenticity. The specific steps are as follows: S2.1, the denoised dataset (denoted as...) x '1, x '2, ..., x ' n ), calculate the mean of the dataset (denoted as ). ) and standard deviation (denoted as s'); For each data point x' k Calculate its Grubbs statistic (denoted as ). G ( k The formula is: , in, G ( k The larger the value, the greater the deviation of the data point from the overall trend of the dataset, and the more likely it is to be an outlier.
[0021] S2.2. Predetermine the Grubbs critical value (denoted as Gα, obtained by looking up the Grubbs critical value table based on the data sample size n and the confidence level α (e.g., 95%)). If G(k) of a certain data point is greater than Gα, then the data point is determined to be an outlier.
[0022] S2.3. For data points identified as outliers, repair them using the method of "linear interpolation between two adjacent normal data points"—assuming the outlier is located at the index. k Its preceding normal data point is x'{ k -1} (index) k -1), the next normal data point is x '{ k +1} (index) k +1), the corrected outlier (denoted as ) x '' k The calculation formula is: , Among them, based on the assumption of data continuity, it is ensured that the repaired data is consistent with the overall data trend, and outliers are avoided from interfering with subsequent attitude calculations.
[0023] In the above embodiments, the original motion data format is unified by normalization, eliminating data adaptation problems caused by differences in data formats of different types of sensors and ensuring the consistency of subsequent data processing. At the same time, the window length is dynamically adjusted based on the motion state for denoising, and outliers are repaired by combining the Grubbs criterion, effectively filtering out data noise and correcting abnormal data, significantly improving the purity and accuracy of target motion data, providing high-quality data input for subsequent human and exoskeleton collaborative posture calculation, avoiding posture calculation deviations caused by poor original data quality, and ensuring the accuracy of subsequent processes.
[0024] Preferably, the motion state is determined based on joint contact pressure and standardized raw motion data. A corresponding window length is set according to the determined motion state. The standardized raw motion data is then denoised using a sliding window with the set length. Outliers are then repaired using the Grubbs criterion to obtain the target motion data, including: Acquire joint contact pressure data from the energy button F p Compared with the triaxial angular velocity data in the standardized raw motion data Set pressure threshold F th With angular velocity threshold Calculate the magnitude of the triaxial angular velocity data: , like and If the exoskeleton is determined to be in a static state, or The exoskeleton is determined to be in motion. If the state is determined to be static, set the sliding window length. M =10, if determined to be in motion state, set the sliding window length. M =5, given the standardized original motion data sequence, based on the set window length. M The moving average filtering algorithm is used to denoise the data. The filtered data The calculation formula is: , in, k Index the current data point. The original motion data sequence, ,and This formula is used to suppress high-frequency vibration noise (such as muscle tremors and exoskeleton mechanical vibrations).
[0025] The mean and standard deviation of the denoised dataset are calculated, and the Grubbs statistic for each data point in the denoised dataset is also calculated. : , in, The mean, Standard deviation, Grubbs statistic for each data point With Grubbs critical value ,like If the data point is determined to be an outlier, it is repaired by linear interpolation of two adjacent normal data points. The target motion data is obtained based on the normal value and the repaired value corresponding to the data point.
[0026] In the above embodiments, the data optimization process is further refined. By combining joint contact pressure data with triaxial angular velocity data, the static or dynamic state of the exoskeleton is accurately determined, providing a scientific basis for dynamically setting the sliding window length. Adjusting the sliding window length according to different states makes sliding mean filtering and denoising more targeted, improving data smoothness when stationary and ensuring data timeliness during movement, thus balancing denoising effectiveness and real-time data. Simultaneously, outliers are accurately identified by comparing Grubbs' statistic with critical values, and repaired using linear interpolation of adjacent normal data points, ensuring the rationality of outlier repair and preserving data authenticity to the greatest extent. This embodiment makes data optimization processing more refined, further improving the quality of target motion data and providing more reliable data support for attitude calculation.
[0027] Preferably, the target motion data is processed by human and exoskeleton coordinated posture calculation to obtain the actual posture data of the exoskeleton joints, including: Extract triaxial acceleration data and triaxial magnetic field strength data in a static state from the target motion data. Normalize the triaxial acceleration data to obtain the gravity vector, and normalize the triaxial magnetic field strength data to obtain the geomagnetic vector. The Euler angles of the exoskeleton joint's initial posture are calculated based on the gravity vector and the geomagnetic vector, resulting in initial Euler angles. These initial Euler angles include a roll angle about the x-axis, used to represent joint tilt. The roll angle is expressed as: , It also includes the pitch angle about the y-axis, used to describe the joint pitch, which is expressed as: , It also includes the yaw angle about the z-axis, used to describe joint deflection, and calculates the projection of the geomagnetic vector onto the horizontal plane. , , for projection Perform normalization processing, and then base the normalized projection values on the normalized values. The yaw angle is expressed as: ; Roll angle Pitch angle and yaw angle Convert to initial attitude quaternion The initial attitude quaternion is represented as: , in, q 00 For the real part, q 01 , q 02 and q 03 It is the imaginary part; Extract triaxial angular velocity data from the target motion data, and filter the triaxial angular velocity data to obtain filtered triaxial angular velocity data; With the initial attitude quaternion q Using 0 as a baseline, attitude updates are performed on the filtered triaxial angular velocity data using quaternion differential equations. The quaternion differential equations are as follows: , in, This is the quaternion multiplication operator. The filtered angular velocity data, This is the antisymmetric matrix corresponding to the angular velocity; The discretized attitude quaternion at time t is solved by first-order Euler integral. : , The predicted attitude quaternion at time t Convert the data to Euler angles and bind the Euler angle data corresponding to each joint to the joint ID to obtain the actual posture data of the exoskeleton joints. The data format is as follows: .
[0028] In the above embodiments, high-precision human and exoskeleton collaborative posture calculation is achieved. Through a two-step strategy of "initial posture calibration + real-time posture update", the initial stage calculates the initial Euler angles based on the gravity vector and geomagnetic vector and converts them into quaternions to avoid the Euler angle gimbal lock problem and ensure the accuracy of the initial posture calibration. In the real-time update stage, the posture is dynamically updated based on the filtered triaxial angular velocity data through quaternion differential equations and first-order Euler integrals, and the joint Euler angles are bound to the ID to achieve a precise correspondence between posture data and specific joints. It can capture the actual posture of the exoskeleton joints in real time and accurately, and obtain accurate actual posture data of the exoskeleton joints. This provides a reliable actual data benchmark for subsequent comparison with the posture data of the virtual model to calculate the deviation, and avoids deviation calculation errors caused by inaccurate posture calculation.
[0029] Preferably, virtual model joint posture data is acquired based on a digital twin platform, and posture deviation is calculated between the actual posture data of the exoskeleton joints and the virtual model joint posture data to obtain posture deviation values, including: Based on the posture data interface of the digital twin platform, real-time posture data of the virtual model joints is extracted according to the exoskeleton joint identifiers. The data format is virtual posture quaternions. ,in, q v0 For the real part, q v1 , q v2 and q v3 The imaginary part is used, and the timestamp corresponding to the data is recorded. ; like Select with timestamp Two adjacent actual attitude data and , Calculated using interpolation formula Actual attitude quaternion at time 1 The interpolation formula is: , For actual attitude quaternions With virtual pose quaternions Normalization is performed separately, and the normalization formula is as follows: , in, q For quaternions to be normalized, The normalized unit quaternion; The attitude deviation quaternion is calculated based on the attitude deviation formula using the normalized unit quaternion. The attitude deviation formula is expressed as: , in, This is the quaternion multiplication operator. The conjugate of the virtual pose unit quaternion; attitude deviation quaternions After normalization, the unit deviation quaternion is obtained. ; Based on the conversion formula, the unit deviation quaternion Convert the attitude deviation value to Euler angles. , in, This is an angle conversion factor used to convert radian values to angle values. This is the roll angle deviation about the x-axis. This represents the pitch angle deviation around the y-axis. This represents the yaw angle deviation around the z-axis.
[0030] In the above embodiments, the deviation between the actual posture data of the exoskeleton joints and the posture data of the virtual model joints is accurately quantified. The virtual model joint posture data is obtained through a digital twin platform, and the interference of time misalignment and data standardization issues on deviation calculation is eliminated by combining timestamp matching and quaternion normalization processing. Then, the posture deviation quaternion is calculated by the posture deviation formula and converted into the posture deviation value in Euler angle form. The posture difference between the physical entity of the exoskeleton and the virtual model in the three dimensions of roll angle, pitch angle and yaw angle can be quantified intuitively and accurately. This provides a clear and accurate basis for the subsequent construction of error compensation model and targeted error compensation, avoiding the problem of directionless compensation due to the ambiguity of deviation quantification.
[0031] Preferably, an error compensation model is constructed, and error compensation is performed on the error compensation model based on the attitude deviation value, including: Define the state vector of the error compensation model. , in, These are the pose quaternions of the joints of the exoskeleton virtual model at time t, used to describe the current pose of the virtual model. These are the roll, pitch, and yaw angle compensation values to be solved at time t, used to correct the deviation between the virtual model and the actual attitude. And initialize the state vector of the error compensation model; The attitude deviation value is used as the observation vector. Establish the correlation between observed values and state vectors: , in, The observation matrix is constructed by extracting elements from the state vector that are related to the observed values. The observed noise vector follows a pattern with a mean of 0 and a covariance of . Gaussian distribution, Based on the error compensation formula and the correlation between observations and state vectors The error compensation model is corrected to obtain the optimal state by revising the predicted state. The error compensation formula is as follows: , in, for t The optimal state at any given moment. To observe the residuals.
[0032] In the above embodiments, a scientific error compensation model is constructed to achieve dynamic adaptive error compensation. A state vector containing the virtual model's attitude quaternion and attitude compensation amount is defined. The attitude deviation value is used as the observation vector and a correlation is established. The model's predicted state is corrected by combining the error compensation formula. It can respond to attitude deviation changes in real time and calculate and adjust the roll angle, pitch angle, and yaw angle compensation amount in a targeted manner. This model effectively offsets the dynamic time-varying errors caused by factors such as simulation delay, model simplification, and physical loss. Compared with traditional offline calibration or simple linear compensation, it has higher compensation accuracy and stronger adaptability, successfully solving the problem of virtual-physical attitude misalignment of the exoskeleton. This provides a key guarantee for the subsequent generation of precise control commands and improvement of exoskeleton control accuracy.
[0033] like Figure 2 As shown, this embodiment of the invention also provides a device for compensating for motion posture errors by combining an energy button and an exoskeleton, comprising: The acquisition module is used to collect raw motion data of the coordinated movement of the human body and the exoskeleton through energy buttons deployed on key joints of the exoskeleton; The standardization processing module is used to standardize the data format of the original motion data and optimize the standardized original motion data to obtain the target motion data. The error compensation module is used to: perform human and exoskeleton coordinated posture calculation on the target motion data to obtain the actual posture data of the exoskeleton joints; Based on the digital twin platform, virtual model joint posture data is obtained, and posture deviation is calculated between the actual posture data of the exoskeleton joint and the virtual model joint posture data to obtain the posture deviation value. An error compensation model is constructed, and error compensation is performed on the error compensation model based on the attitude deviation value; The control module is used to correct the motion data of the exoskeleton simulation control module in real time through the compensated error compensation model, generate control commands, and drive the movement of the exoskeleton physical entity through the control commands.
[0034] Preferably, the data format of the original motion data is standardized, and the standardized original motion data is optimized to obtain the target motion data, including: The original motion data is standardized using a normalization method. The motion state is determined based on the joint contact pressure and the standardized original motion data. A corresponding window length is set according to the determined motion state. The standardized original motion data is then denoised using a sliding window with the set length. Outliers are then repaired using the Grubbs criterion to obtain the target motion data.
[0035] This invention also provides a device for compensating for motion posture errors in a combined simulation of an energy button and an exoskeleton, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the method for compensating for motion posture errors in a combined simulation of an energy button and an exoskeleton as described above.
[0036] This invention also provides a computer-readable storage medium storing a computer program that, when executed by a processor, implements the above-described method for compensating for motion posture errors in the joint simulation of the energy button and exoskeleton.
[0037] It should be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.
[0038] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described apparatus and unit can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.
[0039] In the several embodiments provided in this application, it should be understood that the disclosed apparatus and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed.
[0040] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of the embodiments of the present invention, depending on actual needs.
[0041] Furthermore, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit. The integrated unit can be implemented in hardware or as a software functional unit.
[0042] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.
Claims
1. A method for compensating for motion posture errors in a combined simulation of an energy button and an exoskeleton, characterized in that, include: Raw motion data of the coordinated movement of the human body and exoskeleton is collected by energy buttons deployed in key joints of the exoskeleton. The data format of the raw motion data is standardized, and the standardized raw motion data is optimized to obtain the target motion data. The target motion data is processed by human body and exoskeleton coordinated posture calculation to obtain the actual posture data of the exoskeleton joints. Based on the digital twin platform, virtual model joint posture data is obtained, and posture deviation is calculated between the actual posture data of the exoskeleton joint and the virtual model joint posture data to obtain the posture deviation value. An error compensation model is constructed, and error compensation is performed on the error compensation model based on the attitude deviation value; The motion data of the exoskeleton simulation control module is corrected in real time by the compensated error compensation model, and control commands are generated to drive the movement of the exoskeleton physical entity.
2. The method for compensating for motion posture errors in the combined simulation of energy buttons and exoskeletons according to claim 1, characterized in that, The original motion data is standardized, and the standardized original motion data is then optimized to obtain the target motion data, including: The original motion data is standardized using a normalization method. The motion state is determined based on the joint contact pressure and the standardized original motion data. A corresponding window length is set according to the determined motion state. The standardized original motion data is then denoised using a sliding window with the set length. Outliers are then repaired using the Grubbs criterion to obtain the target motion data.
3. The method for compensating for motion posture errors in the combined simulation of energy buttons and exoskeletons according to claim 2, characterized in that, The motion state is determined based on joint contact pressure and standardized raw motion data. A corresponding window length is set according to the determined motion state. Noise is then denoised using a sliding window with the set length, and outliers are repaired using the Grubbs criterion to obtain the target motion data, including: Acquire joint contact pressure data from the energy button F p Compared with the triaxial angular velocity data in the standardized raw motion data Set pressure threshold F th With angular velocity threshold Calculate the magnitude of the triaxial angular velocity data: ， like and If the exoskeleton is determined to be in a static state, or The exoskeleton is determined to be in motion. If the state is determined to be static, set the sliding window length. M =10, if determined to be in motion state, set the sliding window length. M =5, given the standardized original motion data sequence, based on the set window length. M The moving average filtering algorithm is used to denoise the data. The filtered data The calculation formula is: ， in, k Index the current data point. The original motion data sequence, ,and ; The mean and standard deviation of the denoised dataset are calculated, and the Grubbs statistic for each data point in the denoised dataset is also calculated. : ， in, The mean, Standard deviation, Grubbs statistic for each data point With Grubbs critical value ,like If the data point is determined to be an outlier, it is repaired by linear interpolation of two adjacent normal data points. The target motion data is obtained based on the normal value and the repaired value corresponding to the data point.
4. The method for compensating for motion posture errors in the combined simulation of energy buttons and exoskeletons according to claim 3, characterized in that, The target motion data is processed by human and exoskeleton coordinated posture calculation to obtain the actual posture data of the exoskeleton joints, including: Extract triaxial acceleration data and triaxial magnetic field strength data in a static state from the target motion data. Normalize the triaxial acceleration data to obtain the gravity vector, and normalize the triaxial magnetic field strength data to obtain the geomagnetic vector. The Euler angles of the exoskeleton joint's initial posture are calculated based on the gravity vector and the geomagnetic vector, resulting in initial Euler angles. These initial Euler angles include a roll angle about the x-axis, used to represent joint tilt. The roll angle is expressed as: , It also includes the pitch angle about the y-axis, used to describe the joint pitch, which is expressed as: , It also includes the yaw angle about the z-axis, used to describe joint deflection, and calculates the projection of the geomagnetic vector onto the horizontal plane. , , for projection Perform normalization processing, and then base the normalized projection values on the normalized values. The yaw angle is expressed as: ; Roll angle Pitch angle and yaw angle Convert to initial attitude quaternion The initial attitude quaternion is represented as: ， in, q 00 For the real part, q 01 , q 02 and q 03 It is the imaginary part; Extract triaxial angular velocity data from the target motion data, and filter the triaxial angular velocity data to obtain filtered triaxial angular velocity data; With the initial attitude quaternion q Using 0 as a baseline, attitude updates are performed on the filtered triaxial angular velocity data using quaternion differential equations. The quaternion differential equations are as follows: ， in, This is the quaternion multiplication operator. This is the filtered angular velocity data. This is the antisymmetric matrix corresponding to the angular velocity; The discretized attitude quaternion at time t is solved by first-order Euler integral. : ， The predicted attitude quaternion at time t Convert the data to Euler angles and bind the Euler angle data corresponding to each joint to the joint ID to obtain the actual posture data of the exoskeleton joints. The data format is as follows: .
5. The method for compensating for motion posture errors in the combined simulation of energy buttons and exoskeletons according to claim 4, characterized in that, Based on the digital twin platform, virtual model joint posture data is acquired. Posture deviation is calculated between the actual posture data of the exoskeleton joints and the virtual model joint posture data to obtain posture deviation values, including: Based on the posture data interface of the digital twin platform, real-time posture data of the virtual model joints is extracted according to the exoskeleton joint identifiers. The data format is virtual posture quaternions. ,in, q v0 For the real part, q v1 , q v2 and q v3 The imaginary part is used, and the timestamp corresponding to the data is recorded. ; like Select with timestamp Two adjacent actual attitude data and , Calculated using interpolation formula Actual attitude quaternion at time 1 The interpolation formula is: ， For actual attitude quaternions With virtual pose quaternions Normalization is performed separately, and the normalization formula is as follows: ， in, q For quaternions to be normalized, The normalized unit quaternion; The attitude deviation quaternion is calculated based on the attitude deviation formula using the normalized unit quaternion. The attitude deviation formula is expressed as: ， in, This is the quaternion multiplication operator. The conjugate of the virtual pose unit quaternion; attitude deviation quaternions After normalization, the unit deviation quaternion is obtained. ; Based on the conversion formula, the unit deviation quaternion Convert the attitude deviation value to Euler angles. ， in, This is an angle conversion factor used to convert radian values to angle values. This is the roll angle deviation about the x-axis. This represents the pitch angle deviation around the y-axis. This represents the yaw angle deviation around the z-axis.
6. The method for compensating for motion posture errors in the combined simulation of energy buttons and exoskeletons according to claim 1, characterized in that, Constructing an error compensation model, and performing error compensation on the error compensation model based on the attitude deviation value, including: Define the state vector of the error compensation model. ， in, These are the pose quaternions of the joints of the exoskeleton virtual model at time t, used to describe the current pose of the virtual model. These are the roll, pitch, and yaw angle compensation values to be solved at time t, used to correct the deviation between the virtual model and the actual attitude. And initialize the state vector of the error compensation model; The attitude deviation value is used as the observation vector. Establish the correlation between observed values and state vectors: ， in, The observation matrix is constructed by extracting elements from the state vector that are related to the observed values. The observed noise vector follows a pattern with a mean of 0 and a covariance of . Gaussian distribution, Based on the error compensation formula and the correlation between observations and state vectors The error compensation model is corrected to obtain the optimal state by revising the predicted state. The error compensation formula is as follows: ， in, for t The optimal state at any given moment. To observe the residuals.
7. A device for compensating for motion posture errors by combining an energy button and an exoskeleton, characterized in that, include: The acquisition module is used to collect raw motion data of the coordinated movement of the human body and the exoskeleton through energy buttons deployed on key joints of the exoskeleton; The standardization processing module is used to standardize the data format of the original motion data and optimize the standardized original motion data to obtain the target motion data. The error compensation module is used to: perform human and exoskeleton coordinated posture calculation on the target motion data to obtain the actual posture data of the exoskeleton joints; Based on the digital twin platform, virtual model joint posture data is obtained, and posture deviation is calculated between the actual posture data of the exoskeleton joint and the virtual model joint posture data to obtain the posture deviation value. An error compensation model is constructed, and error compensation is performed on the error compensation model based on the attitude deviation value; The control module is used to correct the motion data of the exoskeleton simulation control module in real time through the compensated error compensation model, generate control commands, and drive the movement of the exoskeleton physical entity through the control commands.
8. The energy button and exoskeleton combined simulation motion posture error compensation device according to claim 7, characterized in that, The original motion data is standardized, and the standardized original motion data is then optimized to obtain the target motion data, including: The original motion data is standardized using a normalization method. The motion state is determined based on the joint contact pressure and the standardized original motion data. A corresponding window length is set according to the determined motion state. The standardized original motion data is then denoised using a sliding window with the set length. Outliers are then repaired using the Grubbs criterion to obtain the target motion data.
9. A device for compensating for motion posture errors by combining an energy button and an exoskeleton, characterized in that, The device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the energy button and exoskeleton joint simulation motion posture error compensation method as described in any one of claims 1 to 6.
10. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the method for compensating for motion posture errors in the joint simulation of energy buttons and exoskeleton as described in any one of claims 1 to 6.