An action-capture-based aerobic fitness assisted training system

Abstract

This invention relates to the field of assisted training technology and discloses an aerobic fitness assisted training system based on motion capture. The system includes a motion capture module, a state detection module, a data processing module, an equipment control module, and a file management module. The motion capture module generates left and right gait cycles and parameters of the same origin. The state detection module generates the actual stable parameter time range and parameter state validity indicators. The data processing module determines the effective gait cycle and target movement set, and uses paired, responsive, and reproducible methods to complete candidate generation, verification determination, and control eligibility assessment, and selects either equipment parameter correction or movement prompt correction channels. The equipment control module performs parameter switching and returns the actual execution results. The file management module updates the training assistance file and outputs the movement prompt correction content. This invention evaluates movement deviations after the actual speed of the treadmill and the actual incline of the treadmill have stabilized, and selects either equipment parameter correction or movement prompt correction after withdrawal and reproduction.

Description

Technical Field

[0001] This invention relates to the field of assisted training technology, specifically to an aerobic fitness assisted training system based on motion capture. Background Technology

[0002] In treadmill aerobic training, cameras or wearable devices are typically used to collect the positions of the body's joints, identify gait events such as ground contact and lift-off, and calculate motion parameters such as support time, swing time, stride length, ground contact position, and joint angles. Based on this, left-right asymmetry or movement deviation is determined, and the running belt speed and treadmill incline are adjusted, or movement prompts are output. Some solutions send the target speed and target incline to the treadmill controller, which is then executed by the running belt drive motor and incline adjustment mechanism.

[0003] Existing solutions often collect and evaluate motion data immediately after control commands are issued, without defining a stable range based on the actual speed of the running belt and the actual incline of the treadmill. Gait data during parameter changes still enters the deviation assessment process. Deviation identification often involves directly comparing left and right side motion amounts or using a fixed template as a reference, failing to establish a centered deviation relative to an individual baseline and making it difficult to obtain the target motion set over continuous effective gait cycles. Speed ​​and incline adjustments lack candidate parameter generation under nominal load tolerance conditions, as well as verification and screening based on parameter response residuals and comprehensive corrected responses. Before adjusting equipment parameters or outputting motion prompts, there is a lack of withdrawal and reproduction verification, and the resulting control qualification criteria. There is also a lack of a unified selection basis between equipment parameter correction and motion prompt correction. Summary of the Invention

[0004] To address the shortcomings of existing technologies, this invention provides an aerobic fitness assisted training system based on motion capture, which solves the problems of existing technologies in aligning motion data with the actual stable state of the equipment, identifying deviations relative to personal benchmarks and evaluating candidate responses, and selecting correction channels after withdrawal and reproduction.

[0005] To achieve the above objectives, the present invention provides the following technical solution: This invention provides a motion capture-based aerobic fitness assisted training system, comprising: The motion capture module is used to establish a unified time reference and form left and right gait cycles and parameters of the same motion source; The status detection module is used to detect the actual speed of the treadmill belt and the actual incline of the running platform, forming a time range of actual stable parameters and a valid indicator of parameter status. The data processing module is used to determine the effective gait cycle based on the left and right gait cycles, homogeneous motion parameters, actual stable parameter time range, and parameter status validity flags. It forms a personal baseline, a set of centered deviations and target motions, generates candidate parameter groups in a paired manner, and forms parameter exchange instructions. Based on the actual execution results returned, it determines the parameter group to be verified in a responsive manner and forms the withdrawal and reproduction execution order. Then, based on the actual execution results, it forms a control qualification flag in a reproduction manner to select the equipment parameter correction channel or the motion prompt correction channel and forms identification and verification results. The equipment control module is used to execute parameter exchange commands, withdraw and reproduce the execution sequence and equipment parameter correction commands, form actual execution results and send them back to the data processing module; the file management module is used to generate or update training auxiliary files based on the identification and verification results and output the action prompt correction content through the treadmill display screen.

[0006] Preferably, the motion capture module includes: The calibration synchronization unit is used to convert the camera frame time, treadmill controller time, incremental encoder detection time and tilt sensor detection time to a unified time reference, establish the treadmill coordinate system and determine the range of motion acquisition capability; the gait parameter unit is used to generate left and right gait cycles, homologous motion parameters, joint confidence, pelvic center position, handrail grip status and left and right gait pairing status based on the synchronously acquired trainee images.

[0007] Preferably, the status detection module includes: The actual state unit is used to detect the actual speed of the running belt and the actual slope of the running platform, forming the actual speed sequence of the running belt and the actual slope sequence of the running platform; the effective segmentation unit is used to form the actual stable parameter segment, the actual stable parameter segment time range, and the effective parameter state flag when the actual speed sequence of the running belt and the actual slope sequence of the running platform belong to the preset speed stability tolerance range and the preset slope stability tolerance range centered on the target speed and target slope of the equipment control module, respectively.

[0008] Preferably, the data processing module includes a deviation identification unit; The deviation identification unit is used to determine the effective gait cycle when the start and end times of the left and right gait cycles all fall within the actual stable parameter time range, the parameter status is marked as valid, the joint confidence output by the motion capture module is not less than the preset minimum joint confidence, the pelvic center position is within the preset running belt safe area, and the handhold status and the left and right gait pairing status are respectively "not passed" and "passed". The deviation identification unit is also used to calculate the motion asymmetry, individual benchmark, centering deviation and summary centering deviation based on the same motion parameters using lateral difference. When the absolute value of the centering deviation is greater than the motion deviation threshold and the sign is consistent within several effective gait cycles of the most recent consecutive preset continuous judgment cycle, a continuous motion deviation category and target motion set are formed.

[0009] Furthermore, the data processing module also includes a response evaluation unit; The response evaluation unit is used to form an original parameter set based on the median of the actual speed of the running belt and the actual slope of the running platform within the current actual stable parameter time range. It uses a paired approach to generate different candidate parameter sets for speed and slope when the absolute value of the difference between the nominal loads is not greater than the preset nominal load tolerance. When the candidate parameter set belongs to the preset equipment range, the preset safety range, and the motion acquisition capability range, the candidate parameter is marked as valid. If there are two candidate parameter groups with valid flags indicating they have passed, the system enters a two-way parameter exchange process. If there is only one candidate parameter group with valid flags indicating they have passed, the system enters a one-way verification process. If neither passes, the original parameter group is maintained and the system proceeds to an action prompt correction, thus generating a parameter exchange instruction.

[0010] Furthermore, the response evaluation unit is also used for: After receiving the actual execution results, only the parameter segments whose parameter exchange completion flag is "passed" and parameter exchange exception flag is "failed" are processed. Candidate parameter segments are aligned with the preceding and following original parameter baseline segments. Left and right source action differences are calculated, and individual baseline and parameter condition differences, as well as time drift, are deducted. A response-based approach is used to calculate the parameter response residuals and the comprehensive corrected response based on the summarized centralized deviation. The parameter group to be verified is determined using this response-based approach. An effective response is formed when the absolute value of the residual of at least one type of parameter response is greater than the upper limit of the measurement uncertainty range and the comprehensive corrected response is greater than the preset response threshold; if only one candidate parameter group has an effective response, it is selected; if the effective responses of two candidate parameter groups are formed and the absolute value of the difference between the comprehensive corrected responses is greater than the preset discrimination tolerance, the larger one is selected; otherwise, the candidate is canceled and the action prompt correction is initiated.

[0011] Among them, the parameter condition difference is a comparison between the candidate parameter conditions and the original parameter conditions.

[0012] Furthermore, the data processing module also includes a channel determination unit; The channel determination unit is used to form a first verification segment, a withdrawal segment, and a second verification segment when the actual execution result satisfies the parameter exchange completion flag as pass, the parameter exchange abnormality flag as fail, and the paired left and right gait cycles in the corresponding segment meet the determination conditions for valid gait cycles. The comprehensive deviation of the actual stable parameter segment corresponding to the most recent original parameter group before the first verification segment is used as the pre-verification benchmark, and the control qualification flag is formed in a reproducible manner. When there is no parameter group to be verified, an action prompt correction instruction is generated and the action prompt correction channel is selected; when the control qualification flag is passed, an equipment parameter correction instruction is generated and the equipment parameter correction channel is selected; when it fails and the difference between the comprehensive deviation between the pre-verification benchmark and the withdrawal segment is not less than the preset retention judgment threshold, the action retention flag is passed and the original parameter group is maintained. If the action fails and the difference between the overall deviation between the pre-verification baseline and the withdrawal segment is less than the preset retention judgment threshold, the action retention flag is set as failed, an action prompt correction instruction is generated, and the action prompt correction channel is selected; if the action cannot be reproduced or the verification segment is invalid, the parameter group to be verified is canceled; the identification and verification results are output.

[0013] Preferably, the equipment control module includes: The drive execution unit is used to execute the target speed and target slope according to the preset speed change rate limit and the preset slope change rate limit; the parameter switching unit is used to switch according to the parameter exchange instruction after the current paired left and right gait cycle ends: when there are two candidates, the sequence is the original parameter group, the first candidate parameter group, the original parameter group, the second candidate parameter group, and the original parameter group; when there is a single candidate, the sequence is the original parameter group, the corresponding candidate parameter group, and the original parameter group; it forms a parameter segment identifier, parameter segment start and end time, parameter exchange completion flag, and parameter exchange exception flag as the actual execution result to be returned; it is also used to execute the equipment parameter correction instructions formed by the withdrawal and reproduction execution sequence and the channel determination unit.

[0014] Preferably, the document management module includes a document generation unit; The file generation unit is used to generate or update training auxiliary files based on the identification and verification results, and write correction attributes according to the control qualification flag and action retention flag: when passing, write device parameter correction attributes and device correction parameter group; when failing and the action retention flag is passed, write action retention monitoring attributes; when selecting the action prompt correction channel, write action prompt correction attributes.

[0015] Furthermore, the file management module also includes a monitoring and alerting unit; The monitoring and prompting unit is used to read training aid files and continuously monitor the target action set. When a new continuous action deviation category is formed, it triggers the data processing module to re-identify and verify it. According to the action prompt correction instruction, the action prompt correction content is retrieved and output through the treadmill display screen, and then written into the action prompt correction correspondence by the file generation unit.

[0016] This invention provides an aerobic fitness assisted training system based on motion capture. It has the following beneficial effects: 1. This invention forms an actual stable parameter time range based on the actual speed of the running belt and the actual slope of the running platform, and determines the effective gait cycle within this range, so that the motion data used for personal benchmark, centering deviation and parameter response evaluation correspond to the stable interval after the target speed and target slope are reached.

[0017] 2. This invention uses a side difference method to calculate the personal baseline and the central deviation from the same source motion parameters, forming a target motion set on a continuous effective gait cycle, and then generating candidate parameter groups in a paired manner. The parameter response residuals obtained by the response method and the comprehensive correction response are combined to determine the parameter group to be verified.

[0018] 3. This invention switches parameters according to the order of withdrawal and reproduction, and uses reproduction to form control qualification mark; when the control qualification mark is passed, a device parameter correction instruction is generated and a device parameter correction channel is selected; when it fails, the original parameter group is maintained or an action prompt correction channel is selected and the action prompt correction content is output. Attached Figure Description

[0019] Figure 1 This is a system architecture diagram of the present invention; Figure 2 This is a flowchart of the method of the present invention; Figure 3 This is a working logic diagram of the motion capture module of the present invention; Figure 4 This is a working logic diagram of the state detection module of the present invention; Figure 5 This is a working logic diagram of the data processing module of the present invention; Figure 6 This is a working logic diagram of the device control module of the present invention; Figure 7 This is a working logic diagram of the file management module of the present invention; Figure 8 Box plots of the absolute values ​​of parameter response residuals obtained by two segmentation methods in the application embodiments of the present invention; Figure 9 This is a graph showing the change in the overall deviation during the withdrawal and reproduction verification process in an application embodiment of the present invention. Detailed Implementation

[0020] The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0021] Please see the appendix Figure 1 This invention provides a motion capture-based aerobic fitness assisted training system, comprising: The motion capture module is used to establish the treadmill coordinate system and a unified time reference to obtain the left and right gait cycles, homogeneous motion parameters, joint confidence, and pelvic center position.

[0022] The status detection module is used to detect the actual speed of the running belt and the actual slope of the running platform, and to determine the actual stable parameter segment, the time range of the actual stable parameter segment, and the valid flag of the parameter status.

[0023] The data processing module is used to filter effective gait cycles, identify continuous motion deviations using a side-difference method, generate candidate parameter groups using a paired method, complete three-level difference and response evaluation using a responsive method, and perform withdrawal and reproduction verification using a reproducibility method to form the corrected channel selection result.

[0024] The equipment control module is used to perform parameter exchange, withdrawal and reproduction of corresponding parameter switching, and correction of parameter switching. It outputs the target parameter group, control command issuance time and parameter segment identifier to the status detection module, and sends the actual execution results back to the data processing module.

[0025] The file management module is used to generate or update training auxiliary files based on the identification and verification results and action prompt correction instructions, continuously monitor the training status and output action prompt correction content.

[0026] See appendix Figure 2 This invention provides a motion capture-based aerobic fitness assisted training method, comprising the following steps: S1, the motion capture module establishes the treadmill coordinate system and a unified time reference, and obtains the left and right gait cycles, homogeneous motion parameters, joint confidence and pelvic center position.

[0027] S2, the status detection module detects the actual speed of the running belt and the actual slope of the running platform, and determines the actual stable parameter segment, the time range of the actual stable parameter segment, and the valid flag of the parameter status.

[0028] S3, the data processing module filters valid gait cycles, uses a side difference method to identify continuous motion deviations, uses a paired method to generate candidate parameter groups, uses a response method to complete three-layer difference and response evaluation, and uses a reproducibility method to perform withdrawal and reproduction verification, forming the corrected channel selection result.

[0029] S4, the equipment control module performs parameter exchange, withdrawal and reproduction corresponding parameter switching and correction channel corresponding parameter switching, outputs the target parameter group, control command issuance time and parameter segment identifier to the status detection module, and sends the actual execution result back to the data processing module.

[0030] S5, the file management module generates or updates training auxiliary files based on the identification and verification results and action prompt correction instructions, continuously monitors the training status and outputs action prompt correction content.

[0031] The technical solutions in the embodiments of the present invention will be further explained below in conjunction with the specific processing procedures.

[0032] In this embodiment, the treadmill is an adjustable speed and adjustable incline treadmill, including a running belt, a running belt drive motor, a running platform, an incline adjustment mechanism, a treadmill controller, a speed control interface, and an incline control interface. The treadmill controller receives the target speed and target incline via the speed control interface and the incline control interface, and controls the running belt drive motor and the incline adjustment mechanism to operate according to preset speed change rate limits and preset incline change rate limits.

[0033] An incremental encoder is installed on the front roller to detect the actual speed of the running belt; an incline sensor is installed on the running table to detect the actual slope of the running table. A timestamp generator assigns timestamps to the actual speed of the running belt and the actual slope of the running table. The outputs of the incremental encoder and the incline sensor are connected to the status detection module.

[0034] The treadmill is equipped with two cameras: a first camera positioned at the front and a second camera positioned on the side. Both cameras are synchronized global shutter cameras and are connected to a camera synchronization controller and camera calibration components. Visual markers are placed at the hip, knee, ankle, heel, toe, and wrist positions of the user. Pre-defined handrail areas are set in the treadmill coordinate system to determine the handrail grip position.

[0035] The data processing module communicates with the treadmill controller, and the treadmill display screen is used to output motion prompts and corrections.

[0036] The treadmill is also equipped with an independent emergency stop device, an independent safety key, an emergency stop button, and a controller watchdog, and features a switching process for constraint parameters such as speed limit, incline limit, speed change rate limit, and incline change rate limit. The training prompt library stores the correspondence between the main target movement parameters and the corresponding movement prompt corrections.

[0037] See appendix Figure 3 In this embodiment, step S1 is executed by the motion capture module, which includes a calibration synchronization unit and a gait parameter unit, used to establish a treadmill coordinate system and a unified time reference, and to obtain the left and right gait cycles, homogeneous motion parameters, joint confidence and pelvic center position.

[0038] S110, the calibration synchronization unit connects the first camera, the second camera, the camera synchronization controller, and the camera calibration component. A treadmill coordinate system is established with the running belt's forward direction as the longitudinal direction, the running belt's width direction as the transverse direction, the direction perpendicular to the running belt surface as the normal direction, and the calibration position on the center line of the running belt surface as the origin.

[0039] The camera calibration component is set in the common acquisition area of ​​the first and second cameras. The calibration synchronization unit obtains the intrinsic and extrinsic parameters of the first and second cameras. The intrinsic parameters are used to convert the image position of the visual marker into the camera imaging ray, and the extrinsic parameters are used to convert the camera imaging ray into the treadmill coordinate system. Based on the spatial intersection position of the corresponding imaging rays of the two cameras, the position of the visual marker in the treadmill coordinate system is determined.

[0040] For the calibration of camera intrinsic and extrinsic parameters, those skilled in the art can use calibration boards and binocular geometric correction. The solution of the calibration matrix is ​​a well-known technique in the field and will not be elaborated here.

[0041] The camera synchronization controller provides synchronization trigger signals to the first and second cameras. The calibration synchronization unit converts the camera frame times, treadmill controller times, incremental encoder detection times, and tilt sensor detection times to a unified time reference, and periodically corrects the clock offset of each device according to the synchronization trigger signal. When the difference in acquisition times between two trainee images exceeds a preset camera synchronization error, a camera synchronization failure state is formed, and the corresponding trainee image is not used for human joint position calculation. The unified time reference is used to align the start and end times of the left and right gait cycles with the actual stable parameter time range in subsequent processing.

[0042] The calibration synchronization unit determines the maximum automatic training speed based on the camera sampling frequency, the shortest foot support time corresponding to each speed level, and the preset minimum number of effective frames.

[0043] The camera sampling frequency is set according to the camera's rated acquisition capability, for example, 60fps to 120fps; the minimum foot support time is set according to the preset allowable training range, training mode, and lower limit of foot support time at the corresponding speed, for example, 0.10s to 0.30s; the minimum number of effective frames is the sum of the preset ground contact confirmation frames and the preset ground lift confirmation frames. The ground contact confirmation frames and ground lift confirmation frames are set according to the continuous frame confirmation requirements of ground contact events and ground lift events, respectively, and the minimum number of effective frames is, for example, 4 frames to 8 frames.

[0044] Within the allowed training range, the camera sampling frequency is multiplied by the shortest foot support time for the corresponding speed level in ascending order of speed, and the result is rounded down to obtain the effective number of frames for the corresponding speed level.

[0045] When the number of effective frames is not less than the minimum number of effective frames, the corresponding speed setting meets the motion capture requirements; the maximum speed that meets the motion capture requirements is determined as the maximum automatic training speed and is used as the upper limit of the speed range of motion capture capability, and the upper limit of the preset slope allowable range is used as the upper limit of the slope of the motion capture capability range.

[0046] The preset allowable training range is set based on the treadmill's maximum speed, training mode, and the speed range confirmed by the trainee before starting training, for example, 1km / h to 18km / h; the allowable incline range is set based on the incline's maximum speed, training mode, and the incline range confirmed by the trainee before starting training. If there is no speed setting that meets the motion capture requirements, the maximum automatic training speed and motion capture capability range are not established, and the data processing module does not enter parameter exchange.

[0047] The calibration synchronization unit outputs the treadmill coordinate system and unified time reference to the gait parameter unit, and outputs the maximum automatic training speed and motion acquisition capability range to the response evaluation unit for range checking of candidate parameter groups.

[0048] S120, the gait parameter unit simultaneously acquires images of the trainee through the first and second cameras, and uses visual landmark positioning to obtain the positions of the human joints of the left and right hips, left and right knees, left and right ankles, left and right heels, left and right toes, and left and right wrists. The gait parameter unit extracts the contour of the same visual landmark from the two trainee images, determines the image position based on the center of the contour, and then determines the position of the human joint in the treadmill coordinate system based on the intrinsic and extrinsic parameters of the first and second cameras. The human joint positions are then arranged according to timestamps to form joint time-series data. For the contour extraction and center positioning of the visual landmark, those skilled in the art can use color segmentation, connected component filtering, and contour center calculation. The image processing process is well-known in the field and will not be described in detail here.

[0049] The pelvic center is located at the midpoint between the left and right hip joints. The trunk center is determined based on the centroid of the trainee's trunk region in the trainee image and transformed to the treadmill coordinate system; the trunk tilt angle is determined based on the angle between the line connecting the pelvic center and the trunk center and the normal to the treadmill coordinate system.

[0050] The confidence score for key points ranges from 0 to 1. The gait parameter unit calculates the visual landmark contour integrity, dual-camera reprojection consistency, and adjacent frame position continuity, respectively.

[0051] The contour integrity is determined by the ratio of the identified contour length to the calibrated contour length. When the calculation result is greater than 1, it is taken as 1. When the reprojection error is equal to 0, the reprojection consistency of the two cameras is taken as 1. When the reprojection error is within the allowable range of the reprojection error, the reprojection consistency of the two cameras is taken as 1 minus the ratio of the reprojection error to the upper limit of the allowable range of the reprojection error. When the reprojection error is greater than the upper limit, it is taken as 0. When the positional change between adjacent frames is 0, the positional continuity between adjacent frames is 1. When the positional change between adjacent frames is within the allowable range, the positional continuity between adjacent frames is calculated as 1 minus the ratio of the positional change between adjacent frames to the upper limit of the allowable range. When the positional change between adjacent frames is greater than this upper limit, the value is 0. All values ​​are limited to 0 to 1.

[0052] The gait parameter unit takes the minimum value among contour integrity, dual-camera reprojection consistency and adjacent frame position continuity as the joint confidence of the corresponding human joint in the current sampling frame.

[0053] The reprojection error tolerance range is set based on the camera calibration residual, and the preset tolerance range for adjacent frame position changes is set based on the camera sampling frequency and the preset allowable training range. When a visual landmark is not simultaneously recognized by both cameras, the confidence level of the corresponding joint point is recorded as 0 and the joint point is missing; adjacent frame data is not used to replace the human joint point position in this frame.

[0054] The gait parameter unit obtains the normal and longitudinal coordinates of the left and right heels and toes from the joint timing data. The normal distance of the foot relative to the running belt is the difference between the normal coordinate of the heel or toe and the normal coordinate of the running belt surface; the longitudinal velocity is determined based on the difference in the longitudinal coordinates of the foot in adjacent valid frames and the corresponding time interval; the direction of movement is determined based on the sign of the longitudinal velocity. When the longitudinal velocity is equal to 0, the direction of movement of the previous valid frame is used. If there is no previous valid frame, the current frame is not used for gait event recognition.

[0055] The ground contact distance range is set based on the calibration error of the running belt surface, the size of the visual markers, and the fluctuation range of the joint positions, for example, 0mm to 30mm. When the foot is in the swing phase, the normal distance between the foot and the running belt continuously falls within the ground contact distance range, and the direction of movement changes from forward to backward with the running belt. After the number of consecutive frames that meet the conditions reaches the ground contact confirmation frame number, the moment when the first frame meets the conditions is determined as the ground contact event, and the foot state changes to the support phase.

[0056] When the foot is in the support phase, the normal distance between the foot and the running belt is continuously greater than the upper limit of the ground contact distance range, and the direction of movement changes from backward with the running belt to forward. After the number of consecutive frames that meet the conditions reaches the preset number of ground clearance confirmation frames, the moment when the first frame meets the conditions is determined as the ground clearance event, and the foot state changes to the swing phase.

[0057] If the determination condition does not reach the corresponding number of confirmation frames, the original gait phase is maintained to avoid repeated ground contact or ground lift events; if the continuous absence of joints causes the gait event to be unconfirmed, the current ipsilateral gait cycle is invalid.

[0058] The data between two consecutive same-side ground contact events constitutes a same-side gait cycle. The gait parameter unit pairs the left and right data according to the alternating left and right ground contact events, forming left and right gait cycles and recording their start and end times. If the left and right ground contact events do not alternate, any same-side gait cycle lacks a ground-leaning event, or the gait cycle duration does not fall within the preset gait cycle duration range, the left and right gait pairing status is considered failed, resulting in an invalid cycle. The preset gait cycle duration range is set based on the gait cycle boundaries within the preset allowable training range, for example, 0.4s to 2s.

[0059] The gait parameter unit calculates six types of homologous motion parameters based on left and right gait events. Left and right support times are the time differences between the same-side ground contact event and the same-side ground lift event, respectively; left and right swing times are the time differences between the same-side ground lift event and the next same-side ground contact event, respectively; left and right stride lengths are the absolute values ​​of the difference between the longitudinal coordinate of the heel of the current grounding foot and the longitudinal coordinate of the heel of the opposite supporting foot at the same moment; and the ground contact positions of the left and right feet relative to the pelvis are the position differences between the heel position and the pelvic center position in the treadmill coordinate system at the time of ground contact.

[0060] The knee joint angle at left and right ground contact is determined by the angle between the spatial vector from hip to knee and the spatial vector from ankle to knee at the time of ground contact; the hip joint angle at left and right ground departure is determined by the angle between the spatial vector from the center of the torso to the hip and the spatial vector from knee to hip at the time of ground departure. The joint angle is obtained by dividing the inner product of the two spatial vectors by the product of their magnitudes, and then performing an inverse cosine operation on the result. If any human joint point involved in the calculation is missing, the corresponding joint angle is not formed, and the current gait cycle is invalid.

[0061] The gait parameter unit determines the handrail grip state based on the continuous positional relationship between the left and right wrist joints and the preset handrail area. The handrail grip state is considered successful if any wrist joint continuously belongs to the preset handrail area for a duration not less than the preset grip judgment duration; otherwise, it is considered unsuccessful. The preset handrail area is set according to the actual boundary of the handrail in the treadmill coordinate system, and the preset grip judgment duration is set according to the upper limit of the duration of normal arm swing across the handrail area, for example, 1s to 3s.

[0062] The motion capture module outputs left and right gait cycles with a unified time reference, homogeneous motion parameters, joint confidence, pelvic center position, trunk tilt angle, handhold status, and left and right gait pairing status to the data processing module. The pelvic center position, trunk tilt angle, joint confidence, and handhold status are only used for safety and data validity assessment and are not substituted into the lateral differential method.

[0063] See appendix Figure 4 In this embodiment, step S2 is executed by the state detection module, which includes an actual state unit and an effective segmentation unit. It is used to detect the actual speed of the running belt and the actual slope of the running platform, form the actual speed sequence of the running belt and the actual slope sequence of the running platform, and determine the actual stable parameter segment, the time range of the actual stable parameter segment, and the effective flag of the parameter state.

[0064] S210, the actual state unit connects to an incremental encoder, an inclinometer, and a timestamp generator. The incremental encoder is located on the front roller of the running belt and outputs pulse signals corresponding to the roller's rotation. The inclinometer is located on the running platform and outputs the running platform's inclinometer angle relative to the horizontal plane. The actual state unit reads the pulse increment and running platform inclinometer angle within a preset sampling period. The sampling period is set based on the number of pulses per cycle, the treadmill control cycle, and the speed change rate limit, for example, 10ms to 50ms.

[0065] The actual speed of the running belt is calculated based on the effective roller diameter, the number of pulses per revolution, the pulse increment within the sampling interval, and the sampling duration. The effective roller diameter is obtained by measuring the actual distance the running belt travels when the roller rotates a predetermined number of times, and then dividing the actual travel distance by the number of revolutions and pi. The product of the effective roller diameter and pi is the distance the running belt travels for one revolution of the roller. The pulse increment within the sampling interval is divided by the number of pulses per revolution to obtain the number of revolutions of the roller within the sampling interval. The number of revolutions of the roller and the distance the running belt travels are multiplied and then divided by the sampling duration to obtain the actual speed of the running belt expressed in m / s, which is then converted to km / h according to standard unit conversion relationships.

[0066] The actual state unit performs low-pass filtering on the continuous speed data and uses the filtered data as the actual speed of the running belt. For low-pass filtering of the speed signal, those skilled in the art can select a digital low-pass filter based on the sampling frequency and the response frequency of the running belt drive system. The determination of the filter order and coefficients is a well-known technique in the art and will not be elaborated upon here. The continuous speed data is retained in the sensor detection record for sensor failure status determination.

[0067] After the tilt sensor is installed, the output when the treadmill is in a horizontal position is used as the zero point. The actual state unit subtracts the zero-point deviation from the current output of the tilt sensor to obtain the treadmill tilt angle, and then takes the tangent of the treadmill tilt angle to obtain the actual slope ratio. The actual slope of the treadmill is expressed as the actual slope ratio; when it needs to be displayed as a percentage, the actual slope ratio is multiplied by 100. The response evaluation unit uses the actual slope ratio when calculating the nominal load.

[0068] The timestamp generator assigns timestamps to each actual running belt speed and actual treadmill incline. The timestamp for the actual running belt speed is taken from the midpoint of the sampling interval, and the timestamp for the actual treadmill incline is taken from the moment the tilt sensor completes sampling, both converted according to a unified time base. The actual state unit arranges the detection results according to the timestamps, forming a timestamped sequence of actual running belt speed and actual treadmill incline, allowing the time when the control command is issued and the time when the treadmill reaches the target state to be recorded separately.

[0069] If the target speed is greater than 0, and the incremental encoder does not output a valid pulse within the preset sensor data waiting time limit, or if the speed corresponding to the pulse increment in the sampling interval is greater than the speed limit, the sensor failure state corresponding to the incremental encoder is "passed". If the target speed is equal to 0 and the incremental encoder does not output a pulse, no sensor failure state is formed.

[0070] If the tilt sensor fails to update data within the preset sensor data waiting time limit, outputs data beyond the measurement range, or the actual slope change of the running platform at adjacent sampling times is greater than the product of the slope change rate limit and the sampling time, the corresponding sensor failure status of the tilt sensor is "pass".

[0071] If the timestamps do not increment sequentially, the corresponding detection data is invalid, the sensor failure status corresponding to the timestamp is "pass," and the sensor failure status is output to the valid segmentation unit. The preset sensor data waiting time limit is set according to the preset sampling period and the communication period of the treadmill controller, for example, 0.1s to 0.5s. When the incremental encoder, tilt sensor, and timestamps all meet the detection requirements, the sensor failure status is "fail."

[0072] The actual state unit outputs the actual speed sequence of the running belt, the actual slope sequence of the running platform, and the sensor failure status to the effective segmentation unit, and outputs the actual speed sequence of the running belt and the actual slope sequence of the running platform to the data processing module.

[0073] S220: The effective segmentation unit receives the actual speed sequence of the running belt, the actual slope sequence of the treadmill, and the sensor failure status. It also obtains the target speed, target slope, control command issuance time, and parameter segment identifier corresponding to the current target parameter group from the equipment control module. The effective segmentation unit uses the target parameter group, control command issuance time, and parameter segment identifier as associated information, and divides the actual stable parameter segments only based on the actual speed of the running belt, the actual slope of the treadmill, and their timestamps, without using the left and right gait cycles as the segmentation condition.

[0074] The effective segmentation unit compares the actual speed of the running belt with a preset speed stability tolerance range centered on the target speed, and compares the actual slope of the treadmill with a preset slope stability tolerance range centered on the target slope. When both the actual speed of the running belt and the actual slope of the treadmill fall within their respective tolerance ranges, the current sampling time meets the stability condition; when either data point falls outside its respective tolerance range, the current sampling time does not meet the stability condition.

[0075] The preset speed stability tolerance range is set based on the incremental encoder measurement error, the running belt speed control accuracy, and the filtered speed fluctuation range. For example, it is a closed interval of ±0.1km / h to ±0.3km / h centered on the target speed. The preset slope stability tolerance range is set based on the tilt sensor measurement error, the slope adjustment mechanism positioning accuracy, and the running platform static fluctuation range. For example, it is a closed interval of ±0.002 to ±0.01 centered on the target slope, where the slope is expressed as the actual slope ratio.

[0076] When the speed and gradient conditions continuously meet the preset stability confirmation time, the parameter stability state is considered satisfied, and the first sampling time that continuously meets the stability conditions is taken as the start time of the actual stable parameter segment. The preset stability confirmation time is set based on the running belt speed fluctuation period, the stabilization time after the gradient adjustment mechanism stops, and the sampling period, for example, 0.5s to 2s. After the parameter stability state is satisfied, if any detection data does not fall within the corresponding tolerance range, the last sampling time that meets the stability condition is taken as the end time of the actual stable parameter segment, and the parameter stability state changes to unsatisfied.

[0077] The actual stable parameter segment consists of the start time, the end time, and the corresponding time range of the actual speed sequence of the running belt and the actual slope sequence of the treadmill. The time range from the start time to the end time is the actual stable parameter segment's time range. When the target parameter set changes, the original actual stable parameter segment terminates at the start of the parameter switch, and the actual stable parameter segment corresponding to the new target parameter set is redefined. Parameter change phases, time ranges where parameter stability conditions are not met, and time ranges where sensor failure conditions are met are not included in the actual stable parameter segment.

[0078] When an actual stable parameter segment is formed, the parameter status validity flag is "passed". Once an actual stable parameter segment has been formed, its parameter status validity flag remains "passed" after the normal operation ends. Before an actual stable parameter segment is formed, the parameter status validity flag for the current acquisition interval is "failed". After an actual stable parameter segment is formed, the parameter status validity flag for subsequent parameter change phases and unstable phases is "failed". If the duration of the stable condition is less than the preset stability confirmation duration, an actual stable parameter segment is not formed.

[0079] From the moment the control command is issued, if the parameter stability state is not met, the effective segment unit continues to receive the actual speed of the running belt and the actual slope of the running platform. If the waiting time exceeds the preset parameter state waiting period and the parameter stability state is still not met, the data acquisition corresponding to the current target parameter group is terminated, and a rollback request is generated. The preset parameter state waiting period is set based on the acceleration and deceleration time of the running belt drive motor, the travel time of the slope adjustment mechanism, the speed change rate limit, and the slope change rate limit, for example, 5s to 30s.

[0080] When the sensor failure status is "pass", the effective segmentation unit takes the last valid timestamp before the time of the anomaly as the end time of the current actual stable parameter segment. The valid flag of the parameter status corresponding to the already formed actual stable parameter segment is kept as "pass", and the valid flag of the parameter status corresponding to the acquisition interval since the time of the anomaly is "fail", and a rollback request is generated.

[0081] The rollback request includes the current target parameter group, the time of the anomaly, and the anomaly conditions, and is output to the channel determination unit, which combines the safety status to form a rollback command or a stop command.

[0082] The effective segmentation unit writes the parameter segment identifier, the actual stable parameter segment time range, the target parameter group, and the control command issuance time into the parameter segment record, and outputs the actual stable parameter segment, the actual stable parameter segment time range, the parameter status validity flag, and the parameter segment identifier to the data processing module.

[0083] The deviation identification unit determines the inclusion relationship between the start and end times of the left and right gait cycles and the actual stable parameter segment time range based on a unified time reference. The response evaluation unit and the channel determination unit establish the correspondence between the actual execution results and the actual stable parameter segments based on the parameter segment identifier, and call the parameter status validity flag in the candidate parameter response evaluation and withdrawal and reproduction verification. Data with the parameter status validity flag failing will not be included in subsequent calculations.

[0084] See appendix Figure 5 In this embodiment, step S3 is executed by the data processing module, which includes a deviation identification unit, a response evaluation unit, and a channel determination unit, and is used to complete effective gait cycle screening, continuous motion deviation identification, candidate parameter response evaluation, withdrawal and reproduction verification, and correction channel selection.

[0085] S310, the deviation identification unit compares the start and end times of the left and right gait cycles with the actual stable parameter segment time range according to a unified time reference. If the start and end times of both left and right gait cycles fall within the actual stable parameter segment time range, the parameter status is valid (passed), the confidence level of each human joint required to form the six types of homogeneous motion parameters is not less than the preset minimum joint confidence level within the corresponding gait event sampling frame, the pelvic center position is within the preset running belt safe area, the handrail gripping status is failed, and the left and right gait pairing status is passed, then the corresponding left and right gait cycles are considered valid gait cycles and written to the motion data cache. All other left and right gait cycles are invalid, and the reasons for invalid cycles are recorded in the motion quality log. The preset minimum joint confidence level is set based on the joint position error, and the preset treadmill safety zone is set based on the treadmill boundary and safety distance.

[0086] A stable segment is formed when there are continuous and effective gait cycles within the actual stable parameter segment, and when there are no safety anomalies or parameter switching. The initial parameter set is formed by the target speed and target incline selected by the trainee before starting training; after the treadmill reaches the target speed and target incline, the median of the actual speed of the running belt and the actual incline of the treadmill within the first actual stable parameter segment is taken to confirm the actual execution status of the initial parameter set.

[0087] Within the stable segment corresponding to the initial parameter set, when the number of effective gait cycles is not less than the preset number of baseline cycles, a personal baseline segment is formed; the preset number of baseline cycles is set according to the gait cycle fluctuation range, for example, 10 to 30 cycles.

[0088] The deviation identification unit uses a side-biased calculation method to calculate the normalized scale, action asymmetry, individual baseline, centered deviation, and aggregated centered deviation. ; In the formula: This refers to the category number of the parameters of the same source action; This refers to the sequence number of the gait cycle; and These are the sets of indices of effective gait cycles within the individual baseline segment and the current stable segment, respectively. and The parameters of the same motion on the left and right sides are respectively. In order to be with the first The preset minimum normalization scale corresponding to similar source action parameters; For normalization; For asymmetric motion quantities; For individual benchmarks; This is a deviation from centralization; To summarize the centralization deviation; , and These are operations for median, larger value, and absolute value, respectively.

[0089] The preset minimum normalization scale is set based on the measurement resolution of various homogeneous motion parameters. The initially formed personal baseline serves as both the original baseline and the current personal baseline. With fixed treadmill parameters and repeated sampling of the same motion state, the absolute value of the difference between the summed center deviation and the median of repeated sampling is calculated for each type of homogeneous motion parameter. The 95th percentile of this absolute value sequence is taken as the measurement uncertainty for that type. If the number of repeated samplings is insufficient to determine the 95th percentile, the maximum value in the absolute value sequence is taken as the measurement uncertainty for that type. A corresponding measurement uncertainty range is formed with the negative of the measurement uncertainty as the lower bound and the measurement uncertainty itself as the upper bound.

[0090] The number of repeated acquisitions is set based on the number of effective gait cycles obtainable under the same parameter set, for example, 10 to 30 times. The median of the absolute value of the centered deviation within the individual baseline segment is multiplied by a constant of 1.4826 to obtain the baseline dispersion; the baseline dispersion is multiplied by a preset motion deviation threshold coefficient and then added to the corresponding measurement uncertainty to obtain the motion deviation threshold.

[0091] If, within a few valid gait cycles of a certain type of centering deviation, the absolute value of such deviation is greater than the corresponding action deviation threshold and the signs are consistent, then the parameters of the same type of homogeneous action are added to the target action set and recorded in the continuous action deviation category. If the action deviation threshold is not exceeded, the cycle is invalid, or the sign is reversed, then the continuous counting restarts.

[0092] When the target action set is empty, the current training parameters are maintained. When the target action set is not empty, preset base weights that correspond one-to-one with the parameters of each type of homogeneous action and whose values ​​are not less than 0 are called. The preset base weights of each type in the set are divided by the sum of the base weights to obtain the action parameter weights, so that the sum of the action parameter weights in the target action set is equal to 1. When the sum of the base weights in the target action set is equal to 0, the set parameters are recorded as abnormal, and the candidate parameter generation is not initiated.

[0093] S320, the response evaluation unit takes the median of the actual speed of the running belt and the actual incline of the treadmill within the current stable parameter segment to form the current treadmill parameter set, and records the current treadmill parameter set as the original parameter set. Based on the training mode, the ratio of the current nominal load to the candidate incline is calculated using a paired approach. ; In the formula: The treadmill speed is expressed in km / h. The speed is expressed in m / min; 1000 and 60 represent the number of meters per kilometer and the number of minutes per hour, respectively. This represents the current actual slope ratio; 3.5 represents the nominal load; 3.5 is a constant term. For the velocity term coefficient; The coefficient of the product term of speed and gradient; The index of the candidate parameter group; , and The first One candidate speed, one candidate slope ratio, and one nominal load; This is the current nominal load; This is the preset nominal load tolerance.

[0094] In fast walking mode, It can be taken as 0.1. You can select 1.8; in jogging mode or endurance running mode, It can be taken as 0.2. You can take 0.9.

[0095] Within the preset speed allowable range, the response evaluation unit obtains candidate speeds that are greater than and less than the current speed according to the preset speed step size. When a complete speed step size cannot be formed, the corresponding boundary speed is taken. When the boundary speed is equal to the current speed or the candidate speed is equal to 0, no candidate parameter group is generated in the corresponding direction.

[0096] If a candidate parameter group simultaneously falls within the preset equipment range, preset safety range, and motion acquisition capability range, and the candidate slope ratio is within the allowable slope range, and the absolute value of the difference in nominal load is not greater than the preset nominal load tolerance, the candidate parameter is marked as valid. The candidate parameter group with higher speed and lower slope is designated as the first candidate parameter group, and the candidate parameter group with lower speed and higher slope is designated as the second candidate parameter group.

[0097] When two candidate parameter sets have valid flags indicating they have passed, bidirectional parameter exchange is initiated; when only one candidate parameter set has valid flags indicating they have passed, unilateral verification is initiated. If no candidate parameter sets have valid flags indicating they have passed, the original parameter set is maintained, the reason for candidate parameter generation failure is recorded, and action prompts for correction are initiated. The candidate parameter set and the original parameter set are written into the parameter exchange queue. The response evaluation unit generates a parameter exchange instruction, which is output to the parameter switching unit. After completing the parameter exchange, the parameter switching unit sends the actual execution result back to the response evaluation unit.

[0098] After obtaining the actual execution results, only the parameter segments whose parameter exchange completion flag is "passed" and whose parameter exchange exception flag is "failed" are processed.

[0099] When the parameter exchange completion flag is "not passed" or the parameter exchange exception flag is "passed", the corresponding candidate parameter group is cancelled, the exception condition is written into the candidate parameter cancellation reason, and a rollback request is reported to the channel determination unit.

[0100] Based on the parameter segment identifier, the candidate parameter segment is aligned with the preceding and following original parameter reference segments; the left and right homogeneous motion difference and the subtraction of the individual reference constitute the first layer of difference; the comparison between the candidate parameter conditions and the original parameter conditions is used as the parameter condition difference, which constitutes the second layer of difference; the time drift subtraction of the preceding and following original parameter reference segments constitutes the third layer of difference.

[0101] The interpolation baseline, parameter response residuals, and comprehensive corrected response are calculated using a response-based method. These are then written into the parameter response record, and the correlation between the comprehensive corrected response and the corresponding candidate parameter group is written into the candidate parameter record. ; In the formula: , and These are the summative centering deviations of the pre-base segment, candidate parameter segment, and post-base segment, respectively. , and These are the center times of the corresponding parameter segments; As the interpolation reference; For parameter response residuals; For the set of target actions; Weights for action parameters; To comprehensively correct the response; For summation operations.

[0102] If a preceding or following reference segment is missing, the center times of the three parameter segments are not arranged in chronological order, or the center time of the preceding reference segment is equal to the center time of the following reference segment, the original parameter reference segment is re-acquired. If, after the original parameter group reaches a stable state again, a preceding or following reference segment that meets the requirements still cannot be formed within an acquisition time that can accommodate approximately several gait cycles of the preset reference cycle, the corresponding candidate parameter group is cancelled, and the overdue acquisition of the reference segment is written into the reason for the cancellation of the candidate parameter. The acquisition time is determined based on the preset number of reference cycles and the upper limit of the preset gait cycle duration range.

[0103] A valid response is formed when the absolute value of the response residual of at least one type of parameter in the target action set is greater than the upper bound of the corresponding measurement uncertainty range, and the overall corrected response is greater than a preset response threshold; otherwise, a valid response is not formed. The preset response threshold is set based on the measurement uncertainty ranges of various types and the weights of the action parameters.

[0104] If only one candidate parameter group has a valid response and does not generate any new non-target action deviation, it is identified as the parameter group to be verified. If both candidate parameter groups have valid responses, and the absolute value of the difference between the comprehensive correction responses is greater than the preset discrimination tolerance, the one with the larger comprehensive correction response is selected. Otherwise, the reason for the cancellation of the candidate parameter is recorded and the process is switched to action prompt correction.

[0105] The preset discrimination tolerance is set based on the repeated acquisition fluctuation range of the comprehensive correction response. New non-target action deviations are determined by the deviation identification unit according to the continuous period of persistent action deviations, action deviation thresholds, and sign consistency rules. When a candidate parameter group causes a new non-target action deviation or reports an abnormal condition, the corresponding candidate parameter group is canceled and a rollback request is reported.

[0106] The response evaluation unit generates candidate parameter selection results based on the above branches, and writes the candidate parameter selection results and the reasons for canceling candidate parameters into the candidate parameter record; the parameter group to be verified is written into the verification task queue. If there is no parameter group to be verified, the candidate parameter selection results are output to the channel determination unit, which generates an action prompt correction instruction.

[0107] S330, the channel determination unit receives the candidate parameter selection result. When the candidate parameter selection result contains a parameter group to be verified, the channel determination unit forms a withdrawal and reproduction execution order, outputs the withdrawal and reproduction execution order to the parameter switching unit, and switches between the parameter group to be verified, the original parameter group, and the parameter group to be verified, and sends the actual execution result back to the channel determination unit.

[0108] If the candidate parameter selection result does not contain the parameter group to be verified, the withdrawal and reproduction verification will not be performed; the channel determination unit will enter the action prompt correction branch according to the candidate parameter selection result.

[0109] The first verification segment, the withdrawal segment, and the second verification segment are formed only when the parameter exchange completion flag is passed, the parameter exchange abnormality flag is failed, the parameter status valid flag is passed, the start and end times of the paired left and right gait cycles all fall within the corresponding actual stable parameter time range, the joint confidence of each human joint required to form the six types of homogeneous motion parameters in the corresponding gait event sampling frame is not less than the preset minimum joint confidence, the pelvic center position is within the preset running belt safe area, and the handrail gripping status is failed while the left and right gait pairing status is passed.

[0110] If any condition is not met, the corresponding verification data will not be substituted into the reproducibility formula.

[0111] Before verification, the baseline is the comprehensive deviation of the most recent stable segment of the original parameter group before the first verification segment.

[0112] The channel determination unit uses a repetitive method to determine control eligibility criteria: ; In the formula: This represents the overall deviation of the corresponding segment; , , and These are the combined deviations from the first verification segment, the withdrawal segment, the second verification segment, and the pre-verification baseline, respectively. To control qualification marks; This is the preset response confirmation threshold; This is the preset reproduction consistency tolerance; This is the preset withdrawal and recovery tolerance; 1 indicates success, and 0 indicates failure.

[0113] , and The measurement uncertainty range of the comprehensive deviation, the repeated application fluctuation range, and the repeated acquisition fluctuation range of the original parameter group are set respectively.

[0114] When there is a parameter group to be verified and the control qualification flag is passed, the action retention flag is failed, a device parameter correction instruction is generated, the parameter group to be verified is determined as the current training parameter, the device parameter correction channel is selected, and the device parameter correction attribute is generated.

[0115] If there is a parameter group to be verified, the control qualification flag is "failed", and the difference between the baseline before verification and the comprehensive deviation of the withdrawal segment is not less than the preset retention judgment threshold, the action retention flag is "passed", the original parameter group is maintained, and action retention monitoring is initiated. If there is a parameter group to be verified, the control qualification flag is "failed", and the action retention judgment condition is not met, the action retention flag is "failed", the original parameter group is restored, and the main target action parameters are determined by the product of the action parameter weight and the difference exceeding the action deviation threshold. If the results are the same, they are determined according to the preset priority, forming an action prompt correction instruction, an action prompt correction channel, and an action prompt correction attribute.

[0116] When there is no parameter group to be verified, the channel determination unit restores or maintains the original parameter group and determines the main target action parameters based on the target action set; when the effective response of the candidate parameter group is not formed, an action prompt correction instruction, an action prompt correction channel, and an attribute of not forming an effective response are formed, and no action prompt correction attribute is formed; when there is no candidate parameter group or the two candidate parameter groups cannot be distinguished, an action prompt correction instruction, an action prompt correction channel, and an action prompt correction attribute are formed.

[0117] If the verification segment does not meet the admission conditions, the verification segment is invalid and must be collected again. If a valid verification segment cannot be formed within the preset verification collection period, the withdrawal and reproduction verification will be terminated, the control qualification flag and the action retention flag will be determined as failing, the parameter group to be verified will be canceled, and the reason for the verification failure will be recorded.

[0118] If the response of the second verification segment relative to the withdrawal segment does not meet the response confirmation threshold, or if the difference between the two responses is greater than the reproduction consistency tolerance, it is determined that the second application cannot reproduce the first response. The control qualification flag and the action retention flag are both determined to be unsuccessful, the parameter group to be verified is cancelled, and the reason for verification failure is recorded. The channel determination unit summarizes the safety status based on human body position, parameter stability, sensor failure status, camera synchronization failure status, handrail grip status, and emergency stop conditions.

[0119] When the safety status is abnormal but the emergency stop condition is not met, a rollback command is generated; when the emergency stop condition is met, a stop command is generated. The channel determination unit generates identification and verification results and outputs them to the file generation unit. The reason for canceling candidate parameters is written into the candidate parameter record, and the reason for verification failure is written into the training event record.

[0120] See appendix Figure 6In this embodiment, step S4 is executed by the device control module, which includes a drive execution unit and a parameter switching unit. The device control module controls the treadmill speed and incline according to parameter exchange instructions, the order of withdrawal and reproduction, device parameter correction instructions, rollback instructions, and stop instructions, and generates the actual execution result. The device control module initiates parameter switching after the left and right gait cycles end, and uses the treadmill's actual state reaching the target parameter set as the condition for completing the parameter exchange.

[0121] S410, the drive actuator unit is communicatively connected to the treadmill controller. The treadmill controller receives the target speed via the speed control interface and the target incline via the incline control interface. It controls the running belt drive motor based on the difference between the target speed and the actual speed of the running belt, and controls the incline adjustment mechanism based on the difference between the target incline and the actual incline of the running platform.

[0122] For closed-loop control of treadmill speed and incline, those skilled in the art can use proportional-integral-derivative control. The tuning of its controller parameters is a well-known technique in the field and will not be elaborated here.

[0123] After receiving the control command, the drive execution unit checks whether the target speed and slope are within the upper limit limits, and whether the target parameter group is within both the preset equipment range and the preset safety range. If the control command lacks a target parameter group, the number of candidate parameter groups is inconsistent with the switching order, or any parameter does not belong to the corresponding range, the control command is not executed, the parameter exchange completion flag is set to fail, the parameter exchange exception flag is set to pass, and the exception condition is fed back to the channel determination unit.

[0124] When the target parameter set meets the execution conditions, the drive unit adjusts the actual speed of the running belt according to the preset speed change rate limit and the actual incline of the running platform according to the preset incline change rate limit. The preset speed change rate limit is set based on the acceleration / deceleration capability of the running belt drive motor, the running belt load, and training safety requirements, for example, 0.1 km / h / s to 0.5 km / h / s; the preset incline change rate limit is set based on the operating speed of the incline adjustment mechanism and the limitations of the running platform structure, for example, 0.005 / s to 0.02 / s. The speed limit, incline limit, preset equipment range, and preset safety range are set based on the treadmill's rated parameters and the parameter range confirmed by the trainee before starting training.

[0125] An independent safety key, emergency stop button, and emergency stop device constitute a stop path independent of the data processing module. When the independent safety key is disengaged, the emergency stop button is triggered, or the emergency stop condition is met, the emergency stop device stops the running belt drive motor. A controller watchdog monitors the communication status between the treadmill controller and the data processing module. If the communication interruption time exceeds the preset controller watchdog monitoring period, the treadmill controller stops receiving incomplete control commands and triggers the emergency stop device. The controller watchdog monitoring period is set based on the communication cycle and control command update cycle, for example, 0.2s to 1s.

[0126] S420, the parameter switching unit receives the parameter exchange command, the candidate parameter group set, and the original parameter group, and obtains the end time of the currently paired left and right gait cycles from the data processing module. The parameter exchange command is initiated after the currently paired left and right gait cycles end to avoid motion data within the same left and right gait cycle corresponding to different target parameter groups.

[0127] If the end time of the left and right gait cycles is not obtained within the preset parameter switching waiting period, the parameter switching unit cancels the current parameter exchange, the parameter exchange completion flag is marked as failed, the parameter exchange exception flag is marked as passed, and the exception condition is fed back to the channel determination unit. The preset parameter switching waiting period is set according to the preset gait cycle duration range, for example, 1 to 3 times the upper limit of the range.

[0128] When a first candidate parameter group and a second candidate parameter group exist, the parameter switching unit performs bidirectional parameter exchange in the order of original parameter group, first candidate parameter group, original parameter group, second candidate parameter group, and original parameter group. When only one candidate parameter group exists, unilateral verification is performed in the order of original parameter group, candidate parameter group, and original parameter group. Unilateral verification is only used to determine whether a candidate parameter group has obtained equipment control qualification, and is not used to determine the relative dominance relationship between speed and gradient.

[0129] At the start of each target parameter group switch, the parameter switching unit generates a parameter segment identifier and sends the target parameter group, parameter segment identifier, and control command issuance time to the status detection module. The parameter segment identifier corresponds to the original parameter group, the first candidate parameter group, the second candidate parameter group, the parameter group to be verified, or the original parameter group that has been withdrawn from use. The control command issuance time serves as the parameter segment start time, and the parameter segment end time is formed when the next target parameter group switch begins or when the current parameter segment terminates abnormally.

[0130] When the parameter stability condition is not met, the equipment control module only performs parameter switching, does not collect action data for response evaluation, and pauses target action prompts. The parameter stability condition is met when the actual speed of the running belt and the actual slope of the running platform belong to the speed stability tolerance range and slope stability tolerance range corresponding to the target parameter group, respectively, and the status detection module completes stability confirmation and forms a valid parameter status flag indicating a passed actual stable parameter segment. The parameter exchange completion flag is then marked as passed.

[0131] The parameter exchange completion flag indicates that the target parameter group has reached a stable state; the data processing module filters action data based on the actual stable parameter time range.

[0132] When the parameter stability condition is not met, the parameter switching unit maintains the current target parameter group and waits for the status detection result within the preset parameter status waiting period. If the waiting time exceeds the preset parameter status waiting period and the parameter stability condition is still not met, the current parameter segment is terminated, the parameter exchange completion flag is marked as failed, the parameter exchange exception flag is marked as passed, and the abnormal condition of not reaching the target parameter group is reported.

[0133] When a parameter segment ends normally, the parameter exchange anomaly flag is set to "failed". If the target parameter group is not reached, a communication error occurs, a control command error occurs, or a rollback command aborts the parameter switching, the parameter exchange anomaly flag is set to "passed", and the time of the anomaly is taken as the parameter segment end time. The parameter segment identifier, parameter segment start and end times, parameter exchange completion flag, and parameter exchange anomaly flag are collectively referred to as the actual execution result. The parameter switching unit writes the actual execution result into the parameter segment record and sends it back to the response evaluation unit and the channel determination unit. The parameter segment start and end times include the parameter segment start time and the parameter segment end time.

[0134] When the parameter switching unit performs withdrawal and reproduction verification, it switches in the order of parameter group to be verified, original parameter group, and parameter group to be verified, and forms a parameter segment identifier for each switch. After each parameter segment is aligned with the time range and effective gait period of the actual stable parameter segment, it is used by the channel determination unit to form the first verification segment, withdrawal segment, and second verification segment; if the parameter exchange completion flag of any parameter segment is "failed" or the parameter exchange abnormality flag is "passed", the corresponding verification segment is not formed.

[0135] Upon receiving a device parameter correction instruction, the parameter switching unit determines the candidate parameter group verified through withdrawal and reproduction as the current training parameters. Upon receiving a rollback instruction, it cancels any unexecuted control instructions in the parameter exchange queue and restores the most recently stable parameter group according to the speed change rate limit and the slope change rate limit. The most recently stable parameter group is the most recent target parameter group whose parameter state was valid before the anomaly occurred.

[0136] After restoring the most recently stable parameter set, if a stable parameter state is not achieved within the preset parameter state waiting period, the parameter switching unit executes controlled deceleration, reducing the target speed to 0 km / h according to the speed change rate limit; after the running belt stops, the target incline is restored to a horizontal position on the running platform according to the incline change rate limit. If an emergency stop condition is met during controlled deceleration, the emergency stop device stops the treadmill. When the parameter switching unit receives a stop command, it triggers the emergency stop device and writes the stop time and abnormal conditions into the training event log.

[0137] See appendix Figure 7 In this embodiment, step S5 is executed by the file management module, which includes a file generation unit and a monitoring and prompting unit. The file management module generates or updates training auxiliary files based on the identification and verification results and action prompt correction instructions, and performs continuous monitoring, action prompt output, data recording, and safety anomaly reporting. The training auxiliary files serve as the basis for identification and verification in subsequent training phases and do not replace the real-time control instructions received by the device control module.

[0138] S510, the document generation unit receives the identification and verification results output by the channel judgment unit. The identification and verification results include individual baselines, target action sets, parameter response residuals, comprehensive corrected responses, control qualification indicators, action retention indicators, withdrawal and reproduction verification records, and correction channel selection results. Correction channel selection results include equipment parameter correction channels and equipment correction parameter groups, or action prompt correction channels and main target action parameters.

[0139] After the first round of identification and verification is completed, the file generation unit writes the personal baseline into the current personal baseline, writes the continuous movement deviation category corresponding to the target movement set into the training auxiliary file, and forms the correction attribute based on the control qualification mark, movement retention mark and correction channel selection results.

[0140] If no valid response is formed for a candidate parameter group, only the "no valid response formed" attribute is generated. If a valid response is formed for a candidate parameter group, and the control qualification flag is passed, then a device parameter correction attribute is generated and written into the device correction parameter group. If the control qualification flag is failed but the action hold flag is passed, then an action hold monitoring attribute is generated. If the control qualification flag is failed, the action hold flag is failed, and an action prompt correction instruction has already been generated, then an action prompt correction attribute is generated. Each continuous action deviation category corresponds to only one correction attribute in the same round of identification and verification.

[0141] The training support file should include at least the current personal baseline, the category of persistent movement deviation and its correction attributes, the equipment correction parameter group, the movement cue correction correspondence, the withdrawal and reproduction verification record, and the periodic personal baseline update record. The original baseline is stored in the personal baseline record and corresponds to the training support file. When the movement cue correction correspondence has not yet been formed, it is not replaced by the main target movement parameters; after the monitoring cue unit completes the movement cue output, the file generation unit appends the movement cue correction correspondence.

[0142] When a training aid file already exists, the file generation unit reads the current personal baseline, the original baseline, persistent motion deviation categories, and existing withdrawal and reproduction verification records. Newly discovered persistent motion deviation categories are added to the target motion set, and a new round of parameter response residuals, comprehensive corrected responses, withdrawal and reproduction verification records, correction channel selection results, and correction attributes are appended and written. When the same persistent motion deviation category reappears, the original record is retained, without overwriting the original baseline, existing withdrawal and reproduction verification records, or interim personal baseline update records.

[0143] The file generation unit provides training auxiliary files to the monitoring and prompting unit for reading. For the storage of training auxiliary files and various training records, those skilled in the art can use relational databases or structured files; the storage media and database field organization are well-known technologies in the field and will not be elaborated upon here.

[0144] S520: After the correction channel selection is completed and the training auxiliary file is formed or updated, the monitoring prompt unit reads the current personal baseline, continuous movement deviation category, correction attribute, equipment correction parameter group and movement prompt correction correspondence, and continuously monitors the target movement parameters, non-target movement parameters, pelvic center position, gait cycle stability, actual running belt speed, actual treadmill slope and joint confidence according to the effective gait cycle.

[0145] Gait cycle stability is satisfied when the gait cycle duration falls within the preset range, the left-right gait pairing is successful, and all consecutive cycles are valid gait cycles; otherwise, gait cycle stability is unsatisfactory. When the treadmill maintains current training parameters, the parameter status is valid (passed), gait cycle stability is satisfied, the number of consecutive valid gait cycles is not less than the preset stage baseline number, there are no safety anomalies, parameter exchanges, withdrawals, or replication verifications, and all types of centralization deviations do not meet the continuous movement deviation judgment conditions, the monitoring and prompting unit measures the median of various movement asymmetries within the corresponding stable segment to form a stage-based personal baseline. The preset stage baseline number is set based on the number of valid gait cycles and the gait cycle fluctuation range of the personal baseline segment, for example, 10 to 30 cycles.

[0146] The archive generation unit updates the current personal benchmark with a phased personal benchmark and writes the current personal benchmark before the update, the phased personal benchmark, and the corresponding current training parameters into the phased personal benchmark update record. The original benchmark and the withdrawal and reproduction verification records are retained. The phased personal benchmark is used for centering deviation calculation in subsequent training phases and does not change the already completed candidate parameter response evaluation and withdrawal and reproduction verification results.

[0147] The monitoring and prompting unit monitors the centralization deviation of recent effective gait cycles relative to the current personal baseline according to the judgment rules of continuous movement deviation in step S310. If, within several effective gait cycles of the most recent consecutive continuous judgment cycle, the absolute value of a certain type of centralization deviation is greater than the corresponding movement deviation threshold and the signs are consistent, and the homogeneous movement parameters of this type are not included in the current continuous movement deviation category, it is determined as a new continuous movement deviation.

[0148] The monitoring and prompting unit outputs the recent effective gait cycles, the current personal baseline, and the corresponding persistent movement deviation category to the data processing module. This triggers the deviation identification unit to re-identify and verify the deviations, forming a target movement set, and continuing to perform candidate parameter generation, response evaluation, and withdrawal and reproduction verification. The file generation unit updates the training aid file based on the results of the new round of identification and verification.

[0149] After the channel determination unit generates an action cue correction instruction, the monitoring cue unit retrieves the action cue correction content from the training cue database based on the category of the main target action parameters and the centering deviation sign. The training cue database stores the correspondence between six categories of homogeneous action parameters, the centering deviation direction, and the action cue correction content.

[0150] When the corresponding content is retrieved, the treadmill display screen outputs the action prompt correction content, and forms an action prompt correction correspondence containing the main target action parameters, the centering deviation direction, and the output content; the file generation unit appends the action prompt correction correspondence to the training assistance file after the prompt output.

[0151] If the training cue library does not contain any corrective action cue content corresponding to the main target action parameters and the centering deviation direction, the monitoring cue unit will not output the corrective action cue content, will write the reason for the retrieval failure into the training event log, and will continue monitoring while maintaining the original parameter set. After the correspondence in the training cue library is confirmed and updated, the monitoring cue unit can re-retrieve in subsequent valid gait cycles; updating the training cue library does not change the established control qualification flags and withdrawal and reproduction verification records.

[0152] The monitoring and prompting unit is equipped with personal baseline records, gait cycle records, parameter segment records, parameter response records, candidate parameter records, parameter eligibility tables, movement quality records, training event records, and training auxiliary files.

[0153] The Individual Benchmark Record stores the individual benchmark, original benchmark, current individual benchmark, and interim individual benchmark; the Gait Cycle Record stores the movement asymmetry, centering deviation, and effective gait cycle determination results; the Parameter Segment Record stores the actual stable parameter segment time range, summarized centering deviation, and actual execution results; the Parameter Response Record stores the interpolation benchmark, parameter response residuals, and comprehensive corrected response; the Candidate Parameter Record stores the original parameter set, candidate parameter validity flags, reasons for candidate parameter generation failure, candidate parameter selection results, and reasons for candidate parameter cancellation; the Parameter Qualification Table stores the correspondence between candidate parameter sets and control qualification flags; the Movement Quality Record stores the reasons for invalid cycles; and the Training Event Record stores the reasons for verification failure, retrieval failure, and safety anomaly conditions.

[0154] If the following conditions occur: the actual speed of the running belt does not fall within the preset speed stability tolerance range corresponding to the target speed and the duration exceeds the preset parameter status waiting period; the actual slope of the running platform does not fall within the preset slope stability tolerance range corresponding to the target slope and the duration exceeds the preset parameter status waiting period; the sensor failure status is "pass"; the camera synchronization failure status is "pass"; the number of consecutive invalid cycles is not less than the preset number of continuous judgment cycles; the pelvic center position does not fall within the preset running belt safety area; the handrail grip status is "pass"; the number of consecutive invalid cycles caused by the inability to pair left and right gaits is not less than the preset number of continuous judgment cycles; the candidate parameter group generates new non-target action deviations; or the user performs a stop operation, the monitoring prompt unit terminates the current parameter identification and reports the abnormal conditions to the channel judgment unit.

[0155] When there are no candidate parameter groups, the effective response of a candidate parameter group is not formed, or the two candidate parameter groups cannot be distinguished, the channel determination unit cancels the corresponding candidate parameter group based on the candidate parameter selection result and generates an action prompt correction instruction. The monitoring prompt unit executes the prompt retrieval and output based on the action prompt correction instruction. When the previous or subsequent reference segment is missing, the response evaluation unit re-collects the original parameter reference segment; when the parameter stability state is not satisfied, no action data for response evaluation is collected.

[0156] When a sensor fails, its parameters become unstable after a certain period, or parameter exchange cannot continue, the monitoring and prompting unit generates a rollback request and outputs it to the channel determination unit. The channel determination unit summarizes the safety status based on the human body position, parameter stability, sensor failure status, camera synchronization failure status, handrail status, and emergency stop conditions, and generates a rollback command or stop command. The equipment control module restores the most recently stable parameter set, executes controlled speed reduction, or stops the treadmill according to the corresponding command.

[0157] To further aid in understanding the technical solutions of the embodiments of the present invention, a specific application example is given below.

[0158] This application example uses repeatable exemplary test data to illustrate the processing procedure. The trainee selects jogging mode, with an initial treadmill speed of 8 km / h and an actual incline ratio of 0.020. The camera sampling frequency is 100 fps, the lowest joint confidence level is 0.80, the speed stability tolerance range is target speed ±0.2 km / h, the incline stability tolerance range is target incline ±0.005, and the continuous judgment cycle count is 8. The response evaluation unit uses a paired conversion to calculate the current speed: ; The converted speed is 133.333 m / min. In jogging mode, the speed coefficient is 0.2, the coefficient for the speed-gradient product is 0.9, and the current nominal load is: ; Using 8.5 km / h and 7.5 km / h as candidate speeds, the converted speeds are 141.667 m / min and 125 m / min, respectively. A paired inverse solution is used to determine the candidate slope ratios. ; ; The first candidate parameter group consists of 8.5 km / h and 0.00575, and the second candidate parameter group consists of 7.5 km / h and 0.03615. Both candidate parameter groups fall within the equipment's range, safety range, and motion acquisition capability range, and meet the nominal load tolerance requirements.

[0159] Within an effective gait cycle, the left and right support times are 0.34s and 0.38s, respectively, with a normalized scale of 0.36s and a personal baseline of -0.020. Lateral deviation calculations are used. ; ; The median absolute value of the centering deviation within the individual baseline segment is 0.018, the motion deviation threshold coefficient is 1.5, and the measurement uncertainty is 0.012. The baseline dispersion is 0.0267, and the motion deviation threshold is 0.052. When the centering deviation for eight consecutive effective gait cycles is less than -0.052, the deviation identification unit adds the left and right support times to the target motion set, and the corresponding motion parameter weight is set to 1.

[0160] The summative centering deviation of the first candidate parameter segment is -0.090 for the preceding baseline segment and -0.086 for the following baseline segment. The center time of the candidate parameter segment is located at the midpoint between the center times of the two original parameter baseline segments, with an interpolation baseline of -0.088. The summative centering deviation of the first candidate parameter segment is -0.050, and the parameter response residual and the comprehensive corrected response are both 0.038. The summative centering deviation of the second candidate parameter segment is -0.079, and the parameter response residual and the comprehensive corrected response are both 0.009. The upper bound of the measurement uncertainty range is 0.012, and the response threshold is 0.020. Therefore, the effective response of the first candidate parameter group is formed and it is determined to be the parameter group to be verified.

[0161] The combined deviations from the pre-verification baseline, the first verification segment, the withdrawn segment, and the second verification segment obtained from the withdrawal and reproduction verification were 0.087, 0.052, 0.089, and 0.055, respectively. The response confirmation threshold was 0.025, the reproduction consistency tolerance was 0.010, and the withdrawal recovery tolerance was 0.010. The corresponding judgment results are as follows: ; ; ; ; If all four conditions are met simultaneously, the control qualification flag is passed. The channel determination unit generates a device parameter correction instruction, and the parameter switching unit determines the first candidate parameter group as the current training parameters.

[0162] Further, using the same training mode and candidate parameter groups, 12 rounds of parameter exchange data were organized. The absolute values ​​of the parameter response residuals were calculated according to two methods: segmentation based on the actual stable parameter segment and segmentation based on the control command issuance time. The average absolute value of the parameter response residuals obtained by the segmentation method based on the actual stable parameter segment was 0.0382, with a standard deviation of 0.0026; the average value obtained by the segmentation method based on the control command issuance time was 0.0278, with a standard deviation of 0.0158. In this set of data, the results obtained by the segmentation method based on the actual stable parameter segment showed less dispersion, indicating that after excluding the parameter change phase, the parameter response residuals in each round had good consistency.

[0163] See appendix Figure 8 The gray boxes represent the interquartile range of the data, the horizontal lines inside the boxes represent the median, the upper and lower whiskers represent the range of non-outlier data, and the black diamonds and the average values ​​in the legend represent the average values ​​of the corresponding segmentation methods.

[0164] See appendix Figure 9 Hollow circles, solid squares, hollow triangles, and solid rhombuses represent the four data segments mentioned above, respectively.

[0165] While specific embodiments of the present invention have been described above, those skilled in the art should understand that these specific embodiments are merely illustrative. Those skilled in the art can omit, substitute, and modify the details of the above methods and systems in various ways without departing from the principles and essence of the present invention. For example, combining the above method steps to perform substantially the same function and achieve substantially the same result using substantially the same method falls within the scope of the present invention.

Claims

1. An action-capture-based aerobic fitness assistant training system, characterized by, include: The motion capture module is used to establish a unified time reference and form left and right gait cycles and parameters of the same motion source; The status detection module is used to detect the actual speed of the treadmill belt and the actual incline of the running platform, forming a time range of actual stable parameters and a valid indicator of parameter status. The data processing module is used to determine the effective gait cycle based on the left and right gait cycles, the homogeneous motion parameters, the actual stable parameter time range, and the parameter state validity flag, to form a personal benchmark, a centralized deviation, and a target motion set, to generate candidate parameter groups in a paired manner and to form parameter exchange instructions, to determine the parameter group to be verified based on the actual execution results and to form a withdrawal and reproduction execution order, and then to form a control qualification flag based on the actual execution results in a reproduction manner, so as to select the device parameter correction channel or the motion prompt correction channel and form an identification and verification result; The equipment control module is used to execute the parameter exchange command, the withdrawal and reproduction execution sequence and the equipment parameter correction command, form the actual execution result and send it back to the data processing module; The file management module is used to generate or update training assistance files based on the identification and verification results and output the corrected content of the action prompts on the treadmill display screen.

2. The motion capture-based aerobic fitness assistant training system according to claim 1, wherein, The motion capture module includes: The calibration synchronization unit is used to convert the camera frame time, treadmill controller time, incremental encoder detection time and tilt sensor detection time to the unified time reference, establish the treadmill coordinate system and determine the motion acquisition capability range. The gait parameter unit is used to form the left and right gait cycles, the homologous motion parameters, joint confidence, pelvic center position, handhold status and left and right gait pairing status based on synchronously acquired trainee images.

3. The aerobic fitness assisted training system based on motion capture according to claim 1, characterized in that, The status detection module includes: The actual state unit is used to detect the actual speed of the running belt and the actual slope of the running platform, and to form a sequence of actual running belt speed and a sequence of actual running platform slope. An effective segmentation unit is used to form an actual stable parameter segment, the time range of the actual stable parameter segment, and a valid flag of the parameter state when the actual speed sequence of the running belt and the actual slope sequence of the running platform belong to a preset speed stability tolerance range and a preset slope stability tolerance range centered on the target speed and target slope of the equipment control module, respectively.

4. The aerobic fitness assisted training system based on motion capture according to claim 1, characterized in that, The data processing module includes a deviation identification unit; The deviation identification unit is used to determine the effective gait cycle when the start and end times of the left and right gait cycles all fall within the time range of the actual stable parameter segment, the parameter status is valid (pass), the joint confidence output by the motion capture module is not less than the preset minimum joint confidence, the pelvic center position is within the preset running belt safe area, and the handhold state and the left and right gait pairing state are respectively "pass" and "pass". The deviation identification unit is also used to calculate the motion asymmetry, the individual baseline, the centralized deviation, and the aggregated centralized deviation using a side-difference method based on the homologous motion parameters. If, within a few consecutive preset determination periods of the effective gait cycles, the absolute value of the centering deviation is greater than the action deviation threshold and the signs are consistent, a continuous action deviation category and the target action set are formed.

5. The motion capture-based aerobic fitness assisted training system according to claim 4, characterized in that, The data processing module also includes a response evaluation unit; The response evaluation unit is used to form an original parameter set based on the median of the actual speed of the running belt and the actual slope of the running platform within the current actual stable parameter time range. The paired method is used to generate different candidate parameter sets of speed and slope when the absolute value of the difference between the nominal loads is not greater than the preset nominal load tolerance. When the candidate parameter group belongs to the preset equipment range, preset safety range, and preset motion acquisition capability range, the candidate parameter is marked as valid. If there are two candidate parameter groups with valid flags indicating they have passed, bidirectional parameter exchange is initiated. If there is only one candidate parameter group with valid flags indicating they have passed, unilateral verification is initiated. If neither passes, the original parameter group is maintained and the action prompt correction is initiated, thus forming the parameter exchange instruction.

6. The motion capture-based aerobic fitness assisted training system according to claim 5, characterized in that, The response evaluation unit is also used for: After receiving the actual execution result, only the parameter segments whose parameter exchange completion flag is passed and parameter exchange exception flag is failed are processed. The candidate parameter segments are aligned with the original parameter reference segments before and after. The left and right homogeneous action difference is completed and the personal reference and parameter condition difference and time drift are deducted. The response formula is used to calculate the parameter response residual and comprehensive correction response based on the summary central deviation. The set of parameters to be verified is determined by combining the response expression: An effective response is formed when the absolute value of the response residual of at least one type of parameter is greater than the upper bound of the measurement uncertainty range and the overall corrected response is greater than the preset response threshold. If only one candidate parameter group has a valid response, it is selected. If both candidate parameter groups have valid responses and the absolute value of the difference between the combined corrected responses is greater than the preset discrimination tolerance, the larger one is selected. Otherwise, the candidate is canceled and the process proceeds to the action prompt correction.

7. The motion capture-based aerobic fitness assisted training system according to claim 6, characterized in that, The data processing module also includes a channel determination unit; The channel determination unit is used to form a first verification segment, a withdrawal segment, and a second verification segment when the actual execution result satisfies the parameter exchange completion flag as passed, the parameter exchange abnormality flag as failed, and the paired left and right gait cycles in the corresponding segment satisfy the determination condition of the effective gait cycle. The comprehensive deviation of the actual stable parameter segment corresponding to the most recent original parameter group before the first verification segment is used as the pre-verification benchmark, and the control qualification mark is formed by the reproducibility formula. If the parameter group to be verified is not available, generate an action prompt correction instruction and select the action prompt correction channel; When the control qualification flag is passed, the device parameter correction instruction is generated and the device parameter correction channel is selected; If the verification fails and the difference between the pre-verification baseline and the withdrawal segment is not less than the preset retention judgment threshold, the action retention flag is passed and the original parameter set is maintained. If the verification fails and the difference between the pre-verification benchmark and the withdrawal segment is less than the preset retention judgment threshold, the action retention flag is failed, an action prompt correction instruction is generated, and the action prompt correction channel is selected. If the problem cannot be reproduced or the verification segment is invalid, the parameter group to be verified is cancelled. Output the identification and verification results.

8. The motion capture-based aerobic fitness assisted training system according to claim 7, characterized in that, The device control module includes: The drive execution unit is used to execute the target speed and target slope according to the preset speed change rate limit and the preset slope change rate limit; A parameter switching unit, used to switch according to the parameter exchange instruction after the currently paired left and right gait cycles have ended: When there are two candidates, the order is: original parameter set, first candidate parameter set, original parameter set, second candidate parameter set, original parameter set; When there is only one candidate, the order is: original parameter set, corresponding candidate parameter set, and original parameter set. The parameter segment identifier, parameter segment start and end time, parameter exchange completion flag, and parameter exchange exception flag are generated and sent back as the actual execution result. It is also used to execute the device parameter correction instructions formed by the withdrawal and reproduction execution sequence and channel determination unit.

9. The aerobic fitness assisted training system based on motion capture according to claim 1, characterized in that, The document management module includes a document generation unit; The file generation unit is used to generate or update the training assistance file based on the identification and verification results, and to write correction attributes according to the control qualification flag and the action retention flag therein: Write the device parameter correction attribute and device correction parameter group during the process; If the action fails but the action-holding flag is set to pass, write the action-holding monitoring attribute. When the aforementioned action prompt correction channel is selected, the action prompt correction attribute is written.

10. The motion capture-based aerobic fitness assisted training system according to claim 9, characterized in that, The file management module also includes a monitoring and alerting unit; The monitoring and prompting unit is used to read the training assistance file and continuously monitor the target action set. When a new continuous action deviation category is formed, the data processing module is triggered to re-identify and verify it. The motion prompt correction content is retrieved according to the motion prompt correction instruction and output through the treadmill display screen, and then written into the motion prompt correction correspondence by the file generation unit.

Images