Granted Patent B2

US 12,566,501 · App. 18/183,923 · Granted Mar 3, 2026

Inventors: Lei Su (Shenzhen, CN); Xin Zhou (Shenzhen, CN); Meiqi Li (Shenzhen, CN); Fengyun Liao (Shenzhen, CN)

Assignee: SHENZHEN SHOKZ CO., LTD.

G06F3/015A61B5/1116A61B5/296A61B5/397A61B5/7225G06F18/2415G16H40/67G16H50/20

Abstract

Claims (98)

1 . A motion monitoring method, comprising:

obtaining a movement signal of a user during motion, the movement signal comprising at least an electromyographic signal and an attitude signal; and

segmenting, based on feature information corresponding to the electromyographic signal and feature information corresponding to the attitude signal, the movement signal, wherein the feature information corresponding to the electromyographic signal includes at least frequency information or amplitude information, and the feature information corresponding to the attitude signal includes at least one of an angular velocity direction, an angular velocity value, an acceleration of an angular velocity, an angle, displacement information, and stress;

determining, based on at least one segment of the movement signal through a movement recognition model, movement-related information, wherein the movement recognition model is a trained machine learning model, the movement-related information includes movement quality and the movement quality includes a standard movement or a wrong movement; and

in responding to determining that the movement quality is the wrong movement, controlling an electromyography sensor to generate an electrical stimulation signal to prompt the user to make a motion adjustment.

2 . The motion monitoring method of claim 1 , wherein the segmenting, based on the feature information corresponding to the electromyographic signal and the feature information corresponding to the attitude signal, the movement signal includes:

segmenting the attitude signal the feature information corresponding to the attitude signal according to operations including:

determining, based on a time domain window of the attitude signal, at least one target feature point from the time domain window according to a preset condition, wherein the at least one target feature point includes one of a movement start point, a movement middle point, and a movement end point; and

segmenting, based on the at least one target feature point, the attitude signal.

3 . The motion monitoring method of claim 2 , wherein the preset condition includes one or more of a change in the angular velocity direction corresponding to the attitude signal; the angular velocity corresponding to the attitude signal being greater than or equal to an angular velocity threshold; a changed value of the angular velocity value corresponding to the attitude signal being an extreme value; and the angle corresponding to the attitude signal reaching an angular threshold;

wherein the preset condition further includes the acceleration of the angular velocity corresponding to the attitude signal being continuously greater than or equal to an acceleration threshold of the angular velocity for a first specific time range.

4 . The motion monitoring method of claim 1 , wherein segmenting, based on the feature information corresponding to the electromyographic signal and the feature information corresponding to the attitude signal, the movement signal includes:

segmenting the electromyographic signal based on the feature information corresponding to the electromyographic signal according to operations including:

pre-processing the electromyographic signal in a frequency domain or a time domain;

obtaining, based on the pre-processed electromyographic signal, the feature information corresponding to the electromyographic signal; and segmenting the electromyographic signal based on the feature information corresponding to the electromyographic signal.

5 . The motion monitoring method of claim 4 , wherein the pre-processing the electromyographic signal in a frequency domain or a time domain includes:

filtering the electromyographic signal to select components of the electromyographic signal in a specific frequency range in the frequency domain.

6 . The motion monitoring method of claim 4 , wherein the pre-processing the electromyographic signal in a frequency domain or a time domain includes:

performing a signal correction processing on the electromyographic signal in the time domain.

7 . The motion monitoring method of claim 6 , wherein the performing a signal correction processing on the electromyographic signal in the time domain includes:

determining a singularity in the electromyographic signal, wherein the singularity corresponds to an abrupt signal of the electromyographic signal; and

performing the signal correction processing on the singularity in the electromyographic signal;

wherein the performing the signal correction processing on the singularity in the electromyographic signal includes:

removing the singularity, or correcting the singularity according to a signal around the singularity.

8 . The motion monitoring method of claim 7 , wherein the singularity includes a burr signal, the determining the singularity in the electromyographic signal includes:

selecting, based on the time domain window of the electromyographic signal, different time windows from the time domain window of the electromyographic signal, wherein the different time windows respectively cover different time ranges; and

determining, based on the feature information corresponding to the electromyographic signal in the different time windows, the burr signal.

9 . The motion monitoring method of claim 8 , wherein the determining, based on the feature information corresponding to the electromyographic signal in the different time windows, the burr signal, includes:

determining first amplitude information corresponding to the electromyographic signal within a first time window and second amplitude information corresponding to the electromyographic signal within a second time window, wherein the first time window and the second time window are two adjacent time windows;

determining whether a ratio of the second amplitude information to the first amplitude information is greater than a threshold:

in response to determining that the ratio of the second amplitude information to the first amplitude information is greater than the threshold, performing a signal correction processing on the electromyographic signal within the second time window;

in response to determining that the ratio of the second amplitude information to the first amplitude information is not greater than the threshold, retaining the electromyographic signal within the second time window.

10 . The motion monitoring method of claim 1 , further comprising determining, based on the attitude signal, the feature information corresponding to the attitude signal, wherein the attitude signal comprises coordinate information in at least one original coordinate system; and

determining, based on the attitude signal, the feature information corresponding to the attitude signal comprises:

obtaining a target coordinate system and a conversion relationship between the target coordinate system and the at least one original coordinate system;

converting, based on the conversion relationship, the coordinate information in the at least one original coordinate system to coordinate information in the target coordinate system; and

determining, based on the coordinate information in the target coordinate system, the feature information corresponding to the attitude signal;

wherein the target coordinate system changes as an orientation of the user changes.

11 . The motion monitoring method of claim 10 , wherein the attitude signal includes coordinate information generated by at least two sensors, the at least two sensors are located at different motion parts of the user and correspond to different original coordinate systems, the determining, based on the attitude signal, the feature information corresponding to the attitude signal includes:

determining feature information corresponding to each of the at least two sensors based on the conversion relationship between different original coordinate systems and the target coordinate system; and

determining, based on the feature information respectively corresponding to the at least two sensors, a relative motion between the motion parts of the user.

12 . The motion monitoring method of claim 10 , wherein the conversion relationship between the at least one original coordinate system and the target coordinate system is obtained by a calibration process including:

constructing a specific coordinate system, the specific coordinate system being related to an orientation of the user during the calibration process;

obtaining first coordinate information of the at least one original coordinate system when the user is in a first pose;

obtaining second coordinate information of the at least one original coordinate system when the user is in a second pose; and

determining the conversion relationship between the at least one original coordinate system and the specific coordinate system according to the first coordinate information, the second coordinate information, and the specific coordinate system;

wherein the calibration process further includes:

obtaining a conversion relationship between the specific coordinate system and the target coordinate system; and

determining, according to the conversion relationship between the at least one original coordinate system and the specific coordinate system and the conversion relationship between the specific coordinate system and the target coordinate system, the conversion relationship between the at least one original coordinate system and the target coordinate system.

13 . The motion monitoring method of claim 10 , wherein the attitude signal includes coordinate information generated by at least two sensors, the at least two sensors are located at different locations of a same motion part of the user and correspond to different original coordinate systems, the determining, based on the coordinate information in the target coordinate system, the feature information corresponding to the attitude signal includes:

determining, based on the coordinate information in original coordinate systems corresponding to multiple sensors of different types by Kalman filtering, the coordinate information in the target coordinate system.

14 . The motion monitoring method of claim 1 , wherein the movement signal further comprising: an electro-cardio signal, a temperature signal, a humidity signal, a blood oxygen concentration, and a respiration rate; and the movement recognition model includes one or more of a machine learning model that classifies the movement signal of the user, a machine learning model that recognizes the movement quality of the user, a machine learning model that recognizes a number of movements of the user, and a machine learning model that recognizes a fatigue level of the user performing the movement.

15 . The motion monitoring method of claim 1 , wherein a training of the motion recognition model comprising:

determining a confidence level of the movement signal by evaluating the movement signal, and

in response to determining the confidence level being greater than or equal to a confidence level threshold, training the motion recognition model in real time by using the motion signal as sample data.

16 . The motion monitoring method of claim 1 , wherein the segmenting, based on feature information corresponding to the electromyographic signal and feature information corresponding to the attitude signal, comprises:

pre-processing the electromyographic signal in a frequency domain or a time domain, according to operations including:

determining a singularity in the electromyographic signal, wherein the singularity corresponds to an abrupt signal of the electromyographic signal; and

performing the signal correction processing on the singularity in the electromyographic signal;

wherein the performing the signal correction processing on the singularity in the electromyographic signal includes removing the singularity or performing the signal correction processing on the singularity according to a signal around the singularity includes:

correcting the singularity according to a signal around the singularity;

obtaining, based on the pre-processed electromyographic signal, the feature information corresponding to the electromyographic signal; and

segmenting, based on feature information corresponding to the electromyographic signal and feature information corresponding to the attitude signal, the movement signal.

17 . The motion monitoring method of claim 16 , wherein the singularity includes a burr signal, the determining the singularity in the electromyographic signal includes:

selecting, based on the time domain window of the electromyographic signal, different time windows from the time domain window of the electromyographic signal, wherein the different time windows respectively cover different time ranges; and

determining, based on the feature information corresponding to the electromyographic signal in the different time windows, the burr signal;

wherein the determining, based on the feature information corresponding to the electromyographic signal in the different time windows, the burr signal, includes:

determining first amplitude information corresponding to the electromyographic signal within a first time window and second amplitude information corresponding to the electromyographic signal within a second time window, wherein the first time window and the second time window are two adjacent time windows;

determining whether a ratio of the second amplitude information to the first amplitude information is greater than a threshold:

in response to determining that the ratio of the second amplitude information to the first amplitude information is greater than the threshold, performing a signal correction processing on the electromyographic signal within the second time window;

in response to determining that the ratio of the second amplitude information to the first amplitude information is not greater than the threshold, retaining the electromyographic signal within the second time window.

18 . The motion monitoring method of claim 17 , further comprising determining, based on the attitude signal, the feature information corresponding to the attitude signal, wherein the attitude signal comprises coordinate information in at least one original coordinate system; and

determining, based on the attitude signal, the feature information corresponding to the attitude signal comprises:

obtaining a target coordinate system and a conversion relationship between the target coordinate system and the at least one original coordinate system;

converting, based on the conversion relationship, the coordinate information in the at least one original coordinate system to coordinate information in the target coordinate system; and

determining, based on the coordinate information in the target coordinate system, the feature information corresponding to the attitude signal;

wherein the target coordinate system changes as an orientation of the user changes;

the attitude signal includes coordinate information generated by at least two sensors, the at least two sensors are located at different motion parts of the user and correspond to different original coordinate systems, the determining, based on the attitude signal, the feature information corresponding to the attitude signal includes:

determining feature information corresponding to each of the at least two sensors based on the conversion relationship between different original coordinate systems and the target coordinate system; and

determining, based on the feature information respectively corresponding to the at least two sensors, a relative motion between the motion parts of the user.

19 . The motion monitoring method of claim 18 , wherein the conversion relationship between the at least one original coordinate system and the target coordinate system is obtained by a calibration process including:

constructing a specific coordinate system, the specific coordinate system being related to an orientation of the user during the calibration process;

obtaining first coordinate information of the at least one original coordinate system when the user is in a first pose;

obtaining second coordinate information of the at least one original coordinate system when the user is in a second pose; and

determining the conversion relationship between the at least one original coordinate system and the specific coordinate system according to the first coordinate information, the second coordinate information, and the specific coordinate system;

wherein the calibration process further includes:

obtaining a conversion relationship between the specific coordinate system and the target coordinate system; and

determining, according to the conversion relationship between the at least one original coordinate system and the specific coordinate system and the conversion relationship between the specific coordinate system and target coordinate system, the conversion relationship between the at least one original coordinate system and the target coordinate system.

20 . A motion monitoring and feedback method, comprising:

obtaining movement signal of a user during motion, wherein the movement signal includes an electromyographic signal and an attitude signal;

segmenting, based on feature information corresponding to the electromyographic signal, the electromyographic signal to obtain electromyographic signal segments;

segmenting, based on feature information corresponding to the attitude signal, the attitude signal to obtain attitude signal segments, comprising:

determining, based on a time domain window of the attitude signal, at least one target feature point from the time domain window according to a preset condition, wherein the preset condition includes one or more of a change in an angular velocity direction corresponding to the attitude signal; an angular velocity corresponding to the attitude signal being greater than or equal to an angular velocity threshold; a changed value of an angular velocity value corresponding to the attitude signal being an extreme value; and an angle corresponding to the attitude signal reaching an angular threshold; and

segmenting, based on the at least one target feature point, the attitude signal;

determining, based on the electromyographic signal segments and the attitude signal segment through a movement recognition model, a movement quality, wherein the movement quality includes a standard movement or a wrong movement, the movement recognition model is a trained machine learning model, a training of the motion recognition model comprising:

determining a confidence level of the movement signal by evaluating the movement signal, and

in response to determining the confidence level being greater than or equal to a confidence level threshold, training the motion recognition model in real time by using the motion signal as sample data;

in response to determining the movement quality being the wrong movement, controlling a electromyography sensor to generate an electrical stimulation signal to prompt the user to make a motion adjustment.

Assignments (2)

Continuity (2)

References Cited (80)