Health monitoring method based on robot dog
By integrating accelerometers and gyroscopes on the robot dog, collecting user data and calculating characteristic indicators, the problem of poor monitoring of existing wearable devices in complex environments is solved, and accurate monitoring of user motion status and abnormal warning are achieved.
Patent Information
- Application Number
- CN202510656211.9
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-05-21
- Publication Date
- 2025-06-27
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
Existing wearable devices have problems such as poor user experience and limited monitoring effects in health monitoring, especially in complex environments, which leads to misjudgment.
Using a health monitoring method based on robot dogs, users' acceleration and angular velocity data are collected through robot dogs' accelerometers and gyroscopes, dynamic coefficients and angular velocity characteristics are calculated, rotation degree and motion degree are generated, and alarms are made when these data exceed the preset threshold.
It realizes accurate analysis and evaluation of user's movement status, can improve the accuracy and sensitivity of monitoring in complex environments, promptly conduct fall alarms, and provide real-time health status monitoring and abnormal warnings.
Smart Images
Figure CN120203569A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of data processing, and particularly relates to a health monitoring method based on a robotic dog. Background Art
[0002] Existing wearable devices such as smart bracelets and smart watches, although to a certain extent, have realized the monitoring of users' daily activities and some health indicators, still have some limitations. For example, these devices usually need to be worn by users on specific parts such as the wrist for a long time. For user groups with mobility difficulties, cognitive impairments, or those who have a resistance to wearing devices, the user experience and monitoring effects are greatly affected. At the same time, the development of robot technology, especially quadruped robot (robotic dog) technology, has made remarkable progress. Robotic dogs have high mobility, flexibility, and environmental adaptability, and can move freely and perform tasks in various complex terrains and scenarios. The various sensors carried by them, such as accelerometers, gyroscopes, and cameras, can real-time obtain the surrounding environment information and their own motion state data. These characteristics make robotic dogs have great application potential in the field of health monitoring. However, complex backgrounds or dynamically changing environments may affect the monitoring effect of robotic dogs, resulting in misjudgments. For example, the optical flow method is sensitive to background changes and may generate noise; the frame difference method is difficult to detect small-scale movements. In real life, the environments where users are located are often complex and changeable, and robotic dogs are difficult to adapt to these changes, thus affecting the accuracy of monitoring. Summary of the Invention
[0003] Aiming at the problems existing in the current use of wearable devices for health monitoring, the present invention integrates the advantages of robotic dogs and proposes a health monitoring method based on a robotic dog, and the method includes:
[0004] Step S1, using the accelerometer of the robotic dog to collect the acceleration data of the user in a preset time period, and using the gyroscope of the robotic dog to collect the angular velocity data of the user in the preset time period;
[0005] Step S2, according to the angular velocity data of the user in the preset time period, determine the dynamic coefficient and angular velocity characteristics, and generate the rotation degree;
[0006] Step S3, generate the movement degree according to the acceleration data of the user in the preset time period;
[0007] Step S4, when the rotation degree exceeds the rotation threshold or the movement degree exceeds the movement threshold, an alarm is given.
[0008] Further, step S2 includes the following sub-steps:
[0009] Step S21, calculate the segment coefficient of the user at the i-th moment according to the angular velocity of the user at the i-th moment in the preset time period;
[0010] Step S22: Subtract the segment coefficient of the user at time i from 1 to obtain the dynamic coefficient of the user at time i.
[0011] Step S23: Construct the angular velocity feature of the user in a preset time period.
[0012] Step S24: Obtain the rotation degree according to the dynamic coefficient of the user at time i and the angular velocity feature in the preset time period.
[0013] In the present invention, by calculating the segment coefficient of the user at each moment (time i), the local detailed features of the user's movement can be captured, avoiding the feature loss caused by a single global parameter. Similar to "video frame analysis", the segment coefficient at each moment is equivalent to a single-frame feature, and the dynamic coefficient integrates the segment changes, improving the parsing ability for complex movements. Through feature construction, the system can distinguish normal movements (such as walking) from abnormal movements (such as falling, convulsions, etc.), improving the sensitivity to abnormal behaviors.
[0014] Further, in step S21, the segment coefficient K of the user at time i i is calculated as follows: , where p i-1 represents the angular velocity value of the user at time i - 1, p i+1 represents the angular velocity value of the user at time i + 1, p i represents the angular velocity value of the user at time i, represents the average angular velocity up to the j-th segment after dividing the preset time period into several segments, and J represents the number of segments divided.
[0015] Further, step S23 includes the following sub-steps:
[0016] Step S231: Convert the angular velocity of the user at time i into an angular velocity vector.
[0017] Step S232: Process the angular velocity vectors at all times to obtain a feature representation sequence.
[0018] Step S233: Generate a random number, and use the random number and the feature representation sequence to determine the angular velocity feature.
[0019] In the present invention, converting the angular velocity from a scalar form to a vector form, generating a feature representation sequence by processing the angular velocity vectors at all times, can capture the time dynamic characteristics of the user's movement. The feature representation sequence can compress the high-dimensional vector data into low-dimensional features through a dimensionality reduction algorithm, reducing the computational complexity. Perturbing or sampling the feature representation sequence with a random number can improve the robustness of the algorithm to noise and outliers, avoiding overfitting.
[0020] Further, in step S232, the angular velocity vectors at all times are processed using a max-pooling operation, and after processing, they are converted into a feature representation sequence. The processing method of the max-pooling operation is as follows: , where G represents the feature representation sequence, w1 represents the angular velocity vector of the user at the first time, w2 represents the angular velocity vector of the user at the second time, and w I represents the angular velocity vector of the user at the I-th time.
[0021] Further, in step S233, the calculation method of the angular velocity feature W is as follows: , where T k represents the k-th value of the feature representation sequence, ε represents a random number, K represents the number of the feature representation sequence, and max(·) represents taking the maximum value.
[0022] Further, in step S24, the calculation method of the rotation degree X is as follows: , where W represents the angular velocity feature and A represents the average value of the dynamic coefficients of the user in a preset time period.
[0023] Further, step S3 includes the following sub-steps:
[0024] Step S31, calculate the fusion index of the user at the i-th time according to the acceleration of the user at the i-th time in a preset time period;
[0025] Step S32, take the average value of the fusion indices at all times as the motion degree.
[0026] In the present invention, through the fusion index, the acceleration data of the user in the X, Y, and Z axes can be integrated into a single index. The calculation of the fusion index takes into account the time window, and the threshold is generated using the fusion indices at all times, which is equivalent to using the "global mode" of the user's movement rather than the outliers at a single time, thereby reducing false alarms caused by noise or short-term fluctuations.
[0027] Further, in step S31, the fusion index R of the user at the i-th time i is calculated as follows: , where f xi represents the acceleration in the x direction of the user at the i-th time, f yi represents the acceleration in the y direction of the user at the i-th time, f zi represents the acceleration in the z direction of the user at the i-th time, log2(·) represents the logarithmic function, exp(·) represents the exponential function, max(·) represents taking the maximum value, and min(·) represents taking the minimum value.
[0028] The beneficial technical effects of the present invention are as follows: The present invention makes full use of the sensor advantages of the robotic dog. By collecting the acceleration data and angular velocity data of the user, it accurately analyzes and evaluates the user's motion state. By determining the dynamic coefficient and angular velocity characteristics and generating the rotation degree, and generating the motion degree according to the acceleration data, it can comprehensively and meticulously reflect the user's motion situation and physical activity state. When the present invention detects that the rotation degree exceeds the preset rotation threshold or the motion degree exceeds the preset motion threshold, it promptly gives a fall alarm, thereby realizing real-time monitoring of the user's health status and abnormal early warning, and providing a more effective, convenient and intelligent solution for the user's health management. Brief Description of the Drawings
[0029] In order to more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the following will briefly introduce the drawings required for use in the description of the embodiments or the prior art. Obviously, the drawings in the following description are some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.
[0030] Figure 1 It is a flowchart of a health monitoring method based on a robotic dog provided by an embodiment of the present invention. Detailed Embodiments
[0031] The following will clearly and completely describe the technical solutions in the embodiments of the present invention with reference to the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.
[0032] As Figure 1 shown, the present invention provides a health monitoring method based on a robotic dog, including the following steps:
[0033] S1. Use the accelerometer of the robotic dog to collect the acceleration data of the user in a preset time period, and use the gyroscope of the robotic dog to collect the angular velocity data of the user in a preset time period;
[0034] S2. Determine the dynamic coefficient and angular velocity characteristics according to the angular velocity data of the user in a preset time period, and generate the rotation degree;
[0035] S3. Generate the motion degree according to the acceleration data of the user in a preset time period;
[0036] S4. Give an alarm when the rotation degree exceeds the rotation threshold or the motion degree exceeds the motion threshold.
[0037] In the embodiment of the present invention, S2 includes the following sub-steps:
[0038] S21. Calculate the segment coefficient of the user at the i-th moment within a preset time period according to the angular velocity of the user at the i-th moment;
[0039] S22. Subtract the segment coefficient of the user at the i-th moment from 1 to obtain the dynamic coefficient of the user at the i-th moment;
[0040] S23. Construct an angular velocity feature for the user within a preset time period;
[0041] S24. Obtain the rotation degree according to the dynamic coefficient of the user at the i-th moment and the angular velocity feature within the preset time period.
[0042] In the present invention, by calculating the segment coefficient of the user at each moment (the i-th moment), the local detailed features of the user's movement can be captured, avoiding the feature loss caused by a single global parameter. Similar to "video frame analysis", the segment coefficient at each moment is equivalent to a single-frame feature, and the dynamic coefficient integrates the segment changes, improving the parsing ability for complex movements. Through feature construction, the system can distinguish normal movements (such as walking) from abnormal movements (such as falling, twitching, etc.), and improve the sensitivity to abnormal behaviors.
[0043] In the embodiment of the present invention, in S21, the segment coefficient K of the user at the i-th moment i has the following calculation formula: ; where p i-1 represents the angular velocity value of the user at the (i - 1)-th moment, p i+1 represents the angular velocity value of the user at the (i + 1)-th moment, p i represents the angular velocity value of the user at the i-th moment, represents the average angular velocity up to the j-th segment after dividing the preset time period into several segments, and J represents the number of divided segments.
[0044] In the embodiment of the present invention, S23 includes the following sub-steps:
[0045] S231. Convert the angular velocity of the user at the i-th moment into an angular velocity vector;
[0046] S232. Process the angular velocity vectors at all moments to obtain a feature representation sequence;
[0047] S233. Generate a random number, and use the random number and the feature representation sequence to determine the angular velocity feature.
[0048] In the present invention, the angular velocity is converted from a scalar form to a vector form. By processing the angular velocity vectors at all times, a feature representation sequence is generated, which can capture the temporal dynamic characteristics of the user's movement. The feature representation sequence may compress the high-dimensional vector data into low-dimensional features through a dimensionality reduction algorithm, reducing the computational complexity. By perturbing or sampling the feature representation sequence with random numbers, the robustness of the algorithm to noise and outliers can be improved, avoiding overfitting.
[0049] In an embodiment of the present invention, in S232, after processing the angular velocity vectors at all times using the max pooling operation, it is converted into a feature representation sequence. The processing method of the max pooling operation is as follows: ; where G represents the feature representation sequence, w1 represents the angular velocity vector of the user at the first time, w2 represents the angular velocity vector of the user at the second time, and w I represents the angular velocity vector of the user at the I-th time.
[0050] In an embodiment of the present invention, in S233, the calculation formula of the angular velocity feature W is: ; where T k represents the k-th value of the feature representation sequence, ε represents a random number, K represents the number of the feature representation sequence, and max(·) represents taking the maximum value.
[0051] In an embodiment of the present invention, in S24, the calculation formula of the rotation degree X is: ; where W represents the angular velocity feature, and A represents the mean value of the dynamic coefficients of the user in a preset time period.
[0052] In an embodiment of the present invention, S3 includes the following sub-steps:
[0053] S31. Calculate the fusion index of the user at the i-th time according to the acceleration of the user at the i-th time within a preset time period;
[0054] S32. Take the mean value of the fusion indexes at all times as the degree of movement.
[0055] In the present invention, through the fusion index, the acceleration data of the user in the X, Y, and Z axes may be integrated into a single index. The calculation of the fusion index takes into account the time window, and using the fusion indexes at all times to generate a threshold is equivalent to using the "global mode" of the user's movement, rather than the outliers at a single time, thereby reducing false alarms caused by noise or short-term fluctuations.
[0056] In an embodiment of the present invention, in S31, the fusion index R of the user at the i-th time i is calculated as follows: ; where f xi represents the acceleration in the x direction of the user at the i-th time, and f yiDenote the acceleration of the user in the y direction at time i, f zi Denote the acceleration of the user in the z direction at time i, log2(·) represents the logarithmic function, exp(·) represents the exponential function, max(·) represents taking the maximum value, and min(·) represents taking the minimum value.
[0057] Finally, it should be noted that: the above embodiments are only used to illustrate the technical solutions of the present invention, rather than limiting it; although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that they can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some or all of the technical features; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the scope of the technical solutions of the embodiments of the present invention.
Claims
1. A health monitoring method based on a robot dog, characterized in that: The method comprises: Step S1, using the accelerometer of the robot dog to collect acceleration data of the user in a preset time period, and using the gyroscope of the robot dog to collect angular velocity data of the user in a preset time period; Step S2, determining the dynamic coefficient and angular velocity characteristics according to the angular velocity data of the user in a preset time period, and generating a rotation degree; Step S3, generating a degree of motion according to the acceleration data of the user in a preset time period; Step S4, when the rotation degree exceeds the rotation threshold or the movement degree exceeds the movement threshold, an alarm is issued.
2. The method according to claim 1, characterized in that The step S2 further comprises: Step S21, calculating the fragment coefficient of the user at time i according to the angular velocity of the user at time i within a preset time period; Step S22, subtracting the fragment coefficient of the user at time i from 1 to obtain the dynamic coefficient of the user at time i; Step S23, constructing an angular velocity feature for the user in a preset time period; Step S24, obtaining the rotation degree according to the dynamic coefficient of the user at time i and the angular velocity characteristics in a preset time period.
3. The method according to claim 2, characterized in that In step S21, the user's fragment coefficient K at time i i The calculation method is: , where p i-1 represents the angular velocity value of the user at time i-1, p i+1 represents the angular velocity value of the user at time i+1, p i Indicates the angular velocity value of the user at time i, It represents the average angular velocity up to the jth segment after the preset time period is divided into a number of segments, and J represents the number of divided segments.
4. The method according to claim 2, characterized in that: The step S23 further comprises: Step S231, converting the angular velocity of the user at time i into an angular velocity vector; Step S232, processing the angular velocity vectors at all times to obtain a feature representation sequence; Step S233, generate a random number, and use the random number and the feature representation sequence to determine the angular velocity feature.
5. The method according to claim 4, characterized in that In step S232, the angular velocity vectors at all times are processed by the maximum pooling operation and converted into a feature representation sequence. The processing method of the maximum pooling operation is: , where G represents the feature representation sequence, w1 represents the angular velocity vector of the user at the first moment, w2 represents the angular velocity vector of the user at the second moment, and wI represents the angular velocity vector of the user at the I moment.
6. The method according to claim 4, characterized in that In step S233, the angular velocity characteristic W is calculated as follows: , where T k represents the kth value of the feature representation sequence, ε represents a random number, K represents the number of feature representation sequences, and max(·) represents the maximum value.
7. The method according to claim 2, characterized in that In step S24, the rotation degree X is calculated as follows: , where W represents the angular velocity feature and A represents the average value of the dynamic coefficient of the user in a preset time period.
8. The method according to claim 1, characterized in that The step S3 comprises the following sub-steps: Step S31, calculating the fusion index of the user at time i according to the acceleration of the user at time i within a preset time period; Step S32: taking the average of the fusion indexes at all times as the degree of motion.
9. The method according to claim 8, characterized in that In step S31, the fusion index R of the user at time i i The calculation method is: , where f xi represents the acceleration of the user in the x direction at time i, f yi represents the user's acceleration in the y direction at time i, f zi represents the acceleration of the user in the z direction at time i, log2(·) represents the logarithmic function, exp(·) represents the exponential function, max(·) represents the maximum value, and min(·) represents the minimum value.