A human posture recognition system based on wearable sensors

Abstract

The application discloses a human posture recognition system based on a wearable sensor and belongs to the technical field of sensor measurement. The human posture recognition system collects foot bottom pressure through a pressure sensor module, filters and denoises and performs smooth processing through a signal processing module, and wirelessly transmits sensor signals to an upper computer through a wireless transmission module. The upper computer adopts a hidden Markov algorithm and a random forest fusion algorithm to recognize postures. The system comprises a pair of flexible pressure sensor arrays, the sensor arrays are worn on the foot bottom to collect foot bottom pressure signals, the signal processing module is completed by the upper computer, the wireless transmission module selects a Bluetooth module, and the upper computer performs real-time recognition after training the model. The system improves the recognition rate, is high in adaptability, and is suitable for medical treatment, human-computer interaction, sports training and other fields.

Description

Technical Field

[0001] This invention relates to the field of sensor measurement technology, specifically to a human posture recognition system based on a flexible plantar pressure sensor array. Background Technology

[0002] In today's era of rapid scientific and technological development, human posture recognition technology plays a vital role in numerous fields. In sports science, analyzing athletes' postures can improve training efficiency and athletic performance; in augmented reality (AR) and virtual reality (VR) fields, accurate human posture tracking can provide a more immersive experience; in health and medical systems, posture recognition helps monitor patients' rehabilitation processes and prevent falls; in human-computer interaction, posture recognition can simplify user interfaces and enhance user experience; and in gaming and entertainment systems, posture recognition provides players with a more natural way to interact.

[0003] Despite the numerous advantages of computer vision-based human pose recognition technology, such as non-contact operation and wide range of applications, challenges remain in practical applications. These challenges include sensitivity to environmental conditions such as lighting and occlusion, and hardware dependence on computationally intensive algorithms such as deep learning. Furthermore, traditional pose recognition methods, such as optical sensors and inertial measurement units (IMUs), while offering an alternative solution, also have limitations, such as dependence on lighting and angle, and inconvenience in wearing them.

[0004] Therefore, developing a new human posture recognition technology is particularly important. This technology should be more adaptable, able to work stably in different environments, and also cost-effective, easy to promote and apply.

[0005] The emergence of flexible pressure sensors has brought a new breakthrough to this field. These sensors possess excellent flexibility and wearability, conforming to the curves of the human body to provide continuous and accurate pressure data. From this data, human posture and movement can be accurately inferred. Compared to traditional visual and inertial sensors, flexible pressure sensors are not limited by light or viewing angle and have a significant advantage in wearing comfort. Furthermore, they typically have low power consumption, making them suitable for extended use.

[0006] In conclusion, flexible pressure sensors have enormous application potential in the field of human posture recognition. Through in-depth research and optimization of the performance and integration methods of these sensors, more efficient, reliable, and user-friendly human posture recognition systems can be developed, thereby promoting technological advancements and expanding applications in related fields. Summary of the Invention

[0007] To address the shortcomings of existing technologies, this invention provides a human posture recognition system based on a flexible foot pressure sensor array. This system uses the flexible foot pressure sensor array to capture foot pressure information of different human postures, and processes the signals through an improved hidden Markov algorithm and a random forest fusion algorithm to achieve human posture recognition, thus solving the accuracy and comfort problems existing in the background technology.

[0008] To achieve the above objectives, the present invention specifically adopts the following technical solution:

[0009] A human posture recognition system based on wearable sensors, characterized in that it includes a sensor module, a wireless transmission module and a host computer, wherein the host computer includes a signal processing module and a recognition module;

[0010] The sensor module collects plantar pressure signals from both feet;

[0011] The wireless transmission module wirelessly transmits the plantar pressure signals collected by the two sensors to the subsequent module for further processing.

[0012] The host computer receives data from the wireless transmission module, processes the data to extract features, and finally uses an improved hidden Markov algorithm and a random forest fusion algorithm to identify human posture.

[0013] Furthermore, the sensor module includes a flexible pressure sensor array.

[0014] 2. A flexible plantar pressure sensor array is fixed to the two soles of the human body to detect changes in plantar pressure. The plantar pressure sensor array module uses a 6*4 flexible pressure sensor array patch, which contains 18 sensor array patch units. The flexible array pressure sensor adopts the piezoresistive sensing mechanism. The resistance value of the sensitive unit changes with the contact force. The object being measured is the resistance change of each sensitive point array.

[0015] Furthermore, the wireless transmission module is composed of a Bluetooth module;

[0016] The wireless transmission module uses a Bluetooth transmission module, which is connected to the sensor acquisition module. Each flexible sensor has a Bluetooth transmitter module to wirelessly transmit the collected plantar pressure data to the host computer. The host computer uses a CH340USB to serial port module as the Bluetooth receiver module with a baud rate of 115200.

[0017] Furthermore, the host computer uses an improved hidden Markov algorithm and a random forest fusion algorithm to identify poses;

[0018] The host computer performs data filtering and preprocessing on the received data, PCA dimensionality reduction, and then calculates the trajectory of the center of pressure on the soles of both feet as one feature. At the same time, the data is input into a hidden Markov model to calculate the maximum likelihood of the hidden state as the second feature. Finally, the two features are concatenated and input into a random forest for supervised learning. After obtaining the trained model, the model is loaded for real-time recognition.

[0019] Compared with the prior art, the present invention provides a human posture recognition system based on a flexible foot pressure sensor array, which has the following advantages:

[0020] This invention uses the calculated plantar pressure center and the likelihood numbers of the Hidden Markov Model (HMM) for different postures as features. Spatial dynamics are represented by the Co-op trajectory, which directly reflects the body's balance control strategy on the support surface, including geometric features of pressure distribution (such as trajectory length, ellipse area, forward / backward / left / right displacement) and temporal dynamic features (such as velocity and acceleration). The HMM models the phased changes of movement through implicit state sequences (such as the support phase and swing phase in the gait cycle), and its output likelihood number represents the probability that the current observed sequence (pressure signal) belongs to a certain posture type, implying the temporal dependence of the movement and the state transition rules. The HMM reduces the influence of individual differences (such as stride length) through probabilistic modeling, while the random forest selects universal features based on feature importance; the two work together to enhance the model's adaptability to users who have never been before.

[0021] This invention achieves robustness and efficiency in human pose recognition by fusing COP trajectories with HMM likelihood features and using random forests for classification, while requiring fewer samples compared to deep learning. Attached Figure Description

[0022] Figure 1 This is a schematic diagram of the flexible pressure sensor array structure according to an embodiment of the present invention;

[0023] Figure 2 This is a structural diagram of the sensor module according to an embodiment of the present invention;

[0024] Figure 3 This illustrates the relationship between pressure and resistance in a flexible sensor according to an embodiment of the present invention.

[0025] Figure 4 This is a flowchart illustrating the system workflow of an embodiment of the present invention.

[0026] Figure 5 This is a waveform diagram of data collected from 18 channels on the sole of the foot according to an embodiment of the present invention;

[0027] Figure 6 This is a COP trajectory distribution diagram calculated for an embodiment of the present invention. Detailed Implementation

[0028] The technical solutions of 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.

[0029] Example

[0030] like Figure 1-6 As shown in the figure, an embodiment of the present invention proposes a human posture recognition system based on wearable sensors. The system mainly consists of the following parts: a flexible pressure sensor array, a signal processing module, a wireless transmission module, and a host computer.

[0031] 1. Sensor Module

[0032] (1) Flexible pressure sensor array:

[0033] The flexible pressure sensor array is a highly sensitive sensing system composed of multiple small sensor units designed to be easily worn as insoles in shoes. Each sensor possesses extremely high sensitivity, capable of accurately capturing a series of subtle pressure changes generated by the human body in motion or at rest, and converting them into electrical signals for subsequent signal analysis and processing.

[0034] This sensor array employs an advanced manufacturing process called precision printing, which precisely prints nanoscale force-sensitive materials and conductive silver paste onto a flexible polymer film substrate. After printing, these materials undergo a thermosetting process to ensure they are firmly bonded to the substrate and maintain long-term stability and reliability.

[0035] In terms of physical properties, these flexible pressure sensors exhibit a significant piezoresistive effect. When the sensor is subjected to external pressure, its resistance decreases. This change follows a power function pattern, indicating a non-linear but predictable correlation between resistance and applied pressure. More importantly, the relationship between the reciprocal of resistance (i.e., conductivity) and pressure is nearly linear. This is crucial for the processing and interpretation of sensor signals, as it simplifies the mathematical models and algorithms for inferring the original pressure value from the electrical signal.

[0036] In summary, this flexible pressure sensor array not only provides real-time monitoring of changes in human posture, but also demonstrates a high degree of flexibility and accuracy in design and manufacturing, making it an ideal choice for developing next-generation human-computer interaction interfaces and health monitoring devices.

[0037] 2. Wireless transmission module:

[0038] The wireless transmission module is responsible for transmitting the processed signals from the signal processing module to the receiving end of the host computer. The system uses the BLE02 (i.e., ATK-MW579) Bluetooth module as its wireless transmission module. Its operating voltage is between 3.3V and 5V, it supports Bluetooth 4.2 protocol, operates at a 2.4GHz ISM frequency, and features a master-slave architecture. It supports multiple operating modes, including a low-power mode, data transmission, and one-to-many data broadcasting. The Bluetooth module is connected to the flexible sensors on both feet. The host computer uses a CH340USB-to-serial module as the Bluetooth receiver module, with a baud rate of 115200.

[0039] 4. Host computer:

[0040] The host computer includes data processing and a fusion algorithm of Hidden Markov Model and Random Forest.

[0041] Hidden Markov Models (HMMs) are algorithms that infer hidden sequences from observable sequences. This algorithm can handle the prediction of nonlinear time series, calculating the state probability at the next time step from the transition probability matrix and initial probability matrix of each observable sequence. The dynamic time warping algorithm module is used for time-series matching and recognition of the feature information output by the signal processing module. HMMs are a powerful time-series signal processing technique; the pressure signal from the sole of the foot is also a time-series signal. By treating pressure as an observable sequence and posture as a hidden sequence, prediction of different human postures can be achieved.

[0042] The purpose of fusing the random forest algorithm is to reduce the uncertainty of the output probability of the HMM algorithm. The output of the HMM is used as a feature input into the random forest model, and the human pose can be obtained after decision-making.

[0043] The Hidden Markov Model (HMM) and Random Forest (RBS) algorithms are fused together. First, the original data is smoothed and cleaned, and the COP (CoP) pressure trajectory distribution of plantar pressure is calculated as a feature. Second, the processed data undergoes PCA (Polygonometric Analysis) for dimensionality reduction, and then the data is input into the HMM. Finally, the optimal output sequence is used as a feature input into the RBS for final classification. This approach compensates for the instability of HMM in pose prediction, improving both the recognition rate and speed.

[0044] Once the host computer has trained the model, the trained model can be used directly by inputting data. The test set and training set are divided in advance to output the human posture recognition.

[0045] A wearable sensor-based human posture recognition system measures the pressure changes on the soles of a subject's feet at a given moment using a flexible pressure sensor array. It then uses a fusion algorithm of Hidden Markov Model and Random Forest to identify and output the subject's movements. The system workflow is as follows: Figure 4 As shown.

[0046] Since each unit of the flexible pressure sensor array outputs the pressure change of each unit in the form of an electrical signal, the electrical signal will change to different degrees during one acquisition cycle. Since the acquisition is carried out using an array, the sensor array of one foot will collect different data changes of 18 units. Different actions cause the pressure of each unit on the sole of the foot to appear in a certain order. Therefore, this invention proposes to use a method to calculate the distribution of the center of plantar pressure to help identify the pressure.

[0047] The center of pressure is the point of application of the resultant force of the vertical reaction forces (pressures) at all contact points when an object comes into contact with a supporting surface (such as the ground or platform). It characterizes the spatial concentration of pressure distribution and is a core parameter in biomechanics, robotics, and engineering for analyzing balance, stability, and motion patterns.

[0048] The trajectory refers to the path that the CoP moves along the support surface over time, reflecting pressure changes during dynamic processes (such as the migration of plantar pressure during walking).

[0049] The CoP trajectory of this invention is realized through a flexible plantar pressure sensor array, and the steps are as follows:

[0050] 1. Sensor calibration:

[0051] (1) Determine the position coordinates of each sensor ( , ).

[0052] (2) Calibrate the force-electric signal relationship to ensure The measurements are accurate.

[0053] 2. Data Acquisition: High-frequency sampling (typically 100-1000Hz) is used to acquire transient vertical forces from each sensor. (t).

[0054] 3. Real-time calculation: (1) Calculate the resultant force: = ,

[0055] (2) Calculate the weighted position sum: (t)=

[0056] (t)=

[0057] 4. Instantaneous Output CoP: (t)= , (t)=

[0058] Application areas:

[0059] The human posture recognition system of this invention has extremely broad application prospects. In the medical field, this system can monitor changes in a patient's posture, such as gait monitoring and rehabilitation training postures, providing doctors with valuable information and helping them to more comprehensively understand the patient's condition and rehabilitation status. In the rehabilitation field, it can be used to evaluate the effectiveness of a patient's rehabilitation training, provide personalized training suggestions, and improve the effectiveness and quality of rehabilitation training.

[0060] In the field of human-computer interaction, this system can be applied to smart home and smart office scenarios to achieve a more intelligent and convenient human-computer interaction method. For example, by recognizing the user's posture, the smart home system can automatically adjust environmental parameters such as lighting and temperature to provide a more comfortable living experience; smart office equipment can automatically adjust the screen angle and height according to changes in the user's posture to improve work efficiency and comfort.

[0061] In addition, this human posture recognition system has broad application potential in fields such as sports training, virtual reality / augmented reality, and security monitoring, bringing more convenience and security to people's lives and work.

[0062] Finally, it should be noted that the above descriptions are merely preferred embodiments of the present invention and are not intended to limit the present invention. Although the present invention has been described in detail with reference to the foregoing embodiments, those skilled in the art can still modify the technical solutions described in the foregoing embodiments or make equivalent substitutions for some of the technical features. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A human posture recognition system based on wearable sensors, characterized in that: It includes a sensor module, a wireless transmission module, and a host computer, wherein the host computer includes a signal processing module and a recognition module; The sensor module collects plantar pressure signals from both feet; The wireless transmission module wirelessly transmits the plantar pressure signals collected by the two sensors to the subsequent modules for further processing. The host computer receives data from the wireless transmission module, processes the data to extract features, and finally uses a fusion algorithm of Hidden Markov Model and Random Forest to identify human posture. The host computer uses a fusion algorithm of an improved Hidden Markov Model and a random forest algorithm to identify poses; The host computer performs data filtering and preprocessing on the received data, PCA dimensionality reduction, and then calculates the plantar pressure center trajectory of both feet. The plantar pressure center trajectory of both feet is used as the first feature. At the same time, the data is input into a hidden Markov model to calculate the maximum likelihood of the hidden state as the second feature. Finally, the two features are concatenated and input into a random forest for supervised learning. After obtaining the trained model, the model is loaded for real-time recognition.

2. The human posture recognition system based on wearable sensors according to claim 1, characterized in that: The sensor module includes a flexible pressure sensor array.

3. The human posture recognition system based on wearable sensors according to claim 1, characterized in that: A flexible plantar pressure sensor array is fixed to the soles of both feet to detect changes in plantar pressure.

4. The human posture recognition system based on wearable sensors according to claim 3, characterized in that: The plantar pressure sensor array module uses 6 The flexible pressure sensor array patch consists of 4 units, which contain 18 sensor array patch units. The flexible array pressure sensor adopts the piezoresistive sensing mechanism. The resistance value of the sensitive unit changes with the contact force. The object being measured is the resistance change of each sensitive point array.

5. The human posture recognition system based on wearable sensors according to claim 1, characterized in that: The wireless transmission module consists of a Bluetooth module.

6. The human posture recognition system based on wearable sensors according to claim 5, characterized in that: The Bluetooth transmission module connects to the sensor acquisition module. Each foot is equipped with a Bluetooth transmitter module to wirelessly transmit the collected plantar pressure data to the host computer. The host computer uses a CH340USB to serial port module as the Bluetooth receiver module with a baud rate of 115200.

Images