CN116959317A - A human body model driving system - Google Patents

Abstract

The invention discloses a human body model driving system. The system includes a wearable inertial measurement unit, a data transfer router, a data receiving and model display host computer. The present invention realizes a system that uses a wearable inertial measurement unit to drive a human body model in real time. The wearable inertial measurement unit collects and uploads data, forwards data through a data transfer router, and uses a host computer to perform data reception, time alignment, and attitude analysis. calculation, coordinate axis conversion, human body model driving and display, and finally realize a system that can drive the host computer human body model in real time from external human body movements. The invention realizes a system that uses a wearable inertial measurement unit to drive the human body model in real time, solves the problems of inconsistent multi-node communication standards, host computer data reception, attitude calculation, and coordinate axis correspondence. At the same time, it also realizes various functions during use. The inertial measurement units are independent of each other and connected without cables.

Description

A human body model driving system

Technical field

The embodiment of the present invention relates to the field of motion measurement, and mainly proposes a human body model driving system.

Background technique

The performance of my country's sports competitions has made remarkable progress in the past ten years. At present, the level of Chinese athletes is not far behind that of European and American powers, and some high-level athletes are only a few centimeters away from the world's top levels. Both professional athletes and sports enthusiasts hope to improve their sports level by observing their own shortcomings. Therefore, how to measure the movements during exercise and conduct inversion is crucial.

Driven by the above needs, a lot of technical research and research work has been carried out on sports measurement needs and key technologies. This system has been designed for motion inversion. The main purpose is to intuitively display posture during exercise and facilitate athletes to locate their own shortcomings.

Contents of the invention

Based on the above requirements, the present invention designs a real-time driven human body model driving system.

The technical solution of the present invention is: a human body model driving system, including a wearable inertial measurement unit, a data transfer router, a data receiving and model display host computer;

The wearable inertial measurement unit is used to measure the acceleration and angular velocity information of the preset position of the human body, and ensure that the measured data is transmitted to the data transfer router in real time;

A wireless link is established between the data transfer router, the wearable inertial measurement unit, the data receiver and the model display host computer through WIFI signals. Through the wireless link, the received acceleration and angular velocity information is uploaded to the data receiving and model in real time. Display the host computer and transfer the instructions issued by the data receiver model display host computer to the wearable inertial measurement unit;

The data reception and model display host computer inverts human body motion from random, noisy original motion signals.

Preferably, the wearable inertial measurement unit consists of 9 IMU sensors, which are respectively worn on the left thigh, right thigh, left calf, right calf, left upper arm, left lower arm, right upper arm, right lower arm and waist of the human body. It is used to collect, save and upload the acceleration and angular velocity parameters of the sensitive limbs and waist in real time.

Preferably, the IMU sensor has a built-in data storage module and a repeatedly rechargeable polymer physical battery.

Preferably, after the IMU sensor receives the "real-time test start command", it starts measuring and replies in real time with the "real-time test data frame"; after receiving the "real-time test end command", the IMU sensor replies with the "real-time test end reply frame" and then keeps WIFI is on; if the "real-time test end command" is not received but the data storage module is full at this time and the real-time test cannot continue, the IMU sensor will actively send a "real-time test" according to the preset protocol after sending the last frame of data. "Test end reply frame", informing the data receiving and model display host computer IMU sensor that the real-time test has stopped, and then exits the real-time test mode.

Preferably, the data receiving and model display host computer includes a data receiving module and a data analysis module;

The data receiving module stores the received IMU sensor data in a local file. After the storage is completed, the file path is inserted into the task queue;

The data analysis module pulls data from the task queue, analyzes the corresponding files and stores the calculation results in the database;

The analysis is responsible for parsing the received data and determining whether the frame header checksum of each frame of data is correct and the frame sequence numbers are continuous without missing frames. If it is found that the frame header checksum is incorrect and the frame sequence numbers are discontinuous, If there is a problem, a "data is incomplete" prompt will be given. After ensuring that there are no problems, the data of each sensor will be aligned in time, and the motion status will be determined based on the data for attitude calculation.

Preferably, the attitude calculation uses the SINS strapdown inertial navigation update algorithm to update the attitude, speed and position information of the human body during movement or walking; the speed error and heading angle error in the zero speed interval are used as observation quantities to establish the Karl Mann filter is used to estimate the velocity error, position error and attitude angle error, and then compensate the estimated errors to the corresponding variables to obtain an estimate close to the true value of the state variable.

Preferably, in the swing phase interval, the Kalman filter only updates time and does not update observations; when zero speed is detected, that is, in the support phase interval, the Kalman filter updates time and measurements.

Preferably, the data analysis module determines that the human body is in a stationary state based on multiple IMU sensor data, then performs initial calibration, calculates the attitude angle of each IMU sensor through acceleration information and magnetic field strength information, and then obtains the initial attitude from the carrier coordinate system to the navigation coordinate system. The matrix solves the problem of human body wearing and installation and ensures the correspondence between the human body model and the external human body.

Preferably, the data receiving and model display host computer uses the global coordinate system to drive all joints, achieving decoupling between each joint. The specific implementation method is as follows:

The turning point during walking is detected based on the angular velocity in the reference coordinate system obtained in the attitude calculation, and the data is segmented according to the turning point. The movement direction of each segment of data is a straight line;

Calculate the rotation matrix from the reference coordinate system to the global coordinate system for each piece of data, and then transform the data into a unified global coordinate system.

Preferably, the present invention realizes the foot displacement driving of the human body model with the back as the core, and the data accuracy is high. The specific implementation method is as follows: the data analysis module uses the constraints of the back and the length of the lower limbs to correct the estimated position of the calf sensor, And the position of the thigh sensor is inferred from the position of the calf sensor, and the position and rotation angle data are stored in the database.

Compared with the prior art, the beneficial effects of the present invention are:

Through wireless WIFI communication, the TCP protocol is used to send online check-in information to the data transfer router, thereby establishing a stable and reliable wireless data transmission link with it. All instructions are issued and measurement data are uploaded through wireless WIFI, which greatly reduces the complexity of device connection, wearing, and information transmission in the gait measurement system that uses traditional wired communication methods, and greatly improves the system's efficiency. Application convenience.

Analyze the received data to determine whether the frame header checksum of each frame of data is correct and the frame sequence number is continuous without frame loss. If it is found that there is an error in the frame header checksum and the frame sequence number is discontinuous, a " "Incomplete data" prompt, ensure that there are no problems before aligning the data time of each sensor data.

Detailed ways

The present invention will be described in detail below with reference to the accompanying drawings.

The invention finally realizes a system that can drive the host computer human body model in real time by external human body movements. The invention realizes a system that uses a wearable inertial measurement unit to drive the human body model in real time, solves the problems of inconsistent multi-node communication standards, host computer data reception, attitude calculation, and coordinate axis correspondence. At the same time, it also realizes various functions during use. The inertial measurement units are independent of each other and connected without cables.

The working process of this system is shown in Figure 1, which is a human body model driving system, a wearable inertial measurement unit, a data transfer router, a data receiving and model display host computer (referred to as the host computer).

The wearable inertial measurement unit consists of 9 IMU sensors, which are respectively worn on the left thigh, right thigh, left calf, right calf, left upper arm, left lower arm, right upper arm, right lower arm and waist of the human body, and are used to measure sensitive to The acceleration and angular velocity parameters of the limbs and waist are used for data collection, data saving, data uploading and other functions to ensure that the data is transmitted to the data transfer router in real time.

Multiple IMU sensors form a wearable sensor data collection module. The accelerometer and gyroscope in the sensor are used to measure the acceleration and angular velocity information of the corresponding position, and collect movement data of different wearing parts of the lower limbs. The wearable sensor uses wireless sensing technology, has a built-in data storage module and a rechargeable polymer lithium battery. It does not require any cables when used, puts almost no burden on the wearer, and is simple to wear and easy to operate.

The data transfer router will receive the acceleration and angular velocity information of the left thigh, right thigh, left calf, right calf, left upper arm, left lower arm, right upper arm, right lower arm and waist through WIFI signals to establish a wireless link for data collection. Data upload and other functions can be uploaded to the data receiving and model display host computer in real time.

Wireless data is transmitted in real time, each sensor is independent of each other, and data can be stored at the same time to ensure data integrity. The specific implementation method is as follows:

Send online check-in information to the data transfer router through wireless WIFI communication, thereby establishing a stable and reliable wireless data transmission link with it. All instructions are issued and measurement data are uploaded through wireless WIFI, which greatly reduces the complexity of device connection, wearing, and information transmission in the gait measurement system that uses traditional wired communication methods, and greatly improves the system's efficiency. Application convenience.

Data reception and model display The host computer is responsible for providing data reception, data analysis, data time alignment, attitude calculation, initial deviation compensation, coordinate axis conversion, and model-driven display functions, that is, from the random, noisy original motion signal, the Act out human movement.

The data receiving and model display host computer includes a data receiving module and a data analysis module;

The data receiving module receives IMU sensor data and stores it in a local file. After the storage is completed, the module inserts the file path into the task queue. The data analysis module pulls data from the task queue, parses the corresponding files and stores the calculation results in the database. When the number of tasks increases, we can configure multiple data analysis modules to process tasks in parallel according to the number of server cores.

The data receiving module can receive and process data from multiple devices at the same time, identify data correctness, perform data analysis, and packet forwarding. The specific implementation method is as follows:

After the wearable sensor (i.e. IMU sensor) receives the "real-time test start command", it starts measuring and responds to the "real-time test data frame" in real time. After receiving the "real-time test end command", the wearable sensor replies with the "real-time test end reply frame" and then keeps the WIFI on. If the host computer does not send a "real-time test end command", but the Flash is full at this time and the real-time test cannot continue, the wearable sensor will actively send a "real-time test end reply frame" according to the protocol after sending the last frame of data. , inform the host computer that the wearable sensor has stopped real-time testing, and then exit the real-time testing mode.

Human motion parameters are obtained by fusing and interpreting the measurement data of different IMU sensors at the same time. Therefore, inconsistent time bases will directly lead to inaccurate measurement of gait parameters, or even unavailability of measurement data. Therefore, wireless signals are transmitted via WiFi and use the TCP protocol to ensure that each wearable sensor receives time synchronization instructions at the same time, and the time base between each wearable sensor is consistent.

The data analysis module can analyze and calculate multiple packets of data at the same time and perform time alignment to ensure the spatiotemporal correctness of the data. The specific implementation method is as follows:

If it is found that there are problems such as incorrect frame header checksums and discontinuous frame serial numbers, a prompt "Incomplete data reception (specific number of lost frames)" will pop up after all reception is completed, and the user can choose whether to manually upload it again. If the user clicks the button to send the "stored data upload command", the host computer will re-receive the real-time test data and overwrite the original data with this data. The uploaded data also includes the first frame and data frame; if real-time data is not received for a continuous period of time, the wearable sensor is considered to be full and the real-time test is exited. At this point, the training process ends and the data integrity is verified as above. It is up to the user to choose whether to manually upload it again.

The host computer completes the attitude calculation function through human body status recognition and wearing and installation error calibration.

Furthermore, the host computer system will accurately determine whether the human body is in a stationary state, and then through initial moment calibration, it solves the problem of human body wearing and installation and ensures the correspondence between the human body model and the external human body.

Furthermore, when the human body is moving, the human body model driving system performs a targeted coordinate system conversion for the problem that the human body model coordinate system does not correspond to the external sensor coordinate system, and realizes the use of geographical system angles to drive the human body model.

The attitude calculation process includes a state recognition program, which can realize the recognition of human body state for posture correction, and solves the problem of cumulative error in inertial human body posture calculation. The specific implementation method is as follows:

The strapdown inertial navigation system (SINS) selects the "east-north-sky" geographical coordinate system as the navigation coordinate system to implement the recursive update algorithm of strapdown inertial navigation. Using the speed error and heading angle error in the zero speed interval as observation quantities, a Kalman filter is established to estimate the speed error, position error and attitude angle error of the system, and then compensate the estimated errors into the corresponding variables. Get an estimate close to the true value of the state variable. The algorithm flow is shown in Figure 2. The algorithm part mainly uses the strapdown inertial navigation system navigation algorithm, zero speed detection algorithm, and Kalman filter and zero speed correction algorithm. The following is the implementation of these algorithms.

·SINS navigation algorithm

The strapdown inertial navigation system (SINS) selects the "east-north-sky" geographical coordinate system as the navigation coordinate system to implement the recursive update algorithm of strapdown inertial navigation. The strapdown inertial navigation update algorithm is divided into three parts: attitude, speed and position update. The attitude update algorithm is the core.

From the angular velocity equation we get

Since the MEMS sensor has low accuracy and cannot be sensitive to the earth's rotation angular velocity, it is ignored In general sports scenes or walking scenes, the speed of people is less than 10m/s, the radius of the earth R=6371393m, and/> So/> It is on the order of 10 ^-7 ~ 10 ^-6 and can also be ignored. Therefore, for MEMS sensors, the above equation can be equivalent to:

and Q are the attitude matrix and attitude quaternion. The initial value is calculated from the initial attitude angle θ ₀ , γ ₀ , ψ ₀ obtained from the initial alignment, and then calculated from the continuously updated quaternion. Confirmed/> Finally, the quaternion update equation is as follows:

Calculated by Q The formula is as follows

after updated by Three attitude angles can be calculated.

The specific force equation is

in Calculated from formula (1),/> is the accelerometer measurement value,/> and/> can be ignored, g is the acceleration due to gravity, so it can be calculated/> That is, the acceleration of the human body relative to the earth, and then get the speed update function:

Then we get the position update equation:

In summary, the posture, speed and position information of the human body during movement or walking can be obtained.

·Zero speed detection algorithm

The low accuracy of MEMS inertial sensors is the main error factor affecting the system navigation accuracy. When used for a long time, navigation errors will accumulate over time and seriously affect the accuracy of the final measurement results. Through different zero-speed detection algorithms, the stationary interval of the human body in motion is detected, and then parameter correction is performed in the zero-speed interval, which can effectively eliminate speed errors and constrain position and heading errors.

During walking, as the foot lifts, steps, lands, and stops, IMU sensors worn on different parts of the human body can also be sensitive to the periodic changes in the corresponding parts. Periodic zero-velocity intervals in different parts of the human body can be detected through different algorithms. According to the motion data characteristics of different parts of the human body during exercise, the foot zero-speed detection algorithm uses a generalized likelihood ratio detection algorithm. Through the zero-speed detection algorithm, the zero-speed interval of the corresponding part can be effectively detected.

·Kalman filter and zero speed correction algorithm

The principle of Kalman filter is to use the speed error and heading angle error in the zero speed interval as observation quantities to establish a Kalman filter to estimate the speed error, position error and attitude angle error of the system, and then compensate the estimated errors into the corresponding variables to obtain an estimate close to the true value of the state variable. The state variables of the Kalman filter include speed error, position error and attitude error. Therefore, it is necessary to establish an appropriate state equation based on the error equation of inertial navigation, MEMS sensor characteristics and human motion characteristics.

error equation

(a) Attitude error equation

Take the "Northeast Sky" coordinate system as the navigation coordinate system n. In actual navigation calculations, the navigation coordinate system given by the navigation computer simulation digital platform is recorded as n', and its error angle is called the misalignment angle.

The attitude differential equation is:

consider And/> and/> can be ignored, so the simplified attitude differential equation is:

After considering the error, the attitude differential equation is:

After sorting and converting, the MEMS attitude error equation is obtained as follows:

(b) Speed error equation

Considering the MEMS sensor accuracy, and/> is not considered, so the simplified velocity differential equation is:

The differential equation of velocity after taking the error into account is:

After sorting out the conversion, the MEMS speed error equation is obtained as follows:

(c) Position error equation

The position differential equation is:

The differential equation after considering the error is

in:

so

After sorting and converting, the MEMS position error equation is obtained as follows:

Correction Algorithm and Quantity Measurement

(a)ZUPT

When motion is detected to be in the stationary phase, its true speed should theoretically be zero. However, due to the large measurement error of MEMS sensors, the speed calculated by MEMS inertial navigation is not actually zero. The ZUPT method treats the speed calculated by the MEMS inertial navigation during the stationary phase as the speed error, and uses this speed error as a quantity measurement for Kalman filter estimation to achieve the purpose of suppressing navigation parameter errors.

Therefore, the quantity based on the ZUPT algorithm is measured as ΔV, and

(b)ZIHR

In the stationary phase, theoretically the heading angle at the two moments before and after will not change. Also due to the large measurement error of the MEMS sensor, the heading angle difference calculated at the two moments before and after is not zero. Therefore, the heading angle difference between the two moments before and after the zero speed interval can be measured as a quantity to suppress the heading error angle.

Therefore, the quantity measurement based on the ZIHR algorithm is and

Kalman filter

(a) Equation of state

Combining the attitude error equation, velocity error equation and position error equation, the state equation expression can be obtained as

in:

The state vector is

The one-step transfer matrix is

The system noise matrix is

W ₁ = [w _gx w _gy w _gz w _ax w _ay w _az ] ^T

The system noise distribution matrix is

(b) Measurement equation

Combining ZUPT and ZIHR, the measurement equation expression can be obtained as

Among them, the quantity measurement is

The measurement matrix is

H ₂₄ = [0 secγsinθΔt secγcosθΔt]

The measurement noise matrix is

(c)Filtering algorithm

According to the Kalman filter algorithm, the continuous equation is discretized and brought into the following formula:

(a) One-step prediction of state

(b) State one-step prediction mean square error matrix

(c) Filter gain

(d)State estimation

(e)State estimation mean square error matrix

P _k =(IK _k H _k )P _k/k-1

Since there is zero speed measurement only in the zero speed interval, in the swing phase interval, the Kalman filter only performs time updates and does not perform measurement updates. When zero speed is detected, that is, in the support phase interval, the filter performs time updates. Updates and measurement updates.

After completing the attitude calculation, the turning point during walking is detected based on the angular velocity in the reference coordinate system, and the data is segmented according to the turning point. The movement direction of each segment of data is a straight line. Calculate the rotation matrix from the reference coordinate system to the global coordinate system for each piece of data. On this basis, convert the data to a unified global coordinate system. Use zero-speed correction and acceleration quadratic integration to calculate the displacement of each sensor. Through the displacement data Drive the human body model that has been modeled in the unity software to complete the action display.

Furthermore, when the human body model driving system was demonstrated, it realized the foot displacement driving of the human body model with the back as the core, and the data accuracy was high. The specific implementation method is as follows:

The gait tracking thread will use the back constraints and lower limb length to correct the estimated position of the calf sensor, infer the position of the thigh sensor from the position of the calf sensor, and store the position and rotation angle data in the database.

The parts not described in detail in the present invention belong to the common knowledge of those skilled in the art.

Claims (10)

1. A human body model driving system, characterized by: including a wearable inertial measurement unit, a data transfer router, a data receiving and model display host computer; The wearable inertial measurement unit is used to measure the acceleration and angular velocity information of the preset position of the human body, and ensure that the measured data is transmitted to the data transfer router in real time; A wireless link is established between the data transfer router, the wearable inertial measurement unit, the data receiver and the model display host computer through WIFI signals. Through the wireless link, the received acceleration and angular velocity information is uploaded to the data receiving and model in real time. Display the host computer and transfer the instructions issued by the data receiver model display host computer to the wearable inertial measurement unit; The data reception and model display host computer inverts human body motion from random, noisy original motion signals. 2. The system according to claim 1, characterized in that: the wearable inertial measurement unit is composed of 9 IMU sensors, which are respectively worn on the left thigh, right thigh, left calf, right calf, left upper arm, and left lower arm of the human body. The arms, right upper arm, right lower arm and waist are used to collect, save and upload data in real time to the acceleration and angular velocity parameters of the sensitive limbs and waist. 3. The system according to claim 1, characterized in that the IMU sensor has a built-in data storage module and a repeatedly rechargeable polymer physical battery. 4. The system according to claim 3, characterized in that: after the IMU sensor receives the "real-time test start command", it starts measuring and replies in real time the "real-time test data frame"; after the IMU sensor receives the "real-time test end command" , reply "Real-time test end reply frame", and then keep the WIFI on; if the "Real-time test end command" is not received but the data storage module is full at this time and the real-time test cannot continue, the IMU sensor will send the last frame after sending After receiving the data, it actively sends a "real-time test end reply frame" according to the preset protocol to inform the data receiving and model display host computer that the IMU sensor has stopped real-time testing, and then exits the real-time test mode. 5. The system according to claim 1, characterized in that: the data receiving and model display host computer includes a data receiving module and a data analysis module; The data receiving module stores the received IMU sensor data in a local file. After the storage is completed, the file path is inserted into the task queue; The data analysis module pulls data from the task queue, analyzes the corresponding files and stores the calculation results in the database; The analysis is responsible for parsing the received data and determining whether the frame header checksum of each frame of data is correct and the frame sequence numbers are continuous without missing frames. If it is found that the frame header checksum is incorrect and the frame sequence numbers are discontinuous, If there is a problem, a "data is incomplete" prompt will be given. After ensuring that there are no problems, the data of each sensor will be aligned in time, and the motion status will be determined based on the data for attitude calculation. 6. The system according to claim 5, characterized in that: the attitude calculation adopts the SINS strapdown inertial navigation update algorithm to update the attitude, speed and position information of the human body during movement or walking; using the zero speed interval The speed error and heading angle error are used as observed quantities. A Kalman filter is established to estimate the speed error, position error and attitude angle error. Then the estimated errors are compensated into the corresponding variables to obtain a state variable close to the true value. estimate. 7. The system according to claim 6, characterized in that: in the swing phase interval, the Kalman filter only performs time updates and does not perform observation quantity updates; when zero speed is detected, that is, in the support phase interval, the Kalman filter The device performs time updates and measurement updates. 8. The system according to claim 5, characterized in that: the data analysis module determines that the human body is in a stationary state based on multiple IMU sensor data, then performs initial calibration, and calculates the attitude angle of each IMU sensor through acceleration information and magnetic field strength information. Then the initial posture matrix from the carrier coordinate system to the navigation coordinate system is obtained, which solves the problem of human body wearing and installation and ensures the correspondence between the human body model and the external human body. 9. The system according to claim 5, characterized in that: the data receiving and model display host computer uses the global coordinate system to drive all joints, realizing decoupling between each joint. The specific implementation method is as follows: The turning point during walking is detected based on the angular velocity in the reference coordinate system obtained in the attitude calculation, and the data is segmented according to the turning point. The movement direction of each segment of data is a straight line; Calculate the rotation matrix from the reference coordinate system to the global coordinate system for each piece of data, and then transform the data into a unified global coordinate system. 10. The system according to claim 5, characterized in that: it realizes the foot displacement driving of the human body model with the back as the core, and the data accuracy is high. The specific implementation method is as follows: The data analysis module uses the back constraints and lower limb length to correct the estimated position of the calf sensor, infers the position of the thigh sensor from the position of the calf sensor, and stores the position and rotation angle data into the database.