KR102479536B1 - Calibration system and method for measuring joint motion, and wearable joint motion measuring apparatus using the same - Google Patents

Abstract

Calibration system and method for measuring joint motion It relates to a wearable joint motion measuring device using the same, and in detail, a calibration system that generates a calibration model that infers joint motion from a signal of a soft sensor, wherein angle change of a wearer's joint part a sensor length change extractor extracting a length change of the soft sensor from a signal of the soft sensor generated by; a pre-learning unit for generating and learning a pre-calibration model inferring a change in angle of the joint from a change in length of the soft sensor based on musculoskeletal simulation data; and a transfer learning unit that adjusts the pre-calibration model based on motion data collected for angle changes of the wearer's joints.

Description

Calibration system and method for measuring joint motion, and wearable joint motion measuring apparatus using the same {Calibration system and method for measuring joint motion, and wearable joint motion measuring apparatus using the same}

The present invention relates to a calibration system and method for measuring joint motion, and a wearable joint motion measuring device for measuring joint motion of a wearer using the calibration system and method.

Wearable robots have been introduced to assist the posture or movement of people with lower extremity lesions due to accidents or aging and workers who move heavy loads, and are mainly used as a means for assisting users' muscular strength. Since such a wearable robot should be provided based on the wearer's physical disability or physical characteristics, measurement of the wearer's gait motion should be preceded in order to prescribe the robot.

To this end, it is important to check the movement of each joint of the user in various environments and to understand the motion intention. Various motion measurement methods have been introduced to identify such motion intention. Motion analysis is a method that measures changes in the musculoskeletal structure of the human body during a specific motion and measures information such as joint and muscle load, joint motion angle, and rotational torque to check information such as gait type, stride length, and gait speed. means

Conventionally introduced motion measurement devices use a plurality of cameras installed in a specific space, a plurality of markers attached to a specific part of the body, and a ground reaction force measuring device to measure the movement of joints in space, the gait state based on the rotation angle and ground reaction force. constructed in a way that measures

However, in the conventional method, it is very expensive to install the motion measuring device, and it is difficult to measure and analyze various motions due to the spatial limitations of the motion analysis device itself, and the respective markers are rearranged in space in real time. Since the coordinate values must be matched, complicated signal processing is required, and thus there is a problem in that it is difficult to measure motions in real life or in various environments.

Meanwhile, Xsens Motion Technol BV, no February, pp 1-7, 2009 discloses a wearable sensing device based on an inertial measurement unit. Inertial measurement devices have a compact sensor size and excellent measurement performance, but have the disadvantage that signal drift occurs when used for a long time, reducing the reliability of the signal and compromising the safety of the wearer because the sensor is made of a solid material. there is. In order to improve these disadvantages, research using advanced signal processing technology has been published, but this technology can compensate for drift only in the direction of gravity, so the advantages of the technology are limited. There are practical downsides to being taller.

In order to solve the disadvantages of wearable sensing suits based on these inertial measurement devices, Adv. Funct. Mater., vol. 26, no. 11, p. 1678-1698, Mar. 2016. and Int. J. Rob. Res., vol. 33, no. 14, p. 1748-1764, Dec. In 2014, a wearable sensing device using a soft sensor that senses various physical quantities based on flexible sensors such as cloth, rubber, and fluid was disclosed.

Wearable sensing devices using these soft sensors increase the wearing comfort and convenience of putting on and taking off clothes, secure user safety through softness, and have the advantage of being excellent in price competitiveness. However, in the case of wearable motion based on the conventional soft sensor, the nonlinearity and hysteresis of the measurement signal of the soft sensor are high, and it is difficult to map to the wearer's joint angle. It has been mainly applied only to the same limited single axis (single DOF) joint behavior.

Accordingly, IEEE/ASME Trans. Mechatronics, vol. 24, no. 1, p. 56?66, Feb. As in 2019, attempts to overcome the complexity of these signals through artificial intelligence technology (Deep Neural Network) are being carried out at the basic stage. A study on a full body suit capable of simultaneously measuring various joints by using an advanced calibration technique has been reported in an early form, but has disadvantages in that it requires a large amount of measurement data and can measure only a few predefined motions.

Therefore, in order to measure various daily motions in real life, a high-speed, high-performance calibration technique that can accurately measure the angle of each joint with only a small amount of data is required by overcoming these limitations, and based on this high-performance calculation model, complex joint behavior It is required to develop a device design that can comprehensively measure .

Xsens Motion Technol BV, no February, pp 1-7, 2009. Adv. Funct. Mater., vol. 26, no. 11, p. 1678-1698, Mar. 2016. Int. J. Rob. Res., vol. 33, no. 14, p. 1748-1764, Dec. 2014. IEEE/ASME Trans. Mechatronics, vol. 24, no. 1, p. 56?66, Feb. 2019.

An object of one aspect is to provide a calibration system and method for measuring joint motion.

In addition, another aspect is to provide a wearable joint motion measurement device using the calibration system.

To achieve the above purpose,

On one side,

A calibration system that creates a calibration model that infers joint motion from a signal from a soft sensor,

a sensor length change extractor extracting a length change of the soft sensor from a signal of the soft sensor generated by an angle change of a wearer's joint;

a pre-learning unit for generating and learning a pre-calibration model inferring a change in angle of a joint from a change in length of the soft sensor based on musculoskeletal simulation data; and

A transfer learning unit adjusting the pre-calibration model based on motion data collected for angle changes of the wearer's joints; a calibration system including a is provided.

In this case, the soft sensor may be a capacitive stretch sensor that measures a change in length of the soft sensor as a change in capacitance of the sensor.

The calibration system may further include a simulation data collection unit that collects musculoskeletal system simulation data, and may further include a motion data collection unit that collects motion data about angle changes of joints of the wearer.

The calibration system may infer angle changes for multiple degrees of freedom of a joint part from sensor signals transmitted from a plurality of soft sensors.

The calibration system may simultaneously infer angle changes of a plurality of joint parts from sensor signals transmitted from a plurality of soft sensors, and may infer predetermined motions and non-set motions of joints.

On the other side,

A calibration method for generating a calibration model that infers joint motion from a signal of a soft sensor,

extracting a length change of the soft sensor from a signal of the soft sensor generated by an angle change of a wearer's joint;

generating and learning a pre-calibration model inferring a change in angle of a joint from a change in length of the soft sensor based on musculoskeletal system simulation data; and

A calibration method comprising the steps of collecting motion data for angle change of the wearer's joint and adjusting the pre-calibration model based on the collected motion data.

On another aspect

a measuring unit disposed to be spaced apart from the wearer's joint and including a soft sensor generating a sensor signal according to a change in angle of the joint; and

There is provided a wearable joint motion measurement device including a; signal processing unit for inferring angle changes of joint parts from sensor signals of the measuring unit using the calibration system.

The measurement unit may include a first fixing member extending from one end of the soft sensor; and a second fixing member extending from the other end of the soft sensor).

The measuring unit may include a plurality of soft sensors.

The wearable joint motion measuring device may further include a signal transmitting unit for transmitting a signal of the measuring unit to the signal processing unit.

In addition, the wearable joint motion measuring device may further include a wearing unit worn on at least a part of the user's body and connected to the measuring unit.

In addition, the wearable joint motion measuring device may further include a signal output unit outputting a joint angle change generated by the signal processing unit.

A calibration system and method of a soft sensor for detecting joint motion provided in one aspect can measure angles for all degrees of freedom of a wearer's joints using a soft sensor, and generate a pre-calibration model from abundant data through musculoskeletal simulation and By learning and performing customized calibration for each wearer to adjust the pre-calibration model, it is possible to complete high-performance calibration with only a small amount of motion data of the wearer, and can perform not only preset motions but also Various motions can be measured quickly and accurately.

In addition, the wearable joint motion measuring device according to another aspect can measure a wearer's joint motion in real time with high performance and high speed without space limitation through wearing.

1 is a schematic diagram schematically showing a calibration system provided in one aspect;
2 is a schematic diagram showing a measurement unit of a wearable joint motion measuring device provided in another aspect;
3 is a schematic diagram showing an embodiment of the stretch sensor of the static capacitance measurement method,
4 is a photograph of a measuring unit placed on a wearer's ankle to measure ankle joint motion;
5 is a photograph and graph showing the experimental results of measuring the 2DOF behavior of the ankle joint using a joint motion measuring device including a measuring unit arranged as shown in FIG. 4;
6 is a graph comparing measurement performance of ankle joint motion using a calibration system provided in one aspect and a calibration system based on simulation data.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

on one side

A calibration system that creates a calibration model that infers joint motion from a signal from a soft sensor,

a sensor length change extractor extracting a length change of the soft sensor from a signal of the soft sensor generated by an angle change of a wearer's joint;

a pre-learning unit for generating and learning a pre-calibration model inferring a change in angle of a joint from a change in length of the soft sensor based on musculoskeletal simulation data; and

A transfer learning unit adjusting the pre-calibration model based on motion data collected for angle changes of the wearer's joints; a calibration system including a is provided.

1 is a schematic diagram schematically illustrating a calibration system provided in one aspect.

Hereinafter, a calibration system provided in one aspect will be described in detail with reference to the drawings.

The calibration system provided in one aspect is a calibration system that generates a calibration model that infers joint motion from a signal from a soft sensor, and more specifically, generates a calibration model that infers angle changes of the joint of a wearer wearing a soft sensor It may be a system that

The calibration system may further include a motion data collection unit that collects motion data of a wearer wearing the soft sensor to generate a calibration model.

The motion data is signal change of the soft sensor and angle change data of the joint part according to the joint motion of the wearer wearing the soft sensor, and is collected through a motion capture experiment of the wearer wearing the soft sensor. It can be.

The motion capture experiment may be measured using a motion capture system including a motion capture camera. For example, by wearing a soft sensor on the wearer's body and measuring the wearer's joint motion, that is, the angle change of the joint, through a motion capture camera, the angle change of the joint part according to the signal change of the soft sensor can be measured. there is.

In this case, the soft sensor is a sensor having flexibility and/or elasticity, and may be a stretch sensor (or strain sensor) that measures deformation such as tension or bending, and more preferably, capacitance changes according to a change in length of the sensor. It may be a capacitive measurement type stretch sensor (or strain sensor).

Among soft sensors, the capacitive stretch sensor has excellent linearity and repeatability and low hysteresis. Thus, as a soft sensor, it may be easier to map joint angles to sensor signals.

The calibration system includes a sensor length change extractor that extracts a length change of the soft sensor from a signal of the soft sensor generated by an angle change of a wearer's joint.

The sensor length variation extraction unit may be referred to as a sensor calibration map (S-map) unit.

The S-map unit extracts the length change of the soft sensor from the soft sensor signal generated according to the joint motion of the wearer wearing the soft sensor so that the length of the soft sensor changes according to the angle change of the wearer's joint. can be

The S-map unit may further include a signal conversion unit that compensates for nonlinearity and/or hysteresis of the soft sensor, and through this, compensates for nonlinearity and/or hysteresis of the sensor signal, thereby increasing linearity. can be increased and hysteresis can be reduced.

Accordingly, the calibration system provided in one aspect further includes a configuration for extracting a change in length of the soft sensor from a signal of the soft sensor, in contrast to a calibration system that directly infers a change in angle of a joint from a signal of the soft sensor. The change in length of the sensor can be more accurately mapped from the signal of the soft sensor.

The S-map unit, for example, converts the change in static capacitance sensed from the sensor into a change in length of the sensor using a linear calculation method or a calculation method for compensating hysteresis using a capacitive stretch sensor as the soft sensor. can make it

The calibration system includes a pre-learning unit that generates and learns a pre-calibration model inferring a change in angle of a joint from a change in length of the soft sensor based on musculoskeletal system simulation data.

The pre-learning unit may hereinafter be referred to as a motion decoding net (MD-net) unit.

The MD-net unit is a configuration for a step performed after the S-map unit, and can convert a change in length of a sensor extracted from the S-map unit into an angle change of a joint based on simulation data, and pre-calibration It may be a configuration that creates and learns a model.

To this end, the calibration system may further include a simulation data collection unit that collects musculoskeletal system simulation data.

The simulation data collection unit may collect data from OpenSim 4.0 software, as an example of storage where simulation data for inferring motions of the wearable robot is stored.

The pre-calibration model generated and learned through more simulation data for complex joint motions in the MD-net unit can infer various unset joint motions in addition to preset joint motions.

In this case, learning of the MD-net unit may be performed using a deep learning loss function (MSE, mean squared error) and adam (adaptive moment estimation) optimizer algorithm based on the musculoskeletal system simulation data, but is not limited thereto, and the simulation data Various algorithms for learning based on may be used.

The calibration system includes a transfer learning unit that adjusts the pre-calibration model based on motion data collected for angle changes of the wearer's joints.

The transfer learning unit may hereinafter be referred to as a T-net (Transfer-net) unit.

The T-net unit fine-adjusts the pre-calibration model learned based on the simulation to fit the wearer based on the motion data, and fine-tunes the pre-calibration model through the T-net unit to obtain a customized color for each wearer. A calibration model can be created.

6 (a) shows a calibration model (Tranfer model) generated through a calibration system provided in one aspect and a calibration model (Direct model) generated based on musculoskeletal simulation data for ankle joint motion used for model construction. It is a graph comparing the result measured using motion capture with the actual foot movement (Reference) measured, and FIG. 6 (b) is a calibration provided on one side for any ankle joint movement not used for model construction This is a graph comparing the results measured using the calibration model (Tranfer model) created through the system and the calibration model (Direct model) created based on the musculoskeletal simulation data with the actual foot movement (Reference) measured by motion capture.

As shown in FIG. 6 (a), both the transfer model and the direct model showed similar results to the actual ankle motion for the ankle joint motion used for model construction, while as shown in FIG. 6 (b), the model Regarding the arbitrary ankle movements not used in the construction, the transfer model is similar to the actual ankle movements, whereas the direct model has low similarity with the actual ankle movements.

As the above result, the calibration system provided in one aspect can infer various non-preset joint motions in addition to the preset joint motions by using the pre-calibration model based on rich simulation data, and at the same time, the pre-calibration model By being finely adjusted based on the wearer's motion data, there is an advantage in generating a calibration model capable of performing height calibration with high accuracy even with a small amount of motion data.

The T-net unit may linearly map an angle change on simulation data and a wearer's angle change based on motion data collected for angle change of a wearer's joint.

The relationship between the angle change on the simulation data and the wearer's angle change is non-linear. In order to map this linearly, a method of converting each angular change into a coordinate value of the Cartesian coordinate system of the toe and then linearly mapping it may be performed. By dealing with linear relationships in this way, there is an advantage in that relationships can be easily built with little data and at high speed.

In the calibration system provided in one aspect, the system can infer angle changes for multiple degrees of freedom of a joint part from sensor signals transmitted from a plurality of soft sensors, and can simultaneously infer angle changes of a plurality of joint parts.

In addition, in the case of the prior art, since an end-to-end method of directly mapping joint angles using a soft sensor is used, the accuracy and While there is a problem of low reliability, the calibration system provided in one aspect creates a calibration model using a modularization method divided into a plurality of steps, so that each step can be easily adjusted, and the linearity and hysteresis of the soft sensor can be lowered. It has the advantage of being able to perform calibration at high speed while increasing accuracy.

On the other hand, on the other side,

A calibration method for generating a calibration model that infers joint motion from a signal of a soft sensor,

extracting a length change of the soft sensor from a signal of the soft sensor;

generating and learning a pre-calibration model inferring a change in angle of a joint from a change in length of the soft sensor based on musculoskeletal system simulation data; and

A calibration method comprising the steps of collecting motion data for angle change of a wearer's joint and adjusting the pre-calibration model based on the collected motion data is provided.

Hereinafter, the calibration method will be described in detail for each step.

The calibration method is a calibration method for generating a calibration model for inferring joint motion from a signal of a soft sensor, and may be a method for generating a calibration model using the aforementioned calibration system.

The calibration method may perform a step of collecting motion data of a wearer wearing the soft sensor to generate a calibration model that infers a joint motion from a signal of the soft sensor.

The motion data may be collected through a motion capture experiment of a wearer wearing the soft sensor.

Thereafter, the calibration method includes extracting a length change of the soft sensor from a signal of the soft sensor.

The step may be a step of processing a signal of the soft sensor generated during the motion capture experiment.

This step may be referred to as a sensor calibration map (S-map) step.

The S-map step extracts the length change of the soft sensor from the soft sensor signal generated according to the joint motion of the wearer wearing the soft sensor so that the length of the soft sensor changes according to the angle change of the wearer's joint. may be a step.

The S-map step may further include a signal conversion unit that compensates for nonlinearity and/or hysteresis of the soft sensor, and through this, the nonlinearity and/or hysteresis of the sensor signal is compensated for, thereby compensating for the soft sensor. From the signal, the length change of the sensor can be more accurately mapped.

The S-map unit, for example, converts the change in static capacitance sensed from the sensor into a change in length of the sensor using a linear calculation method or a calculation method for compensating hysteresis using a capacitive stretch sensor as the soft sensor. can make it

Next, the calibration method includes generating and learning a pre-calibration model inferring a change in angle of a joint from a change in length of the soft sensor based on musculoskeletal system simulation data.

This step may hereinafter be referred to as a motion decoding net (MD-net) step.

The MD-net step is a step performed after the S-map step, and the length change of the sensor extracted in the S-map step can be converted into an angle change of the joint based on simulation data, and a pre-calibration model It may be a stage of creation and learning.

To this end, the calibration method may further include a simulation data collection step of collecting musculoskeletal simulation data.

The simulation data may be collected from OpenSim 4.0 software, which is an example of storage where simulation data for inferring motions of the wearable robot is stored.

In the MD-net step, the pre-calibration model generated and learned through more simulation data for complex joint motions may infer various unset joint motions in addition to preset joint motions.

In this case, the learning of the MD-net step may be performed using a deep learning loss function (MSE, mean squared error) and adam (adaptive moment estimation) optimizer algorithm based on the musculoskeletal system simulation data, but is not limited thereto, and simulation Various algorithms for learning based on data may be used.

Next, the calibration method includes collecting motion data about angle changes of the wearer's joints and adjusting the pre-calibration model based on the collected motion data.

This step may be referred to as a T-net (Transfer-net) step.

The T-net step is a step of fine-tuning a pre-calibration model learned based on simulation to suit the wearer based on the motion data, and a customized calibration model suitable for the wearer can be formed through the T-net unit. .

The T-net step may linearly map the angular change on the simulation data and the wearer's angular change based on the motion data collected for the angular change of the wearer's joint.

The relationship between the angle change on the simulation data and the wearer's angle change is non-linear. In order to map this linearly, a method of converting each angular change into a coordinate value of the Cartesian coordinate system of the toe and then linearly mapping it may be performed. By dealing with linear relationships in this way, there is an advantage in that relationships can be easily built with little data and at high speed.

Also, on the other side

a measuring unit disposed to be spaced apart from the wearer's joint and including a soft sensor generating a sensor signal according to a change in angle of the joint; and

There is provided a wearable joint motion measurement device including a; signal processing unit for inferring angle changes of joint parts from sensor signals of the measuring unit using the calibration system.

Hereinafter, a wearable joint motion measurement device provided in another aspect will be described in detail with reference to the drawings.

2 is a schematic diagram showing a measurement unit 100 of a wearable joint motion measuring device provided in another aspect.

A wearable joint motion measurement device according to another aspect may be a device for measuring joint motion, preferably a change in angle of a joint through wearing, and more preferably a device for measuring angle change for multiple degrees of freedom of a joint. can be

In this case, the joint may be a knee joint, an ankle joint, a wrist joint, a hip joint, or the like, and may be a joint having one degree of freedom or a joint having multiple degrees of freedom of two or more degrees of freedom.

In addition, the wearable joint motion measurement device according to another aspect can quickly and accurately measure various undefined daily motions other than a preset motion by using the transfer-learned calibration system based on musculoskeletal system simulation data.

The measuring unit 100 includes a soft sensor 10 .

The soft sensor 10 is a sensor having flexibility and/or elasticity, and may be a stretch sensor (or strain sensor) that measures deformation such as tension or bending. It may be a stretch sensor (or strain sensor) of a changing capacitance measurement method.

Among soft sensors, the capacitive stretch sensor has excellent linearity and repeatability and low hysteresis. Thus, as a soft sensor, it may be easier to map joint angles to sensor signals.

The capacitance measurement type stretch sensor may have a structure in which a conductor layer is formed between insulator layers.

3 is a schematic diagram showing an embodiment of the static capacitance measurement method stretch sensor. As shown in FIG. 3, the static capacitance measurement method stretch sensor includes three conductor layers (first to third conductor layers). It may be formed in a 5-layer structure with two insulator layers (first and second insulator layers) disposed therebetween, but is not limited thereto, and is configured in various other forms exhibiting characteristics in which capacitance changes according to a change in the length of the sensor. It can be.

Meanwhile, the soft sensor 10 may be preferably disposed to be spaced apart from the wearer's joints when measuring joint motion. This is to prevent a problem of deteriorating the measurement reliability of the angle change of the joint due to additional deformation in addition to the length change of the sensor, if the soft sensor 10 is disposed on the surface of the wearer's joint. Additional deformations such as , pressing, and bending may occur.

Accordingly, the soft sensor 10 may be disposed on a first link extending in a first direction from a joint of the wearer or a second link extending in a second direction from the joint when measuring joint motion. there is.

In this case, the first link and the second link may mean a part connected to a joint. For example, as shown in FIG. 1 , when the joint is a wearer's knee joint, the first link may be the shin and the second link may be the thigh.

Meanwhile, the measuring unit 100 arranges the soft sensor 10 to be spaced apart from the wearer's joint, and the length of the soft sensor 10 changes according to the angle change of the joint during the user's joint movement. As a configuration for connecting the soft sensor 10 and a joint, a first fixing member 20 extending from one end of the soft sensor 10 and a second fixing member extending from the other end of the soft sensor 10 ( 30) may be further included.

The first fixing member 20 may fix one end of the soft sensor 10 to the first link, and the second fixing member 30 may move the other end of the soft sensor 10 away from the joint. It may be fixed to the second link extending in the second direction.

In addition, the first fixing member 20 and the second fixing member 30 connect the soft sensor to the wearer's joint, but are made of a material with insufficient or no elasticity and/or flexibility so as not to expand or contract according to the angle change of the joint. It may be desirable to configure

Meanwhile, the measuring unit 100 may include a plurality of soft sensors.

A plurality of soft sensors may be disposed for each joint as many as the number of degrees of freedom of the joint, and through this, each sensor detects a change in the joint angle of each degree of freedom of the joint, thereby measuring angle change for all degrees of freedom in the joint. can For example, in order to measure the two-DOF joint of the ankle, two soft sensors may be disposed on the shin in a direction perpendicular to the joint axis of each DOF.

In addition, by disposing the soft sensors on a plurality of joints, angle changes of a plurality of joints can be simultaneously measured.

The measuring unit 100 fixes each of the plurality of soft sensors to a part of the wearer's body and extends from one end and the other end of each of the plurality of soft sensors to connect each of the plurality of soft sensors to a joint. A first fixing member and a second fixing member may be further included.

FIG. 4 is a photograph in which the measurement unit 100 is arranged along one side to measure ankle joint motion, and FIG. 5 measures the 2DOF behavior of the ankle joint by disposing the measurement unit 100 as shown in FIG. 4 . These are photos and graphs showing the results of one experiment.

As shown in FIG. 4 , in order to measure the motion of the ankle joint, the measuring unit 100 may include four soft sensors 10, and the four soft sensors 10 may include a talocrural joint axis. It may include a first soft sensor (DF sensor) for measuring dorsiflexion disposed on a joint axis and a second soft sensor (PF sensor) for measuring plantar flexion, and the subtalar joint It may include a third soft sensor (SU sensor) for measuring supination and a fourth soft sensor (PF sensor) for measuring pronation on a subtalar joint axis, as shown in FIG. 5 . As such, joint motion of multiple degrees of freedom can be measured through the measurement unit 100 .

Meanwhile, the wearable joint motion measuring device provided in another aspect includes a signal processing unit for inferring an angle change of a joint part from a sensor signal of the measuring unit using the calibration system.

The signal processing unit can measure various motions for multiple degrees of freedom within one joint and simultaneously measure motions for a plurality of joints by using the calibration model generated using the calibration system.

The calibration system may include some or all of the components of the aforementioned calibration system.

The wearable joint motion measuring device may further include a signal transmitting unit for transmitting a signal of the measuring unit to the signal processing unit.

The signal transmitting unit may transfer the signal generated by the measurement unit to the signal transmitting unit through a wired or wireless communication method.

The signal transfer unit may be worn on a part of the wearer's body, for example, on the wearer's waist, but is not limited thereto, and may be communicatively connected to the measurement unit and the signal processing unit in a worn or unworn state.

The wearable joint motion measurement device may further include a wearable part worn on at least a part of the user's body and connected to the measurement part 100 .

The wearing unit may be configured to be worn on a part of the user's body while being connected to the measurement unit, and may be in the form of a suit.

In this case, the suit may be in the form of a full-body suit worn on the user's entire body, or may be in the form of a partial suit that is modularized to suit each part of the user's body.

For example, in order to measure the motion of the ankle joint, the wearable part is in the form of a shoe worn on the foot, and one end of the measuring part 100 is connected to the wearing part in the form of a shoe, and the other end of the measuring part 100 is connected to the wearing part of the wearer's body. It may be fixed to a part of the body, and by moving the ankle joint while wearing the wearable part in the form of a shoe, the movement of the ankle joint is controlled through a soft sensor connected to the ankle joint and a signal processing unit communicatively connected to the soft sensor. can be measured

The wearable joint motion measuring device may further include a signal output unit outputting a joint angle change generated by the signal processing unit.

The signal output unit outputs the signal processed by the signal processing unit to the user, and may be an image output unit such as a display, but is not limited thereto, and various signal output devices that process digital signals may be used.

Since the present invention can have various changes and various embodiments, specific embodiments will be described in detail with reference to the drawings. However, this is not intended to limit the present invention to specific embodiments, and should be understood to include all modifications, equivalents, or substitutes included in the spirit and scope of the present invention. Like reference numbers have been used for like elements in describing each figure.

Terms such as first, second, A, and B may be used to describe various components, but the components should not be limited by the terms. These terms are only used for the purpose of distinguishing one component from another. For example, a first element may be termed a second element, and similarly, a second element may be termed a first element, without departing from the scope of the present invention.

In addition, when an element is referred to as being “connected” to another element, it should be understood that although it may be directly connected or connected to the other element, other elements may exist in the middle. On the other hand, when an element is referred to as “directly connected” or “directly connected” to another element, it should be understood that no other element exists in the middle.

Terms used herein are only used to describe specific embodiments, and are not intended to limit the present invention. Singular expressions include plural expressions unless the context clearly dictates otherwise. In this application, the terms "include" or "have" are intended to designate that there is a feature, number, step, operation, component, part, or combination thereof described in the specification, but one or more other features It should be understood that the presence or addition of numbers, steps, operations, components, parts, or combinations thereof is not precluded.

10: soft sensor
20: first fixing member
30: second fixing member
100: measurement unit of the wearable joint motion measurement device

Claims (15)

A system for generating a calibration model that infers joint motion from a signal of a soft sensor,
a sensor length change extractor extracting a length change of the soft sensor from a signal of the soft sensor generated by an angle change of a wearer's joint;
a pre-learning unit for generating and learning a pre-calibration model inferring a change in angle of the joint from a change in length of the soft sensor based on musculoskeletal simulation data; and
and a transfer learning unit configured to adjust the pre-calibration model by linearly mapping the angular change inferred by the pre-learning unit and the angular change obtained from motion data of the wearer with respect to the joint region.
According to claim 1,
The system of claim 1 , wherein the soft sensor is a capacitive stretch sensor that measures a change in length of the soft sensor as a change in capacitance of the sensor.
According to claim 1,
The system
A system further comprising a motion data collection unit configured to collect motion data for angle changes of the wearer's joints.
According to claim 1,
The system
A system further comprising a simulation data collection unit for collecting musculoskeletal simulation data.
According to claim 1,
The system infers an angle change for multiple degrees of freedom of a joint from sensor signals transmitted from a plurality of soft sensors.
According to claim 1,
The system simultaneously infers angle changes of a plurality of joint parts from sensor signals transmitted from a plurality of soft sensors.
According to claim 1,
The system
A system that infers preset motions and non-set motions of joints.
A method of generating a calibration model that is performed by the system of claim 1 and infers joint motion from a signal of a soft sensor,
extracting a length change of the soft sensor from a signal of the soft sensor generated by an angle change of a wearer's joint, performed by a sensor length change extractor;
generating and learning a pre-calibration model that infers an angular change of a joint from a length change of the soft sensor based on musculoskeletal system simulation data, which is performed by a pre-learning unit; and
Performed by a transfer learning unit, adjusting the pre-calibration model by linearly mapping the angle change obtained from the wearer's motion data and the angle change inferred by the pre-learning unit with respect to the joint region; Calibration method comprising a.
According to claim 8,
The calibration method
Collecting motion data for angle changes of the wearer's joints; further comprising a calibration method.
According to claim 8,
The calibration method
Further comprising, a calibration method comprising: collecting simulation data for collecting musculoskeletal system simulation data.
a measurement unit disposed to be spaced apart from a wearer's joint and including a soft sensor generating a sensor signal according to a change in angle of the joint; and
A wearable joint motion measurement device comprising a; signal processing unit for inferring angle change of a joint from a sensor signal of the measuring unit using the system of claim 1 .
According to claim 11,
The measuring part
Further comprising a fixing member connecting the soft sensor to the wearer's joint,
the fixing member
a first fixing member fixing one end of the soft sensor to a body part; and
A wearable joint motion measuring device further comprising a; second fixing member connecting the other end of the soft sensor to a joint.
According to claim 11,
The wearable joint motion measurement device
A wearable joint motion measurement device further comprising a; signal transmission unit for transmitting the signal of the measurement unit to the signal processing unit.
According to claim 11,
The wearable joint motion measurement device
A wearable joint motion measuring device further comprising a; wearing part on which at least a part of the user's body is worn and connected to the measuring part.
According to claim 11,
The wearable joint motion measurement device
A wearable joint motion measurement device further comprising a; signal output unit for outputting a joint angle change generated by the signal processing unit.

Images