KR20090060755A - Method for building a human body model for human motion analysis - Google Patents

Abstract

A human model production method for human action analysis is provided to realize stable actions during unfolding, lifting up/down, and tilting actions, thereby being applicable to both large and small displays. A lot of location information of a body of a testee and a ground reaction force of the testee are obtained(S10). A human model is applied to one of the testee's actions, and each size of a figure is determined by using the location information(S40). A ZMP(Zero Moment Point) of the human model is calculated(S50). A substantial ZMP of the testee is calculated by using the ground reaction force(S60). The human model is compensated, so that an error between the ZMP of the human model and the substantial ZMP of the testee is minimized.

Description

METHODO FOR BUILDING A HUMAN BODY MODEL FOR HUMAN MOTION ANALYSIS

The present invention relates to a method for generating a human body model used to analyze human kinematic movements, and more particularly, to a method for generating a human body model that humanoid robot can be used to mimic human kinematic movements. will be.

Humanoid robots not only look at humans in appearance but also act like humans. These humanoid robots must mimic not only the external movements of the human body but also the kinematic movements to move similarly to humans.

In order to obtain information for mimicking the human's appearance, an optical motion capturing method is used in which a plurality of markers are attached to a test subject's body and information is obtained by the movement of these markers. This method is widely used because it can easily acquire human motion.

In order to acquire information to mimic human kinematic movements, the trace of the center of mass (hereinafter simply referred to as 'COM') or the zero moment point (hereinafter referred to as 'ZMP') We need to analyze the trajectory of. In order to obtain a human COM or ZMP trajectory, it is necessary to measure the mass and movement of each part of the human body. For this reason, there is a need for a method of generating a human model that divides the human body into several parts and predicts the mass of each part. However, since the material constituting the human body is not uniform, the density of each part of the human body is different, and the ratio of bones and muscles of each part also varies from person to person, in order to accurately consider these differences, a human body model suitable for each subject is required. Must be created.

Many researchers have proposed several human models.

Hanvan (EPHanavan, “A mathematical model of the human body”, Report No. AMRLTR 64-102, AD-608-463. Ohio: Aerospace Medical Research Laboratories , October 1964.), Wieden (MRWeadon, “The simulation of aerial movement: Ii.a mathematical inertia model of the human body ”, Journal of Biomechanics , 1990, vol. 23, pp. 75-88), lower body (H.Hatze,“ A mathematical model for the computational determination of parameter values of anthropometric segments ”, Journal of Biomechanics , 1980, vol. 13, pp. 833-843), RK Jensen,“ Estimation of the biomechanical properties of three body types using a photogrammic method ”, Journal of Biomechanics , 1978, vol. 11, pp. 349-358) suggested models that can represent the human body. However, they relied on manual work to determine the size of each part of the human body model and did not suggest an automatic method.

Zashosky (VMZatsiorsky, Kinetics of Human Motion , Human Kinetics, Champaign, IL, USA, 2002) measured the inertial characteristics of each part of the human body in 100 men. He acquired the center of gravity (COM), mass, and moment of inertia of each part of the human body.

Popovic, M. Popovic, A. Hofmann, and H. Herr, “Angular momentum regulation during human walking: biomechanics and control” in Proc. Of IEEE Int. Conf. On Robotics and Automation , 2004, vol. 3, pp. 2405-2411.) Developed a human body model to measure the distribution of human masses. They measured the human body directly or used the data referring to the measured literature. However, this measurement is very time consuming.

Determining the size of each part of the human body model using the conventional human body models mentioned is time-consuming and complicated. In addition, since an average or standard human model is obtained instead of the human model of the test subject, it is difficult to obtain accurate information of the test subject. Therefore, it is necessary to make a human body model that can best express the subject, and furthermore, a human body model that can easily obtain the ZMP of the kinematic behavior of the subject only by such a human body model is needed.

The present invention was devised to solve the above problems, and an object of the present invention is to provide a method for generating a human body model in which the dimensions of each body part can represent the dimensions of the corresponding body part of the subject relatively accurately.

In order to achieve the above object and other objects, the human body model generating method of the present invention is a method for generating a human body model composed of a plurality of figures, a plurality of the subject's body at a predetermined time interval during the operation of the subject Acquiring the location information and the ground reaction force of the test subject; Applying a human body model to one of the subject's motions and determining the dimensions of each figure using position information; Calculating a ZMP of the human body model; Calculating the actual ZMP of the test subject using the ground reaction force; Compensating the human model so that the error between the ZMP of the human model and the actual ZMP of the test subject is minimized within the constraints.

The dimensioning step of the figure includes determining fixed and variable dimensions of the figure, and the human body model correcting step includes optimizing the variable dimension so that the error is minimized within the constraint, and using the optimized variable dimension Modifying the dimensions of the.

The variable dimension optimization step includes calculating an error, checking whether the variable used in calculating the error satisfies the constraint, and if the constraint is satisfied, replacing the variable used in the error calculation with the optimized variable. It includes a step.

The constraint may be a condition in which the variable dimension of the figure of the human body model is larger or smaller than the variable dimension of the other of the figures.

The figure of the human body model is any one of a sphere, a cylinder and a cube, and the fixed dimension is the length of the cylinder, the horizontal and vertical length of the cube, and the variable dimension is the radius of the sphere, the radius of the cylinder and the height of the cube.

The position information and ground reaction acquiring step may be performed in the motion capture system. The ground reaction obtaining step may include installing a ground reaction meter in the motion capture system and obtaining ground reaction force from the ground reaction meter.

In addition, the present invention comprises the steps of preparing a human body model consisting of a plurality of figures; Acquiring the ground reaction force of the test subject from the ground reaction force and a plurality of pieces of position information of the subject's body for each frame while the test subject is operating on the ground reaction gauge in the motion capture system; Determining a fixed dimension and a variable dimension of each figure by applying the human body model to the test subject's motion for each frame; Calculating the ZMP of the human body model for each frame; Calculating the actual ZMP of the test subject by using the ground reaction force for each frame; Optimizing the variable dimensions such that the sum of the errors between the ZMP of the human model and the actual ZMP of the subject over the total number of frames is minimized within the constraints; And calibrating the human body model using the optimized variable dimensions.

The variable dimension optimization step includes calculating a sum of an error, checking whether a variable dimension used when calculating the sum of an error satisfies a constraint, and optimizing the variable dimension used when calculating a sum of an error if the constraint is satisfied. And substituting the variable variable.

The constraint is a condition where the variable dimension of one of the figures of the human body model is larger or smaller than the variable dimension of the other one of the figures.

The figure is any one of a sphere, a cylinder and a cuboid, the fixed dimension is the length of the cylinder and the horizontal and vertical lengths of the cube, and the variable dimensions are the radius of the sphere, the radius of the cylinder and the height of the cube.

According to the method for generating a human body model of the present invention, a test subject can complete a customized human body model for each subject by a simple operation performed by the motion capture system and the ground reaction force measuring instrument. Using the human body model generated by the present invention, it is possible to easily calculate a COM trajectory or a ZMP trajectory for analyzing various motions of the subject without measuring ground reaction force. The ground reaction force gauge used to measure ground reaction force is quite expensive, so it is quite difficult to install it widely to analyze the wide range of motion of the subject. However, if the human body model according to the present invention is used, the ZMP trajectory or the COM trajectory can be easily obtained using only the marker data of the test subject's motion obtained in the motion capture system.

Hereinafter, with reference to the accompanying drawings will be described in detail a method for generating a human body model of the present invention.

The present invention provides a method for generating a human body model composed of a plurality of figures. The figure constituting the human body model of the present invention shapes each body part of the human body. By determining the dimensions of the figure, the human body model can be completed. The optimal dimensions of the figures are obtained when the human body model of the present invention is applied to the subject's motion, the ZMP of the human body model conforming to the motion is obtained, and the error of both is minimized compared to the actual ZMP in the motion. The human body model is completed by calibrating the human body model using the optimal dimensions.

1 is a flowchart illustrating a method of generating a human body model of the present invention.

Referring to FIG. 1, the method for generating a human body model of the present invention includes: obtaining a plurality of pieces of position information of a subject's body at predetermined time intervals during a subject's operation (S10) and obtaining a ground reaction force of the subject (S20). ; Applying a human body model to one of the subject's motions (S30); Determining an initial dimension of each figure based on the position information (S40); And calibrating the human model such that the error between the ZMP of the human model and the actual ZMP of the test subject is minimized within the constraints. Determining the size of the figure includes determining fixed and variable dimensions of the figure. The human body model correction step includes: calculating a ZMP of the human body model (S50); Calculating an actual ZMP of the test subject based on the ground reaction force (S60); Confirming whether an error between the ZMP of the human body model and the actual ZMP of the test subject is minimum within a constraint (S70); If not the minimum step of modifying the dimensions of the figure (S80). The anatomical model is completed when the optimal geometrical dimension is obtained such that the error between the ZMP of the anatomical model and the actual ZMP of the test subject is minimized within the constraints.

Hereinafter, each process of the method of generating a human body model of the present invention will be described.

Modeling of the human body

Figure 2 shows a human body model employed in the method for generating a human body model of the present invention.

The human body model 100 according to the present invention includes a head portion 101, a neck portion 102, an upper body portion 103, an upper arm portion 104, a lower arm portion 105, a hand portion 106, and a lower torso. It consists of ten body parts, such as the part 107, the upper leg part 108, the lower leg part 109, and the foot part 110, and each body part is simplified by the simple figure. The head portion 101 and the hand portion 106 are simplified into spheres, the foot portion 110 is simplified into a cuboid, and the remaining portions (neck portion 102, upper torso portion 103, upper arm portion 104, lower arm portion) 105, lower body portion 107, upper leg portion 108, and lower leg portion 109 are simplified to cylinders.

The density of each body part is assumed to be the same. Therefore, when the specific dimensions of each body part are determined, not only the mass of each body part of the human body model 100 can be obtained, but also the ZMP of the human body model 100 when the human body model 100 performs a specific operation. Or you can get COM.

Body movement acquisition

Acquire various motions of the human body for applying the human body model 100. In an embodiment of the present invention, to obtain a human body motion for applying the human body model 100, a motion capture system is used.

3 illustrates human body motion acquisition. Acquisition of the human body motion is performed by using a motion capture system and ground force plates 21 and 22. Ground reaction force measuring instruments (21, 22) are installed in the motion capture system. The test subject smokes on the ground reaction force measuring instruments 21 and 22 to obtain the test subject's motion (motion) through the motion capture system.

In the motion capture system, a marker (not shown) is attached to a plurality of surfaces of the subject's body, and the human body's motion (motion) as shown by using position information (marker data) obtained from the marker when the subject is operating and Location information of each part of the human body is obtained. Preferably, the ground reaction force is measured by the ground reaction meter while each motion is measured by the motion capture system. That is, the motion capture system and the ground reaction force measuring instruments 21 and 22 are preferably synchronized.

The test subject postponed on two ground reaction meters (21, 22). The test subject evenly moves each part of the test subject's body by shaking his arms, legs, trunk, and head. Specifically, it is preferable to postpone so that all other parts of the subject's body move while one or both feet are supported on the ground reaction force measuring instruments 21 and 22.

The motion capture system calculates the subject's motion and position information every predetermined time when the subject acts. That is, when one motion calculated by the motion capture system is one frame, a plurality of continuous motions are calculated by capturing the subject's motions at regular time intervals during the motion capture system during the subject's operation. The motion capture system in this embodiment captures the subject's motion in 1 / 120th of a second. For example, if the subject is operated for 10 seconds, 1200 frames of motion are obtained. Further, as described above, since the motion capture system and the ground reaction meter are synchronized, the motion of the test subject and the ground reaction are acquired every frame of the motion capture system.

Application of the human body model

The subject's motion shape is obtained from the marker data obtained through the motion capture system. By applying the human body model 100 to the motion shape formed based on the marker data, it is possible to calculate the ZMP of the human body model 100 in each operation situation of the test subject.

Figure 4 shows an example of applying the human body model 100 to the motion of one frame obtained in the motion capture system. In detail, FIG. 4 illustrates an example in which the upper left arm portion 104 and the lower arm portion 105 of the human body model 100 are applied to the motion of the test subject.

The upper arm of the motion shape 11 of one frame obtained by the motion capture system coincides with the center line of the cylinder corresponding to the upper arm portion 104 of the human body model 100 in the same direction as the line connecting the shoulder and elbow. Similarly, the lower arm is aligned with the centerline of the lower arm portion 105 of the human body model to the line connecting the elbow and the wrist.

In addition, the upper part of the body of the motion shape 11 of one frame obtained by the motion capture system matches the center of the cylinder of the upper body part 103 to a line connecting the neck and waist center. In the lower part of the torso of the motion shape 11, the center line of the cylinder of the lower torso part 107 is matched with the line connecting the end point of the waist and the upper leg of both legs. In the upper leg of the motion shape 11, the center line of the cylinder of the upper leg part 108 coincides with the line which connects an upper end point with a knee. The lower leg of the motion shape 11 is made to match the centerline of the cylinder of the lower leg part 109 with the line which connects a knee and an ankle. In addition, the head and both hands of the motion shape 11 match the center of the sphere of the head 101 and the hand 106 of the human body model to the center of the head and the center of the hand.

Determination of the dimensions of the body parts of the human body model

When acquiring the subject's motion, the length of each part of the subject's body may be known using position information of a plurality of markers attached to the subject's body surface. In addition, it is possible to know some dimensions of figures constituting each body part of the human body model 100 in which the subject body is simplified into figures.

For example, based on the positional information of the marker, in the case of a rectangular parallelepiped (foot portion 110), the horizontal length and the vertical length can be determined, and the cylinder (neck portion 102, upper body portion 103, upper arm portion 104). In the case of the lower arm portion 105, the lower trunk portion 107, the upper leg portion 108, and the lower leg portion 109, the length may be determined. That is, fixed dimensions such as the length of the cylinder and the horizontal and vertical lengths of the rectangular parallelepiped can be determined.

However, the radius of the sphere (head 101, hand 106) and the cylinder (neck 102, upper torso 103, upper arm 104, lower arm 105, lower torso) The radius of the portion 107, the upper leg portion 108, the lower leg portion 109, and the height of the cuboid (foot portion 110) may be determined from the marker data (position information of the subject body) to match the characteristics of the subject. Can't. Therefore, the radius of the sphere, the radius of the cylinder, the height of the cuboid, etc. become variable dimensions for correcting the anatomical model. These variable dimensions are optimized according to the characteristics of the test subject by the correction steps S60 and S70, and the size of each body part is determined in such a state, thereby completing the human body model 100.

Acquisition of ZMP Trajectory of Human Body Model

As described above, the human body model 100 of the present invention is applied to the motion of one frame obtained by the motion capture system. Then, the position of each body part of the human body model 100 is determined in accordance with any one point of the operation of the test subject. In this state, the ZMP of the human body model 100 is calculated. Then, the ZMP of the human body model 100 corresponding to any point in time of the test subject's operation is calculated.

M. Vukobratovic, B. Borovac, proposed a formula to summarize the concept of ZMP and obtain ZMP (“Zero-moment point-thirty five years of its life”, Int. Journal of Humanoid Robotics, vol.1, pp.157-173, 2004) Using the proposed method, the x coordinate and y coordinate of the ZMP of the human body model of the present invention can be obtained from the following equation (1) and (2).

Here, the subscript i indicates each body part of the human body model 100 (eg, the head part 101, the neck part 102, the upper torso part 103, the upper arm part 104, the lower arm part 105, The hand part 106, the lower torso part 107, the upper leg part 108, the lower leg part 109, and the foot part 110 are shown in turn, and n represents the total number of body parts. x _i , y _i , z _i , I _i and m _i are the coordinate values, moments of inertia and mass of x, y and z of the center of the i-th body part of the anatomical model, respectively.

The x, y and z coordinate values of the center of each body part are related to the fixed dimensions as described above. However, the moment of inertia and mass vary with variable dimensions (eg, the radius of the sphere, the radius of the cylinder, the height of the cuboid). That is, in Equations 1 and 2, m _i and I _i vary according to each body part of the human body model 100 that varies according to the variable dimensions. When calibrating the anatomical model 100, these values continue to change.

Actual ZMP Calculation

The actual ZMP of the subject can be calculated using the ground reaction force. As described above, the ground reaction force may be obtained by using the ground reaction measuring instruments 21 and 22 while acquiring the motion shape of the test subject using the marker. 5 illustrates the forces and moments measured by ground reaction measurements.

Equation 3 below can be derived from the ZMP definition.

는 모션캡쳐시스템의 고정좌표계(도 5의 바닥에 정의한 좌표계)에서의 i번째 반발력의 위치벡터,

는 ZMP의 위치벡터,

는 i번째 지면반력벡 터이다.here,

Is the position vector of the i th repulsive force in the fixed coordinate system (coordinate system defined at the bottom of FIG. 5) of the motion capture system,

Is the position vector of ZMP,

Is the i th ground reaction vector.

From the above Equation 3, the x coordinate and the y coordinate of the actual ZMP can be derived as Equations 4 and 5 below.

In Equations 3 to 5, i is the order of the ground reaction force, and n is the number of the ground reaction force. In addition, the subscripts x, y and z of the variables represent the x, y and z direction components of the corresponding respective vectors.

From Equations 4 and 5, when the test subject operates on the ground reaction force meter in the motion capture system, the actual ZMP coordinates of the test subject per frame of the motion capture system can be obtained. During the time of executing the motion capture, the coordinates of the actual ZMP may be obtained to obtain a trace of the actual ZMP of the test subject.

Calibration of the human body model

As described above, ZMP of the non-optimized human body model 100 is calculated by applying each body part of the human body model 100 to the motion of the test subject obtained per frame of the motion capture system. By comparing the calculated ZMP and the actual ZMP of the test subject measured from the ground reaction force measuring instruments 21 and 22, the variable dimensions of the body parts 101 to 110 of the human body model 100 are adjusted to match the characteristics of the test subject. Optimize.

To optimize variable dimensions, we use a constrained nonlinear multivairiable optimization algorithm.

The above optimization algorithm will be described.

The error of the ZMP of the human body model 100 obtained in each frame of the motion capture system and the actual ZMP of the test subject is summed by the total number of frames of the motion capture system that acquires the test subject's motion shape. When the error sum is first obtained, the value of the variable dimension input for calculating the ZMP of the human body model 100 is set to an arbitrary value. The variable dimension is obtained when the obtained total error is minimized while satisfying the constraint. By modifying the dimensions of each body part of the human body model 100 using this variable dimension, the human body model 100 matches the characteristics of the test subject.

The equation for calculating the sum error may be expressed as in Equation 6 below.

은 모션캡쳐시스 템의 일 프레임 당 캡쳐된 모션의 데이터,

는 가변치수들(즉, 구의 반지름, 원기둥의 반지름, 직육면체의 높이)의 값을 가지는 벡터,

는 시간 t에서 가변치수벡터 일 때 일 프레임의 모션에 인체모델(100)을 적용하여 계산한 ZMP의 값,

는 모션캡쳐시스템의 일 프레임 당 지면반력측정기(21, 22)로부터 측정된 지면반력을 사용하여 계산된 피실험자의 실제 ZMP의 값이다.Where t is time, n _t is the total number of frames of body motion captured,

Is the data of motion captured per frame of the motion capture system,

Is a vector with values of variable dimensions (ie sphere radius, cylinder radius, cuboid height),

Is the variable dimension vector at time t. When ZMP value calculated by applying the human body model 100 to the motion of one frame,

Is the value of the actual ZMP of the subject calculated using the ground reaction force measured from the ground reaction force meters 21 and 22 per frame of the motion capture system.

를 최소화하는 가변치수벡터

를 구하는 것을 목적으로 한다. 가변치수벡터

의 값은 머리 부분(101)의 반지름, 목 부분(102)의 반지름, 윗몸통 부분(103)의 반지름, 위팔 부분(104)의 반지름, 아래팔 부분(105)의 반지름, 손 부분(106)의 반지름, 아랫몸통 부분(107)의 반지름, 윗다리 부분(108)의 반지름, 아랫다리 부분(109)의 반지름, 발 부분(110)의 높이가 된다. The optimization of variable dimensions is the objective function of Equation 6.

Variable dimension vector to minimize

To obtain. Variable dimension vector

The value of is the radius of the head portion 101, the radius of the neck portion 102, the radius of the upper body portion 103, the radius of the upper arm portion 104, the radius of the lower arm portion 105, the hand portion 106 Radius of, the radius of the lower body portion 107, the radius of the upper leg 108, the radius of the lower leg 109, the height of the foot portion 110.

In addition, the following equations 7 to 9 are used as constraints used in the optimization algorithm.

The constraint is a condition in which one variable dimension of the figures constituting the human body model 100 is larger or smaller than the other variable dimension. These constraints are related to the size between the body parts and are defined by intuitive consideration of the normal human body. For example, the condition that the radius of the head portion 101 is greater than or equal to the radius of the neck portion 102 and the condition that the radius of the upper leg portion 108 is greater than or equal to the radius of the lower leg portion 109 are employed as constraint conditions. do.

의 값들의 순서를 윗다리 부분(108)의 반지름, 아랫다리 부분(109)의 반지름, 아랫몸통 부분(107)의 반지름, 윗몸통 부분(103)의 반지름, 목 부분(102)의 반지름, 머리 부분(101)의 반지름, 위팔 부분(104)의 반지름, 아래팔 부분(105)의 반지름, 손 부분(106)의 반지 름, 발 부분(106)의 높이로 했을 때의 값이다. A행렬의 첫 번째 행에서 알 수 있는 바와 같이, 윗다리 부분(101)의 반지름은 아랫다리 부분(102)의 반지름보다 크거나 최소한 같다는 것을 의미한다.The values of the matrix A shown in Equation 8 are variable dimension vectors

The order of the values of the radius of the upper leg 108, the radius of the lower leg 109, the radius of the lower body 107, the radius of the upper body 103, the radius of the neck 102, the head It is a value when it is set as the radius of 101, the radius of the upper arm part 104, the radius of the lower arm part 105, the radius of the hand part 106, and the height of the foot part 106. FIG. As can be seen in the first row of matrix A, the radius of upper leg portion 101 means that it is greater than or at least equal to the radius of lower leg portion 102.

The matrix A shown in Equation 8 is exemplary, and the values of the matrix A may vary when the human body model 100 is considered in more detail.

를 최소화 하는 가변치수벡터

를 찾는 최적화를 수행하는 것이다. 이렇게 하여, 최적의 가변치수벡터

의 값들을 계산해 낼 수 있다. 이렇게 계산된 가변치수를 사용해 인체모델(100)의 각 신체부분의 치수를 수정한다. 수정된 인체모델(100)로부터 계산된 ZMP는 피실험자의 실제 ZMP와 동일하거나 거의 유사하게 된다. 따라서, 수정된 인체모델(100)은 피실험자의 특성에 맞는 인체모델로 보정될 수 있다.In summary, using the optimization algorithm, the objective function of Equation 6 while satisfying the constraints Equation 7 to 9

Variable dimension vector to minimize

Is to perform an optimization to find. In this way, the optimal variable dimension vector

Can be calculated. The calculated dimensions are used to modify the dimensions of each body part of the human body model 100. The ZMP calculated from the modified human model 100 is the same as or almost similar to the actual ZMP of the test subject. Therefore, the modified human body model 100 may be corrected to a human body model suitable for the test subject's characteristics.

The optimization algorithm has been described as an example, and other algorithms may be used to solve optimization problems of variable dimensions.

The present inventors conducted an experiment using a motion capture system and a ground reaction force meter to verify the accuracy of the above-described human body model.

6 illustrates an operation postponed by the test subject in the above experiment. 7 (a) and 7 (b) show the ZMP trajectory in the X direction and the ZMP trajectory in the Y direction and ground reaction force calculated by applying the human body model of the present invention when the motion capture system performs the operations as shown in FIG. 6, respectively. It is a graph showing the actual ZMP trajectory in the X direction and the actual ZMP trajectory in the Y direction obtained from the measuring instrument.

7 (a) and 7 (b), the ZMP trajectory of the human body model of the present invention is indicated by a solid line, and the actual ZMP trajectory of the test subject obtained from the ground reaction force meters 21 and 22 is indicated by a dashed line. . 7 (a) and 7 (b), it can be seen that the actual ZMP trajectory obtained through the ZMP trajectory and the ground reaction force meter of the human body model of the present invention is almost similar. Although a slight error is seen between the two ZMP trajectories, such an error can be eliminated by further subdividing the human body model 100.

The present invention described above is not limited to the above-described embodiment and the accompanying drawings, and it is common in the field of the present invention that various substitutions, modifications, and changes can be made without departing from the technical spirit of the present invention. It will be evident to those who have knowledge of.

The present invention proposes a simple human model capable of analyzing the kinematic movement of human motion, and proposes a method for determining the size of each body part of the human body model. According to the method of generating a human body model of the present invention, it is possible to easily analyze the kinematic motion of a human, and simply obtain a COM or ZMP trajectory used by the humanoid robot to mimic a human kinematic motion.

1 is a flowchart illustrating a method of generating a human body model of the present invention.

2 shows an example of a human body model employed in the method for generating a human body model of the present invention.

3 shows the motion acquisition of the subject using the motion capture system and the ground reaction force meter.

Figure 4 shows an example of applying the human body motion obtained in the motion capture system to the human body model of the present invention.

5 illustrates the forces and moments measured by ground reaction measurements.

6 illustrates an operation performed by the test subject to verify the accuracy of the human body model according to the present invention.

7 (a) and 7 (b) show the ZMP trajectory in the X direction and the ZMP trajectory in the Y direction and ground reaction force calculated using the human body model of the present invention, respectively, when the motion capture system performs the operations as shown in FIG. It is a graph showing the actual ZMP trajectory in the X direction and the actual ZMP trajectory in the Y direction obtained from the measuring device.

Claims (11)

It is a method of creating a human body model composed of a plurality of figures, Acquiring a plurality of position information of the test subject's body and the ground reaction force of the test subject at predetermined time intervals during the test subject's operation; Applying the anatomical model to one of the subject's motions and determining the dimensions of each figure using the position information; Calculating a ZMP of the human body model; Calculating an actual ZMP of the test subject using the ground reaction force; And Correcting the anatomical model so that an error between the ZMP of the anatomical model and the actual ZMP of the test subject is minimized within a constraint. Human body model generation method comprising a. The method of claim 1, Determining the size of the figure includes the step of determining the fixed and variable dimensions of the figure, The human body model correcting step includes optimizing the variable dimension such that the error is minimized within the constraint, and correcting the size of the figure using the optimized variable dimension. How to create a human body model. The method of claim 2, The variable dimension optimization step includes calculating the error, checking whether the variable dimension used in the error calculation satisfies the constraint, and if the constraint condition is satisfied, the variable dimension used in the error calculation; Replacing with an optimized variable dimension How to create a human body model. The method of claim 3, The constraint is a condition in which the variable dimension of any of the figures of the human body model is larger or smaller than the variable dimension of the other of the figures. How to create a human body model. The method of claim 4, wherein The figure is any one of a sphere, a cylinder and a cuboid, The fixed dimension is the length of the cylinder, the horizontal and vertical length of the rectangular parallelepiped, The variable dimension is the radius of the sphere, the radius of the cylinder and the height of the cuboid How to create a human body model. The method of claim 1, The position information and ground reaction acquiring step are performed in a motion capture system. How to create a human body model. The method of claim 6, The ground reaction acquiring step may include installing a ground reaction meter in the motion capture system and obtaining ground reaction force from the ground reaction meter. How to create a human body model. Preparing a human body model composed of a plurality of figures; Acquiring a plurality of pieces of position information of the subject's body and the ground reaction force from the ground reaction force measurer for each frame while the subject is operating on the ground reaction force meter in the motion capture system; Applying the human body model to the test subject's motion every frame and determining fixed and variable dimensions of each of the figures using the position information; Calculating a ZMP of the anatomical model for each frame; Calculating the actual ZMP of the test subject by using the ground reaction force for each frame; Optimizing the variable dimension such that the sum of errors between the ZMP of the human model and the actual ZMP of the test subject is minimized within a constraint over the total number of frames; And Correcting the anatomical model using the optimized variable dimensions Human body model generation method comprising a. The method of claim 8, The step of optimizing the variable dimension includes calculating the sum of the errors, checking whether the variable dimension used in calculating the sum of the errors satisfies the constraint, and calculating the sum of the errors if the constraint is satisfied. Replacing the used variable dimension with the optimized variable dimension. How to create a human body model. The method of claim 9, The constraint is a condition in which the variable dimension of any of the figures of the human body model is larger or smaller than the variable dimension of the other of the figures. How to create a human body model. The method of claim 10, The figure is any one of a sphere, a cylinder and a cuboid, The fixed dimension is the length of the cylinder and the horizontal and vertical length of the rectangular parallelepiped, The variable dimension is the radius of the sphere, the radius of the cylinder and the height of the cuboid How to create a human body model.

Images