KR100813693B1 - Method for controlling arm movement of robot and robot thereof - Google Patents

Abstract

A method for controlling arm movement of a robot and a robot thereof are provided to generate motion of a predetermined arm of the robot by applying a lifting angle for an elbow using operation database of a human arm. A method for controlling arm movement of a robot comprises the steps of: obtaining(S102) an angle for the ground surface with the wrist position from the shoulder of human and the normal direction of the palm to reproduce arm operation of human; calculating(S103) lifting angle for the elbow through an angle for the obtained wrinkle position and the normal direction of the palm with the ground surface; calculating(S104) each joint value of the robot by using the lifting angle of the elbow, the angle for the wrinkle position and the normal direction of the palm with the ground surface; and controlling(S105) the joint of the robot by using the calculated each joint value.

Description

METHOD FOR CONTROLLING ARM MOVEMENT OF ROBOT AND ROBOT THEREOF}

1 is a flowchart illustrating a method for controlling arm motion of a robot according to a first exemplary embodiment of the present invention.

2 is a block diagram showing a robot control apparatus according to a second embodiment of the present invention.

3 is a block diagram showing a robot control apparatus according to a third embodiment of the present invention.

4 is an exemplary diagram showing an experiment for building a database for various arm motions in order to express the characteristics of human arm motion.

5 is a diagram showing elbow lift angles for expressing features of human arm movement.

6 is a view showing the coordinate system of the arm of a human and a robot.

7 is an illustration of a robot with an arm of six degrees of freedom.

8 is a view showing the coordinate system of the left arm of the robot.

** Major Code Description **

200: robot control apparatus according to the first embodiment of the present invention

210: network interface 220: storage device

230: processor

300: a robot control apparatus according to a second embodiment of the present invention

310: network interface 320: storage device

330: processor

The present invention relates to a method for controlling a robot's arm motion, and more particularly, to a method for controlling a humanoid robot to reproduce motion similar to a human's arm motion and a robot.

Humanoid robots aimed at providing useful services have recently been developed and released. In order for the humanoid robots thus developed to provide useful services to humans, humanoid robots must be able to operate like humans.

In order to enable a humanoid robot to operate like a human, recently, a method of measuring human motion and adjusting and applying the measured data to a humanoid robot has been used. Here, an optical motion capture system is frequently used to measure a human motion, and the optical motion capture system allows a marker to be attached to a site representing a human body motion, that is, a joint, and at a time of the marker. It is to measure the trajectory according to.

This approach to measuring and applying human behavior has been tried by several researchers. Among them, ChangHwan Kim, Doik Kim, and Yonghwan Oh, "Solving an inverse kinematics problem for a humanoid robots imitation of human motions using optimization," in Second International Conference on Informatics in Control, Automation and Robotics, September 2005, pp 85–92, Barcelona, Spain) proposed a method of converting marker data attached to a human arm into an action that can be applied to the arm of a humanoid robot using an optimization technique. This study not only allows the humanoid robot to mimic the entire human arm motion by imitating the position and direction of the human hand and the direction of the upper arm, but also considers the operating range and speed limits of the arm joint motor of the humanoid robot. It was. However, this method could not produce a new, random human behavior.

Pollard (Nancy S. Pollard, Jessica K. Hodgins, Marcia J. Riley, and Christopher G. Atkeson, "Adapting human motion for the control of a humanoid robot," in IEEE International Conference on Robotics and Automation, May 2002, 2, pp. 1390–1397, Washington, DC, USA) have developed a method that enables the application of captured human motion to an upper body-oriented humanoid robot. The robot mimics human motion by minimizing the difference between the captured human upper body motion and the humanoid robot motion to mimic it.

Nakaoka (Shinichiro Nakaoka, Atsushi Nakazawa, Kazuhito Yokoi, Hirohisa Hirukawa, and Katsushi Ikeuchi, "Generating whole body motions for a biped humanoid robot from captured human dances," in International Conference on Robotics and Automation, September 2003, pp. 3905 3910, Tapei, Taiwan.) Proposed a systematic procedure and an imitation algorithm to capture traditional Japanese dances with a motion capture system and to mimic them by a humanoid robot (model HRP-1S). Such a method extracts only a few of the main movements from the captured traditional dance, and constitutes some representative movements. And the joint displacement of the humanoid robot which can mimic each representative motion was calculated | required. Subsequently, the entire dance motion was continuously expressed using the joint displacement of the robot for each representative motion. However, this method has a problem in that the body motion has to be modified in consideration of the zero moment point (ZMP) in order to satisfy the dynamic stability of the robot in the process of calculating the displacement of the joint.

The results of the bibliographic studies so far have been on methods of direct imitation for a given human behavior. However, these studies are inadequate for generating any of a variety of humanistic behaviors from the human motion database.

Meanwhile, Asfour and Dillmann (T. Asfour and R. Dillmann, "Human-like motion of a humanoid robot arm based on a closed-form solution of the inverse kinematics problem," in IEEE / RSJ International Conference on Intelligent Robots and Systems, October 2003, vol. 2, pp. 1407-1412., Proposed an alternative approach for generating human-like arm movements. Another proposed approach is Soechting and Flanders ((a) JF Soechting and M. Flanders, "Errors in pointing are due to approximations in targets in sensorimotor transformations," in Journal of Neurophysiology, October 1989, vol. 62, pp. 595-608; (b) JF Soechting and M. Flanders, "Sensorimotor representations for pointing to targets in three-dimensional space," in Journal of Neurophysiology, October 1989, vol. 62, pp. 582-594. We use a model that modifies the characteristics of the arm motion. That is, an arm motion mathematical model composed of four parameters consisting of terms of hand coordinates with respect to a coordinate system defined on a human shoulder was used to provide an approximated motion for a given target arm motion. However, this method also has a large error in the target position, phase and the actual robot hand. In addition, the four defined parameters do not have a special physical meaning, and thus did not greatly assist in understanding human behavior.

The previous findings suggest that not only mimicking one human motion, but also the development of a new method that allows humanoid robots to generate human-like motion in real time at any moment.

Accordingly, the present invention proposes a method of characterizing human arm motion from a human arm motion database measured several times using a motion capture device, by which a humanoid robot can arbitrarily generate human-like arm motion at any time. The purpose is to make sure.

In order to achieve the above object, the present invention grasps the human arm movement through the "Elbow Elevation Angle" proposed in the present invention, and the robot uses the elbow lift angle to control the arbitrary arm of the human being. Allows you to reproduce the behavior.

In order to achieve the above object, the present invention includes the steps of acquiring the position of the wrist from the human shoulder and the angle between the normal direction of the palm and the ground to reproduce human arm motion; Calculating an elbow lift angle through the obtained position of the wrist and the angle formed between the normal direction of the palm and the ground; Calculating each joint value of the robot by using the position of the wrist, the angle between the normal direction of the palm and the ground, and the elbow lift angle; Using the calculated each joint value, it provides a method for controlling the arm motion of the robot comprising the step of controlling the joint of the robot.

Example

Hereinafter, with reference to the accompanying drawings an embodiment according to the present invention will be described in detail.

1 is a flowchart illustrating a method for controlling arm motion of a robot according to a first embodiment of the present invention, and FIG. 4 is an exemplary view showing an experiment for constructing a database for various arm motions in order to express characteristics of human arm motion. 5 is a view showing the elbow lift angle for expressing the characteristics of the human arm movement. And, Figure 6 is a view showing the coordinate system of the arm of the human and humanoid robot, Figure 7 is an illustration of a humanoid robot having an arm of six degrees of freedom, Figure 8 is a view showing the coordinate system of the left arm of the humanoid robot.

Hereinafter, with reference to FIG. 1, FIGS. 4 to 8 will be described together.

As can be seen with reference to Figure 1, the arm motion control method of the robot according to the present invention is characterized by controlling the arm motion of the robot using the elbow lifting angle. Specifically, it is as follows.

First, a trajectory of capturing a human arm motion is obtained (S101). As can be seen from FIG. 4, this can be obtained from a database in which various human arm movements are constructed using an optical motion capture system. At this time, the human arm motion is captured through a marker attached to the human arm and built into a database.

와

, 그리고 구좌표계 원점에서 손목까지의 거리

로 표현될 수 있다. 그리고, 상기 손바닥의 법선 방향과 지면 간의 각도는 손바닥 평면의 법선 벡터를 먼저 구한 후, 상기 법선벡터가 지면과 이루는 각

으로 표현될 수 있다.Next, the wrist position from the human shoulder is obtained through the trajectory. Then, the angle between the palm normal direction and the ground is obtained through the trajectory (S102). At this time, the position of the wrist, as can be seen through Figure 6, two phase angles of the spherical coordinate system defined in the shoulder

Wow

, And the distance from the spherical coordinates origin to the wrist

It can be expressed as. The angle between the palm normal direction and the ground is obtained by first obtaining a normal vector of the palm plane, and then forming an angle between the palm vector and the ground.

It can be expressed as.

Subsequently, the elbow lift angle is calculated based on the position of the wrist and the angle between the normal direction of the palm and the ground. At this time, the elbow lift angle is between the surface perpendicular to the ground (surface consisting of blue lines in FIG. 5) and the wrist, elbow and shoulder (surface composed of red lines in FIG. 5) as shown in FIG. Can be expressed in degrees. At this time, the elbow lift angle is defined as "0" when the arm posture is perpendicular to the ground, that is, when the blue plane and the red plane overlap in FIG. 5.

값이다. 이때, 로봇의 각 관절은 모터로 구현될 수 있으므로, 상기 관절 값은 상기 모터의 회전각 변이일 수 있다. 한편, 상기 로봇은 또한 손목이 1개의 자유도를 갖도록 1개의 관절을 추가적으로 포함할 수 있다. 그러나, 손목의 관절은 전체적인 팔 움직임과 큰 개연성이 없으므로, 임의적으로 지정될 수 있다.Next, each joint value of the robot is calculated using the position of the wrist, the normal direction of the palm, and the elbow lift angle (S104). In this case, the arm of the robot may have at least five degrees of freedom, as shown in FIGS. It may consist of two joints. Therefore, each joint value to be calculated is a value for at least five joints, that is, as shown in FIG.

Value. In this case, since each joint of the robot may be implemented by a motor, the joint value may be a rotation angle variation of the motor. On the other hand, the robot may further include one joint so that the wrist has one degree of freedom. However, the joints of the wrist may be arbitrarily designated since there is no great probability of arm movement and overall movement.

Finally, the joints of the robot are controlled using the respective joint values (S105). In this case, since each joint of the robot may be implemented by a motor as described above, the joints of the robot may be controlled by controlling the rotation angle of the motor.

On the other hand, Figure 2 is a block diagram showing a robot control apparatus according to a second embodiment of the present invention. As can be seen with reference to Figure 2, the robot control apparatus according to the second embodiment of the present invention is built in the robot, to calculate the joint value using the above-described elbow lift angle, and outputs a joint control signal It is characterized by.

In detail, the robot control apparatus 200 includes a network interface 210, a storage device 220, and a processor 220.

,

,

) 및 손바닥의 법선 방향과 지면이 이루는 각도(

)를 외부로 부터 수신한다. The network interface 210 is a wrist position from the human shoulder described above (

,

,

) And the angle between the palm's normal and the ground (

) Is received from the outside.

,

,

) 및 손바닥의 법선 방향과 지면이 이루는 각도(

)를 저장한다.The storage device 220 is the wrist position (received through the network interface 210)

,

,

) And the angle between the palm's normal and the ground (

Save).

,

,

) 및 손바닥의 법선 방향과 지면이 이루는 각도(

)를 이용하여, 팔꿈치 들림각을 산출하고, 그리고 상기 손목의 위치, 상기 손바닥의 법선 방향과 지면이 이루는 각도, 및 팔꿈치 들림각을 이용하여 로봇의 각 관절 값을 산출한 후, 상기 로봇의 관절 제어 신호를 출력한다. The processor 230 stores the wrist position in the storage device 220

,

,

) And the angle between the palm's normal and the ground (

Calculate the elbow lift angle, and calculate the joint values of the robot using the position of the wrist, the angle between the normal direction of the palm and the ground, and the elbow lift angle. Output a control signal.

On the other hand, Figure 3 is a block diagram showing a robot control apparatus according to a third embodiment of the present invention.

As can be seen with reference to Figure 3, the robot control apparatus 300 according to the third embodiment of the present invention to control the robot remotely, calculates the joint value by using the elbow lift angle described above, It is characterized by outputting a control signal.

In detail, the robot control apparatus 300 includes a network interface 310, a storage device 320, and a processor 330.

The storage device 320 stores a trajectory for capturing a human arm motion.

,

,

) 및 손바닥의 법선 방향과 지면이 이루는 각도(

)를 획득하고, 상기 획득된 손목 위치 및 손바닥의 법선 방향과 지면이 이루는 각도를 이용하여, 팔꿈치 들림각을 산출한다. 그리고, 상기 프로세서(330)는 상기 손목의 위치, 상기 손바닥의 법선 방향과 지면이 이루는 각도, 및 팔꿈치 들림각을 이용하여 로봇의 각 관절 값을 산출한 후, 상기 로봇의 관절 제어 신호를 출력한다.The processor 330 is a wrist position from the human shoulder through the trajectory in the storage device 320 (

,

,

) And the angle between the palm's normal and the ground (

), And the elbow lift angle is calculated using the obtained wrist position and the angle between the normal direction of the palm and the ground. The processor 330 calculates each joint value of the robot using the position of the wrist, the angle between the normal direction of the palm and the ground, and the elbow lift angle, and then outputs a joint control signal of the robot. .

The network interface 310 transmits a joint control signal of the robot to the robot.

So far, the arm motion control method and robot control apparatus of the robot which concern on this invention were demonstrated.

Hereinafter, the basis for defining the human arm motion through the above-described motion capturing through the elbow lift angle will be described, and the elbow lift angle will be described in more detail through an equation. In addition, obtaining the joint value of the robot through the elbow lift angle will be described in more detail with reference to the following equation.

Motion capturing

In order to obtain various arm motions of a human, an experiment using a motion capture system was performed as shown in FIG. 4. At this time, the performer during the experiment was free to take the position of the arm. The area where the arm of the performer could reach was divided equally into six planes perpendicular to the ground, and the arm motion experiment was performed several times as shown in FIG. The actor experimented with five different diameter circles for five seconds, and repeated this experiment several times, at least three times, with the palms changing from side to side.

elbow Lifting angle ( Elbow Elevation Angle ) Justice

After constructing the human arm motion database through the above experiment, the hand and arm postures, which constitute various motions, can be observed by the overall position of the arm due to the spatial position of the wrist, the phase of the hand, and the elbow posture. It can be seen that the decision.

Among these, the spatial position of the wrist can be obtained by using the trajectory of the marker attached to the human hand. The phase of the hand can be obtained using the normal direction of the palm.

In addition, the elbow posture is an angle between a surface perpendicular to the ground (a surface composed of blue lines in FIG. 5) and a wrist, elbow, and shoulder connecting surface (a surface composed of red lines in FIG. 5) as shown in FIG. 5. It can be expressed as The angle between them is defined as "Elbow Elevation Angle". In this case, as described above, when the arm posture is perpendicular to the ground, that is, when the blue plane and the red plane overlap in FIG. 5, the elbow lift angle is defined as “0”.

After acquiring the spatial position of the wrist, the phase of the hand, and the elbow lift angle through a database, when closely observed, it can be seen that the elbow lift angle alone represents the wrist position and the hand phase. have.

Therefore, by using the elbow lift angle to control the robot's arm, it is possible to control the robot to operate naturally like a human.

elbow Lifting angle  Characteristic equation

Mathematical modeling is introduced to calculate the elbow lift angle through the wrist position and palm orientation that can be seen from a database constructed as in FIG. 4.

In this case, the mathematical modeling is the response surface method (Myers, Raymond H. and DC Montgomery (2002), Response Surface Methodology, John Wiley & Sons, Inc., 605 3th Avenue, New York, NY 10158-0012) Can be achieved through This response surface technique is one technique that represents the relationship between an adjustable input value and its output for a system.

Therefore, in the present invention, the position of the wrist and the normal direction of the palm are used as the input value through the reaction function, and the elbow lift angle is calculated as the output.

(1)

(One)

는 실험 결과이고

은 실험결과

의 미지의 반응함수이다. 그리고

은 인간의 실제 팔꿈치 들림각과 반응함수를 통해 산출되는 팔꿈치 들림각 간의 오차를 나타낸다.

은 조절 가능한 입력 변수 벡터로, 여기서는 손목의 위치 벡터 성분과 손바닥의 법선 벡터가 지면과 이루는 각으로 이루어진다. 이때, 손목의 위치는 마커 데이터를 이용하여 어께에 정의한 구좌표계의 두 위상각

와

, 그리고 구좌표계 원점에서 손목까지의 거리

로 표현될 수 있다. 또한, 손바닥의 법선 벡터는 손에 부착된 세 개의 마커 데이터를 이용하여 손바닥 평면의 법선벡터를 구하고 이 법선벡터가 지면과 이루는 각

를

의 나머지 성분으로 정의함으로써, 얻어질 수 있다: 즉,

here

Is the result of the experiment

Results

This is an unknown reaction function. And

Represents the error between the actual elbow lift angle and the elbow lift angle calculated by the response function.

Is an adjustable input variable vector, where the position vector component of the wrist and the normal vector of the palm are the angles that make up the ground. At this time, the position of the wrist is two phase angles of the spherical coordinate system defined in the shoulder using marker data.

Wow

, And the distance from the spherical coordinates origin to the wrist

It can be expressed as. In addition, the palm normal vector uses three marker data attached to the hand to obtain the normal vector of the palm plane, and the angle formed by the normal vector to the ground.

To

By defining the remaining components of, it can be obtained:

은 다음과 같은 근사함수를 이용하여 정의될 수 있다.The reaction function

Can be defined using the following approximation function:

(2)

(2)

는 반응함수의 항의 수를 나타내고,

은 형상함수(혹은 기본함수)라고 부르는 근사식의 형태를 결정 짖는 함수를 나타낸다. 또한,

는 각 형상함수의 미지의 계수이다. 이들 계수들은 주어진 실험 결과를 이용하여 결정될 수 있다. here

Represents the number of terms in the reaction function,

Represents a function that determines the form of the approximation called a shape function (or basic function). Also,

Is the unknown coefficient of each shape function. These coefficients can be determined using the given experimental results.

는 수학식(1)과 수학식(2)에 따라 다음과 같이 나타낼 수 있다.As described above, the error when each result is given in the experiment of several experiments

Can be expressed as follows according to equations (1) and (2).

(3)

(3)

은 j번째 실험에서 얻을 수 있는 팔꿈치 들림각의 실험치이고,

은 이 j번째 실험치와 그에 대응하는 반응함수

와의 오차를 의미한다.

는 j번째 실험의 입력 변수 벡터이다. 수학식(3)을 행렬 형식으로 다시 정리하면,Where N is the number of experiments.

Is the experimental elbow lift angle obtained from the jth experiment,

Is the jth experimental value and the corresponding response function

It means the error between and.

Is the input variable vector of the j th experiment. If we rearrange Equation (3) in matrix form,

(4)

(4)

를 요소로 갖는 N×NB 행렬이다. 미지 벡터

는 오차 벡터

의 크기의 제곱, 즉,

을 최소화함으로써 다음과 같이 결정된다.Get Where X is

N × NB matrix with. Unknown vector

Error vector

Square of the magnitude of,

By minimizing this, it is determined as follows.

(5)

(5)

Thus, the response function of equation (2) can be determined.

의 성분을 정규화하는 것이 효과적이다. 각 성분들은 크기에 있어 서로 간에 큰 차이를 가지므로 입력 변수의 영향을 최소화하고 반응함수의 근사화에 따른 오차를 줄이기 위해 정규화할 필요가 있다. On the other hand, in applying the response surface method (Response Surface Method) input variable vector

It is effective to normalize the components of. Since each component has a large difference in size, it needs to be normalized to minimize the influence of the input variables and reduce the error due to the approximation of the response function.

In the present invention, the size of each component is normalized to have a dimensionless size of 2 or less. Since this normalization process is simply a multiplication by a constant, a detailed description thereof will be omitted. In addition, there is no particular limitation on the size. The normalized input variable vector is

(6)

(6)

Here, the '-' at the head of every input variable indicates that it is a normalized variable.

는 수학식(6)에 따라서 정규화한 후, 수학식(5)를 사용하여 미지 계수벡터

를 얻는다. The quadratic equation is used as the shape function of Equation (2), which is the most used function in using the response surface method. It is also possible to use a shape function other than the quadratic equation. Actually, the input variable vector obtained in the experiments from Equation (2) to Equation (5).

Is normalized according to Equation (6), and then unknown coefficient vector using Equation (5).

Get

Finally, the response function for elbow lift angle is defined using four normalized input variables.

(7)

(7)

은 정규화된 입력변수로 구성된 팔꿈치 들림각의 반응함수이고, 입력변수는

이다. here

Is the response function of the elbow lift angle composed of normalized input variables.

to be.

를 구하여 인간의 팔 움직임을 특징짓는 특성방정식인 팔꿈치 들림각 식을 완성할 수 있다.Unknown coefficient by applying Equation (5) to Equation (7)

We can complete the elbow lift angle equation, a characteristic equation that characterizes human arm movement.

,

,

)와 손바닥의 방향(

)이 주어지면 이에 대응하는 로봇에 적합한 인간다운 팔꿈치의 들림각을 수학식(7)로부터 얻을 수 있다. 물론 이때, 입력 변수 값은 수학식(6)에 설명한 대로 정규화시켜야한다.Once the elbow lift angle expression is completed, it can be applied directly to humanoid robots. That is, the wrist position of the humanoid robot

,

,

) And palm orientation (

), The lifting angle of the human-like elbow suitable for the corresponding robot can be obtained from Equation (7). Of course, at this time, the input variable value should be normalized as described in Equation (6).

Inverse kinematics  Human elbow Lifting angle  avatar

In order for the humanoid robot to actually generate and operate an elbow lift angle similar to that obtained in Equation (7), inverse kinematics should be applied to properly move the other joints of the arm except the hand.

As can be seen with reference to FIGS. 7 and 8, a robotic arm with six degrees of freedom is contemplated in the present invention. That is, they have three degrees of freedom on their shoulders, two degrees of freedom on their elbows, and one degree of freedom on their wrists.

To solve the inverse kinematics for these degrees of freedom, six conditional expressions are required. Typically, three conditions at the hand position and three conditions at the hand phase are used to solve the inverse kinematics problem. However, this method does not guarantee human arm movement.

)을 계산할 수 있는 방법을 설명한다. 다만, 도 8에서의 로봇의 손목굽힘각(Hand Stool angle)(도 8에서의

)은 인간다운 팔 자세에는 큰 영향을 미치지 않아 사용자에 의해 정의하도록 한다.Therefore, in the present invention, given the three-dimensional position coordinates of the hand, the palm direction, and the elbow lift angle calculated from these values, five robot arm joint values (in FIG. 8).

Explain how you can calculate However, the hand stool angle of the robot in FIG. 8 (in FIG.

) Does not significantly affect the human arm posture and is defined by the user.

As mentioned above, six conditions are required to apply inverse kinematics to a six degree of freedom robotic arm.

,

,

) 3개 조건과 손바닥의 방향과 지면이 이루는 각(

)의 1개 조건, 그리고 수학식8의 팔꿈치 들림각을 포함하여 일단 5개의 조건식을 얻을 수 있다. 다만, 마지막 조건인 손목굽힘각(도 8에서의

)은 사용자에 의해 정의된다. In the present invention, these six conditions are input values for the arm posture that the robot targets, and the wrist position as seen from the spherical coordinate system of the shoulder (

,

,

) 3 conditions and the angle between the direction of the palm and the ground (

Five conditions can be obtained once, including one condition of), and the elbow lift angle of Equation (8). However, the wrist bending angle which is the last condition (in FIG.

) Is defined by the user.

These six conditional formulas solve the inverse kinematics problem based on the geometric approach by following these steps:

계산 (One)

Calculation

,

,

)와 손바닥의 방향각(

)이 주어지면 인간다운 팔꿈치 들림각을 수학식(7)로부터 알 수 있다. 도 8의

은 도 6의 (a)와 (b)에서 알 수 있는 바와 같이 단지 어깨와 손목까지의 거리에 대한 함수로 표현됨을 알 수 있다. 따라서,

은First, wrist position (

,

,

) And palm orientation angle (

) Can be known from Equation (7). Of FIG. 8

As can be seen in (a) and (b) of Figure 6 it can be seen that it is expressed only as a function of the distance to the shoulder and wrist. therefore,

silver

(8)

(8)

와

은 도 6에서와 같이 로봇의 위팔과 아래팔의 길이이고,

은 손목위치의 구좌표계 값이다.to be. here,

Wow

Is the length of the upper and lower arms of the robot, as shown in Figure 6,

Is the spherical coordinates of the wrist position.

과

계산 (2)

and

Calculation

과

은 도 6의 (b)에서 알 수 있는 바와 같이 관절 벡터

에 의해 결정된다. 도 6의 (a)와 (b)에서 구좌표계의 위상각치

,

, 그리고 팔꿈치 들림각

가 "0"의 값을 가지는 경우의

를 벡터

로 정의한다. 이때, 벡터

와 어깨에서 손목으로의 위치 벡터 (도 6(b)에서

)에 의해 구성되는 면은 어깨에 위치한 좌표계의 x-z면과 평행이다(도 6 (b) 참조). 이러한 기하학적인 관계를 이용하여 벡터

의 성분을 다음과 같이 얻는다.Of FIG. 8

and

Is a joint vector as can be seen in FIG.

Determined by Phase angle values of the spherical coordinate system in (a) and (b) of FIG.

,

, And elbow lift angle

Has a value of "0"

Vector

Defined as Vector

Position vector from shoulder to wrist (in Figure 6 (b)

) Is parallel to the xz plane of the coordinate system located on the shoulder (see FIG. 6 (b)). Vector of these geometric relationships

The component of is obtained as follows.

(9)

(9)

은 수학식(7)의 팔꿈치 들림각

과 손목의 구좌표계 값

,

에 의해 얻어진다.vector

Is the elbow lift angle in equation (7)

And spherical coordinates of the wrist

,

Obtained by

(10)

10

,

그리고

은

,

그리고

에 의해 정의되는 오일러(Euler) 각의 회전행렬을 나타낸다. 일단,

을 구한 후에는 도 6(a) 및 (b)에서 알 수 있는 바와 같이 기하학적으로 손쉽게 관절 각

과

을 구할 수 있다.here ,

,

And

silver

,

And

Represents the rotation matrix of the Euler angle defined by. First,

After obtaining, as shown in FIGS. 6 (a) and (b), the joint angles are geometrically easy.

and

Can be obtained.

(11)

(11)

(12)

(12)

,

그리고

은

의 x, y, z 성분을 나타낸다.here

,

And

silver

X, y, and z components are shown.

계산 (3)

Calculation

Wrist position with respect to the Cartesian coordinate system defined on the shoulder is as shown in Figure 6 (b).

(13)

(13)

는 i번째 좌표계에서 j번째 좌표계로 변환시켜주는 변환행렬(Homogeneous Transformation Matrix)이다.

은 어깨로부터 4번째 좌표계에서 본 손목의 위치 벡터이다: 즉,

.here

Is a transformation matrix (Homogeneous Transformation Matrix) that transforms from the i th coordinate system to the j th coordinate system.

Is the position vector of the wrist in the fourth coordinate system from the shoulder:

.

,

,

의 값을 수학식(13)에 대입하면

을 얻을 수 있다.Position vector of wrist and obtained from equation (8), equation (11) and equation (12)

,

,

Substituting the value of into Equation (13)

Can be obtained.

(14)

(14)

here,

(15)

(15)

(16)

계산 (4)

Calculation

를 구하기 위해서는 도 6(b)에서 알 수 있는 바와 같이먼저

를 다음과 같이 계산하여야 한다.Finally,

To obtain, as can be seen in Figure 6 (b) first

Should be calculated as follows.

(17)

(17)

은 팔꿈치에서 손목으로의 위치 벡터와 지면으로부터 위로 향하는 벡터에 의해 형성되는 면의 법선 벡터이다.

은 어깨, 손목 그리고 팔꿈치로 이루어진 면의 법선 벡터이다. 따라서 주어진 손바닥과 지면과의 사이각

와

의 차이가 구하려는

이다. here

Is the normal vector of the face formed by the position vector from the elbow to the wrist and the vector pointing up from the ground.

Is the normal vector of the face of the shoulders, wrists and elbows. Thus the angle between the given palm and the ground

Wow

Want to get the difference

to be.

는 다음의 수학식 19를 통해 구해질 수 있다.In other words,

Can be obtained through Equation 19 below.

(18)

(18)

를 포함하여

의 모든 6개 관절 값을 구할 수 있다. 또한, 인간다운 팔꿈치 들림각이 사용되었으므로 생성된 로봇의 팔 동작은 인간과 유사한 형태를 갖게 된다.The joint values obtained in steps 1 to 4 and given by the user

Including

All six joints can be found. In addition, since a human-like elbow lift angle is used, the generated arm motion of the robot has a shape similar to that of a human.

In the above description of the preferred embodiments of the present invention by way of example, the scope of the present invention is not limited only to these specific embodiments, the present invention is in various forms within the scope of the spirit and claims of the present invention Can be modified, changed, or improved.

As described above, the present invention proposed an elbow lift angle as a mathematical model that can characterize human arm movement using a human arm motion database. This invention can generate any arm movement in real time for any robotic arm having at least 5 degrees of freedom (shoulder 3, elbow 1, wrist 1).

In addition, since the present invention does not affect the movement of the wrist to the target position at all due to the elbow lift angle, there is no error about the target position in the inverse kinematics problem. Therefore, when the arm is moved to grab an object on a shelf or a predetermined point, there is an advantage that it can be actively utilized for generating an arm movement when holding and manipulating an object.

Claims (10)

Wrist position from a human shoulder obtainable via a trajectory capturing human arm motion; And controlling the arm movement of the robot using an elbow lift angle that can be expressed by an angle formed by the normal direction of the palm of the human hand and an angle formed by the trajectory. delete delete Acquiring the position of the wrist from the human shoulder and the angle between the normal direction of the palm and the ground to reproduce human arm motion; Calculating an elbow lift angle through the obtained position of the wrist and the angle formed between the normal direction of the palm and the ground; Calculating each joint value of the robot by using the position of the wrist, the angle between the normal direction of the palm and the ground, and the elbow lift angle; And controlling the joints of the robot by using the calculated values of the joints. The method of claim 4, wherein the obtaining step Obtaining a trajectory for capturing a human arm motion; And obtaining an angle between the wrist position from the human shoulder and the normal direction of the palm and the ground through the trajectory. delete The method of claim 4, wherein calculating each joint value of the robot At least three angles for the shoulder joint of the robot; And calculating at least two or more angles for the elbow joint of the robot. A storage device for storing the position of the wrist from the human shoulder and the angle between the normal direction of the palm and the ground to reproduce human arm motion; The elbow lift angle is calculated using the wrist position in the storage device and the angle formed by the normal direction of the palm and the ground, and the position of the wrist, the angle formed by the normal direction of the palm and the ground, and the elbow lift angle are calculated. And a processor configured to output the joint control signal of the robot after calculating each joint value of the robot using the robot. A storage device for storing the position of the wrist from the human shoulder and the angle between the normal direction of the palm and the ground to reproduce human arm motion; The elbow lift angle is calculated using the wrist position in the storage device and the angle formed by the normal direction of the palm and the ground, and the position of the wrist, the angle formed by the normal direction of the palm and the ground, and the elbow lift angle are calculated. And calculating a joint value of the robot using the processor, and outputting a joint control signal of the robot. A storage device for storing a trajectory for capturing a human arm motion; The angle between the wrist position from the human shoulder and the normal direction of the palm and the ground is obtained through the trajectory within the storage device, and the elbow lift angle is obtained by using the obtained wrist position and the angle between the normal direction of the palm and the ground. And calculating each joint value of the robot using the position of the wrist, the angle formed by the normal direction of the palm and the ground, and the elbow lift angle, and then outputting a joint control signal of the robot; And a network interface for outputting a joint control signal of the robot.

Images