KR101288149B1 - Device and method for controlling the balance of biped walking robots - Google Patents

Abstract

The present invention discloses an apparatus and method for balancing a robot. The present invention provides a walking command receiver for receiving a command for a walking pattern of a robot; A path generation unit configured to generate a nominal angle, which is a time history of a rotation angle at which each joint of the robot should move using a command for the walking pattern; The position of the robot's Zero Moment Point (ZMP) is measured using force / torque data acting on the robot's foot, and the rotation angle of the pelvis and ankle joint of the robot is measured using the ZMP position. A balance controller for generating a correction angle; And a joint controller for controlling the movement of the robot joint using the nominal angle and the correction angle. According to the present invention, the balance control can be precisely performed using only ZMP sensor data measured by the robot. In addition, there is an advantage that the same balance control technique is applied to both the robot's foot support state and the one-foot support state.

Description

DEVICE AND METHOD FOR CONTROLLING THE BALANCE OF BIPED WALKING ROBOTS}

Embodiments of the present invention relate to a balance control device and method of the robot, and more particularly to a balance control device and method for controlling the balance of the robot using the data measured in the force / torque sensor.

In the case of biped walking robots with small legs, such as walking robots, especially humanoid robots, securing stability during walking is an absolute requirement for speeding up walking. For this purpose, a smooth walking pattern is basically used, but it is difficult to have sufficient walking speed alone. In general, the robot is biased toward the left / right side instead of the center due to manufacturing tolerances, and even the walking surface is not a perfect horizontal plane due to the irregularities of the ground.

Accordingly, large robots in motion walking, which are relatively large in stationary or walking, especially inertial forces, lose their balance, safety, and are easily shaken. In this state, they cannot continue walking.

Therefore, in order to improve the walking speed, research on the balance control of the robot is required. Publications Loffler, K., Gienger, M., and Pfeiffer, F., 2003, "Sensor and Control Design of a Dynamically Stable Biped Robot" Proc. of the IEEE Int. Conf. on Robotics and Automation, pp.484 ~ 490, shows how to use multiple sensors based on the inverted pendulum model. However, the method presented in this paper has the disadvantage that the control method is complicated because the body orientation sensor must be used in addition to the force / torque sensor used for the ZMP sensor to control the position of the joint.

Publications Kim, JH and Oh, JH, 2003, "Torque Feedback Control of the Humanoid Platform KHR-1," Proc. of the 3 ^rd IEEE Int Conf. on Humanoid Robots proposes a balance control method using a 3-axis F / T sensor under the robot's ankle. However, this paper is also based on the inverted pendulum model, so the information on the bearing of the body is required. In the proposed balance control method, the balance control algorithms of the robot's foot support and foot support state are different and the control method is complicated. there was.

In order to solve the problems of the prior art as described above, the present invention is to propose a method and apparatus for controlling the balance of the robot using the ZMP sensor data.

Other objects of the present invention may be derived by those skilled in the art through the following examples.

According to a preferred embodiment of the present invention to achieve the above object, a walking command receiving unit for receiving a command for the walking pattern of the robot;

A path generation unit configured to generate a nominal angle, which is a time history of a rotation angle at which each joint of the robot should move using a command for the walking pattern;

The position of the robot's Zero Moment Point (ZMP) is measured using force / torque data acting on the robot's foot, and the rotation angle of the pelvis and ankle joint of the robot is measured using the ZMP position. A balance controller for generating a correction angle; And

And a joint control unit for controlling the movement of the robot joint using the nominal angle and the correction angle.

The path generator may generate the nominal angle using an inverse kinematics model.

The force / torque data may be measured using an F / T sensor installed under the robot ankle.

Correction angles of the pelvis and the ankle joint generated by the balance controller may have the same magnitude and may be opposite in sign to each other.

The joint controller rotates a roll direction joint of the pelvis and ankle of the robot when controlling the left / right direction of the robot, and a pitch of the pelvis and ankle of the robot when controlling the front / rear direction of the robot. You can rotate the directional joint.

The joint control unit may control the joint of the robot by using proportional derivative (PD) control.

According to another embodiment of the invention, the step of receiving a command for the walking pattern of the robot; Generating a nominal angle, which is a time history of a rotation angle at which each joint of the robot should move using the command for the walking pattern; The position of ZMP of the robot is measured by using force / torque data acting on the foot of the robot measured by the robot, and the rotation angle of the pelvis and ankle joint of the robot is determined by using the ZMP position. Generating a correction angle; And controlling the movement of the robot joint using the nominal angle and the correction angle.

According to the present invention, the balance control can be precisely performed using only ZMP sensor data measured by the robot. In addition, there is an advantage that the same balance control technique is applied to both the robot's foot support state and the one-foot support state.

1 is a view showing an example of a balance control device of a biped walking robot according to an embodiment of the present invention.
2 is a view showing an example of a robot standing on two feet according to an embodiment of the present invention.
3 is a view showing the results of an experiment in which the body of the stationary robot according to an embodiment of the present invention is pushed backward and tilted.
4 is a diagram illustrating the trajectory of the ZMP position signal when the robot 109 stops after moving forward 8 steps in the stop state according to an embodiment of the present invention.
5 is a view showing the movement of the robot according to an embodiment of the present invention.
Figure 6 is a flow chart showing the overall flow for the balance control method of the robot according to an embodiment of the present invention.

While the invention is susceptible to various modifications and alternative forms, specific embodiments thereof are shown by way of example in the drawings and will herein be described in detail. It should be understood, however, that the invention is not intended to be limited to the particular embodiments, but includes all modifications, equivalents, and alternatives falling within the spirit and scope of the invention. Like reference numerals are used for like elements in describing each drawing.

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

1 is a view showing an example of a balance control device of a biped walking robot according to an embodiment of the present invention.

Referring to FIG. 1, the balance control apparatus 100 may include a walking command receiver 101, a path generator 103, a balance controller 105, and a joint controller 107.

The walking command receiving unit 101 receives a command for the movement of the robot 109 from a PC or a remote controller. The command for movement may include a command for a walking pattern of the robot 109 such as moving forward / backward / left / right.

을 생성한다.The path generation unit 103 uses a proper inverse kinematics model according to the walking pattern command received by the walking command receiving unit 101, and is a nominal angle that is a time history of the rotation angle at which each joint of the robot 109 should move.

.

Inverse kinematics refers to a process of determining a rotation angle of a joint corresponding to a position and an attitude of a distal end of a robot manipulator.

It will be apparent to those skilled in the art that there may be a variety of inverse kinematic models, and that various inverse kinematic models may be used to determine the angle of rotation at which the joint should move. Accordingly, detailed description of the inverse kinematics model will be omitted.

를 생성한다.When the robot 109 is biased to one side due to the disturbance, the balance controller 105 uses the data measured by the F / T (force / torque) sensor installed under the robot 109 ankle to determine the actual position of the ZMP. Correction angle, which is the rotational angle of the pelvis and ankle joint for grasping and balance control

.

는 로봇(109)의 균형 제어를 위해 로봇(109)의 발목 각을 변화시켜 로봇(109)의 몸통을 기울어진 방향으로 더 이동시키기 위한 각을 의미한다. Where the correction angle

Denotes an angle for further moving the torso of the robot 109 in an inclined direction by changing the ankle angle of the robot 109 to control the balance of the robot 109.

를 통해 로봇(109)의 몸통은 지면에 대하여 항상 평행하도록 제어된다. 또한, 좌/우 방향 제어 시에는 로봇(109)의 골반과 발목의 롤(roll) 방향 관절을, 전/후 방향 제어 시에는 로봇(109)의 골반과 발목의 피치(pitch) 방향 관절을 보정각

만큼 회전 시킨다. 이때, 로봇(109)의 골반과 발목에 적용되는 보정각

는 크기는 같으나 부호가 반대가 되도록 출력된다.At this time, the correction angle output from the balance controller 105

Through the body of the robot 109 is controlled to be always parallel to the ground. In addition, in the left / right direction control, the pelvis and ankle joints of the pelvis of the robot 109 are corrected, and in the forward / rear direction control, the pitch direction joints of the pelvis and ankle of the robot 109 are corrected. bracket

Rotate as much as At this time, the correction angle applied to the pelvis and ankle of the robot 109

Outputs the same size but with the opposite sign.

와 2개의 수평 방향의 토크

,

이다. According to an embodiment of the present invention, the F / T (force / torque) sensor may include a kind of three-axis load cell sensor and a microprocessor mounted under the ankle of the robot 109. The three-axis load cell sensor senses force / torque data applied to each foot, and the microprocessor uses the sensed data to measure the position of the zero moment point (ZMP). Here, ZMP means a point where the sum of all moments acting on the robot 109 becomes zero. The measured physical quantity measured by the load cell sensor is the vertical force

And two horizontal torques

,

to be.

Although the load cell sensor has been described above, it will be apparent to those skilled in the art that various sensors capable of measuring force / torque may be applied to the present invention.

The joint controller 107 controls the movement of the robot 109 joint. According to an embodiment of the present invention, the joint controller 107 may control the joint of the robot 109 by using proportional derivative (PD) control, but is not limited thereto.

및 균형 제어부(105)에서 생성된 보정각

가 합산된 관절의 기준각

이 입력된다. 관절 제어부(107)는 기준각

을 이용하여 로봇(109) 관절의 움직임을 제어하여 로봇(109)의 균형 제어를 수행한다The joint controller 107 has a nominal angle generated by the path generator 103.

And the correction angle generated by the balance controller 105

Reference angle of the summation

Is input. Joint control unit 107 is a reference angle

By controlling the movement of the robot 109 joint using the balance control of the robot 109

2 is a view showing an example of a robot standing on two feet according to an embodiment of the present invention.

Referring to FIG. 2, the robot 109 in the yz plane can be simplified to a single mass that performs two-dimensional translational motion. This is because the body always causes translational movement in the y or z direction in the case of a general walking pattern in which the robot 109 body stops or moves with the body upright with respect to the ground.

은 yz 좌표계의 원점이면서 안정영역의 중앙인 O점으로부터 무게중심 G점까지의 거리,

는 수평방향의 외란 힘,

는 무게중심의 수평 좌표,

는 몸통 운동에 의하여 선분 OG가 수직선과 이루게 되는 각도로서 보다 상세하게는 공칭각

와 외란 작용 시에 로봇(109) 링크, 관절, F/T 센서 등의 유연성에 기인하여 수동적으로 발생하는 경사각

, 균형제어를 위하여 발목과 골반의 관절이 회전하는 각인 보정각

의 합이다. 다만, 설명의 편의를 위하여 도 2에서의

은 0인 것으로 가정한다.In Figure 2 m is the magnitude of the robot 109 mass,

Is the distance from the point O to the center of gravity G to the origin of the yz coordinate system

Is the horizontal disturbance force,

Is the horizontal coordinate of the center of gravity,

Is the angle that the line segment OG makes with the vertical line by the body movement.

Tilt angle generated manually due to flexibility of robot 109 link, joint, F / T sensor, etc.

Angle of rotation of ankle and pelvis joint for balance control

. However, for convenience of description

Is assumed to be zero.

가 작다고 가정하면 하기의 수학식 1과 같은 경사각에 관한 선형 운동 방정식을 얻을 수 있다.Apply Newton's law to this model of robot 109

Assuming that is small, a linear equation of motion regarding an inclination angle as shown in Equation 1 below can be obtained.

[ Equation 1 ]

는 관성력으로서 하기의 수학식 2와 같이 표현될 수 있다.Here, k means the horizontal rigidity due to the flexibility of the robot link, joint, F / T sensor. c denotes a horizontal component of attenuation shown in PD joint control or the like.

May be expressed as Equation 2 below as an inertia force.

& Quot; (2 ) & quot ;

는 관성력으로서 F/T 센서에 의하여 감지된 ZMP 값을 이용하여 생성된 보정각을 이용하여 간접적으로 생성된다. 보정각은 하기의 수학식 3과 같이 표현될 수 있다.here,

Is indirectly generated using the correction angle generated using the ZMP value sensed by the F / T sensor as the inertial force. The correction angle may be expressed as in Equation 3 below.

& Quot; (3 ) & quot ;

는 로봇의 균형제어를 위한 보정각,

은 목표 감쇠율, m은 로봇의 질량, k는 수평 방향 강성, c는 수평 방향 감쇠,

은 로봇 무게 중심의 높이, g는 중력가속도를 의미한다.here,

Is the correction angle for the balance control of the robot,

Is the target damping rate, m is the mass of the robot, k is the horizontal stiffness, c is the horizontal damping,

Is the height of the robot's center of gravity, and g is the acceleration of gravity.

,

는 제어 이득 상수로서, 순수한 감쇠 제어를 위해 설정되며

가

보다 크도록 설정된다.Also,

,

Is the control gain constant, which is set for pure attenuation control

end

Is set to be greater than.

는 y축 방향의 ZMP의 위치를 의미하며, 하기의 수학식 4와 같이 표현될 수 있다.Also,

Denotes the position of ZMP in the y-axis direction and may be expressed as in Equation 4 below.

[ Equation 4 ]

성분에 관한 연립 미분 방정식을 얻을 수 있으며,

,

라고 정의할 때

를 충분히 크게 하면

에 관한 식은 하기의 수학식 5와 같이 정리된다.Combining Equations 1 to 4, the inclination angle

You can get simultaneous differential equations about the components,

,

When you define

If you make it big enough

Is related to Equation 5 below.

[ Equation 5 ]

,

의 설정 시 균형 제어가 댐핑 제어로 수렴할 수 있도록

를 크게 하는 동시에 폐루프 감쇠 값이 적절하도록

를 결정하는 것이 바람직하다.Thus, the control gain constant

,

So that balance control can converge to damping control

To increase the closed loop attenuation

It is desirable to determine.

According to an embodiment of the present invention, the correction angle generation of the balance controller 105 may be generated by applying the above equation even in a state where the robot 109 is in one-foot support.

이어서

가 0 대신에 지지 발의 중심 위치로 수렴하는 차이 외에는 동일한 균형 제어 효과를 얻을 수 있다. 또한, 로봇(109)의 전후 반향 제어에 대해서도 동일한 로봇의 균형제어를 위한 보정각 생성방법을 동일하게 적용할 수 있다.Nominal angle when robot 109 is in one-foot support

next

The same balance control effect can be obtained except that the difference converges to the center position of the support foot instead of zero. In addition, the correction angle generation method for the balance control of the same robot can be applied to the front and rear echo control of the robot 109 in the same manner.

3 is a view showing the results of an experiment in which the body of the stationary robot according to an embodiment of the present invention is pushed backward and tilted.

Referring to FIG. 3, when the force applied to the robot 109 is removed, the robot 109 starts to shake in a free vibration state, and the vibration decreases little by little during unbalance control and lasts about 15 seconds. In case of applying the balance control, it can be seen that the value of the ZMP position in the same direction rapidly decreases and converges to zero.

4 is a diagram illustrating the trajectory of the ZMP position signal when the robot 109 stops after moving forward 8 steps in the stop state according to an embodiment of the present invention.

Referring to FIG. 4, it can be seen that the balance control is activated to correct the pelvis, that is, the hip joint angle trajectory, in real time, and the ZMP trajectory is accompanied by some irregularity, but always in the stable region regardless of the supporting state.

5 is a view showing the movement of the robot according to an embodiment of the present invention.

Referring to Figure 5, it can be seen that the room is stably moved as a picture of the robot taken from the front and side at regular time intervals.

Figure 6 is a flow chart showing the overall flow for the balance control method of the robot according to an embodiment of the present invention. Hereinafter, a process performed at each step will be described with reference to FIG. 6.

In order to control the balance of the robot, in step S600, the walking command receiving unit 101 receives a command for the walking pattern of the robot 109 from a PC or a remote controller.

Subsequently, in step S605, the path generation unit 103 generates a nominal angle that is a time history of the rotation angle at which each joint of the robot 109 should move using a walking pattern command. Inverse kinematic models can be used to generate nominal angles.

In step S610, the balance controller 105 determines the ZMP position of the robot 109 by using force / torque data acting on the foot of the robot 109 measured by the robot 109, and uses the ZMP position. Create a correction angle, which is the angle of rotation of the robot's pelvis and ankle joints. In this case, the force / torque data may be measured using an F / T sensor installed under the ankle of the robot 109.

In step S615, the joint controller 107 controls the movement for controlling the balance of the robot using the nominal angle and the correction angle.

Embodiments of the balance control method of the robot according to the present invention have been described so far, and the configuration of the balance control device 100 described with reference to FIGS. 1 to 5 can be applied to the present embodiment as it is. Detailed description thereof will be omitted.

As described above, the present invention has an advantage of precisely controlling balance using only ZMP sensor data measured by a robot. In addition, there is an advantage that the same balance control technique is applied to both the robot's foot support state and one foot support state.

As described above, the present invention has been described by specific embodiments such as specific components and the like. For those skilled in the art, various modifications and variations are possible from these descriptions. Accordingly, the spirit of the present invention should not be construed as being limited to the embodiments described, and all of the equivalents or equivalents of the claims, as well as the following claims, belong to the scope of the present invention .

100: balance control device 101: walk command receiver
103: path generation unit 105: balance control unit
107: joint control unit 109: robot

Claims (13)

A walking command receiver configured to receive a command for a walking pattern of a robot;
A path generation unit configured to generate a nominal angle, which is a time history of a rotation angle at which each joint of the robot should move using a command for the walking pattern;
The position of the robot's Zero Moment Point (ZMP) is measured using force / torque data acting on the robot's foot, and the rotation angle of the pelvis and ankle joint of the robot is measured using the ZMP position. A balance controller for generating a correction angle; And
And including a joint controller for controlling the movement of the robot joint using the nominal angle and the correction angle,
The joint control unit rotates the roll direction joints of the pelvis and the ankle of the robot when controlling the left / right direction of the robot, and controls the pelvis and the ankle of the robot when controlling the forward / rear direction of the robot. A balance control device for a robot, characterized by rotating a pitch direction joint. The method of claim 1,
And the path generation unit generates the nominal angle by using an inverse kinematics model. The method of claim 1,
And the force / torque data is measured using an F / T sensor installed under the robot ankle. The method of claim 1,
Correction angle of the pelvis and the ankle joint generated by the balance control has the same size, and the sign of the robot, characterized in that the signs are opposite to each other. delete The method of claim 1,
The joint control unit is a balance control device of the robot, characterized in that for controlling the joint of the robot using a proportional derivative (PD) control. The method of claim 1,
The correction angle is a balance control device for a robot, characterized in that expressed as the following equation.

here,

Is the correction angle for the balance control of the robot,

And

Is the control gain constant,

Denotes the position of ZMP in the y-axis direction, respectively. The method of claim 7, wherein
The control gain constant is a balance control apparatus of a robot, characterized in that expressed by the following equation.

here,

Is the target damping rate, m is the mass of the robot, k is the horizontal stiffness, c is the horizontal damping,

Is the height of the robot's center of gravity, and g is the acceleration of gravity, respectively. Receiving a command for a walking pattern of the robot;
Generating a nominal angle, which is a time history of a rotation angle at which each joint of the robot should move using the command for the walking pattern;
The position of the robot's Zero Moment Point (ZMP) is measured by using force / torque data acting on the robot's foot measured by the robot, and the pelvis and ankle joint of the robot using the ZMP position. Generating a correction angle that is a rotating angle of the; And
Controlling the movement of the robot joint using the nominal angle and the correction angle;
The controlling of the movement of the joint may include rotating the pelvis and the ankle joints of the ankle in the left / right direction of the robot and controlling the forward / rear direction of the robot. The balance control method of the robot, characterized in that for rotating the pelvis and the pitch joint of the ankle. 10. The method of claim 9,
And the correction angles of the pelvis and the ankle joint generated in the step of generating the correction angles are the same in size and opposite in sign to each other. delete 10. The method of claim 9,
The correction angle is a balance control method of the robot, characterized in that expressed as the following equation.

here,

Is the correction angle for the balance control of the robot,

And

Is the control gain constant,

Denotes the position of ZMP in the y-axis direction, respectively. The method of claim 12,
The control gain constant is a balance control method of a robot, characterized in that expressed by the following equation.

here,

Is the target damping rate, m is the mass of the robot, k is the horizontal stiffness, c is the horizontal damping,

Is the height of the robot's center of gravity, and g is the acceleration of gravity, respectively.

Images