KR20130066947A - Vertical center of mass motion generation method using evolutionary optimized central pattern generator for human-like stable bipedal walking - Google Patents

Abstract

PURPOSE: A vertical direction gravity center movement generating method using an evolutionarily optimized center pattern generator for stable biped walking is provided to change a walking figure by a variable walking pattern generator. CONSTITUTION: A gravity center path of a three-dimensional linear pendulum model in a vertical direction is generated through a center pattern generator. Disturbance by the movement of a gravity center in a vertical direction is compensated through a sensor feedback of the center pattern generator. The center pattern generator is optimized as an evolutionary algorithm. [Reference numerals] (AA) Start; (BB) Finish; (S11) Generate a vertical gravity center pattern of 3D linear inverted pendulum model by using a central pattern generator; (S12) Compensate distortion by the vertical movement of gravity through a sensor feedback of a central pattern generator; (S13) Optimize a central pattern generator for generating a vertical gravity path with an evolutionary algorithm; (S14) Produce μ0, r,r' of a neural scillator and K_f of a sensory feedback by optimizing a central pattern generator

Description

VERTICAL CENTER OF MASS MOTION GENERATION METHOD USING EVOLUTIONARY OPTIMIZED CENTRAL PATTERN GENERATOR FOR HUMAN-LIKE STABLE BIPEDAL WALKING}

The present invention relates to the walking of bipedal robots, and in particular, to a vertical center of gravity using an evolutionary optimized central pattern generator (CPG) for stable bipedal walking such as humans. center of mass).

Development of a gait generation method for stable bipedal walking is one of the key research topics for bipedal robots. In particular, the 3-D linear inverted pendulum model (3-D LIPM) is the most widely used method for generating the gait of biped walking robots.

This method assumes that the height of the center of gravity is constant to decouple the equation of movement of the center of gravity in the sagittal and lateral directions of the biped walking robot.

A method of generating a gait for various biped walking based on the 3-D LIPM has been developed. Many are open to the public.

The foregoing patent discloses a ZMP (Zero) measured for a biped robot whose gait is controlled according to a COM (Center Of Mass) error value (formerly COM error value) used for gait control of a biped robot at a previous time point and a previous COM error value. Generating a COM value used to generate a COM error value (current COM error value) for gait control of the biped robot at a current point of time by performing Kalman filtering on a Moment Point) value; Converting the generated COM value into a ZMP value to generate a predicted ZMP value used to generate a current COM error value; Generating an error value (ZMP error value) between the predicted ZMP value and the target ZMP value of the biped robot; Converting a ZMP error value to generate a current COM error value; And controlling the gait of the biped robot by applying a current COM error value to the biped robot.

However, keeping the height of the center of gravity still has the problem of causing the biped walking robot to walk unnaturally with a small stride.

The present invention has been made in view of the above problems, and aims at generating a center of mass motion in a vertical direction using an evolution optimized optimized central pattern generator (CPG).

In addition, the present invention uses a variable walking pattern generator (MWPG) to generate the gait of the biped walking robot, and generates a trajectory of the center of gravity in the vertical direction using CPG for the movement of the center of gravity in the vertical direction The purpose is to.

In addition, the present invention optimizes the CPG through an evolutionary algorithm, thereby deriving a desired output signal from the CPG, and disturbance due to vertical center of gravity movement through the sensory feedback of the CPG. Its purpose is to compensate for the distance.

In order to achieve this technical problem, a vertical center of gravity motion generation method using an evolution optimized center pattern generator for stable biped walking of the present invention is an evolution optimized optimized central pattern generator (CPG). It includes; (a) generating a center of gravity movement in the vertical direction using.

According to the present invention as described above, by using a method of generating a center of mass movement in the vertical direction using an evolution optimized optimized central pattern generator (CPG), the biped walking robot is fixed ( constant) Compared to biped walking at the center of gravity, it is possible to walk naturally like a human with a larger stride.

And according to the present invention, the variable step generator while compensating for disturbance (sensory feedback) of the CPG, while the biped walking robot walks on the vertical center of gravity trajectory generated by the evolution-optimized CPG It is also possible to change the pace by a (modifiable walking pattern generator).

1 is an exemplary view showing a 3-D linear inverted pendulum model (3D-LIPM) in a plane according to the present invention.
2 is a flowchart illustrating a method for generating a center of gravity movement in a vertical direction using an evolution-optimized center pattern generator for stable biped walking according to the present invention.
3 is a flowchart illustrating a detailed process of step S10 of the method for generating the center of gravity movement in the vertical direction using an evolution-optimized center pattern generator for stable biped walking according to the present invention.

Specific features and advantages of the present invention will become more apparent from the following detailed description based on the accompanying drawings. Prior to this, terms and words used in the present specification and claims are to be interpreted in accordance with the technical idea of the present invention based on the principle that the inventor can properly define the concept of the term in order to explain his invention in the best way. It should be interpreted in terms of meaning and concept. It is to be noted that the detailed description of known functions and constructions related to the present invention is omitted when it is determined that the gist of the present invention may be unnecessarily blurred.

Hereinafter, with reference to the accompanying drawings for the configuration and operation of the embodiment of the present invention will be described in detail.

Bipedal walking consists of a single support phase and a double support phase. The main dynamics of the humanoid robot in one-foot support state are the 3D linear inverted pendulum model (3D-LIPM) and the supporting legs are weightless rods as shown in FIG. It is assumed that mass is gathered at one point.

Conventional 3-D LIPM does not consider the center of mass movement in the vertical direction. That is, it is possible to set the height of the center of gravity to a constant value and consequently decouple the center of gravity equation of motion in the sagittal and lateral directions of the biped walking robot. The dynamic equation of 3-D LIPM is as follows.

와

는 각각 무게 중심 위치와 ZMP(zero moment point)를 나타낸다. [수학식 1]로부터 3D-LIPM의 무게 중심의 움직임과 ZMP 사이의 관계를 얻을 수 있다.here,

Wow

Denotes the center of gravity position and the zero moment point (ZMP), respectively. From Equation 1, the relationship between the movement of the center of gravity of the 3D-LIPM and the ZMP can be obtained.

By solving the solution of Equation 1 using an inverse Laplace transform, the equation of motion of the center of gravity of the 3D-LIPM can be obtained as follows. [Equation 2] is the equation of motion of the center of gravity in the sagittal direction of the robot, and [Equation 3] is the equation of motion of the center of gravity in the lateral direction of the robot.

와

는 각각 정면 방향과 측면 방향으로의 무게 중심의 처음/마지막의 위치와 속도를 나타낸다. 그리고

와

는 각각

와

를 나타내고,

는

를 나타낸다.

는 남아있는 한발지지 시간을 나타내고,

와

는 각각 정면 방향과 측면 방향의 ZMP 함수를 나타내며, 여기서

이고,

이다.here,

Wow

Are the first and last positions and speeds of the center of gravity in the front and lateral directions, respectively. And

Wow

Respectively

Wow

Lt; / RTI >

The

Indicates.

Represents the remaining support time,

Wow

Represents the ZMP function in the front and side directions, respectively

ego,

to be.

와

를 이용하여 ZMP궤적을 변화시킴으로써 더욱 확장시킨다.The first and second terms on the right side of [Equation 2] and [Equation 3] represent the homogeneous solution and the specific solution of Equation 1, respectively. The special solution is the movement of the center of gravity in the front and side directions.

Wow

Expand further by changing the ZMP trajectory using.

The conventional 3-D LIPM does not consider this special solution but considers only a homogeneous solution. As a result, in the conventional 3-D LIPM, the movement of the center of gravity in a single support state is predetermined and not variable. That is, in the existing 3-D LIPM, the biped walking robot cannot independently change the gait consisting of one foot support time, two foot support time, stride length in the front direction and the lateral direction, and foot direction.

However, by considering special solutions in a variable walking pattern generator (MWPG), the biped walking robot can independently change the pace by the ZMP change.

Hereinafter, the method for generating the center of gravity movement in the vertical direction using the evolution-optimized central pattern generator for stable biped walking according to the present invention will be described with reference to FIG. 2. In operation S10, the center of gravity movement in the vertical direction is generated using an evolution optimized optimized central pattern generator (CPG).

At this time, the CPG internally generates a multi-dimensional rhythmic signal, and modifies the signal generated by the CPG according to environmental perturbation using sensory feedback. .

The rhythmic signal is generated using a neural oscillator as shown in [Equation 4].

와

는 각각 i번째 뉴런(neuron)의 내부 상태(inner state)와 자가 억제 상태(self-inhibition state)를 나타낸다.

은 출력신호(output signal)의 크기(amplitude)에 영향을 주는 외부 입력 신호(external input signal)를 나타내고

은 출력 신호(output signal)를 나타낸다.

는 i번째 뉴런과 j번째 뉴런 사이의 위상(phase) 차를 결정하는 연결 가중치(connecting weight)를 나타낸다.

는 자가 억제의 가중치를 나타내고

은 안정적인 이족 보행을 위해 필요한 센서 피드백 신호를 나타낸다.

와

은 출력 신호의 모양(shape)와 주기(period)에 영향을 주는 시간 상수(time constant)를 나타낸다.here,

Wow

Represents the inner state and the self-inhibition state of the i-th neuron, respectively.

Denotes an external input signal that affects the amplitude of the output signal.

Denotes an output signal.

Denotes a connecting weight that determines a phase difference between the i th neuron and the j th neuron.

Represents the weight of self-inhibition

Represents the sensor feedback signal required for stable biped walking.

Wow

Denotes a time constant that affects the shape and period of the output signal.

Biological rhythmic locomotion is then derived from a series of muscle relaxation and flexion, with one relax and the other constrict and the process repeats.

CPG structures have been developed for the modeling of such biological systems and are used in the present invention to generate trajectories of vertical centers of gravity. In this structure, basic rhythmic movements are generated by neural oscillators, which consist of relaxation neurons and contractile neurons.

Hereinafter, referring to FIG. 3, a detailed process of step S10 of the method for generating a center of gravity movement in the vertical direction using an evolution optimized center pattern generator for stable biped walking according to the present invention will be described below.

First, the present invention generates a center of mass trajectory in the vertical direction of a 3-D linear inverted pendulum model using CPG as shown in [Equation 5] (S11).

는 크기 조정 인수(amplitude scaling factor)이고

은 수직방향의 무게 중심 움직임을 위한 CPG에 의해 생성된 궤적을 나타낸다.

과

는 각각 이완과 수축 뉴런의 역할을 수행한다.here,

Is the amplitude scaling factor

Represents the trajectory generated by the CPG for vertical center of gravity movement.

and

Plays the role of relaxation and contraction neurons, respectively.

After the step S11, the disturbance caused by the movement of the center of gravity in the vertical direction is compensated for through sensory feedback of the central pattern generator (S12).

At this time, the sensor feedback is defined as shown in [Equation 6] to minimize the oscillation of the ground reaction force (GRF) of the biped walking robot to improve the stability of the biped walking robot. .

는 조정 인수이고,

와

은 각각 왼발과 오른발의 지면 반발력(GRF)를 나타내며,

은 로봇의 질량을 나타낸다.here,

Is the adjustment factor,

Wow

Represents the ground reaction force (GRF) of the left and right feet, respectively.

Represents the mass of the robot.

After step S12, a central pattern generator for generating a center of mass trajectory in the vertical direction is optimized with an evolutionary algorithm (S13).

At this time, the optimization of the center of mass trajectory in the vertical direction minimizes the vibration of the ground reaction force (GRF) while the biped walking robot walks through the evolutionary optimization algorithm of the center pattern generator (CPG). To derive the output signal.

와 같아야 하며, 제S13단계의 최적화(optimization)는 [수학식 7]을 통해 도출한다.In addition, the period of the trajectory generated by the evolution optimized central pattern generator (CPG) is derived to be the same as the walking period of the biped walking robot, and the amplitude of this trajectory at each period is They must be the same size

It should be equal to, and optimization of step S13 is derived through Equation 7.

는 진화 최적화된 CPG에 의해 생성된 궤적의 주기이고,

와

는 각각 이 궤적의 처음과 끝 주기에서의 크기를 나타낸다.

는 걷는 동안 GRF와 로봇의 무게 사이의 차이의 합을 의미하고

는 로봇이 걷는 동안 넘어지면 주어지는 벌칙 값(penalty value)이다.here,

Is the period of the trajectory generated by the evolutionary optimized CPG,

Wow

Are the magnitudes at the beginning and end of the trajectory, respectively.

Is the sum of the differences between the GRF and the weight of the robot while walking

Is the penalty value given when the robot falls down while walking.

,

,

와 센서 피드백(sensory feedback)의

값을 도출한다(S14).In addition, the central pattern generator (CPG) for generating the center of mass trajectory in the vertical direction in step S13 may be evolutionarily optimized to achieve a neural oscillator.

,

,

Of sensory feedback

A value is derived (S14).

While the present invention has been particularly shown and described with reference to preferred embodiments thereof, it will be understood by those skilled in the art that various changes in form and details may be made therein without departing from the spirit and scope of the invention as defined by the appended claims. It will be appreciated by those skilled in the art that numerous changes and modifications may be made without departing from the invention. And all such modifications and changes as fall within the scope of the present invention are therefore to be regarded as being within the scope of the present invention.

10: flat
11: center of mass of 3-D LIPM
12: Rod that supports the center of gravity of 3D-LIPM

Claims (7)

A method of generating vertical center of mass motion for a stable bipedal robot,
generating a center of gravity movement in a vertical direction using an evolution optimized optimized central pattern generator (CPG); Method of generating vertical center of gravity movement using pattern generator. The method of claim 1,
The step (a)
(a-1) The center of mass trajectory of the 3-D linear inverted pendulum model is expressed by the central pattern generator (CPG) by Equation 5. Generating using; a vertical center of gravity movement generating method using an evolution-optimized central pattern generator for stable bipedal walking, comprising: a step of generating the same.
&Quot; (5) "

Of Equation 5 above

Is the amplitude scaling factor

Is the trajectory generated by the central pattern generator for vertical center of gravity movement,

and

Are neuron factors according to extension and contraction, respectively. The method of claim 1,
The step (a)
(a-2) compensating for disturbance caused by vertical center of gravity movement through sensory feedback of the central pattern generator; stable bipedal walking, comprising a A method of generating vertical center of gravity motion using an evolution-optimized central pattern generator. The method of claim 3, wherein
Disturbance compensation by the center of mass movement in the vertical direction,
Vertical center of gravity movement using an evolution-optimized central pattern generator for stable bipedal walking, characterized in that the central pattern generator compensates for the sensory feedback by Equation 6 How to produce.
&Quot; (6) "

Of Equation 6 above

Is the scaling factor,

Wow

Are the ground reaction forces of the left and right feet, respectively.

Is the mass of the robot. The method of claim 1,
The step (a)
(a-3) optimizing a central pattern generator for generating a center of mass trajectory in a vertical direction with an evolutionary algorithm; A method of generating vertical center of gravity movement using an evolution-optimized central pattern generator for stable bipedal walking. The method of claim 5, wherein
The optimization of the center of mass trajectory in the vertical direction is
Evolution-optimized central pattern generator for stable bipedal walking characterized by evolutionary optimization of the central pattern generator (CPG) through an evolutionary optimization algorithm according to the objective function of Equation (7). Method of generating vertical center of gravity movement using.
&Quot; (7) "

Of Equation 7 above

Is the period of the trajectory generated by the evolutionary optimized CPG,

Wow

Are the magnitudes at the beginning and end of this trajectory, respectively.

Is the sum of the difference between the ground reaction force and the weight of the robot while walking,

Is the penalty value given when the robot falls down while walking. The method of claim 5, wherein
Evolutionary optimization of a central pattern generator (CPG) for generating the center of mass trajectory in the vertical direction, so that the neural oscillator

,

,

Of sensory feedback

A method of generating vertical center of gravity motion using an evolution-optimized central pattern generator for stable bipedal walking characterized by deriving (scaling factor) values.

Images