CN107315346B - Humanoid robot gait planning method based on CPG model - Google Patents
Humanoid robot gait planning method based on CPG model Download PDFInfo
- Publication number
- CN107315346B CN107315346B CN201710487162.6A CN201710487162A CN107315346B CN 107315346 B CN107315346 B CN 107315346B CN 201710487162 A CN201710487162 A CN 201710487162A CN 107315346 B CN107315346 B CN 107315346B
- Authority
- CN
- China
- Prior art keywords
- osc
- oscillator
- formula
- robot
- model
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Active
Links
Images
Classifications
-
- G—PHYSICS
- G05—CONTROLLING; REGULATING
- G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Manipulator (AREA)
Abstract
The invention relates to the technical field of humanoid robots, in particular to a gait planning method of a humanoid robot based on a CPG model. It comprises the following steps: (1) establishing a corresponding coupled oscillator model according to the concrete hardware parameters of the robot: (2) improving the model, increasing a centroid offset control item, and obtaining an improved oscillator model: (3) and (3) optimizing by using a genetic algorithm by taking the speed as an input condition to obtain the optimal value of the parameter in the step (2), and substituting the obtained optimal value of the parameter into the model in the step (2). By adopting the planning method, when the robot is accelerated rapidly or moves rapidly, the robot is not easy to shake back and forth to cause falling.
Description
Technical Field
The invention relates to the technical field of humanoid robots, in particular to a gait planning method of a humanoid robot based on a CPG model.
Background
A bionic Central Pattern Generator (CPG) is a local oscillation network composed of neurons, and can generate stable phase locking by mutual inhibition between neurons, and generate rhythmic motion of a body-related part by self-oscillation, which is inspired by the fact that some researchers propose gait methods based on bionics. However, the CPG model has a large number of parameters without clear physical significance, and the value is difficult to determine, so that the CPG model is difficult to be directly applied to the gait planning of the robot, and for this reason, the Endo and the like simplify the CPG model and plan the gait of the humanoid robot by using an oscillator with few parameters; ha et al propose a Linear Coupled Oscillator (Linear Coupled Oscillator) model and apply it to a small humanoid robot that can withstand the computational cost limitation.
However, when the linear coupled oscillator model is applied to gait planning of a humanoid robot in the prior art, when the robot is accelerated rapidly or moves rapidly, the robot is still easy to oscillate back and forth, and thus falls down.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: the gait planning method of the humanoid robot based on the CPG model is provided, and when the robot is accelerated rapidly or moves rapidly, the robot is not easy to shake back and forth to cause tumble.
The technical scheme adopted by the invention is as follows: a gait planning method of a humanoid robot based on a CPG model comprises the following steps:
(1) establishing a corresponding coupled oscillator model according to the concrete hardware parameters of the robot:
OSCs(t)=OSCb(t)+OSCm(t);
in the formula, OSCm(t) represents the motion trajectory of the two-leg tip relative to the fixed center of mass, i.e. the output of the motion oscillator; OSCb(t) represents the motion trajectory of the centroid relative to the reference frame, i.e. the output of the balanced oscillator; OSCs(t) is the motion trail of the tail ends of the two legs relative to the actual mass center, namely the output of the coupled oscillator;
(2) improving the model, increasing a centroid offset control item, and obtaining an improved oscillator model:
OSCs(t)=OSCb(t)+OSCm(t)+offect;
in the formula, the offset represents the gain of the coupled oscillator corresponding to the initial position of the robot;
(3) and (3) optimizing by using a genetic algorithm by taking the speed as an input condition to obtain the optimal value of the parameter in the step (2), and substituting the obtained optimal value of the parameter into the model in the step (2).
OSC in step (1)bThe specific formula of (t) is:
OSCb(t)=ρbsin(ωbt+△b)+ub;
in the formula, ρbTo balance the amplitude, ω, of the oscillatorbTo balance the frequency of the oscillator, △bTo balance the initial phase, u, of the oscillatorbIs the offset of the balanced oscillator;
OSC in step (1)mThe specific formula of (t) is:
in the formula, ρmAmplitude, ω, of the moving oscillatormTo move the frequency of the oscillator, △mFor the initial phase of the motion vibrator, T is a walking cycle time, and r represents the proportion of the total cycle of the two-foot supporting time.
The parameter optimized by the genetic algorithm in the step (3) is rhobAnd offect.
The genetic algorithm in the step (3) sets population scale, cross probability, evolution probability and evolution algebra limit number, and then adopts a roulette method to select excellent populations, and the optimization target is given by the following formula:
Object:minimize f(ρb,offect)=max|Xzmp(t)-Xfcenter|+ρb,t∈[0,T],
in the formula, XzmpIs the X-axis coordinate of the zero moment point and has the formulaWherein n is the number of connecting rods of the robot, miMass of connecting rod i, g gravity acceleration, xiAnd ziThe positions of the x axis and the z axis of the connecting rod I are respectively,andthe corresponding acceleration is obtained;
in the formula, XfcenterThe X-axis coordinate of the center point of the robot supporting leg is expressed by the formulaWherein XtipFor supporting the distance from the tip of the leg to the origin,XheelThe distance from the heel of the supporting leg to the origin;
in the formula, max | Xzmp(t)-XfcenterAnd | is the farthest distance between the zero moment point and the central point of the support leg in the X axial direction in one period.
Compared with the prior art, the method has the following advantages that: the method is mainly characterized in that an X-axis ZMP stability margin and the oscillation amplitude of the center of mass are combined to serve as an optimization target in the walking process, a genetic algorithm is used as a solving tool, and corresponding optimal parameters are obtained under different speed inputs, so that the optimal parameters of the robot can be found under different speed inputs, the larger stability margin is ensured, the walking stability of the humanoid robot can be improved, and the probability of falling caused by the divergence of the front and back oscillations of the robot is reduced.
Drawings
Fig. 1 is an explanatory diagram of a coupled oscillator model in a gait planning method of a humanoid robot based on a CPG model.
FIG. 2 is a flow chart of a genetic algorithm in the gait planning method of the humanoid robot based on the CPG model.
Detailed Description
The present invention will be further described with reference to the following detailed description and drawings, but the present invention is not limited to the following detailed description.
A gait planning method of a humanoid robot based on a CPG model comprises the following steps:
(1) establishing a corresponding coupled oscillator model according to the concrete hardware parameters of the robot:
OSCs(t)=OSCb(t)+OSCm(t);
in the formula, OSCm(t) represents the motion trajectory of the two-leg tip relative to the fixed center of mass, i.e., the output of a motion Oscillator (motion Oscillator); OSCb(t) represents the movement locus of the center of mass relative to a reference coordinate system, i.e. the equilibrium vibrationThe output of an Oscillator (Balance Oscillator); OSCs(t) is the motion trail of the tail ends of the two legs relative to the actual mass center, namely the output of the coupled oscillator; as shown in figure 1, the walking experiment of human beings shows that the motion of ankle joints and mass centers of human bodies approaches to a sine fluctuation curve during normal walking, the two vibrators can respectively represent the poking of the ankle joints and the mass centers, and the coupled oscillator is formed by superposing the two vibrators, so that the whole walking curve of a human body can be represented.
OSC in step (1)bThe specific formula of (t) is:
OSCb(t)=ρbsin(ωbt+△b)+ub;
in the formula, ρbTo balance the amplitude, ω, of the oscillatorbTo balance the frequency of the oscillator, △bTo balance the initial phase, u, of the oscillatorbIs the offset of the balanced oscillator;
OSC in step (1)mThe specific formula of (t) is:
in the formula, ρmAmplitude, ω, of the moving oscillatormTo move the frequency of the oscillator, △mFor the initial phase of the motion vibrator, T is a walking cycle time, and r represents the proportion of the total cycle of the two-foot supporting time.
According to the formula, the centroid position c (t), the right foot end position r (t) and the left foot end position l (t) at the time t can be obtained;
during a walking cycle, the two legs are respectively supported for a common timeInner right sideThe legs are supporting points atThe inner left leg is a supporting point, when the right leg is a supporting leg and the left leg is a swinging leg, the coordinate of the right foot is not changed to r0, and thenThe method comprises the following steps:
c(t)=r0-OSC(t),
when the left foot is a supporting leg and the right foot is a swinging leg, the coordinate of the left foot is not changed to l0, and thenThe method comprises the following steps:
after the initial supporting point coordinates are determined, the central point coordinates c (t), the right foot coordinates r (t) and the left foot l (t) of each moment t can be obtained through the formula, the motor values of a plurality of motors of two legs of the robot can be obtained through inverse kinematics after the coordinates of the three points are known, and the motors move to the designated positions at each moment to realize the walking of the two legs of the robot.
(2) Improving the model, and adding a centroid shift control term to obtain an improved oscillator model:
OSCs(t)=OSCb(t)+OSCm(t)+offect;
in the formula, the offset represents the gain of the coupled oscillator corresponding to the initial position of the robot;
from the oscillator model in step (1) 9 parameters can be derived, where pmT and r are given by the velocity task, and ω after T is confirmedm、ωbAlso with the unique determination, △m、△b、ubCan be defined by the required positions of the left and right leg ends at the initial moment of the robot, thus only balanced oscillation is leftAmplitude of the device ρbExperiments show that the larger the moving speed of the robot is, the larger the amplitude of the mass center is, the smaller the moving speed is, and the smaller the amplitude of the mass center is, however, when the amplitude of the mass center is increased, the shaking amplitude of the robot is increased along with the increase of the moving speed, so that the shaking amplitude becomes an unstable walking factor, once disturbance is generated, shaking overshoot is easy to occur, and finally the robot is fallen down, so that the single rho value is independentbNor is it the optimal control factor.
According to the walking rule of human body, the gravity center is always made to be forward when the walking is accelerated suddenly, and is made to be backward when the walking is decelerated suddenly, so that the passenger can take the change of inertia force caused by the change of speed, and the walking stability is kept. By the inspiration, the center of mass offset term offset is added on the coupled oscillator model, so that when the offset is increased, the center of mass of the robot relatively moves backwards, the center of mass of the robot relatively moves forwards when the offset is reduced, the relative position of the center of mass of the robot can be adjusted by adjusting the band-under value of the offset, and the balance of the robot is further controlled. Experiments have found that the rho is adjusted in combinationbAnd the offect value can effectively improve the stability of the robot, however, a proper reference value is difficult to find only through manual adjustment, and the parameter optimization must be carried out on the number of the robot parameters under the condition of taking the speed as an input condition in order to ensure the stability of the robot at different speeds, so that the parameter optimization is carried out by adopting a genetic algorithm.
(3) And (3) optimizing by using a genetic algorithm by taking the speed as an input condition to obtain the optimal value of the parameter in the step (2), and substituting the obtained optimal value of the parameter into the model in the step (2).
The genetic algorithm in the step (3) sets population scale, cross probability, evolution probability and evolution algebra limit number, and then adopts a roulette method to select excellent populations, and the optimization target is given by the following formula: object of minimize f (ρ)b,offect)=max|Xzmp(t)-Xfcenter|+ρb,t∈[0,T],
In the formula, XzmpIs the X-axis coordinate of the zero moment point and has the formulaWherein n is the number of connecting rods of the robot, miMass of connecting rod i, g gravity acceleration, xiAnd ziThe positions of the x axis and the z axis of the connecting rod I are respectively,andthe corresponding acceleration is obtained;
in the formula, XfcenterThe X-axis coordinate of the center point of the robot supporting leg is expressed by the formulaWherein XtipFor supporting the distance of the tip of the leg to the origin, XheelThe distance from the heel of the supporting leg to the origin;
in the formula, max | Xzmp(t)-XfcenterAnd | is the farthest distance between the zero moment point and the central point of the support leg in the X axial direction in one period.
In the present application, the population size M is set to 100, and the poor probability P is setc0.5, evolution probability PeThe evolution algebra is limited to T1500, and the process of the genetic algorithm is shown in fig. 2.
Claims (1)
1. A gait planning method of a humanoid robot based on a CPG model is characterized by comprising the following steps:
(1) establishing a corresponding coupled oscillator model according to the concrete hardware parameters of the robot:
OSCs(t)=OSCb(t)+OSCm(t);
in the formula, OSCm(t) represents the motion trajectory of the two-leg tip relative to the fixed center of mass, i.e. the output of the motion oscillator; OSCb(t) represents the motion trajectory of the centroid relative to the reference frame, i.e. the output of the balanced oscillator; OSCs(t) is the motion trail of the tail ends of the two legs relative to the actual mass center, namely the output of the coupled oscillator;
(2) improving the model, increasing a centroid offset control item, and obtaining an improved oscillator model:
OSCs(t)=OSCb(t)+OSCm(t)+offect;
in the formula, the offset represents the gain of the coupled oscillator corresponding to the initial position of the robot;
(3) optimizing by using a genetic algorithm with the speed as an input condition to obtain the optimal value of the parameter in the step (2), and substituting the obtained optimal value of the parameter into the model in the step (2);
OSC in step (1)bThe specific formula of (t) is:
OSCb(t)=ρbsin(ωbt+△b)+ub;
in the formula, ρbTo balance the amplitude, ω, of the oscillatorbTo balance the frequency of the oscillator, △bTo balance the initial phase, u, of the oscillatorbIs the offset of the balanced oscillator;
OSC in step (1)mThe specific formula of (t) is:
in the formula, ρmAmplitude, ω, of the moving oscillatormTo move the frequency of the oscillator, △mThe initial phase of the motion vibrator, T is a walking cycle time, and r represents the proportion of the supporting time of the feet in the total cycle;
the parameter optimized by the genetic algorithm in the step (3) is rhobAnd an offect;
the genetic algorithm in the step (3) sets population scale, cross probability, evolution probability and evolution algebra limit number, and then adopts a roulette method to select excellent populations, and the optimization target is given by the following formula:
Object:minimize f(ρb,offect)=max|Xzmp(t)-Xfcenter|+ρb,t∈[0,T],
in the formula, XzmpIs the X-axis coordinate of the zero moment point and has the formulaWherein n is the number of connecting rods of the robot, miMass of connecting rod i, g gravity acceleration, xiAnd ziThe positions of the x axis and the z axis of the connecting rod I are respectively,andthe corresponding acceleration is obtained;
in the formula, XfcenterThe X-axis coordinate of the center point of the robot supporting leg is expressed by the formulaWherein XtipFor supporting the distance of the tip of the leg to the origin, XheelThe distance from the heel of the supporting leg to the origin;
in the formula, max | Xzmp(t)-XfcenterAnd | is the farthest distance between the zero moment point and the central point of the support leg in the X axial direction in one period.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710487162.6A CN107315346B (en) | 2017-06-23 | 2017-06-23 | Humanoid robot gait planning method based on CPG model |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710487162.6A CN107315346B (en) | 2017-06-23 | 2017-06-23 | Humanoid robot gait planning method based on CPG model |
Publications (2)
Publication Number | Publication Date |
---|---|
CN107315346A CN107315346A (en) | 2017-11-03 |
CN107315346B true CN107315346B (en) | 2020-01-14 |
Family
ID=60180230
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710487162.6A Active CN107315346B (en) | 2017-06-23 | 2017-06-23 | Humanoid robot gait planning method based on CPG model |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107315346B (en) |
Families Citing this family (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108594661B (en) * | 2018-05-08 | 2021-01-26 | 东南大学 | Bionic motion control method of wheel-leg combined robot based on CPG |
CN108372506B (en) * | 2018-05-16 | 2021-03-02 | 中南大学 | Method for realizing humanoid robot self-adaptive walking frame based on CPG model |
CN112937721B (en) * | 2021-04-18 | 2022-09-30 | 北京工业大学 | Design of seven-connecting-rod biped robot and hybrid control method based on ZMP and CPG |
CN113204892B (en) * | 2021-05-26 | 2023-01-31 | 深圳市优必选科技股份有限公司 | Centroid trajectory generation method and device, computer readable storage medium and robot |
CN114460849B (en) * | 2022-04-12 | 2022-07-12 | 北京晟海汇泽科技有限公司 | Bionic robot fish motion control method and device and bionic robot fish |
Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102147592A (en) * | 2010-02-10 | 2011-08-10 | 中国科学院自动化研究所 | Fuzzy controller for controlling motion of four-footed robot |
CN103197671A (en) * | 2012-01-04 | 2013-07-10 | 中国人民解放军第二炮兵工程学院 | Humanoid robot gait planning and synthesizing method |
CN103345285A (en) * | 2013-06-27 | 2013-10-09 | 山东大学 | Quadruped robot remote control system and remote control method thereof |
CN103376742A (en) * | 2012-04-24 | 2013-10-30 | 中国科学院合肥物质科学研究院 | CPG control system of wall-climbing robot imitating flexible structure of feet of gecko |
CN104331081A (en) * | 2014-10-10 | 2015-02-04 | 北京理工大学 | Gait planning method for walking of biped robot along slope |
CN106292288A (en) * | 2016-09-22 | 2017-01-04 | 同济大学 | Model parameter correction method based on Policy-Gradient learning method and application thereof |
Family Cites Families (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
GB0104995D0 (en) * | 2001-02-28 | 2001-04-18 | Isis Innovation | Artificial morpho-functional multiped and motion controller therefor |
US7703562B2 (en) * | 2007-05-25 | 2010-04-27 | Toyota Motor Engineering & Manufacturing North America, Inc. | Energy efficient robotic system |
-
2017
- 2017-06-23 CN CN201710487162.6A patent/CN107315346B/en active Active
Patent Citations (6)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN102147592A (en) * | 2010-02-10 | 2011-08-10 | 中国科学院自动化研究所 | Fuzzy controller for controlling motion of four-footed robot |
CN103197671A (en) * | 2012-01-04 | 2013-07-10 | 中国人民解放军第二炮兵工程学院 | Humanoid robot gait planning and synthesizing method |
CN103376742A (en) * | 2012-04-24 | 2013-10-30 | 中国科学院合肥物质科学研究院 | CPG control system of wall-climbing robot imitating flexible structure of feet of gecko |
CN103345285A (en) * | 2013-06-27 | 2013-10-09 | 山东大学 | Quadruped robot remote control system and remote control method thereof |
CN104331081A (en) * | 2014-10-10 | 2015-02-04 | 北京理工大学 | Gait planning method for walking of biped robot along slope |
CN106292288A (en) * | 2016-09-22 | 2017-01-04 | 同济大学 | Model parameter correction method based on Policy-Gradient learning method and application thereof |
Non-Patent Citations (2)
Title |
---|
Gait Pattern Generation and Stabilization for Humanoid Robot Based on Coupled Oscillators;Inyong Ha 等;《IEEE Xplore》;20111205;第3207-3212页 * |
基于线性耦合振荡器模型的仿人机器人步态规划算法;汪柳青 等;《中国科学技术大学学报》;20141031;第44卷(第10期);第795-803页 * |
Also Published As
Publication number | Publication date |
---|---|
CN107315346A (en) | 2017-11-03 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107315346B (en) | Humanoid robot gait planning method based on CPG model | |
CN110315543B (en) | Biped robot gait generation and optimization method | |
US8204626B2 (en) | Control device for mobile body | |
KR101100657B1 (en) | Robot and attitude control method of robot | |
KR100718479B1 (en) | Ambulation control apparatus and ambulation control method of robot | |
JP3972854B2 (en) | Robot motion control device | |
JP5052013B2 (en) | Robot apparatus and control method thereof | |
Buschmann et al. | Experiments in fast biped walking | |
Liu et al. | Bipedal walking with dynamic balance that involves three-dimensional upper body motion | |
Park et al. | Evolutionary-optimized central pattern generator for stable modifiable bipedal walking | |
Park et al. | Full-body joint trajectory generation using an evolutionary central pattern generator for stable bipedal walking | |
JP4483254B2 (en) | Robot apparatus and control method of robot apparatus | |
JP2024502726A (en) | How to move an exoskeleton | |
Kajita et al. | Biped gait control based on spatially quantized dynamics | |
Missura et al. | Gradient-driven online learning of bipedal push recovery | |
JP3677623B2 (en) | Real-time optimal control method for legged robot | |
Yu et al. | Biped robot gait planning based on 3D linear inverted pendulum model | |
Song et al. | CPG-based control design for bipedal walking on unknown slope surfaces | |
Liu et al. | Bipedal walking with push recovery balance control involves posture correction | |
CN114700948B (en) | Lower limb exoskeleton robot control system based on divergent motion component | |
Galdeano et al. | Optimal pattern generator for dynamic walking in humanoid robotics | |
Aghaabbasloo et al. | Biped robot joint trajectory generation using PSO evolutionary algorithm | |
CN115202259A (en) | CPG control system of quadruped robot and parameter setting method thereof | |
JP4587699B2 (en) | Robot apparatus and control method thereof | |
Ye et al. | Bipedal walking control by using acceleration factor |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |