CN108537268A - A kind of robot quasi-periodic motion demonstration learning method - Google Patents
A kind of robot quasi-periodic motion demonstration learning method Download PDFInfo
- Publication number
- CN108537268A CN108537268A CN201810295971.1A CN201810295971A CN108537268A CN 108537268 A CN108537268 A CN 108537268A CN 201810295971 A CN201810295971 A CN 201810295971A CN 108537268 A CN108537268 A CN 108537268A
- Authority
- CN
- China
- Prior art keywords
- motion
- periodic motion
- quasi
- periodic
- aperiodic
- 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.)
- Granted
Links
- 230000033001 locomotion Effects 0.000 title claims abstract description 116
- 238000000034 method Methods 0.000 title claims abstract description 38
- 230000000737 periodic effect Effects 0.000 claims abstract description 24
- 238000000354 decomposition reaction Methods 0.000 claims abstract description 8
- 238000000605 extraction Methods 0.000 claims description 8
- 239000000203 mixture Substances 0.000 claims description 8
- 239000004615 ingredient Substances 0.000 claims description 5
- 238000000926 separation method Methods 0.000 claims description 5
- 230000006870 function Effects 0.000 abstract description 6
- 238000005516 engineering process Methods 0.000 abstract description 3
- 230000004888 barrier function Effects 0.000 abstract 1
- 238000001228 spectrum Methods 0.000 description 4
- 238000013528 artificial neural network Methods 0.000 description 3
- 238000012512 characterization method Methods 0.000 description 3
- 230000000694 effects Effects 0.000 description 3
- 238000005498 polishing Methods 0.000 description 3
- 238000005507 spraying Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 2
- 238000010801 machine learning Methods 0.000 description 2
- 238000012545 processing Methods 0.000 description 2
- 230000006978 adaptation Effects 0.000 description 1
- 230000015572 biosynthetic process Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 230000007812 deficiency Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002708 enhancing effect Effects 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 230000003993 interaction Effects 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000008447 perception Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 238000004064 recycling Methods 0.000 description 1
- 238000011160 research Methods 0.000 description 1
- 238000003786 synthesis reaction Methods 0.000 description 1
- 238000012549 training Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/241—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches
- G06F18/2415—Classification techniques relating to the classification model, e.g. parametric or non-parametric approaches based on parametric or probabilistic models, e.g. based on likelihood ratio or false acceptance rate versus a false rejection rate
Abstract
The invention discloses a kind of robot quasi-periodic motions to demonstrate learning method, the study of robot quasi-periodic motion and extensive reproduction function are realized using classification, decomposition, modeling and synthetic technology successively, i.e., determine that movement locus is periodic motion, aperiodic motion or quasi-periodic motion first with sorting technique;Recycle decomposition algorithm that quasi-periodic motion is decomposed into periodic motion and aperiodic motion one by one;Then modeling study and prediction are carried out to periodic motion and aperiodic motion respectively, each component after study is finally synthesized by a new quasi-periodic motion according to the definition of quasi-periodic motion.The present invention successfully characterizes robot motion track quasi-periodic motion complicated under non-structure environment, efficiently solves the technical barriers such as the having a single function of traditional automatic planning system method of such complicated track, universality difference.
Description
Technical field
The present invention relates to robot motion track demonstration learning art field more particularly to a kind of robot quasi-periodic motions
Demonstrate learning method.
Background technology
Demonstration study (Learning from Demonstration (LfD)) is a kind of by human-computer interaction, teaching, general
The technical ability of people imparts to the technology of robot, can simplify programming significantly, improve learning efficiency.The core of LfD is to robot
Movement locus carries out data characterization and study.Most scholars are devoted to building for research robot aperiodic motion track at present
Mould learning method, for example neural network (Artificial Neural Networks (ANNs)), hidden Markov model
(Hidden Markov Model (HMM)) is used for solving the problems, such as perception problems and reproduction in early stage, but both methods needs
Excessive sample data can just train ideal model.Calinon uses gauss hybrid models (Gaussian Mixture
Model (GMM)) track is encoded, and it is extensive using Gaussian Mixture recurrence (Gassian Mixture Regression (GMR))
Reappear the movement locus of robust under varying environment.This method not only can be used for being split movement locus, but also can by improving
With by introducing task parameters, adaptation and generalization ability of the enhancing model under new environment.Ijspeert and Schaal proposes dynamic
State metaaction model (Dynamic Motion Primitive (DMP)) models track, and is existed using non parametric regression
Reappear movement under varying environment.It is dedicated to learning dynamics system rather than motion itself.Therefore its reappear process not only flexibly but also
Stablize, but is difficult that the suitable parameter of adjustment goes to generate new model to adapt to various occasions.Although these methods can solve
The movement locus in the fields LfD characterizes and problem concerning study.But these methods can not characterize and learn non-structure environment (first, ring
The position of object is uncertain in border, second is that object type is various in environment) under Comlex-locus, for example under non-structure environment
The spraying of the acquisition track and curved surface part of peg-in-hole assembly, polishing track.The traditional modeling method of the type games is all based on list
The automatic orbit planning system of one Functional Design.For example the scholars such as Heping Chen propose a kind of based on free form surface
(free-form surfaces), CAD model (CAD modle), tool model (Tool modle) spraying profile advise automatically
The system of drawing.The scholars such as BASANEZ are also based on one Robotic polishing system of part C AD modellings.But these systems lack
Weary universality, has a single function, and programming is complicated.In conclusion current demonstration learning ways and track automatic planning system are not
Such Comlex-locus can fully be characterized.
Invention content
The purpose of the present invention is in view of the deficiencies of the prior art, propose that a kind of robot quasi-periodic motion demonstrates study side
Method.
The purpose of the present invention is achieved through the following technical solutions:
A kind of robot quasi-periodic motion demonstration learning method, including:
Step 1, when determine robot motion track be quasi-periodic motion when, offset is carried out successively to the movement locus
Quasi-periodic motion is decomposed into a series of periodic motions and non-week by extraction, harmonic wave separation, envelope extraction and phase identification one by one
Phase moves;
Step 2 returns GMR respectively to periodic motion and aperiodic motion using gauss hybrid models GMM and Gaussian Mixture
In each component carry out modeling study and prediction;
Each component after study is synthesized new quasi-periodic motion by step 3 according to following definition:
Wherein, γ is offset, ξijFor envelope component, χijFor harmonic component, N be nonlinear combination periodic motion and
The quantity of aperiodic motion, C be periodic motion Fourier expansion formula in harmonic components number.
Based on the above technical solution, the present invention can also be improved as follows.
Further, before the step 1, further include:
Using the spectrogram characteristic of Fast Fourier Transform (FFT) FFT, determine robot motion track be periodic motion, it is aperiodic
Movement or quasi-periodic motion.
Further, the spectrogram characteristic using Fast Fourier Transform (FFT) FFT determines that robot motion track is the period
Movement, aperiodic motion or quasi-periodic motion, specifically include:
As f=0, amplitude is more than predetermined threshold value, then the movement locus is aperiodic motion;
As f=0, amplitude is not present, and works as f=nf0When, amplitude is more than predetermined threshold value, then the movement locus is the period
Movement;
As f=0 and f=nf0When, amplitude is all higher than predetermined threshold value, then the movement locus is quasi-periodic motion;
Wherein, f is frequency, and n is integer, f0For fundamental frequency.
Further, the offset extraction uses empirical mode decomposition EMD methods, obtains offset γ.
Further, the harmonic wave separation use notch filter notch filter methods, obtain only there are one frequency at
The harmonic component ξ with amplitude modulation dividedijχij。
Further, the envelope extraction uses Hilbert transform methods, obtains envelope component ξij。
Further, harmonic component χijExpression formula be:
Wherein, ΩiFor the angular frequency of first harmonic ingredient,For the phase angle of jth order harmonic components.
Further, the phase angleExpression formula be:
Wherein, K is the number of peak value, tpiAnd kiFor harmonic component χijPeak value and the time, ω be harmonic component χijAngle
Frequency, αjFor phase compensation factor.
Compared with prior art, the present invention having an advantageous effect in that:
(1) modeling algorithm selects GMM/GMR models, can learn fortune paracycle of extensive different application occasion different function
Dynamic rail mark, so that the invention successfully solves the single-function problem of traditional quasi-periodic motion planning system.
(2) extensive reproduction ability, the invention successfully solve traditional quasi-periodic motion planning by force for the track of GMM/GMR
System does not have the problem of extensive predictive ability.
(3) current existing demonstration learning ways are directed to, which can handle the movement locus with more general significance,
Solve the problems, such as Comlex-locus characterization and the study under non-structure environment.
Description of the drawings
Fig. 1 is the flow diagram that a kind of robot quasi-periodic motion provided in an embodiment of the present invention demonstrates learning method.
Specific implementation mode
The principle and features of the present invention will be described below with reference to the accompanying drawings, and the given examples are served only to explain the present invention, and
It is non-to be used to limit the scope of the present invention.
The concept of signified quasi-periodic motion in the present invention is according to for example initial ranging campaign of high-precision assembly and song
Complicated track that the polishings of surface parts, spraying operation are formed and propose, movement locus function definition is:
Wherein, u0(t) it is offset,2ui(t) it is periodic component,1ui(t) it is corresponding envelope ingredient.u0(t) and1ui(t)
All it is aperiodic motion,2ui(t) it is periodic motion, N is the quantity of periodic motion and the aperiodic motion of nonlinear combination.
Periodic component2ui(t) it is equivalent to according to Fourier expansion formula:
Wherein, Ai0、Aij、Ωi、It is the coefficient in Fourier expansion formula, A with Ci0-- DC component, Aij-- it shakes
Width, Ωi-- the angular frequency of (fundamental frequency) first harmonic ingredient,-- the phase angle of jth subharmonic ingredient, Fu of C-- periodic motions
In in leaf series expansion harmonic components number.
And then propose that the equivalent model of quasi-periodic motion u (t) is as follows:
Wherein,ξij=1uiAij,That is γ is the offset of equivalent model
Component, ξijFor the envelope component of equivalent model, χijFor harmonic component.
As shown in Figure 1, a kind of demonstration learning method of robot quasi-periodic motion provided in an embodiment of the present invention includes such as
Lower step:
(1) classify:It is right using the spectrogram characteristic of Fast Fourier Transform (FFT) (Fast Fourier transform (FFT))
Track is classified, and determines that robot motion track is periodic motion, aperiodic motion or quasi-periodic motion.If frequency spectrum exists only
With DC component (i.e. as f=0, amplitude is more than predetermined threshold value), then the movement locus is aperiodic motion;If frequency spectrum is not
There are DC component (i.e. as f=0, amplitude are not present), but its frequency spectrum is in frequency f=nf0(wherein, n is integer, f0For base
Frequently amplitude is more than predetermined threshold value at), then the movement locus is periodic motion;If not only there is DC component in frequency spectrum, but also in frequency f
=nf0The amplitude at place is more than predetermined threshold value, then the movement locus is quasi-periodic motion.
For aperiodic motion and periodic motion, directly modeled using existing GMM/GMR methods, for paracycle
Movement then continues subsequent processing.
(2) systematization is decomposed:First with empirical mode decomposition (Empirical Mode Decomposition (EMD)) side
Method extracts offset γ;Notch filter (notch filter) method of recycling carries out harmonic wave separation and obtains ξijχij;Then
Utilize Hilbert transform methods extraction envelope ξij, harmonic component χijFrequency content fijBy Fast Fourier Transform (FFT) (Fast
Fourier transform (FFT)) it determines, in addition, be derived by the calculation formula of phase such as according to harmonic phase principle
Under:
Wherein, tpiAnd kiIt it is peak value and the time of harmonic component, K is the number of peak value, and ω is the angular frequency of the harmonic components
Rate, αjIt is phase compensation factor.
It is offset γ, envelope ξ by a quasi-periodic motion decomposing trajectories by sequence of operations aboveijAnd it is humorous
Wave component χij。
Above-mentioned empirical mode decomposition (Empirical Mode Decomposition (EMD)), notch filter (notch
Filter) and Hilbert transform methods are the prior arts.
(3) model:Gauss hybrid models (Gaussian Mixture Model (GMM)) and Gaussian Mixture return
(Gaussian Mixture Regression (GMR)) is the regression algorithm in a kind of machine learning based on probabilistic model.
GMM/GMR methods regard training data as a sequence of random variables, and assume that it meets normal distribution, consider that same variable faces
Spatial coherence between the nearly moment, to according to the data of multiple teaching come the distribution characteristics of estimated data, and as according to
According to the new continuous movement locus of generation.In short, its input is discrete sample data, output is continuous movement locus.The calculation
The effect of method is the same with the simple regression algorithm of for example polynomial regression etc., can solve continuously asking in machine learning
Topic.But GMM/GMR can overcome the uncertainty in multiple teaching track, reduce the influence caused by bad teaching;Separately
Outside, due to being parameter model, so its model is compact, it is suitble to the movement locus sequence of processing mass data.
Herein, input data is respectively offset γ, envelope ξij, output data is γ ' and ξij' (they when
Between index sequence t ' is become from t).In other words, input data can be modeled and is predicted using GMM/GMR models, export number
According to length can be more than former data length, for example original data length is 200, then can be predicted using GMM/GMR general
Change the data more than 200 length, for example 500 or 1000 etc..
Its periodic motion harmonic componentSince it is with specific formula, therefore its modeling and prediction
A new time series t ' need to be assigned, is obtained
In conclusion obtaining γ ', ξ by the stepij' and χij′。
(4) synthesize:Each component after study is synthesized into a new paracycle according to the equivalent model formula of quasi-periodic motion
U ' is moved, synthesis expression formula is as follows:
Compared with prior art, the present invention having an advantageous effect in that:
(1) modeling algorithm selects GMM/GMR models, can learn fortune paracycle of extensive different application occasion different function
Dynamic rail mark, so that the invention successfully solves the single-function problem of traditional quasi-periodic motion planning system.
(2) extensive reproduction ability, the invention successfully solve traditional quasi-periodic motion planning by force for the track of GMM/GMR
System does not have the problem of extensive predictive ability.
(3) current existing demonstration learning ways are directed to, which can handle the movement locus with more general significance,
Solve the problems, such as Comlex-locus characterization and the study under non-structure environment.
The foregoing is merely presently preferred embodiments of the present invention, is not intended to limit the invention, it is all the present invention spirit and
Within principle, any modification, equivalent replacement, improvement and so on should all be included in the protection scope of the present invention.
Claims (8)
1. a kind of robot quasi-periodic motion demonstrates learning method, which is characterized in that including:
Step 1, when determine robot motion track be quasi-periodic motion when, carry out offset successively to the movement locus and carry
It takes, harmonic wave separation, envelope extraction and phase identification, quasi-periodic motion is decomposed into a series of periodic motions and aperiodic one by one
Movement;
Step 2 returns GMR respectively in periodic motion and aperiodic motion using gauss hybrid models GMM and Gaussian Mixture
Each component carries out modeling study and prediction;
Each component after study is synthesized new quasi-periodic motion by step 3 according to following definition:
Wherein, γ is offset, ξijFor envelope component, χijFor harmonic component, N is the periodic motion of nonlinear combination and aperiodic
The quantity of movement, C be periodic motion Fourier expansion formula in harmonic components number.
2. according to the method described in claim 1, it is characterized in that, before the step 1, further include:
Using the spectrogram characteristic of Fast Fourier Transform (FFT) FFT, determine that robot motion track is periodic motion, aperiodic motion
Or quasi-periodic motion.
3. according to the method described in claim 2, it is characterized in that, the spectrogram using Fast Fourier Transform (FFT) FFT is special
Property, determine that robot motion track is periodic motion, aperiodic motion or quasi-periodic motion, is specifically included:
As f=0, amplitude is more than predetermined threshold value, then the movement locus is aperiodic motion;
As f=0, amplitude is not present, and works as f=nf0When, amplitude is more than predetermined threshold value, then the movement locus is periodic motion;
As f=0 and f=nf0When, amplitude is all higher than predetermined threshold value, then the movement locus is quasi-periodic motion;
Wherein, f is frequency, and n is integer, f0For fundamental frequency.
4. according to the method described in claim 1, it is characterized in that, offset extraction uses the empirical mode decomposition side EMD
Method obtains offset γ.
5. according to the method described in claim 1, it is characterized in that, the harmonic wave separation uses notch filter notch
Filter methods obtain the harmonic component ξ with amplitude modulation only there are one frequency contentijχij。
6. according to the method described in claim 1, it is characterized in that, the envelope extraction is obtained using Hilbert transform methods
Envelope component ξij。
7. according to the method described in claim 1, it is characterized in that, harmonic component χijExpression formula be:
Wherein, ΩiFor the angular frequency of first harmonic ingredient,For the phase angle of jth order harmonic components.
8. the method according to the description of claim 7 is characterized in that the phase angleExpression formula be:
Wherein, K is the number of peak value, tpiAnd kiFor harmonic component χijPeak value and the time, ω be harmonic component χijAngular frequency,
αjFor phase compensation factor.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810295971.1A CN108537268B (en) | 2018-03-30 | 2018-03-30 | Robot quasi-periodic motion demonstration learning method |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810295971.1A CN108537268B (en) | 2018-03-30 | 2018-03-30 | Robot quasi-periodic motion demonstration learning method |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108537268A true CN108537268A (en) | 2018-09-14 |
CN108537268B CN108537268B (en) | 2021-10-15 |
Family
ID=63483044
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810295971.1A Expired - Fee Related CN108537268B (en) | 2018-03-30 | 2018-03-30 | Robot quasi-periodic motion demonstration learning method |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108537268B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111325768A (en) * | 2020-01-31 | 2020-06-23 | 武汉大学 | Free floating target capture method based on 3D vision and simulation learning |
CN113057850A (en) * | 2021-03-11 | 2021-07-02 | 东南大学 | Recovery robot control method based on probability motion primitive and hidden semi-Markov |
CN114601455A (en) * | 2022-05-12 | 2022-06-10 | 电子科技大学 | Motion recognition method based on two-stage neural network |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102004050942B4 (en) * | 2003-10-29 | 2007-03-22 | Tropf, Hermann | Bootstrap method for supervised teach-in of a pattern recognition system |
CN101226398A (en) * | 2008-01-17 | 2008-07-23 | 上海交通大学 | Distributed soldering point quality monitoring system and method |
CN102411776A (en) * | 2011-11-17 | 2012-04-11 | 南京信息工程大学 | Robot visual image segmentation method based on statistics and fractal dimension |
US8904530B2 (en) * | 2008-12-22 | 2014-12-02 | At&T Intellectual Property I, L.P. | System and method for detecting remotely controlled E-mail spam hosts |
CN106414057A (en) * | 2014-01-22 | 2017-02-15 | 3M创新有限公司 | Microoptics for glazing |
-
2018
- 2018-03-30 CN CN201810295971.1A patent/CN108537268B/en not_active Expired - Fee Related
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
DE102004050942B4 (en) * | 2003-10-29 | 2007-03-22 | Tropf, Hermann | Bootstrap method for supervised teach-in of a pattern recognition system |
CN101226398A (en) * | 2008-01-17 | 2008-07-23 | 上海交通大学 | Distributed soldering point quality monitoring system and method |
US8904530B2 (en) * | 2008-12-22 | 2014-12-02 | At&T Intellectual Property I, L.P. | System and method for detecting remotely controlled E-mail spam hosts |
CN102411776A (en) * | 2011-11-17 | 2012-04-11 | 南京信息工程大学 | Robot visual image segmentation method based on statistics and fractal dimension |
CN106414057A (en) * | 2014-01-22 | 2017-02-15 | 3M创新有限公司 | Microoptics for glazing |
Non-Patent Citations (1)
Title |
---|
程红太: "灵长类仿生机器人飞跃轨迹规划及控制策略", 《东北大学学报(自然科学版)》 * |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN111325768A (en) * | 2020-01-31 | 2020-06-23 | 武汉大学 | Free floating target capture method based on 3D vision and simulation learning |
CN113057850A (en) * | 2021-03-11 | 2021-07-02 | 东南大学 | Recovery robot control method based on probability motion primitive and hidden semi-Markov |
CN114601455A (en) * | 2022-05-12 | 2022-06-10 | 电子科技大学 | Motion recognition method based on two-stage neural network |
Also Published As
Publication number | Publication date |
---|---|
CN108537268B (en) | 2021-10-15 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Alexanderson et al. | Style‐controllable speech‐driven gesture synthesis using normalising flows | |
Kadri et al. | Operator-valued kernels for learning from functional response data | |
Calinon et al. | Stochastic gesture production and recognition model for a humanoid robot | |
Wang et al. | Wind speed forecasting based on the hybrid ensemble empirical mode decomposition and GA-BP neural network method | |
CN108537268A (en) | A kind of robot quasi-periodic motion demonstration learning method | |
Feng et al. | Short-term traffic flow prediction based on wavelet function and extreme learning machine | |
Dong et al. | Short-term wind-speed forecasting based on multiscale mathematical morphological decomposition, K-means clustering, and stacked denoising autoencoders | |
Haque et al. | A new strategy for wind speed forecasting using hybrid intelligent models | |
Afanasieva et al. | Time series forecasting using fuzzy techniques | |
Abiyev et al. | Differential evaluation learning of fuzzy wavelet neural networks for stock price prediction | |
Abiyev | Fuzzy wavelet neural network for prediction of electricity consumption | |
Afanasieva et al. | Selection of time series forecasting model, using a combination of linguistic and numerical criteria | |
Zhao et al. | A mid-long term load forecasting model based on improved grey theory | |
Wang et al. | A wind speed prediction method based on improved empirical mode decomposition and support vector machine | |
Eissa et al. | Assessment of wind power prediction using hybrid method and comparison with different models | |
Krishna et al. | Complexity and performance evaluation of segmented and recursive reinforcement learning | |
Boukheddimi et al. | Robot dance generation with music based trajectory optimization | |
Zhou et al. | Hierarchical learning recurrent neural networks for 3D motion synthesis | |
Inamura et al. | Geometric proto-symbol manipulation towards language-based motion pattern synthesis and recognition | |
Milenkovic et al. | Kalman filtering applied to forecasting the demand for railway passenger services | |
Li et al. | Short-term traffic flow forecasting based on SVR | |
CN113506360A (en) | Virtual character expression driving method and system | |
Wang et al. | Mutual information and non-fixed ANNs for daily peak load forecasting | |
Bhagwan et al. | Software modules clustering: an effective approach for reusability | |
Chatterjee et al. | Machine Learning Based Prediction of Energy Consumption |
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 | ||
CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20211015 |
|
CF01 | Termination of patent right due to non-payment of annual fee |