CN108491767A - Autonomous roll response method, system and manipulator based on Online Video perception - Google Patents
Autonomous roll response method, system and manipulator based on Online Video perception Download PDFInfo
- Publication number
- CN108491767A CN108491767A CN201810182168.7A CN201810182168A CN108491767A CN 108491767 A CN108491767 A CN 108491767A CN 201810182168 A CN201810182168 A CN 201810182168A CN 108491767 A CN108491767 A CN 108491767A
- Authority
- CN
- China
- Prior art keywords
- manipulator
- action
- video
- sequence
- anticipation
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/20—Movements or behaviour, e.g. gesture recognition
- G06V40/28—Recognition of hand or arm movements, e.g. recognition of deaf sign language
-
- B—PERFORMING OPERATIONS; TRANSPORTING
- B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
- B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
- B25J9/00—Programme-controlled manipulators
- B25J9/16—Programme controls
- B25J9/1628—Programme controls characterised by the control loop
- B25J9/163—Programme controls characterised by the control loop learning, adaptive, model based, rule based expert control
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/40—Scenes; Scene-specific elements in video content
- G06V20/41—Higher-level, semantic clustering, classification or understanding of video scenes, e.g. detection, labelling or Markovian modelling of sport events or news items
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V40/00—Recognition of biometric, human-related or animal-related patterns in image or video data
- G06V40/10—Human or animal bodies, e.g. vehicle occupants or pedestrians; Body parts, e.g. hands
- G06V40/107—Static hand or arm
Abstract
The present invention provides a kind of autonomous roll response method and system based on Online Video perception, this method and system are applied to manipulator, specially obtain the video actions sequence at user partial position;By in video actions sequence present frame and forward frame information input shift to an earlier date in trained prediction model, obtain the anticipation result to part;The targeted attitude that anticipation result is used to that manipulator to be made to realize for anticipation action.This programme puts forth effort on the fine identification of the action of the part to user in interactive process, attempt in the action of the part of user from starting to during end, carry out dynamic rolling anticipation, and then pointedly the action of manipulator is adjusted into Mobile state, quick, smooth manipulator response is ultimately formed, to improve the interaction success rate of entire interactive process.
Description
Technical field
The present invention relates to robotic technology fields, more particularly to a kind of autonomous roll response based on Online Video perception
Method, system and manipulator.
Background technology
Manipulator have higher degree of freedom and action execute speed, can be effectively carried out interacted with people it is various bionical
Gesture executes scheduled action, is the effective carrier for carrying out various actions.With manipulator production cost it is continuous reduction,
Execution efficiency is increasingly promoted, and the application depth and range of manipulator also constantly extend.
In terms of people and manipulator interaction, one important and common scene is that manipulator is moved according to human hands are different
Make, carries out targetedly gesture and convert, form interesting human-computer interaction, or even complete specific interactive task.Such as people and machine
Finger-guessing game match between tool hand, manipulator dynamically adjust itself posture while perceive human body gesture motion, formed it is quick,
Natural reply action.Above-mentioned scene can carry out and popularize in places such as recreation ground, market or even science and technology centers, to rich
The rich cultural life of the people, and drive the development and transition of people's livelihood recreation industry, Popular Science Education industry.
During above-mentioned human-computer interaction, to autonomous sensing capability, the strain rate effect ability of arm-and-hand system propose compared with
High requirement.
The inventors of the present application found that being followed up according to the gesture motion of external world's perception for manipulator, current method is
Human hands are acted and carry out one-off recognition work, and corresponding action is executed according to recognition result, whole process be it is static,
Disposably, once error causes the interaction success rate of manipulator and user to be lower without the chance of dynamic adjustment.
Invention content
In view of this, the present invention provides a kind of autonomous roll response method, system and machines based on Online Video perception
Tool hand, the interaction success rate for improving manipulator and user.
A kind of autonomous roll response method based on Online Video perception, is applied to manipulator, the autonomous roll response
Method includes step:
In the entire action process of user, the video actions sequence of the part of user is obtained;
By in the video actions sequence present frame and forward frame information input shift to an earlier date in trained prediction model, obtain
To the anticipation result to the part;The anticipation result is for making the manipulator carry out rolling tune to itself action
It is whole, to realize the targeted attitude for anticipation action.
Optionally, the video actions sequence for obtaining user partial position, including:
The video actions sequence of the part is obtained using two-dimensional color camera or gray scale camera.
Optionally, further include step:
The current pose of the manipulator is adjusted according to the anticipation result;
When the anticipation result confidence level reaches preset confidence threshold value, the current pose is adjusted to the mesh
Mark posture.
Optionally, the prediction model is obtained by following step:
Acquire the video sequence of a variety of parts;
The video sequence is handled by presetting method, obtains multiple training datas;
Preset function is trained using the method for supervised learning, and using the multiple training data, is obtained described
Prediction model.
A kind of autonomous roll response system based on Online Video perception, is applied to manipulator, the autonomous roll response
System includes:
Retrieval module, in the entire action process of user, obtaining the video actions of the part of user
Sequence;
Action anticipation module, for by the video actions sequence present frame and forward frame information input train in advance
In good prediction model, the anticipation result to the part is obtained;The anticipation result is for making the manipulator to certainly
Body action carries out rolling adjustment, to realize the targeted attitude for anticipation action.
Optionally, the retrieval module specifically utilizes two-dimensional color camera or gray scale camera to obtain the part portion
The video actions sequence of position.
Optionally, further include:
The first adjustment module, the current pose for adjusting the manipulator according to the anticipation result;
Second adjustment module, for when the anticipation result confidence level reaches preset confidence threshold value, working as by described in
Preceding pose adjustment is the targeted attitude.
Optionally, further include model training module for training the prediction model, the model training module includes:
Sequence acquisition unit, the video sequence for acquiring a variety of parts;
Data processing unit obtains multiple training datas for being handled by presetting method the video sequence;
Function training unit, for the method using supervised learning, and using the multiple training data to preset function
It is trained, obtains the prediction model.
A kind of manipulator, which is characterized in that including autonomous roll response system as described above.
It can be seen from the above technical proposal that the present invention provides a kind of autonomous roll responses based on Online Video perception
Method and system, this method and system are applied to manipulator, specially obtain the video actions sequence at user partial position;It will regard
Present frame and forward frame information input in frequency action sequence shift to an earlier date in trained prediction model, obtain to the pre- of part
Sentence result;The targeted attitude that anticipation result is used to that manipulator to be made to realize for anticipation action.This programme puts forth effort on human-computer interaction
To the fine identification of the action of the part of user in journey, it is intended to act the mistake from starting to end in the part of user
Cheng Zhong carries out dynamic rolling anticipation, and then is pointedly adjusted into Mobile state to the action of manipulator, ultimately forms quick, flat
Sliding manipulator response, to improve the interaction success rate of entire interactive process.
Description of the drawings
In order to more clearly explain the embodiment of the invention or the technical proposal in the existing technology, to embodiment or will show below
There is attached drawing needed in technology description to be briefly described, it should be apparent that, the accompanying drawings in the following description is only this
Some embodiments of invention for those of ordinary skill in the art without creative efforts, can be with
Obtain other attached drawings according to these attached drawings.
The step of Fig. 1 is a kind of autonomous roll response method perceived based on Online Video provided by the embodiments of the present application is flowed
Cheng Tu;
The step of Fig. 2 is another autonomous roll response method perceived based on Online Video provided by the embodiments of the present application
Flow chart;
Fig. 3 is a kind of step flow chart of the training method of prediction model provided in an embodiment of the present invention;
Fig. 4 is a kind of structural frames of autonomous roll response system based on Online Video perception provided by the embodiments of the present application
Figure;
Fig. 5 is the structure of another autonomous roll response system based on Online Video perception provided by the embodiments of the present application
Block diagram;
Fig. 6 is the structure of another autonomous roll response system based on Online Video perception provided by the embodiments of the present application
Block diagram.
Specific implementation mode
Following will be combined with the drawings in the embodiments of the present invention, and technical solution in the embodiment of the present invention carries out clear, complete
Site preparation describes, it is clear that described embodiments are only a part of the embodiments of the present invention, instead of all the embodiments.It is based on
Embodiment in the present invention, it is obtained by those of ordinary skill in the art without making creative efforts every other
Embodiment shall fall within the protection scope of the present invention.
Embodiment one
The step of Fig. 1 is a kind of autonomous roll response method perceived based on Online Video provided by the embodiments of the present application is flowed
Cheng Tu.
Autonomous rolling developing method provided in this embodiment based on Online Video perception is applied to manipulator, specifically
Applied in the control device of the manipulator, manipulator is improved for being realized during being controlled manipulator in control device
With the purpose of the interaction success rate of user.As shown in Figure 1, the autonomous roll response method of the present embodiment specifically includes following step
Suddenly:
S101:Obtain the video actions sequence at user partial position.
Because this programme purpose is that manipulator is made to act for the part especially hand of user, therefore this
In obtain the part of user, the especially action sequence of hand first, specifically obtain the video actions sequence of hand
Row.
It is specifically to be regarded using two-dimensional color camera or gray scale camera to obtain the part of user in the present embodiment
Frequency action sequence.It, can for scheme in compared with the prior art using the 3 d pose of the hand of depth camera capture people
Reduce cost, and can avoid depth camera caused by Image Acquisition rate is relatively low can not fast Acquisition high speed gesture ask
Topic.
S102:It is prejudged according to the action of video actions sequence pair part.
After obtaining above-mentioned video actions sequence, using advance trained prediction model to the part of user, tool
Body is to predict the action of the hand of user.Specifically by above-mentioned video actions sequence inputting prediction model,
Video actions sequence is calculated using prediction model, to obtain corresponding prediction result.
Specific in the present embodiment, because the hand of user is in continuous variation, therefore, it is possible to only be observed currently
To hand partial act in the case of, the molar behavior of hand is prejudged, that is, is prejudged after the hand makes complete action
Final carriage, if determination is " scissors ", " hammer " or " cloth ".
The final carriage of the prediction result, that is, hand motion, for enabling manipulator to realize the mesh for the prediction result
Mark posture, the final carriage for the hand motion corresponding to the user that targeted attitude can execute for manipulator.
It can be seen from the above technical proposal that present embodiments providing a kind of autonomous rolling sound perceived based on Online Video
Induction method, this method are applied to manipulator, specially obtain the video actions sequence at user partial position;By video actions sequence
In present frame and forward frame information input shift to an earlier date in trained prediction model, obtain the anticipation result to part;In advance
Result is sentenced for making manipulator realize the targeted attitude for anticipation action.This programme is put forth effort in interactive process to user
Part action fine identification, it is intended in the action of the part of user from starting to during end, carry out
Dynamic rolling prejudges, and then is pointedly adjusted into Mobile state to the action of manipulator, ultimately forms quick, smooth manipulator
Response, to improve the interaction success rate of entire interactive process.
Further include following steps in the present embodiment, as shown in Figure 2:
S103:The current pose of manipulator is adjusted according to anticipation result.
After obtaining prediction result, based on current prediction result, and based on the anticipation to each possible gesture option
Confidence level as a result, Optimal Decision-making manipulator targeted attitude, and correspondingly adjustment Current mechanical hand posture leaned on to the targeted attitude
Hold together.Current posture is adjusted according to anticipation result, close to the targeted attitude.
It should be noted that the full standard action that current goal posture is often manipulator itself (such as goes out and " cuts
Knife ") process a certain posture in centre (such as two fingers slightly " are bent ", this is a middle attitude for " scissors ").This
One way, had both improved the response speed that manipulator acts human hands, i.e., human body do the midway that completely acts immediately to
Go out response.It in turn ensures that its set complete action will not disposably be finished by manipulator rashly, avoids due to observation
Human action is imperfect and causes it to provide errored response and moves.
In the present embodiment the determination mechanism of above-mentioned targeted attitude be based in advance structure maximum confidence action anticipation result with
The rule list realization of posture is responded, as shown in table 1.
Human body hand practical posture online | Maximum confidence prejudges result | Manipulator responds targeted attitude |
It clenches fist | Stone | Five fingers open slightly |
Five fingers open slightly | Cloth | Two fingers open slightly |
Two fingers open slightly | Scissors | It clenches fist |
Table 1
S104:When anticipation result reaches confidence threshold value, current pose is adjusted.
With observation video frame number accumulation, anticipation result gradually converges to final action, when dynamic to some human hands
At the time of work judges that confidence level is sufficiently high or a set of hand motion of human body will be finished, final corresponding actions are determined simultaneously
Completely executed.
The action of above-mentioned human hands is online roll anticipation and according to the process of anticipation result dynamic adjusting machine tool hand posture such as
Shown in table 2.
Table 2
Prediction model in the present embodiment is trained to obtain by following step, as shown in Figure 3:
S5001:Acquire the video sequence of a variety of activities.
Specifically, the video sequence for acquiring multiclass hand instruction action acts for every class, acquires and preserve more parts not
With the video sequence data of external shooting environmental and different action executors.In the present embodiment, action classification include " stone ",
Three kinds of action sequences of " scissors ", " cloth ", each action sequence midway contains from the initial attitude clenched fist to arms swing, to micro-
It is micro- spread one's fingers, certain fingers open etc. part son action, ultimately form complete action sequence.
S5002:By obtaining multiple training datas to the processing of video sequence.
To the training video sequence of above-mentioned steps acquisition, using sides such as affine transformation, contrast stretching, hand skin color transformation
Formula carries out further data enhancing, expands training data, obtains multiple training datas, constitute corresponding training dataset, to carry
The universality of high training set.
S5003:It is trained to obtain prediction model using multiple training datas.
After obtaining corresponding multiple training datas, temporally list entries grader frame by frame from front to back, and give simultaneously
Fixed corresponding action classification label so that sorter model acts the dynamic of integrity degree in input different length sequence, and difference
When making sequence, corresponding complete action can be prejudged and be identified.Ideally, the sequence frame observed with model
Increase, observation is more complete, and action anticipation accuracy rate is higher.
In the present embodiment, we are used as using depth of round neural network (recurrent neural network) and are connect
It is incremented by video sequence by dynamic and its corresponding complete action classification marks fallout predictor way of realization as input.The network passes through
The loop structure of orientation is introduced, it being capable of associated Series Modeling and forecasting problem before and after the processing.Here pass through the side of supervised learning
Formula trains network parameter using category label as supervision message, so that network has the video sequence of receiving portion observation
Under the conditions of action prediction ability, be suitable for the present embodiment the problem of scene.
Embodiment two
Fig. 4 is a kind of structural frames of autonomous roll response system based on Online Video perception provided by the embodiments of the present application
Figure.
Autonomous rolling toning system provided in this embodiment based on Online Video perception is applied to manipulator, specifically
Applied in the control device of the manipulator, manipulator is improved for being realized during being controlled manipulator in control device
With the purpose of the interaction success rate of user.As shown in figure 4, the autonomous roll response system of the present embodiment specifically includes retrieval
Module 10 and action anticipation module 20.
Retrieval module is used to obtain the video actions sequence at user partial position.
Because this programme purpose is that manipulator is made to act for the part especially hand of user, therefore this
In obtain the part of user, the especially action sequence of hand first, specifically obtain the video actions sequence of hand
Row.
The module is specifically that the local portion of user is obtained using two-dimensional color camera or gray scale camera in the present embodiment
The video actions sequence of position.Compared with the prior art in using depth camera capture people hand 3 d pose scheme come
It says, cost can be reduced, and can avoid depth camera can not fast Acquisition high speed hand caused by Image Acquisition rate is relatively low
The problem of gesture.
Action anticipation module according to the action of video actions sequence pair part for being prejudged.
After obtaining above-mentioned video actions sequence, using advance trained prediction model to the part of user, tool
Body is to predict the action of the hand of user.Specifically by above-mentioned video actions sequence inputting prediction model,
Video actions sequence is calculated using prediction model, to obtain corresponding prediction result.
Specific in the present embodiment, because the hand of user is in continuous variation, therefore, it is possible to only be observed currently
To hand partial act in the case of, the molar behavior of hand is prejudged, that is, is prejudged after the hand makes complete action
Final carriage, if determination is " scissors ", " hammer " or " cloth ".
The final carriage of the prediction result, that is, hand motion, for enabling manipulator to realize the mesh for the prediction result
Mark posture, the final carriage for the hand motion corresponding to the user that targeted attitude can execute for manipulator.
It can be seen from the above technical proposal that present embodiments providing a kind of autonomous rolling sound perceived based on Online Video
System is answered, which is applied to manipulator, specially obtains the video actions sequence at user partial position;By video actions sequence
In present frame and forward frame information input shift to an earlier date in trained prediction model, obtain the anticipation result to part;In advance
Result is sentenced for making manipulator realize the targeted attitude for anticipation action.This programme is put forth effort in interactive process to user
Part action fine identification, it is intended in the action of the part of user from starting to during end, carry out
Dynamic rolling prejudges, and then is pointedly adjusted into Mobile state to the action of manipulator, ultimately forms quick, smooth manipulator
Response, to improve the interaction success rate of entire interactive process.
System in the present embodiment further includes the first adjustment module 30 and second adjustment module 40, as shown in Figure 5:
The first adjustment module is used to be adjusted the current pose of manipulator according to anticipation result.
After obtaining prediction result, based on current prediction result, and based on the anticipation to each possible gesture option
Confidence level as a result, Optimal Decision-making manipulator targeted attitude, and correspondingly adjustment Current mechanical hand posture leaned on to the targeted attitude
Hold together.Current posture is adjusted according to anticipation result, close to the targeted attitude.
It should be noted that the full standard action that current goal posture is often manipulator itself (such as goes out and " cuts
Knife ") process a certain posture in centre (such as two fingers slightly " are bent ", this is a middle attitude for " scissors ").This
One way, had both improved the response speed that manipulator acts human hands, i.e., human body do the midway that completely acts immediately to
Go out response.It in turn ensures that its set complete action will not disposably be finished by manipulator rashly, avoids due to observation
Human action is imperfect and causes it to provide errored response and moves.
In the present embodiment the determination mechanism of above-mentioned targeted attitude be based in advance structure maximum confidence action anticipation result with
The rule list realization of posture is responded, as shown in table 1.
Human body hand practical posture online | Maximum confidence prejudges result | Manipulator responds targeted attitude |
It clenches fist | Stone | Five fingers open slightly |
Five fingers open slightly | Cloth | Two fingers open slightly |
Two fingers open slightly | Scissors | It clenches fist |
Table 1
Second adjustment module is used to, when anticipation result reaches confidence threshold value, adjust current pose.
With observation video frame number accumulation, anticipation result gradually converges to final action, when dynamic to some human hands
At the time of work judges that confidence level is sufficiently high or a set of hand motion of human body will be finished, final corresponding actions are determined simultaneously
Completely executed.
The action of above-mentioned human hands is online roll anticipation and according to the process of anticipation result dynamic adjusting machine tool hand posture such as
Shown in table 2.
Table 2
System in the present embodiment further includes model training module 50, and the module is for training prediction model, the prediction mould
Type specifically includes sequence acquisition unit 51, data processing unit 52 and function training unit 53, as shown in Figure 6:
Sequence acquisition unit is used to acquire the video sequence of a variety of activities.
Specifically, the video sequence for acquiring multiclass hand instruction action acts for every class, acquires and preserve more parts not
With the video sequence data of external shooting environmental and different action executors.In the present embodiment, action classification include " stone ",
Three kinds of action sequences of " scissors ", " cloth ", each action sequence midway contains from the initial attitude clenched fist to arms swing, to micro-
It is micro- spread one's fingers, certain fingers open etc. part son action, ultimately form complete action sequence.
Data processing unit is used for by obtaining multiple training datas to the processing of video sequence.
To the training video sequence of above-mentioned steps acquisition, using sides such as affine transformation, contrast stretching, hand skin color transformation
Formula carries out further data enhancing, expands training data, obtains multiple training datas, constitute corresponding training dataset, to carry
The universality of high training set.
Function training unit using multiple training datas for being trained to obtain prediction model.
After obtaining corresponding multiple training datas, temporally list entries grader frame by frame from front to back, and give simultaneously
Fixed corresponding action classification label so that sorter model acts the dynamic of integrity degree in input different length sequence, and difference
When making sequence, corresponding complete action can be prejudged and be identified.Ideally, the sequence frame observed with model
Increase, observation is more complete, and action anticipation accuracy rate is higher.
In the present embodiment, we are used as using depth of round neural network (recurrent neural network) and are connect
It is incremented by video sequence by dynamic and its corresponding complete action classification marks fallout predictor way of realization as input.The network passes through
The loop structure of orientation is introduced, it being capable of associated Series Modeling and forecasting problem before and after the processing.Here pass through the side of supervised learning
Formula trains network parameter using category label as supervision message, so that network has the video sequence of receiving portion observation
Under the conditions of action prediction ability, be suitable for the present embodiment the problem of scene.
Embodiment three
A kind of manipulator is present embodiments provided, which is a complete system, and minimum includes corresponding movement
Component and control device, the control device are provided with the self-service rolling corresponding system provided in embodiment as above.This is independently rolled
Dynamic response system is specifically used for obtaining the video actions sequence at user partial position;By the present frame in video actions sequence with before
Shift to an earlier date in trained prediction model to frame information input, obtains the anticipation result to part;Anticipation result is for making machine
Tool hand realizes the targeted attitude for anticipation action.This programme puts forth effort on the dynamic of the part to user in interactive process
The fine identification made, it is intended in the action of the part of user from starting to during end, dynamic rolling anticipation is carried out, into
And pointedly the action of manipulator is adjusted into Mobile state, quick, smooth manipulator response is ultimately formed, so as to carry
The interaction success rate of high entire interactive process.
Each embodiment in this specification is described in a progressive manner, the highlights of each of the examples are with
The difference of other embodiment, the same or similar parts between the embodiments can be referred to each other.
It should be understood by those skilled in the art that, the embodiment of the embodiment of the present invention can be provided as method, apparatus or calculate
Machine program product.Therefore, the embodiment of the present invention can be used complete hardware embodiment, complete software embodiment or combine software and
The form of the embodiment of hardware aspect.Moreover, the embodiment of the present invention can be used one or more wherein include computer can
With in the computer-usable storage medium (including but not limited to magnetic disk storage, CD-ROM, optical memory etc.) of program code
The form of the computer program product of implementation.
The embodiment of the present invention be with reference to according to the method for the embodiment of the present invention, terminal device (system) and computer program
The flowchart and/or the block diagram of product describes.It should be understood that flowchart and/or the block diagram can be realized by computer program instructions
In each flow and/or block and flowchart and/or the block diagram in flow and/or box combination.These can be provided
Computer program instructions are set to all-purpose computer, special purpose computer, Embedded Processor or other programmable data processing terminals
Standby processor is to generate a machine so that is held by the processor of computer or other programmable data processing terminal equipments
Capable instruction generates for realizing in one flow of flow chart or multiple flows and/or one box of block diagram or multiple boxes
The device of specified function.
These computer program instructions, which may also be stored in, can guide computer or other programmable data processing terminal equipments
In computer-readable memory operate in a specific manner so that instruction stored in the computer readable memory generates packet
The manufacture of command device is included, which realizes in one flow of flow chart or multiple flows and/or one side of block diagram
The function of being specified in frame or multiple boxes.
These computer program instructions can be also loaded into computer or other programmable data processing terminal equipments so that
Series of operation steps are executed on computer or other programmable terminal equipments to generate computer implemented processing, thus
The instruction executed on computer or other programmable terminal equipments is provided for realizing in one flow of flow chart or multiple flows
And/or in one box of block diagram or multiple boxes specify function the step of.
Although the preferred embodiment of the embodiment of the present invention has been described, once a person skilled in the art knows bases
This creative concept, then additional changes and modifications can be made to these embodiments.So the following claims are intended to be interpreted as
Including preferred embodiment and fall into all change and modification of range of embodiment of the invention.
Finally, it is to be noted that, herein, relational terms such as first and second and the like be used merely to by
One entity or operation are distinguished with another entity or operation, without necessarily requiring or implying these entities or operation
Between there are any actual relationship or orders.Moreover, the terms "include", "comprise" or its any other variant meaning
Covering non-exclusive inclusion, so that process, method, article or terminal device including a series of elements not only wrap
Those elements are included, but also include other elements that are not explicitly listed, or further include for this process, method, article
Or the element that terminal device is intrinsic.In the absence of more restrictions, being wanted by what sentence "including a ..." limited
Element, it is not excluded that there is also other identical elements in process, method, article or the terminal device including the element.
Technical solution provided by the present invention is described in detail above, specific case used herein is to this hair
Bright principle and embodiment is expounded, the explanation of above example is only intended to help understand the present invention method and its
Core concept;Meanwhile for those of ordinary skill in the art, according to the thought of the present invention, in specific implementation mode and application
There will be changes in range, in conclusion the content of the present specification should not be construed as limiting the invention.
Claims (9)
1. a kind of autonomous roll response method based on Online Video perception, is applied to manipulator, which is characterized in that described autonomous
Roll response method includes step:
In the entire action process of user, the video actions sequence of the part of user is obtained;
By in the video actions sequence present frame and forward frame information input shift to an earlier date in trained prediction model, obtain pair
The anticipation result of the part;The anticipation result carries out rolling adjustment for making the manipulator act itself, with
Realize the targeted attitude for anticipation action.
2. autonomous roll response method as described in claim 1, which is characterized in that the video for obtaining user partial position
Action sequence, including:
The video actions sequence of the part is obtained using two-dimensional color camera or gray scale camera.
3. autonomous roll response method as described in claim 1, which is characterized in that further include step:
The current pose of the manipulator is adjusted according to the anticipation result;
When the anticipation result confidence level reaches preset confidence threshold value, the current pose is adjusted to the target appearance
State.
4. such as the autonomous roll response method of claims 1 to 3 any one of them, which is characterized in that the prediction model passes through
Following step obtains:
Acquire the video sequence of a variety of parts;
The video sequence is handled by presetting method, obtains multiple training datas;
Preset function is trained using the method for supervised learning, and using the multiple training data, obtains the prediction
Model.
5. a kind of autonomous roll response system based on Online Video perception, is applied to manipulator, which is characterized in that described autonomous
Roll response system includes:
Retrieval module, in the entire action process of user, obtaining the video actions sequence of the part of user;
Action anticipation module, for by the video actions sequence present frame and forward frame information input it is trained in advance
In prediction model, the anticipation result to the part is obtained;The anticipation result is for keeping the manipulator dynamic to itself
Rolling adjustment is carried out, to realize the targeted attitude for anticipation action.
6. autonomous roll response system as claimed in claim 5, which is characterized in that the retrieval module specifically utilizes two
Dimension color camera or gray scale camera obtain the video actions sequence of the part.
7. autonomous roll response system as claimed in claim 5, which is characterized in that further include:
The first adjustment module, the current pose for adjusting the manipulator according to the anticipation result;
Second adjustment module, for when the anticipation result confidence level reaches preset confidence threshold value, by the current appearance
State is adjusted to the targeted attitude.
8. such as the autonomous roll response system of claim 5~7 any one of them, which is characterized in that further include for training
The model training module of prediction model is stated, the model training module includes:
Sequence acquisition unit, the video sequence for acquiring a variety of parts;
Data processing unit obtains multiple training datas for being handled by presetting method the video sequence;
Function training unit carries out preset function for the method using supervised learning, and using the multiple training data
Training, obtains the prediction model.
9. a kind of manipulator, which is characterized in that including the autonomous roll response system of such as claim 5~8 any one of them.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810182168.7A CN108491767B (en) | 2018-03-06 | 2018-03-06 | Autonomous rolling response method and system based on online video perception and manipulator |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201810182168.7A CN108491767B (en) | 2018-03-06 | 2018-03-06 | Autonomous rolling response method and system based on online video perception and manipulator |
Publications (2)
Publication Number | Publication Date |
---|---|
CN108491767A true CN108491767A (en) | 2018-09-04 |
CN108491767B CN108491767B (en) | 2022-08-09 |
Family
ID=63341354
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201810182168.7A Active CN108491767B (en) | 2018-03-06 | 2018-03-06 | Autonomous rolling response method and system based on online video perception and manipulator |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN108491767B (en) |
Citations (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009045714A (en) * | 2007-08-22 | 2009-03-05 | Ntn Corp | Method and device for teaching attitude of robot arm |
US20120130355A1 (en) * | 2009-07-31 | 2012-05-24 | Dexterite Surgical | Manipulator with manual hold and comfortable articulation |
CN102831439A (en) * | 2012-08-15 | 2012-12-19 | 深圳先进技术研究院 | Gesture tracking method and gesture tracking system |
CN103240746A (en) * | 2013-04-18 | 2013-08-14 | 塔米智能科技(北京)有限公司 | Finger-guessing robot with image recognition system, and gesture recognition method |
CN103419204A (en) * | 2012-05-22 | 2013-12-04 | 林其禹 | Finger guessing game robot system |
JP2014104527A (en) * | 2012-11-27 | 2014-06-09 | Seiko Epson Corp | Robot system, program, production system, and robot |
US20140161310A1 (en) * | 2012-12-07 | 2014-06-12 | Pixart Imaging Inc. | Device and Method for Determining Gesture and Operation Method of Gesture Determining Device |
EP2803202A1 (en) * | 2012-12-11 | 2014-11-19 | Unify GmbH & Co. KG | Method of processing video data, device, computer program product, and data construct |
CN104915009A (en) * | 2015-06-25 | 2015-09-16 | 深圳先进技术研究院 | Gesture prediction method and system |
CN104992171A (en) * | 2015-08-04 | 2015-10-21 | 易视腾科技有限公司 | Method and system for gesture recognition and man-machine interaction based on 2D video sequence |
WO2015182561A1 (en) * | 2014-05-27 | 2015-12-03 | 並木精密宝石株式会社 | Manipulator |
US9381426B1 (en) * | 2013-03-15 | 2016-07-05 | University Of Central Florida Research Foundation, Inc. | Semi-automated digital puppetry control |
CN105818129A (en) * | 2016-04-12 | 2016-08-03 | 华南理工大学 | Humanoid hand control system based on data glove |
CN105930784A (en) * | 2016-04-15 | 2016-09-07 | 济南大学 | Gesture recognition method |
CN106022294A (en) * | 2016-06-01 | 2016-10-12 | 北京光年无限科技有限公司 | Intelligent robot-oriented man-machine interaction method and intelligent robot-oriented man-machine interaction device |
CN106272409A (en) * | 2016-08-03 | 2017-01-04 | 北京航空航天大学 | Mechanical arm control method based on gesture identification and system |
US20170064212A1 (en) * | 2015-08-31 | 2017-03-02 | Orcam Technologies Ltd. | Systems and methods for identifying exposure to a recognizable item |
WO2017156835A1 (en) * | 2016-03-18 | 2017-09-21 | 深圳大学 | Smart method and system for body building posture identification, assessment, warning and intensity estimation |
CN107291232A (en) * | 2017-06-20 | 2017-10-24 | 深圳市泽科科技有限公司 | A kind of somatic sensation television game exchange method and system based on deep learning and big data |
US20170334066A1 (en) * | 2016-05-20 | 2017-11-23 | Google Inc. | Machine learning methods and apparatus related to predicting motion(s) of object(s) in a robot's environment based on image(s) capturing the object(s) and based on parameter(s) for future robot movement in the environment |
CN107688391A (en) * | 2017-09-01 | 2018-02-13 | 广州大学 | A kind of gesture identification method and device based on monocular vision |
-
2018
- 2018-03-06 CN CN201810182168.7A patent/CN108491767B/en active Active
Patent Citations (21)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
JP2009045714A (en) * | 2007-08-22 | 2009-03-05 | Ntn Corp | Method and device for teaching attitude of robot arm |
US20120130355A1 (en) * | 2009-07-31 | 2012-05-24 | Dexterite Surgical | Manipulator with manual hold and comfortable articulation |
CN103419204A (en) * | 2012-05-22 | 2013-12-04 | 林其禹 | Finger guessing game robot system |
CN102831439A (en) * | 2012-08-15 | 2012-12-19 | 深圳先进技术研究院 | Gesture tracking method and gesture tracking system |
JP2014104527A (en) * | 2012-11-27 | 2014-06-09 | Seiko Epson Corp | Robot system, program, production system, and robot |
US20140161310A1 (en) * | 2012-12-07 | 2014-06-12 | Pixart Imaging Inc. | Device and Method for Determining Gesture and Operation Method of Gesture Determining Device |
EP2803202A1 (en) * | 2012-12-11 | 2014-11-19 | Unify GmbH & Co. KG | Method of processing video data, device, computer program product, and data construct |
US9381426B1 (en) * | 2013-03-15 | 2016-07-05 | University Of Central Florida Research Foundation, Inc. | Semi-automated digital puppetry control |
CN103240746A (en) * | 2013-04-18 | 2013-08-14 | 塔米智能科技(北京)有限公司 | Finger-guessing robot with image recognition system, and gesture recognition method |
WO2015182561A1 (en) * | 2014-05-27 | 2015-12-03 | 並木精密宝石株式会社 | Manipulator |
CN104915009A (en) * | 2015-06-25 | 2015-09-16 | 深圳先进技术研究院 | Gesture prediction method and system |
CN104992171A (en) * | 2015-08-04 | 2015-10-21 | 易视腾科技有限公司 | Method and system for gesture recognition and man-machine interaction based on 2D video sequence |
US20170064212A1 (en) * | 2015-08-31 | 2017-03-02 | Orcam Technologies Ltd. | Systems and methods for identifying exposure to a recognizable item |
WO2017156835A1 (en) * | 2016-03-18 | 2017-09-21 | 深圳大学 | Smart method and system for body building posture identification, assessment, warning and intensity estimation |
CN105818129A (en) * | 2016-04-12 | 2016-08-03 | 华南理工大学 | Humanoid hand control system based on data glove |
CN105930784A (en) * | 2016-04-15 | 2016-09-07 | 济南大学 | Gesture recognition method |
US20170334066A1 (en) * | 2016-05-20 | 2017-11-23 | Google Inc. | Machine learning methods and apparatus related to predicting motion(s) of object(s) in a robot's environment based on image(s) capturing the object(s) and based on parameter(s) for future robot movement in the environment |
CN106022294A (en) * | 2016-06-01 | 2016-10-12 | 北京光年无限科技有限公司 | Intelligent robot-oriented man-machine interaction method and intelligent robot-oriented man-machine interaction device |
CN106272409A (en) * | 2016-08-03 | 2017-01-04 | 北京航空航天大学 | Mechanical arm control method based on gesture identification and system |
CN107291232A (en) * | 2017-06-20 | 2017-10-24 | 深圳市泽科科技有限公司 | A kind of somatic sensation television game exchange method and system based on deep learning and big data |
CN107688391A (en) * | 2017-09-01 | 2018-02-13 | 广州大学 | A kind of gesture identification method and device based on monocular vision |
Non-Patent Citations (4)
Title |
---|
YEN-TING CHEN 等: "Multiple-angle Hand Gesture Recognition by Fusing SVM Classifiers", 《2007 IEEE INTERNATIONAL CONFERENCE ON AUTOMATION SCIENCE AND ENGINEERING》 * |
何瑞青: "助老助残移动机器人手势交互控制研究", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
程坤: "基于嵌入式平台的智能移动机器人系统设计", 《中国优秀硕士学位论文全文数据库 信息科技辑》 * |
齐静 等: "机器人视觉手势交互技术研究进展", 《机器人》 * |
Also Published As
Publication number | Publication date |
---|---|
CN108491767B (en) | 2022-08-09 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Li et al. | Human-centered reinforcement learning: A survey | |
KR102425578B1 (en) | Method and apparatus for recognizing an object | |
CN106139564B (en) | Image processing method and device | |
CN108983979A (en) | A kind of gesture tracking recognition methods, device and smart machine | |
CN111191599B (en) | Gesture recognition method, device, equipment and storage medium | |
Cruz et al. | Multi-modal integration of dynamic audiovisual patterns for an interactive reinforcement learning scenario | |
CN106445156A (en) | Method, device and terminal for intelligent home device control based on virtual reality | |
JP2020102194A (en) | System, method and program for context based deep knowledge tracking | |
Koenig et al. | Robot life-long task learning from human demonstrations: a Bayesian approach | |
CN109685037B (en) | Real-time action recognition method and device and electronic equipment | |
CN105042789B (en) | The control method and system of a kind of intelligent air condition | |
CN106022305A (en) | Intelligent robot movement comparing method and robot | |
CN103760970A (en) | Wearable input system and method | |
US10326928B2 (en) | Image processing apparatus for determining whether section of target area matches section of person area and control method thereof | |
CN111027403A (en) | Gesture estimation method, device, equipment and computer readable storage medium | |
CN108985262A (en) | Limb motion guidance method, device, server and storage medium | |
CN112720453A (en) | Method and apparatus for training manipulation skills of a robotic system | |
CN108115678B (en) | Robot and motion control method and device thereof | |
Kim et al. | Using human gaze to improve robustness against irrelevant objects in robot manipulation tasks | |
Meng et al. | Robots learn to dance through interaction with humans | |
CN109471954A (en) | Content recommendation method, device, equipment and storage medium based on mobile unit | |
WO2018150654A1 (en) | Information processing device, information processing method, and program | |
US20220277833A1 (en) | Method and apparatus for determining psychological counseling training scheme | |
CN108491767A (en) | Autonomous roll response method, system and manipulator based on Online Video perception | |
CN112507166B (en) | Intelligent adjustment method and related device for motion course |
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 |