CN105931218B - The intelligent sorting method of modular mechanical arm - Google Patents
The intelligent sorting method of modular mechanical arm Download PDFInfo
- Publication number
- CN105931218B CN105931218B CN201610212575.9A CN201610212575A CN105931218B CN 105931218 B CN105931218 B CN 105931218B CN 201610212575 A CN201610212575 A CN 201610212575A CN 105931218 B CN105931218 B CN 105931218B
- Authority
- CN
- China
- Prior art keywords
- mechanical arm
- solution
- instruction
- sentence
- intelligent sorting
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/04—Inference or reasoning models
- G06N5/046—Forward inferencing; Production systems
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2200/00—Indexing scheme for image data processing or generation, in general
- G06T2200/04—Indexing scheme for image data processing or generation, in general involving 3D image data
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computing Systems (AREA)
- Evolutionary Computation (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Computational Linguistics (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Quality & Reliability (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Manipulator (AREA)
Abstract
The intelligent sorting method of modular mechanical arm provided by the invention, it has the feature that, comprising the following steps: step 1 completes object detection and recognition by body-sensing sensor, the exact space position information for obtaining each object in scene, obtains three-dimensional scenic semanteme map and describes file；Step 2 determines in the form of interactive and is intended to that reasoning obtains sorting rule；And step 3, solution is received, solution is programmed for by robot instruction, parsing Complied executing instruction by natural language programming, and control mechanical arm and carry out intelligent sorting.Method proposed by the present invention realizes the control of modular mechanical arm, while by Intellisense and identification, and human-computer dialogue has incorporated control system with interacting, and successfully completes intelligent sorting task, improves the intelligence of modular mechanical arm.
Description
Technical field
The present invention relates to the control field of modular mechanical arm, in particular to a kind of intelligent sorting side of modular mechanical arm
Method.
Background technique
Modular mechanical arm occupies little space relative to traditional mechanical arm, at low cost, vdiverse in function, strong flexibility
Advantage.In recent years, more and more company, robot or robot research have been carried out to modular mechanical arm both at home and abroad
It studies and achieves certain achievement.
Three-dimension object recognition, human-computer interaction and reasoning are incorporated on modular mechanical arm sorting operation platform, are greatly improved
The level of intelligence of sorting system.
Related patents are found in literature search: application No. is the hairs of CN201410723309.3 disclosed on April 22nd, 2015
Bright patent " intelligent sorting system and method for sorting ", discloses a kind of intelligent sorting method, this invention can identify a variety of face
The object of color, and can be carried out guidance inquiry learning, the object of different colours is put into designated position automatically.
But above-mentioned patent relates only to object color identification, does not identify that the three-dimensional of size, shape and object is sat
Mark；Above-mentioned patent is to guide inquiry learning in advance using study module simultaneously, the then movement of repetitive learning, to a certain degree
On reduce the level of intelligence of system, the method for not using reasoning is not used rhetorical question guidance and is analyzed with expectation.
Summary of the invention
The present invention is to carry out to solve the above-mentioned problems, it is therefore intended that Three-dimension object recognition, human-computer interaction and is pushed away
Reason is dissolved on modular mechanical arm sorting operation platform, proposes a kind of intelligent sorting method of modular mechanical arm.
The intelligent sorting method of modular mechanical arm provided by the invention, has the feature that, comprising the following steps:
Step 1 completes object detection and recognition by body-sensing sensor, obtains the exact space position of each object in scene
Confidence breath, obtains three-dimensional scenic semanteme map and describes file；
Step 2 determines in the form of interactive and is intended to that reasoning obtains sorting rule；And
Step 3 receives solution, and solution is programmed for robot language instruction, solution by natural language programming
Complied executing instruction is analysed, and controls mechanical arm and carries out intelligent sorting.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, step 1,
The collected map depth information of body-sensing sensor and colouring information are generated into three dimensional point cloud after fusion treatment, passed through
It after computer obtains data, extracted by pretreatment, key point, calculate Feature Descriptor, by obtained Feature Descriptor and mould
Type library is matched, generates transformation hypothesis and verify, and is obtained three-dimensional scenic semanteme map and is described file.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, step 2,
Human-computer dialogue is by speech recognition part, inference machine part and speech synthesis part composition, first in speech recognition part, microphone
The voice signal that array inputs user carries out noise reduction process, and carries out feature extraction using pre-defined algorithm, then in conjunction with HMM sound
Model and N-gram language model are learned, text is converted for voice signal by tone decoding searching algorithm and is sent to inference machine portion
Point, inference machine part receives text, using predetermined inference mechanism, text is carried out retrieval with the case in case library and finds most phase
As case, in conjunction with three-dimensional scenic semanteme map describe file carry out map match, expectation analysis and guidance, to improve user
Expectation, ultimately produce solution, the guidance information of user is sent to speech synthesis part in a text form, the part will
Obtained text generates corresponding voice signal by three text analyzing, prosody modeling and speech synthesis steps and exports interaction language
Sound.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, step 3,
The solution that module obtains natural language is obtained by natural language first, then by natural language explanation module solution party
Case is construed to robot language instruction, robot language instruction is sent to parsing collector, parsing collector is by predetermined
Sequence is parsed, and parsing has compiled robot language instruction, finally by executor module, is received and is executed executable finger
It enables.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, pre-defined algorithm
For MFCC algorithm.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, by predetermined
Inference mechanism is improved CBR-BDI inference mechanism.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, speech synthesis
Use TTS technology.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, natural language
It obtains module and realizes communication by the transmission of UDP, obtain solution, have natural language instructions, object inside solution
Coordinate and end effector posture.
The intelligent sorting method of modular mechanical arm provided by the invention also has a feature in that wherein, natural language
Explanation module is to segment natural language instructions the specific step that solution is construed to robot language instruction, word
Method analysis, syntactic analysis obtain robot language instruction after semantic analysis.
Invention effect and effect
The intelligent sorting method of related modular mechanical arm according to the present invention is obtained by body-sensing sensor real first
When three-dimensional scenic semanteme map file is described, then in the form of a dialog establish sorting rule, finally will be in image file
Object coordinates are sent respectively to modular mechanical arm, and mechanical arm grabs object after receiving object coordinates, and object is put into correspondence
Basket inside, so that intelligent sorting is realized, with the development of the modular mechanical arm of mechanical arm especially lightweight, module
The control system for changing mechanical arm is more and more important, and as modular mechanical arm sorting system, this system has visual performance and pushes away
Function is managed, possesses human-computer interaction and improves desired function, while this system can be by the solution of natural language come real
The control of existing mechanical arm, to realize automated programming.
Detailed description of the invention
The step of Fig. 1 is the intelligent sorting method of the modular mechanical arm of the present invention in embodiment figure；
Fig. 2 is the overall construction drawing of point cloud acquisition and object identification in embodiment of the invention；
Fig. 3 is the spatial point cloud object identification of the multiple target scene of the present invention in embodiment and the ensemble stream for understanding system
Cheng Tu；
Fig. 4 is the structural schematic diagram of the speech recognition part of the present invention in embodiment；
Fig. 5 is the structural schematic diagram of the improved CBR-BDI inference mechanism of the present invention in embodiment；
Fig. 6 is the structural schematic diagram of the speech synthesis unit of the present invention in embodiment；
Fig. 7 is the structural schematic diagram of the mechanical arm control module of the present invention in embodiment；And
Fig. 8 is the flow chart of the parsing collector of the present invention in embodiment.
Specific embodiment
Referring to the drawings and embodiment makees in detail the intelligent sorting method of modular mechanical arm according to the present invention
Description.
The step of Fig. 1 is the intelligent sorting method of the modular mechanical arm of the present invention in embodiment figure.
As shown in Figure 1, the intelligent sorting method of modular mechanical arm has follow steps:
Step 1: object detection and recognition is completed by body-sensing sensor, obtains the exact space position of each object in scene
Confidence breath, obtains three-dimensional scenic semanteme map and describes file, enter step two.
Fig. 2 is the overall construction drawing of point cloud acquisition and object identification in embodiment of the invention.
As shown in Fig. 2, the collected map depth information of body-sensing sensor and colouring information is raw after fusion treatment
At three dimensional point cloud, after obtaining data by computer, extracted by pretreatment, key point, calculate Feature Descriptor, will
To Feature Descriptor matched with model library, generate transformation assume and verify, obtain three-dimensional scenic semanteme map description text
Part.
Fig. 3 is the spatial point cloud object identification of the multiple target scene of the present invention in embodiment and the ensemble stream for understanding system
Cheng Tu.
Object identification is mainly made of offline and online two parts with understanding system.
Off-line procedure: it is an offline process that model library, which is established, the key technology being related to have pretreatment, image segmentation,
Feature description, the ego-motion estimation of body-sensing sensor, dense three-dimensional point cloud model generate.Data filtering pretreatment is carried out first,
Then object detection is carried out, that is, is partitioned into the single cluster at each visual angle from scene, then extracts characteristic point and feature
Description.By gridiron pattern calibration algorithm, 4 × 4 variation in rigidity squares between each frame data are obtained using the matching of Feature Descriptor
Battle array, the data under different perspectives to surprise and is added up, so that complete cloud of three-dimension object is obtained, for obtaining the geometry of object
Shape, and object model mark is manually set.
In line process: object identification and pose estimation carry out online.The key technology being related to has based on apart from threshold
The characteristic matching of value, transformation are assumed to convert with verifying, position auto―control coordinate, by the frame point for obtaining body-sensing sensor in real time
Cloud data distribute a suitable object model to object each in scene by the 3D recognizer based on local surfaces feature
Class, in addition it can obtain object model to scene corresponding points relative pose transformation matrix.Finally by the three of the object of identification
It ties up geometrical characteristic and xml document is written in image texture information, building three-dimensional scenic semanteme map describes file.
Step 2: being determined in the form of interactive and be intended to, and reasoning obtains sorting rule, enters step three.
Human-computer dialogue is made of speech recognition part, inference machine part and speech synthesis part.
First in speech recognition part, the voice signal that microphone array inputs user carries out noise reduction process, and uses
MFCC algorithm carries out feature extraction and passes through tone decoding searching algorithm then in conjunction with HMM acoustic model and N-gram language model
Text is converted by voice signal and is sent to inference machine part, and inference machine part receives text, using improved CBR-BDI reasoning
Text match with the case in case library and finds most like case, describes in conjunction with three-dimensional scenic semanteme map by mechanism
File carries out map match, expectation analysis and guidance, to improve the expectation of user, ultimately produces solution.Wherein to
The guidance information at family is sent to speech synthesis part in a text form, which passes through text analyzing, rhythm for obtained text
Three steps of rule modeling and speech synthesis generate corresponding voice signal and export interactive voice.
Fig. 4 is the structural schematic diagram of the speech recognition part of the present invention in embodiment.
As shown in figure 4, when user speaks to robot, microphone first receives voice inside speech recognition part
Signal, then the preprocessing part of system carries out noise reduction process to voice signal, and carries out feature extraction using MFCC algorithm, it
System combination acoustic model and language model afterwards convert text sentence for voice signal by tone decoding searching algorithm.
Fig. 5 is the structural schematic diagram of the improved CBR-BDI inference mechanism of the present invention in embodiment.
As shown in figure 5, inference machine part is using CBR-BDI inference mechanism as core, while being added in inference machine part
Map match, expectation analysis and guidance, to realize the improvement of CBR-BDI inference mechanism.
After inference machine unit receives text sentence, by semantic similarity and sentence structure similarity calculation, from case
If obtaining similar cases in example library, system will carry out task attribute calculating to the case binding rule, differentiate task attribute
Whether complete, if task attribute is complete, i.e. m_num > 0 then enters in next step, otherwise proposes rhetorical question by certain rule by system
Sentence is putd question into guidance；
The complete case of task attribute is matched to obtain m_match, decision rule with current real-time map file
It is as follows:
M_match=0: indicate there is no satisfactory object in scene；
0 < m_match < 1: indicate that the physical quantities are less than the desired quantity of user in scene；
M_match=1: indicate that the two quantity is just equal in scene；
M_match > 1: indicate that the physical quantities are more than the desired quantity of user in scene.
Only as m_match >=1, show that case task can be carried out in current environment, into the next step phase
It hopes analysis, otherwise carries out corresponding rhetorical question guidance by rule；
Case task and the current necessary entry of setting operating rule are subjected to the matching analysis (expectation is analyzed), differentiate that user anticipates
Scheme for relatively current operating rule entry, if it is feasible, if feasible, then natural language solution is generated, otherwise, system is led
It is asked in reply to rules guide, it is desirable that user's completion rule.
Fig. 6 is the structural schematic diagram of the speech synthesis unit of the present invention in embodiment.
As shown in fig. 6, the guidance rhetorical question information of inference machine will be transmitted to speech synthesis unit in a text form, finally with language
The form of sound signal exports.
Step 3 receives solution, and solution is programmed for robot language instruction, solution by natural language programming
Complied executing instruction is analysed, and controls mechanical arm and carries out intelligent sorting.
Fig. 7 is the structural schematic diagram of the mechanical arm control module of the present invention in embodiment.
As shown in fig. 7, the nature of modular mechanical arm next will be completed after receiving the solution of step 2
Programming with Pascal Language automatically parses execution and motion control.The automated programming of modular mechanical arm and parsing motion control one are divided into 4
A module, natural language obtain module, natural language explanation module, parsing collector and executor module.
The solution that module obtains natural language is obtained by natural language first, natural language obtains module and passes through
The transmission of UDP communicates to realize, obtains solution, has natural language instructions, the coordinate of object and end inside solution
Actuator posture.
Then solution is construed to robot language instruction by natural language explanation module, robot language is instructed
It is sent to parsing collector, parsing collector is parsed in a predetermined order, instructs parsing to compile the robot language
It translates robot language instruction to instruct at executable CAN, finally by executor module, receives and execute executable after compiling
CAN instruction.
Natural language explanation module is construed to solution the specific step of robot language instruction are as follows: by nature language
Speech order is segmented, morphological analysis, syntactic analysis, obtains robot language instruction after semantic analysis.
Fig. 8 is the flow chart of the parsing collector of the present invention in embodiment.
As shown in figure 8, sending it to parsing collector after obtaining robot language instruction, parsing collector is pressed
Sequence is parsed, and compiling statement text, parsing compiling robot sentence, if it is the language for having move are read and parsed by row
Sentence calls inverse solution function to find out the joint shaft degree of respective coordinates, is judged by movep and movel using curve interpolating function
Or linear interpolation function；It instructs, then the subsequent coordinate of movep is inverted solution if it is movep, progress curve interpolating；If
It is movel instruction, then the subsequent coordinate of movel is inverted solution, progress linear interpolation；Then it is corresponding for compiling each interpolated point
CAN instruction；If it is the sentence for having hand, end effector control function is called, end is judged by hand on and hand off
End actuator opens or closes, and be compiled into end effector opens or closes CAN instruction；If it is the sentence for having round,
The first joint shaft rotation function is called, the subsequent angle of round is rotated, each joint shaft degree after rotation is compiled into accordingly
CAN instruction, calls normal solution function to find out position after rotation；It is then do-nothing operation if it is the sentence for having Nop；If it is having Time's
Sentence just reads the subsequent time, compiles this section of delay time；If it is there is end sentence to indicate that parsing terminates, compiling terminates mark
Know symbol, returns.
Parsing compile robot language and has instructed, and finally by executor module, receives and executes executable CAN and refer to
It enables, if it is the sentence of move, round, then sends corresponding CAN and instruct to mechanical arm, if it is hand sentence, then send phase
The CAN answered is instructed to end effector, is then delayed the corresponding time if it is Time sentence, is then executed if it is end identifier
It finishes, returns.
The action and effect of embodiment
The intelligent sorting method of the modular mechanical arm according to involved by the present embodiment is obtained by body-sensing sensor real first
When three-dimensional scenic semanteme map file is described, then in the form of a dialog establish sorting rule, finally will be in image file
Object coordinates are sent respectively to modular mechanical arm, and mechanical arm grabs object after receiving object coordinates, and object is put into correspondence
Basket inside, so that intelligent sorting is realized, with the development of the modular mechanical arm of mechanical arm especially lightweight, module
The control system for changing mechanical arm is more and more important, and as modular mechanical arm sorting system, this system has visual performance and pushes away
Function is managed, possesses human-computer interaction and improves desired function, while this system can be by the solution of natural language come real
The control of existing mechanical arm, to realize automated programming.
Above embodiment is preferred case of the invention, the protection scope being not intended to limit the invention.
Claims (6)
1. a kind of intelligent sorting method of modular mechanical arm, which comprises the following steps:
Step 1 completes object detection and recognition by body-sensing sensor, obtains the exact space position letter of each object in scene
Breath, obtains three-dimensional scenic semanteme map and describes file；
Step 2 determines in the form of interactive and is intended to that reasoning obtains sorting rule；And
Step 3 receives solution, solution is programmed for robot language instruction by natural language programming, parsing is compiled
It translates and executes instruction, and control mechanical arm and carry out intelligent sorting；
In the step 1, by the collected map depth information of the body-sensing sensor and colouring information after fusion treatment
Generate three dimensional point cloud, by computer obtain data after, by pretreatment, key point extract, calculate Feature Descriptor, will
The obtained Feature Descriptor is matched with model library, generates transformation hypothesis and verify, and it is semantic to obtain the three-dimensional scenic
Map describes file；
In the step 2, the human-computer dialogue is made of speech recognition part, inference machine part and speech synthesis part；
First in the speech recognition part, the voice signal that microphone array inputs user carries out noise reduction process, and uses
Pre-defined algorithm carries out feature extraction and passes through tone decoding searching algorithm then in conjunction with HMM acoustic model and N-gram language model
Text is converted by the voice signal and is sent to the inference machine part, and the inference machine part receives the text, uses
The text match with the case in case library and finds most like case, in conjunction with the three-dimensional by predetermined inference mechanism
Scene Semantics map describes file and carries out map match, expectation analysis and guidance, to improve the expectation of the user, finally gives birth to
At solution；
The speech synthesis part is sent in the form of the text to the guidance information of the user, which will obtain
The text generates the corresponding voice signal output interaction by three text analyzing, prosody modeling and speech synthesis steps
Voice；
In the step 3, the solution that module obtains natural language is obtained by natural language first, then by natural language
The solution is construed to the robot language and instructed by speech explanation module, and robot language instruction is sent to solution
Collector is analysed, the parsing collector is parsed in a predetermined order, and robot language instruction parsing is compiled into
Executable CAN instruction robot language instruction；
Specific parsing Compilation Method is as follows: compiling statement text, parsing compiling robot sentence are read and parsed by row；If
It is the sentence for having move, calls inverse solution function to find out the joint shaft degree of respective coordinates, judged to use by movep and movel
Curve interpolating function or linear interpolation function；It instructs, then the subsequent coordinate of movep is inverted solution if it is movep, march
Line interpolation；It instructs, then the subsequent coordinate of movel is inverted solution if it is movel, progress linear interpolation；Then compiling is each inserts
Complement point is corresponding CAN instruction；If it is the sentence for having hand, end effector control function is called, by hand on and hand
Off judges opening or closing for end effector, and be compiled into end effector opens or closes CAN instruction；If it is having
The sentence of round calls the first joint shaft rotation function, rotates the subsequent angle of round, each joint shaft degree after rotation
It is compiled into corresponding CAN instruction, normal solution function is called to find out position after rotation；It is then do-nothing operation if it is the sentence for having Nop；Such as
Fruit is that have the sentence of Time just to read the subsequent time, compiles this section of delay time；It is tied if it is there is end sentence to indicate to parse
Beam compiles end identifier, returns；
Parsing has compiled robot language instruction, finally by executor module, receives and executes the executable CAN after compiling
Instruction；If it is the sentence of move, round, then sends corresponding CAN and instruct to mechanical arm；If it is hand sentence, then send
Corresponding CAN is instructed to end effector；It is then delayed the corresponding time if it is Time sentence；It is then held if it is end identifier
Row finishes, and returns.
2. the intelligent sorting method of modular mechanical arm according to claim 1, it is characterised in that:
Wherein, the pre-defined algorithm is MFCC algorithm.
3. the intelligent sorting method of modular mechanical arm according to claim 1, it is characterised in that:
It wherein, is improved CBR-BDI inference mechanism by the predetermined inference mechanism.
4. the intelligent sorting method of modular mechanical arm according to claim 1, it is characterised in that:
Wherein, speech synthesis uses TTS technology.
5. the intelligent sorting method of modular mechanical arm according to claim 1, it is characterised in that:
Wherein, the natural language obtains module and realizes communication by the transmission of UDP, obtains the solution,
There are natural language instructions, the coordinate and end effector posture of object inside the solution.
6. the intelligent sorting method of modular mechanical arm according to claim 1, it is characterised in that:
Wherein, natural language explanation module is construed to the solution the specific step of the robot language instruction
Are as follows: natural language instructions are segmented, morphological analysis, syntactic analysis, obtain robot language instruction after semantic analysis.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610212575.9A CN105931218B (en) | 2016-04-07 | 2016-04-07 | The intelligent sorting method of modular mechanical arm |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201610212575.9A CN105931218B (en) | 2016-04-07 | 2016-04-07 | The intelligent sorting method of modular mechanical arm |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN105931218A CN105931218A (en) | 2016-09-07 |
| CN105931218B true CN105931218B (en) | 2019-05-17 |
Family
ID=56840150
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201610212575.9A Active CN105931218B (en) | 2016-04-07 | 2016-04-07 | The intelligent sorting method of modular mechanical arm |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN105931218B (en) |
Families Citing this family (16)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN106681323B (en) * | 2016-12-22 | 2020-05-19 | 北京光年无限科技有限公司 | Interactive output method for robot and robot |
| CN107127757B (en) * | 2017-05-24 | 2023-03-31 | 西安科技大学 | Dynamic task allocation method for multi-robot cooperation flexible cable driven gangue picking equipment |
| CN108044621B (en) * | 2017-09-20 | 2020-01-10 | 广东拓斯达科技股份有限公司 | Computer readable storage medium and robot using the same |
| CN107622523B (en) * | 2017-09-21 | 2018-08-21 | 石器时代（内蒙古）智能机器人科技有限公司 | A kind of intelligent robot |
| CN107742311B (en) * | 2017-09-29 | 2020-02-18 | 北京易达图灵科技有限公司 | Visual positioning method and device |
| CN108247601A (en) * | 2018-02-09 | 2018-07-06 | 中国科学院电子学研究所 | Semantic crawl robot based on deep learning |
| JP2021530794A (en) * | 2018-07-17 | 2021-11-11 | アイ・ティー スピークス エル・エル・シーiT SpeeX LLC | Methods, systems, and computer program products for interacting with intelligent assistants and industrial machinery |
| CN109146163B (en) * | 2018-08-07 | 2021-12-07 | 上海大学 | Method and equipment for optimizing sorting distance of automatic sorting system and storage medium |
| CN110253588A (en) * | 2019-08-05 | 2019-09-20 | 江苏科技大学 | A kind of New Type of Robot Arm dynamic grasping system |
| CN110666806B (en) * | 2019-10-31 | 2021-05-14 | 湖北文理学院 | Article sorting method, article sorting device, robot and storage medium |
| CN113021333A (en) * | 2019-12-25 | 2021-06-25 | 沈阳新松机器人自动化股份有限公司 | Object grabbing method and system and terminal equipment |
| CN111260761B (en) * | 2020-01-15 | 2023-05-09 | 北京猿力未来科技有限公司 | Method and device for generating mouth shape of animation character |
| CN112232141B (en) * | 2020-09-25 | 2023-06-20 | 武汉云极智能科技有限公司 | Mechanical arm interaction method and equipment capable of identifying object space position |
| CN112667823B (en) * | 2020-12-24 | 2022-11-01 | 西安电子科技大学 | Semantic analysis method and system for task execution sequence of mechanical arm and computer readable medium |
| CN112809689B (en) * | 2021-02-26 | 2022-06-14 | 同济大学 | Language-guidance-based mechanical arm action element simulation learning method and storage medium |
| CN113742458B (en) * | 2021-09-18 | 2023-04-25 | 苏州大学 | Natural language instruction disambiguation method and system oriented to mechanical arm grabbing |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1293752A (en) * | 1999-03-19 | 2001-05-02 | 松下电工株式会社 | Three-D object recognition method and pin picking system using the method |
| WO2014140129A1 (en) * | 2013-03-12 | 2014-09-18 | Centre National D'etudes Spatiales | Method of measuring the direction of a line of sight of an imaging device |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102615052B (en) * | 2012-02-21 | 2013-12-25 | 上海大学 | Machine visual identification method for sorting products with corner point characteristics |
| CN103324938A (en) * | 2012-03-21 | 2013-09-25 | 日电（中国）有限公司 | Method for training attitude classifier and object classifier and method and device for detecting objects |
-
2016
- 2016-04-07 CN CN201610212575.9A patent/CN105931218B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN1293752A (en) * | 1999-03-19 | 2001-05-02 | 松下电工株式会社 | Three-D object recognition method and pin picking system using the method |
| WO2014140129A1 (en) * | 2013-03-12 | 2014-09-18 | Centre National D'etudes Spatiales | Method of measuring the direction of a line of sight of an imaging device |
Non-Patent Citations (4)
| Title |
|---|
| 一种实时的三维语义地图生成方法;吴凡 等;《计算机工程与应用》;20151109;第53卷(第6期);68页，图1，图2 |
| 基于Julius的机器人语音识别系统构建;付维 等;《单片机与嵌入式系统应用》;20110801(第8期);42-43页，图1 |
| 基于自然语言的分拣机器人解析器技术研究;熊志恒 等;《计算机工程与应用》;20151221;第58卷(第3期);115-119页，图4 |
| 改进的分拣作业机械臂基于范例推理 - 信念期望意图推理机制;周昊天 等;《计算机应用》;20160310;第36卷(第3期);748-749页，图3，图4，图6 |
Also Published As
| Publication number | Publication date |
|---|---|
| CN105931218A (en) | 2016-09-07 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105931218B (en) | The intelligent sorting method of modular mechanical arm | |
| US11056096B2 (en) | Artificial intelligence (AI)-based voice sampling apparatus and method for providing speech style in heterogeneous label | |
| CN101101752B (en) | Monosyllabic language lip-reading recognition system based on vision character | |
| CN111325817A (en) | Virtual character scene video generation method, terminal device and medium | |
| Bastianelli et al. | A discriminative approach to grounded spoken language understanding in interactive robotics | |
| US20100082345A1 (en) | Speech and text driven hmm-based body animation synthesis | |
| Sargin et al. | Analysis of head gesture and prosody patterns for prosody-driven head-gesture animation | |
| CN105787471A (en) | Gesture identification method applied to control of mobile service robot for elder and disabled | |
| Françoise et al. | A hierarchical approach for the design of gesture-to-sound mappings | |
| CN106985137A (en) | Multi-modal exchange method and system for intelligent robot | |
| CN106457563A (en) | Method of performing multi-modal dialogue between a humanoid robot and user, computer program product and humanoid robot for implementing said method | |
| CN104123939A (en) | Substation inspection robot based voice interaction control method | |
| CN112101045B (en) | Multi-mode semantic integrity recognition method and device and electronic equipment | |
| JP5141876B2 (en) | Orbit search device | |
| CN106557165B (en) | The action simulation exchange method and device and smart machine of smart machine | |
| CN113936637A (en) | Voice self-adaptive completion system based on multi-mode knowledge graph | |
| Kruijff-Korbayová et al. | An event-based conversational system for the nao robot | |
| CN109741751A (en) | Intension recognizing method and device towards intelligent sound control | |
| Kibria | Speech Recognition for Robotic Control | |
| CN116934926B (en) | Recognition method and system based on multi-mode data fusion | |
| CN114360485A (en) | Voice processing method, system, device and medium | |
| CN107123420A (en) | Voice recognition system and interaction method thereof | |
| KR101727306B1 (en) | Languange model clustering based speech recognition apparatus and method | |
| Fardana et al. | Controlling a Mobile Robot with Natural Commands based on Voice and Gesture | |
| Chen et al. | A novel approach of system design for dialect speech interaction with NAO robot |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |