CN108108759B - Dynamic grouping method for multiple intelligent agents - Google Patents

Dynamic grouping method for multiple intelligent agents Download PDF

Info

Publication number
CN108108759B
CN108108759B CN201711376482.0A CN201711376482A CN108108759B CN 108108759 B CN108108759 B CN 108108759B CN 201711376482 A CN201711376482 A CN 201711376482A CN 108108759 B CN108108759 B CN 108108759B
Authority
CN
China
Prior art keywords
agent
information
similarity function
similarity
agents
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
Application number
CN201711376482.0A
Other languages
Chinese (zh)
Other versions
CN108108759A (en
Inventor
高晓利
王维
李捷
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Sichuan Jiuzhou Electric Group Co Ltd
Original Assignee
Sichuan Jiuzhou Electric Group Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Sichuan Jiuzhou Electric Group Co Ltd filed Critical Sichuan Jiuzhou Electric Group Co Ltd
Priority to CN201711376482.0A priority Critical patent/CN108108759B/en
Publication of CN108108759A publication Critical patent/CN108108759A/en
Application granted granted Critical
Publication of CN108108759B publication Critical patent/CN108108759B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/23Clustering techniques
    • G06F18/232Non-hierarchical techniques
    • G06F18/2321Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions
    • G06F18/23211Non-hierarchical techniques using statistics or function optimisation, e.g. modelling of probability density functions with adaptive number of clusters
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/25Fusion techniques
    • G06F18/251Fusion techniques of input or preprocessed data
    • GPHYSICS
    • G06COMPUTING; CALCULATING OR COUNTING
    • G06FELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00Pattern recognition
    • G06F18/20Analysing
    • G06F18/25Fusion techniques
    • G06F18/254Fusion techniques of classification results, e.g. of results related to same input data

Landscapes

  • Engineering & Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Probability & Statistics with Applications (AREA)
  • Stored Programmes (AREA)
  • Information Retrieval, Db Structures And Fs Structures Therefor (AREA)

Abstract

The invention discloses a dynamic grouping method of multiple intelligent agents, which comprises the following steps: step one, receiving multi-source information; step two, data preprocessing; step three, selecting a clustering center; step four, constructing a similarity function; and step five, autonomous dynamic grouping. According to the invention, through data preprocessing, the unified quantification of numerical information and non-numerical information and the processing problem of characteristic data are realized, and the problem that a good division result is difficult to obtain when a fuzzy C mean value meets an isolated point and a noise point is solved; the design of a non-fixed center selection strategy solves the problem that the requirement of the unmanned intelligent agent for autonomy cannot be met by manually setting an initialization clustering center; and finally, the constraint of membership is weakened through the similarity function construction of a high-dimensional space and the autonomous dynamic grouping based on the internal acting force, and the dynamic grouping of the intelligent agent is realized substantially.

Description

Dynamic grouping method for multiple intelligent agents
Technical Field
The invention relates to the field of multi-sensor information fusion, in particular to a dynamic grouping method of multiple intelligent agents.
Background
The essence of target clustering is clustering, while the fuzzy C-means clustering method is a typical clustering analysis method, which is to divide target objects into a plurality of corresponding categories according to a certain measurement standard, and ensure that better similarity exists in the same category and obvious difference exists among the categories.
However, the conventional fuzzy C-means based clustering method has the following disadvantages: firstly, only influence factors such as position and pose quantifiable information are adapted, non-quantifiable information is not adapted, and the multi-source information utilization rate of an intelligent agent is not high; secondly, when isolated points and noise points are encountered, good division results are difficult to obtain; thirdly, an initialization clustering center needs to be set manually, and the requirement of autonomy of the unmanned intelligent body is not met; in addition, due to the limitation of membership degree constraint, other sample points are easily interfered by other clustering centers except the type, and the actual problem of intelligent agent dynamic grouping cannot be reflected.
Disclosure of Invention
The technical problem to be solved by the invention is as follows: aiming at the existing problems, an autonomous multi-agent dynamic grouping method is provided, a non-fixed center is dynamically selected based on different motion states of agents, different similarity measures are designed according to different information and are used as internal acting force between the agents to realize dynamic grouping of the multi-agents, further, the polysemy processing of grouping is realized according to a voting mechanism and attribute uniqueness, and the multi-agent grouping efficiency is improved.
The invention provides a dynamic grouping method of multi-agent, which is characterized by comprising the following steps:
step one, receiving multi-source information;
step two, data preprocessing;
step three, selecting a clustering center;
step four, constructing a similarity function;
and step five, autonomous dynamic grouping.
Further, the multi-source information comprises position information PI, attitude information SI, attribute information AI and tactical motion characteristics TPI.
Further, the data preprocessing comprises non-quantization information processing or special data processing, wherein the non-quantization information comprises attribute information and tactical action characteristics, and the special data comprises missing values, isolated points and noise point data.
Further, the non-quantization information processing means digitizing non-quantization information according to a category of the non-quantization information.
Further, the special data processing includes: processing missing values based on probability theory, processing noise points based on a box separation method or processing isolated points based on a filtering method.
Further, the processing of the isolated points based on the filtering method includes: and judging whether the current trace point is an isolated point or not through a Kalman filtering algorithm, if so, directly compiling the current trace point into a unit group, and caching the unit group.
Further, the selecting of the clustering center as the non-fixed center is to select the clustering center from the micro unmanned intelligent agent clustering without organization and fixed center, and the selecting of the non-fixed center includes selecting the non-fixed center in a random motion state and a class rule motion state.
Further, the non-fixed center selection under the random motion state is realized based on the regional fuzzy C-means method, or the non-fixed center selection under the class regular motion state is realized based on the uniform distribution and the bionic idea.
Further, the similarity function includes a similarity function of the position information, a similarity function of the attitude information, a similarity function of the attribute information, a similarity function of the tactical action characteristics, and a comprehensive similarity function.
Furthermore, dynamic grouping of the multiple intelligent agents is realized based on the internal acting force among the multiple intelligent agents and by combining a bionic theory.
Further, when agent j belongs to a plurality of different groups at the same time, processing is performed according to the following steps:
calculating included angles between the agent j and the non-fixed centers of the different groups;
comparing the sizes of the included angles;
checking according to a voting mechanism;
validation is based on the uniqueness of the group attributes.
In summary, due to the adoption of the technical scheme, the invention has the beneficial effects that: by data preprocessing, the problems of unified numeralization of quantized information and unquantized information and processing of characteristic data are realized, and the problem that a good division result is difficult to obtain when a fuzzy C mean value encounters an isolated point and a noise point is solved; the design of a non-fixed center selection strategy solves the problem that the requirement of the unmanned intelligent agent for autonomy cannot be met by manually setting an initialization clustering center; and finally, the constraint of membership is weakened through the similarity function construction of a high-dimensional space and the autonomous dynamic grouping based on the internal acting force, and the dynamic grouping of the intelligent agent is realized substantially.
Drawings
The invention will now be described, by way of example, with reference to the accompanying drawings, in which:
FIG. 1 is a flow chart of an autonomic multi-agent dynamic clustering method.
Fig. 2 is a non-fixed center selection flow diagram.
Fig. 3 is a flowchart of the integrated similarity matrix calculation.
FIG. 4 is a flow chart of a location similarity function construction.
FIG. 5 is a flow diagram of a grouping ambiguity process.
Detailed Description
All of the features disclosed in this specification, or all of the steps in any method or process so disclosed, may be combined in any combination, except combinations of features and/or steps that are mutually exclusive.
Any feature disclosed in this specification may be replaced by alternative features serving equivalent or similar purposes, unless expressly stated otherwise. That is, unless expressly stated otherwise, each feature is only an example of a generic series of equivalent or similar features.
Fig. 1 is a flow chart of an autonomous multi-agent dynamic grouping method implemented by the present invention. The method comprises five steps of multi-source information receiving, data preprocessing, non-fixed center selection, similarity function construction of a high-dimensional space and autonomous dynamic grouping.
And receiving multi-source information. The invention can receive multi-source information observed, detected and received by multi-sensors with different types, different working systems and different data rates, mainly comprises four types of multi-source information for multi-agent dynamic grouping, namely position information, attitude information, attribute information and tactical action characteristics, wherein the position information comprises positions in three coordinate axis directions of an x axis, a y axis and a z axis, and is represented by PI (x, y, z); the attitude information includes the speed and direction in the three-coordinate direction, expressed as SI ═ Vxy, Vyz, Vxz, Fxy, Fyz, Fxz); the attribute information comprises friend or foe attribute, type and model, and the attribute information is AI (AI)attr,AIclass,AItype) Represents; tactical movement characteristics are represented by TPI, which may take the values of surveillance, reconnaissance, countermeasure, supportInterference, spoofing, attacks.
And (4) preprocessing data. The data preprocessing comprises two aspects of non-quantization information digitization and special data processing, wherein the non-quantization information comprises attribute information and tactical action characteristics, and the special data comprises missing values, isolated points and noise point data. Through data preprocessing, the invention realizes the unified numeralization of quantized information and unquantized information and the processing of characteristic data, and solves the problem that a good division result is difficult to obtain when a fuzzy C mean value meets isolated points and noise points.
Based on classification and segmentation ideas, non-quantized information is defined numerically as follows:
friend or foe attribute AIattr: 1-enemy, 2-my, 3-friend, 4-neutral, 5-civil, 0-unknown;
type AIclass: 1-land, 2-sea, 3-empty, 0-unknown;
model AItype: 1-199-land platform model, 200-;
TPI: 1-surveillance, 2-reconnaissance, 3-countermeasure, 4-support, 5-interference, 6-spoofing, 7-attack.
And (4) special data processing. The missing values are filled by adopting conditional probability distribution, the noise points are used for smoothing data by adopting a box separation method, and the isolated points are processed by adopting a filtering method.
Missing value processing based on conditional probability distribution. The filling is based on the statistical distribution characteristic of the missing values, for which the missing values are not replaced by 0 or statistical mean values, but by randomly drawn values of their distribution. Let xcomRepresenting the complete feature vector and assuming that some of the elements are missing xmisThe rest to-be-observed value xobsThen the complete feature vector can be expressed as:
Figure GDA0003269126030000051
the probability of the missing value is assumed to be irrelevant to the value, and the missing value is filled by the distribution mean of the conditional probability density function, which is specifically expressed as follows:
Figure GDA0003269126030000052
wherein p (x)obs;θ)=∫p(xcom;θ)dxmisAnd theta denotes the unknown set of parameters whose estimated values are first derived from the remaining observations xobsAnd obtaining the parameter by adopting an EM expectation maximization method.
And (4) noise point processing based on a box separation method. Firstly, mapping symbolic data of target mixed data into numerical data, and carrying out unified data transformation; and then sorting the data in an ascending order, dividing the sorted data, and setting the number of the data in the box. The data is smoothed using a constant frequency or median, i.e., the ordered data values are smoothed by looking at the close-neighbor characteristics of the data.
And processing isolated points based on a filtering method. Judging whether the current trace point is an isolated point by using a Kalman filtering algorithm, if so, directly compiling the current trace point into a unit group, caching the unit group, and performing proofreading through multiple cycles.
Non-fixed center selection strategy, see fig. 2. It is assumed that n existing agents perform non-fixed center selection according to their motion states, including non-fixed center selection in random motion states and quasi-regular motion states. Through the design of a non-fixed center selection strategy, the method solves the problem that the requirement of the unmanned intelligent agent for autonomy cannot be met by manually setting the initialization clustering center. There are no assumptions about m fusion centers, C each1,C2,…,Cm
Non-stationary center selection in random motion state. When the unmanned multi-agent is thrown out to execute an initial task, the motion state of the unmanned multi-agent can be assumed to be random motion, at the moment, if the action range of the agent is MaxAffRan and the battle range AreaCov, m areas are determined based on floor (AreaCov/MaxAffRan), and then a clustering center is selected in each area based on a fuzzy C mean value method.
And selecting a non-fixed center in a regular motion-like state. The unmanned multi-Agent can also move in a regular manner like a straight line, and at the moment, starting from the first Agent1, based on the idea of uniform distribution and bionics, the fusion center is sequentially set as an unfixed center for the agents which are multiple of MaxAffRan and are away from the Agent 1.
The similarity function in the high dimensional space is constructed, see fig. 3. Assume that agents i and j have information at time k of
Figure GDA0003269126030000061
According to the characteristics of the multi-source information types, different similarity functions are designed, including a similarity function of position information, a similarity function of attitude information, a similarity function of attribute information, a similarity function of tactical action characteristics and a comprehensive similarity function. According to the invention, different similarity functions are constructed according to the characteristics of the multi-source information types, and the problem of quantification of internal acting force among the multi-agents is solved.
The similarity function of location information between agent i and agent j is defined as
Figure GDA0003269126030000062
The calculation steps are shown in fig. 4, and specifically are as follows:
Figure GDA0003269126030000063
wherein
Figure GDA0003269126030000064
Represents the distance between agent i and agent j; [ MinDis, Maxis]Indicating the possible existence of the agent, the value can be set according to the performance and the combat range of the unmanned agent, for example, if the agent is a micro unmanned plane with a wing span less than 15 cm, then MaxDis can be set to 10 km, and minidis can be set to 0.5 km. The similarity function indirectly and truly reflects the distribution state between the intelligent agents from the distance dimension, namely the farther the distance between the two intelligent agents is, the smaller the similarity is, and otherwise, the greater the similarity is; and the value of which does not appear infinitesimal or infiniteAnd the condition is convenient for calculating the similarity between every two intelligent agents subsequently under the same frame.
Gesture information similarity function between agent i and agent j
Figure GDA0003269126030000071
Is calculated by the formula
Figure GDA0003269126030000072
Jxy、JyzAnd JxzRespectively representing the attitude similarity of the agent i and the agent J in an XOY plane, a YOZ plane and an XOZ plane, Jxy、JyzAnd JxzThe calculation formulas of (A) and (B) are respectively as follows:
Figure GDA0003269126030000073
Figure GDA0003269126030000074
Figure GDA0003269126030000075
wherein, max VxyAnd min VxyRepresenting a maximum velocity and a minimum velocity of the agent in the XOY plane; max VyzAnd min VyzRepresenting the maximum and minimum velocities of the agent in the YOZ plane; max VxzAnd min VxzRepresenting the maximum and minimum velocities of the agent in the XOZ plane.
Attribute information similarity function between agent i and agent j
Figure GDA0003269126030000076
Is calculated by the formula
Figure GDA0003269126030000077
Wherein
Figure GDA0003269126030000078
And
Figure GDA0003269126030000079
respectively representing the similarity of the attributes of the enemy and the my, the similarity of the types and the similarity of the models, which are defined as follows:
Figure GDA00032691260300000710
Figure GDA00032691260300000711
Figure GDA0003269126030000081
similarity function of tactical motion characteristics
Figure GDA0003269126030000082
The definition is defined based on a binary method if
Figure GDA0003269126030000083
And
Figure GDA0003269126030000084
same, then
Figure GDA0003269126030000085
Equal to 0, otherwise equal to 1.
Constructing a comprehensive similarity matrix P ═ (P)ij)i,,j=1,2,…,nThe calculation formula is as follows:
Figure GDA0003269126030000086
wherein wPI、wSI、wAIAnd wTPIRespectively representing position information, attitude information, attribute information and tactical action characteristic weight.
And (4) autonomous dynamic grouping. Assume similarity threshold is SimiDegree and fusion center is CiIf p isijIf the content is less than or equal to SimleDegreee, the fact that the agent j belongs to C is indicatediOtherwise, it indicates
Figure GDA0003269126030000087
Suppose in CiHaving NumNoC in the regioniIndividual agent not belonging to group CiIn this case, it can be regarded as NumNoCiAnd (4) entering a next round of grouping circulation for discrete subgroups. However, during the grouping process, j ∈ C may occuriAnd ClI ≠ l, which requires clustering ambiguity processing.
Grouping ambiguity handling, see fig. 5. Combining agent j with non-fixed center CiAnd ClAngle theta therebetweenj,iAnd thetaj,lThe attribute and history clustering results are performed, if θ isj,iLess than threshold and thetaj,ij,lThen it indicates that the agent is not fixed center C from both distance and directioniThe positions are relatively close; and then, the grouping result is corrected according to a voting mechanism and combined with the historical grouping result, and finally, the grouping result is confirmed based on the uniqueness of the group attribute, and the grouping result is output.
The invention is not limited to the foregoing embodiments. The invention extends to any novel feature or any novel combination of features disclosed in this specification and any novel method or process steps or any novel combination of features disclosed.

Claims (12)

1. A multi-agent dynamic grouping method is characterized by comprising the following steps:
step one, receiving multi-source information; the multi-source information comprises position information, posture information, attribute information and tactical action characteristics, and is used for multi-agent dynamic grouping; it is composed ofThe position information includes positions in three coordinate axis directions of an x axis, a y axis and a z axis, and is represented by PI ═ x, y, z; the attitude information includes the speed and direction in the three-coordinate direction, expressed as SI ═ Vxy, Vyz, Vxz, Fxy, Fyz, Fxz); the attribute information comprises friend or foe attribute, type and model, and the attribute information is AI (AI)attr,AIclass,AItype) Represents; tactical action characteristics are represented by TPI, which may take on values of surveillance, reconnaissance, countermeasure, support, interference, fraud, attack;
step two, data preprocessing;
step three, selecting a clustering center;
step four, constructing a similarity function; the similarity function comprises a similarity function of position information, a similarity function of posture information, a similarity function of attribute information, a similarity function of tactical action characteristics and a comprehensive similarity function;
similarity function of the location information
Figure FDA0003269126020000011
The calculation formula of (2) is as follows:
Figure FDA0003269126020000012
wherein i and j represent agents i and j; k represents the k time;
Figure FDA0003269126020000013
represents the distance between agent i and agent j; [ MinDis, Maxis]Representing areas where agents may be present;
similarity function of the pose information
Figure FDA0003269126020000014
The calculation formula of (2) is as follows:
Figure FDA0003269126020000015
where i and j represent agents i and j; k represents the k time; attitude information SI ═ (Vxy, Vyz, Vxz, Fxy, Fyz, Fxz), including velocity and direction in the three-coordinate direction; j. the design is a squarexy、JyzAnd JxzRespectively representing the attitude similarity of the agent i and the agent J in an XOY plane, a YOZ plane and an XOZ plane, Jxy、JyzAnd JxzThe calculation formulas of (A) and (B) are respectively as follows:
Figure FDA0003269126020000021
Figure FDA0003269126020000022
Figure FDA0003269126020000023
wherein, max VxyAnd min VxyRepresenting a maximum velocity and a minimum velocity of the agent in the XOY plane; max VyzAnd min VyzRepresenting the maximum and minimum velocities of the agent in the YOZ plane; max VxzAnd min VxzRepresenting the maximum speed and the minimum speed of the agent on the XOZ plane;
similarity function of the attribute information
Figure FDA0003269126020000024
The calculation formula of (2) is as follows:
Figure FDA0003269126020000025
where i and j represent agents i and j; k represents the k time;
Figure FDA0003269126020000026
and
Figure FDA0003269126020000027
respectively representing the similarity of the attributes of the enemy and the my, the similarity of the types and the similarity of the models, which are defined as follows:
Figure FDA0003269126020000028
Figure FDA0003269126020000029
Figure FDA00032691260200000210
similarity function of the tactical motion characteristics
Figure FDA00032691260200000211
Based on a binary method, if
Figure FDA00032691260200000212
And
Figure FDA00032691260200000213
same, then
Figure FDA00032691260200000214
Equal to 0, otherwise,
Figure FDA00032691260200000215
equal to 1, where i and j denote agents i and j; k represents the k time;
the integrated similarity function pijThe calculation formula of (a) is as follows:
Figure FDA0003269126020000031
where i and j represent agents i and j; k represents the k time; w is aPI、wSI、wAIAnd wTPIRespectively representing position information, attitude information, attribute information and tactical action characteristic weight;
and step five, autonomous dynamic grouping.
2. The method of claim 1, wherein said multi-source information includes position information PI, attitude information SI, attribute information AI, and tactical motion characteristics TPI.
3. The method of claim 1, wherein said pre-processing of data comprises non-quantitative information processing or special data processing, wherein non-quantitative information comprises attribute information and tactical motion characteristics, and special data comprises missing values, outliers and noise point data.
4. The method as claimed in claim 3, wherein said non-quantized information processing means to quantize non-quantized information according to its category.
5. A method for dynamic grouping of multi-agents as claimed in claim 3, wherein said special data handling comprises: processing missing values based on probability theory, processing noise points based on a box separation method or processing isolated points based on a filtering method.
6. The method of claim 5, wherein said filtering-based approach to outlier handling comprises: and judging whether the current trace point is an isolated point or not through a Kalman filtering algorithm, if so, directly compiling the current trace point into a unit group, and caching the unit group.
7. The method as claimed in claim 1, wherein said selecting clustering centers as non-fixed center selection means selecting clustering centers from unorganized, non-fixed center micro-unmanned agent clusters, including non-fixed center selection in random motion state and quasi-regular motion state.
8. The method as claimed in claim 7, wherein the non-stationary center selection in random motion state is implemented based on a partitioned fuzzy C-means method, or the non-stationary center selection in regular motion state is implemented based on uniform distribution and bionic idea.
9. The method of claim 8, wherein said partitioned fuzzy C-means based approach to non-stationary center selection in random motion comprises:
determining m regions based on floor (AreaCov/MaxAffRan);
selecting a cluster center in each region based on a fuzzy C-means method,
wherein MaxAffRan is the action range of the intelligent agent, and AreaCov is the combat range.
10. The method as claimed in claim 8, wherein said achieving non-fixed center selection in a regular motion-like state based on uniform distribution and bionic idea comprises:
determining a first Agent 1;
and sequentially setting an Agent with a distance of multiple MaxAffRan from the Agent1 as a non-fixed center, wherein MaxAffRan is the action range of the Agent.
11. The method as claimed in claim 1, wherein the dynamic grouping of the multi-agents is achieved based on the intrinsic forces between the multi-agents in combination with the bionic theory.
12. The method of claim 11, wherein when agent j belongs to a plurality of different groups simultaneously, the process is performed according to the following steps:
calculating included angles between the agent j and the non-fixed centers of the different groups;
comparing the sizes of the included angles;
checking according to a voting mechanism;
validation is based on the uniqueness of the group attributes.
CN201711376482.0A 2017-12-19 2017-12-19 Dynamic grouping method for multiple intelligent agents Active CN108108759B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711376482.0A CN108108759B (en) 2017-12-19 2017-12-19 Dynamic grouping method for multiple intelligent agents

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711376482.0A CN108108759B (en) 2017-12-19 2017-12-19 Dynamic grouping method for multiple intelligent agents

Publications (2)

Publication Number Publication Date
CN108108759A CN108108759A (en) 2018-06-01
CN108108759B true CN108108759B (en) 2021-11-02

Family

ID=62210144

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711376482.0A Active CN108108759B (en) 2017-12-19 2017-12-19 Dynamic grouping method for multiple intelligent agents

Country Status (1)

Country Link
CN (1) CN108108759B (en)

Families Citing this family (1)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN108921298B (en) * 2018-06-12 2022-04-19 中国科学技术大学 Multi-agent communication and decision-making method for reinforcement learning

Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6865408B1 (en) * 2001-04-11 2005-03-08 Inlight Solutions, Inc. System for non-invasive measurement of glucose in humans
CN101697167A (en) * 2009-10-30 2010-04-21 邱建林 Clustering-decision tree based selection method of fine corn seeds
CN102215260A (en) * 2011-06-02 2011-10-12 北京理工大学 Multi-agent system cooperative control method capable of maintaining connectivity
CN102831474A (en) * 2012-08-06 2012-12-19 江南大学 Improved fuzzy C-mean clustering method based on quantum particle swarm optimization
CN104200295A (en) * 2014-05-29 2014-12-10 南京邮电大学 Partition based multi-police-intelligent-agent task allocation method in RCRSS (Robo Cup Rescue Simulation System)
CN104573038A (en) * 2015-01-19 2015-04-29 合肥工业大学 Guide method for group behavior in social network
CN104899229A (en) * 2014-03-07 2015-09-09 上海市玻森数据科技有限公司 Swarm intelligence based behavior clustering system
CN106447542A (en) * 2016-08-29 2017-02-22 江苏大学 Active traveling service system for Internet of Vehicles and service need dynamic acquisition and construction method

Patent Citations (8)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US6865408B1 (en) * 2001-04-11 2005-03-08 Inlight Solutions, Inc. System for non-invasive measurement of glucose in humans
CN101697167A (en) * 2009-10-30 2010-04-21 邱建林 Clustering-decision tree based selection method of fine corn seeds
CN102215260A (en) * 2011-06-02 2011-10-12 北京理工大学 Multi-agent system cooperative control method capable of maintaining connectivity
CN102831474A (en) * 2012-08-06 2012-12-19 江南大学 Improved fuzzy C-mean clustering method based on quantum particle swarm optimization
CN104899229A (en) * 2014-03-07 2015-09-09 上海市玻森数据科技有限公司 Swarm intelligence based behavior clustering system
CN104200295A (en) * 2014-05-29 2014-12-10 南京邮电大学 Partition based multi-police-intelligent-agent task allocation method in RCRSS (Robo Cup Rescue Simulation System)
CN104573038A (en) * 2015-01-19 2015-04-29 合肥工业大学 Guide method for group behavior in social network
CN106447542A (en) * 2016-08-29 2017-02-22 江苏大学 Active traveling service system for Internet of Vehicles and service need dynamic acquisition and construction method

Non-Patent Citations (2)

* Cited by examiner, † Cited by third party
Title
"Clustering by group consensus of unstable dynamic linear high-order multi-agent systems";Cai Ning等;《2015 34th Chinese Control Conference (CCC)》;20150914;第7212-7216页 *
"基于智能体的组合服务动态编排";欧毓毅等;《计算机应用研究》;20080531;第25卷(第5期);第1425-1428页 *

Also Published As

Publication number Publication date
CN108108759A (en) 2018-06-01

Similar Documents

Publication Publication Date Title
Borrmann et al. The 3d hough transform for plane detection in point clouds: A review and a new accumulator design
JP5759161B2 (en) Object recognition device, object recognition method, learning device, learning method, program, and information processing system
JP5800494B2 (en) Specific area selection device, specific area selection method, and program
CN101986328A (en) Local descriptor-based three-dimensional face recognition method
Carmichael et al. Large data sets and confusing scenes in 3-d surface matching and recognition
CN113484875B (en) Laser radar point cloud target hierarchical identification method based on mixed Gaussian ordering
Hermann et al. Anticipate your surroundings: Predictive collision detection between dynamic obstacles and planned robot trajectories on the GPU
CN111024085A (en) Unmanned aerial vehicle track planning method with end point direction and time constraints
CN114387462A (en) Dynamic environment sensing method based on binocular camera
Kim et al. Planning for grasp selection of partially occluded objects
CN108108759B (en) Dynamic grouping method for multiple intelligent agents
CN109840558B (en) Self-adaptive clustering method based on density peak value-core fusion
Patil et al. A survey on joint object detection and pose estimation using monocular vision
Cao et al. Density-based fuzzy C-means multi-center re-clustering radar signal sorting algorithm
Zeng et al. Parametricnet: 6dof pose estimation network for parametric shapes in stacked scenarios
CN114565861A (en) Airborne downward-looking target image positioning method based on probability statistic differential homoembryo set matching
CN110119751A (en) Laser radar point cloud Target Segmentation method, target matching method, device and vehicle
CN111986299B (en) Point cloud data processing method, device, equipment and storage medium
Wang et al. An efficient scene semantic labeling approach for 3D point cloud
Javakhishvili et al. A method of open cluster membership determination
CN111476321B (en) Air flyer identification method based on feature weighting Bayes optimization algorithm
Jie A Chaos Particle Swarm Optimization Based on Adaptive Inertia Weight
Srivastava et al. Large scale novel object discovery in 3d
Wu et al. An optimized K-means clustering algorithm based on BC-QPSO for remote sensing image
CN111765882B (en) Laser radar positioning method and related device thereof

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