CN101303589A - Multi-agent dynamic multi-target collaboration tracking method based on finite-state automata - Google Patents
Multi-agent dynamic multi-target collaboration tracking method based on finite-state automata Download PDFInfo
- Publication number
- CN101303589A CN101303589A CNA2008100315434A CN200810031543A CN101303589A CN 101303589 A CN101303589 A CN 101303589A CN A2008100315434 A CNA2008100315434 A CN A2008100315434A CN 200810031543 A CN200810031543 A CN 200810031543A CN 101303589 A CN101303589 A CN 101303589A
- Authority
- CN
- China
- Prior art keywords
- state
- zhen
- information
- target
- input
- 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
Images
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/02—Total factory control, e.g. smart factories, flexible manufacturing systems [FMS] or integrated manufacturing systems [IMS]
Landscapes
- Image Analysis (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a multi-agent dynamic multi-objective cooperative tracking method which is based on a finite state automaton, which is characterized in that a combined agent selects one finite state automaton from a plurality of finite state automatons as the finite state automaton to maintain the behavior state model of the combined agent according to self-detected environmental information I, task information M which is sent by a server, other agents or a manager of agent groups and needs to be fulfilled, and/or artificially designated information H which is sent by the server. The agent is driven by behavior state, emotion information and other factors and can carry out centralized control or agent individual information interaction control by information interaction or by combining the server for team coordination. The method is suitable for different architectures such as centralized type, distributed type and hybrid type.
Description
Technical field
The invention belongs to robot navigation and application, relate to a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata.
Background technology
Complicated and the diversified feasible higher demand of having finished of modern task to multirobot combination and team's task, collaborative polynary Ai Zhen body is the video capture device by having possessed how, by information interaction and fusion, environment is scouted, instructed the ability of behavior with regard to particular importance by vision.But the present tracking that does not have a highly versatile in this respect can supply widespread use.
In a circumstances not known, use the observation of team collaboration of large-scale robot and tracking dynamic multi-objective with visual apparatus, need the asynchronous Ai Zhen phantom type that passes through on the multirobot that environment is observed in real time, and solution local problem, need simultaneously between whole many Ai Zhen system synchronously with communicate by letter, the real-time of guarantee information and accuracy, and make a strategic decision according to global information.Synchronization aspects can be by the timestamp in the agreement, many Ai Zhen body community information is mutual, perhaps carry out synchronously in conjunction with the global monitoring and the management of server end, the asynchronous Ai Zhen phantom type that then uses based on finite-state automata, by the location of state, be independent of action control platform, can transfer heterogeneous many Ai Zhen body of not being with video equipment, Distributed Calculation information, collaborative finishing the work.The two combination, compromise has preferably solved the synchronous and asynchronous problem of information and decision-making, increases the adaptivity of robotic team to environment, helps the efficient cooperative cooperating of robotic team, comprehensive, the accuracy of assurance scouting information and rapidity.
Summary of the invention
Technical matters to be solved by this invention is that a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata is provided.
The present invention solves the problems of the technologies described above the technical scheme that is adopted to be:
A kind of many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata, it is characterized in that: the artificial appointed information H that mission bit stream M that the needs that combined type Ai Zhen body is passed on according to environmental information I, server or other Ai Zhen bodies surveyed by self or Ai Zhen body population management person are finished and/or server are passed on, in a plurality of finite automaton state machines, select a finite automaton state machine conduct in order to keep the finite automaton state machine of this combined type Ai Zhen body behavior state model.
Described finite automaton state machine is full-automatic state machine and semi-automatic state machine; Full-automatic state machine is:
By waiting for, observe, follow the tracks of, lose and 5 states of having much to do being formed, the concrete transformational relation of each state is as follows:
(a), be input as " forbid connect " then to maintain the original state for " wait " state; Be input as " connection " and then switch to " observation " state;
(b), be input as " lose objects " and then maintain the original state for " observation " state; Being input as " forbid connect " then switches to " wait " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " losing " state; Be input as " discovery target " and then switch to " tracking " state;
(c), be input as " discovery target " and then maintain the original state for " tracking " state; Be input as " lose objects " and then switch to " losing " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " observation " state; Being input as " forbid connect " then switches to " wait " state;
(d), be input as " lose objects " and then maintain the original state for " losing " state; Be input as " discovery target " and then switch to " tracking " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " observation " state;
(e), be input as " forbid connect " then to switch to " wait " state for " having much to do " state; Be input as " task is finished " then according to record, be returned to the state before receiving an assignment, can be " observation " state, " losing " state or " tracking " state;
Semi-automatic state machine is:
By waiting for, observe, follow the tracks of, lose and 5 states of having much to do being formed, the concrete transformational relation of each state is as follows:
(a), be input as " forbid connect " then to maintain the original state for " wait " state; Be input as " connection " and then switch to " observation " state;
(b), be input as " lose objects " and then maintain the original state for " observation " state; Being input as " forbid connect " then switches to " wait " state; Be input as " task " and then switch to " having much to do " state; Be input as " discovery target " and then switch to " tracking " state;
(c), be input as " discovery target " and then maintain the original state for " tracking " state; Be input as " lose objects " and then switch to " losing " state; Be input as " task " and then switch to " having much to do " state;
Be input as " order " and then switch to " observation " state; Being input as " forbid connect " then switches to " wait " state;
(d), be input as " lose objects " and then maintain the original state for " losing " state; Be input as " discovery target " and then switch to " tracking " state; Be input as " task " and then switch to " having much to do " state;
Be input as " order " and then switch to " observation " state;
(e), be input as " forbid connect " then to switch to " wait " state for " having much to do " state; Be input as " task is finished " then according to record, be returned to the state before receiving an assignment, can be " observation " state, " losing " state or " tracking " state.
The behavior of described combined type Ai Zhen body can be expressed as M=(Q, ∑, δ, q
0, F), be a mathematical model that consists of the following components:
The finite set Q={ of a state waits for, observation is followed the tracks of, and loses, and is busy }, promptly
Q＝{Wait，Detect，Track，Lost，Busy}；
Acceptable input set ∑, it has indicated all and has allowed the symbol of input, and the finite automaton state machine carries out the variation of state according to the input of the symbol in this set, and it is expressed as follows:
∑=and connect, find target, lose objects forbids connecting, task, task is finished, order },
Be ∑={ connect, findobj, lostobj, unconnect, work, finishwork, order}
Initial state q
0={ wait }, i.e. q
0=Wait}, first state after Ai Zhenti opens under the situation that can't connect video equipment, will maintain this state;
Done state F={ waits for }, promptly F={Wait} finishes under the situation of cooperation tracking when Ai Zhenti can't continue initiatively participation, enter this state after disconnecting video equipment, before Ai Zhen body physical equipment finishes all tasks, also at first disconnect video equipment, Ai Zhen body colony is left in declaration;
Transfer function δ is the mapping of Q * ∑ → Q, is discerned by described finite automaton state machine.
Keep behavior state model and the real-time task and the constraint of environment by combined type Ai Zhen body finite automaton state machine, a plurality of combined type Ai Zhen bodies are separated into some colonies, colony's inner combined type Ai Zhen body can directly carry out the interchange of information; There is a Ai Zhen body population management person in each colony, can directly or indirectly exchange by Ai Zhen body population management person between each colony; Direct interchange between the colony is that mutual and/or timing information carries out alternately by the information updating excitation information, and described indirect communication is in many Ai Zhen system of server is arranged, and exchanges by server between colony.
Ai Zhen body population management person obtains three category informations from combined type Ai Zhen body individuality:
The first kind is a target information, is used to guarantee that the information of the object library on Ai Zhen body population management person or the server is up-to-date information;
Second class is the details of video equipment and the status information of combined type Ai Zhen body individuality; Ai Zhen body population management person is having under the server situation regularly from the new information of Ai Zhen body individual reception, and Ai Zhen body population management person is according to the content of blackboard in the update service device of institute's lastest imformation;
The 3rd category information is a solicited message; When combined type Ai Zhen body individuality runs into emergency condition, handle under the invalid situation in accident, can help request signal, write request sequence, excite request processing module; Request processing module can allow the colony at place carry out module schedules, satisfies the request that certain combined type Ai Zhen body sends; When single colony task can't be finished, this single colony dispatched other colonies again and assists.
Described combined type Ai Zhen body comprises following several sections:
1) a behavior state model of being kept with the finite automaton state machine comprises programme in the behavior state model, and the behavior state model selects appropriate module to carry out next step calculating from the module library according to current state and programme decision;
2) module library comprises video processing module and MBM, and module can be expanded according to demand; Video processing module comprise based on the target detection of three frame differences, morphology denoising, target cut apart, target is screened, target information is calculated, target merging and extraction and mean shift algorithm keeps track; MBM comprises the unexpected memory process of handling, forecasting process and self-shield processing;
3) communication module is used for and other combined types Ai Zhen body, Ai Zhen body population management person and/or server communication.
The negotiation scheme that is adopted comprises agreement protocol, machinery of consultation and three parts of negotiation approach:
1) agreement protocol
The form of agreement protocol comprises that multidigit starting and ending sign, instruction length consult first language, message numbering and message content; The first language of described negotiation comprises that request, order and information command, described message content comprise message recipient type, effective information content and message transmitting time;
2) machinery of consultation
A kind of machinery of consultation is the real-time change of the robotary tabulation that provides by Ai Zhen body population management person or server and global object storehouse, selection, task list and the request sequence information of making a strategic decision, activate a task allocation algorithm, competition negotiation algorithm, information calculations part and request processing module, obtain optimum negotiation scheme, with write-back task list as a result; The supvr sends to relevant combined type Ai Zhen body individuality with highest priority with instruction according to task list then.
Another machinery of consultation is in combined type Ai Zhen body colony, interactive information between the combined type Ai Zhen body individuality, and direct information merges mutually, consults to finish the work;
3) negotiation approach
Carry out communication by reconfigurable multiple mobile robot's point-to-point communication platform, combined type Ai Zhen body individuality can send to current information Ai Zhen body population management person when state variation, Ai Zhen body population management person interactive refreshing information, or update service device information; Do not having under the situation of state variation, upgrading once Ai Zhen body population management person's information, upgrading once the state and the information of other combined types Ai Zhen body individuality in the colony that Ai Zhen body population management person safeguarded simultaneously every time T.
Described many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata may further comprise the steps:
The first step, combined type Ai Zhen body is prepared: after combined type Ai Zhen body was opened, the behavior state model of keeping by the finite automaton state machine carried out activity, or exchanges with other combined types Ai Zhen body; When this combined type Ai Zhen body is opened, be in waiting status, promptly the wait state after combined type Ai Zhen body can connect video equipment, just leaves this state; Maintain the combined type Ai Zhen body individuality of this state, may can't oneself's observation external environment change, only wait for that receiving extraneous instruction carries out activity because the video equipment fault does not have video equipment or is designated as the activity of carrying out investigations that do not allow;
In second step, appointed task is finished: after combined type Ai Zhen body is opened,, then begin regularly to upgrade the information on the combined type Ai Zhen body population management person if in colony, if server is arranged, and update service device information simultaneously then; When receiving an appointed task, combined type Ai Zhen body is if full-automatic state then can convert semi-automatic state to; Accepting of task is if formation is arranged, and then all combined type Ai Zhen bodies that are in non-busy condition all enter task group, and task is distributed according to current combined type Ai Zhen body number of finishing the work, and and guide combined type Ai Zhen body to finish; If accepting of task is that intended target is searched or followed the tracks of, then all currently are in robot observer state or that certain is regional and enter task group and receive an assignment; Finding target with the some combined type Ai Zhen bodies that receive an assignment is sign, and the expression task is finished, and other combined types Ai Zhen body is abandoned this task in the notice task group; After combined type Ai Zhen body received an assignment, its state was a busy condition; The conversion of state is changed according to the instruction that distributes, and under the situation of the state that does not have appointment to change, then automatically returns to and enters busy condition state before, recovers the task point of preservation, continues to finish interrupted work;
The 3rd step, environment investigation: combined type Ai Zhen body is individual connect video equipment after, enter the observer state activity that carries out investigations automatically; Be in the combined type Ai Zhen body individuality of observer state, move individuality in the field of view scope, calculate and write down their information; If relate to server or other combined types Ai Zhen body, then with information sharing and notice; If full-automatic state is then chosen and is followed the tracks of target according to given rule, and change tracking mode over to, be i.e. the Track state; If semi-automatic state moves according to mandate; At full-automatic state, after target is extracted out, if traced into target, then arrive first first tracking, follow the tracks of effective area maximum, target and the nearest standard of combined type Ai Zhen body individual distance to satisfy, choose optimal objective and move and be transformed into tracking mode and follow the tracks of, other targets can be carried out basic vision and be followed the tracks of; If do not trace into target, then get back to observer state and continue observation;
The 4th step, target following: combined type Ai Zhen body individuality obtains mass data information by perceptron from external environment, by video processing module in the module library data are analyzed, the target information that obtains following the tracks of, and by MBM the information after analyzing is stored and calculated; Follow the tracks of and mainly realize by mean shift algorithm based on color; After the track rejection of being followed the tracks of, then target approach lost condition, i.e. Lost state;
The 5th step, the unexpected processing: comprise memory process and forecasting process;
1) memory process, memory be combined type Ai Zhen body individuality in tracking mode, a process with Given information is preserved in storage is divided into short-term memory and long-term memory; Short-term memory can be remembered the nearest action of being done, and long-term memory can be remembered in the whole process of following the tracks of a target, the path that resulting combined type Ai Zhen body is individual and target moves; According to the information of obtaining, information individual to combined type Ai Zhen body by curve fitting and the path that target moves is concluded; Carry out match in client and only consider the forefield data, it is carried out the curve fitting of segmentation; Combined type Ai Zhen body supvr or server end can calculate overall data;
2) forecasting process after the area that calculates target is less than certain threshold value, is thought track rejection; Then enter lost condition, i.e. the Lost state; Short-term memory generally is no more than three actions, and what play a major role is nearest action, other an auxiliary role of inspection; If by short-term memory guiding, do not find target once more, then the target trajectory that goes out of the polynomial curve fitting by long-term memory instructs combined type Ai Zhen body to forward the angle that is predicted to and observes; When combined type Ai Zhen body does not also find target in a specified time, think that target lost really, then send track rejection information.
Beneficial effect of the present invention has:
Many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata proposed by the invention is independent of action control platform, supports centralized control and Ai Zhen body individual information to control alternately.The proposition of many Ai Zhen of combined type body will be applicable to different many Ai Zhen system architectures, and the multiple goal cooperation that is fit to multirobot team is followed the tracks of.Each Ai Zhen physical efficiency works alone, and can also work in coordination with and finish the work, owing to be divided into different Ai Zhen body colonies according to task, so work efficiency is higher.When having server to participate in, server is total commander's maincenter, from the overall situation each Ai Zhen body of commander or colony's cooperation.
Description of drawings
Fig. 1 among the present invention based on the abstract application of many Ai Zhen body dynamic multi-objective collaboration tracking method of finite-state automata;
Fig. 2 is combined type Ai Zhen body Model Design framework among the present invention;
Fig. 3 among the present invention based on the state of many Ai Zhen body dynamic multi-objective collaboration tracking method behavior model of finite-state automata and the detailed call relation synoptic diagram between the module;
Fig. 4 the present invention is based on the finite automaton state machine of behavior model under full-automatic state in many Ai Zhen body dynamic multi-objective collaboration tracking method of finite-state automata;
Fig. 5 the present invention is based on the finite automaton state machine of behavior model under semi-automatic state in many Ai Zhen body dynamic multi-objective collaboration tracking method of finite-state automata;
Fig. 6 is the form of Ai Zhen body team agreement protocol among the present invention;
Fig. 7 is the Track state of behavior model and the module process flow diagram between the Lost state in many Ai Zhen body dynamic multi-objective collaboration tracking method among the present invention;
Embodiment
The invention will be further described below in conjunction with the drawings and specific embodiments.
Embodiment 1:
The present invention proposes a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata, and its application foundation is Ai Zhen system more than (Multi-Agent System is called for short MAS).This method is applied on the sustainable Ai Zhenti who independently plays a role, it selects a finite automaton state machine (Deterministic Finite Automation according to environment and mission requirements, be called for short DFA) keep the behavior state model of combined type Ai Zhen body, then in conjunction with the environmental information of video equipment sensor institute perception and the shared resource in the Ai Zhen body colony, cooperate and negotiation with other Ai Zhen body communications, carry out modeling, prediction, planning, decision-making, instruct Ai Zhen body intrasubject action control actuator necessarily to move.The present invention with the Ai Zhen body individual abstract be a combined type model, in this model, to Ai Zhen body self, the finite automaton state machine can freely select to keep behavior state according to environment and mission requirements, to Ai Zhen body social groups, can manage heterogeneous many Ai Zhen body team from an abstraction hierarchy.This Ai Zhenti is with behavior state and emotion information, tracking fatigue strength as Ai Zhenti, the factors such as individual character preference that with the evaluation function are the Ai Zhenti that reflects of measurement are driving, coordinate by information interaction or in conjunction with server team, carry out centralized control or Ai Zhen body individual information is controlled alternately.This multiple goal collaboration tracking method is applicable to different architecture such as centralized, distributed, hybrid.
Follow the tracks of the technical matters that exists in order to solve existing heterogeneous many Ai Zhen body multiple goal cooperation, the present invention has proposed a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata according to the characteristic of finite-state automata at this problem, this method is independent of action control platform, supports centralized control and Ai Zhen body individual information to control alternately.The proposition of this many Ai Zhen of combined type body will be applicable to different many Ai Zhen system architectures, has provided the scheme that the multiple goal cooperation of a kind of suitable multirobot team is followed the tracks of.
Each Ai Zhen body colony gets access to active data from each Ai Zhen body, with its fusion, share to all Ai Zhenti, and according to the task requests of updated information and Ai Zhen body individuality, call different decision-makings, realize Task Distribution, the cooperation of collaborative good colony reaches cooperative detection and the purpose of cooperating and following the tracks of at circumstances not known dynamic background dynamic multi-objective; In order to realize this function, on the leader Ai Zhenti of Ai Zhen body colony, comprise the Task Distribution part in the decision-making generation module, knowledge resources such as modeling target prediction instruct colony of whole robot to move.
Have a combined type Ai Zhen phantom type that can in the collaborative environment that calculates, continue independently to play a role on the Ai Zhen body individuality, by its environment of video equipment sensor senses, and by moving control action in this environment.The environmental information that it obtains by itself, and cooperate and negotiation by the shared resource in the Ai Zhen body colony with other Ai Zhen body communications, select self to use the finite automaton state machine to keep the behavior state model, carry out modeling, prediction, planning, decision-making instructs action control actuator to move, and acts on environment, if server is arranged, the respective resources on the then synchronous update service device.
The present invention is further illustrated below in conjunction with accompanying drawing.
1. Ai Zhen body colony is abstract
This method has been carried out the abstract of two levels by the state location based on the Ai Zhen phantom type of finite-state automata.The first, by the Ai Zhen phantom type that is independent of the motion control platform heterogeneous many Ai Zhen body is abstracted into the individual model of consistent Ai Zhen body; Second, keep the behavior state model with a plurality of Ai Zhen body individualities by the finite automaton state machine in the Ai Zhen phantom type, by the real-time task and the constraint of environment, should be separated into some groups by a plurality of individualities, there is a supvr that performance is stronger in each group, can directly or indirectly exchange by the supvr between each group then.And inner directly each the Ai Zhen body of group can directly carry out the interchange of information.Whole abstract form as shown in Figure 1.
Ai Zhen body population management person obtains three category informations from Ai Zhen body individuality.
The first kind is a target information, be used to guarantee that the information of the object library on population management person or the server is up-to-date information, one side trigger message calculating section, it carries out the polynomial curve piecewise fitting to the Given information of the target effective that is updated, and round-off error, on the other hand, according to the selection of decision-making, trigger task allocation algorithms and produce new tracking decision-making, tabulation sends instruction to relevant Ai Zhen body individuality according to robot task.
Second class is the details of video equipment and the status information of Ai Zhen body individuality.Population management person is regularly from the new information of Ai Zhen body individual reception, if server is arranged, then population management person is according to the content of the renewal blackboard of institute's lastest imformation.
The 3rd category information is a solicited message.When Ai Zhen body individuality runs into emergency condition, handle under the invalid situation in accident, can help request signal, write request sequence, excite request processing module.The Request Processing mould can allow the Ai Zhen body colony at place carry out module schedules, satisfies the request that certain Ai Zhen body sends.When single colony task can't be finished, dispatch other colonies again and assist.
2. the individual model of combined type Ai Zhen body
Combined type Ai Zhen body has carried out the design as Fig. 2 according to the method that proposes, and comprises following several sections:
1) one comprises programme with the behavior state model that DFA kept in the model, and model selects appropriate module to carry out next step calculating from the module library according to current state and programme decision;
2) module library mainly contains video processing module and MBM, and module can be expanded according to demand; Video processing module comprises the target detection based on three frame differences, the morphology denoising, and target is cut apart, and target is screened, and target information is calculated, and target merges and extracts Meanshift tracking (mean shift algorithm) etc.; MBM comprises the unexpected memory process of handling, and forecasting process and self-shield are handled.
3) communication module
3. the behavior state model of Ai Zhen body
The realization entity of behaviour decision making layer has been formed in the behavior state model that DFA keeps and the module library of video processing module and MBM, can be expressed as M=(Q, ∑, δ, a q
0, mathematical model F) is described in detail as follows.
1) the finite set Q={Wait of a state, Detect, Track, Lost, five states are arranged: waiting status: Wait, observer state: Detect, tracking mode: Track in the Busy} state set, lose tracking mode, abbreviate tracking mode as: Lost, busy condition: Busy.Detailed introduction is referring to the detailed introduction of behavior state model.Detect state in the state set, the programme in Track state and the Lost state is to the complete covering of independent behaviour planning having carried out of Ai Zhenti.The programme of these three kinds of states, video processing module that provides in the binding modules storehouse and MBM are handled the information that obtains, and the behavior of Ai Zhenti are provided the decision-making of self according to the result.The module that these three kinds of states and their pairing programmes will be called concerns as shown in Figure 3 in detail.
2) acceptable input set ∑, it has indicated all and has allowed the symbol of input, and the finite automaton state machine carries out the variation of state according to the input of the symbol in this set, is expressed as follows:
∑=and connect, find target, lose objects forbids connecting, task, task is finished, order }, promptly
∑＝{connect，findobj，lostobj，unconnect，work，finishwork，order}
Comprise 7 symbols that can import in this set, their representatives generation of corresponding physical event in practice is described in detail as follows:
Connect connect: the available video equipment that carries out of the individual success of expression Ai Zhen body connects;
Find target findobj: expression Ai Zhen body individuality in the scope that the visual field can reach, searches the unknown or known target by the current video image that obtains under existing state;
Lose objects lostobj: expression Ai Zhen body individuality can reach in the visual field in the scope by the current video image that obtains under standing state, can't find one and the identical target of historical information;
Forbid connecting unconnect: expression Ai Zhen body individuality stipulates that it can not use video equipment to obtain environmental information and initiatively participates in collaborative tracking activity when physical condition restriction or integral body are carried out resource allocation;
Work: expression Ai Zhen body individuality is accepted specified command in current state, distributes new task, and after accepting the work instruction, Ai Zhenti enters the Busy state, does not allow self to carry out new task allotment.
Task is finished finishwork: expression Ai Zhen body individuality is finished tasks all in the instruction queue.After finishing the work, the state before Ai Zhenti returns to and takes orders continues to finish the task that designated order is interrupted;
Order order: expression Ai Zhen body individuality is accepted instruction, is transformed into new state, finishes the task of appointment.
3) initial state q
0={ Wait} after the Ai Zhen body is opened, directly enters this state, if can't connect video equipment, then maintains this state.
4) done state F={Wait} when Ai Zhenti can't continue initiatively to participate in finishing the cooperation tracking, then disconnects video equipment, enter this state, before before Ai Zhen body physical equipment finishes all tasks, also at first disconnect video equipment, Ai Zhen body colony is left in declaration.
5) transfer function δ is the mapping of Q * ∑ → Q, is discerned by the finite automaton state machine.The finite automaton state machine is divided into two kinds according to the behavior state of Ai Zhen body individuality, and a kind of is finite automaton state machine under the full-automatic state, and a kind of is finite automaton state machine under the semi-automatic state.
Full-automatic state: in Ai Zhen body action process, under the situation that does not have other Ai Zhen bodies and server info to support, also can finish autonomous discovery target, carry out tracing task, the state of work such as other Ai Zhen body individualities of independent searching is called as full-automatic state.
Semi-automatic state: in Ai Zhen body action process, under the situation of specifying cooperation, Ai Zhenti accepts instruction, follow the tracks of and search intended target, carry out communication with known Ai Zhenti, cooperation is finished the work, and can not arbitrarily abandon existing task, carry out other unauthorized independent behaviours, be called as semi-automatic state.
The state set of the finite automaton state machine under full-automatic state and the semi-automatic state is consistent, and two states can freely be changed according to actual environment and mission requirements.Freely change the environmental information I that surveys by self, the artificial appointed information H that mission bit stream M that the needs that server or other Ai Zhenti pass on are finished and server are passed on, in multiple finite automaton state machine, with N=f (I, M, H) for selecting index, select optimum automatic state machine.If comprise target information in the environmental information that current Ai Zhen body self is surveyed, and Ai Zhen body population management person or server do not pass on assignment instructions, and then this Ai Zhenti moves with full-automatic state; And if Ai Zhen body population management person or server have been passed on assignment instructions, then select semi-automatic state, with the cooperation purpose, the instruction of preferentially executing the task after task is finished, is got back to full-automatic state, carries out oneself's decision-making then; If receive manual command's information, then be in semi-automaticly, unconditional conformance instruction is finished up to task, receives till other instructions.
The concrete transfer function of finite automaton state machine can be referring to Fig. 4 and Fig. 5.
4. negotiation scheme
The negotiation scheme that this method adopts comprises agreement protocol, machinery of consultation and three parts of negotiation approach.
1) agreement protocol
Communication is the key link that the present embodiment method realizes cooperation, and it has played function served as bridge in the process that realizes cooperation.The form of agreement protocol as shown in Figure 6 in this method.Multidigit starting and ending sign, instruction length all is in order to guarantee in the transmission of network, instruction of parsing that can be complete because when information package is transmitted, not necessarily once can provide a complete information, therefore in the information that receives continuously, need the instruction reduction that can will be split.Message numbering and message transmitting time be for when network blockage occurring and retransmitting, and assurance need not repeat or out-of-date information is upgraded data in Ai Zhen body population management person or server and the Ai Zhen body individuality.
2) machinery of consultation
The present invention can adopt following two kinds of machineries of consultation, a kind of is the real-time change of the robotary tabulation that provides by Ai Zhen body population management person or server and global object storehouse, the information such as selection, task list, request sequence of making a strategic decision, knowledge resources such as activate a task allocation algorithm, competition negotiation algorithm, information calculations part and request processing module, obtain optimum negotiation scheme, with write-back task list as a result.The supvr sends to relevant Ai Zhen body individuality with highest priority with instruction according to task list then.Another machinery of consultation is an interactive information between Ai Zhen body individuality and the Ai Zhen body individuality, and direct information merges mutually, consults to finish the work.Selection between the Ai Zhen body individuality be by after the information interaction between the Ai Zhen body population management person or server carry out group's division result according to task.The Ai Zhen body individuality that satisfies simulated condition will be divided into a Ai Zhen body group (i.e. Ai Zhen body colony), the mutual information needed of other members in Ai Zhen body individuality and the Ai Zhen body group.
3) negotiation approach
This heterogeneous many Ai Zhen body carries out communication by reconfigurable multiple mobile robot's point-to-point communication platform, Ai Zhen body individuality can send to current information Ai Zhen body population management person when state variation, Ai Zhen body population management person interactive refreshing information, or update service device information (existing under the situation of server).Do not having under the situation of state variation, upgrading once Ai Zhen body population management person's information, upgrading once the state and the information of other Ai Zhen body individualities in the group that it safeguards simultaneously every time T.
The present invention is applied in a circumstances not known, uses on the large-scale robotic team with visual apparatus, be used to cooperate observation with follow the tracks of dynamic multi-objective, key step is as follows:
The first step, Ai Zhenti prepares: after a Ai Zhen body was opened, the behavior state model that it is just kept by DFA carried out activity, or exchanges with other Ai Zhen bodies.When this Ai Zhen body is opened, be in the Wait state, after Ai Zhenti can connect video equipment, just leave this state.Maintain the Ai Zhen body individuality of this state, may can't the oneself detect the external environment variation because the video equipment fault does not have video equipment or is designated as the activity of carrying out investigations that do not allow.Only wait for that receiving extraneous instruction carries out activity;
In second step, appointed task is finished: after the Ai Zhen body is opened,, then begin regularly to upgrade the information on the Ai Zhen body population management person if in colony, if server is arranged, and update service device information simultaneously then.Server is designed to accept the information of manual intervention part.When receiving an appointed task, Ai Zhenti is if full-automatic state then can convert semi-automatic state to.Accepting of task is if formation is arranged, and all Ai Zhenti that are in non-Busy state enter task group, and task is distributed according to current Ai Zhen body number of finishing the work, and and guide them to finish.If accepting of task is that intended target is searched or followed the tracks of, then all the current Detect of being in states or that certain is regional robot enters task group, receives an assignment.Finding target with the some Ai Zhen bodies that receive an assignment is sign, and the expression task is finished, and other machines people in the notice task group, abandons this task.If after Ai Zhenti receives an assignment, then will be in a highest state of state medium priority of Ai Zhen body individuality: the Busy state.The individual current work that is assigned of finishing of Ai Zhen body is described, can not be interrupted that other order should be carried out after it finishes this work by the instruction of self.The conversion of state is changed according to the instruction that distributes, and under the situation of the state that does not have appointment to change, then automatically returns to and enters busy condition state before, recovers the task point of preservation, continues to finish interrupted work.
In the 3rd step, the environment investigation: behind the individual connection of the Ai Zhen body video equipment, enter observer state automatically, activity carries out investigations.Be in the Ai Zhen body individuality of observer state, move individuality in the field of view scope, calculate and write down their information.If relate to server or other Ai Zhen bodies, then with information sharing and notice.If full-automatic state is then chosen and is followed the tracks of target according to given rule, and change the Track state over to, if semi-automatic state moves according to mandate; For moving individual investigation within sweep of the eye, the technology that adopts is that elder generation's use three frame difference background methods of wiping out are come the extraction to target, then by morphology denoising and gaussian filtering, the separating part that obtains in the space is made mask, carrying out target cuts apart, target merges and extracts, and carries out iterative original object information searching at last, and uses watershed algorithm to obtain single target information.By the principle that the target of definition is screened they are screened, the principle of its target screening is as follows:
1. target focus point position is close, and the aim colour tone pitch is similar, and the area size is similar, then think it be cut apart by same target mistake due to, choose a bigger reservation of area, get the focus point average, the tone average.
2. realistic objective observation, can not occur rgb value substantially and be 0 or the H value be 0 situation.
3. the general area of noise that the light influence is measured during driftlessness is very big, but when having target, because the light influence, the noise information of measuring is less.The physical action of Ai Zhen body individuality also may produce part small size information, need such pinpoint target be screened.
Needing to target designation after the target screening to carry out, wherein need be by the distribution of Ai Zhen body population management person or server to target designation.After the definition screening target, earlier according to tone value, give Target Assignment tone numbering, server sends to each Ai Zhen body individuality with information encoding then, and whether check in object library has this tone numbering target, if do not have, can be 1 to target designation then.If the target of existing this tone numbering then checks whether be same target.By the target Given information, the position of target center of gravity in image, the position and the direction of Ai Zhen body individuality, the data messages of sonar etc. can draw the position or the direction of target.If in global map, this target is similar with known target information, then thinks a known target, distributes known numbering, and its information is notified to the Ai Zhen body individuality that other follow the tracks of this target.Follow the tracks of several Ai Zhen body individualities of same target then, after process in, several Ai Zhen body individualities are revised mutual exchange of information to information.If not same target then according to known numbering, continues to distribute next numbering.
At full-automatic state, after target is extracted out, if traced into target, then arrive first first tracking to satisfy, tracking effective area maximum, target and the nearest standard of Ai Zhen body individual distance are chosen optimal objective and are moved and be transformed into the Track state and follow the tracks of, and other targets can be carried out basic vision and be followed the tracks of.If do not trace into target, then get back to the Detect state and continue observation.
The 4th the step, target following: in the Ai Zhen body action process and the maximum state of environmental interaction be exactly Track, tracking mode.In this state, Ai Zhen body individuality obtains mass data information by perceptron from external environment, by video processing module in the module library data are analyzed, the target information that obtains following the tracks of, and by MBM the information after analyzing is stored and calculated.Because two invariants have been adopted in the description of target: colouring information and profile information and a variable area information represent that therefore following the tracks of the algorithm that adopts is mainly to realize by the Meanshift algorithm based on color.If work as after the track rejection of being followed the tracks of, then enter Lost, the track rejection state.On the theoretical analysis, the conversion determinacy between Track state and the Lost state is very strong, when target within sweep of the eye, just can not lose and state exchange.But in a mobile Ai Zhenti and unknown environment, can produce certain inertia because Ai Zhen body physics moves, the shadow of environment can move and changes along with Ai Zhenti, and in the environment of many Ai Zhen body, blocking also is very important problem.When the state exchange determinacy was strong excessively, Ai Zhenti will be too high to the Loss Rate of target following.Therefore, in method, designed the process of unexpected processing.When target is lost, at first, instruct and carry out certain action by the memory models in unexpected the processing, repeatedly search and confirm, can't tracking target if this process of track rejection is repeatedly confirmed still, think that then this target finally loses, carry out next step action;
In the 5th step, the unexpected processing: at light, under the various unpredictable condition influence such as physics inertia, it is the very possible thing that occurs that the accident of target is lost.Therefore in order to handle this accident, for Ai Zhen body individual design memory and forecast function.
(1) memory process, memory be Ai Zhen body individuality in the Track state, a process with Given information is preserved in container is divided into short-term memory and long-term memory.Short-term memory can be remembered the nearest action of being done, and long-term memory can be remembered in the whole process of following the tracks of a target, the path that resulting Ai Zhen body is individual and target moves.After having got access to a certain amount of information, can use curve fitting information individual to the Ai Zhen body and the path that target moves to conclude.In client, because individual performance limitations of Ai Zhen body and view data are handled, considerable tasks such as communication, and the track uncertainty of target travel carry out match and only consider the forefield data, and it is carried out the curve fitting of segmentation.Ai Zhen body supvr or server end can calculate in more detail by overall data
(2) forecasting process after the area that calculates target is less than certain threshold value, can be thought track rejection.Then enter the Lost state.If after the lose objects, do not move by short-term memory temporarily, then finish an interim action earlier, observe again.Short-term memory generally is no more than three actions, and what play a major role is nearest action, other an auxiliary work for inspection.If by short-term memory guiding, do not find target once more, then the target trajectory that goes out of the polynomial curve fitting by long-term memory instructs Ai Zhenti to forward the angle that is predicted to and observes.When Ai Zhenti does not also find target in a specified time, think that then target lost really, then send information to server, server will instruct Ai Zhen body colony to make corresponding decision and action.
Claims (8)
1, a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata, it is characterized in that: the artificial appointed information H that mission bit stream M that the needs that combined type Ai Zhen body is passed on according to environmental information I, server or other Ai Zhen bodies surveyed by self or Ai Zhen body population management person are finished and/or server are passed on, in a plurality of finite automaton state machines, select a finite automaton state machine conduct in order to keep the finite automaton state machine of this combined type Ai Zhen body behavior state model.
2, a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method as claimed in claim 1 based on finite-state automata, it is characterized in that: described finite automaton state machine is full-automatic state machine and semi-automatic state machine;
Full-automatic state machine is:
By waiting for, observe, follow the tracks of, lose and 5 states of having much to do being formed, the concrete transformational relation of each state is as follows:
(a), be input as " forbid connect " then to maintain the original state for " wait " state; Be input as " connection " and then switch to " observation " state;
(b), be input as " lose objects " and then maintain the original state for " observation " state; Being input as " forbid connect " then switches to " wait " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " losing " state; Be input as " discovery target " and then switch to " tracking " state;
(c), be input as " discovery target " and then maintain the original state for " tracking " state; Be input as " lose objects " and then switch to " losing " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " observation " state; Being input as " forbid connect " then switches to " wait " state;
(d), be input as " lose objects " and then maintain the original state for " losing " state; Be input as " discovery target " and then switch to " tracking " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " observation " state;
(e), be input as " forbid connect " then to switch to " wait " state for " having much to do " state; Be input as " task is finished " then according to record, be returned to the state before receiving an assignment, can be " observation " state, " losing " state or " tracking " state;
Semi-automatic state machine is:
By waiting for, observe, follow the tracks of, lose and 5 states of having much to do being formed, the concrete transformational relation of each state is as follows:
(a), be input as " forbid connect " then to maintain the original state for " wait " state; Be input as " connection " and then switch to " observation " state;
(b), be input as " lose objects " and then maintain the original state for " observation " state; Being input as " forbid connect " then switches to " wait " state; Be input as " task " and then switch to " having much to do " state; Be input as " discovery target " and then switch to " tracking " state;
(c), be input as " discovery target " and then maintain the original state for " tracking " state; Be input as " lose objects " and then switch to " losing " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " observation " state; Being input as " forbid connect " then switches to " wait " state;
(d), be input as " lose objects " and then maintain the original state for " losing " state; Be input as " discovery target " and then switch to " tracking " state; Be input as " task " and then switch to " having much to do " state; Be input as " order " and then switch to " observation " state;
(e), be input as " forbid connect " then to switch to " wait " state for " having much to do " state; Be input as " task is finished " and then switch to " observation " state; Be input as " task is finished " then according to record, be returned to the state before receiving an assignment, can be " observation " state, " losing " state or " tracking " state.
3, a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method based on finite-state automata as claimed in claim 2, it is characterized in that: the behavior of described combined type Ai Zhen body can be expressed as M=(Q, ∑, δ, q
0, F), be a mathematical model that consists of the following components:
The finite set Q={ of a state waits for, observation is followed the tracks of, and loses, and is busy }, promptly
Q＝{Wait，Detect，Track，Lost，Busy}；
Acceptable input set ∑, it has indicated all and has allowed the symbol of input, and the finite automaton state machine carries out the variation of state according to the input of the symbol in this set, and it is expressed as follows:
∑=and connect, find target, lose objects forbids connecting, task, task is finished, order },
Be ∑={ connect, findobj, lostobj, unconnect, work, finishwork, order}
Initial state q
0={ wait }, i.e. q
0=Wait}, first state after Ai Zhenti opens under the situation that can't connect video equipment, will maintain this state;
Done state F={ waits for }, promptly F={Wait} finishes under the situation of cooperation tracking when Ai Zhenti can't continue initiatively participation, enter this state after disconnecting video equipment, before Ai Zhen body physical equipment finishes all tasks, also at first disconnect video equipment, Ai Zhen body colony is left in declaration;
Transfer function δ is the mapping of Q * ∑ → Q, is discerned by described finite automaton state machine.
4, a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method as claimed in claim 1 based on finite-state automata, it is characterized in that: keep behavior state model and the real-time task and the constraint of environment by combined type Ai Zhen body finite automaton state machine, a plurality of combined type Ai Zhen bodies are separated into some colonies, and colony's inner combined type Ai Zhen body can directly carry out the interchange of information; There is a Ai Zhen body population management person in each colony, can directly or indirectly exchange by Ai Zhen body population management person between each colony; Direct interchange between the colony is that mutual and/or timing information carries out alternately by the information updating excitation information, and described indirect communication is in many Ai Zhen system of server is arranged, and exchanges by server between colony.
5, a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method as claimed in claim 4 based on finite-state automata, it is characterized in that: Ai Zhen body population management person obtains three category informations from combined type Ai Zhen body individuality:
The first kind is a target information, is used to guarantee that the information of the object library on Ai Zhen body population management person or the server is up-to-date information;
Second class is the details of video equipment and the status information of combined type Ai Zhen body individuality; Ai Zhen body population management person is having under the server situation regularly from the new information of Ai Zhen body individual reception, and Ai Zhen body population management person is according to the content of blackboard in the update service device of institute's lastest imformation;
The 3rd category information is a solicited message; When combined type Ai Zhen body individuality runs into emergency condition, handle under the invalid situation in accident, can help request signal, write request sequence, excite request processing module; Request processing module can allow the colony at place carry out module schedules, satisfies the request that certain combined type Ai Zhen body sends; When single colony task can't be finished, this single colony dispatched other colonies again and assists.
6. a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method as claimed in claim 4 based on finite-state automata, it is characterized in that: described combined type Ai Zhen body comprises following several sections:
1) a behavior state model of being kept with the finite automaton state machine comprises programme in the behavior state model, and the behavior state model selects appropriate module to carry out next step calculating from the module library according to current state and programme decision;
2) module library comprises video processing module and MBM, and module can be expanded according to demand; Video processing module comprise based on the target observation of three frame differences, morphology denoising, target cut apart, target is screened, target information is calculated, target merging and extraction and mean shift algorithm keeps track; MBM comprises the unexpected memory process of handling, forecasting process and self-shield processing;
3) communication module is used for and other combined types Ai Zhen body, Ai Zhen body population management person and/or server communication.
7. a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method as claimed in claim 4 based on finite-state automata, it is characterized in that: the negotiation scheme that is adopted comprises agreement protocol, machinery of consultation and three parts of negotiation approach:
1) agreement protocol
The form of agreement protocol comprises that multidigit starting and ending sign, instruction length consult first language, message numbering and message content; The first language of described negotiation comprises that request, order and information command, described message content comprise message recipient type, effective information content and message transmitting time;
2) machinery of consultation
A kind of machinery of consultation is the real-time change of the robotary tabulation that provides by Ai Zhen body population management person or server and global object storehouse, selection, task list and the request sequence information of making a strategic decision, activate a task allocation algorithm, competition negotiation algorithm, information calculations part and request processing module, obtain optimum negotiation scheme, with write-back task list as a result; The supvr sends to relevant combined type Ai Zhen body individuality with highest priority with instruction according to task list then.
Another machinery of consultation is in combined type Ai Zhen body colony, interactive information between the combined type Ai Zhen body individuality, and direct information merges mutually, consults to finish the work;
3) negotiation approach
Carry out communication by reconfigurable multiple mobile robot's point-to-point communication platform, combined type Ai Zhen body individuality can send to current information Ai Zhen body population management person when state variation, Ai Zhen body population management person interactive refreshing information, or update service device information; Do not having under the situation of state variation, upgrading once Ai Zhen body population management person's information, upgrading once the state and the information of other combined types Ai Zhen body individuality in the colony that Ai Zhen body population management person safeguarded simultaneously every time T.
8. as each described a kind of many Ai Zhen body dynamic multi-objective collaboration tracking method of claim 1 ~ 7, it is characterized in that, may further comprise the steps based on finite-state automata:
The first step, combined type Ai Zhen body is prepared: after combined type Ai Zhen body was opened, the behavior state model of keeping by the finite automaton state machine carried out activity, or exchanges with other combined types Ai Zhen body; When this combined type Ai Zhen body is opened, be in waiting status, promptly the Wait state after combined type Ai Zhen body can connect video equipment, just leaves this state; Maintain the combined type Ai Zhen body individuality of this state, may can't oneself's observation external environment change, only wait for that receiving extraneous instruction carries out activity because the video equipment fault does not have video equipment or is designated as the activity of carrying out investigations that do not allow;
In second step, appointed task is finished: after combined type Ai Zhen body is opened,, then begin regularly to upgrade the information on the combined type Ai Zhen body population management person if in colony, if server is arranged, and update service device information simultaneously then; When receiving an appointed task, combined type Ai Zhen body is if full-automatic state then can convert semi-automatic state to; Accepting of task is if formation is arranged, and then all combined type Ai Zhen bodies that are in non-busy condition all enter task group, and task is distributed according to current combined type Ai Zhen body number of finishing the work, and and guide combined type Ai Zhen body to finish; If accepting of task is that intended target is searched or followed the tracks of, then all currently are in robot observer state or that certain is regional and enter task group and receive an assignment; Finding target with the some combined type Ai Zhen bodies that receive an assignment is sign, and the expression task is finished, and other combined types Ai Zhen body is abandoned this task in the notice task group; After combined type Ai Zhen body received an assignment, its state was a busy condition; The conversion of state is changed according to the instruction that distributes, and under the situation of the state that does not have appointment to change, then automatically returns to and enters busy condition state before, recovers the task point of preservation, continues to finish interrupted work;
The 3rd step, environment investigation: combined type Ai Zhen body is individual connect video equipment after, enter the observer state activity that carries out investigations automatically; Be in the combined type Ai Zhen body individuality of observer state, move individuality in the field of view scope, calculate and write down their information; If relate to server or other combined types Ai Zhen body, then with information sharing and notice; If full-automatic state is then chosen and is followed the tracks of target according to given rule, and change tracking mode over to, be i.e. the Track state; If semi-automatic state moves according to mandate; At full-automatic state, after target is extracted out, if traced into target, then arrive first first tracking, follow the tracks of effective area maximum, target and the nearest standard of combined type Ai Zhen body individual distance to satisfy, choose optimal objective and move and be transformed into tracking mode and follow the tracks of, other targets can be carried out basic vision and be followed the tracks of; If do not trace into target, then get back to observer state and continue observation;
The 4th step, target following: combined type Ai Zhen body individuality obtains mass data information by perceptron from external environment, by video processing module in the module library data are analyzed, the target information that obtains following the tracks of, and by MBM the information after analyzing is stored and calculated; Follow the tracks of and mainly realize by mean shift algorithm based on color; After the track rejection of being followed the tracks of, then target approach lost condition, i.e. Lost state;
The 5th step, the unexpected processing: comprise memory process and forecasting process;
1) memory process, memory be combined type Ai Zhen body individuality in tracking mode, a process with Given information is preserved in storage is divided into short-term memory and long-term memory; Short-term memory can be remembered the nearest action of being done, and long-term memory can be remembered in the whole process of following the tracks of a target, the path that resulting combined type Ai Zhen body is individual and target moves; According to the information of obtaining, information individual to combined type Ai Zhen body by curve fitting and the path that target moves is concluded; Carry out match in client and only consider the forefield data, it is carried out the curve fitting of segmentation; Combined type Ai Zhen body supvr or server end can calculate overall data;
2) forecasting process after the area that calculates target is less than certain threshold value, is thought track rejection; Then enter lost condition, i.e. the Lost state; Short-term memory generally is no more than three actions, and what play a major role is nearest action, other an auxiliary role of inspection; If by short-term memory guiding, do not find target once more, then the target trajectory that goes out of the polynomial curve fitting by long-term memory instructs combined type Ai Zhen body to forward the angle that is predicted to and observes; When combined type Ai Zhen body does not also find target in a specified time, think that target lost really, then send track rejection information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008100315434A CN101303589B (en) | 2008-06-20 | 2008-06-20 | Multi-agent dynamic multi-target collaboration tracking method based on finite-state automata |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN2008100315434A CN101303589B (en) | 2008-06-20 | 2008-06-20 | Multi-agent dynamic multi-target collaboration tracking method based on finite-state automata |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN101303589A true CN101303589A (en) | 2008-11-12 |
| CN101303589B CN101303589B (en) | 2011-04-13 |
Family
ID=40113518
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN2008100315434A Expired - Fee Related CN101303589B (en) | 2008-06-20 | 2008-06-20 | Multi-agent dynamic multi-target collaboration tracking method based on finite-state automata |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN101303589B (en) |
Cited By (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101937341A (en) * | 2010-09-09 | 2011-01-05 | 西北工业大学 | Autonomic multi-mode management method |
| CN102176122A (en) * | 2011-03-07 | 2011-09-07 | 华中科技大学 | Automatic control system based on finite automaton and control method thereof |
| WO2014101558A1 (en) * | 2012-12-27 | 2014-07-03 | 深圳先进技术研究院 | Processing method and system for controlling service amounts of multiple robots |
| CN104076800A (en) * | 2014-07-10 | 2014-10-01 | 哈尔滨工程大学 | UUV group task coordination system and method |
| CN104246749A (en) * | 2012-04-13 | 2014-12-24 | Infnis网络公司 | Method, server, terminal device, and computer-readable recording medium for selectively removing nondeterminism of nondeterministic finite automata |
| CN105637432A (en) * | 2013-08-30 | 2016-06-01 | 慧与发展有限责任合伙企业 | Identifying anomalous behavior of a monitored entity |
| CN106681322A (en) * | 2016-12-21 | 2017-05-17 | 华东师范大学 | Ground autonomous mobile robot safety navigation method based on formal description |
| CN109416867A (en) * | 2016-07-04 | 2019-03-01 | 索尼公司 | Robot, robot system and recording medium |
| CN111596691A (en) * | 2020-07-07 | 2020-08-28 | 福州大学 | Decision modeling and cooperative control method and system of multi-robot system based on human-in-loop |
-
2008
- 2008-06-20 CN CN2008100315434A patent/CN101303589B/en not_active Expired - Fee Related
Cited By (14)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN101937341B (en) * | 2010-09-09 | 2014-07-30 | 西北工业大学 | Autonomic multi-mode management method |
| CN101937341A (en) * | 2010-09-09 | 2011-01-05 | 西北工业大学 | Autonomic multi-mode management method |
| CN102176122A (en) * | 2011-03-07 | 2011-09-07 | 华中科技大学 | Automatic control system based on finite automaton and control method thereof |
| CN104246749B (en) * | 2012-04-13 | 2016-01-13 | Infnis网络公司 | Probabilistic method of non deterministic finite automaton, server and end device is eliminated for selectivity |
| CN104246749A (en) * | 2012-04-13 | 2014-12-24 | Infnis网络公司 | Method, server, terminal device, and computer-readable recording medium for selectively removing nondeterminism of nondeterministic finite automata |
| WO2014101558A1 (en) * | 2012-12-27 | 2014-07-03 | 深圳先进技术研究院 | Processing method and system for controlling service amounts of multiple robots |
| CN105637432A (en) * | 2013-08-30 | 2016-06-01 | 慧与发展有限责任合伙企业 | Identifying anomalous behavior of a monitored entity |
| CN104076800A (en) * | 2014-07-10 | 2014-10-01 | 哈尔滨工程大学 | UUV group task coordination system and method |
| CN109416867A (en) * | 2016-07-04 | 2019-03-01 | 索尼公司 | Robot, robot system and recording medium |
| US11200787B2 (en) | 2016-07-04 | 2021-12-14 | Sony Corporation | Robot, robot system, and storage medium |
| US11837064B2 (en) | 2016-07-04 | 2023-12-05 | Sony Group Corporation | Robot, robot system, and storage medium |
| CN106681322A (en) * | 2016-12-21 | 2017-05-17 | 华东师范大学 | Ground autonomous mobile robot safety navigation method based on formal description |
| CN111596691A (en) * | 2020-07-07 | 2020-08-28 | 福州大学 | Decision modeling and cooperative control method and system of multi-robot system based on human-in-loop |
| CN111596691B (en) * | 2020-07-07 | 2021-08-31 | 福州大学 | Decision modeling and cooperative control method and system of multi-robot system based on human-in-loop |
Also Published As
| Publication number | Publication date |
|---|---|
| CN101303589B (en) | 2011-04-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN101303589B (en) | Multi-agent dynamic multi-target collaboration tracking method based on finite-state automata | |
| CN111369181B (en) | Train autonomous scheduling deep reinforcement learning method and device | |
| CN101615265B (en) | Intelligent decision simulating experimental system based on multi-Agent technology | |
| CN110209716A (en) | Intelligent internet of things water utilities big data processing method and system | |
| CN109995546A (en) | The intelligent plant automatic system architecture that edge calculations are cooperateed with cloud computing | |
| CN108229685A (en) | A kind of unmanned Intelligent Decision-making Method of vacant lot one | |
| CN110533685B (en) | Object tracking method and device, storage medium and electronic device | |
| Liu et al. | A proposed radicality evaluation method for design ideas at conceptual design stage | |
| CN109086550A (en) | The evacuation emulation method and system of Q study are shared based on multi-Agent | |
| Yang et al. | An information fusion approach to intelligent traffic signal control using the joint methods of multiagent reinforcement learning and artificial intelligence of things | |
| CN101083019A (en) | Rapid evaluating system based on roomage state sensing | |
| CN110084405A (en) | Flexible intelligent of handling up assembles logistics route planing method | |
| CN108449383A (en) | Distributed thin cloud computing system mobile in real time | |
| CN110222924A (en) | Multimode track traffic station passenger flow control system and control method | |
| CN106127365A (en) | Quantitative remote sensing On-line Product interactive mode autonomous production method | |
| Darmoul et al. | Artificial immunity to control disturbances in public transportation systems: concepts, mechanisms and a prototype implementation of a knowledge based decision support system | |
| CN108647886A (en) | Scientific algorithm process management system | |
| TW202013304A (en) | A system of virtual and physical integrated network service fulfillment and monitor based on artificial intelligence | |
| TW202019133A (en) | Software defined driven ict service provider system based on end to end orchestration | |
| Zhu et al. | An agent-based support system for railway station dispatching | |
| Berger et al. | Semi-Heterarchical allocation and routing processes in FMS control: a stigmergic approach | |
| Wang et al. | Human-drone collaborative spatial crowdsourcing by memory-augmented and distributed multi-agent deep reinforcement learning | |
| CN109947130A (en) | A kind of transmission of electricity multiple aircraft formation tour management method and system based on grid-based management | |
| CN109450587A (en) | Spectral integration processing method, device and storage medium | |
| CN110782039B (en) | Artificial intelligence instant combat guide platform based on layered structure and multiple modules |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| C06 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C14 | Grant of patent or utility model | ||
| GR01 | Patent grant | ||
| C17 | Cessation of patent right | ||
| CF01 | Termination of patent right due to non-payment of annual fee |
Granted publication date: 20110413 Termination date: 20110620 |