CN108985580A - Multirobot disaster based on improved BP searches and rescues method for allocating tasks - Google Patents
Multirobot disaster based on improved BP searches and rescues method for allocating tasks Download PDFInfo
- Publication number
- CN108985580A CN108985580A CN201810673369.7A CN201810673369A CN108985580A CN 108985580 A CN108985580 A CN 108985580A CN 201810673369 A CN201810673369 A CN 201810673369A CN 108985580 A CN108985580 A CN 108985580A
- Authority
- CN
- China
- Prior art keywords
- task
- robot
- auctioneer
- auction
- rescue
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06312—Adjustment or analysis of established resource schedule, e.g. resource or task levelling, or dynamic rescheduling
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
- G06N3/084—Backpropagation, e.g. using gradient descent
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/06—Resources, workflows, human or project management; Enterprise or organisation planning; Enterprise or organisation modelling
- G06Q10/063—Operations research, analysis or management
- G06Q10/0631—Resource planning, allocation, distributing or scheduling for enterprises or organisations
- G06Q10/06316—Sequencing of tasks or work
Abstract
The invention discloses the multirobot disasters based on improved BP to search and rescue method for allocating tasks.When auctioneer does not have enough abilities to complete current search and rescue task, the mode that the use of information of the task is broadcasted is distributed to bidder；Bidder responds according to oneself current state；If bidder is idle or can complete search and rescue task, is made to auctioneer and receive the corresponding of this task；Otherwise, bidder, which will not participate in, this time auctions；If auctioneer does not receive bid during entire auction, auction process terminates；Otherwise, these tender prices will be normalized in auctioneer, and then using improved BP, to treated, tender price is trained.The present invention can effectively be such that critical the wounded is preferentially given treatment to, and method has the robustness to robot fault.Computer Simulation demonstrates the training effect of improved BP.
Description
Technical field
The invention belongs to disaster assistance technical fields, are related to a kind of based on improvement BP (Back Propagation) nerve net
The multirobot disaster of network searches and rescues method for allocating tasks.
Background technique
The multi-robot system that the present invention is formed with multiple wheeled autonomous mobile robots with search and rescue aid function
(multirobot team) is research object, the multi-robot system it is unknown it is toxic, cave in there are radioactive source, structural body etc.
Implement life search and rescue in hazardous environment.In this process, the task distribution between robot is that guarantee that critical the wounded obtains excellent
First give treatment to and efficiently complete the key of rescue.The method being widely used at present in multi-robotic task distribution specifically includes that market
Auction system and its improvement, the method for Behavior-based control, the method based on linear programming etc..The prior art proposes a kind of improved
Multiple agent task allocation algorithms based on auction, make it is mutually coordinated between multirobot, in nobody complicated dynamic environment
Task is completed in the shortest possible time.The prior art also combines the multirobot of exchange tree on the basis of contract net agreement
Coordination approach proposes a kind of multirobot dynamic task allocation algorithm based on Pareto improvement, improves task distribution
Efficiency and the time for reducing completion task.The method of Behavior-based control is to find robot-task pair with maximum utility, then
The task is distributed into corresponding robot, this method real-time and fault-tolerance are strong；But this method can only obtain local optimum
Solution.In the various practical applications based on auction system, when each robot never proposes side (distance, speed, energy consumption respectively
Deng) provide multiple tender prices to a certain task when, be difficult to find a fusion with decision-making technique to determine bid victor.
On the other hand, neural network can be used for information fusion.Based on gradient decline BP neural network theory of algorithm according to
It is versatile according to solid, but exist be easily trapped into that local minimum, study convergence process be slow and network structure be difficult to it is determining not
Foot.Therefore, the BP neural network of application enhancements of the present invention realizes the bid information fusion of robot, and the task from disaster environment
It sets out, is devised with the method for allocating tasks of preferential treatment of severe the wounded with roboting features.
Summary of the invention
It is an object of the invention to overcome defect existing in the prior art, propose a kind of based on improved BP
Multirobot disaster search and rescue method for allocating tasks.When there is new task to need to complete, auctioneer's release task message first,
It is merged, is then realized using contract net agreement auction algorithm more by bid information of the neural network to each robot
The task of robot is distributed.Task is found in auction process or is initiated the robot of auction while being used as auctioneer and auction
Person, other robots are only bidder.
Its technical solution is as follows:
Multirobot disaster based on improved BP searches and rescues method for allocating tasks, comprising the following steps:
1) each robot of multi-robot system (or multirobot team) starts pair according to the principle of depth-first
Environment scans for.
2) when some robot is when set direction finds new rescue task, which is auctioned.The auction machine
Device person is auctioneer.
3) auctioneer appoints all robots publication of the mode for the use of information broadcast for needing to complete task into team
Be engaged in information, this mission bit stream include the geographical location of the wounded, task deadline for submission of tenders, required by task machine number nc、
The priority etc. of task.
4) each robot in team is all potential bidder, they receive the search and rescue mission bit stream of auctioneer's publication
Afterwards, it is responded according to itself current state:
1. participating in submitting a tender if current robot is in idle condition；
2. if current robot is carrying out task, but performed task priority is lower than the task of auction, then
It participates in submitting a tender；
3. if current robot is carrying out task, but performed task priority is higher than the task of auction, then
It is not involved in bid.
5) conflict if there is auction, then call auction conflict-solving strategy.
6) if the bid machine number that auctioneer receives before auctioning deadline is wanted less than the required by task is completed
Machine number, then auction process terminates, which is stored in respective unfinished task list by all robots in team.
7) otherwise, these tender prices will be normalized in auctioneer, then using the improvement BP mind after training
The multiple tender prices progress provided through network to the quasi- robot for participating in the task is integrated ordered, selects ncA victor.
8) auctioneer notifies this ncA victor will execute task, send contract to auctioneer after bidder is notified
Confirmation message.Otherwise, auctioneer will be considered to bidder because failure and other reasons abandon this task, then from other machines of bid
Assessing size according to bid amounts in people successively selects other robots to participate in the task.
9) if auctioneer receives the minimum machine number that contract confirmation message number is greater than the completion required by task, contract
It establishes and starts to execute rescue task.
10) if auctioneer receives the minimum machine number that contract confirmation message number is less than the completion required by task, no
Contract can be established；The task is stored in by all robots does not complete task list, and idle machine people continues to search for predetermined direction,
The robot rescued continues to execute former rescue task.
If 11) contract is established, some robot to receive an assignment is carrying out the lower task of priority, then pause should
The task of the lower priority is simultaneously stored in unfinished task list by the task of lower priority.
12) when some robot complete undertaking rescue task after, if do not complete task list be it is empty, turn
Step 1)；Otherwise, it selects the task of highest priority to be auctioned in not completing task list, goes to step 3).
Further, the auction conflict is a robot 1 during auction task 1 in steps of 5, has been connected to machine
At this moment there is the case where two auctioneers fight for idle machine people in the information for the task 2 that people 2 issues in system, in order to avoid
There is deadlock, the resolution policy that the present invention takes is the priority that two auctioneers compare two tasks, the high task of priority
Continue to distribute, the low task of priority is stored into unfinished task list, the corresponding robot for auctioning high-priority task
Auctioneer is served as in subsequent assigning process.
In step 7, the improved BP neural network is referred to and is calculated based on Levenberg-Marquardt (abbreviation L-M)
The BP neural network of method, which not only has the local convergence of Gauss-Newton method, but also there are also the global convergences of gradient method
Property.If x(k)It is the network weight vector of kth time iteration, kth+1 time weight vector x(k+1)It can be obtained by following formula
Are as follows:
x(k+1)=x(k)+Δx(k+1) (1)
In Gauss-Newton method computation rule
Δx(k+1)=-[JT(x)J(x)]-1J(x)e(x) (2)
E (x)=[e in formula1(x) … eN(x)]TIt is output error vector, x represents x herek, do not causing the feelings obscured
Superscript is omitted under condition, similarly hereinafter；N is the dimension of neural network output, and J (x) is Jacobian matrix of the error to weight, its meter
Calculating formula is
In weight updated value formula (2), J (x)TThere are irreversible situations by J (x), therefore the algorithm is possible to not restrain,
To solve this problem, using the right value update of following forms
Δx(k+1)=-[JT(x)J(x)+λI]-1J(x)e(x) (4)
λ is a constant in formula, and I is n × n unit matrix.
The invention has the benefit that
The present invention is a kind of multi-robot Task Allocation based on contract net agreement in conjunction with improved Back Propagation, with
Traditional method for allocating tasks is compared, it has advantage following aspects: completing to appoint first is that this method considers robot
The multiple parameters being related to when business, such as distance, speed, energy consumption, task priority；Secondly this method is by neural network to each
The fusion and sequence of robot bid amounts, can capable of completing to individual machine people for task are allocated, can also be to multiple
The task that robot could complete is allocated；Again, integrated ordered to also ensure method for allocating tasks pair proposed by the present invention
The robustness of robot fault；Finally, this method is realized each robot and is being rescued by the priority of each task of setting
The dynamic for the task of having promised to undertake is adjusted in the process, ensure that the available preferential treatment of the big personnel of injured degree.Traditional BP
The neural network deficiency slow there are convergence rate, simulation result show that the training process of improved BP is significantly accelerated.This
The fusion problem of multiple bidding parameters can be effectively treated in invention, and the rescue task allocation strategy proposed can guarantee critical the wounded
It is given treatment in the shortest time.
Detailed description of the invention
Fig. 1 is task distribution and rescue work flow chart of the auctioneer robot during disaster is searched and rescued；
Fig. 2 is auction conflict-solving strategy；
Fig. 3 is task response and rescue work flow chart of the bidder robot during disaster is searched and rescued；
Fig. 4 is traditional BP neural network training process；
Fig. 5 is based on Levenberg-Marquardt Algorithm BP neural network training process.
Specific embodiment
Technical solution of the present invention is described in more detail with reference to the accompanying drawings and detailed description.
The bid of 1 rescue task is merged with bid information
1.1BP neural network
BP neural network, that is, back-propagating network is a kind of feed-forward type network using error back propagation training algorithm, is
One of current most widely used neural network model.The bid information B of multiple robots collectively forms BP nerve net in team
The input of network, B=[B1 B2 … BM], Bi(i=1 ..., M) is the tender price of robot i, takes Bi=[di si ei], wherein
diFor the shortest distance of robot i to target point, siFor the most fast movement speed of robot i, eiFor current remaining of robot i
Electricity, M are the machine numbers for participating in submitting a tender.The present invention use three layers of BP neural network, 15 neurons of input layer, hidden layer 10
Neuron, 5 neurons of output layer；Namely for a task, only receive the bid for responding 5 most fast robots.
In order to reduce the complexity of calculating, hidden layer uses unipolarity Log-Sigmoid function, and output layer uses linear function.
1.2 normalization
Influence in order to avoid each input data amplitude difference of BP network and different physical significances to output, to input
Data be normalized.When needing input data transforming to [0,1], following formula is used:
When needing to transform to input data [- 1,1], following formula is used:
Wherein, xi(i=1 ..., 15) represents the bid amounts of different robots, xminAnd xmaxDistribution represents all bid amounts
In minimum value and maximum value, x 'tIt is the data after normalization, that is, the input value of neural network.
1.3Levenberg-Marquardt BP algorithm
Improved BP neural network refers to the BP nerve net based on Levenberg-Marquardt (abbreviation L-M) algorithm
Network, which not only has the local convergence of Gauss-Newton method, but also there are also the global convergences of gradient method.If x(k)It is kth
The network weight vector of secondary iteration, kth+1 time weight vector x(k+1)It can be obtained by following formula are as follows:
x(k+1)=x(k)+Δx(k+1) (7)
In Gauss-Newton method computation rule
Δx(k+1)=- [JT(x)J(x)]-1J(x)e(x) (8)
E (x)=[e in formula1(x) … eN(x)]TIt is output error vector, x represents x herek, do not causing the feelings obscured
Superscript is omitted under condition, similarly hereinafter；N is the dimension of neural network output, and J (x) is Jacobian matrix of the error to weight, its meter
Calculating formula is
In weight updated value formula (2), J (x)TThere are irreversible situations by J (x), therefore the algorithm is possible to not restrain,
To solve this problem, using the right value update of following forms
Δx(k+1)=-[JT(x)J(x)+λI]-1J(x)e(x) (10)
λ is a constant in formula, and I is n × n unit matrix.
Neural network needs the data of reason manual construction to be trained before merging for bid information.
The method for allocating tasks of 2 searching rescues
Contract net agreement is one kind of auction algorithm.Three kinds of auction systems primarily now have First-Price sealing
Auction, Vickrey auction, Dutch auction.Traditional contracts network protocol uses First-Price sealed auction system, works as auction
Person submits to the bid of auctioneer one sealing, and bidder is unknown to the tender price of other bidders at this time.The present invention
The task point distribution of multirobot is realized using contract net agreement auction algorithm.The robot of task is found in auction process
Or the artificial auctioneer of machine of auction is initiated, the artificial bidder of machine of task may be completed.But there can only be a machine
People serves as single auctioneer, while it is also possible to bidder.Task allocation process diagram is as shown in Figure 1.
A kind of multirobot disaster assistance method for allocating tasks based on improved BP is proposed, passes through mind first
It is merged through bid information of the network to each robot, multirobot is then realized using contract net agreement auction algorithm
Task point distribution.
Its technical solution is as follows:
A kind of multirobot disaster assistance method for allocating tasks based on improved BP, comprising the following steps:
1) each robot of multi-robot system (or multirobot team) starts pair according to the principle of depth-first
Environment scans for.
2) when some robot is when set direction finds new rescue task, which is auctioned.The auction machine
Device person is auctioneer.
3) auctioneer appoints all robots publication of the mode for the use of information broadcast for needing to complete task into team
Be engaged in information, this mission bit stream include the geographical location of the wounded, task deadline for submission of tenders, required by task machine number nc、
The priority etc. of task.
4) each robot in team is all potential bidder, they receive the search and rescue mission bit stream of auctioneer's publication
Afterwards, it is responded according to itself current state:
1. participating in submitting a tender if current robot is in idle condition；
2. if current robot is carrying out task, but performed task priority is lower than the task of auction, then
It participates in submitting a tender；
3. if current robot is carrying out task, but performed task priority is higher than the task of auction, then
It is not involved in bid.
5) conflict if there is auction, then call auction conflict-solving strategy.
6) if the bid machine number that auctioneer receives before auctioning deadline is wanted less than the required by task is completed
Machine number, then auction process terminates, which is stored in respective unfinished task list by all robots in team.
7) otherwise, these tender prices will be normalized in auctioneer, then using the improvement BP mind after training
The multiple tender prices progress provided through network to the quasi- robot for participating in the task is integrated ordered, selects ncA victor.
8) auctioneer notifies this ncA victor will execute task, send contract to auctioneer after bidder is notified
Confirmation message.Otherwise, auctioneer will be considered to bidder because failure and other reasons abandon this task, then from other machines of bid
Assessing size according to bid amounts in people successively selects other robots to participate in the task.
9) if auctioneer receives the minimum machine number that contract confirmation message number is greater than the completion required by task, contract
It establishes and starts to execute rescue task.
10) if auctioneer receives the minimum machine number that contract confirmation message number is less than the completion required by task, no
Contract can be established；The task is stored in by all robots does not complete task list, and idle machine people continues to search for predetermined direction,
The robot rescued continues to execute former rescue task.
If 11) contract is established, some robot to receive an assignment is carrying out the lower task of priority, then pause should
The task of the lower priority is simultaneously stored in unfinished task list by the task of lower priority.
12) when some robot complete undertaking rescue task after, if do not complete task list be it is empty, turn
Step 1)；Otherwise, it selects the task of highest priority to be auctioned in not completing task list, goes to step 3).
The course of work and conflict-solving strategy of auctioneer is as depicted in figs. 1 and 2, and the response of bidder and the course of work are such as
Shown in Fig. 3.
3 method comparative analyses
Table 1 searches and rescues method for allocating tasks comparison
The multirobot disaster based on improved BP that the invention proposes a kind of searches and rescues method for allocating tasks, and existing
Some disaster assistance method for allocating tasks ratios, the Task Assigned Policy that the present invention designs have the following aspects significantly excellent
Gesture:
(1) the most important target of operation is exactly to save the life security of the wounded in disaster search and rescue, because each the wounded receives
The degree of injury is different, it is necessary to give treatment to critical the wounded preferentially, the present invention is guaranteed by setting task priority method
This target priority is realized；
(2) disaster field situation is multifarious, and one robot of some rescue tasks can be completed, and some rescue tasks need
Want multiple robot work compounds that can complete, the prior art has ignored this feature, but the present invention can be effectively treated
Such case；
(3) rescue site harsh environmental conditions, search and rescue robot failure are may to occur at any time, and of the invention appoints
Business allocation strategy can be most suitable robot and substitute to failed machines people, realize designed method to robot event
The robustness of barrier；
(4) conventional method is submitted a tender from the difference of benefit and cost, but the conversion factor between benefit and cost
It is very dark to determine, it is difficult to ensure that the reasonability submitted a tender, robot also need to consider machine in search operation in order to smoothly complete task
The many factors such as the remaining energy of device people, therefore the present invention establishes multi-parameter Bidding system.
(5) share tasks processing the characteristics of being all rescue tasks, conventional method will be apart from as costs, in order to as early as possible
Treatment of severe the wounded, the present invention take into account the maximum speed of robot when robot submits a tender to task；
(6) process of the robot discovery task in team has asynchronism and concurrency, and default treatment of the present invention is just
It is asynchronism, when multiple robots find different tasks simultaneously, the present invention takes the limited processing of method of priority comparison
Critical the wounded gives treatment to the distribution of task.
(7) auction conflict has solution in existing technology, but only simple time delay method, the present invention from
Personage's priority set out solution conflict, for life rescue meaning clearly.
The simulation comparison of traditional BP neural network and the BP neural network based on L-M algorithm is carried out below.Emulator: pen
Remember this computer, CPU:i5-4210U, memory: 4G, system Windows8.1 professional version, simulation software are Matlab 2015b.It is defeated
Enter the bid amounts that data are normalized each robots, wherein the robot that each output neuron corresponds to identical label completes
The performance evaluation of the task, the value show more greatly corresponding more advantageous completion task of robot.In Matlab 2015b
Using newff function creation BP neural network, network training then is carried out to the data after normalization using train function.It passes
System BP neural network algorithm is trained input data using traingd function, and the BP neural network based on L-M algorithm uses
Self-editing function is trained, shown in training result Fig. 4, Fig. 5, based on L-M Algorithm BP neural network in training accuracy and instruction
It will be better than conventional method in terms of practicing number.
When some robot current search direction find new task after, which is auctioned；Or complete one
After task, the task for the highest priority being stored in unfinished task list is auctioned；By auctioneer when auction starts
It is released news in the way of broadcast to potential bidder；Bidder is according to oneself current state and undertaking for task
Priority respond；If bidder is that the priority of rescue task that is idle or undertaking is lower than times auctioned
Business is then made to auctioneer and receives the response of this task and submitted a tender according to current state；Otherwise, bidder will not participate in
This time auction.If auctioneer receives bid number before auctioning deadline is less than the minimum robot for completing the required by task
Number, then auction process terminates；Otherwise, these tender prices will be normalized in auctioneer, after then using training
Improved BP is integrated ordered to bid information progress after processing, and the machine for executing the task is selected according to sequence size
People.The present invention can effectively be such that critical the wounded is preferentially given treatment to, and method has the robustness to robot fault.Computer
The simulating, verifying training effect of improved BP.
The foregoing is only a preferred embodiment of the present invention, the scope of protection of the present invention is not limited to this, it is any ripe
Know those skilled in the art within the technical scope of the present disclosure, the letter for the technical solution that can be become apparent to
Altered or equivalence replacement are fallen within the protection scope of the present invention.
Claims (3)
1. the multirobot disaster based on improved BP searches and rescues method for allocating tasks, which is characterized in that including following step
It is rapid:
1) each robot of multi-robot system starts to scan for environment according to the principle of depth-first；
2) when some robot is when set direction finds new rescue task, which is auctioned；The auction robot
Referred to as auctioneer；
3) auctioneer believes all robot release tasks of the mode for the use of information broadcast for needing to complete task into team
Breath；
4) each robot in team is all potential bidder, after they receive the search and rescue mission bit stream that auctioneer issues,
According to itself, current state is responded:
1. participating in submitting a tender if current robot is in idle condition；
2. if current robot is carrying out task, but performed task priority is then participated in lower than the task of auction
It submits a tender；
3. if current robot is carrying out task, but performed task priority is higher than the task of auction, then does not join
With bid；
5) conflict if there is auction, then call auction conflict-solving strategy；
6) if the bid machine number that auctioneer receives before auctioning deadline is less than the machine that the required by task is wanted of completing
Device number, then auction process terminates, which is stored in respective unfinished task list by all robots in team；
7) otherwise, these tender prices will be normalized in auctioneer, then using the improvement BP nerve net after training
Multiple tender prices progress that network provides the quasi- robot for participating in the task is integrated ordered, selects ncA victor；
8) auctioneer notifies this ncA victor will execute task, send contract confirmation to auctioneer after bidder is notified
Information；Otherwise, auctioneer will be considered to bidder and abandon this task because of the reason of failure, then from other robots of bid
Assessing size according to bid amounts successively selects other robots to participate in the task；
If 9) auctioneer receives the minimum machine number that contract confirmation message number is greater than the completion required by task, contract is established
And start to execute rescue task；
10) if auctioneer receives the minimum machine number that contract confirmation message number is less than the completion required by task, cannot build
It concludes a contract；The task is stored in by all robots does not complete task list, and idle machine people continues to search for predetermined direction,
The robot of rescue continues to execute former rescue task；
If 11) contract is established, some robot to receive an assignment is carrying out the lower task of priority, then it is lower to suspend this
The task of the lower priority is simultaneously stored in unfinished task list by the task of priority；
12) when some robot complete undertaking rescue task after, if do not complete task list be it is empty, go to step
1)；Otherwise, it selects the task of highest priority to be auctioned in not completing task list, goes to step 3).
2. the multirobot disaster according to claim 1 based on improved BP searches and rescues method for allocating tasks,
It is characterized in that, in step 4, the improved BP refers to the BP nerve net based on Levenberg-Marquardt algorithm
Network, if x(k)It is the network weight vector of kth time iteration, kth+1 time weight vector x(k+1)It is obtained by following formula are as follows:
x(k+1)=x(k)+Δx(k+1) (1)
In Gauss-Newton method computation rule
Δx(k+1)=-[JT(x)J(x)]-1J(x)e(x) (2)
E (x)=[e in formula1(x) … eN(x)]TIt is output error vector, x represents x herek, in the case where not causing to obscure
Superscript is omitted, similarly hereinafter；N is the dimension of neural network output, and J (x) is Jacobian matrix of the error to weight, its calculating is public
Formula is
In weight updated value formula (2), J (x)TThere are irreversible situations by J (x), using the right value update of following forms:
Δx(k+1)=-[JT(x)J(x)+λI]-1J(x)e(x) (4)
λ is a constant in formula, and I is n × n unit matrix.
3. the multirobot disaster according to claim 1 based on improved BP searches and rescues method for allocating tasks,
It is characterized in that, in step 3), the task that auctioneer announces includes the geographical location of the wounded, task deadline for submission of tenders, task institute
The machine number n neededc, task priority information.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810673369.7A CN108985580B (en) | 2018-06-16 | 2018-06-16 | Multi-robot disaster search and rescue task allocation method based on improved BP neural network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201810673369.7A CN108985580B (en) | 2018-06-16 | 2018-06-16 | Multi-robot disaster search and rescue task allocation method based on improved BP neural network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN108985580A true CN108985580A (en) | 2018-12-11 |
| CN108985580B CN108985580B (en) | 2022-09-02 |
Family
ID=64538937
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201810673369.7A Active CN108985580B (en) | 2018-06-16 | 2018-06-16 | Multi-robot disaster search and rescue task allocation method based on improved BP neural network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN108985580B (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919431A (en) * | 2019-01-28 | 2019-06-21 | 重庆邮电大学 | Heterogeneous multi-robot method for allocating tasks based on auction algorithm |
| CN111461488A (en) * | 2020-03-03 | 2020-07-28 | 北京理工大学 | Multi-robot distributed cooperative task allocation method facing workshop carrying problem |
| CN112070328A (en) * | 2019-06-11 | 2020-12-11 | 哈尔滨工业大学（威海） | Multi-water-surface unmanned search and rescue boat task allocation method with known environmental information part |
| CN112862270A (en) * | 2021-01-20 | 2021-05-28 | 西北工业大学 | Individual task selection method, device and system for distributed multiple robots |
| CN115609608A (en) * | 2022-12-02 | 2023-01-17 | 北京国安广传网络科技有限公司 | All-weather health management robot |
Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102915465A (en) * | 2012-10-24 | 2013-02-06 | 河海大学常州校区 | Multi-robot combined team-organizing method based on mobile biostimulation nerve network |
| EP2645196A1 (en) * | 2012-03-30 | 2013-10-02 | The Boeing Company | Network of unmanned vehicles |
| 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) |
| CN105069530A (en) * | 2015-08-13 | 2015-11-18 | 肇庆学院 | Multi-robot task allocation method based on multi-objective optimization |
| US20150355639A1 (en) * | 2006-07-05 | 2015-12-10 | Battelle Energy Alliance, Llc | Real time explosive hazard information sensing, processing, and communication for autonomous operation |
| CN105203097A (en) * | 2015-10-14 | 2015-12-30 | 中国矿业大学 | Multi-robot multi-target point rescue route planning method fit for after-calamity environment |
| CN105843227A (en) * | 2016-04-15 | 2016-08-10 | 上海大学 | Task intensity dynamic adjustment based multiple robots cooperating task hunting allocation algorithm |
| CN106875090A (en) * | 2017-01-09 | 2017-06-20 | 中南大学 | A kind of multirobot distributed task scheduling towards dynamic task distributes forming method |
| CN107589758A (en) * | 2017-08-30 | 2018-01-16 | 武汉大学 | A kind of intelligent field unmanned plane rescue method and system based on double source video analysis |
-
2018
- 2018-06-16 CN CN201810673369.7A patent/CN108985580B/en active Active
Patent Citations (9)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US20150355639A1 (en) * | 2006-07-05 | 2015-12-10 | Battelle Energy Alliance, Llc | Real time explosive hazard information sensing, processing, and communication for autonomous operation |
| EP2645196A1 (en) * | 2012-03-30 | 2013-10-02 | The Boeing Company | Network of unmanned vehicles |
| CN102915465A (en) * | 2012-10-24 | 2013-02-06 | 河海大学常州校区 | Multi-robot combined team-organizing method based on mobile biostimulation nerve network |
| 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) |
| CN105069530A (en) * | 2015-08-13 | 2015-11-18 | 肇庆学院 | Multi-robot task allocation method based on multi-objective optimization |
| CN105203097A (en) * | 2015-10-14 | 2015-12-30 | 中国矿业大学 | Multi-robot multi-target point rescue route planning method fit for after-calamity environment |
| CN105843227A (en) * | 2016-04-15 | 2016-08-10 | 上海大学 | Task intensity dynamic adjustment based multiple robots cooperating task hunting allocation algorithm |
| CN106875090A (en) * | 2017-01-09 | 2017-06-20 | 中南大学 | A kind of multirobot distributed task scheduling towards dynamic task distributes forming method |
| CN107589758A (en) * | 2017-08-30 | 2018-01-16 | 武汉大学 | A kind of intelligent field unmanned plane rescue method and system based on double source video analysis |
Non-Patent Citations (4)
| Title |
|---|
| XIN YI,ETC: "A Bio-Inspired Approach to Task Assignment of Swarm Robots in 3-D Dynamic Environments", 《IEEE TRANSACTIONS ON CYBERNETICS ( VOLUME: 47, ISSUE: 4, APRIL 2017)》 * |
| 姜来浩等: "基于改进捆绑拍卖多机器人任务分配研究", 《齐齐哈尔大学学报(自然科学版)》 * |
| 张帆: "基于MAS的多机器人搜救系统及任务分配方法研究", 《中国优秀博硕士学位论文全文数据库(硕士) 信息科技辑》 * |
| 梁志伟等: "RoboCup机器人救援仿真中基于拍卖的任务分配算法", 《机器人》 * |
Cited By (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109919431A (en) * | 2019-01-28 | 2019-06-21 | 重庆邮电大学 | Heterogeneous multi-robot method for allocating tasks based on auction algorithm |
| CN112070328A (en) * | 2019-06-11 | 2020-12-11 | 哈尔滨工业大学（威海） | Multi-water-surface unmanned search and rescue boat task allocation method with known environmental information part |
| CN112070328B (en) * | 2019-06-11 | 2023-06-27 | 哈尔滨工业大学（威海） | Multi-water surface unmanned search and rescue boat task allocation method with partially known environmental information |
| CN111461488A (en) * | 2020-03-03 | 2020-07-28 | 北京理工大学 | Multi-robot distributed cooperative task allocation method facing workshop carrying problem |
| CN111461488B (en) * | 2020-03-03 | 2022-03-11 | 北京理工大学 | Multi-robot distributed cooperative task allocation method facing workshop carrying problem |
| CN112862270A (en) * | 2021-01-20 | 2021-05-28 | 西北工业大学 | Individual task selection method, device and system for distributed multiple robots |
| CN112862270B (en) * | 2021-01-20 | 2023-08-11 | 西北工业大学 | Individual task selection method, device and system for distributed multi-robot |
| CN115609608A (en) * | 2022-12-02 | 2023-01-17 | 北京国安广传网络科技有限公司 | All-weather health management robot |
Also Published As
| Publication number | Publication date |
|---|---|
| CN108985580B (en) | 2022-09-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN108985580A (en) | Multirobot disaster based on improved BP searches and rescues method for allocating tasks | |
| Zhao et al. | A heuristic distributed task allocation method for multivehicle multitask problems and its application to search and rescue scenario | |
| Sousa et al. | A distributed architecture and negotiation protocol for scheduling in manufacturing systems | |
| Vig et al. | Coalition formation: From software agents to robots | |
| Shi et al. | An auction-based rescue task allocation approach for heterogeneous multi-robot system | |
| Goldberg et al. | Cloud robotics and automation: A survey of related work | |
| CN104899760B (en) | A kind of motivational techniques in time correlation movement gunz sensory perceptual system | |
| Trigui et al. | A distributed market-based algorithm for the multi-robot assignment problem | |
| CN111784211B (en) | Cluster-based group multitask allocation method and storage medium | |
| Lim et al. | Incentive mechanism design for resource sharing in collaborative edge learning | |
| CN107103408A (en) | Complex task distribution method under a kind of mass-rent environment | |
| Tang et al. | Auction-based cooperation mechanism to parts scheduling for flexible job shop with inter-cells | |
| Zhu et al. | Agent-based dynamic scheduling for earth-observing tasks on multiple airships in emergency | |
| Das et al. | Game theoretic equilibrium analysis of energy auction in microgrid | |
| Yuan et al. | Multi-robot task allocation using CNP combines with neural network | |
| Wang et al. | A task allocation algorithm based on market mechanism for multiple robot systems | |
| Zhang et al. | An innovation approach for optimal resource allocation in emergency management | |
| Su et al. | Dynamic task allocation for heterogeneous agents in disaster environments under time, space and communication constraints | |
| Eijyne et al. | Development of a task-oriented, auction-based task allocation framework for a heterogeneous multirobot system | |
| Lee et al. | A preference-based task allocation framework for multi-robot coordination | |
| Choi et al. | Performance optimisation of mobile robots for search-and-rescue | |
| Dias et al. | Market-based multirobot coordination for complex space applications | |
| DeGroote et al. | Efficient and Adaptable Task Assignment for UAS Considering the MDMTSP | |
| Suárez et al. | Improving rescue operations in disasters: approaches about task allocation and re-scheduling | |
| Kang et al. | A Multi-Agent Management System for Grid Resource Allocation |
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 |