CN108985580B - Multi-robot disaster search and rescue task allocation method based on improved BP neural network - Google Patents

Multi-robot disaster search and rescue task allocation method based on improved BP neural network Download PDF

Info

Publication number
CN108985580B
CN108985580B CN201810673369.7A CN201810673369A CN108985580B CN 108985580 B CN108985580 B CN 108985580B CN 201810673369 A CN201810673369 A CN 201810673369A CN 108985580 B CN108985580 B CN 108985580B
Authority
CN
China
Prior art keywords
task
robot
auctioneer
robots
auction
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Active
Application number
CN201810673369.7A
Other languages
Chinese (zh)
Other versions
CN108985580A (en
Inventor
戴学丰
严浙平
郝冰
张辉
张宏民
赵岩
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Qiqihar University
Original Assignee
Qiqihar University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Qiqihar University filed Critical Qiqihar University
Priority to CN201810673369.7A priority Critical patent/CN108985580B/en
Publication of CN108985580A publication Critical patent/CN108985580A/en
Application granted granted Critical
Publication of CN108985580B publication Critical patent/CN108985580B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

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

Landscapes

  • Business, Economics & Management (AREA)
  • Human Resources & Organizations (AREA)
  • Engineering & Computer Science (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • Economics (AREA)
  • Entrepreneurship & Innovation (AREA)
  • Strategic Management (AREA)
  • General Physics & Mathematics (AREA)
  • Game Theory and Decision Science (AREA)
  • Educational Administration (AREA)
  • Quality & Reliability (AREA)
  • Tourism & Hospitality (AREA)
  • Marketing (AREA)
  • Operations Research (AREA)
  • General Business, Economics & Management (AREA)
  • Development Economics (AREA)
  • Biomedical Technology (AREA)
  • Evolutionary Computation (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Data Mining & Analysis (AREA)
  • Health & Medical Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Management, Administration, Business Operations System, And Electronic Commerce (AREA)

Abstract

The invention discloses a multi-robot disaster search and rescue task allocation method based on an improved BP neural network. When the auctioneer does not have enough capacity to complete the current search and rescue task, the information of the task is issued to the auctioneer in a broadcasting mode; the auction participants respond according to the current state of the auction participants; if the auctioneer is idle or can complete the search and rescue task, the auctioneer is correspondingly provided with a task for receiving the search and rescue task; otherwise, the bidder will not participate in the auction; if the auctioneer does not receive a bid throughout the auction process, the auction process ends; otherwise, the auctioneer normalizes the bid prices and then trains the processed bid prices using the modified BP neural network. The invention can effectively help the critically ill patients with priority, and the method has robustness to the robot fault. Computer simulation verifies the training effect of the improved BP neural network.

Description

Multi-robot disaster search and rescue task allocation method based on improved BP neural network
Technical Field
The invention belongs to the technical field of disaster rescue, and relates to a multi-robot disaster search and rescue task allocation method based on an improved BP (Back propagation) neural network.
Background
The invention takes a multi-robot system (multi-robot team) consisting of a plurality of autonomous wheel type mobile robots with searching and rescuing functions as a research object, and the multi-robot system implements life search and rescue in unknown dangerous environments with toxicity, radioactive sources, structural body collapse and the like. In the process, task distribution among the robots is the key to ensure that critically ill patients are treated preferentially and the critical patients are treated efficiently. The method widely applied to the task allocation of multiple robots at present mainly comprises the following steps: market auction methods and improvements thereof, behavior-based methods, linear programming-based methods, and the like. The prior art provides an improved auction-based multi-agent task allocation algorithm, so that multiple robots are coordinated with each other, and tasks are completed in the shortest time in a complex unmanned dynamic environment. In addition, in the prior art, a multi-robot dynamic task allocation algorithm based on pareto improvement is provided by combining a multi-robot coordination method of a switching tree on the basis of a contract network protocol, so that the task allocation efficiency is improved, and the time for completing tasks is shortened. The behavior-based method is to find the robot-task pair with the maximum utility and then distribute the task to the corresponding robot, and the method has strong real-time performance and fault tolerance; but this method can only yield a locally optimal solution. In various auction-based practical applications, when each robot gives a plurality of bid prices for a certain task from the negative side (distance, speed, energy consumption, etc.), respectively, it is difficult to find a fusion and decision method to determine the bid winner.
On the other hand, neural networks may be used for information fusion. The BP neural network algorithm based on gradient descent is firm in theoretical basis and strong in universality, but has the defects of easiness in falling into local minimum, slow learning convergence process and difficulty in determining a network structure. Therefore, the invention applies the improved BP neural network to realize the fusion of the bidding information of the robot, and designs a task allocation method for preferentially treating the critically ill patients from the tasks and the characteristics of the robot in the disaster environment.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provides a multi-robot disaster search and rescue task allocation method based on an improved BP neural network. When a new task needs to be completed, the auctioneer firstly releases task information, fuses the bidding information of each robot through a neural network, and then realizes the task allocation of multiple robots by adopting a contract network agreement auction algorithm. During the auction process, the robot which finds the task or initiates the auction can be used as the auctioneer and the auction participant at the same time, and other robots can only be used as the auction participants.
The technical scheme is as follows:
the multi-robot disaster search and rescue task allocation method based on the improved BP neural network comprises the following steps:
1) each robot of the multi-robot system (or called multi-robot team) starts to search the environment according to the depth-first principle.
2) When a certain robot finds a new rescue task in a given direction, an auction is conducted on the task. The auction robot is called an auctioneer.
3) The auctioneer distributes the information needed to complete the task to all the robots in the team in a broadcasting mode, and the task information comprises the geographical position of the wounded, the task bidding deadline time and the number n of the robots needed by the task c The priority of the task, etc.
4) Each robot in the team is a potential auction player, and after receiving search and rescue task information issued by the auction player, the robots respond according to the current state of the robots:
if the current robot is in an idle state, participating in bidding;
if the current robot is executing the task, but the priority of the executed task is lower than that of the auction task, the robot participates in bidding;
and if the current robot executes the task but the executed task has higher priority than the auction task, the robot does not participate in the bidding.
5) If an auction conflict occurs, an auction conflict resolution policy is invoked.
6) If the auctioneer receives less bidding robots than are needed to complete the task before the auction deadline, the auction process ends and the task is deposited into its respective list of incomplete tasks by all robots in the team.
7) Otherwise, the auctioneer normalizes the bidding prices, then uses the trained improved BP neural network to comprehensively sort the bidding prices given by the robot to participate in the task, and selects n c A winner.
8) Auctioneer informing of this n c The winner will execute the task, and the auction receives the notice and sends contract confirmation information to the auctioneer. Otherwise, the auctioneer considers that the auctioneer abandons the task due to failure and the like, and then sequentially selects other robots to participate in the task from other bidding robots according to the evaluation size of the bid valueAnd (5) performing tasks.
9) And if the number of the auctioneers receiving the contract confirmation information is larger than the minimum number of robots required for completing the task, establishing the contract and starting to execute the rescue task.
10) If the number of the messages confirmed by the auctioneer after receiving the contract is less than the minimum number of robots required for completing the task, the auctioneer cannot establish the contract; all robots store the tasks into an uncompleted task list, the idle robots continue to search in the preset direction, and the rescue robots continue to execute the original rescue tasks.
11) If the contract is established and a certain robot receiving the task is executing the task with lower priority, suspending the task with lower priority and storing the task with lower priority into an uncompleted task list.
12) After a certain robot finishes the undertaken rescue task, if the uncompleted task list is empty, turning to step 1); otherwise, selecting the task with the highest priority from the incomplete task list for auction, and turning to the step 3).
Further, in the auction conflict in step 5, the robot 1 receives the information of the task 2 issued by the robot 2 in the process of the auction task 1, and at this time, two auctioneers compete for the idle robot in the system, so as to avoid deadlock, the invention adopts a solution strategy that the two auctioneers compare the priorities of the two tasks, the task with the higher priority is continuously allocated, the task with the lower priority is stored in the incomplete task list, and the robot corresponding to the auction task with the higher priority acts as an auctioneer in the subsequent allocation process.
In step 7, the improved BP neural network refers to a BP neural network based on a Levenberg-Marquardt (L-M for short) algorithm, and the algorithm not only has local convergence of a Gaussian-Newton method, but also has global convergence of a gradient method. Let x (k) Is the network weight vector of the kth iteration, the weight vector x of the (k + 1) th iteration (k+1) Can be obtained by the following formula:
x (k+1) ＝x (k) +Δx (k+1) (1)
in the calculation rule of Gauss-Newton method
Δx (k+1) ＝-[J T (x)J(x)] -1 J(x)e(x) (2)
Wherein e (x) is [ e ] 1 (x) … e N (x)] T Is the output error vector, where x represents x k The upper corner marks are omitted under the condition of not causing confusion, and the same is applied below; n is the dimension of the neural network output, J (x) is the Jacobian matrix of error versus weight, which is calculated as
Figure BSA0000166050280000041
In the weight update value formula (2), J (x) T J (x) has irreversible condition, so the algorithm may not converge, and to solve the problem, the following form of weight updating is adopted
Δx (k+1) ＝-[J T (x)J(x)+λI] -1 J(x)e(x) (4)
In the formula, λ is a constant, and I is an n × n unit matrix.
The invention has the beneficial effects that:
the invention relates to a multi-robot task allocation method based on the combination of a contract network protocol and an improved BP algorithm, which has the following advantages compared with the traditional task allocation method: firstly, the method considers a plurality of parameters related to the completion of tasks by the robot, such as distance, speed, energy consumption, task priority and the like; secondly, the method fuses and sequences the scaling values of all robots through a neural network, so that tasks which can be finished by a single robot can be distributed, and tasks which can be finished by a plurality of robots can also be distributed; thirdly, comprehensive sequencing also ensures the robustness of the task allocation method provided by the invention to the robot fault; finally, the method realizes the dynamic adjustment of each robot on the committed tasks in the rescue process by setting the priority of each task, and ensures that the personnel with large injury degree can be treated in priority. The traditional BP neural network has the defect of slow convergence speed, and simulation results show that the training process of the improved BP neural network is obviously accelerated. The invention can effectively solve the fusion problem of a plurality of bidding parameters, and the proposed rescue task allocation strategy can ensure that the critically ill can be rescued in the shortest time.
Drawings
FIG. 1 is a flow chart of task allocation and rescue operation of an auctioneer robot during disaster search and rescue;
FIG. 2 is an auction conflict resolution strategy;
FIG. 3 is a flow chart of task response and rescue operations of a bidder robot during a disaster search and rescue process;
FIG. 4 is a conventional BP neural network training process;
FIG. 5 is a BP neural network training process based on the Levenberg-Marquardt algorithm.
Detailed Description
The technical solutions of the present invention will be described in further detail with reference to the accompanying drawings and the detailed description.
1 rescue task bidding and bidding information fusion
1.1BP neural network
The BP neural network, namely the back propagation network, is a feedforward network using an error back propagation training algorithm, and is one of the most widely applied neural network models at present. The bidding information B of a plurality of robots in a team jointly forms the input of a BP neural network, and B ═ B 1 B 2 … B M ]，B i (i 1, …, M) is the bid price of robot i, and B is taken i ＝[d i s i e i ]Wherein d is i Is the shortest distance, s, from the robot i to the target point i For the robot i fastest moving speed, e i The current remaining power of the robot i, M is the number of robots participating in the bid. The invention adopts three layers of BP neural networks, 15 neurons of an input layer, 10 neurons of a hidden layer and 5 neurons of an output layer; that is, only the 5 robots that responded the fastest bid for a task are accepted. In order to reduce the complexity of calculation, a unipolar Log-Sigmoid function is adopted by the hidden layer, and a linear function is adopted by the output layer.
1.2 normalization
In order to avoid the influence of different amplitudes and different physical meanings of input data of the BP network on output, the input data is normalized. When it is desired to transform the input data to [0, 1], the following formula is used:
Figure BSA0000166050280000061
when it is desired to transform the input data to [ -1, 1], the following formula is used:
Figure BSA0000166050280000062
wherein x is i (i-1, …, 15) represents the bid values for different robots, x min And x max The distribution represents the minimum and maximum values, x 'of all the bid values' t Is the normalized data, i.e., the input values to the neural network.
1.3Levenberg-Marquardt BP Algorithm
The improved BP neural network is based on a Levenberg-Marquardt (L-M for short) algorithm, and the algorithm not only has local convergence of a Gaussian-Newton method, but also has global convergence of a gradient method. Let x (k) Is the network weight vector of the kth iteration, the weight vector x of the (k + 1) th iteration (k+1) Can be obtained by the following formula:
x (k+1) ＝x (k) +Δx (k+1) (7)
in the calculation rule of Gauss-Newton method
Δx (k+1 )＝-[J T (x)J(x)] -1 J(x)e(x) (8)
Wherein e (x) is [ e ] 1 (x) … e N (x)] T Is the output error vector, where x represents x k The upper corner marks are omitted under the condition of not causing confusion, and the same is applied below; n is the dimension of the neural network output, J (x) is the Jacobian matrix of error versus weight, which is calculated as
Figure BSA0000166050280000071
In the weight update value formula (2), J (x) T J (x) has irreversible condition, so the algorithm may not converge, and to solve the problem, the following form of weight updating is adopted
Δx (k+1) ＝-[J T (x)J(x)+λI] -1 J(x)e(x) (10)
In the formula, λ is a constant, and I is an n × n unit matrix.
Neural networks need to be trained by artificially constructed data before being used for bid information fusion.
2 search and rescue task allocation method
The contract network agreement is one of the auction algorithms. The main three auction systems today are the First-Price sealed auction, the Vickrey auction, the Dutch auction. The traditional contract network agreement adopts a First-Price sealed auction system, and when an auction participant submits a sealed bid to the auction participant, the bid Price of the auction participant to other auction participants is unknown. The invention adopts a contract network protocol auction algorithm to realize the task point distribution of multiple robots. Robots that find a task during the auction process or those that initiate the auction are auctioneers, and robots that may complete a task are auctioneers. But only one robot can act as a single auctioneer and it can also be an auction master. The task allocation flow chart is shown in fig. 1.
The method comprises the steps of firstly fusing bidding information of each robot through the neural network, and then realizing task point distribution of the multiple robots by adopting a contract network agreement auction algorithm.
The technical scheme is as follows:
a multi-robot disaster rescue task allocation method based on an improved BP neural network comprises the following steps:
1) each robot of the multi-robot system (or called multi-robot team) starts to search the environment according to the depth-first principle.
2) When a robot finds a new rescue task in a given direction, the task is auctioned. The auction robot is called an auctioneer.
3) The auctioneer distributes the information needed to complete the task to all the robots in the team in a broadcasting mode, and the task information comprises the geographical position of the wounded, the task bidding deadline time and the number n of the robots needed by the task c The priority of the task, etc.
4) Each robot in the team is a potential auctioneer, and after receiving search and rescue task information issued by the auctioneer, the robots respond according to the current state of the robots:
if the current robot is in an idle state, participating in bidding;
if the current robot is executing the task, but the priority of the executed task is lower than that of the auction task, the robot participates in bidding;
and if the current robot executes the task but the executed task has higher priority than the auction task, the robot does not participate in the bidding.
5) If an auction conflict occurs, an auction conflict resolution policy is invoked.
6) If the auctioneer receives less bidding robots than are needed to complete the task before the auction deadline, the auction process ends and the task is deposited into its respective list of incomplete tasks by all robots in the team.
7) Otherwise, the auctioneer normalizes the bidding prices, then uses the trained improved BP neural network to comprehensively sort the bidding prices given by the robot to participate in the task, and selects n c And (6) a winner.
8) Auctioneer informing of this n c The winner will execute the task, and the auction receives the notice and sends contract confirmation information to the auctioneer. Otherwise, the auctioneer considers that the auctioneer abandons the task due to a fault or the like, and then sequentially selects other robots to participate in the task from the other robots bidding according to the evaluation size of the bid value.
9) And if the number of the auctioneers receiving the contract confirmation information is larger than the minimum number of robots required for completing the task, establishing the contract and starting to execute the rescue task.
10) If the number of the messages confirmed by the auctioneer after receiving the contract is less than the minimum number of robots required for completing the task, the auctioneer cannot establish the contract; all the robots store the tasks into an incomplete task list, the idle robots continue to search in the preset direction, and the rescue robots continue to execute the original rescue tasks.
11) If the contract is established and a certain robot receiving the task is executing the task with lower priority, suspending the task with lower priority and storing the task with lower priority into an uncompleted task list.
12) After a certain robot finishes the undertaken rescue task, if the uncompleted task list is empty, turning to step 1); otherwise, selecting the task with the highest priority from the incomplete task list for auction, and turning to the step 3).
The auctioneer's working process and conflict resolution strategy are shown in fig. 1 and 2, and bidder's responses and working processes are shown in fig. 3.
3 method comparative analysis
TABLE 1 search and rescue task assignment method comparison
Figure BSA0000166050280000091
Compared with the conventional disaster rescue task allocation method, the task allocation strategy designed by the invention has the following obvious advantages in several aspects:
(1) the most important objective of the operation in disaster search and rescue is to save the life safety of the wounded, because each wounded has different injury degrees, the critically wounded must be treated by priority, the invention sets the priority method of the task, guarantee this goal to be realized preferentially;
(2) disaster site conditions are different, some rescue tasks can be completed by one robot, some rescue tasks can be completed by the cooperation of a plurality of robots, and the characteristic is ignored in the prior art, but the disaster site condition can be effectively processed by the invention;
(3) the rescue site environment condition is severe, the fault of the search and rescue robot is possible to occur at any time, the task allocation strategy of the invention can be the most suitable robot to replace the fault robot, and the robustness of the designed method to the robot fault is realized;
(4) the traditional method carries out bidding based on the difference between benefit and cost, but the conversion coefficient between the benefit and the cost is determined darkly, the rationality of the bidding is difficult to ensure, and the robot needs to consider various factors such as the residual energy of the robot in order to successfully complete tasks in search and rescue operations, so the multi-parameter bidding method is established.
(5) Task dispersion processing is the characteristic of all rescue tasks, the distance is taken as the cost in the traditional method, and in order to cure the critically ill patient as soon as possible, the maximum speed of the robot is also taken into consideration when the robot throws the task;
(6) the process of finding tasks by robots in a team is asynchronous and concurrent, the default treatment of the invention is asynchrony, and when a plurality of robots find different tasks at the same time, the invention adopts a priority comparison method to limit the distribution of the treatment tasks of the critically ill patients.
(7) The auction conflict is solved by the existing method in the prior art, but only by a simple time delay method, and the invention solves the conflict from the character priority, thereby having obvious significance for life rescue.
The following is a simulation comparison between the traditional BP neural network and the BP neural network based on the L-M algorithm. Simulation equipment: notebook computer, CPU: i5-4210U, memory: 4G, system Windows8.1 professional edition, simulation software is Matlab 2015 b. The input data is normalized scaled values of all robots, wherein each output neuron corresponds to a robot with the same label to complete benefit evaluation of the task, and the larger the value is, the more advantageous the corresponding robot to complete the task is. A BP neural network was created using the newff function in Matlab 2015b, and then the normalized data was network trained using the train function. The traditional BP neural network algorithm trains input data by using a trailing function, the BP neural network based on the L-M algorithm trains by using a self-programmed function, and the training results are shown in fig. 4 and 5.
When a certain robot finds a new task in the current search direction, the robot carries out auction on the task; or after completing a task, performing auction on the task with the highest priority stored in the incomplete task list; when the auction starts, the auctioneer issues information to potential auctioneers in a broadcasting mode; the auction participants respond according to the current state of the auction participants and the priority of the tasks being undertaken; if the auctioneer is idle or the priority of the undertaken rescue task is lower than the auctioning task, making a response for receiving the task to the auctioneer and bidding according to the current state; otherwise, the bidder will not participate in the auction. If the auctioneer receives a bid amount less than the minimum number of machines required to complete the task before the auction deadline, the auction process ends; otherwise, the auctioneer normalizes the bid prices, then comprehensively sorts the processed bid information by adopting the trained improved BP neural network, and selects the robot for executing the task according to the sorting size. The invention can effectively make the critically ill patients get the first treatment, and the method has the robustness to the robot fault. Computer simulation verifies the training effect of the improved BP neural network.
The above description is only a preferred embodiment of the present invention, and the scope of the present invention is not limited thereto, and any simple modifications or equivalent substitutions of the technical solutions that can be obviously obtained by those skilled in the art within the technical scope of the present invention are within the scope of the present invention.

Claims (2)

1. The multi-robot disaster search and rescue task allocation method based on the improved BP neural network is characterized by comprising the following steps of:
1) each robot of the multi-robot system starts to search the environment according to a depth-first principle;
2) when a certain robot finds a new rescue task in a given direction, performing auction on the task; the auction robot is called an auctioneer;
3) the auctioneer distributes the task information to all robots in the team by broadcasting the information needing to complete the task;
4) each robot in the team is a potential auctioneer, and after receiving search and rescue task information issued by the auctioneer, the robots respond according to the current state of the robots:
if the current robot is in an idle state, participating in bidding;
② if the robot is currently performing a task, but the task performed has a lower priority than the auctioned task,
participation in bidding;
third, if the robot is currently performing a task, but the task performed has a higher priority than the auctioned task,
then not participate in bidding;
5) if the auction conflict occurs, calling an auction conflict resolution strategy;
6) if the number of the bidding robots received by the auctioneer before the auction deadline is less than the number of the robots required for completing the task, the auction process is ended, and the task is stored in respective incomplete task lists by all the robots in the team;
7) otherwise, the auctioneer normalizes the bid price, then uses the trained improved BP neural network to comprehensively sort the multiple bid prices given by the robot to participate in the task, and selects n c A winner;
8) the auctioneer notifies n c The winner will execute the task, and the auction receives the notice and sends contract confirmation information to the auctioneer; otherwise, the auctioneer considers that the auctioneer abandons the task because of the fault, and then sequentially selects other robots to participate in the task from other bidding robots according to the evaluation size of the bid value;
9) if the number of the auctioneers receiving the contract confirmation information is larger than the minimum number of robots required for completing the task, establishing a contract and starting to execute the rescue task;
10) if the number of the messages confirmed by the auctioneer after receiving the contract is less than the minimum number of robots required for completing the task, the auctioneer cannot establish the contract; all robots store the tasks into an incomplete task list, the idle robots continue to search in the preset direction, and the rescue robots continue to execute the original rescue tasks;
11) if the contract is established and a certain robot receiving the task is executing the task with lower priority, suspending the task with lower priority and storing the task with lower priority into an uncompleted task list;
12) after a certain robot finishes the undertaken rescue task, if the uncompleted task list is empty, turning to step 1); otherwise, selecting the task with the highest priority from the uncompleted task list for auction, and turning to the step 3);
in the step 7), improving the BP neural network, namely the BP neural network based on the Levenberg-Marquardt algorithm, and setting x (k) Is the network weight vector of the kth iteration, the weight vector x of the (k + 1) th iteration (k+1) Obtained by the following formula:
x (k+1) ＝x (k) +Δx (k+1) (1)
in the calculation rule of Gauss-Newton method
Δx (k+1) ＝-[J T (x)J(x)] -1 J(x)e(x) (2)
Wherein e (x) is [ e ] 1 (x)…e N (x)] T Is the output error vector, where x represents x k The upper corner marks are omitted under the condition of not causing confusion, and the same is applied below; n is the dimension of the neural network output, J (x) is the Jacobian matrix of error versus weight, which is calculated as
Figure FDA0003749574180000031
In the weight update value formula (2), J (x) T J (x) presence of irreversibleThe following form of weight update is used:
Δx (k+1) ＝-[J T (x)J(x)+λI] -1 J(x)e(x) (4)
in the formula, λ is a constant, and I is an n × n unit matrix.
2. The method for distributing the multi-robot disaster search and rescue task based on the improved BP neural network as claimed in claim 1, wherein in step 3), the tasks published by the auctioneer include the geographical location of the wounded, the task bid deadline time, and the number of robots n required by the task c Information of the priority of the task.
CN201810673369.7A 2018-06-16 2018-06-16 Multi-robot disaster search and rescue task allocation method based on improved BP neural network Active CN108985580B (en)

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 CN108985580A (en) 2018-12-11
CN108985580B true 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)

Families Citing this family (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109919431B (en) * 2019-01-28 2023-04-07 重庆邮电大学 Heterogeneous multi-robot task allocation method based on auction algorithm
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
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
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
CN115609608B (en) * 2022-12-02 2023-03-10 北京国安广传网络科技有限公司 All-weather health management robot

Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8965578B2 (en) * 2006-07-05 2015-02-24 Battelle Energy Alliance, Llc Real time explosive hazard information sensing, processing, and communication for autonomous operation
EP2645196B1 (en) * 2012-03-30 2018-12-12 The Boeing Company Network of unmanned vehicles
CN102915465B (en) * 2012-10-24 2015-01-21 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)
CN105069530B (en) * 2015-08-13 2019-02-26 肇庆学院 A kind of multi-robot Task Allocation based on multiple-objection 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
CN105843227B (en) * 2016-04-15 2018-10-23 上海大学 A kind of multi-robot Cooperation of task based access control closeness dynamic adjustment surrounds and seize method for allocating tasks

Patent Citations (2)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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

Also Published As

Publication number Publication date
CN108985580A (en) 2018-12-11

Similar Documents

Publication Publication Date Title
CN108985580B (en) Multi-robot disaster search and rescue task allocation method based on improved BP neural network
Zhao et al. A heuristic distributed task allocation method for multivehicle multitask problems and its application to search and rescue scenario
CN105843227B (en) A kind of multi-robot Cooperation of task based access control closeness dynamic adjustment surrounds and seize method for allocating tasks
CN110053052B (en) Task allocation and path planning method for multilayer SOM heterogeneous welding robot
CN111784211B (en) Cluster-based group multitask allocation method and storage medium
Vig et al. Market-based multi-robot coalition formation
CN113671987B (en) Multi-machine distributed time sequence task allocation method based on non-deadlock contract net algorithm
Ahmed et al. Task assignment for a physical agent team via a dynamic forward/reverse auction mechanism
CN109919431A (en) Heterogeneous multi-robot method for allocating tasks based on auction algorithm
Li et al. Multi-AUV cooperative task allocation based on improved contract network
CN114154819A (en) Task similarity-based multi-AGV distributed scheduling method and system
CN111798097B (en) Autonomous mobile robot task allocation processing method based on market mechanism
Zhu et al. Agent-based dynamic scheduling for earth-observing tasks on multiple airships in emergency
Dai et al. Research on multi-robot task allocation based on BP neural network optimized by genetic algorithm
CN109886574B (en) Multi-robot task allocation method based on improved threshold method
CN115113626B (en) Robot distributed control system and method thereof
Zhou et al. A task assignment strategy for multi-AUV collaborative hunting problem
Song et al. A distributed bidirectional auction algorithm for multirobot coordination
Huang et al. Multi robot cooperative rescue based on two-stage task allocation algorithm
Zheng et al. Frontier Point Allocation in Multi-Robot Collaborative Exploration with a Hybrid Auction Algorithm
Cui et al. A dynamic task equilibrium allocation algorithm based on combinatorial auctions
Wang et al. A distributed task assignment algorithm for multi-robots in plant factory
Liang et al. Dynamic Task Allocation Method for Heterogenous Multiagent System in Uncertain Scenarios of Agricultural Field Operation
Busquets et al. Learning when to auction and when to bid
Zhao et al. A Dynamic Task Assignment Method Based on Auction Algorithm for Dual-arm Robots

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