CN112181608A - Distributed distribution algorithm for multipoint dynamic aggregation tasks based on local information - Google Patents

Distributed distribution algorithm for multipoint dynamic aggregation tasks based on local information Download PDF

Info

Publication number
CN112181608A
CN112181608A CN201910592037.0A CN201910592037A CN112181608A CN 112181608 A CN112181608 A CN 112181608A CN 201910592037 A CN201910592037 A CN 201910592037A CN 112181608 A CN112181608 A CN 112181608A
Authority
CN
China
Prior art keywords
task
agent
communication range
intelligent
intelligent agent
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
Application number
CN201910592037.0A
Other languages
Chinese (zh)
Other versions
CN112181608B (en
Inventor
裘智峰
张亮
管建锋
王一帆
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.)
Central South University
Original Assignee
Central South 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 Central South University filed Critical Central South University
Priority to CN201910592037.0A priority Critical patent/CN112181608B/en
Publication of CN112181608A publication Critical patent/CN112181608A/en
Application granted granted Critical
Publication of CN112181608B publication Critical patent/CN112181608B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/48—Program initiating; Program switching, e.g. by interrupt
    • G06F9/4806—Task transfer initiation or dispatching
    • G06F9/4843—Task transfer initiation or dispatching by program, e.g. task dispatcher, supervisor, operating system
    • G06F9/4881—Scheduling strategies for dispatcher, e.g. round robin, multi-level priority queues
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F9/00—Arrangements for program control, e.g. control units
    • G06F9/06—Arrangements for program control, e.g. control units using stored programs, i.e. using an internal store of processing equipment to receive or retain programs
    • G06F9/46—Multiprogramming arrangements
    • G06F9/50—Allocation of resources, e.g. of the central processing unit [CPU]
    • G06F9/5005—Allocation of resources, e.g. of the central processing unit [CPU] to service a request
    • G06F9/5027—Allocation of resources, e.g. of the central processing unit [CPU] to service a request the resource being a machine, e.g. CPUs, Servers, Terminals
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N20/00—Machine learning

Landscapes

  • Engineering & Computer Science (AREA)
  • Software Systems (AREA)
  • Theoretical Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Engineering & Computer Science (AREA)
  • General Physics & Mathematics (AREA)
  • Artificial Intelligence (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Medical Informatics (AREA)
  • Computing Systems (AREA)
  • Mathematical Physics (AREA)
  • Multi Processors (AREA)

Abstract

The invention provides a distributed distribution algorithm of a multipoint dynamic aggregation task based on local information, which mainly solves the problem of distributed distribution of the multipoint dynamic aggregation task. And researching a multi-agent distributed scheduling algorithm based on local information. The intelligent agent carries out autonomous decision making according to local information and forms a preliminary distribution scheme through pre-distribution; and dynamically adjusting by taking the time for completing the optimized task as a necessary condition, communicating by the intelligent agent at intervals of a certain time, and judging whether to dynamically adjust according to the local information. The task allocation method provided by the invention takes the shortest overall task completion time as an optimization target, realizes the rapid allocation of dynamic multiple tasks, reduces the total time for task completion, can fully mobilize the intelligent agent in the system to participate in the task completion through a multi-stage allocation strategy, and improves the overall efficiency of the system.

Description

Distributed distribution algorithm for multipoint dynamic aggregation tasks based on local information
Technical Field
The invention relates to the technical field of multi-agent task allocation algorithms, in particular to a multi-point dynamic aggregation task distributed allocation algorithm based on local information.
Background
In recent years, more and more researchers at home and abroad research the task allocation problem of multiple intelligent agents, and rich technical development experience and theoretical achievement are obtained. People began to research the task allocation problem of the multi-agent system from the 20 th century and the 80 th century, the research on the task allocation problem is developed from a simple small-scale static task allocation problem to a complex large-scale dynamic task allocation problem, and the research on the task allocation problem is developed from a small-scale centralized task allocation which has global optimization capability but huge calculation amount to a large-scale distributed task allocation which has high solving speed but is difficult to realize global optimization.
The idea of the current centralized task allocation method is derived from an optimization control theory, but multiple agents are only used as executors rather than decision makers, and the adopted methods include an integer programming method, a branch definition method, a cut plane method, a hidden enumeration method, an intelligent optimization algorithm and the like, wherein the intelligent optimization algorithm has high convergence rate in solving the problem of large-scale centralized task allocation, but easily falls into the problems of local optimization, incapability of performing multiple allocation and the like, and can not meet the requirement of real-time performance while wasting the resources of the agents, and seriously influences the task completion degree.
Disclosure of Invention
The invention provides a local information-based multipoint dynamic clustering task distributed distribution method which overcomes or at least partially solves the problems, mainly considers the communication constraint of an intelligent agent, and designs a local information-based multi-intelligent agent distributed scheduling algorithm. The algorithm is divided into a pre-allocation part and a dynamic adjustment part, an intelligent body carries out autonomous decision making according to local information, and a preliminary task allocation scheme is formed through the pre-allocation. Because the intelligent agent can only obtain local information, the task allocation scheme obtained by pre-allocation is not necessarily optimal, and therefore, a dynamic adjustment part is added to dynamically adjust the preliminary task allocation scheme formed by pre-allocation under the condition that the task completion time is optimized. The intelligent agent carries out communication once at regular intervals and determines whether to dynamically adjust or not according to the latest local information.
In order to achieve the above object, the present invention provides a distributed distribution method for multipoint dynamic rendezvous tasks based on local information, which comprises:
s1: the pre-allocation is that each intelligent agent carries out autonomous decision according to the local information obtained in the initial period to obtain a primary task allocation scheme;
s2: the dynamic adjustment is that the intelligent agent judges whether the target task exceeds the optimal execution capacity according to local information obtained by the intelligent agent, finds out the task needing to be supported in the communication range and makes a decision on whether the intelligent agent is to support the task;
in step S1, the method further includes:
s11: the intelligent agent counts task information in a communication range, sends local information obtained by the intelligent agent to the intelligent agent in the communication range, processes the obtained data packet and updates a database;
s12: the intelligent agent calculates the necessary execution capacity of each task according to the information in the database of the intelligent agent, counts the sum of the execution capacity of the intelligent agent for executing each task in the communication range, and screens out the tasks which do not meet the necessary execution capacity. If all tasks in the communication range meet the necessary execution capacity, jumping to step S14;
s13, selecting an optimal task as a task target by the intelligent agent for the tasks which are screened out in the step S12 and do not meet the necessary execution capacity;
s14, the intelligent agent calculates the optimal execution capacity of all tasks in the communication range and counts all tasks which do not meet the optimal execution capacity in the communication range;
s15, the intelligent agent updates the target task information in the database, packs the database information and sends the information to other intelligent agents in the communication range;
and S16, the intelligent agent does not perform initial allocation after selecting the target task, the intelligent agent without the target task jumps to S11, and all the intelligent agents finish the target task to form an initial task allocation scheme.
Further, in step S11, the information acquiring part mainly acquires the task information in the communication range.
Each agent counts tasks in the communication range, and obtains specific information of all tasks in the communication range, including serial numbers of the tasks, positions of the tasks in the working environment, real-time task amount and fire growth speed.
And comparing the real-time task quantity of each task with a task threshold, if the real-time task quantity of the task is lower than the task threshold, indicating that the task is completed, recording the task as 1 in a corresponding completion item of the task in the task information matrix, and otherwise, recording the task as-1, indicating that the task is not completed. And simultaneously, recording the acquired task information of the tasks in the communication range in a task information matrix in the intelligent agent database, and recording the data updating time.
Further, in the step S12, β is definedjTo perform task tjThe set of agents of (3) then acts on task tjSum of execution capabilities of agents βjCan be expressed as:
Figure RE-GDA0002309125450000031
in order to ensure that each task can be completed, a necessary execution capacity is set for each task, and when the tasks meet the necessary execution capacity, the tasks can be ensured to be finally completed.
We can calculate task tjNecessary performance capability ofj minComprises the following steps:
Figure RE-GDA0002309125450000041
we set task tjOptimum performance capability β ofj maxTwice the necessary execution capacity, then βj maxCan be expressed as:
Figure RE-GDA0002309125450000042
definition of TjAs task tjFinal time to be completed, final time to be completed for all tasks T (T)1,t2,...,tM) Determined by the task that was last completed, then T (T)1,t2,...,tM) Can be expressed as:
T(t1,t2,...,tM)＝max{T1,T2,...,TM}. (4)
further, in step S14, we can calculate the task t without interferencejCombustion area increase amount deltas at time deltatj1(t) the following:
Δsj1＝lj(t)·Δr (5)
assuming that the agent has certain execution capacity beta, the beta is the workload of the agent in unit time, and at the moment t, the task tjHas a total execution capacity of betajThen task tjThe total number of tasks Δ s that can be completed within the time Δ tj2Can be calculated as:
Δsj2＝βj(t)·Δt (6)
we can calculate the task t over time Δ tjTask variation amount Δ s ofjComprises the following steps:
Δsj＝Δsj1-Δsj2＝lj(t)·Δr-βj·Δt (7)
dividing both sides of the formula (2-3) by delta t simultaneously to obtain the task tjRate of change of combustion area at time t
Figure RE-GDA0002309125450000043
Comprises the following steps:
Figure RE-GDA0002309125450000051
suppose alphajAs task tjThe rate of increase of fire is then alphajCan be expressed as:
Figure RE-GDA0002309125450000052
the task is assumed hereThe combustion area is approximately a perfect circle, task tjCombustion area sj(t) can be expressed as:
sj(t)＝πr2. (10)
then one can get:
Figure RE-GDA0002309125450000053
we can get the task t under the action of the agentjThe fire spread model of (1) is as follows:
Figure RE-GDA0002309125450000054
further, in step S15, the information exchange part mainly communicates with other agents within the communication range to exchange local information obtained by the agents themselves.
The intelligent agent packs the local information in the database into a data packet, acquires the condition of the intelligent agent in the communication range, and sends the data packet containing the local information to other intelligent agents.
The intelligent agent sends data packets to other intelligent agents and receives data packets from other intelligent agents, and for a plurality of data packets sent by the same intelligent agent, the obtained data packets are deleted, and only the latest obtained data packets are reserved.
Further, in step S16, after the agent selects the target task, the agent does not perform the initial assignment, but the process goes to step S11 where the agent is not assigned to the target, and the calculation is performed again until all agents are assigned to the task, and then the assignment is finished, so as to form the initial task assignment scheme. In step S2, the method further includes:
s21; the intelligent agent counts task information in a communication range, sends local information obtained by the intelligent agent to the intelligent agent in the communication range, processes the obtained data packet and updates a database;
s22, the intelligent agent calculates the optimal execution capacity of each task in the communication range and the sum of the execution capacity of the intelligent agent actually executing the task according to the local information;
s23, the agent judges whether the target task meets the best executing ability, if not, quits the dynamic adjustment;
s24, the agent counts the position information of other executing agents of the target task, if the agent is farthest from the target, the agent becomes a support agent, otherwise, the agent exits the dynamic adjustment;
s25, screening out tasks which do not meet the optimal execution capacity in the communication range by the intelligent agent according to the information calculated in the step 2, calculating the difference value between the optimal execution capacity and the actual execution capacity of each task, and selecting the task with the largest execution capacity difference value as a rescue task;
and S26, the intelligent agent calculates the completion time of the original task and the rescue task before and after adjustment respectively. If the whole task completion time is shortened after the adjustment, the dynamic adjustment is carried out, the target task of the task is changed into a rescue task, and otherwise, the dynamic adjustment is not carried out.
Further, in step S21, the agent packs the local information in its database into a data packet, obtains the condition of the agent within the communication range, and sends the data packet containing the local information to other agents. The intelligent agent sends data packets to other intelligent agents and receives data packets from other intelligent agents, and for a plurality of data packets sent by the same intelligent agent, the obtained data packets are deleted, and only the latest obtained data packets are reserved.
Further, in the step S22, the task tjSum of execution capabilities of agents βjCan be expressed as:
Figure RE-GDA0002309125450000071
in order to ensure that each task can be completed, a necessary execution capacity is set for each task, when the tasks meet the necessary execution capacity, the tasks can be ensured to be finally completed, and the task t can be calculatedjNecessary performance capability ofj minComprises the following steps:
Figure RE-GDA0002309125450000072
we set task tjOptimum performance capability β ofj maxTwice the necessary execution capacity, then βj maxCan be expressed as:
Figure RE-GDA0002309125450000073
further, in step S23, for the constraint on the execution capabilities of the agents allocated to all tasks, the sum of the execution capabilities of the agents allocated to all tasks must be smaller than the sum of the execution capabilities of all agents:
Figure RE-GDA0002309125450000074
further, in step S24, it is set that at most one agent exists among agents acting on the same task at the same time as a rescue agent, and an agent farthest from the task is selected as the rescue agent.
The agent therefore only dynamically adjusts when the sum of the performance capabilities of the target task acting on itself is higher than the optimal performance capability of that task and is itself furthest away from the target task.
The task which does not meet the optimal execution capacity is taken as a rescue task in priority, according to the obtained local information, the intelligent agent calculates the difference value between the sum of the execution capacity of the intelligent agent acting on each task which does not meet the optimal execution capacity and the optimal execution capacity of the intelligent agent in the communication range, and the task with the largest execution capacity difference value is selected as the rescue task.
Then the intelligent agent calculates the completion time of the original target task and the rescue task before and after dynamic adjustment, if the completion time of the whole task after adjustment is shortened, the adjustment is carried out, the target task of the intelligent agent is changed into the rescue task, otherwise, the adjustment is not carried out, and the specific calculation method comprises the following steps:
let us assume at tkDynamically adjusting the time according to a formula
Figure RE-GDA0002309125450000081
Can get the task tjThe equation of state of (a) is:
Figure RE-GDA0002309125450000082
βjto act on task tjSum of execution capabilities of Agents of (1), sj(tk) Is task tjAt tkThe amount of tasks at a time.
By solving task tjThe state equation of (1), we can calculate the task tjIs completed by time TjComprises the following steps:
Figure RE-GDA0002309125450000083
suppose an agent riThe original target task of (1) is Yt (i), the agent riHas a task capability of betaiThe sum of the execution capabilities of the agents acting on task Yt (i) is βYt(i)Then, the completion time of the task yt (i) before the dynamic adjustment can be calculated as:
Figure RE-GDA0002309125450000084
suppose an agent riThe rescue task is Jt (i), and the intelligence of the task Jt (i) is actedThe sum of the performance capabilities of the energy bodies is betaJt(i)Then, the completion time T of the task Jt (i) before the dynamic adjustment can be calculatedJt(i)Comprises the following steps:
Figure RE-GDA0002309125450000091
let us assume at tkDynamically adjusting time of day, agent riTo execute the rescue task Jt (i), the sum of the execution capacities of the agents acting on the original target task Yt (i) is decreased, and then the sum β 'of the execution capacities of the agents of the adjusted original target task Yt (i) can be calculated'Yt(i)Comprises the following steps:
β′Yt(i)＝β′Yt(i)-βi (22)
we can calculate task Yt (i) task completion time T 'after dynamic adjustment'Yt(i)Comprises the following steps:
Figure RE-GDA0002309125450000092
if the Euclidean distance between tasks is set as the distance between the tasks, the distance between the original target task Yt (i) and the rescue task Jt (i) can be calculated as follows:
Figure RE-GDA0002309125450000093
wherein the position coordinate of the original target task Yt (i) in the working environment is (x)Yt(i),yYt(i)) The position coordinate of the rescue task jt (i) in the work environment is (x)Jt(i),yJt(i))。
Intelligent body riHas a movement speed viThen agent riThe time t (i) spent on the link to support task jt (i) is:
Figure RE-GDA0002309125450000094
intelligent body riUpon reaching the location of rescue task Jt (i), the performance of agents acting on rescue task Jt (i) will increase, and the sum of the performance of agents acting on rescue task Jt (i) adjusted to β'Jt(i)Can be calculated as:
β′Jt(i)＝βJt(i)+βi (26)
we can calculate a task completion time T 'of rescue task Jt (i)'Jt(i)Comprises the following steps:
Figure RE-GDA0002309125450000101
intelligent body riComparing the completion time of the original target task Yt (i) and the rescue task Jt (i) before and after adjustment, if yes
max(T′Yt(i),T′Jt(i))＜max(TYt(i),TJt(i)) (28)
Then the agent riIf the target task of (1) is changed to Jt (i), otherwise, no adjustment is carried out, and the agent riThe target task of (a) is still yt (i).
Drawings
FIG. 1 is a flow diagram of a pre-allocation of the present invention.
Fig. 2 is a flow chart of dynamic adjustment.
Fig. 3 is an environment map established by a cartesian coordinate system.
FIG. 4 shows the results of an example simulation.
FIG. 5 shows the state quantity variation trend of all tasks in the calculation example.
Detailed Description
The invention is described in detail below with reference to the attached drawing, which is a preferred example of various embodiments of the invention.
The pre-allocation is that each intelligent body makes an autonomous decision according to the local information obtained initially to obtain a preliminary task allocation scheme, and the dynamic adjustment is that the intelligent body judges whether the self target task exceeds the optimal execution capacity according to the local information obtained by the intelligent body, finds out the task needing support in the communication range and makes a decision on whether the intelligent body is to support the task.
Assuming that the working environment of the agent is an environment map established based on a cartesian coordinate system in a two-dimensional plane, as shown in fig. 3, in a given environment map, there are some agents, fires and obstacles, in which small circles with numbers represent the agent, small squares with numbers represent the task, and large squares with numbers represent the obstacle. We assume that there are 3 tasks in the environment map, and for convenience of description we number the tasks as task 1, task 2, and task 3. The attribute parameters of the task comprise a task number, a position abscissa X of the task in the environment, a position ordinate Y of the task in the environment, an initial task amount s (0) and a fire growth speed alpha, and the specific parameters are shown in the table.
Figure RE-GDA0002309125450000111
Assume that there are 10 agents in the environment map, numbered 1 to 10, that are performing fire suppression tasks. The attribute parameters of the agent include agent number, position abscissa X in the environment, position ordinate Y in the environment, fire extinguishing capability beta and moving speed v,
Figure RE-GDA0002309125450000112
the results of the example simulation are shown in FIG. 4. In order to simulate the execution scheme of the whole multi-agent system, the path of the agent is planned by using an A-x algorithm, and the dotted line in FIG. 4 is the driving path of the agent. For example, the target task of the agent 1 is task 1, and after the task 1 is completed, task 2 is executed, and then the path of the agent 1 is the initial position of the agent 1, the position of the task 1, and the position of the task 2. Due to communication constraints, the communication range of the agent is limited. The communication range of the intelligent agent is set to be a circular area which takes the intelligent agent as the center and takes the maximum communication distance of a straight line as the radius. As shown in fig. 4, the dotted circle represents the communication range of the agent, and it is set that other agents can receive the information sent by the agent in the dotted circle, and agents outside the dotted circle cannot receive the information sent by the agent. For example, agent 9 may be able to communicate with agent 10 and obtain real-time information for task 3, and other agents and tasks may not be within communication range of agent 9, and therefore agent 9 may not obtain relevant information. The task allocation scheme obtained by pre-allocation in the simulation example is shown in table 3, where 1 in the setting table indicates that the agent executes the task, and 0 indicates that the agent does not execute the task. Only one target task of each intelligent agent at the same time is needed, and the same task can be completed by the cooperation of a plurality of intelligent agents. As can be seen from tables 3-3, the pre-allocation results show that the target tasks of agents 1, 2, 5, 6, and 7 are task 1, the target tasks of agents 3, 4, and 8 are task 2, and the target tasks of agents 9 and 10 are task 3.
Figure RE-GDA0002309125450000121
Figure RE-GDA0002309125450000131
The state quantity change trend of all tasks in the calculation example is shown in fig. 5, task 3 is completed firstly, and the consumed time is 29 seconds; then task 1 is completed, consuming 63 seconds; task 2 is completed last, consuming 78 seconds, and the overall task completion time is 78 seconds. When t is 35 seconds, the agent 7 performs dynamic adjustment to change the target task to task 2, and it can be seen from fig. 5 that the task amount of the original target task of the agent decreases after 35 seconds, and the speed of decreasing task 2 is slightly increased. When t is 63 seconds, task 1 is completed, the agent which originally executes task 1 changes the target task into task 2 through dynamic adjustment, after the agent reaches task 2, we can find that the task amount of task 2 is rapidly reduced, and when t is 78 seconds, task 2 is completed.
The invention is described above by way of example with reference to the accompanying drawings, and it is obvious that the invention specifically implements a distributed allocation algorithm for multipoint dynamic aggregation tasks based on local information. And finally, carrying out simulation analysis on the multi-agent distributed scheduling algorithm facing the multi-point dynamic aggregation task and based on the local information, and verifying the feasibility of the multi-agent distributed scheduling algorithm.

Claims (9)

1. A distributed distribution method for multipoint dynamic aggregation tasks based on local information is characterized by comprising the following steps:
s1: the pre-allocation is that each agent obtains a task allocation scheme according to the initial local information;
s2: the dynamic adjustment is that the intelligent agent decides whether to support or not according to the target task obtained by the intelligent agent.
2. The method according to claim 1, wherein in the step S1, the method further comprises:
s11: the intelligent agent counts task information in a communication range, sends local information obtained by the intelligent agent to the intelligent agent in the communication range, processes an obtained data packet, updates a database, calculates the necessary execution capacity of each task, counts the sum of the execution capacity of the intelligent agent for executing each task in the communication range, screens out tasks which do not meet the necessary execution capacity, and selects an optimal task as a task target;
s12, the intelligent agent calculates the optimal execution capacity of all tasks in the communication range, counts all tasks which do not meet the optimal execution capacity in the communication range, does not meet the optimal execution capacity, and selects an optimal task target from all unsatisfied tasks in the communication range;
s13, selecting a nearest task as a target task by the task of the intelligent agent outside the communication range; the intelligent agent updates the target task information in the database, packs the database information of the intelligent agent and sends the packed database information to other intelligent agents in the communication range;
and S14, the intelligent agent does not perform initial distribution after selecting the target task, the intelligent agent without the target task performs redistribution, and all the intelligent agents finish the distribution after having the target task to form an initial task distribution scheme.
3. The method according to claim 2, wherein in step S11, the agent packages the local information in its database into a data packet to be sent to other agents, and only other agents within the communication range of the agent can receive the data packet of the agent due to the limited communication range of the agent. And updating the obtained data packet.
4. The method according to claim 2, wherein in step S12, the necessary execution capacity of each task is calculated, and when the task satisfies the necessary execution capacity, it is ensured that the task can be finally completed.
5. The method according to claim 2, wherein in step S13, the optimal performance capabilities of all tasks in the communication range are calculated, if the optimal performance capabilities are not satisfied, one of all tasks that do not satisfy the optimal performance capabilities is selected as a task target, and if both the optimal performance capabilities and the task that does not satisfy the optimal performance capabilities are satisfied, the task in the communication range is selected as a target task.
6. The method according to claim 1, wherein in the step S2, the method further comprises:
s21: the intelligent agent counts task information in a communication range, sends local information obtained by the intelligent agent to the intelligent agent in the communication range, processes the obtained data packet and updates a database;
s22, the intelligent agent calculates the optimal execution capacity of each task in the communication range and the sum of the execution capacity of the intelligent agent actually executing the task according to the local information, and selects the rescue task;
s23, the agent counts the position information of other executing agents of the target task, if the agent is farthest from the target, the agent becomes a support agent, otherwise, the agent exits the dynamic adjustment;
and S24, the intelligent agent calculates the completion time of the original task and the rescue task before and after adjustment respectively. It is decided not to make dynamic adjustments.
7. The method according to claim 6, wherein in step S22, the agent calculates the sum of the optimal performance capability of each task in the communication range and the performance capability of the agent actually executing the task according to the local information, the agent determines whether its target task satisfies the optimal performance capability, if not, the agent selects the task that does not satisfy the optimal performance capability in the communication range, calculates the difference between the optimal performance capability and the actual performance capability of each task, and selects the task with the largest difference in performance capability as the rescue task.
8. The method of claim 6, wherein in step S23, since the computing power of the agent is limited, in order to reduce the computing burden of the agent, we set the agent to consider the dynamic adjustment when the sum of the performance capabilities of the agent acting on the target task is higher than the optimal performance capability of the task.
9. The method according to claim 6, wherein in step S24, if the overall task completion time after the adjustment is shortened, the dynamic adjustment is performed, and the target task of the task is changed to a rescue task, otherwise the dynamic adjustment is not performed.
CN201910592037.0A 2019-07-03 2019-07-03 Distributed distribution method for multipoint dynamic aggregation tasks based on local information Active CN112181608B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201910592037.0A CN112181608B (en) 2019-07-03 2019-07-03 Distributed distribution method for multipoint dynamic aggregation tasks based on local information

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201910592037.0A CN112181608B (en) 2019-07-03 2019-07-03 Distributed distribution method for multipoint dynamic aggregation tasks based on local information

Publications (2)

Publication Number Publication Date
CN112181608A true CN112181608A (en) 2021-01-05
CN112181608B CN112181608B (en) 2023-10-31

Family

ID=73915923

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201910592037.0A Active CN112181608B (en) 2019-07-03 2019-07-03 Distributed distribution method for multipoint dynamic aggregation tasks based on local information

Country Status (1)

Country Link
CN (1) CN112181608B (en)

Cited By (3)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113095645A (en) * 2021-03-31 2021-07-09 中国科学院自动化研究所 Heterogeneous unmanned aerial vehicle task allocation method for emergency scene with unevenly distributed tasks
CN113112079A (en) * 2021-04-19 2021-07-13 中国人民解放军96901部队26分队 Task allocation method based on heuristic dynamic deepening optimization algorithm
CN114595971A (en) * 2022-03-09 2022-06-07 中国电子科技集团公司第五十四研究所 Distributed decision-making method for local task parameter sharing

Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN104166750A (en) * 2014-06-18 2014-11-26 南京邮电大学 Robocup rescue cooperation method based on weighing cooperation algorithm
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)
CN105975332A (en) * 2016-05-03 2016-09-28 北京理工大学 Method for forming multi-agent distributed union
US20170131727A1 (en) * 2015-11-06 2017-05-11 Massachusetts Institute Of Technology Dynamic task allocation in an autonomous multi-uav mission
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
CN106886459A (en) * 2017-01-24 2017-06-23 浙江工商大学 A kind of multiple agent internet data acquisition tasks distribution method based on actual measurement bandwidth
CN106897129A (en) * 2017-01-24 2017-06-27 浙江工商大学 A kind of multiple agent internet data acquisition tasks dispatching method based on region
CN107820276A (en) * 2017-10-27 2018-03-20 北京邮电大学 A kind of wireless senser method for allocating tasks
CN108009012A (en) * 2017-12-14 2018-05-08 中南大学 A kind of multiple agent dynamic task allocation method of task based access control model
CN108334986A (en) * 2018-02-06 2018-07-27 东华大学 A kind of multiple agent Multipurpose Optimal Method of the rescheduling method based on plastic mechanism
CN108416488A (en) * 2017-12-21 2018-08-17 中南大学 A kind of more intelligent robot method for allocating tasks towards dynamic task
US20180326583A1 (en) * 2017-05-11 2018-11-15 King Fahd University Of Petroleum And Minerals Dynamic multi-objective task allocation
US20190049975A1 (en) * 2017-08-11 2019-02-14 Tata Consultancy Services Limited Method and system for optimally allocating warehouse procurement tasks to distributed robotic agents

Patent Citations (13)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
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)
CN104166750A (en) * 2014-06-18 2014-11-26 南京邮电大学 Robocup rescue cooperation method based on weighing cooperation algorithm
US20170131727A1 (en) * 2015-11-06 2017-05-11 Massachusetts Institute Of Technology Dynamic task allocation in an autonomous multi-uav mission
CN105975332A (en) * 2016-05-03 2016-09-28 北京理工大学 Method for forming multi-agent distributed union
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
CN106897129A (en) * 2017-01-24 2017-06-27 浙江工商大学 A kind of multiple agent internet data acquisition tasks dispatching method based on region
CN106886459A (en) * 2017-01-24 2017-06-23 浙江工商大学 A kind of multiple agent internet data acquisition tasks distribution method based on actual measurement bandwidth
US20180326583A1 (en) * 2017-05-11 2018-11-15 King Fahd University Of Petroleum And Minerals Dynamic multi-objective task allocation
US20190049975A1 (en) * 2017-08-11 2019-02-14 Tata Consultancy Services Limited Method and system for optimally allocating warehouse procurement tasks to distributed robotic agents
CN107820276A (en) * 2017-10-27 2018-03-20 北京邮电大学 A kind of wireless senser method for allocating tasks
CN108009012A (en) * 2017-12-14 2018-05-08 中南大学 A kind of multiple agent dynamic task allocation method of task based access control model
CN108416488A (en) * 2017-12-21 2018-08-17 中南大学 A kind of more intelligent robot method for allocating tasks towards dynamic task
CN108334986A (en) * 2018-02-06 2018-07-27 东华大学 A kind of multiple agent Multipurpose Optimal Method of the rescheduling method based on plastic mechanism

Non-Patent Citations (3)

* Cited by examiner, † Cited by third party
Title
FAROUQ ZITOUNI ETAL: "\"An adaptive protocol for dynamic allocation of tasks in a multi-robot system\"", 《2016 INTERNATIONAL CONFERENCE ON ADVANCED ASPECTS OF SOFTWARE ENGINEERING (ICAASE)》 *
JIANFENG GUAN等: ""Vehicles for Material Distribution Based on Multi-stage Auction Algorithm"", 《SCHEDULING OF MULTI-LOAD AUTOMATED GUIDED VEHICLES FOR MATERIAL DISTRIBUTION BASED ON MULTI-STAGE AUCTION ALGORITHM》 *
陈夏冰;刘国栋;刘丽娟;: "基于分区的多机器人任务分配", 江南大学学报(自然科学版), no. 04 *

Cited By (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN113095645A (en) * 2021-03-31 2021-07-09 中国科学院自动化研究所 Heterogeneous unmanned aerial vehicle task allocation method for emergency scene with unevenly distributed tasks
CN113095645B (en) * 2021-03-31 2023-06-23 中国科学院自动化研究所 Heterogeneous unmanned aerial vehicle task allocation method aiming at emergency scene with uneven task distribution
CN113112079A (en) * 2021-04-19 2021-07-13 中国人民解放军96901部队26分队 Task allocation method based on heuristic dynamic deepening optimization algorithm
CN113112079B (en) * 2021-04-19 2022-11-15 中国人民解放军96901部队26分队 Task allocation method based on heuristic dynamic deepening optimization algorithm
CN114595971A (en) * 2022-03-09 2022-06-07 中国电子科技集团公司第五十四研究所 Distributed decision-making method for local task parameter sharing

Also Published As

Publication number Publication date
CN112181608B (en) 2023-10-31

Similar Documents

Publication Publication Date Title
CN112181608A (en) Distributed distribution algorithm for multipoint dynamic aggregation tasks based on local information
WO2022000924A1 (en) Double-resource die job shop scheduling optimization method based on ammas-ga nested algorithm
CN106875090B (en) Dynamic task-oriented multi-robot distributed task allocation forming method
US9131529B1 (en) System and method for demand driven network topology management
CN108681787A (en) Based on the unmanned plane method for optimizing route for improving the two-way random tree algorithm of Quick Extended
CN109634310B (en) Self-adaptive multi-robot-based optimized formation control method and system
CN111432433B (en) Unmanned aerial vehicle relay intelligent flow unloading method based on reinforcement learning
CN112068586B (en) Space-time joint optimization four-rotor unmanned aerial vehicle trajectory planning method
CN111766784B (en) Iterative optimization method for multi-robot pattern composition in obstacle environment
CN106154836B (en) A kind of online dynamic particles group PID optimization method
CN104063541A (en) Hierarchical decision making mechanism-based multirobot cooperation method
CN114115329B (en) Relay collaborative unmanned aerial vehicle task planning method and device
CN113281993B (en) Greedy K-mean self-organizing neural network multi-robot path planning method
WO2018086041A1 (en) Method and device for dynamically adjusting flight position of aerial vehicle
CN112731942A (en) Multi-AUV formation control method based on improved navigator virtual structure method
CN105160433A (en) Assembly line multi-target modeling method, particle swarm algorithm and optimization scheduling method
CN110632940B (en) Active anti-interference time-varying track tracking control method for multiple unmanned aerial vehicles with hybrid quantizers
CN113159369B (en) Multi-forest-area scheduling route planning method based on optimized genetic algorithm
CN117434838A (en) Variable time domain event trigger intersection butt joint collaborative prediction control method
CN106440844B (en) A kind of grate-cooler scraper velocity control method
CN116857778A (en) Automatic control method of subway station air conditioning system based on reinforcement learning
CN112184400B (en) Heterogeneous multi-agent multi-stage distributed auction algorithm based on local information
CN112734136B (en) Particle swarm optimization-based rotation irrigation group optimization method and system
CN108805445B (en) Grouping sequence scheduling method for providing rotary standby for air conditioner load group
CN115857354A (en) Method for optimizing plantar force distribution and trajectory tracking of quadruped robot

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