CN108416488B - Dynamic task-oriented multi-intelligent-robot task allocation method - Google Patents

Dynamic task-oriented multi-intelligent-robot task allocation method Download PDF

Info

Publication number
CN108416488B
CN108416488B CN201711394738.0A CN201711394738A CN108416488B CN 108416488 B CN108416488 B CN 108416488B CN 201711394738 A CN201711394738 A CN 201711394738A CN 108416488 B CN108416488 B CN 108416488B
Authority
CN
China
Prior art keywords
task
intelligent robot
intelligent
tasks
time
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
CN201711394738.0A
Other languages
Chinese (zh)
Other versions
CN108416488A (en
Inventor
裘智峰
陈杰
杨宁
管建锋
郭宇骞
桂卫华
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 CN201711394738.0A priority Critical patent/CN108416488B/en
Publication of CN108416488A publication Critical patent/CN108416488A/en
Application granted granted Critical
Publication of CN108416488B publication Critical patent/CN108416488B/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
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/123—DNA computing
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
    • G06N3/00—Computing arrangements based on biological models
    • G06N3/12—Computing arrangements based on biological models using genetic models
    • G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming

Abstract

The invention provides a dynamic task-oriented multi-intelligent-robot task allocation method, which mainly solves the problem of multi-task allocation of a task state measuring tool with time-varying characteristics. The method comprises the following steps: firstly, acquiring dynamic task characteristic parameters, and establishing a characteristic equation of task point state quantities by combining intelligent robot capability parameters; designing a revenue function of the intelligent robot according to a characteristic equation; secondly, designing a genetic algorithm fitness function according to the income function; further designing a genetic algorithm difference selection operator and a local mutation operator, and proposing an algorithm repair strategy; and finally, generating an intelligent robot task allocation scheme by using a genetic algorithm to complete multi-task allocation. The task allocation method provided by the invention aims at obtaining the maximum benefit of the system, realizes the rapid allocation of dynamic multi-tasks, solves the problem of chromosome deadlock of the algorithm, and avoids the search from falling into local optimum; through the multi-stage allocation strategy, the intelligent robot in the system can be fully mobilized to participate in completing tasks, and the overall efficiency of the system is improved.

Description

Dynamic task-oriented multi-intelligent-robot task allocation method
Technical Field
The invention relates to the technical field of intelligent robot task allocation algorithms, in particular to a dynamic task-oriented multi-intelligent robot task allocation method.
Background
In recent years, with the deep understanding of people on artificial intelligence and complex systems, research in the field of intelligent multi-robot systems has achieved abundant results in the aspects of theory and practical systems, and intelligent robots can replace human beings to complete certain work under some severe environments, such as aerospace, deep sea exploration, exploration and disaster relief and the like, wherein the environments are often accompanied by substances harmful to human bodies, such as toxicity, oxygen-free, high temperature and high pressure, strong radiation and the like, and exceed the limit of human beings.
In the current multi-robot system, the oriented tasks are large in quantity and dynamically change, the traditional genetic algorithm can generate deadlock, easily get into local optimization and cannot carry out repeated distribution when solving the problem of dynamic task distribution, the intelligent robot resources are wasted, the real-time requirement cannot be met, and the task completion degree is seriously influenced.
Disclosure of Invention
The invention provides a dynamic task oriented multi-intelligent robot task allocation method which overcomes or at least partially solves the problems, and provides an intelligent multi-robot dynamic task allocation method based on an improved genetic algorithm according to the characteristics of the dynamic task allocation problem.
In order to achieve the above object, the present invention provides a dynamic task-oriented intelligent multi-robot task allocation method, which comprises:
s1: acquiring capability parameters, environment information, task initial characteristic parameters and related constraint conditions of the intelligent robot according to the environment information;
s2: designing a dynamic profit model according to the change of the state quantity of the dynamic task;
s3: calculating a profit value obtained by distributing each intelligent robot by the tasks;
s4: when a task occurs, a task allocation scheme is generated based on an improved genetic algorithm, and the intelligent robot executes a corresponding target task according to the task allocation scheme.
In step S1, the method further includes:
s11: the method comprises the steps of modeling an environment by establishing a coordinate system, wherein the environment is distributed in a plurality of intelligent robots with task execution capacity, tasks with attributes changing along with time and a plurality of static obstacles;
s12: and analyzing factors needing to be considered in task allocation and sorting out relevant constraint conditions according to the information of the environment.
Further, in step S11, the environment is an intelligent robot work environment, and it is assumed that the intelligent robot in the environment is an intelligent robotThe robot, the task point and the barrier are all located on the same plane, a rectangular coordinate system is established, and N is distributed (N belongs to Z)+) A task point, M (M ∈ Z)+) An intelligent robot and B (B belongs to Z)+) A static obstacle.
Obtaining an approximate coordinate of (x) for each task pointj,yj) J-1, 2, …, N, approximate coordinates (x) of the intelligent roboti,yi) I ═ 1,2, …, M and the approximate coordinates (x) of the obstacleb,yb),b＝1,2,…,B。
The linear distance formula between the jth task and the intelligent robot i is as follows:
Figure BDA0001518198630000031
let the coordinates of task point j and task point j' be (x)j,yj)，(xj′,yj′) Then, the distance between two points is expressed as:
Figure BDA0001518198630000032
further, in step S12, the state quantity of the task point j (j ═ 1,2, …, N) changes with time, and the state quantity of the task point j (j ═ 1,2, …, N) is expressed as:
Figure BDA0001518198630000033
in the formula:
Figure BDA0001518198630000034
is the amount of state change, alpha, of task j per unit timejThe state growth rate of task point j.
The intelligent robot i has an execution capability betaiThe set of intelligent robots working on task j is lambdajThe state quantity of the task j is expressed by the following equation:
Figure BDA0001518198630000035
this is a dynamic task allocation problem. The state quantity of the task changes along with time, and the growth speed of the task is alpha without external interventionjIt is influenced by how many intelligent robots are performing the task (the sum of the performance capabilities of the intelligent robots).
Assume that the m robot performing the target task j at time t performs the sum of
Figure BDA0001518198630000036
When in use
Figure BDA0001518198630000037
In time, it indicates that the m intelligent robots cannot complete the task j, sj(t) exhibits an ascending trend; when in use
Figure BDA0001518198630000038
Then, it indicates that m intelligent robots can complete the task, sj(t) shows a tendency to decline.
Further, in step S12, the system task allocation needs to consider factors including task point characteristic parameters and intelligent robot characteristic parameters.
The characteristic parameters of the task point j (j ═ 1,2, …, N) include: state quantity S of taskj(t) indicating the task state quantity at time t, and when t is 0, the task state quantity is the initial state quantity S of the taskj(0) (ii) a Rate of growth alphajAnd the coordinate position (x) in the environmentj,yj)。
The intelligent robot i (i ═ 1,2, …, M) capability parameters include: velocity v of movementiThe path length of the intelligent robot i in unit time is represented; performance capability βi(βi> 0) and the coordinate position (x) of the location in the environmentj,yj)。
Further, in the step S12, there are many constraints, and the function H is used to avoid collision of the intelligent robot during the forward target task1Description of the inventionThe situation that the intelligent robot collides with the obstacle:
Figure BDA0001518198630000041
in the formula: zi(t) represents the position state of the intelligent robot i at time t, ObIndicating the location area of the obstacle, H1(Zi(t),Zb) 1 means that the intelligent robot i collides with at least one obstacle.
Function H for the same reason2The situation of collision between intelligent robots is described as follows:
Figure BDA0001518198630000042
in the formula: dsafeIs the safe distance between the intelligent robots.
Therefore, the constraint conditions of the intelligent robot in the forward target process are as follows:
Figure BDA0001518198630000043
Figure BDA0001518198630000044
given the execution capacity and the requirement of the environment intelligent robot to be higher than the maximum growth speed in the task, the method is expressed by the following formula:
Figure BDA0001518198630000051
when the tasks are distributed, the sum of the execution capacity of the intelligent robot cannot be lower than the growth speed of the corresponding target task without external intervention, otherwise, the system has global deadlock, and the overall deadlock is expressed by a formula:
Figure BDA0001518198630000052
Figure BDA0001518198630000053
representing the sum of the intelligent robot performance capabilities on task j.
In step S2, in order to distinguish the importance of different tasks at time t, each task has a corresponding revenue function:
Figure BDA0001518198630000054
wherein IjIs the maximum benefit that can be gained to complete task j when
Figure BDA0001518198630000055
dφj(t)/dt > 0, wherein
Figure BDA0001518198630000056
As a function of phij(t) a time threshold increasing with increasing time t, in which casej(t) is a "price-rising function" of the revenue obtained by completing task j over time; when in use
Figure BDA0001518198630000057
dφj(t)/dt < 0, wherein
Figure BDA0001518198630000058
As a function of phij(t) a time threshold decreasing with increasing time t, in which casej(t) is a "discounting function" of the revenue obtained by completing task j over time; when in use
Figure BDA0001518198630000059
φj(t) ═ a, where a is a set constant value.
Considering that the robot reaching the target task needs to spend time on the road, setting
Figure BDA00015181986300000510
To complete task jTo obtain the maximum time of interest, wherein
Figure BDA00015181986300000511
Suppose that
Figure BDA00015181986300000512
For the time consumed on the road by the intelligent robot which reaches the target task j first, the price rising function can be set to
Figure BDA00015181986300000513
Figure BDA00015181986300000514
A discount function is set as
Figure BDA00015181986300000515
The formula for obtaining the maximum profit and the change of the task state quantity given to the completion of the task j is as follows:
Figure BDA0001518198630000061
in the formula, w1Is a weight coefficient, Sj(0) Is the initial state quantity of task j.
In step S3, it is assumed that the time required to complete task j is
Figure BDA0001518198630000062
The obtained benefit is
Figure BDA0001518198630000063
In order to reasonably allocate the intelligent robot for executing the task, earnings are allocated according to a labor allocation principle, and the method specifically comprises the following steps:
the income obtained by the ith intelligent robot when the task j is executed at the time t is expressed by the following formula:
Figure BDA0001518198630000064
in the formula, xij(t) epsilon {0,1} represents whether the ith intelligent robot executes the task j at the time t, if yes, the task j is 1, otherwise, the task j is 0, wherein
Figure BDA0001518198630000065
Which represents the sum of the performance capabilities of the intelligent robot performing task j at time t.
And allocating the income obtained by completing the task j to the intelligent robot i, wherein the income is expressed by the formula:
Figure BDA0001518198630000066
after the task of the whole system is completed, the total income obtained by the ith intelligent robot is expressed by a formula as follows:
Figure BDA0001518198630000067
in the formula IiIndicates the income, x, available to the intelligent robot i to complete all tasksijAnd e {0,1} represents whether the ith intelligent robot executes the task j, if so, the ith intelligent robot is 1, and otherwise, the ith intelligent robot is 0.
The total gain achieved by completing the entire system task is thus formulated as:
Figure BDA0001518198630000071
the goal of the overall system is to make the final task profitable
Figure BDA0001518198630000072
To a maximum, i.e.
Figure BDA0001518198630000073
In the formula (I), the compound is shown in the specification,
Figure BDA0001518198630000074
indicating the time at which task j is completed.
In step S4, the method further includes:
s41: initializing genetic parameters, coding and setting a fitness function;
s42: calculating the fitness value of each population;
s43: selecting operators, wherein the invention provides difference assessment selection, firstly performing pre-selection, secondly performing formal selection, and selecting S chromosomes of different templates to form a group k;
s44: carrying out single-point crossing, wherein the dynamic task is oriented to the invention, whether each chromosome is trapped in a global deadlock condition needs to be detected in real time, and if the chromosome is trapped in the global deadlock condition, a repair strategy is adopted;
s45: and (3) mutation, designing a new mutation operator, keeping the gene code with the maximum profit value and the position unchanged on the basis of a non-deadlock task on a chromosome, and randomly selecting two other genes for exchanging positions.
S46: generating a group k +1, judging whether the genetic algorithm evolution algebra reaches the set maximum evolution algebra, if so, determining an optimal distribution scheme, and starting the intelligent robot to execute a target task; otherwise, go back to step S32
S47: once the tasks are completed, the idle intelligent robot without target work exists, whether the number of the completed tasks reaches the given number of the tasks in the system is judged, if yes, the optimal solution is output, otherwise, the step S31 is returned, the genetic algorithm is applied again to allocate new tasks to the idle intelligent robot, and the system tasks are finished after all the tasks are executed.
Further, in step S41, the initialized genetic parameters include: and setting a maximum evolution algebra G, intelligent robot characteristic parameters and task characteristic parameters.
The attribute parameters of the task point j (j ═ 1,2, …, N) include: initial state quantity S of taskj(0) Growth rate alphajAnd the approximate coordinate position (x) in the environmentj,yj)。
Intelligent robot i: (i-1, 2, …, M) characteristic parameters include: velocity v of movementiPerformance betaiAnd the approximate coordinate position (x) of where in the environmenti,yi)
With machine coding, decoding operations can be avoided, the locus sequence value of the chromosome represents the number of the intelligent robot, and the string value represents the target number to be executed with a task, forming a mapping relationship, assuming that there are 6 intelligent robots and 4 task points, as shown in table 1.
Table 1 chromosomal coding pattern:
Figure BDA0001518198630000081
the task is assigned with a chromosome [231323], which means that the 1 st and 5 th intelligent robots execute task 2, the 2 nd, 4 th and 6 th intelligent robots execute task 3, the 3 rd intelligent robot executes target task 1, and task 4 on the chromosome is executed without the intelligent robot.
And (3) performing task allocation each time, and selecting a fitness function as follows:
Figure BDA0001518198630000082
further, in step S43, the selection operator of the present invention adopts a difference examination admission, the number of candidate individuals before each crossing is greater than the number S of chromosomes selected finally, and first performs examination selection on a part of individuals, then formally selects the number S of crossing individuals satisfying the requirement from the preselected individuals, and the eliminated individuals enter the next candidate list;
firstly, pre-selecting, evaluating the number C of candidate individuals for each evolution, wherein the evaluation index is the size of a fitness value, selecting H (H is more than S) individuals with the top rank of the fitness value as a candidate list, and sorting the H (H is more than S) individuals from large to small according to the fitness value;
and then formally selecting, removing individuals with the same encoding genes with the number 1 individual as a template, sequentially selecting S individuals with different templates as templates to form a population P (k), and removing the individuals to sequentially supplement the number of the lacking populations according to the size of the fitness value if the population specification number S meeting the requirement cannot be obtained, wherein k is 0 initially.
Here, similarity is used to judge whether templates are similar, and the formula is as follows:
Cii′＝kii′n (i, i ' is E (i, i ' is less than or equal to H) ^ i ≠ i ') (seventeen)
In the formula: in the formula: k is a radical ofii′The number of common gene values at the corresponding positions of the chromosome i and the chromosome i', and n is the length of a chromosome coding string; given a threshold value mu of a selection operator, when Cii′Mu is larger than or equal to mu, which means that the chromosome i is similar to the chromosome i', otherwise, the chromosome i is not similar.
Further, in step S44, there may be a large number of deadlocked chromosomes after crossing, and the conventional method is to discard the chromosomes, which is feasible for the weak constraint problem, but the problem faced by the present invention has many constraints, and the conventional method is not suitable.
When detecting that the gene codes on the chromosome are trapped in global deadlock, namely the intelligent robots are improperly distributed after the gene codes on the chromosome are coded, the sum of the execution capacity of the intelligent robot distributed by each task is smaller than the growth rate of the task without external intervention, which can cause that all tasks can not be completed, the invention adopts a repair strategy, which specifically comprises the following steps:
counting codes corresponding to the intelligent robot on the chromosome once the chromosome is detected to have global deadlock;
calculating the execution capacity of the intelligent robot at each task point and the growth rate ratio P of the intelligent robot to the taskj(0 < j < N) and sorting
Figure BDA0001518198630000101
Will be provided with
Figure BDA0001518198630000102
To intelligent robots with low on-task execution capacity
Figure BDA0001518198630000103
Task of (1), if
Figure BDA0001518198630000104
The task(s) has not yet been deadlocked, and continues to be
Figure BDA0001518198630000105
To intelligent robots with low on-task execution capacity
Figure BDA0001518198630000106
…, up to
Figure BDA0001518198630000107
The task of (1) removes the deadlock.
Through the technical scheme, a set of complete execution scheme is generated, and the time required for completing all tasks and the time required for completing the tasks of the whole system are obtained.
Based on the technical scheme, the invention provides a genetic algorithm facing to dynamic tasks, which mainly solves the problem of multi-task allocation of time-varying characteristics of a task state measuring tool; the algorithm designs a fitness function of the genetic algorithm, further designs a difference selection operator and a local mutation operator of the genetic algorithm, and provides an algorithm repair strategy, the algorithm can solve the problem of chromosome deadlock, avoids search from being trapped in local optimum, and can fully mobilize an intelligent robot in the system to participate in completing tasks through a multi-stage distribution strategy, so that the overall efficiency of the system is improved.
Drawings
FIG. 1 is a flow chart of the scheme of the invention.
FIG. 2 is an environmental model diagram of the present invention.
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.
In a preferred embodiment, the invention provides a dynamic task-oriented genetic algorithm, and solves the problems of multiple dynamic tasks in an environment through cooperative cooperation between intelligent robots.
The intelligent robot is embodied in a cooperative way as follows: modeling the environment, acquiring information in the environment, distributing tasks based on an improved genetic algorithm according to the acquired information when the tasks occur, once the tasks are completed, the intelligent robot in the system reuses the improved genetic algorithm to distribute the tasks for the idle intelligent robot through communication, and the intelligent robot is cooperated with other intelligent robots to execute unfinished tasks until all the tasks are completed.
The improved genetic algorithm employs the following steps: the method comprises the steps of obtaining a parameter set of problems, coding chromosomes, selecting operators through difference assessment, adopting single-point crossing, solving the chromosomes suffering from deadlock through a recovery strategy, detecting whether the completed tasks meet requirements or not when tasks are completed in a task space, finishing if yes, outputting the time for completing each task, and otherwise, performing task allocation on the idle intelligent robot by using an improved genetic algorithm again.
Referring to fig. 1, the population number C of the genetic algorithm candidates is 50, the pre-selected population number H is 35, the formally selected population number S is 30, the maximum iteration number G is 150, and w is w11.5; the number M of the intelligent robots is equal to 8, n is equal to 8, and the execution capacities of the intelligent robots are respectively as follows: β ═ β1,β2,…,β80.3,0.4,0.3,0.2,0.3,0.1,0.2, and the motion speeds of the intelligent robot are: v. ofi＝{v1,v2,…, v 82,3,1,2,4,2,1, 3; the number N of tasks is 3, and the growth speed alpha is { alpha ] without external intervention1,α2,α30.8,0.9,0.7, and the initial state quantity is Sj(0)＝{S1(0),S2(0),S3(0) 1.5,2,2.5, and 0.9 as the threshold μ; set limit time
Figure BDA0001518198630000111
The maximum value of the task is respectively as follows:
Figure BDA0001518198630000112
in the same way, can obtain I2＝7.05， I35.85; the price function can be set as
Figure BDA0001518198630000113
A discount function is set as
Figure BDA0001518198630000121
a＝0.5。
And (3) selecting an operator, firstly, pre-selecting, and then formally selecting satisfactory S chromosomes as a group P (k), wherein k is 0 initially.
Assume that the gene encoding chromosome A is [ 12132112 ]]The gene encoding chromosome B is [ 13123322 ]]If the number of the position coincidence genes corresponding to the two chromosomes is 3, then CAB0.375 < mu 0.9, so chromosomes a and B are dissimilar and can intersect.
By adopting single-point crossing, when the fact that the gene codes on the chromosome are trapped in the global deadlock is detected, the execution capacity and the value of the intelligent robot distributed by each task are smaller than the growth speed without external intervention, all tasks cannot be completed, and the deadlock problem is solved through a recovery strategy.
Two chromosomal a and B genes were assumed to encode as shown in table 2:
table 2:
Figure BDA0001518198630000122
the parent chromosome a and B gene codes did not create a deadlock problem, and the 3 gene coding bits of the two chromosomes crossed to generate two rudiments as shown in table 3:
table 3:
the young child of a:
Figure BDA0001518198630000126
b of the young children：
Figure BDA0001518198630000127
Then for the prototype child of B, the sum of the performance capabilities of the intelligent robot to perform task 1 is:
Figure BDA0001518198630000123
the sum of the performance capabilities of performing task 2 is:
Figure BDA0001518198630000124
the sum of the performance capabilities of performing task 3 is:
Figure BDA0001518198630000125
then the chromosome satisfies the deadlock constraint, i.e.
Figure BDA0001518198630000131
At this time, P needs to be calculatedj(j is more than 0 and less than N), and the values are respectively 0.75,1 and 0.71. Therefore, the intelligent robot 6 needs to be assigned task 2 for the times, the deadlock is removed, and finally the prototype child of B is coded as:
Figure BDA0001518198630000134
assuming that the chromosomal gene is encoded as [ 12132112 ]]Intelligent robot execution capability on task 1 and
Figure BDA0001518198630000132
by the same method
Figure BDA0001518198630000133
The task 1 of the chromosome is not deadlocked, the corresponding task 1 gene and the position are kept unchanged, two genes are randomly selected for transposition, and the No. 2 and No. 4 genes are supposed to be transposed to generate a new gene code as follows:
the novel gene is encoded as [ 13122112 ].
And when the genetic iteration times exceed the set maximum iteration times G, the conditions are met, the optimal task allocation scheme is generated, and the intelligent robot executes respective target tasks according to the task allocation scheme.
Once a task is completed, the system re-uses the improved genetic algorithm to perform task allocation for the idle intelligent robot until all tasks are completed and the time required for all tasks to be completed and the time required for completing the tasks of the whole system are obtained.
The present invention has been described in connection with the accompanying drawings, and it is to be understood that the invention is not limited to the specific embodiments described above, and that various modifications may be made without departing from the spirit and scope of the invention.

Claims (8)

1. A dynamic task-oriented multi-intelligent-robot task allocation method is characterized by comprising the following steps:
s1: acquiring capability parameters, environment information, initial characteristic parameters of a target task and related constraint conditions of the intelligent robot;
s2: designing a dynamic profit model according to the change of the state quantity of the dynamic task;
s3: generating a dynamic task allocation scheme based on an improved genetic algorithm, and executing tasks by the intelligent robot according to the task allocation scheme; the step S3 includes:
s31: initializing genetic parameters, coding and setting a fitness function;
s32: calculating the fitness value of each population;
s33: selecting operators, and providing difference assessment selection: firstly, pre-selecting, then formally selecting, and selecting S chromosomes of different templates to form a population P (k);
s34: carrying out single-point crossing, detecting whether each chromosome is in a global deadlock condition in real time, and if so, adopting a repair strategy;
s35: mutation, designing a mutation operator, not randomly mutating a certain gene position, but keeping the gene code with the maximum profit value and the position unchanged on the basis of a non-deadlock task on a chromosome, and randomly selecting two genes from the rest genes for position exchange;
s36: generating a group k +1, judging whether the genetic algorithm evolution algebra reaches the set maximum evolution algebra, if so, determining an optimal distribution scheme, and starting the intelligent robot to execute a target task; otherwise, go back to step S32;
s37: once the tasks are completed, the idle intelligent robot without the target tasks exists, whether the number of the completed tasks reaches the given number of the tasks in the system is judged, if yes, the optimal solution is output, otherwise, the step S31 is returned, the genetic algorithm is applied again to allocate new tasks to the idle intelligent robot, and the method is ended until all the system tasks are executed.
2. The method according to claim 1, wherein in the step S1, the method further comprises:
s11: establishing a coordinate system to model an environment, wherein a plurality of intelligent robots with task execution capacity, task points with time-varying attributes and state quantities and a plurality of static obstacles are distributed in the environment;
s12: and analyzing factors needing to be considered in task allocation and sorting out relevant constraint conditions according to the environment information.
3. The method according to claim 2, wherein in step S11, the environment is an intelligent robot work environment, N (N e Z +) task points, M (M e Z +) intelligent robots and B (B e Z +) static obstacles are distributed in the environment, and approximate coordinates of each task point, intelligent robot and obstacle are determined.
4. The method according to claim 2, wherein in step S12, according to the change characteristics of the task point state quantities, a task point state quantity model is established, and factors to be considered for task allocation, including task point characteristic parameters, intelligent robot capability parameters, and static obstacle attribute parameters, are analyzed;
the task j (j ═ 1,2, …, N) point characteristic parameters include: initial state quantity Sj(0) Growth rate alphajAnd the approximate coordinate position (x) in the environmentj,yj)；
The intelligent robot i (i ═ 1,2, …, M) capability parameters include: velocity v of movementiThe amount of tasks performed per unit time is betaiAnd the approximate coordinate position (x) of where in the environmenti,yi)；
Assuming that the state quantities of the tasks are variable, the state quantities of the task j (j ═ 1,2, …, N) are expressed by the equation:
Figure FDA0003418780160000021
in the formula:
Figure FDA0003418780160000022
is the amount of state change, alpha, of task j per unit timejThe state growth rate of the task point j;
the set of intelligent robots working on task j is lambdajWithout external intervention, the state quantity of task j (j ═ 1,2, …, N) is expressed by the equation:
Figure FDA0003418780160000023
from the perspective of a single intelligent robot, each intelligent robot needs to avoid collision with other intelligent robots and also avoid collision with an obstacle, and the collision condition of the intelligent robot with the obstacle is described by a function H1:
Figure FDA0003418780160000024
in the formula: zi(t) represents the position state of the intelligent robot i at time t, ObIndicating a disorderLocation area of object, H1(Zi(t),Zb) 1 represents that the intelligent robot i collides with at least one obstacle;
function H for the same reason2The situation of collision between intelligent robots is described as follows:
Figure FDA0003418780160000025
in the formula: dsafeIs the safe distance between intelligent robots;
therefore, the constraint conditions of the intelligent robot in the forward target task process are as follows:
Figure FDA0003418780160000026
5. the method according to claim 1, wherein in the step S2, the method further comprises:
in order to distinguish the importance of different target tasks at time t, each task has a corresponding revenue function Ijφj(t) in which IjIs the maximum benefit that can be gained to complete task j when
Figure FDA0003418780160000027
dφj(t)/dt > 0, wherein
Figure FDA0003418780160000028
As a function of phij(t) a cutoff time increasing with time t, where φj(t) is a "price-rising function" of the revenue obtained by completing task j over time; when in use
Figure FDA0003418780160000029
d phi j (t)/dt is less than 0, wherein
Figure FDA00034187801600000210
As a function of phij(t) a decreasing cut-off time as the time t increases, in which casej(t) is a "discounting function" of the revenue obtained by completing task j over time; when in use
Figure FDA00034187801600000211
φj(t) ═ a, where a is a set constant value; the goal of the overall system task is to make the final benefits obtained
Figure FDA00034187801600000212
To a maximum, i.e.
Figure FDA00034187801600000213
Wherein
Figure FDA00034187801600000214
The time required to complete task j;
setting considering that the intelligent robot reaching the target task needs to spend time on the road
Figure FDA00034187801600000215
Time to get maximum benefit to complete task j, where
Figure FDA0003418780160000031
Suppose that
Figure FDA0003418780160000032
For the time consumed on the road by the intelligent robot which reaches the target task j first, the price rising function can be set to
Figure FDA0003418780160000033
A discount function is set as
Figure FDA0003418780160000034
6. The method as claimed in claim 1, wherein in step S32, when each task is executed, the earnings obtained need to be reasonably allocated to the intelligent robot executing the task, and the allocation is performed according to the amount of labor; assuming that the ith intelligent robot executes the task j at the time t, the income obtainable by the ith intelligent robot at the time t is expressed by the following formula:
Figure FDA0003418780160000035
in the formula, xij(t) epsilon {0,1} represents whether the ith intelligent robot executes the task j at the time t, if yes, the task j is 1, otherwise, the task j is 0, wherein
Figure FDA0003418780160000036
Figure FDA0003418780160000037
Represents the sum of the intelligent robot's performance capabilities to perform task j at time t,
Figure FDA0003418780160000038
representing the revenue obtained by completing task j;
the total income obtained by the intelligent robot i after the task j is completed is Iij, and the total income is expressed by a formula as follows:
Figure FDA0003418780160000039
after the task of the whole system is completed, the total income obtained by the ith intelligent robot is expressed by a formula as follows:
Figure FDA00034187801600000310
in the formula: i isiIndicates the income, x, available to the intelligent robot i after completing all tasksijE {0,1} represents whether the intelligent robot i executes the task j, if yes, the task j is 1, and if not, the task j is 0; the fitness value may be calculated from a fitness function, and thus the fitness function may be chosen as:
Figure FDA00034187801600000311
in the formula: i is the total gain obtained by all robots after completing all tasks.
7. The method of claim 1, wherein in step S33, a new selection operator is designed, a difference assessment selection is performed, the number of candidate individuals before each crossover is greater than the number S of final selected chromosomes, a part of individuals are first assessed and selected, then the number S of crossover individuals satisfying the requirement is formally selected from the preselected individuals, and the eliminated individuals enter the next candidate list, and the specific steps are as follows:
s81: pre-selecting, wherein the number C of candidate individuals is assessed in each evolution, the assessment index is the size of a fitness value, H (H & gt S) individuals with the highest ranking of the fitness value are selected as a candidate list, and the H individuals are ranked from large to small according to the fitness value;
s82: formally selecting, firstly removing individuals coded with similar gene values of No. 1 individuals by using the individuals as templates, removing individuals with similar coding genes by using individuals with high fitness values as templates successively, selecting S chromosomes of different templates to form a population P (k), and removing the number of the individuals missing from the population according to the fitness value size sequence if a population specification S meeting the requirement cannot be obtained, wherein k is 0 initially, and the similarity formula adopts the following steps: cii′＝kii′N (i, i ' is epsilon (i, i ' is less than or equal to H) ^ i ≠ i '), wherein: k is a radical ofii′The number of common gene values at the corresponding positions of the chromosome i and the chromosome i', and n is the length of a chromosome coding string; setting a threshold value mu of a selection operator when Cii′Mu is larger than or equal to mu, which means that the chromosome i is similar to the chromosome i', otherwise, the chromosome i is not similar.
8. Root of herbaceous plantThe method of claim 1, wherein in step S34, when it is detected that the genetic code on the chromosome is involved in the global deadlock, it indicates that the intelligent robot tasks are improperly allocated after the genetic code on the chromosome is encoded, and the sum of the execution capacity of the intelligent robot allocated to each task is less than the task growth speed αjTherefore, the repair strategy is adopted, and specifically includes:
s91, counting codes corresponding to the intelligent robot on the chromosome once the chromosome is detected to have global deadlock;
s92, calculating the execution capacity of the intelligent robot at each task point and the growth speed ratio P of the intelligent robot to the taskj(j < N > 0) and sorting
Figure FDA0003418780160000041
S93, mixing
Figure FDA0003418780160000042
Assignment of the corresponding Intelligent robot with minimal on-task execution capability
Figure FDA0003418780160000043
Corresponding task if
Figure FDA0003418780160000044
The corresponding task is not yet deadlocked, and continues to be executed
Figure FDA0003418780160000045
Corresponding intelligent robot with minimum current execution capacity on task to be allocated
Figure FDA0003418780160000046
…, up to
Figure FDA0003418780160000047
The task of (1) removes the deadlock.
CN201711394738.0A 2017-12-21 2017-12-21 Dynamic task-oriented multi-intelligent-robot task allocation method Active CN108416488B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201711394738.0A CN108416488B (en) 2017-12-21 2017-12-21 Dynamic task-oriented multi-intelligent-robot task allocation method

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201711394738.0A CN108416488B (en) 2017-12-21 2017-12-21 Dynamic task-oriented multi-intelligent-robot task allocation method

Publications (2)

Publication Number Publication Date
CN108416488A CN108416488A (en) 2018-08-17
CN108416488B true CN108416488B (en) 2022-05-03

Family

ID=63125617

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201711394738.0A Active CN108416488B (en) 2017-12-21 2017-12-21 Dynamic task-oriented multi-intelligent-robot task allocation method

Country Status (1)

Country Link
CN (1) CN108416488B (en)

Families Citing this family (14)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN109460294A (en) * 2018-10-13 2019-03-12 中国人民解放军战略支援部队信息工程大学 A kind of reconfigurable task layout method based on area-efficient and dynamic genetic algorithm
US10867263B2 (en) * 2018-12-04 2020-12-15 Afiniti, Ltd. Techniques for behavioral pairing in a multistage task assignment system
CN109886574B (en) * 2019-02-20 2023-02-14 哈尔滨工程大学 Multi-robot task allocation method based on improved threshold method
CN110046795B (en) * 2019-03-01 2021-11-05 斯坦德机器人（深圳）有限公司 Task allocation method and device for robot
CN110084512B (en) * 2019-04-26 2022-04-22 河海大学常州校区 Multi-robot task allocation method for intelligent warehousing system
CN109986563B (en) * 2019-05-01 2022-03-15 湖南大学 Multi-mobile-robot cooperative operation method and system
CN112181608B (en) * 2019-07-03 2023-10-31 中南大学 Distributed distribution method for multipoint dynamic aggregation tasks based on local information
CN110501905B (en) * 2019-08-27 2022-02-08 中国人民解放军国防科技大学 Multi-agent system self-adaptive method and system based on packing model
CN111027875B (en) * 2019-12-17 2023-09-26 鲁东大学 Intelligent warehouse multi-robot task allocation method based on self-adaptive task pool
CN113050549B (en) * 2019-12-26 2022-06-10 炬星科技（深圳）有限公司 Method, equipment and storage medium for executing task by robot capable of freely networking
CN111325454A (en) * 2020-02-11 2020-06-23 广州地铁设计研究院股份有限公司 Designer role control method
CN114137955B (en) * 2021-10-26 2023-04-28 中国人民解放军军事科学院国防科技创新研究院 Multi-robot rapid collaborative mapping method based on improved market method
CN114330978B (en) * 2021-11-11 2022-08-09 深圳大学 Air-ground robot task dynamic allocation method, storage medium and terminal equipment
CN114924513B (en) * 2022-06-07 2023-06-06 中迪机器人(盐城)有限公司 Multi-robot cooperative control system and method

Family Cites Families (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US8874477B2 (en) * 2005-10-04 2014-10-28 Steven Mark Hoffberg Multifactorial optimization system and method
US9446512B1 (en) * 2014-03-26 2016-09-20 Stc.Unm iAnt swarm robotic platform and evolutionary algorithms
CN105069530B (en) * 2015-08-13 2019-02-26 肇庆学院 A kind of multi-robot Task Allocation based on multiple-objection optimization
CN106020230B (en) * 2016-05-20 2019-01-15 武汉科技大学 A kind of multiple no-manned plane method for allocating tasks under power consumption constraint
CN106654987B (en) * 2016-11-18 2018-06-19 华北电力大学（保定） Power circuit multi-robot Cooperation method for inspecting
CN106875090B (en) * 2017-01-09 2020-05-08 中南大学 Dynamic task-oriented multi-robot distributed task allocation forming method
CN107168054B (en) * 2017-05-10 2020-11-10 沈阳工业大学 Multi-robot task allocation and path planning method

Also Published As

Publication number Publication date
CN108416488A (en) 2018-08-17

Similar Documents

Publication Publication Date Title
CN108416488B (en) Dynamic task-oriented multi-intelligent-robot task allocation method
CN110632907B (en) Scheduling optimization method and system for distributed assembly type replacement flow shop
Burnwal et al. Scheduling optimization of flexible manufacturing system using cuckoo search-based approach
CN107895225B (en) Multi-Agent conflict-free cooperative task allocation method
CN104616062B (en) A kind of Nonlinear System Identification planned based on multi-objective Genetic
CN113792924A (en) Single-piece job shop scheduling method based on Deep reinforcement learning of Deep Q-network
CN105005301A (en) Method for planning operation point sequence and path of industrial robot based on swarm intelligence algorithm
CN109472362B (en) AGV dynamic scheduling method and device based on variable task window
CN103942610A (en) Reconfigurable manufacturing system polymorphic configuration optimization method based on tasks
CN113139710B (en) Multi-resource parallel task advanced plan scheduling method based on genetic algorithm
CN107728639B (en) Heterogeneous multi-AUV system task coordination method under time window constraint
Rodzin et al. Theory of bioinspired search for optimal solutions and its application for the processing of problem-oriented knowledge
CN111915011A (en) Single-amplitude quantum computation simulation method
CN108664021A (en) Robot path planning method based on genetic algorithm and quintic algebra curve interpolation
Baizid et al. Genetic algorithms based method for time optimization in robotized site
CN104281917A (en) Fuzzy job-shop scheduling method based on self-adaption inheritance and clonal selection algorithm
CN113848904A (en) Method for optimizing task allocation of multiple mobile robots based on punished energy consumption
Mutingi et al. Integrated cellular manufacturing system design and layout using group genetic algorithms
CN116880424A (en) Multi-robot scheduling method and device based on multi-objective optimization
CN112699544A (en) Multi-target flexible job shop scheduling method
CN112783210B (en) Multi-target control parameter optimization method of unmanned aerial vehicle cluster control system
CN103246923A (en) Self-adaptive genetic algorithm-based flexible workshop dispatching technology
CN103383752A (en) Assembly scheduling method of aircraft
CN111489035A (en) Mold combination distribution method based on combination of machine learning and backtracking method
CN112632777A (en) II-type bilateral assembly line balancing method and system for household appliance product assembly line

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