CN101618543A - Task allocation method of heterogeneous multi-robot system - Google Patents

Abstract

The invention belongs to the field of automatic control, in particular relates to a task allocation method of a heterogeneous multi-robot system, which comprises the following steps: S1, initializing robots and subtask parameters; S2, executing subtasks with high information concentration near the robots and close to self-capacity and subtask ability requirement by the robots, wherein the robots are skilled in the subtasks. The invention fully considers the influence of a functional structure of the multi-robot system to the task allocation of the multi-robot system, and realizes the task allocation of the heterogeneous multi-robot system.

Description

The method for allocating tasks of heterogeneous multi-robot system

Technical field

The invention belongs to automation field, be specifically related to the method for allocating tasks of heterogeneous multi-robot system.

Background technology

The variation of making rapid progress has taken place as one of human greatest invention of 20th century in robot in short more than 40 years.Robot had become representative strategic objective in the high-tech sector in recent years.The appearance of Robotics and development not only make traditional industrial production looks generation essence change, and the mankind's society is produced far-reaching influence.Along with the develop rapidly of social production technology, the application of robot is constantly expanded.Exploration from the automatic production line to marine resources, and even field such as space operation, robot is ubiquitous.Yet with regard to present Robotics level, single robot all is limited at aspects such as the obtaining of information, processing and control abilities, and for the task of complexity and changeable working environment, the ability of single robot is more inadequate.Multi-robot system provides solution route with its exclusive superiority for this problem.

The research of multi-robot system starts from the seventies in 20th century.From then on, domestic and international many colleges and universities and scientific research institution have carried out extensive studies to multi-robot system.Along with multi-robot system is subjected to the increase of attention degree, distribute as the task of a multi-robot coordination part also more and more to cause people's attention.Although it is the part that multi-robot system is coordinated that task is distributed, but because task is distributed the high level that occupy in the multi-robot system, in the process of research, can ignore the bottom details of system, thus the researcher usually task is distributed as one independently subproblem study.Multi-robot system task distribution has at present obtained certain achievement in research, is a NP difficult problem yet find the solution multi-robot system task allocative decision, and the research that multimachine device system task is distributed still is in initial stage, has got long long way to go.The multi-robot system task is assigned one of problem to be solved at present: consider the influence of the functional structure of robot system to the distribution of multi-robot system task.Robot system functional structure difference is also different to the requirement that task is distributed, and robot capability differs in the isomery robot system, need consider when allocating task what ability is different tasks respectively need, and needs which robot to cooperate and finishes.Because the isomery robot system has generality, therefore must consider the influence that the robot system functional structure is distributed the multi-robot system task, and design corresponding task allocation algorithms on this basis, give full play to the ability of robot.

Summary of the invention

In view of this,, the present invention proposes a kind of method for allocating tasks of heterogeneous multi-robot system, adopt improved ant group algorithm, coordinate the execution of heterogeneous multi-robot at least one subtask in order to address the above problem.

The object of the present invention is achieved like this: the method for allocating tasks of heterogeneous multi-robot system comprises the steps:

S1: initialization robot and subtask parameter;

S2: robot selects oneself to be good at, self-ability with the subtask ability need is approaching, from own near, subtask execution that information concentration is high;

Further, among the described step S1, robot and subtask parameter comprise that device people's ability disposes vectorial C _i, the subtask the ability need vector M _j, the capabilities match parameter K _Ij, the horizontal parameter L of robot executive capability _Ij, the quantity upper limit N of the required robot in subtask _jThe quantity A of robot with current subtasking _j

Further, among the described step S2, the i of robot specifically carries out following steps;

S21: upgrade oneself state s _i, check the ability configuration of self;

S22: the state and the ability configuration of obtaining the other machines people;

S23: the pheromone concentration that obtains all subtasks;

S24: the selection probability that calculates the subtask;

S25: select the highest subtask of probability to carry out;

Further, step S24 specifically comprises the steps:

Robot quantity also many more, therefore keep robot quantity to be directly proportional with the subtask workload, can improve the average efficiency of multi-robot system; 3) continuity of assurance robot subtasking: a lot of robots are when subtasking, may receive " invitation " of new subtask, among the present invention, have only after robot finishes the work, go again to carry out new subtask, thereby guarantee the performance level maximum of overall task.

Other advantages of the present invention, target, to set forth in the following description to a certain extent with feature, and to a certain extent,, perhaps can obtain instruction from the practice of the present invention based on being conspicuous to those skilled in the art to investigating hereinafter.Target of the present invention and other advantages can be passed through following specification, claims, and the specifically noted structure realizes and obtains in the accompanying drawing.

Description of drawings

In order to make the purpose, technical solutions and advantages of the present invention clearer, the present invention is described in further detail below in conjunction with accompanying drawing:

Fig. 1 shows the method for allocating tasks schematic flow sheet of heterogeneous multi-robot system;

Fig. 2 shows the schematic flow sheet that subtasking is selected by robot;

Fig. 3 shows the embodiment task and distributes initial time figure;

Fig. 4 shows the embodiment task and distributes t ₁Time chart;

Fig. 5 shows the embodiment task and distributes t ₂Time chart;

Fig. 6 shows the embodiment task and distributes t ₃Time chart;

Fig. 7 shows the embodiment task and distributes t ₄Time chart;

Fig. 8 shows the embodiment task and distributes t ₅Time chart;

Fig. 9 shows the embodiment task and distributes t ₆Time chart;

Figure 10 shows the embodiment task and distributes t ₇Time chart;

Figure 11 shows the embodiment task and distributes t ₈Time chart;

Figure 12 shows the embodiment task and distributes t ₉Time chart.

The specific embodiment

Below will be described in detail the preferred embodiments of the present invention.

At first to following parameter declaration:

τ _j: the pheromones on the j of subtask;

η _Ij=1/d _Ij: the inverse of distance between i of robot and the subtask j;

C _i: robot capability configuration vector refers to the function that the i of robot itself is possessed;

M _j: the ability need vector of subtask refers to the needed function of subtasking j;

K _Ij: the capabilities match parameter, whether the expression i of robot is fit to carry out a certain subtask j;

L _Ij: the i of robot is to the horizontal parameter of the executive capability of subtask j;

p _Ij: the i of robot is to the selection probability variable of subtask j;

A _j: the robot quantity of current subtasking j;

N _j: the required robot of the subtask j quantity upper limit;

T _j: subtask j total amount is constant;

I _i: the i of the robot amount of executing the task, this parameter are constant;

R _j: the quantity upper limit sign of the required robot in subtask;

F _j: the subtask complement mark;

Task _Ij: switching variable.

Be located in the working region of task the unknown, the set that contains n robot is R={r ₁, r ₂, r ₃..., r _nAnd the set of tasks that contains k subtask be T={t ₁, t ₂..., t _k.N robot at random search mission in the working region, in the time of initial corresponding to all task j (j=1,2 ...) and " hormone " τ _jBe τ (τ is a constant) that the state set of n robot is S={s ₁, s ₂..., s _n, 0 is idle condition, 1 is duty.

Robot possesses multiple different function, as perception (vision, sense of smell, the sense of hearing, infrared, ultrasonic etc.), executive capability (move, carrying, processing are handled etc.), operational capability, communication capacity etc., and, also may need multiple different ability to finish for a task.These Capability Categories are refined into single ability, and these all single abilities are formed set.

If by k single ability c _jThe competence set of forming is:

S241 compute switch variable task _Ij,

task _ij＝F _j·R _j·s _i·K _ij；

Wherein, F _jBe subtask complement mark, s _iThe state of the expression i of robot, K _IjBe the i of robot capabilities match parameter, R _jBe the quantity upper limit sign of the required robot of subtask j, its expression formula is:

$R_j=\left\{\begin{array}{cc}0,& A_j={\mathrm{<figure-callout\; id="1"\; label="N\; j"\; filenames="A2009101044200015H3.png,A2009101044200015H4.png"\; state="\{\{state\}\}">N</figure-callout>}}_j\\ 1,& A_j<N_j\end{array}\right.;$

S242: according to the switching variable task of step S21 gained _Ij, calculate the selection probability of subtask j by following formula:

p _ij＝p′ _ij·task _ij；

${p^{\&prime;}}_{\mathrm{ij}}=\frac{{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}{\underset{j}{\mathrm{\&Sigma;}}{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}\&CenterDot;(1-\mathrm{\&omega;})+\mathrm{\&omega;}\&CenterDot;L_{\mathrm{ij}};$

In the formula, τ _jBe the pheromones on the j of subtask, η _IjBe the inverse of distance between i of robot and the subtask j, L _IjFor the i of robot to the horizontal parameter of subtask j executive capability, ω is the constant between 0 to 1;

Further, also comprise the steps: after the step S2

S3: judge whether subtask j finishes, as not, the plain τ of lastest imformation then _j, and redirect execution in step S3; In this way, execution in step S4 then;

S4: with robotary s _iWith pheromones τ _jMake zero.

The method for allocating tasks of the heterogeneous multi-robot system that the present invention proposes takes into full account the influence of the functional structure of robot system to the distribution of multi-robot system task, has realized that the task of heterogeneous multi-robot system is distributed; Specifically comprise following advantage: can realize that 1) optimum robot shines upon to the subtask: the subtask in overall task to robot capability require variant, the ability configuration of each robot is also variant, therefore allow each robot select the subtask of mating most, can improve systematic function with self-ability; 2) the robot quantity that guarantees the chooser task is directly proportional with the subtask workload: because working environment space, place, subtask is limited, the robot quantity in the space more at most time of the mutual collision prevention consumption of robot many more, the average efficiency of robot is low more; Simultaneously, the subtask workload is big more, and then working space is relatively also big more, can hold

C＝{c _j}，1≤j≤k

1) ability of robot configuration vector

Robot capability configuration vector is meant the function that robot itself is possessed.

Be provided with n the r of robot with difference in functionality _i, 1≤i≤n.For the r of robot _i, define its ability and dispose vectorial C _iFor:

C _i＝diag{α _i1，α _i2，…，α _ik}·[c ₁，c ₂，…，c _k] ^T (1)

Wherein, α _IkThe expression r of robot _iWhether has ability c _k, if robot possesses ability c _k, α _Ik=1; If robot does not possess ability c _k, α then _Ik=0.

2) the ability need vector M of subtask _j

The ability need vector of subtask is meant the needed function of subtasking.

Be provided with m task t _j, 1≤j≤m.For task t _j, define its ability need vector and be:

M _j＝diag{β _j1，β _j2，…，β _jk}·[c ₁，c ₂，…，c _k] ^T (2)

β _JkThe correspondence t that finishes the work _jWhether need ability c _k, if finish the work t _jNeed ability c _k, β _Jk=1, do not need this ability, then β if finish the work _Jk=0.

3) capabilities match parameter K _Ij

Match parameter K _IjBe the ability whether the expression i of robot is fit to carry out a certain task j, set up for realizing the 1st principle.

Whether robot is fit to carry out a certain task, disposes vectorial C with robot capability _iAnd the ability need vector M of subtask _jRelevant, if C _i〉=M _j, the r of robot _iBe fit to subtasking j; Otherwise, the r of robot _iTo be not suitable for subtasking j.If K _IjBe the capabilities match parameter, its expression formula is:

$K_{\mathrm{ij}}=\left\{\begin{array}{ccc}1& \mathrm{if}& C_i\&GreaterEqual;{\mathrm{<figure-callout\; id="0"\; label="M\; j"\; filenames="A2009101044200014H1.png"\; state="\{\{state\}\}">M</figure-callout>}}_j\\ 0& \mathrm{if}& C_i<M_j\end{array}\right.---\left(3\right)$

4) the horizontal parameter L of robot executive capability _Ij

This parameter is a variable, sets up for realizing the 2nd principle, and its span is between 0-1, and the ability of robot and the ability need of subtask are approaching more, then L _IjBig more.Its expression formula is:

$L_{\mathrm{ij}}=\left\{\begin{array}{cc}0& C_i<M_j\\ \frac{M_j}{C_i}& C_i\&GreaterEqual;M_j\end{array}\right.---\left(4\right)$

5) probability variable p is selected in the subtask _Ij

This parameter is set up for realizing the 3rd, 4 principle.Its expression formula is:

$p_{\mathrm{ij}}=\frac{{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}{\underset{j}{\mathrm{\&Sigma;}}{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}---\left(5\right)$

Utilize the horizontal parameter L of robot executive capability _IjTo p _IjHandle:

p′ _ij＝(1-ω)·p _ij+ω·L _ij (6)

6) the quantity upper limit N of the required robot in subtask _j, this parameter is a constant, its value is according to the needs decision of subtask.

7) the quantity A of robot of current subtasking _j, this parameter is a variable

8) the quantity upper limit sign R of the required robot in subtask _j, its expression formula is:

$R_j=\left\{\begin{array}{cc}0,& A_j={\mathrm{<figure-callout\; id="1"\; label="N\; j"\; filenames="A2009101044200015H3.png,A2009101044200015H4.png"\; state="\{\{state\}\}">N</figure-callout>}}_j\\ 1,& A_j<N_j\end{array}\right.---\left(7\right)$

9) subtask complement mark F _j

10) switching variable task _Ij

task _ij＝F _j·R _j·s _i·K _ij (9)

Wherein, s _iThe state of the expression i of robot, its expression formula is:

Utilize task _IjTo p ' _IjHandle, select probability variable to be write as again the subtask:

p _ij＝p′ _ij·task _ij (11)

Probability variable p selects by the subtask in robot _IjThe chooser task, if the subtask remains unfulfilled,, press following formula:

τ _j＝τ+Δτ＞0 (12)

Upgrade the pheromone concentration of subtask, assist to obtain more robot.

Robot satisfies the functional requirement of subtask, and the robot quantity upper limit of subtask does not reach again, and the state of robot is free time, then task _Ij=1, switch opens, robot can the chooser task; Otherwise switch cuts out, and robot can not the chooser task.

Referring to Fig. 1, the method for allocating tasks of the heterogeneous multi-robot system of present embodiment comprises the steps:

S1: initialization robot and subtask parameter comprise that device people's ability disposes vectorial C _i, the subtask the ability need vector M _j, the capabilities match parameter K _Ij, the horizontal parameter L of robot executive capability _Ij, the quantity upper limit N of the required robot in subtask _jThe quantity A of robot with current subtasking _j

S2: robot selects oneself to be good at, self-ability with the subtask ability need is approaching, from own near, subtask execution that information concentration is high; The i of robot specifically carries out following steps;

S21: upgrade oneself state s _i, check the ability configuration of self;

S22: the state and the ability configuration of obtaining the other machines people;

S23: the pheromone concentration that obtains all subtasks;

S24: the selection probability that calculates the subtask; Specifically comprise the steps:

S241: compute switch variable task _Ij,

task _ij＝F _j·R _j·s _i·K _ij；

Wherein, F _jBe subtask complement mark, s _iThe state of the expression i of robot, K _IjBe the i of robot capabilities match parameter, R _jBe the quantity upper limit sign of the required robot of subtask j, its expression formula is:

$R_j=\left\{\begin{array}{cc}0,& A_j={\mathrm{<figure-callout\; id="1"\; label="N\; j"\; filenames="A2009101044200015H3.png,A2009101044200015H4.png"\; state="\{\{state\}\}">N</figure-callout>}}_j\\ 1,& A_j<N_j\end{array}\right.;$

S242: according to the switching variable task of step S21 gained _Ij, calculate the selection probability of subtask j by following formula:

p _ij＝p′ _ij·task _ij；

${p^{\&prime;}}_{\mathrm{ij}}=\frac{{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}{\underset{j}{\mathrm{\&Sigma;}}{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}\&CenterDot;(1-\mathrm{\&omega;})+\mathrm{\&omega;}\&CenterDot;L_{\mathrm{ij}};$ (ω is the constant between 0 to 1)

In the formula, τ _jBe the pheromones on the j of subtask, η _IjBe the inverse of distance between i of robot and the subtask j, L _IjFor the i of robot to the horizontal parameter of subtask j executive capability.

S25: select the highest subtask of probability to carry out;

S3: judge whether subtask j finishes, as not, the plain τ of lastest imformation then _j, and redirect execution in step S2; In this way, execution in step S4 then;

S4: with robotary s _iWith pheromones τ _jMake zero.

Referring to Fig. 2 to 11, be that example is tested with the method for present embodiment with the background that clears land mines, establishing working space is 10 meters * 10 meters, among the figure, counterclockwise to the upper left corner, what occur successively is robot, r by the lower left corner ₁, r ₂, r ₃, r ₄The value of each subtask parameter: t ₁: M ₁=1, N ₁=2, T ₁=1; t ₂: M ₂=2, N ₂=5, T ₂=4; t ₃: M ₃=1, N ₃=3, T ₃=3.

The parameter value of each robot: r ₁: C ₁=2, I ₁=1.5; r ₂: C ₂=2, I ₁=1.5; r ₃: C ₃=1, I ₁=1.5; r ₄: C ₄=3, I ₁=1.5.Adopt " blackboard " mode to communicate between the robot, α=1, β=1, initial time, the pheromones on each subtask is: τ=1.

Initial time, r ₁, r ₂, r ₄Search task t ₁, t ₂, t ₃, advance to impact point with the speed of 1m/s and to execute the task.As shown in Figure 3;

The t1 moment: r ₄T executes the task after the arrival task ₃, to attempt independently finishing the work, the discovery task can not independently be finished by it, increases task t ₃Pheromones τ ₃, attract the other machines people to cooperate to execute the task, as shown in Figure 4.

T2 is in the r of idle condition constantly ₃Be subjected to task t ₃On the attraction of pheromones, r ₃To t ₃Direction walking precedingly removes to assist r ₄Carry out t ₃, as shown in Figure 5.

The t3 moment: r ₁Arrive the t that executes the task of task place behind the cut-through thing ₁, attempt independently finishing the work t ₁, r ₁T can accomplish a task all by oneself ₁, after finishing the work, r ₁Be in idle condition, r ₁At task t ₁The place continues to search for other task, as shown in Figure 6.

The t4 moment: r ₂Arrive t ₂, attempt independently finishing the work, can not independently finish, increase pheromones τ ₂, attract the other machines people who is in idle condition to assist, as shown in Figure 7.

T5 is in the r of robot of idle condition constantly ₁Be subjected to " attraction ", assist r ₂Carry out t ₂, as shown in Figure 8.

The t6 moment: r ₃Arrival task T3 is with r ₄Cooperate together to execute the task t ₂The task complexity be that t is finished in 3, two robot cooperations ₂, with task t ₂Telergone τ ₂Zero clearing, r ₃, r ₄Be in idle condition, as shown in Figure 9.

The t7 moment: r ₁Arrive t ₂, r ₂, r ₁Attempt cooperation and finish the work, task still can not be done, so continue to increase t ₂Pheromones τ ₂, attract the other machines people who is in idle condition to assist, as shown in figure 10.

T8 is in the r of robot of idle condition constantly ₄Be subjected to " attraction ", assist r ₂, r ₁Carry out t ₂, as shown in figure 11.

The t9 moment: r ₂, r ₁, r ₄Three robot cooperations t that finishes the work ₂, as shown in figure 12.

From Fig. 3 to 12 as can be seen this this method effectively solved the task assignment problem of heterogeneous multi-robot system.

The above is the preferred embodiments of the present invention only, is not limited to the present invention, and obviously, those skilled in the art can carry out various changes and modification and not break away from the spirit and scope of the present invention the present invention.Like this, if of the present invention these are revised and modification belongs within the scope of claim of the present invention and equivalent technologies thereof, then the present invention also is intended to comprise these changes and modification interior.

Claims (5)

1. the method for allocating tasks of heterogeneous multi-robot system in order to coordinate the execution of heterogeneous multi-robot at least one subtask, comprises the steps:

S1: initialization robot and subtask parameter;

S2: robot selects oneself to be good at, self-ability with the subtask ability need is approaching, from own near, subtask execution that information concentration is high.

2. the method for allocating tasks of heterogeneous multi-robot system as claimed in claim 1, it is characterized in that: among the described step S1, robot and subtask parameter comprise that the ability of robot disposes vectorial C _i, the subtask the ability need vector M _j, the capabilities match parameter K _Ij, the horizontal parameter L of robot executive capability _Ij, the quantity upper limit N of the required robot in subtask _jThe quantity A of robot with current subtasking _j

3. the method for allocating tasks of heterogeneous multi-robot system as claimed in claim 1 or 2, it is characterized in that: among the described step S2, the i of robot specifically carries out following steps;

S21: upgrade oneself state s _i, check the ability configuration of self;

S22: the state and the ability configuration of obtaining the other machines people;

S23: the pheromone concentration that obtains all subtasks;

S24: the selection probability that calculates the subtask;

S25: select the highest subtask of probability to carry out;

4. the method for allocating tasks of heterogeneous multi-robot system as claimed in claim 3, it is characterized in that: step S24 specifically comprises the steps:

S241 compute switch variable task _Ij,

task _ij＝F _j·R _j·s _i·K _ij；

Wherein, F _jBe subtask complement mark, s _iThe state of the expression i of robot, K _IjBe the i of robot capabilities match parameter, R _jBe the quantity upper limit sign of the required robot of subtask j, its expression formula is:

$R_j=\left\{\begin{array}{cc}0,& A_j=N_j\\ 1,& A_j<N_j\end{array}\right.;$

S242: according to the switching variable task of step S21 gained _Ij, calculate the selection probability of subtask j by following formula:

p _ij＝p′ _ij·task _ij；

${p^{\&prime;}}_{\mathrm{ij}}=\frac{{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}{\underset{j}{\mathrm{\&Sigma;}}{\left({\mathrm{\&tau;}}_j\right)}^{\mathrm{\&alpha;}}{{\mathrm{\&eta;}}_{\mathrm{ij}}}^{\mathrm{\&beta;}}}\&CenterDot;(1-\mathrm{\&omega;})+\mathrm{\&omega;}\&CenterDot;L_{\mathrm{ij}};$

In the formula, τ _jBe the pheromones on the j of subtask, η _IjBe the inverse of distance between i of robot and the subtask j, L _IjFor the i of robot to the horizontal parameter of subtask j executive capability, ω is the constant between 0 to 1.

5. the method for allocating tasks of heterogeneous multi-robot system as claimed in claim 4 is characterized in that: also comprise the steps: after the step S2

S3: judge whether subtask j finishes, as not, the plain τ of lastest imformation then _j, and redirect execution in step S2; In this way, execution in step S4 then;

S4: with robotary s _iWith pheromones τ _jMake zero.

Images