CN102901498B - Method for cooperative search and dynamic task allocation of unmanned aerial vehicle teams under uncertain environment - Google Patents

Abstract

The invention relates to a method for cooperative search and dynamic task allocation of unmanned aerial vehicle teams under an uncertain environment, and not only puts forward a cooperative dynamic task allocation strategy of unmanned aerial vehicle teams, but also designs a concrete guidance law. By adopting an arc air route as the air route by which unmanned aerial vehicles performing tasks, the method includes: step 1. determining the data structure of single unmanned aerial vehicle maintenance; step 2. determining the flight mode of unmanned aerial vehicles; step 3. determining the dominant function of task performing of the unmanned aerial vehicles; step 4. determining a dynamic task allocation process; and step 5. determining the guidance law of aerial vehicle search and task performing. Compared with modern optimization algorithm based task allocation methods, the method provided in the invention reduces the computation load of single unmanned aerial vehicle, and is suitable for conditions characterized by strong real-time performance and uncertain environment. Compared with market mechanism auction algorithm based task allocation methods, the method reduces the times of communication performed among unmanned aerial vehicles and the computation load of single unmanned aerial vehicle, and guarantees the real-time performance of broadcast unmanned aerial vehicles during task performing.

Description

Unmanned vehicle formation collaboratively searching under a kind of uncertain environment and dynamic task allocation method

Technical field

The unmanned vehicle that the present invention relates under a kind of uncertain environment is formed into columns and is worked in coordination with dynamic task allocation strategy, belong to flight formation Guidance and control technical field, be specifically related to the unmanned vehicle formation collaboratively searching under a kind of uncertain environment and dynamic task allocation method.

Background technology

Current had nearly more than 30 countries to drop into research and the production that a large amount of manpower and financial resources is engaged in unmanned plane.Through vicennial development, this technology comparative maturity, play in the army and the people every field and act on, however, single rack unmanned plane also exists some problems when carrying out task, and such as single rack unmanned plane may be subject to the restricted number of sensor, can not observing target area from Multi-angle omnibearing, when facing wide area search task, can not effectively cover whole region of search; If what perform is rescue task, single rack unmanned plane is restricted in load, often affects the usefulness of whole rescue, brings larger loss, in addition, once single rack unmanned plane breaks down, must interrupt task return immediately, may incur loss through delay rescue opportunity.

In recent years due to the complicacy of environment and the multiplicity of task, single rack unmanned plane cannot meet the needs of air-to-ground attack and rescue, for the above-mentioned shortcoming of single rack unmanned plane, abroad proposed in recent years multiple no-manned plane jointly carry out task concept and achieve certain achievement in research, jointly the carry out mode of task of multiple no-manned plane can significantly improve operation and the rescue usefulness of unmanned plane entirety, the ability of success ratio and the anti-accident of executing the task can be increased, based on above-mentioned advantage, the multiple no-manned plane aerial mission that jointly carries out will become an important directions of Development of UAV from now on.Wherein multiple no-manned plane works in coordination with dynamic task allocation problem is the key issue that multiple no-manned plane carries out in task jointly, this is a multiple constraint, the Complex multi-target optimization of strong coupling and decision problem in essence, be specifically related to the research fields such as unmanned plane formation communication protocol, collaboratively searching strategy, dynamic task allocation method and unmanned plane kinematics link Guidance Law Design, the quality of these designs directly affects the effect that multiple no-manned plane works in coordination with dynamic task allocation.

The concept that multiple no-manned plane works in coordination with dynamic task allocation is: carry out in task process at multiple no-manned plane, based on certain environmental knowledge and mission requirements, according to quantity or the load of unmanned plane, sequence task (target or locus point) is had for each unmanned plane distributes one or one group, to complete the task of most probable number simultaneously, the efficiency of aircraft entirety is reaching optimum.The research that multiple no-manned plane works in coordination with dynamic task allocation method grew up in recent years, more influential paper is " the Real-Time Planning for Teams ofAutonomous Vehicles in Dynamic Uncertain Environment " of the Pong Punwattana of University of Washington in 2004, and the development through nearly ten years defines following two kinds of main solutions:

Modern optimization algorithm method for allocating tasks: first the method carries out the modeling of Task Allocation Problem, take out the fitness function relevant to task matching effect, then a colony is generated, each individuality in colony represents a kind of task matching scheme, utilizes the optimum solution that in didactic decision search colony, fitness function is the highest.This method utilizes the mode of colony's rolling optimizing to search for the optimal case that multiple no-manned plane works in coordination with dynamic task allocation, and computing time is longer, is not suitable for the comparatively strong and uncertain condition of environment of real-time; And dynamic task allocation process is generally time correlation, a certain moment determines that the optimality of partition can not ensure the optimality of whole process, therefore determines that the consumption plenty of time in moment does this optimization not too large necessity a certain; The problem of research only relates to single Task Allocation Problem mostly, does not embody the feature of movable body multiple constraint, strong coupling and time correlation.The current major applications of modern optimization algorithm is in unmanned plane information and the known static task scheduling process of mission bit stream.

Agent system method: Task Allocation Problem can abstractly be at Agent system (Multi-Agent Systems, MAS) in, there is M different Agent and will perform the different task of N item, how reasonably by Agent different for task matching, thus improve the whole efficiency of system.Agent has two types usually: a kind of is cooperation, and a kind of is from profit.In the previous case, the Agent in system has common target, under centralized management mechanism, coordinate respective behavior, and System benefits is maximized; And in the latter case, Agent has oneself privately owned utility function, be used for instructing its reasoning and decision-making.Auction system based on market mechanism is the conventional method solving dynamic task allocation in Agent system method.The deficiency of this method is, in order to reach optimality, carrying out taking turns communication more, if frequently communicated in hostile area, can increase the probability that unmanned plane is found between the unmanned plane needing to represent each Agent.

Summary of the invention

Take turns the shortcoming of communication more for existing Agent system method, devise the relatively simple synergistic mechanism of one, make just can enter respective pattern (Mode) alternately by three times between unmanned plane, avoid the frequent communication issue between unmanned plane; The guidance law for unmanned aerial vehicle design, unites the coupling of Task Assigned Policy and movable body and temporal correlation, verifies the feasibility of dynamic task allocation strategy in dynamic process.

Unmanned vehicle formation collaboratively searching under uncertain environment and a dynamic task allocation method, is characterized in that: the air route adopting circular arc air route to execute the task as unmanned plane, comprises following step:

Step one: determine the data structure that single rack unmanned plane is safeguarded;

In unmanned plane is formed into columns every frame unmanned plane airborne computer in store:

(1) unmanned plane self sequence number UID;

(2) task number TID performed by each unmanned plane, when unmanned plane is in search pattern, performed by unmanned plane, task number TID is-1;

(3) mission payload that carries of unmanned plane self x is load categories; When middle all elements is 0, and now unmanned plane enters search pattern;

(4) load needed when unmanned plane performs TID task middle element represents different types of load of required by task, and when all elements is all 0, expression task completes; Under search pattern for sky;

Step 2: the offline mode determining unmanned plane;

The offline mode of unmanned plane is divided into search pattern and execution pattern, search pattern is divided into again broadcast mode and response modes, wherein, unmanned plane does not find executable task in search coverage, in communication radius, also do not have other unmanned planes to provide executable task, at this moment unmanned plane is in search pattern; In search procedure, discovery enters broadcast mode after can executing the task, and carries out mission bit stream share with other unmanned planes in communication radius; Under search pattern, if run into the broadcast of other unmanned planes in communication radius, then enter response modes.Unmanned plane finds in self investigative range can complete independently task, or enters execution pattern after response process is carried out in the broadcast receiving other unmanned planes, and now TID is adjusted to from-1 the TID executed the task by unmanned plane;

Step 3: determine the advantage function that unmanned plane is executed the task;

UID load entrained by the unmanned plane of i is made to be tID is that the task of j needs load to be uID is the unmanned plane of i dominance vector A when being the task of j relative to TID _ij=[a _ij1, a _ij2..., a _ijk..., α _ijx] computing formula as follows:

k＝1,2,3...x （1）

When for the UID unmanned plane that is i, to perform TID be the task of j, advantage function computing formula is as follows:

$a_{\mathrm{ij}}={\mathrm{\Σ}}_{k=1}^x{a}_{\mathrm{ijk}}---\left(2\right)$

Step 4: determine dynamic task allocation flow process;

If UID is the unmanned plane of i detect some tasks point, then by step 3 method obtain UID be the unmanned plane of i perform each task point time advantage function, the maximum TID of selective advantage function is the task point of j; If the maximum TID of advantage function is the task point advantage function a of j _ij=0, then TID=-1, unmanned plane proceeds search or waits for that other unmanned planes are broadcasted; If the maximum TID of advantage function is the task point advantage function a of j _ij≠ 0, now, the whether capable complete independently task of this unmanned plane need be judged, if advantage function a _ijduring=2x, then unmanned plane enters the task that execution pattern is executed the task alone a little; If advantage function a _ijduring < 2x, unmanned plane enters broadcast mode, jointly to be finished the work as a supplement j by other unmanned planes;

Step 5: determine the guidance law that aircraft is executed the task;

Make the position of any time unmanned plane for (x _u, y _u), unmanned plane determines the task point j of execution when A point, enter execution pattern, now A point coordinate (x _a, y _a), the coordinate (x of task point j _o, y _o), unmanned plane enters task point O along the circular arc that diameter is OA, now the unmanned plane air route radius of executing the task unmanned plane is at position (x _u, y _u) the desired track drift angle instruction at place:

The feedback that radius limits is added in the instruction of flight path drift angle:

Wherein, z is feedback factor to be adjusted;

The invention has the advantages that:

1, the inventive method is relative to the method for allocating tasks based on modern optimization algorithm, the mode of colony's rolling optimizing is not utilized to search for the optimal case that multiple no-manned plane works in coordination with dynamic task allocation, shorten computing time, reduce the computational load of single rack unmanned plane, be applicable to the comparatively strong and uncertain condition of environment of real-time, can be applicable in the task dynamic allocation procedure of uncertain environment;

2, the inventive method is relative to the method for allocating tasks based on market mechanism auction algorithm, reduces the number of communications of carrying out between unmanned plane, reduces the difficulty of UAV Communication Data-Link design, also improves the survival probability of executing the task in enemy-occupied area simultaneously; Secondly the optimum integer programming algorithm of NP-hard is instead of by the suboptimum PTA method of polynomial time, reduce the computational load of broadcast unmanned plane, ensure that the broadcast unmanned plane real-time of executing the task, this when unmanned plane and task point quantity larger and when executing the task and a little need load more, have very large practical significance;

3, the inventive method with existing can compared with the dynamic assignment data of reference, the solution considered a problem in the dynamic process of complex systematic dynamics, do not only give the collaborative dynamic task allocation strategy of unmanned vehicle formation, and be the concrete guidance law of this Strategy Design, in dynamic process, the computation complexity of verification method and validity, have clear and definite engineering background.

Accompanying drawing explanation

Fig. 1 is that the present invention forms into columns collaboratively searching and dynamic task allocation method overall flow figure;

Fig. 2 to form into columns the description figure that collaboratively searching and dynamic task allocation method will deal with problems for the present invention;

Fig. 3 describes the search of aircraft and the guidance law schematic diagram of task object;

Fig. 4 is the guiding schematic diagram that unmanned plane is executed the task;

Fig. 5 is the Initial situation figures of 5 frame unmanned planes when performing 6 tasks;

When Fig. 6 is for employing existing method, 5 frame unmanned planes perform each unmanned plane trajectory diagram during 6 tasks;

When Fig. 7 is for employing the inventive method, 5 frame unmanned planes perform each unmanned plane trajectory diagram during 6 tasks;

Fig. 8 is the Initial situation figures of 8 frame unmanned planes when performing 10 tasks;

When Fig. 9 is for employing existing method, 8 frame unmanned planes perform each unmanned plane trajectory diagram during 10 tasks;

When Figure 10 is for employing the inventive method, 8 frame unmanned planes perform each unmanned plane trajectory diagram during 10 tasks.

Embodiment

Unmanned vehicle formation collaboratively searching under uncertain environment and a dynamic task allocation method, as shown in Figure 1, comprise following step:

Step one: determine the data structure that single rack unmanned plane is safeguarded;

In order to completing of collaborative dynamic task allocation, need to store following data in the airborne computer of every frame unmanned plane in unmanned plane is formed into columns:

(1) unmanned plane self sequence number UID, is used for confirming the identity of each unmanned plane in unmanned plane is formed into columns, when have broadcast mode unmanned plane broadcast needed for execute the task after, response modes unmanned plane relies on No. UID source confirming broadcast message;

(2) task number TID performed by each unmanned plane, be used for identifying the current pattern be in of unmanned plane and the task that will perform, when unmanned plane is in search pattern, performed by unmanned plane, task number TID is-1;

(3) mission payload that carries of unmanned plane self represent in vector form; X is load categories, therefore it is the vector of 1 × x; Along with unmanned plane carries out task, the load of carrying constantly consumes until vectorial all elements is 0, and now unmanned plane can not perform any goal task and can only enter search pattern;

(4) load needed when unmanned plane performs TID task representing in vector form, is the vector of 1 × x, and the different elements of vector represent different types of load of required by task, constantly carry out task until when vectorial all elements is all 0, represent that this task completes with unmanned plane; This vector under search pattern for sky.

Further illustrate step one below in conjunction with Fig. 2, in figure, pentagon is unmanned plane; Blockage represents the task point on ground, can be emergency place or threat to be destroyed; Peripheral solid line great circle represents the detecting devices investigative range of each unmanned plane, unmanned plane can only find the task point in investigative range, due to unmanned plane form into columns enter this region of search time the task amount of counting to form into columns for unmanned plane with load needed for each task point be all unknown, demarcate so task matching is actual by task point position and distinguish, for convenience of description, the position of task point is represented with No. TID, task point; Suppose that the mission payload kind that the mission payload of UAV flight and impact point need is 3, the digitized representation task number TID outside the bracket of the blockage lower right corner; In bracket 1 × 3 vector for completing the load that this required by task is wanted, the different elements of vector represent different types of load, represent the quantity of various types of load in digital form, such as: expression a little 1 task of will finishing the work needs the first load 0, the second load 1, the third load 2, and load needed for task point constantly carries out task until be 0 with unmanned plane, all elements represent that this task completes when being all 0.Blockage periphery realize the effective radius that small circle represents task point, unmanned plane needs when executing the task to detect the effective radius whether entering task point, if entered, throw in corresponding mission payload, now this unmanned plane completes this task, dotted line great circle represents the communication radius between each unmanned plane, and unmanned plane can only communicate with other unmanned planes be in communication radius.Namely the present invention is unmanned vehicle formation collaboratively searching under such uncertain environment and dynamic task allocation method;

The air route executed the task as unmanned plane in circular arc air route is adopted in the present invention, the reason in this air route is selected to be, first circular arc air route not easily causes unmanned plane overload to exceed restriction when executing the task and switching, topworks is not easily saturated, so can ensure that unmanned plane enters in the circle of task effective radius accurately; Secondly unmanned plane does circular arc search, is convenient to expand hunting zone and is beneficial to find ground-based mission point.

Step 2: the offline mode determining unmanned plane;

The offline mode of unmanned plane is divided into search pattern and execution pattern, and search pattern is divided into again broadcast mode and response modes, and unmanned plane must be in one of following Three models at any time, is introduced respectively below to various pattern.

(1) search pattern (TID=-1): unmanned plane does not find executable task in search coverage, in communication radius, do not have other unmanned planes to provide executable task yet, at this moment unmanned plane is in search pattern, under this pattern, unmanned plane can carry out range searching according to predetermined search pattern, broadcast mode is entered after discovery task in search procedure, carry out information sharing with other unmanned planes in communication radius, jointly complete this task; Under search pattern, if run into the broadcast of other unmanned planes in communication radius, then enter response modes, jointly complete goal task with broadcast unmanned plane.

(2) execution pattern: unmanned plane finds in self investigative range can the task of complete independently, or the pattern entered after the broadcast receiving other unmanned planes carries out responding process, and at this moment TID is adjusted to from-1 TID executed the task by unmanned plane.

Step 3: determine the advantage function that unmanned plane is executed the task;

UID load entrained by the unmanned plane of i is made to be tID is that the task of j needs load to be uID is the unmanned plane of i dominance vector A when being the task of j relative to TID _ij=[a _ij1, a _ij2..., a _ijk..., a _ijx] computing formula as follows:

k＝1,2,3...x （1）

When for the UID unmanned plane that is i, to perform TID be the task of j, advantage function computing formula is as follows:

$a_{\mathrm{ij}}={\mathrm{\&Sigma;}}_{k=1}^x{a}_{\mathrm{ijk}}---\left(2\right)$

In above-mentioned formula (1), Superiority Value a _ijkvalue reason be: work as a _ijkillustrate when=2 that UID is the kth type carriers of the UAV flight of i, meet the demand of task for kth type carriers that TID is j, at this moment this unmanned plane does not need other unmanned planes to assist the kth type carriers task of a j that just can finish the work, and the Superiority Value at this moment had is 2 to the maximum; Work as a _ijk=1 illustrates that UID is that to be not enough to TID be that the task point of j is to the demand of kth type carriers for the kth type carriers of the UAV flight of i, but still there is certain contribution to the demand of this task point to kth type carriers, need to broadcast to other unmanned planes in communication radius, jointly complete this task, the Superiority Value at this moment had is 1; Work as a _ijkthe demand of task to kth type carriers of=0 illustrates UID to be the kth type carriers of the UAV flight of i to TID be j is not contributed, the unmanned plane that may be i due to UID does not have kth type carriers or TID to be that the task of j does not need kth type carriers, at this moment the Superiority Value of the UID unmanned plane that is i should not be greater than calling and the unmanned plane oneself that comes finds the Superiority Value that this task has, otherwise can bring between unmanned plane and communicate frequently, increase the catch probability of ground based threats;

Below in conjunction with Fig. 2, have 3 types with load, i.e. x=3, and the unmanned plane that UID is 1 detect in investigative range TID be 1 and 2 task be example, when executing the task to unmanned plane, the computing method of advantage function carry out a step explanation, wherein:

${\stackrel{\&OverBar;}{R}}_1^u=\left[\begin{array}{ccc}{r}_{11}^u& r_{12}^u& r_{13}^u\end{array}\right]=\left[\begin{array}{ccc}1& 2& 0\end{array}\right]$

${\stackrel{\&OverBar;}{R}}_1^t=\left[\begin{array}{ccc}{r}_{11}^t& r_{12}^t& r_{13}^t\end{array}\right]=\left[\begin{array}{ccc}0& 1& 2\end{array}\right]$

${\stackrel{\&OverBar;}{R}}_2^t=\left[\begin{array}{ccc}{r}_{21}^t& r_{22}^t& r_{23}^t\end{array}\right]=\left[\begin{array}{ccc}2& 0& 1\end{array}\right]$

According to formula (1) can obtain UID be 1 unmanned plane be the dominance vector of the task point of 1 to TID be A ₁₁=[0 2 0], UID be 1 unmanned plane be the dominance vector of the task point of 2 to TID be A ₁₂=[1 0 0], according to formula (2) can obtain UID be 1 unmanned plane be the advantage function a of the task point of 1 to TID ₁₁=2.

Step 4: determine dynamic task allocation flow process;

As shown in Figure 3, UID is that the airborne detecting devices of the unmanned plane of i is with some cycles detection mission dot information, if detect some tasks point, then by step 3 method obtain UID be the unmanned plane of i perform each task point time advantage function, the maximum TID of selective advantage function is the task point of j; If the maximum TID of advantage function is the task point advantage function a of j _ij=0, although illustrate that unmanned plane detects task point, unmanned plane institute bringing onto load has no contribution to this task, at this moment TID=-1, and unmanned plane proceeds search or waits for that other unmanned planes are broadcasted; If the maximum TID of advantage function is the task point advantage function a of j _ij≠ 0, illustrate that this unmanned plane has contribution to detecting of task, select TID to be the tasks carrying of j.

After determining that performing TID is the task of j, the whether capable complete independently task of this unmanned plane need be judged, if during advantage function, then illustrate that the load entrained by this unmanned plane can not broadcast fulfilling the task all by oneself a little of task completely; If a _ijduring < 2x, illustrate that this unmanned plane does not have the task ability that complete independently TID is j, then need broadcast jointly to be finished the work as a supplement j by other unmanned planes, now unmanned plane enters broadcast mode.There are 3 types to further illustrate for load, work as a _ijwhen=6, a is described _ijk=2, wherein N (1,3) represents the integer set of 1 to 3, and at this moment unmanned plane self load of being with just can be executed the task j, then arrange TID=j and enter execution pattern; Work as a _ijduring < 6, illustrate that this unmanned plane does not have the task ability that complete independently TID is j, then need broadcast jointly to be finished the work as a supplement j by other unmanned planes;

Unmanned plane in a broadcast mode, the load of UID entrained by the unmanned plane of i execution TID is that the task of j needs load what through type (3) calculating will be broadcasted needs load vectors:

${\stackrel{\&OverBar;}{S}}_{\mathrm{ij}}^u=[s_{\mathrm{ij}1}^u,s_{\mathrm{ij}2}^u,......,s_{\mathrm{ijk}}^u,......,s_{\mathrm{ijx}}^u]---\left(3\right)$

Wherein,

UID is the load vectors of unmanned plane by self UID and needs of i after broadcast, being in search pattern and containing broadcast vector within communication radius the unmanned plane of middle element, makes its UID be respectively n ₁, n ₂..., n _w..., n _p, all enter response modes, all the other continue to keep search pattern.Enter unmanned plane in response modes by self UID, position and entrained load information send to UID to be the unmanned plane of i, UID is that the unmanned plane information of unmanned plane to the response modes that these are collected of i is handled as follows:

1) set is set as initial sets, wherein, C _ijrepresent the UID set of final response unmanned plane, represent C _ijin final each response unmanned plane entrained by mission payload set;

2) determine that response modes unmanned plane arrives the estimated time of arrival (ETA) (ETA that TID is j task, Estimated Time ofArrival), can determine that when unmanned plane velocity variations is little unmanned plane enters the ETA of task point, is specially easily: set estimated time of arrival (ETA) gather as $\mathrm{ETAs}=\{t_{n_1},t_{n_2},......,t_{n_w},......,t_{n_p}\},$ Then:

$t_{n_w}\&ap;\frac{\mathrm{\&pi;R}-r}{V_u}---\left(4\right)$

Wherein, r represents the effective radius of task point, V _ufor the speed of unmanned plane, the speed of unmanned plane leans on the Throttle Opening Control of unmanned vehicle engine, by Throttle Opening Control at about 50m/s; R is the air route radius that unmanned plane is executed the task;

3) estimated time of arrival (ETA) of response modes unmanned plane is sorted [N, T]=sort (ETAs), wherein $T=\{t_{n_1^{*}},t_{n_2^{*}},......,t_{n_w^{*}},......,t_{n_p^{*}}\}$ For the result after ascending sequence time of arrival, $N=\{n_1^{*},n_2^{*},......,n_w^{*},......,n_p^{*}\}$ For No. UID, corresponding unmanned plane;

4) according to N order to set C _ijsupplement, use according to the order of N simultaneously right supplement, until or whole element adds to C in N _ijin, suppose the response modes unmanned plane set after supplementing $C_{\mathrm{ij}}=\{{\hat{n}}_1^{*},{\hat{n}}_2^{*},{\hat{n}}_w^{*},......,{\hat{n}}_q^{*}\},$ q≤p；

5) C is traveled through _ijin each element inspection whether be true, if very, then will from C _ijmiddle rejecting, will simultaneously from middle rejecting, finally obtains the C removing redundant elements _ijwith do not add to C _ijthe TID of the response modes unmanned plane that neutralization is rejected is set to-1, continues to make it be in search pattern and searches for other task points, final C _ijin the TID of unmanned plane be set to task j and enter execution pattern.

Pass through said method, once after unmanned plane finds the task that can perform, just can determine alternately in three communication works in coordination with the unmanned plane group executed the task and the unmanned plane continuing search pattern, such structure overcomes the shortcoming repeatedly communicated of MAS auction algorithm, has the advantages such as algorithm is simple, real-time is good.

Carry out above-mentioned 3), 4), 5) reason be in order to avoid use integer programming (Linear Programming) method determine C _ij, represent 3 by the thought of integer programming), 4), 5) object as shown in the formula:

Wherein, be and C _ijthe subset of corresponding T, integer programming belongs to NP-hard problem, along with unmanned plane and task quantity increase, the calculated amount of broadcast mode unmanned plane can produce shot array, therefore the shot array problem that eliminate redundancy element two steps in the sequence of the ETAs in (3) and (4) avoid integer programming to produce is used, the computing time of such broadcast mode unmanned plane can drop to sequence and reject the polynomial time of operation, be convenient to the applications that requirement of real-time is stronger, shortcoming is that result at this moment often only has suboptimality.For convenience of description, by the step of above-mentioned (1) ~ (5) referred to as PTA(PolynomialTime Algorithm) method, in order to better description of step four, following table provides the false code form of step 4:

Step 5: determine the guidance law that aircraft is executed the task;

As shown in Figure 4, for the guiding schematic diagram that unmanned plane is executed the task, unmanned plane is all in whole search procedure determines high state, so the guidance law of the search of aircraft and tasks carrying mainly designs the instruction of flight path drift angle, guarantee that each frame unmanned plane that unmanned plane is formed into columns can enter task circular arc accurately.Make the position of any time unmanned plane for (x _u, y _u), unmanned plane determines the task point j of execution at A point, enters execution pattern, now A point coordinate (x _a, y _a), the coordinate (x of task point j _b, y _b), now wish that unmanned plane enters task point B along the circular arc that diameter is BA, the air route radius R that now unmanned plane is executed the task is unmanned plane is at position (x _u, y _u) the desired track drift angle instruction at place:

Wherein, a is the angle of unmanned plane task point line and earth axes x-axis,

The feedback that radius limits is added in the instruction of flight path drift angle:

Wherein, z is feedback factor to be adjusted, is taken as 0.002 in this example.Unmanned plane is in search pattern and also can adopts such guidance law that unmanned plane is moved in a circle along the search center pre-estimated.Above-mentioned guidance law can ensure that unmanned plane enters task point along circular arc track.

Embodiment 1:

Original state when 5 frame unmanned planes perform 6 tasks as shown in Figure 5, wherein, the radius of investigation of each unmanned plane is 800 meters, communication radius is 1600 meters, in order to clear, does not mark in the drawings, when unmanned plane is in search pattern, guidance law guiding unmanned plane does the center of circle (2100,2025), radius is the circular motion of 800 meters, and this belongs to the typical case that unmanned plane number of executing the task is less than number of tasks.The initial information of 5 frame unmanned planes is as shown in the table:

The initial information of 6 task points is as shown in the table:

When adopting existing method to carry out unmanned plane task matching, unmanned plane only selects the task that in self investigative range, advantage function is maximum, does not have information interaction between unmanned plane, and 5 frame unmanned planes perform each unmanned plane track of 6 tasks as shown in Figure 6; Unmanned plane initial information is consistent with Fig. 5 with task point initial information, and UAV Communication radius is 1600 meters, radius of investigation 800 meters, and the full that carries out task needs time 200.8s.

And when adopting the inventive method to carry out unmanned vehicle formation collaboratively searching and dynamic task allocation, the strategy of three communication interactions is have employed between unmanned plane, unmanned plane initial information, task point initial information is consistent with Fig. 5 with the guidance law of search pattern, and each unmanned plane track as shown in Figure 7; UAV Communication radius is 1600 meters, radius of investigation 800 meters, and the full that carries out task needs time 109.2s.

Embodiment 2:

The inventive method is adopted to carry out 8 frame unmanned planes when performing 10 tasks, original state as shown in Figure 8, the radius of investigation of each unmanned plane is 800 meters, communication radius is 1600 meters, in order to clear, do not mark in the drawings, when unmanned plane is in search pattern, guidance law guiding unmanned plane does the center of circle (2100,2025), radius is the circular motion of 800 meters.Wherein, the initial information of 8 frame unmanned planes is as shown in the table:

The initial information of 10 task points is as shown in the table:

When adopting existing method to carry out unmanned plane task matching, unmanned plane only selects the task that in self investigative range, advantage function is maximum, does not have information interaction between unmanned plane, and each unmanned plane track when 8 frame unmanned planes perform 10 goal tasks as shown in Figure 9; Unmanned plane initial information is consistent with Fig. 6 with task point initial information, and UAV Communication radius is 1600 meters, radius of investigation 800 meters, and the full that carries out task needs time 184.8s.

And when adopting the inventive method to carry out unmanned vehicle formation collaboratively searching and dynamic task allocation, each unmanned plane track as shown in Figure 10; The strategy of three communication interactions is have employed between unmanned plane, unmanned plane initial information, task point initial information is consistent with Fig. 8 with the guidance law of search pattern, UAV Communication radius is 1600 meters, radius of investigation 800 meters, and the full that carries out task needs time 149.6s.

Can find out thus, relative to the method for allocating tasks based on modern optimization algorithm, the inventive method does not utilize the mode of colony's rolling optimizing to search for the optimal case that multiple no-manned plane works in coordination with dynamic task allocation, shorten computing time, reduce the computational load of single rack unmanned plane, be applicable to the comparatively strong and uncertain condition of environment of real-time, unmanned plane at most only needs three communication from discovery target to entering execution pattern simultaneously, relative MAS competition auction system reduces the number of communications between unmanned aerial vehicle group, simulation result shows that the method can be applied in the task dynamic allocation procedure of uncertain environment.

Claims (1)

1. the unmanned vehicle formation collaboratively searching under uncertain environment and a dynamic task allocation method, is characterized in that: the air route adopting circular arc air route to execute the task as unmanned plane, comprises following step:

Step one: determine the data structure that single rack unmanned plane is safeguarded;

In unmanned plane is formed into columns every frame unmanned plane airborne computer in store:

(1) unmanned plane self sequence number UID;

(2) task number TID performed by each unmanned plane, when unmanned plane is in search pattern, performed by unmanned plane, task number TID is-1;

(3) mission payload that carries of unmanned plane self x is load categories; When middle all elements is 0, and now unmanned plane enters search pattern;

(4) load needed when unmanned plane performs TID task middle element represents different types of load of required by task, and when all elements is all 0, expression task completes, under search pattern for sky;

Step 2: the offline mode determining unmanned plane;

The offline mode of unmanned plane is divided into search pattern and execution pattern, search pattern is divided into again broadcast mode and response modes, wherein, unmanned plane does not find executable task in search coverage, in communication radius, also do not have other unmanned planes to provide executable task, at this moment unmanned plane is in search pattern; Discovery task in search procedure and entrained by self, load is not enough to finish the work time enter broadcast mode, with other unmanned planes in communication radius carry out mission bit stream share; Under search pattern, if run into the broadcast of other unmanned planes in communication radius, then enter response modes; Unmanned plane finds in self investigative range can the task of complete independently, or enters execution pattern after response process is carried out in the broadcast receiving other unmanned planes, and now TID is adjusted to from-1 the TID executed the task by unmanned plane;

Step 3: determine the advantage function that unmanned plane is executed the task;

UID load entrained by the unmanned plane of i is made to be tID is that the task of j needs load to be uID is the unmanned plane of i dominance vector A when being the task of j relative to TID _ij=[a _ij1, a _ij2..., a _ijk..., a _ijx] computing formula as follows:

When for the UID unmanned plane that is i, to perform TID be the task of j, advantage function computing formula is as follows:

$a_{\mathrm{ij}}={\mathrm{\&Sigma;}}_{k=1}^x{a}_{\mathrm{ijk}}---\left(2\right)$

Step 4: determine dynamic task allocation flow process;

If UID is the unmanned plane of i detect some tasks point, then by step 3 method obtain UID be the unmanned plane of i perform each task point time advantage function, the maximum TID of selective advantage function is the task point of j; If the maximum TID of advantage function is the task point advantage function a of j _ij=0, then TID=-1, unmanned plane proceeds search or waits for that other unmanned planes are broadcasted; If the maximum TID of advantage function is the task point advantage function a of j _ij≠ 0, now, the whether capable complete independently task of this unmanned plane need be judged, if advantage function a _ijduring=2x, then unmanned plane enters the task that execution pattern is executed the task alone a little; If advantage function a _ijduring < 2x, unmanned plane enters broadcast mode, and jointly to be finished the work as a supplement j by other unmanned planes, concrete mode is:

Unmanned plane in a broadcast mode, the load of UID entrained by the unmanned plane of i execution TID is that the task of j needs load then UID be i unmanned plane broadcast need load vectors, ${\stackrel{\&OverBar;}{S}}_{\mathrm{ij}}^u=[s_{\mathrm{ij}1}^u,s_{\mathrm{ij}2}^u,......,s_{\mathrm{ijk}}^u,......,s_{\mathrm{ijx}}^u],$ Wherein,

$s_{\mathrm{ijk}}^u=\left\{\begin{array}{c}{r}_{\mathrm{jk}}^t-r_{\mathrm{ik}}^u,r_{\mathrm{jk}}^t>r_{\mathrm{ik}}^u\\ 0,\mathrm{otherwise}\end{array}\right.k=\mathrm{1,2,3}...x;$

UID is the load vectors of unmanned plane by self UID and needs of i after broadcast, being in search pattern and containing broadcast vector within communication radius the unmanned plane of middle element, makes its UID be respectively n ₁, n ₂..., n _w..., n _p, all enter response modes, and self UID, position and entrained load information send to UID to be the unmanned plane of i, UID is that the unmanned plane information of unmanned plane to the response modes that these are collected of i is handled as follows:

1) set is set as initial sets, wherein, C _ijrepresent the UID set of final response unmanned plane, represent C _ijin final each response unmanned plane entrained by mission payload set;

2) determining that response modes unmanned plane arrives TID is the estimated time of arrival (ETA) ETA of j task, can determine that unmanned plane enters the ETA of task point, is specially when unmanned plane velocity variations is little easily: set estimated time of arrival (ETA) gather as $\mathrm{ETAs}=\{t_{n_1},t_{n_2},......,t_{n_w},......,t_{n_p}\},$ Then:

$t_{n_w}\&ap;\frac{\mathrm{\&pi;R}-r}{V_u}---\left(4\right)$

Wherein, r represents the effective radius of task point, V _ufor the speed of unmanned plane, the speed of unmanned plane leans on the Throttle Opening Control of unmanned vehicle engine, by Throttle Opening Control at about 50m/s; R is the air route radius that unmanned plane is executed the task;

3) estimated time of arrival (ETA) of response modes unmanned plane is sorted wherein $T=\{t_{n_1^{*}},t_{n_2^{*}},......,t_{n_w^{*}},......,t_{n_p^{*}}\}$ For the result after ascending sequence time of arrival, $N=\{n_1^{*},n_2^{*},......,n_w^{*},......,n_p^{*}\}$ For No. UID, corresponding unmanned plane;

4) according to N order to set C _ijsupplement, use according to the order of N simultaneously right supplement, until or whole element adds to C in N _ijin, suppose the response modes unmanned plane set after supplementing $C_{\mathrm{ij}}=\{{\hat{n}}_1^{*},{\hat{n}}_2^{*},......,{\hat{n}}_w^{*},......,{\hat{n}}_q^{*}\},q\&le;p;$

5) C is traveled through _ijin each element inspection whether be true, if very, then will reject from Cij, will simultaneously from middle rejecting, finally obtains the C removing redundant elements _ijwith do not add to C _ijthe TID of the response modes unmanned plane that neutralization is rejected is set to-1, continues to make it be in search pattern and searches for other task point, final C _ijin unmanned plane

TID be set to task j and enter execution pattern; Step 5: the search determining aircraft and the guidance law of executing the task;

Make the position of any time unmanned plane for (x _u, y _u), unmanned plane determines the task point j of execution when A point, enter execution pattern, now A point coordinate (x _a, y _a), the coordinate (x of task point j _o, y _o), unmanned plane enters task point O along the circular arc that diameter is OA, now the unmanned plane air route radius of executing the task unmanned plane is at position (x _u, y _u) the desired track drift angle instruction at place:

The feedback that radius limits is added in the instruction of flight path drift angle:

Wherein, z is feedback factor to be adjusted.