A Consensus Active Neighbor Optimization Method for Large-Scale Distributed UAV Swarms

Abstract

The present invention provides a consensus-based active neighbor optimization method for large-scale distributed UAV clusters, which can solve the problem of information interaction object selection in UAV cluster systems, and is particularly applicable to large-scale distributed UAV clusters. The method includes: for the current UAV i, if the number of its optional neighbors is less than or equal to the neighbor set upper limit n C , then all optional neighbors are added to the neighbor set to obtain the neighbor optimization result; if the number of optional neighbors is greater than n C , then according to the collision safety distance, the optional neighbors are divided into too-close neighbors and non-too-close neighbors, and all the too-close neighbors are added to the neighbor set; when the number of too-close neighbors is less than the neighbor set upper limit n C , filling is required, that is, according to the conditions of the closest distance in space and the closest state, preferred neighbors are selected from the non-too-close neighbors as neighbor supplements and added to the neighbor set to obtain the neighbor optimization result.

Description

Technical Field

[0001] The present invention belongs to the field of UAV swarm control, and particularly relates to a consensus active neighbor selection method for large-scale distributed UAV swarms. Background Art

[0002] With the rapid development of technologies such as artificial intelligence, satellite navigation, and network communication, UAVs applied to various fields have emerged, especially in the fields of agriculture, ground remote sensing, air logistics, military reconnaissance and attack, etc. Compared with traditional aircraft, UAVs have the characteristics of high speed, all-weather, non-contact, zero casualties, etc., so they have been widely used in actual combat. Although UAV technology has developed rapidly and its performance has become increasingly powerful, a single UAV is also difficult to independently complete complex tasks. Especially when solving complex tasks that are distributed in space and need to be processed in parallel in time, a single UAV is difficult to play a role; at the same time, restricted by the platform payload, the resources that a single UAV can be equipped with are limited, and its ability to complete tasks is also limited. To address the above problems, in practical applications, it is often necessary to have multiple UAVs cooperate with each other to form a UAV swarm system.

[0003] For a distributed UAV swarm, each UAV can only perform limited and local information interactions. The selection of neighbors (information interaction objects) determines the order and cohesion of the swarm. Generally speaking, restricted by individual perception, communication, computing power, and motion ability, neighbor selection needs to comprehensively consider influencing factors such as the spatial distance, motion state, computational amount, and information amount between individuals.

[0004] In view of the above influencing factors, as well as the information continuity and reliability required in the information interaction process, it is necessary to design a consensus active selection method for UAV swarms, so that information coordination can be carried out through the autonomy of intelligent individuals in the intelligent group. Summary of the Invention

[0005] In view of this, the present invention provides a consensus active neighbor selection method applicable to large-scale distributed UAV swarms, which can solve the problem of selecting information interaction objects in a UAV swarm system.

[0006] To solve the above technical problems, the present invention is implemented as follows.

[0007] A consensus active neighbor selection method applicable to large-scale distributed UAV swarms includes:

[0008] Step 1: Determine the optional neighbors of the current UAV i;

[0009] Step 2: If the number of optional neighbors is less than or equal to the upper limit n of the neighbor set C , then add all the optional neighbors to the neighbor set to obtain the neighbor selection result of the current UAV i; if the number of optional neighbors is greater than the upper limit n of the neighbor setC , then execute Step Three;

[0010] Step Three: First, reduce the optional neighbors according to the collision safety distance: According to the collision safety distance, divide the optional neighbors into too-close neighbors and non-too-close neighbors, and add all the too-close neighbors to the neighbor set;

[0011] Then, when the current neighbor set is less than the upper limit n of the neighbor set C , perform filling: Screen the preferred neighbors from the non-too-close neighbors as neighbor supplements according to the conditions of the closest distance in space and the closest state, and add them to the neighbor set to obtain the neighbor preference result of the current UAV i.

[0012] Preferably, the performing filling when the neighbor set is less than the upper limit n of the neighbor set C is:

[0013] First, select a threshold r according to the spatial distance S , and screen out the optional neighbors whose distance from the current UAV i is less than r S from the non-too-close neighbors as the position closest neighbors; Take the too-close neighbors in position and the position closest neighbors as the neighbor preference result;

[0014] If the number of position closest neighbors is greater than the supplement upper limit n C,free , then it is necessary to reduce the position closest neighbors; At this time, further select a threshold r according to the state distance V , and screen out the optional neighbors whose state distance from the current UAV i is less than r V from the position closest neighbors as the state closest neighbors; Take all or part of the too-close neighbors and the state closest neighbors as the neighbor preference result;

[0015] wherein, the state refers to the UAV speed; n C,free is the upper limit n of the neighbor set C minus the number of too-close neighbors.

[0016] Preferably, if the number of state closest neighbors is greater than the supplement upper limit n C,free when, then select n C,free state closest neighbors with the closest states and add them to the neighbor set.

[0017] Preferably, Step Two specifically includes Steps 201 to 202, and Step Three includes Steps 203 to 206:

[0018] Step 201: The current UAV i obtains the information of the optional neighbors, including the UAV number set UAV position information set UAV speed information set

[0019] Step 202: Obtain the total number of elements in the UAV number set If is less than or equal to the neighbor set upper limit n C , add all optional neighbors to the neighbor set and determine the neighbor number set Neighbor position set Neighbor speed set This process ends; otherwise, step 203 needs to be performed on the optional neighbors;

[0020] Step 203: Reduce the optional neighbors according to the collision safety distance:

[0021] Let the collision distance threshold be r coll lower than the safety spacing during normal flight of the UAV cluster. According to the UAV position information set calculate the optional neighbors whose distance from UAV i is less than r coll to form the too-close neighbor set N i,collision , and the other optional neighbors form the non-too-close neighbor set; if the total number of elements in the too-close neighbor set |N i,collision | is greater than the neighbor set upper limit n C , all too-close neighbors are added to the neighbor set to determine the neighbor number set N i = N i,collision Neighbor position set Neighbor speed set This process ends; otherwise, neighbors need to be supplemented except for the too-close neighbors, and the supplement upper limit is n C,free = n C -|N i,collision |, and perform step 204; where p j represents the UAV position, and v j represents the UAV speed;

[0022] Step 204: Screen the non-too-close neighbors according to the spatial distance:

[0023] Let the spatial distance selection threshold be r S , and according to the UAV position information set determine the non-too-close neighbors whose distance from UAV i is less than r S to form the nearest neighbor position set N i,spatial , if the total number of elements in the nearest neighbor position set |N i,spatial | is less than or equal to the supplement upper limit n C,free , then add the too-close neighbors and the nearest neighbor positions to the neighbor set and output the neighbor number set N i = N i,spatial + N i,collision Neighbor position set Neighbor speed set This process ends; otherwise, the nearest neighbor positions are more than the supplement upper limit nC,free , perform step 205;

[0024] Step 205: Reduce the position nearest neighbors according to the state distance:

[0025] Set the state distance selection threshold as r V , according to the UAV speed information set Determine the position nearest neighbors whose state distance from UAV i is less than r V to form the state nearest neighbor set N i,status , if the total number of elements in the position nearest neighbor set |N i,spatial | is less than or equal to the replenishment upper limit n C,free , then add the over-nearest neighbors and the state nearest neighbors to the neighbor set, and output the neighbor number set N i = N i,status + N i,collision , the neighbor position set the neighbor speed set This process ends; otherwise, the number of state nearest neighbors is more than the replenishment upper limit n C,free , perform step 206;

[0026] Step 206: Sort the state distances in the state nearest neighbor set, and select the n C,free state nearest neighbors with the smallest state distances to form the preferred neighbor set N i,order , output the neighbor number set N i = N i,order + N i,collision , the neighbor position set the neighbor speed set

[0027] Preferably, in step 1, the optional neighbors of the current UAV i are determined as: the UAVs that the current UAV i can receive information from are used as the optional neighbors.

[0028] Beneficial effects:

[0029] (1) The present invention provides a consensus active neighbor preference method for UAV swarms, which is particularly applicable to large-scale distributed UAV swarms. This method does not simply use all UAVs that the current UAV can receive information from as the information interaction objects, but determines neighbors according to multiple criteria such as safety distance, spatial nearest neighbor, state nearest neighbor, and quantity upper limit, so as to achieve neighbor preference.

[0030] (2) In the neighbor preference method provided by the present invention, special situations where the total number of UAVs within the communication range or neighborhood of the current UAV is less than the neighbor set upper limit are considered. The total number of set elements before and after comparison and screening in each step of selecting neighbors is considered. If it does not exceed the neighbor set upper limit, the screening stops. Only when it exceeds the upper limit will redundant information be filtered according to the spatial distance and state distance.

[0031] (3) When there are many optional neighbors, screening is required. The present invention first screens out the set of overly close neighbors, and the set is set with a distance threshold lower than the safety spacing during the normal flight of the UAV cluster. Objects in the set of overly close neighbors are all regarded as neighbors, and subsequent collision avoidance is achieved through the cluster control algorithm. This can significantly reduce the accident risk caused by collisions between UAVs during the flight of a large-scale UAV group, and improve the safety and reliability of the UAV cluster.

[0032] (4) In a preferred embodiment, when the number of overly close neighbors is less than the required number of neighbors, by adopting the determination order of first spatial distance and then state distance, the special situation where individuals with a close spatial distance but a far state distance may collide but are not regarded as neighbors is avoided. Description of the Drawings

[0033] Figure 1 is a schematic diagram of the consensus active neighbor selection method for the UAV cluster of the present invention.

[0034] Figure 2 is a flowchart of the consensus active neighbor selection method for the UAV cluster provided by the preferred embodiment of the present invention. Detailed Embodiments

[0035] The present invention will be described in detail below with reference to the accompanying drawings and by way of examples.

[0036] First, several important technical terms related to the present invention are explained.

[0037] The consensus active neighbor selection method for the UAV cluster of the present invention is used to determine the neighbors of the current UAV.

[0038] A neighbor is an information interaction object selected by the current UAV.

[0039] The information interaction object is selected from other UAVs that can receive information from the current UAV, that is, other UAVs that can receive information from the current UAV are optional neighbors.

[0040] The consensus active neighbor selection method for the UAV cluster includes three main contents: spatial nearest neighbor, state nearest neighbor, and upper limit on the number.

[0041] Spatial nearest neighbor is the criterion for selecting neighbors based on the spatial distance between UAVs.

[0042] State nearest neighbor is the criterion for selecting neighbors based on the relative speed magnitude between UAVs.

[0043] Upper limit on the number is the upper limit constraint criterion for the number of neighbors selected by the UAV.

[0044] The unmanned aerial vehicle (UAV) adopted in the present invention is equipped with a navigation and positioning device, a communication device, and a control device.

[0045] The core idea of the consensus active neighbor optimization method for the UAV swarm provided by the present invention is as follows:

[0046] Step 1: Determine the optional neighbors of the current UAV i.

[0047] Step 2: If the number of optional neighbors is less than or equal to the upper limit n of the neighbor set C , then add all the optional neighbors to the neighbor set to obtain the neighbor optimization result; if the number of optional neighbors is greater than the upper limit n of the neighbor set C , then execute Step 3;

[0048] Step 3: First, reduce the optional neighbors according to the collision safety distance: according to the collision safety distance, divide the optional neighbors into too-close neighbors and non-too-close neighbors, and add all the too-close neighbors to the neighbor set;

[0049] Then, when the current neighbor set is less than the upper limit n of the neighbor set C , perform filling: screen and select the optimal neighbors from the non-too-close neighbors as neighbor supplements according to the conditions of the closest distance in space and the closest state, and add them to the neighbor set to obtain the neighbor optimization result.

[0050] It can be seen that the present invention does not simply use all the UAVs that the current UAV can receive information from as the information interaction objects, but determines the neighbors according to multiple criteria such as safety distance, spatial nearest neighbor, state nearest neighbor, and upper limit of quantity, so as to achieve neighbor optimization. Secondly, when there are many optional neighbors, screening is required. The present invention first screens out the set of too-close neighbors, and the set distance threshold is lower than the safety distance during the normal flight of the UAV swarm. The objects in the set of too-close neighbors are all regarded as neighbors, and subsequent collision avoidance is achieved through the swarm control algorithm.

[0051] When the number of too-close neighbors is less than the required number of neighbors n C , filling is required. In a preferred embodiment, the filling judgment order of first spatial distance and then state distance is set, avoiding the special situation where individuals with a close spatial distance but a far state distance may collide but are not regarded as neighbors. The specific filling scheme is as follows:

[0052] First, select the threshold r for the spatial distance according to the setting S , screen out the optional neighbors from the non-too-close neighbors whose distance from the current UAV i is less than r S , and use them as the nearest neighbors in position; use the too-close neighbors in position and the nearest neighbors in position as the neighbor optimization result.

[0053] If the number of nearest neighbors in position is greater than the upper limit n for supplementation C,free (n C,freeFor the upper limit n of the neighbor set C (subtracting the number of too-close neighbors), the closest neighbors in terms of position need to be reduced; at this time, further select the threshold r according to the state distance V , and screen out the optional neighbors with a state distance less than r from the closest neighbors in terms of position V as the closest neighbors in terms of state; use all or part of the too-close neighbors and the closest neighbors in terms of state as the preferred neighbor results

[0054] Furthermore, the method for selecting the number of the closest neighbors in terms of state is: if the number of the closest neighbors in terms of state is greater than the replenishment upper limit n C,free , then select n C,free closest neighbors in terms of state with the closest states to join the neighbor set; if the number of the closest neighbors in terms of state is less than or equal to n C,free , then select all the closest neighbors in terms of state as the selected neighbors

[0055] The following combines Figure 2 to describe the preferred embodiments of the present invention in detail

[0056] A1: Set the upper limit n of the neighbor set C , and obtain the information of other drones received by the current drone i, including the set of drone numbers the set of drone position information the set of drone speed information The above-mentioned other drones are the optional neighbors

[0057] A2: Determine whether the number of optional neighbors is less than or equal to the upper limit n of the neighbor set C , if so, use all the optional neighbors as neighbors to form a neighbor set. The set of neighbor numbers, the set of position information, and the set of drone speed information are the three sets corresponding to the received optional neighbors. Otherwise, there are too many optional neighbors, and step A3 needs to be entered for screening

[0058] In this step, obtain the total number of elements in the set of drone numbers If is less than or equal to the upper limit n of the neighbor set C , add all the optional neighbors to the neighbor set and output the set of neighbor numbers the set of neighbor positions the set of neighbor speeds This process ends; otherwise, execute step A3

[0059] A3: Reduce the optional neighbors according to the collision safety distance

[0060] Set the threshold lower than the safety distance during the normal flight of the UAV cluster as the collision distance threshold; according to the UAV position information, screen out the optional neighbors in the optional neighbor set that are less than the collision distance threshold from the current UAV i, which are called overly close neighbors; an overly close neighbor refers to a UAV that is too close to UAV i, less than the safety distance, and is prone to collision. All these overly close neighbors are regarded as neighbors and form the overly close neighbor set N i,collision ; if the number of overly close neighbors is more than n C , it is considered that the current neighbors are sufficient, and the overly close neighbor set N i,collision is used as the neighbors; if the number of overly close neighbors is less than n C , then a part of the UAVs need to be screened from the non-overly close neighbors for supplementation. The required supplementation amount n C,free is n C minus the total number of overly close neighbors; then execute step A4;

[0061] In this step, set the collision distance threshold as r coll , and calculate the overly close neighbor set according to the UAV position information set The other optional neighbors form the non-overly close neighbor set; if the total number of elements in the overly close neighbor set |N | is greater than the upper limit n of the neighbor set i,collision , all overly close neighbors are added to the neighbor set, and the neighbor number set N C is determined as N i = N i,collision , the neighbor position set The neighbor speed set This process ends; otherwise, neighbors need to be supplemented in addition to the overly close neighbors. If the number of overly close neighbors is less than n C , continue to select neighbors from the non-overly close neighbor UAV set. The supplementable upper limit is n C,free = n C - |N i,collision |; calculate the non-overly close neighbor number set The non-overly close neighbor position set The non-overly close neighbor speed set Execute step A4. Among them, p j represents the UAV position, and v j represents the UAV speed;

[0062] A4: Screen the non-overly close neighbors according to the spatial distance:

[0063] According to the UAV position information, screen out the non-overly close neighbors in the non-overly close neighbor position set that are less than the spatial distance selection threshold from the current UAV i, and form the nearest neighbor set N i,spatial ; if the number of nearest neighbors is less than the required supplementation amount n C,free, it indicates that the number of position nearest neighbors is appropriate, and the neighbor set includes over-close neighbors and position nearest neighbors; otherwise, it indicates that the number of position nearest neighbors is excessive and further reduction is required.

[0064] In this step, set the spatial distance selection threshold as r S , according to the set of UAV position information Calculate the set of position nearest neighbors If the total number of elements in the set of position nearest neighbors |N i,spatial | is less than or equal to the replenishment upper limit n C,free , determine the neighbor number set N i = N i,spatial + N i,collision , neighbor position set Neighbor speed set This process ends; otherwise, there are more position nearest neighbors than the replenishment upper limit n C,free , execute step A5.

[0065] A5: Reduce the position nearest neighbors according to the state distance:

[0066] According to the UAV speed information, select the UAVs from the set of position nearest neighbors N i,spatial whose state distance from the current UAV i is less than the state distance selection threshold to form the set of state nearest neighbors N i,status ; if the number of state nearest neighbors is less than or equal to the number of non-over-close neighbor sets n C,free , it indicates that the number is appropriate, and the neighbor set includes over-close neighbors and state nearest neighbors; otherwise, it indicates that the number of state nearest neighbors is excessive and further reduction is required.

[0067] In this step, d s,ij represents the state distance, that is, the relative speed magnitude between UAV i and UAV j. Set the state distance selection threshold as r V , according to the set of UAV speed information Calculate the set of state nearest neighbors N i,status = {j|j ∈ N i,spatial , d s,ij = ||v i - v j || < r V}, if the total number of elements in the set of position nearest neighbors |N i,spatial | is less than or equal to the replenishment upper limit n C,free , determine the neighbor number set N i = N i,status + N i,collision , neighbor position set Neighbor speed set This process ends, otherwise, there are more state nearest neighbors than the replenishment upper limit n C,free , execute step A6.

[0068] A6: Sort the state nearest neighbors according to the state distance, and select the n with the smallest state distance C,free to form the preferred neighbor set N i,order , and form the neighbor set by combining the preferred neighbors and the position too-close neighbors.

[0069] In this step, the UAVs in the state nearest neighbor set are sorted in ascending order according to the state distance, and the numbers of the first n C,free elements are selected to form a set to determine the neighbor number set N i = N i,order + N i,collision , the neighbor position set the neighbor speed set

[0070] So far, this process ends.

[0071] The above specific embodiments only describe the design principle of the present invention. The shapes and names of the components in this description can be different and are not limited. Therefore, those skilled in the art of the present invention can modify or equivalently replace the technical solutions recorded in the foregoing embodiments; and these modifications and replacements do not deviate from the purpose and technical solutions of the present invention, and shall all fall within the protection scope of the present invention.

Claims

1. A consensus active neighbor optimization method for a large-scale distributed UAV cluster, characterized in that, Including: Step 1, determine the optional neighbors of the current UAV i; Step 2: If the number of candidate neighbors is less than or equal to the upper limit n of the neighbor set C , then add all candidate neighbors to the neighbor set to obtain the neighbor preference result of the current UAV i; If the number of optional neighbors is greater than the upper limit n of the neighbor set C , then perform Step 3; Step 3, first reduce the optional neighbors according to the collision safety distance: set the threshold lower than the safety spacing during the normal flight of the UAV cluster as the collision distance threshold; according to the UAV position information, screen out the optional neighbors from the set of optional neighbors whose distance from the current UAV i is less than the collision distance threshold, called too-close neighbors, so as to divide the optional neighbors into too-close neighbors and non-too-close neighbors, and add all the too-close neighbors to the neighbor set; Then, when the current neighbor set does not reach the upper limit n of the neighbor set C fill it up First, select a threshold r according to the spatial distance S , and filter out the optional neighbors whose distance from the current UAV i is less than r S as the nearest neighbors in terms of position; take the over-close neighbors and the nearest neighbors in terms of position as the preferred neighbor results; If the number of position nearest neighbors is greater than the replenishment upper limit n C,free , it is necessary to reduce the position nearest neighbors; at this time, further select the threshold r according to the state distance V , and filter out the optional neighbors whose state distance from the current UAV i is less than r V from the position nearest neighbors as the state nearest neighbors; use all or part of the too-close neighbors and the state nearest neighbors as the preferred neighbor results; if the number of the state nearest neighbors is greater than the replenishment upper limit n C,free when, then select n C,free state nearest state nearest neighbors with the closest states to join the neighbor set; wherein, the state refers to the speed of the drone; n C,free is the upper limit n of the neighbor set C minus the number of too-close neighbors.

2. The method according to claim 1, characterized in that, The specific steps of step 2 include steps 201 to 202, and step 3 includes steps 203 to 206: Step 201: The current drone i obtains information about optional neighbors, including the set of drone numbers The set of drone position information The set of drone speed information Step 202, obtain the total number of elements in the UAV number set If is less than or equal to the neighbor set upper limit n C , add all optional neighbors to the neighbor set and determine the neighbor number set Neighbor position set Neighbor speed set This process ends; otherwise, step 203 needs to be performed on the optional neighbors; Step 203, reduce the optional neighbors according to the collision safety distance; Set the collision distance threshold r coll Lower than the safety distance during the normal flight of the UAV swarm. According to the UAV position information set Calculate the optional neighbors whose distance from UAV i is less than r coll to form the too-close neighbor set N i,collision , and the other optional neighbors form the non-too-close neighbor set; If the total number of elements in the over-neighbor set |N i,collision | is greater than the neighbor set upper limit n C , all over-neighbors are added to the neighbor set, and the neighbor number set N i = N i,collision , the neighbor position set the neighbor speed set This process ends; otherwise, neighbors need to be supplemented in addition to over-neighbors, and the supplement upper limit is n C,free = n C - |N i,collision |, and step 204 is executed; where p j represents the UAV position, and v j represents the UAV speed; Step 204, screen the non-too-close neighbors according to the spatial distance; Set the spatial distance selection threshold as r S , and determine, according to the UAV position information set , the non - too - close neighbors whose distance from UAV i is less than r S to form the nearest - position neighbor set N i,spatial . If the total number of elements in the nearest - position neighbor set |N i,spatial | is less than or equal to the replenishment upper limit n C,free , then add the too - close neighbors and the nearest - position neighbors to the neighbor set, and output the neighbor number set N i = N i,spatial + N i,collision , the neighbor position set the neighbor speed set This process ends; otherwise, the number of nearest - position neighbors is more than the replenishment upper limit n C,free , and step 205 is executed; Step 205, reduce the nearest neighbor in position according to the state distance; Set the state distance selection threshold as r V , according to the UAV speed information set Determine the position nearest neighbor whose state distance from UAV i is less than r V to form the state nearest neighbor set N i,status , if the total number of elements in the position nearest neighbor set |N i,spatial | is less than or equal to the replenishment upper limit n C,free , then add the over-nearest neighbor and the state nearest neighbor to the neighbor set, and output the neighbor number set N i = N i,status + N i,collision , the neighbor position set the neighbor speed set This process ends; otherwise, the number of state nearest neighbors is more than the replenishment upper limit n C,free , execute step 206; Step 206: Sort the state distances in the set of state nearest neighbors, and select the n C,free state nearest neighbors with the smallest state distances to form the preferred neighbor set N i,order , and output the neighbor number set N i = N i,order + N i,collision , the neighbor position set the neighbor velocity set 3. The method according to any one of claims 1-2, characterized in that, The optional neighbors determined in step 1 for the current UAV i are: the UAVs that the current UAV i can receive information from are used as optional neighbors.

Images