CN114043476A - Swarm robot control method based on particle swarm algorithm under rejection environment - Google Patents

Swarm robot control method based on particle swarm algorithm under rejection environment Download PDF

Info

Publication number
CN114043476A
CN114043476A CN202111301771.0A CN202111301771A CN114043476A CN 114043476 A CN114043476 A CN 114043476A CN 202111301771 A CN202111301771 A CN 202111301771A CN 114043476 A CN114043476 A CN 114043476A
Authority
CN
China
Prior art keywords
robot
attacking
enemy
attack
environment
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Granted
Application number
CN202111301771.0A
Other languages
Chinese (zh)
Other versions
CN114043476B (en
Inventor
张军旗
刘欢
王成
臧笛
刘春梅
康琦
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Tongji University
Original Assignee
Tongji University
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by Tongji University filed Critical Tongji University
Priority to CN202111301771.0A priority Critical patent/CN114043476B/en
Publication of CN114043476A publication Critical patent/CN114043476A/en
Application granted granted Critical
Publication of CN114043476B publication Critical patent/CN114043476B/en
Active legal-status Critical Current
Anticipated expiration legal-status Critical

Links

Images

Classifications

    • B—PERFORMING OPERATIONS; TRANSPORTING
    • B25—HAND TOOLS; PORTABLE POWER-DRIVEN TOOLS; MANIPULATORS
    • B25J—MANIPULATORS; CHAMBERS PROVIDED WITH MANIPULATION DEVICES
    • B25J9/00—Programme-controlled manipulators
    • B25J9/16—Programme controls
    • B25J9/1679—Programme controls characterised by the tasks executed
    • B25J9/1682—Dual arm manipulator; Coordination of several manipulators

Landscapes

  • Engineering & Computer Science (AREA)
  • Robotics (AREA)
  • Mechanical Engineering (AREA)
  • Feedback Control In General (AREA)

Abstract

The invention relates to a swarm robot control method based on a particle swarm algorithm in a rejection environment, which comprises the following steps: step 1, establishing an attack and defense confrontation scene in a rejection environment, and initializing parameters of a particle swarm algorithm; step 2, the attacking robot detects surrounding environment information through a sensor, acquires situation information of friend and enemy robots, and calculates the position of enemy territory in real time by using an inertial navigation technology; step 3, constructing a fitness function by the attack robot; step 4, optimizing a fitness function by utilizing a particle swarm algorithm to obtain the optimal occupation of the attacking robot; step 5, moving and attacking the attacking robot; step 6, if any attacking robot enters enemy territory, the task is completed; otherwise, judging whether the maximum operation time is reached, if so, failing the task; otherwise, the iteration of the next time slice is carried out in the step 2. Compared with the prior art, the method avoids global positioning, does not need pre-training in control, and solves the problem of dimension disaster.

Description

Swarm robot control method based on particle swarm algorithm under rejection environment
Technical Field
The invention relates to the field of cooperative and game confrontation control among multi-agent clusters, in particular to a swarm robot control method based on a particle swarm algorithm in a rejection environment.
Background
With the rapid development and mature application of intelligent unmanned technology, cooperative combat among unmanned devices has become possible, and cooperative attack and defense confrontation among unmanned clusters gradually becomes an important mode of future war. As an application carrier of the multi-agent technology, the unmanned cluster judges the surrounding situation through sensing the environment, and takes the actions of fire gathering attack, wounded recoiling, obstacle avoidance, group collision avoidance, dispersion, concentration, cooperation, assistance and the like according to a certain attack and defense strategy to realize attack and defense confrontation.
The cooperative attack and defense confrontation of the group of robots can be described as an optimal decision problem under complex multi-constraint conditions, and the most classical is the problem of earth guard. In this problem, the confrontation environment is composed of two multiple intelligent groups of an intruder and a defender. Where the intruder is intended to try to get as close to and into a territory as possible and the defender is intended to intercept the intruder as far from the territory as possible. The advantages and disadvantages of the situation under the confrontation environment depend on the relationship among an invader, a defender and the territory, and because the state space dimension of the multi-agent attack-defense confrontation task is high, the strategy solving space is exponentially increased along with the increase of the scale of the entity object, the situation is complex and fast to change, attack-defense strategies are various, the solving difficulty is large, and an efficient decision algorithm is needed.
The most popular group confrontation method at present is a multi-agent deep reinforcement learning method. However, such algorithms require extensive pre-training, are limited by dimensional disaster problems, and rely on accurate global positioning and communication, failing to achieve effective collaboration and countermeasures in a denial environment.
Disclosure of Invention
The invention aims to overcome the defects in the prior art and provide a swarm robot control method based on a particle swarm algorithm in a rejection environment, and the control method makes up the bottleneck problem that global positioning and communication are limited by dimension disasters since most of multi-agent algorithms currently exist.
The purpose of the invention can be realized by the following technical scheme:
the invention provides a swarm robot control method based on a particle swarm algorithm in a rejection environment, which comprises the following steps:
step 1, establishing an attack and defense confrontation scene in a rejection environment, and initializing parameters of a particle swarm algorithm;
step 2, the attacking robot detects surrounding environment information through a sensor, acquires situation information of the friend robot and the enemy robot, and calculates the position of the enemy territory in real time by utilizing an inertial navigation technology
Figure BDA0003338687100000021
Step 3, the attacking robot constructs a fitness function containing enemy robot state information, friend robot state information and enemy territory information by using the detected surrounding environment information;
step 4, optimizing the fitness function by utilizing a particle swarm algorithm to obtain the optimal occupation of the attacking robot and guide the attacking robot to move and attack in the next time slice;
step 5, the attacking robot carries out moving and attacking operations;
step 6, if an attacking robot enters enemy territory, the task is completed; otherwise, judging whether the maximum running time is reached, if so, failing the task; otherwise, the iteration of the next time slice is carried out in the step 2.
Preferably, the attack and defense confrontation scene in the rejection environment is established in step 1, and parameters of the particle swarm algorithm are initialized, specifically: initializing the positions of N attacking robots, the positions of M defending robots and a GPS positioning coordinate U of enemy territory; initializing initial number of particles in particle swarm optimization
Figure BDA0003338687100000022
Acceleration factor c1And c2Inertial weight w and dimension of problem D;
all robots have the same attribute and have P-point blood volume, when the robots are attacked, the blood volume is reduced by P, and when the blood volume of the robots is less than or equal to 0, the robots are destroyed.
Preferably, the step 2 specifically comprises: each attacking robot constructs a coordinate system according to the position and the moving direction of the attacking robot; the method comprises the steps of detecting surrounding environment information through a sensor, obtaining coordinates of a friend robot and an enemy robot, and calculating the position of enemy territory in real time by utilizing an inertial navigation technology
Figure BDA0003338687100000023
Preferably, in step 3, the attack robot constructs a fitness function of the attack robot by using the detected ambient environment information, specifically: attacking robot AiConstructing a fitness function F according to the enemy robot state information, the friend robot state information and the enemy territory informationiThe expression is as follows:
Fi＝f1+f2+f3
wherein f is1Is a confrontation fitness f constructed according to the state information of the enemy robot2Is a cooperative fitness f constructed according to the state information of the friend robot3Is the adaptability of enemy to soil.
Preferably, said fitness opposition function f1The expression of (a) is:
Figure BDA0003338687100000031
where Ψ is the attacking robot AiNeighborhood region R1The indices of all enemy robots within,
Figure BDA0003338687100000032
is a sub-fitness function generated according to the state information of the kth enemy robot;
Figure BDA0003338687100000033
is the kth enemyCoordinates of the square robot; x ═ x1，x2) Is an independent variable representing the coordinate of a certain position in the rectangular coordinate system of the attack robot; sigma1And w1Respectively defining the width and the amplitude of the Gaussian-like model; s (i, k) is the measurement of the ith attack robot AiWhether the situation of (2) is better than that of the enemy robot BkDetermines the attacking robot AiIs to enemy robot BkWhether to move or to back is expressed as:
Figure BDA0003338687100000034
the situation information of the robot is obtained by calculation according to the number and the total blood volume of friend robots, wherein '-1' and '1' respectively represent an attacking robot AiIn unfavorable and favorable situations; n is a radical ofiAnd
Figure BDA0003338687100000035
respectively represent attack robots AiAttack range R0The number and total blood volume values of all friend robots within; mkAnd
Figure BDA0003338687100000036
respectively enemy robot BkThe number of all enemy robots and the total blood volume value within the attack range R0;
when S (i, k) — 1, fBkIs a valley-shaped function and represents the attacking robot AiRobot B away from enemykThe farther away, the attacking robot AiThe higher the fitness;
when S (i, k) is 1, fBkIs a peak-shaped function and represents the attacking robot AiWith enemy robot BkThe closer the approach, the higher the fitness of the attacking robot Ai.
Preferably, the attack robot AiIs a cooperative fitness function f2The expression of (a) is:
Figure BDA0003338687100000037
where Φ is the attacking robot AiNeighborhood region R1Indices of all friends in the content; f. ofAkIs a sub-fitness function generated according to the state information of the kth attack robot, wherein
Figure BDA0003338687100000038
Is the coordinates of the kth attacker; x ═ x1，x2) Is an independent variable representing the coordinate of a certain position in the rectangular coordinate system of the attack robot; sigma2And w2Respectively defining the width and the amplitude of the Gaussian-like model;
if attacking robot AiIn the time slices of tau, the attacking robot a is in a disadvantage facing the surrounding enemy robot all the timeiThe cooperative fitness function f needs to be released from the constraint of the fellow and quit the group2Set to 0, attack robot AiAct independently to search for better attack locations;
when A isiFriend with it AkWhen the distance between the two is less than a preset threshold value delta-10 | Ψ |, fAkSet to 0 to avoid collision of two attacking robots.
Preferably, the attack robot AiLand-taking fitness f of enemy3The expression is as follows:
Figure BDA0003338687100000041
wherein U ═ U (U)1，U2) Is the coordinate of the center position of the enemy territory, x ═ x1，x2) Is an independent variable representing the coordinate of a certain position in the rectangular coordinate system of the attack robot; sigma3And w3The width and amplitude of the gaussian-like model are defined separately.
Preferably, the step 4 specifically includes: in each time slice, executing a particle swarm algorithm to optimize the fitness function to obtain an attacking robot AiAt which pointOptimum position p in the former caseg(ii) a The search space of the particle swarm algorithm is an attacking robot AiIs the coordinate of (A) as the center, R1Is within a circular area of radius.
Preferably, the speed and location update expression of the fitness function is:
Figure BDA0003338687100000042
Figure BDA0003338687100000043
wherein, c1，c2Is a constant acceleration factor, w is an inertial weight,
Figure BDA0003338687100000044
represents the velocity of the ith particle in the D-dimension, D ∈ [1]，
Figure BDA0003338687100000045
Indicating the position of the ith particle in the d-dimension,
Figure BDA0003338687100000046
and
Figure BDA0003338687100000047
is a random number vector; d is the dimension of the environment, for a two-dimensional confrontation environment D-2 and a three-dimensional confrontation environment D-3.
Preferably, the step 4 specifically includes:
each attacking robot is oriented to the calculated optimal position pgDirection movement if the robot is in the optimal position pgIs less than the maximum distance of the attack robot moving in a time slice
Figure BDA0003338687100000048
The attacking robot moves to the optimal position pg(ii) a Otherwise, go to the optimal position pgDirection movement
Figure BDA0003338687100000049
A distance; in the moving process of the robot, if other robots exist in the warning area of the position of the next time slice, the moving direction of the robot is rotated anticlockwise by a degrees, and after the rotating times exceed the preset times, if the robot still cannot find a proper collision-free path, the robot stays at the current position until the next time slice;
if an enemy of the attacking robot enters the attacking range in the moving process, the nearest enemy is selected to attack; if the situation of the attacking robot is better than that of the enemy robot, the attacking robot can move towards the enemy robot and attack, namely the closer the attacking robot is to the enemy robot, the higher the fitness of the attacking robot is; conversely, if the attacking robot is inferior to the enemy robot in posture, the closer it is to the enemy robot, the lower its fitness;
the attacking robot and the friend of the attacking robot are in a cooperative relationship and form a group to attack the enemy robot group, and the closer the attacking robot is to the friend of the attacking robot, the higher the fitness of the attacking robot is.
Compared with the prior art, the invention has the following advantages:
1) the group robot senses the motion of the group robot from the surrounding environment by using the sensor carried by the group robot, and does not depend on a global navigation system; each robot constructs a respective coordinate system, and cooperation and confrontation are realized by acquiring relative coordinates of surrounding intelligent bodies without depending on a global positioning system;
2) the fitness function of the robot integrates information of friend, enemy and enemy territory, and the control of the robot in the cooperative and antagonistic environments is realized;
3) compared with a reinforcement learning method, the distributed control of the swarm robots based on the particle swarm algorithm has the advantages of no need of pre-training and strong expandability.
Drawings
FIG. 1 is a flow chart of a swarm robot control method based on a particle swarm algorithm in a rejection environment.
FIG. 2 is an exemplary diagram of a rectangular coordinate system of an attack robot;
FIG. 3 shows an attack robot A1An example of a fitness model of (1);
Detailed Description
The technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, not all, embodiments of the present invention. All other embodiments, which can be obtained by a person skilled in the art without any inventive step based on the embodiments of the present invention, shall fall within the scope of protection of the present invention.
The embodiment provides a swarm robot attacking method based on a particle swarm algorithm in a denial environment, and as shown in fig. 1, the method includes the following steps:
step 1, establishing an attack and defense confrontation scene in a refusal environment, setting the positions of N attack group robots, the positions of M defense group robots and the position coordinates of enemy territory, setting the attributes of all the robots to be the same, wherein the robots have 10-point blood volume, the blood volume is reduced by 1 when the robots are attacked, and the robots are killed when the blood volume is less than or equal to 0; initial number of particles for initializing particle swarm optimization
Figure BDA0003338687100000051
Acceleration factor c1And c2Inertial weight w and dimension D of the problem; all robots have the same attribute and have 10-point blood volume, when the robots are attacked, the blood volume is reduced by 1, and when the blood volume of the robots is less than or equal to 0, the robots are destroyed.
Step 2, the attacking robot detects the surrounding environment through a sensor, acquires situation information of the friend robot and the enemy robot, and calculates the position of the enemy territory in real time by utilizing the inertial navigation technology
Figure BDA0003338687100000052
Step 3, the attacking robot constructs a fitness function containing enemy robot state information, friend robot state information and enemy territory information by using the detected surrounding environment information;
as shown in FIG. 2, each attacking robot AiConstructing a rectangular coordinate system by taking the position of the mobile terminal as the origin of coordinates and the moving direction as the positive direction of the horizontal axis; wherein Qj、PkAnd U is the coordinates of the jth defender, kth attacker and territory respectively, attacking robot AiObtaining Q by sensing environment using its sensorjAnd Pk；
The coordinate U of the enemy territory is known when each attacking robot is at the initial position, and the approximate coordinate of the enemy territory is calculated in real time by applying the inertial navigation technology in the moving process of the robots
Figure BDA0003338687100000066
For each attacking robot, the fitness function comprises three parts of information: state information of the enemy robot, state information of the friend robot and enemy territory information;
the attack robot AiFitness function FiThe calculation formula of (2) is as follows:
Fi＝f1+f2+f3
wherein f is1Is a confrontation fitness f constructed according to the state information of the enemy robot2Is a cooperative fitness f constructed according to the state information of the friend robot3Is the adaptability of enemy to soil.
1) Constructing an confrontation fitness function f of the attack robot1
If the attacking robot is more situationally preferred than the enemy robot, it will move towards the enemy robot and initiate the attack. This situation is mapped into the fitness function model, i.e. the closer the attacking robot is to the enemy robot, the higher its fitness. Conversely, if the attacking robot is less dominant than the enemy robot, the closer it is to the enemy robot, the lower its fitness. The confrontation fitness function model is constructed by utilizing Gaussian-like distribution. Countervailing fitness function f of attack robot1Comprises the following steps:
Figure BDA0003338687100000061
where Ψ is the attacking robot AiNeighborhood region R1Index f of all enemy robots thereinBkIs a sub-fitness function generated from the state information of the kth enemy robot.
Figure BDA0003338687100000062
Is the coordinates of the kth enemy robot; x ═ x1，x2) Is an independent variable representing the coordinate of a certain position in a rectangular coordinate system; sigma1And w1The width and amplitude of the gaussian-like model are defined separately.
Figure BDA0003338687100000063
S (i, k) is the measurement of the ith attack robot AiWhether the situation of (2) is better than that of the enemy robot BkThe index of (1). The situation of the robot is calculated according to the number of friends and the total blood volume, wherein "-1" and "1" represent A respectivelyiIn unfavorable and favorable situations; n is a radical ofiAnd
Figure BDA0003338687100000064
is AiAttack range R0Attack robot (including attacker A)i) Number of and total blood volume values; mkAnd
Figure BDA0003338687100000065
are respectively BkAttack range R0Number of enemy robots within and total blood volume value. When S (i, k) — 1, fBkIs a valley-shaped function. This means that AiFrom BkThe farther away, AiThe higher the fitness. When S (i, k) is 1, fBkIs a peak-shaped function. This means that AiAnd BkThe closer to AiThe higher the fitness.S (i, k) determines AiIs directed to BkWhether to move or to back.
2) Construction of attack robot AiIs a cooperative fitness function f2
The attacking robot and its friend are in a cooperative relationship and form a group to attack the enemy robot population. This situation is mapped into the fitness function model, i.e. the closer the attacking robot is to its friend, the higher its fitness. Attacking robot AiIs a cooperative fitness function f2Comprises the following steps:
Figure BDA0003338687100000071
wherein Φ is AiNeighborhood region R1Index of all friends in the content. f. ofAkIs a sub-fitness function generated according to the state information of the kth attack robot, wherein
Figure BDA0003338687100000072
Is the coordinates of the kth attacker. Notably, according to f2The attacking population may dynamically form subgroups. If A isiWithin a period of tau time, a is always at a disadvantage facing surrounding enemy robots, then aiIt is necessary to escape from the constraints of its partners and exit the population, at which point f will be2Set to 0, Ai acts independently to search for a better attack position. When A isiFriend with it AkWhen the distance therebetween is less than the threshold value Δ ═ 10| Ψ |, fAkSet to 0 to avoid collision of two attacking robots.
3) Constructing a fitness function f according to the position information of enemy territory3
The attacking robot aims to enter enemy territory, and the closer the attacking robot is to the enemy territory, the higher the fitness is. Therefore, the fitness function f3Comprises the following steps:
Figure BDA0003338687100000073
wherein U ═ U (U)1，U2) Is the coordinate of the center position of the enemy territory.
FIG. 3 shows an attacking robot A1An example of a fitness model of (1), where two diamonds represent A1And A2Three triangles are enemy robot B1、B2And B3The star is the enemy territory T. Wherein A is1And A2Are in cooperative relationship with each other, A1And B1、B2And B3The relationship between is a resistance relationship. The fitness model is applied to a particle swarm algorithm to optimize and obtain a group robot attack strategy in a rejection environment so as to guide the movement and attack of an attacking robot.
Step 4, optimizing the fitness function by utilizing a particle swarm algorithm to obtain the optimal occupation of the attacking robot and guide the attacking robot to move and attack in the next time slice;
in each time slice, a particle swarm optimization PSO optimization attack robot A is executediBest position p in its current situationg. Since the attacker's distance of movement is limited within a time slice and the environment is dynamically changing, it only needs to be in its neighborhood R1Finding the position with the best fitness.
Therefore, the search space of the particle swarm algorithm PSO is constrained to attack robot aiIs the coordinate of (A) as the center, R1Is within a circular area of radius.
The speed and location update formula for optimizing the fitness function is as follows:
Figure BDA0003338687100000074
Figure BDA0003338687100000081
wherein, c1，c2Is a constant acceleration factor, w is an inertial weight,
Figure BDA0003338687100000082
represents the velocity of the ith particle in the D-dimension, D ∈ [1]，
Figure BDA0003338687100000083
Indicating the position of the ith particle in the d-dimension,
Figure BDA0003338687100000084
and
Figure BDA0003338687100000085
is a random number vector; d is the dimension of the environment, for a two-dimensional confrontation environment D-2, a three-dimensional confrontation environment D-3;
step 5, moving and attacking the attacking robot;
each robot is directed to the calculated optimal position pgDirection movement if the robot is in the optimal position pgIs less than the maximum distance the robot moves within a time slice
Figure BDA0003338687100000086
The robot moves to the optimal position pg(ii) a Otherwise to pgDirection movement
Figure BDA0003338687100000087
Distance.
In the moving process, if the robot has enemy to enter the attack range in the moving process, the nearest enemy attack to the enemy is selected. During the movement of the robot, if other robots exist in the warning area of the position of the next time slice, the moving direction of the robot is rotated by 15 degrees anticlockwise, and after a series of rotations (23 times), if the robot still cannot find a proper collision-free path, the robot stays at the current position until the next time slice.
Step 6, if an attacking robot enters enemy territory, the task is completed; otherwise, judging whether the maximum running time is reached, if so, failing the task; otherwise, the iteration of the next time slice is carried out in the step 2.
In order to verify the performance of the method in the group robot territory invasion problem more intuitively, the method carries out the antagonism experiment with the following three algorithms applied to the enemy robot group:
(1) based on a rule algorithm, the enemy robot always moves to the middle point position of the attack robot closest to the territory and the enemy robot to intercept the attack robot.
(2) DPSO attack task assignment algorithm, "Cooperative Multi-task assignment for multiple UAVs," Electronics Optics & controls, vol.24, No.1, pp.46-50, 2017.
(3) SDPSO attack task allocation algorithm, "UAV cooperative multiple-task assistance based on discrete particle timing algorithm," Computer Simulation, vol.35, No.2, pp.22-28, 2018.
The challenge results of the algorithm are shown in table 1.
TABLE 1
Figure BDA0003338687100000088
Figure BDA0003338687100000091
It is apparent that the overall odds of the attacking robot group using the PSO-AS method is 100% when the numbers of the attacking robot group and the enemy robot group are the same. When the number of attacking robot groups is only 75% of the enemy robot groups, the proposed method still has a success rate of winning of more than 50%.
While the invention has been described with reference to specific embodiments, the invention is not limited thereto, and various equivalent modifications and substitutions can be easily made by those skilled in the art within the technical scope of the invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.

Claims (10)

1. A swarm robot control method based on a particle swarm algorithm under a rejection environment is characterized by comprising the following steps:
step 1, establishing an attack and defense confrontation scene in a rejection environment, and initializing parameters of a particle swarm algorithm;
step 2, the attacking robot detects surrounding environment information through a sensor, acquires situation information of the friend robot and the enemy robot, and calculates the position of the enemy territory in real time by utilizing an inertial navigation technology
Figure FDA0003338687090000011
Step 3, the attacking robot constructs a fitness function containing enemy robot state information, friend robot state information and enemy territory information by using the detected surrounding environment information;
step 4, optimizing the fitness function by utilizing a particle swarm algorithm to obtain the optimal occupation of the attacking robot and guide the attacking robot to move and attack in the next time slice;
step 5, the attacking robot carries out moving and attacking operations;
step 6, if an attacking robot enters enemy territory, the task is completed; otherwise, judging whether the maximum running time is reached, if so, failing the task; otherwise, the iteration of the next time slice is carried out in the step 2.
2. The method for controlling group robots based on particle swarm optimization in the denial environment according to claim 1, wherein the attack and defense confrontation scene in the denial environment is established in step 1, and parameters of the particle swarm optimization are initialized, specifically: initializing the positions of N attacking robots, the positions of M defending robots and a GPS positioning coordinate U of enemy territory; initializing initial number of particles in particle swarm optimization
Figure FDA0003338687090000013
Acceleration factor c1And c2Inertial weight w and dimension of problem D;
all robots have the same attribute and have P-point blood volume, when the robots are attacked, the blood volume is reduced by P, and when the blood volume of the robots is less than or equal to 0, the robots are destroyed.
3. The swarm robot control method based on the particle swarm algorithm in the denial environment according to claim 1, wherein the step 2 specifically comprises: each attacking robot constructs a coordinate system according to the position and the moving direction of the attacking robot; the method comprises the steps of detecting surrounding environment information through a sensor, obtaining coordinates of a friend robot and an enemy robot, and calculating the position of enemy territory in real time by utilizing an inertial navigation technology
Figure FDA0003338687090000012
4. The swarm robot control method based on the particle swarm algorithm in the denial environment according to claim 1, wherein the attacking robot in step 3 uses the detected ambient environment information to construct a fitness function of the attacking robot, specifically: attacking robot AiConstructing a fitness function F according to the enemy robot state information, the friend robot state information and the enemy territory informationiThe expression is as follows:
Fi＝f1+f2+f3
wherein f is1Is a confrontation fitness f constructed according to the state information of the enemy robot2Is a cooperative fitness f constructed according to the state information of the friend robot3Is the adaptability of enemy to soil.
5. The swarm robot control method based on particle swarm optimization in the rejection environment of claim 4, wherein the countervailing fitness function f is1The expression of (a) is:
Figure FDA0003338687090000021
where Ψ is the attacking robot AiNeighborhood region R1The indices of all enemy robots within,
Figure FDA0003338687090000022
is a sub-fitness function generated according to the state information of the kth enemy robot;
Figure FDA0003338687090000023
is the coordinates of the kth enemy robot; x ═ x1,x2) Is an independent variable representing the coordinate of a certain position in the rectangular coordinate system of the attack robot; sigma1And w1Respectively defining the width and the amplitude of the Gaussian-like model; s (i, k) is the measurement of the ith attack robot AiWhether the situation of (2) is better than that of the enemy robot BkDetermines the attacking robot AiIs to enemy robot BkWhether to move or to back is expressed as:
Figure FDA0003338687090000024
the situation information of the robot is obtained by calculation according to the number and the total blood volume of friend robots, wherein '-1' and '1' respectively represent an attacking robot AiIn unfavorable and favorable situations; n is a radical ofiAnd
Figure FDA0003338687090000025
respectively represent attack robots AiAttack range R0The number and total blood volume values of all friend robots within; mkAnd
Figure FDA0003338687090000026
respectively enemy robot BkAttack range R0All enemy machines withinNumber of people and total blood volume values;
when S (i, k) — 1, fBkIs a valley-shaped function and represents the attacking robot AiRobot B away from enemykThe farther away, the attacking robot AiThe higher the fitness;
when S (i, k) is 1, fBkIs a peak-shaped function and represents the attacking robot AiWith enemy robot BkThe closer the attacking robot A isiThe higher the fitness.
6. The swarm robot control method based on particle swarm optimization in the denial environment of claim 4, wherein the attacking robot AiIs a cooperative fitness function f2The expression of (a) is:
Figure FDA0003338687090000027
where Φ is the attacking robot AiNeighborhood region R1Indices of all friends in the content; f. ofAkIs a sub-fitness function generated according to the state information of the kth attack robot, wherein
Figure FDA0003338687090000028
Is the coordinates of the kth attacker; x ═ x1,x2) Is an independent variable representing the coordinate of a certain position in the rectangular coordinate system of the attack robot; sigma2And w2Respectively defining the width and the amplitude of the Gaussian-like model;
if attacking robot AiIn the time slices of tau, the attacking robot a is in a disadvantage facing the surrounding enemy robot all the timeiThe cooperative fitness function f needs to be released from the constraint of the fellow and quit the group2Set to 0, attack robot AiAct independently to search for better attack locations;
when A isiFriend with it AkWhen the distance between the two is less than a preset threshold value delta-10 | Ψ |, fAkSet to 0 to avoid collision of two attacking robots.
7. The swarm robot control method based on particle swarm optimization in the denial environment as claimed in claim 4, wherein the attacking robot AiLand-taking fitness f of enemy3The expression is as follows:
Figure FDA0003338687090000031
wherein U ═ U (U)1,U2) Is the coordinate of the center position of the enemy territory, x ═ x1,x2) Is an independent variable representing the coordinate of a certain position in the rectangular coordinate system of the attack robot; sigma3And w3The width and amplitude of the gaussian-like model are defined separately.
8. The swarm robot control method based on the particle swarm algorithm in the denial environment according to claim 1, wherein the step 4 specifically comprises: in each time slice, executing a particle swarm algorithm to optimize the fitness function to obtain an attacking robot AiBest position p in its current situationg(ii) a The search space of the particle swarm algorithm is an attacking robot AiIs the coordinate of (A) as the center, R1Is within a circular area of radius.
9. The swarm robot control method based on the particle swarm optimization algorithm in the rejection environment according to claim 8, wherein the speed and location update expression of the fitness function is as follows:
Figure FDA0003338687090000032
Figure FDA0003338687090000033
wherein, c1，c2Is a constant acceleration factor, w is an inertial weight,
Figure FDA0003338687090000034
represents the velocity of the ith particle in the D-dimension, D ∈ [1]，
Figure FDA0003338687090000035
Indicating the position of the ith particle in the d-dimension,
Figure FDA0003338687090000036
and
Figure FDA0003338687090000037
is a random number vector; d is the dimension of the environment, for a two-dimensional confrontation environment D-2 and a three-dimensional confrontation environment D-3.
10. The swarm robot control method based on the particle swarm algorithm in the denial environment according to claim 8, wherein the step 4 specifically comprises:
each attacking robot is oriented to the calculated optimal position pgDirection movement if the robot is in the optimal position pgIs less than the maximum distance of the attack robot moving in a time slice
Figure FDA0003338687090000038
The attacking robot moves to the optimal position pg(ii) a Otherwise, go to the optimal position pgDirection movement
Figure FDA0003338687090000039
A distance; in the moving process of the robot, if other robots exist in the warning area of the position where the next time slice is located, the moving direction of the robot is rotated anticlockwise by a degrees, and after the rotating times exceed the preset times, if the robot still cannot find a proper collision-free path, the robot stays onCurrent position until the next time slice;
if an enemy of the attacking robot enters the attacking range in the moving process, the nearest enemy is selected to attack; if the situation of the attacking robot is better than that of the enemy robot, the attacking robot can move towards the enemy robot and attack, namely the closer the attacking robot is to the enemy robot, the higher the fitness of the attacking robot is; conversely, if the attacking robot is inferior to the enemy robot in posture, the closer it is to the enemy robot, the lower its fitness;
the attacking robot and the friend of the attacking robot are in a cooperative relationship and form a group to attack the enemy robot group, and the closer the attacking robot is to the friend of the attacking robot, the higher the fitness of the attacking robot is.
CN202111301771.0A 2021-11-04 2021-11-04 Swarm robot control method based on particle swarm algorithm under rejection environment Active CN114043476B (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN202111301771.0A CN114043476B (en) 2021-11-04 2021-11-04 Swarm robot control method based on particle swarm algorithm under rejection environment

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN202111301771.0A CN114043476B (en) 2021-11-04 2021-11-04 Swarm robot control method based on particle swarm algorithm under rejection environment

Publications (2)

Publication Number Publication Date
CN114043476A true CN114043476A (en) 2022-02-15
CN114043476B CN114043476B (en) 2023-03-31

Family

ID=80207292

Family Applications (1)

Application Number Title Priority Date Filing Date
CN202111301771.0A Active CN114043476B (en) 2021-11-04 2021-11-04 Swarm robot control method based on particle swarm algorithm under rejection environment

Country Status (1)

Country Link
CN (1) CN114043476B (en)

Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007080584A2 (en) * 2006-01-11 2007-07-19 Carmel-Haifa University Economic Corp. Ltd. Uav decision and control system
CN101908097A (en) * 2010-07-13 2010-12-08 北京航空航天大学 Particle swarm optimization method for air combat decision
CN111176309A (en) * 2019-12-31 2020-05-19 北京理工大学 Multi-unmanned aerial vehicle self-group mutual inductance understanding method based on spherical imaging
CN113052289A (en) * 2021-03-16 2021-06-29 东南大学 Unmanned ship cluster striking position selection method based on game theory
CN113378988A (en) * 2021-07-06 2021-09-10 浙江工业大学 Deep learning system robustness enhancement method and device based on particle swarm optimization

Patent Citations (5)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
WO2007080584A2 (en) * 2006-01-11 2007-07-19 Carmel-Haifa University Economic Corp. Ltd. Uav decision and control system
CN101908097A (en) * 2010-07-13 2010-12-08 北京航空航天大学 Particle swarm optimization method for air combat decision
CN111176309A (en) * 2019-12-31 2020-05-19 北京理工大学 Multi-unmanned aerial vehicle self-group mutual inductance understanding method based on spherical imaging
CN113052289A (en) * 2021-03-16 2021-06-29 东南大学 Unmanned ship cluster striking position selection method based on game theory
CN113378988A (en) * 2021-07-06 2021-09-10 浙江工业大学 Deep learning system robustness enhancement method and device based on particle swarm optimization

Also Published As

Publication number Publication date
CN114043476B (en) 2023-03-31

Similar Documents

Publication Publication Date Title
CN114413906B (en) Three-dimensional trajectory planning method based on improved particle swarm optimization algorithm
CN109144102A (en) A kind of Path Planning for UAV based on improvement bat algorithm
Goldhoorn et al. Searching and tracking people with cooperative mobile robots
CN111381600B (en) UUV path planning method based on particle swarm optimization
EP2010981A2 (en) Method and system for autonomous tracking of a mobile target by an unmanned aerial vehicle
CN113342047A (en) Unmanned aerial vehicle path planning method for improving artificial potential field method based on obstacle position prediction in unknown environment
CN110633857B (en) Autonomous defense clustering algorithm for heterogeneous unmanned aerial vehicle cluster
CN113848984B (en) Unmanned aerial vehicle cluster control method and system
Rubí et al. Quadrotor path following and reactive obstacle avoidance with deep reinforcement learning
CN114063644A (en) Unmanned combat aircraft air combat autonomous decision method based on pigeon flock reverse confrontation learning
Saranovic et al. Interception of automated adversarial drone swarms in partially observed environments
Wen et al. Optimization of the occlusion strategy in visual tracking
Liang et al. Collaborative pursuit-evasion strategy of UAV/UGV heterogeneous system in complex three-dimensional polygonal environment
CN114043476B (en) Swarm robot control method based on particle swarm algorithm under rejection environment
Zheng et al. Coevolving and cooperating path planner for multiple unmanned air vehicles
CN115061495B (en) Unmanned aerial vehicle group countermeasure autonomous control method based on eagle pigeon game
CN115686075A (en) Unmanned aerial vehicle cluster multi-target collaborative search algorithm based on combination of particle swarm and artificial potential field
Liu et al. Trajectories planning for multiple UAVs by the cooperative and competitive PSO algorithm
CN113507339B (en) Unmanned aerial vehicle swarm countering method based on swarm behavior characteristics
CN112925317B (en) AUV path planning method based on improved brainstorming optimization algorithm
Ranade et al. Quadcopter obstacle avoidance and path planning using flood fill method
Gudi et al. Bio-inspired evasive movement of uavs based on dragonfly algorithm in military environment
Simonjan et al. Reinforcement Learning-based Countermeasures against Attacking UAV Swarms
Xi et al. Trajectory generation for quadrotor while tracking a moving target in cluttered environment
CN117313561B (en) Unmanned aerial vehicle intelligent decision model training method and unmanned aerial vehicle intelligent decision method

Legal Events

Date Code Title Description
PB01 Publication
PB01 Publication
SE01 Entry into force of request for substantive examination
SE01 Entry into force of request for substantive examination
GR01 Patent grant
GR01 Patent grant