AUV formation method, device, computer system and readable storage medium

Abstract

The present invention relates to an AUV formation method, device, computer system and readable storage medium, and belongs to the field of automatic control technology. AUV individuals are divided into navigators, coordinators and followers, and the functions and coordination strategies of different roles are designed, which are embedded in the distributed model predictive control (DMPC) framework. The cost functions of different roles are calculated, and the coordination control problem is converted into an optimal control problem. The dispersed AUV groups are gathered to the target area and maintain the desired formation. The distributed coordination strategy does not rely on a fixed formation and can quickly make adjustments to form a new formation. When a member has a problem, it has little impact on the overall function, and the completed new formation can continue to complete the task. The distributed model predictive control significantly reduces the amount of calculation by allocating the overall optimization problem to each subsystem.

Description

Technical Field

[0001] This invention belongs to the field of automatic control technology and relates to a distributed model-based predictive formation control method, device, computer system, and readable storage medium for multiple underwater vehicles. Specifically, it embeds a task role allocation and handover coordination strategy into a distributed model-based predictive formation control framework to converge dispersed groups of vehicles into a target area. Background Art

[0002] Autonomous underwater vehicles (AUVs), freed from the constraints of tethers, have a large range (up to thousands of kilometers) and require no extensive surface support or human intervention, making them economical and safe tools ideal for seafloor exploration, identification, and operations. While individual AUVs offer advantages such as high flexibility and good concealment during deep-sea missions, as exploration missions continue to expand, their individual capabilities are limited, making them incapable of completing complex tasks.

[0003] Compared with a single AUV, a multi-AUV formation can enhance its operational capabilities and improve its operational efficiency through mutual collaboration due to its functional redundancy and spatial distribution, and complete complex operational tasks, thus making up for the shortcomings of a single AUV, such as small carrying capacity, limited energy power, and low exploration efficiency.

[0004] With the increasing complexity of tasks and autonomous capabilities, large robot swarm systems consisting of limited autonomy and local communication have gradually attracted the attention of many researchers and have been applied to various applications such as target search and tracking, surveillance, and collaborative aerial mapping. The main characteristic of robot swarms is their distributed structure, which means that each robot can only interact within a certain range and obtain local information. Therefore, they have the potential to achieve a high degree of autonomous collaboration without the need for centralized control, significantly enhancing flexibility and robustness. To achieve this, the formation control problem is often regarded as a fundamental task for robot swarms to achieve ideal clustering behavior. Summary of the Invention

[0005] Technical problems to be solved

[0006] In order to avoid the shortcomings of the prior art, the present invention provides a distributed model predictive coordinated control method, device, computer system and readable storage medium based on swarm intelligence.

[0007] Technical Solution

[0008] An AUV formation method, characterized by the following steps:

[0009] S1: Design different roles for the AUV swarm and specify coordination mechanisms;

[0010] S2: Establish a distributed model predictive control framework;

[0011] S3: Determine the cost functions of different roles and substitute them into the distributed model predictive control framework to transform the coordinated control problem into an optimal control problem; solve the optimal control input, that is, the current motion trend of the AUV, and use its first step as the control input at the next moment to perform the next optimal control;

[0012] S4: Each AUV continuously updates its state according to the control vector at each moment, forming a motion trajectory in which the state changes over time, and then realizes the clustering task based on swarm intelligence to form the desired structural formation.

[0013] The roles include navigator, coordinator and follower. The navigator focuses on tracking the virtual guidance point and leads the entire formed group so that the entire team can keep up with the virtual guidance point as much as possible; the coordinator coordinates the neighbors around it to move toward the corresponding vertices of the square with itself as the geometric center; the follower finds the coordinator among its neighbors and navigates itself according to the coordinator's position allocation information.

[0014] The coordination mechanism described is as follows: initially all AUVs are set as followers of a virtual navigation point, and then coordinators and navigators are gradually formed according to the role division. The AUV group gathers to the target area and maintains the desired formation.

[0015] The desired formation is a square crystalline lattice structure.

[0016] The strategy of the distributed model predictive control framework is that in each control cycle of the DMPC, the AUVs communicate and search for their neighbors based on the communication distance R. The AUVs then assign roles to themselves according to the coordination mechanism so that multiple AUVs can sail in an orderly group. After transmitting information to their neighbors, each AUV solves its own optimal control problem within the current prediction range and executes the first control variable of the optimal control sequence.

[0017] An AUV formation method and formation device, characterized by comprising:

[0018] Design module, used to design different roles of the AUV group and specify the coordination mechanism;

[0019] Establish a module for building a distributed model predictive control framework;

[0020] The determination module is used to determine the cost functions of different roles and substitute them into the distributed model predictive control framework to transform the coordinated control problem into an optimal control problem;

[0021] The solution module is used to solve the optimal control input, that is, the current motion trend of the AUV, and use its first step length as the control input at the next moment to perform the next optimal control;

[0022] The output module is used to output the distributed model prediction and coordinated control results.

[0023] A computer system, characterized in that it includes: one or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement the above-mentioned method.

[0024] A computer-readable storage medium is characterized by storing computer-executable instructions, which are used to implement the above method when executed.

[0025] A computer program, characterized by comprising computer executable instructions, wherein the instructions are used to implement the method as claimed in claim 1 when the instructions are executed.

[0026] Beneficial effects

[0027] Compared with the existing formation technology, the AUV formation method proposed in this invention has the following beneficial effects:

[0028] 1. High robustness and low computational complexity.

[0029] Because the distributed coordination strategy does not rely on a fixed formation, it can quickly make adjustments to form a new formation. When a problem occurs in one member, it has little impact on the overall function, and the completed new formation can continue to complete the task. Distributed model predictive control significantly reduces the amount of computation by distributing the overall optimization problem to each subsystem.

[0030] 2. Individual scalability

[0031] The formation behavior of AUVs in the present invention is based on dividing roles and then specifying a coordination mechanism, ultimately obtaining a desired formation of a square grid structure, which is conducive to forming an expandable group and no longer limits the number of AUVs in the formation.

[0032] 3. Wide range of applications

[0033] This method is not only applicable to the AUV formation problem based on underwater acoustic communication at any depth, but can also be applied to the intelligent agent formation problem under land or air combat conditions where the communication frequency cannot be too high. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings are only for the purpose of illustrating particular embodiments and are not to be considered limiting of the present invention. Like reference symbols denote like parts throughout the drawings.

[0035] Figure 1 Formation process expected formation example;

[0036] Figure 2 Distributed model predictive control architecture framework;

[0037] Figure 3 AUV neighborhood structure diagram;

[0038] Figure 4 Schematic diagram of the formation of different members of the AUV group;

[0039] Figure 5 Schematic diagram of coordinator position allocation;

[0040] Figure 6 The generation process of the coordinator;

[0041] Figure 7 The process of generating a navigator (I);

[0042] Figure 8 The process of generating a navigator (II);

[0043] Figure 9 A single follower follows the coordinator process;

[0044] Figure 10 A single follower follows the process of followers in a group;

[0045] Figure 11 Coordinator follows coordinator process;

[0046] Figure 12 AUV formation navigation trajectory diagram

[0047] Figure 13 A graph showing the distance changes between the coordinator and its neighbors;

[0048] Figure 14 Structural diagram of the device of the present invention. DETAILED DESCRIPTION

[0049] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below with reference to the accompanying drawings and examples. It should be understood that the specific embodiments described herein are only intended to illustrate the present invention and are not intended to limit the present invention. In addition, the technical features involved in the various embodiments of the present invention described below may be combined with each other as long as they do not conflict with each other.

[0050] The embodiment of the present invention provides a distributed model predictive coordinated control method based on swarm intelligence, which divides AUV individuals into leaders, coordinators and followers, and designs the functions and coordination strategies of different roles, embeds them into the distributed model predictive control (DMPC) framework, calculates the cost functions of different roles, transforms the coordinated control problem into an optimal control problem, and gathers the dispersed AUV groups to the target area and maintains the desired formation. The formation process of the desired formation is as follows: Figure 1 shown.

[0051] The implementation steps are as follows:

[0052] S1: Design AUVs as leaders, coordinators, and followers, and define a coordination mechanism. The leader focuses on tracking a virtual guidance point, leading the entire group and ensuring the entire team can keep up with the virtual guidance point as much as possible. The coordinator coordinates its neighbors to move toward the corresponding vertices of a square with itself as the geometric center. Followers locate the coordinator among their neighbors and navigate themselves based on the coordinator's position information. Initially, all AUVs are configured as followers of the virtual guidance point. Then, based on the role division, the coordinator and leader are gradually formed. The AUV group gathers at the target area and maintains the desired formation.

[0053] S2: Establish a distributed model predictive control framework. The distributed coordination strategy is that in each control cycle of DMPC, AUVs communicate and search for their neighbors based on the communication distance R. Then, AUVs assign themselves roles (leader, coordinator, or follower) according to the coordination mechanism given in step S1 so that multiple AUVs can sail in a group in an orderly manner. After transmitting information to their neighbors, each AUV solves its own optimization control problem within the current prediction range and executes the first control quantity of the optimal control sequence. The distributed model predictive control structure framework is as follows: Figure 2 As shown. At each update time t c , AUV i The current state is used as the initial value of the predicted state in the prediction time domain at this moment, and then according to t c-1 The optimal control input sequence at the time instant is used to obtain the estimated control input sequence within the current prediction time domain. The estimated sequence and the current state value are then exchanged with the adjacent AUVs. After this, each AUV calculates the estimated state variable sequence of its neighboring AUVs within the current prediction period and substitutes it into its own cost function to solve for the optimal control sequence within the current prediction period. The first control input is then used to update the control input for the current control period.

[0054] S3: Determine the cost functions for different roles and substitute them into the model prediction framework to transform the coordinated control problem into an optimal control problem. The optimal control input at that moment is obtained, i.e., the current motion trend of the AUV. The first step length is used as the control input for the next moment to perform the next optimal control.

[0055] S4: Output the distributed model prediction coordinated control results. Each AUV continuously updates its state according to the control vector at each moment, forming a motion trajectory in which the state changes over time, and then realizes the cluster task based on swarm intelligence to form the desired square crystalline lattice structure formation.

[0056] In an optional optimal implementation, the above implementation steps are specifically as follows:

[0057] S1: Design different roles and functions for the AUV swarm and define coordination mechanisms

[0058] S1-1: Define neighbors and their relative expected positions

[0059] Assume that the multi-AUV system consists of N AUVs, and the i-th AUV is denoted as AUV i , suppose AUV i The maximum communication distance is D i , assuming that the multi-AUV system has ideal communication conditions and can communicate with its neighbors in two directions in each control cycle, its communication range is centered on itself, D i At time t, the AUV i The position vector is p i (t) = [x i (t),y i (t)] T , where x i (t) is AUV i The horizontal axis, y i (t) is AUV i The vertical coordinate of AUV can be obtained by the same logic. j The position vector is p j (t), if ||p i (t)-p j (t)||≤D i , then at time t, the AUV i With AUV j They are neighbors and can communicate with each other.

[0060] The desired formation is a square lattice structure, so the AUV can have at most 4 neighbors given the desired distance between neighbors is D R (D R <communication radius R), each AUV generates 4 relative positions to be occupied and numbers them in sequence, such as Figure 3 shown.

[0061] S1-2: Designing different roles for AUV

[0062] Define three different target types of members: leader, coordinator, and follower. Figure 4 A schematic diagram of a group with different members tracking a reference trajectory is given.

[0063] According to the scenario where the group moves from dispersion to gathering and heading to the target area, it is assumed that there is a virtual guide point that maintains constant linear motion in the direction from the initial position of the group to the target area, and the AUV cluster follows.

[0064] Therefore, the update of the virtual guidance point position of the group is described as follows.

[0065] P exp (s;t c )=P exp (t c )+v exp ×(st c ) (1)

[0066] Among them, v exp is the velocity of the virtual guidance point.

[0067] Navigator: Focus on tracking the virtual guidance point, lead the entire formed group, and enable the entire team to keep up with the virtual guidance point as much as possible.

[0068] Coordinator: coordinates its neighbors to move toward the corresponding vertices of the square with itself as the geometric center; Figure 5 In the example, the coordinator will follow the leader, which is the position of neighbor 1, and the other three positions are assigned to the three AUV neighbors to guide them to the designated positions.

[0069] Follower: Find the coordinator among its neighbors and navigate itself based on the coordinator's location distribution information;

[0070] S1-3: Establish coordination mechanisms

[0071] Start by setting all AUVs as followers of the virtual navigation point.

[0072] (1) Generation of Coordinators

[0073] Follower AUV i and follower AUV j There are no neighbors before entering each other's communication range. The follower that is closer to the virtual guidance point is changed to the coordinator, and the other AUV will become its follower. The coordinator will then assign positions to the followers, such as Figure 6As shown in the figure, the coordinator also becomes the local leader of the current group.

[0074] (2) Generation of Navigators

[0075] Designing the AUV i The distance from the virtual guidance point is less than D R / 2, changes to the navigator. Figure 7 As shown, when other AUVs enter the circle with the virtual guidance point as the center and D R / 2 radius area circle will first be with AUV i They communicate and interact with each other, so there will be no situation where two AUVs compete for the leader. Assume that the distance between the two AUVs and the virtual guidance point is D R / 2, such as Figure 8 As shown, the two AUVs are now definitely in a group. According to their following relationship, the followed AUV becomes the leader. Upon encountering the leading AUV, the other AUVs become coordinators and accept the leader's positional arrangements. The leader also sends a special message to the coordinators, indicating its presence in the group, and gradually spreads this message to the other AUVs in the group. When other AUVs encounter the leader, they must accept its leadership.

[0076] (3) A single follower encounters a coordinator

[0077] Once the follower enters the communication range of the coordinator, the coordinator will check the number of its current vacant positions and generate its coordinates relative to the coordinator's current position. Then, the coordinator selects an empty position as the allocation result based on the angle. If there is already an AUV at the assigned position, the AUV will select the closest position among the remaining positions and assign it to the new follower, such as Figure 9 shown.

[0078] (4) A single follower encounters a follower in a group

[0079] When a single follower enters the communication range of a follower in the group, the follower in the group will be changed to a coordinator. Then the follower AUV will be assigned a position, and the coordinator AUV will coordinate the entire formation with the coordinator AUV it originally followed, such as Figure 10 shown.

[0080] (5) Coordinator meets coordinator

[0081] When two coordinators meet, the group with a leader (if neither has a leader, the group whose followers are closer to the virtual guidance point is selected) becomes the leading group and dominates the position allocation. The local navigators in the led group will no longer follow the virtual guidance point, but coordinate the group's formation with other coordinators, such as Figure 11 shown.

[0082] (6) Followers in the group meet followers in the group

[0083] When a follower in a group enters the communication range of a follower in another group, both followers become coordinators. The entire process is then the same as in case 5.

[0084] (7) Followers in the group encounter the coordinator

[0085] When a follower in the group enters the communication range of the coordinator, the follower will change to the coordinator. Then the whole process is the same as the fifth case.

[0086] S2: Building a distributed model predictive control framework

[0087] The distributed coordination strategy is that in each control cycle of the DMPC, the AUVs communicate and search for their neighbors based on the communication distance R, and then the AUVs assign themselves roles (leader, coordinator, or follower) according to certain rules, so that multiple AUVs can sail in an orderly group. After transmitting information to their neighbors, each AUV solves its own optimal control problem within the current prediction range and executes the first control quantity of the optimal control sequence. The distributed model predictive control structure framework is as follows: Figure 2 shown.

[0088] S2-1: Define AUV i Nonlinear system motion model

[0089]

[0090] Among them, t0 is the initial time of the system, z i (t) and u i (t) represents AUV i The state vector and control input vector at time t.

[0091] S2-2: AUV i initialization

[0092] In t c moment, use the previous moment t c-1 The calculated optimal control input is used as the initial control input for the next control cycle, and the currently observed state vector is used as the initial state for the next control cycle.

[0093] S2-3: Identify neighbors and exchange information with them

[0094] Determine the neighboring AUVs based on S2-1 communication.

[0095] Current time t c The state vector observation value of and the estimated control vector for the next control cycle Packing and exchanging information with neighboring AUVs;

[0096]

[0097] Among them, T p To control the cycle, Indicates that at t c-1 The optimal control sequence calculated at each moment; Indicates time s = t c-1 +T p The optimal control input is used as the control input for the next control sequence.

[0098] S2-4: Substitute the relevant information into the optimization function of the MPC sub-controller

[0099]

[0100] Among them, z -i (s;t c ) is the state value of the adjacent AUV received by the controller. In addition to the above optimization function, the controller calculation process also needs to ensure that the system meets the following conditions and constraints

[0101] z i (s;t c ),z -i (s;t c )∈Z (5)

[0102] u i (s;t c )∈U (6)

[0103] z i (t c ;t c )=z i (t c ) (7)

[0104] z i (t c +T p ;t c )∈Φ i (8)

[0105]

[0106]

[0107] Among them, s∈[t c ,t c +T p ],z i (s;t c )∈Z and zi (t c +T p ;t c )∈Φ i represent state constraints and terminal constraints respectively.

[0108] The cost function in the optimization problem is defined as

[0109]

[0110] Among them, F i and Φ i They are

[0111]

[0112] Φ i (z i (t))=γ∑||m(t)|| 2 (13)

[0113] S3: Determine the cost function for different roles

[0114] The cost function of the leader is expressed as follows.

[0115]

[0116] The coordinator’s cost function is expressed as follows.

[0117]

[0118] The follower's cost function is expressed as follows.

[0119]

[0120] Among them, α and γ are the switching parameters of the tracking neighbor item and the virtual guidance point item respectively. and is the expected position vector when following the coordinator, D R is the expected radius.

[0121] Substitute the above cost function into the optimization problem The specific form is (11), where F i According to step S3, we can substitute equations (13), (14), and (15) into (11) to transform the coordinated control problem into an optimal control problem and find the optimal solution. The optimal control sequence at that moment can be obtained by solving That is, the motion trend of each AUV at that moment, and its first step size is used as the control input of the next moment for the next optimal control.

[0122] S4: Output the distributed model prediction coordinated control results

[0123] At each control moment, the state of the AUV at the current moment is calculated based on the optimal control input solved in the previous prediction time domain. Based on step S2-3, information is exchanged with adjacent AUVs to determine the expected relative position of each AUV, and step S2-4 is performed to calculate the optimization problem, solve the optimal control input at that moment, and then update the next optimal control. As time goes by, each AUV continuously updates its state according to the control vector at each moment, forming a motion trajectory of state change over time, thereby realizing a cluster task based on swarm intelligence to form the desired square crystalline lattice structure formation.

[0124] According to another aspect of an embodiment of the present invention, an AUV formation method and formation device is also provided, including a design module, an establishment module, a determination module, a solution module and an output module; the design module is used to design different roles of the AUV group and specify the coordination mechanism; the establishment module is used to establish a distributed model predictive control framework; the determination module is used to determine the cost functions of different roles, substitute them into the distributed model predictive control framework, and transform the coordinated control problem into an optimal control problem; the solution module is used to solve the optimal control input, that is, the current motion trend of the AUV, and use its first step length as the control input at the next moment to perform the next optimal control; the output module is used to output the distributed model predictive coordinated control result.

[0125] According to another aspect of an embodiment of the present invention, a computer system is also provided, comprising one or more processors and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors implement any one of the above-mentioned AUV formation methods.

[0126] According to another aspect of an embodiment of the present invention, a computer-readable storage medium is further provided, wherein the computer-readable storage medium includes a stored program, wherein when the program is run, the device where the computer-readable storage medium is located is controlled to execute any one of the above-mentioned AUV formation methods.

[0127] According to another aspect of an embodiment of the present invention, a computer program is further provided, comprising computer-executable instructions, wherein the instructions are used to implement any one of the above-mentioned AUV formation methods when executed.

[0128] In order to verify the effectiveness of the above-mentioned swarm intelligence-based formation method, the present invention also provides the following embodiments.

[0129] A total of 10 AUVs were used in the MATLAB verification, with the expected distance between neighbors set to 30m and the maximum communication distance set to 35m. Each AUV was designed to follow a virtual guidance point before having neighbors.

[0130] Example 1

[0131] The initial state is that 10 AUVs gather together to verify that the distributed coordination strategy can quickly form a queue. Table 1 lists the initial state of each AUV in Example 1.

[0132] Table 1 Initial state of each AUV in a multi-AUV formation (first group)

[0133]

[0134] Figure 12 The navigation trajectories of each AUV are shown. It can be seen that the entire AUV group formed a formation while being dispersed, and has been maintaining this formation and moving in a straight line as expected.

[0135] Figure 13 The distance between the coordinator and its neighbors is shown, with interaction occurring when the distance is less than 35 meters. The distances between the coordinator and its neighbors are all stable at the desired values, demonstrating that the entire AUV swarm is able to adaptively organize and maintain a formation through the end of the simulation using the distributed model predictive control algorithm under the designed distributed strategy.

[0136] The above description is only a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any technician familiar with this technical field can easily think of various equivalent modifications or replacements within the technical scope disclosed in the present invention, and these modifications or replacements should all be included in the scope of protection of the present invention.

Claims

1. An AUV formation method, characterized in that: Here are the steps: S1: Design different roles and functions for the AUV swarm and define coordination mechanisms S1-1: Define neighbors and their relative expected positions Assume that the multi-AUV system consists of N AUVs, AUV is recorded as AUV i , suppose AUV i The maximum communication distance is , assuming that the multi-AUV system has ideal communication conditions and can communicate with its neighbors in two directions in each control cycle, its communication range is centered on itself, A circle with a radius of t Moment, AUV i The position vector is ,in, For AUV i The horizontal axis, For AUV i The vertical coordinate of AUV j The position vector is , if satisfied , then in t Moment, AUV i With AUV j They are neighbors and can communicate with each other; The desired formation is a square lattice structure, so the AUV can have at most 4 neighbors. The desired distance between given neighbors is D R , D R <communication radius R ,Each AUV generates 4 relative positions to be occupied and numbers them in sequence; S1-2: Designing different roles for AUV Define three different types of members: navigators, coordinators, and followers; According to the scenario where the group moves from dispersion to gathering and heading to the target area, it is assumed that there is a virtual guide point that maintains constant linear motion in the direction from the initial position of the group to the target area, and the AUV cluster follows; Therefore, the update of the virtual guidance point position of the group is described as follows: (1) in, is the speed of the virtual guidance point; Navigator: Focus on tracking the virtual guidance point, lead the entire formed group, and enable the entire team to keep up with the virtual guidance point as much as possible. Coordinator: coordinates its neighbors to move toward the corresponding vertices of the square with itself as the geometric center; the coordinator will follow the leader, that is, the position of neighbor 1, and the other three positions are assigned to the three AUV neighbors to guide them to the designated positions; Follower: Find the coordinator among its neighbors and navigate itself based on the coordinator's location distribution information; S1-3: Establish coordination mechanisms Start setting all AUVs as followers of the virtual navigation point: (1) Generation of coordinators Follower AUV i and follower AUV j There are no neighbors before entering each other's communication range; the follower closer to the virtual guidance point is designed to change to the coordinator, and the other AUV will become its follower; then the coordinator will assign positions to the followers, and the coordinator will also become the local navigator of the current group; (2) Generation of Navigators Designing the AUV i The distance to the virtual guidance point is less than D R / 2, changes to the navigator, when other AUVs enter the circle with the virtual guidance point as the center D R / 2 radius area circle will first be with AUV i They communicate and interact with each other, so there will be no situation where two AUVs compete for the leader; assuming that the distance between the two AUVs and the virtual guidance point is D R / 2, at this time, the two are definitely in a group. According to the following relationship between the two, the followed party changes to the leader. After the other AUVs meet the leader AUV, they all change to coordinators and accept the position arrangement of the leader. At the same time, the leader will send a special message to these coordinators to indicate that the leader is in the group, and gradually send the message to other AUVs in the group. When other groups meet it, they need to accept the leadership of the group. (3) A single follower encounters a coordinator Once a follower enters the communication range of the coordinator, the coordinator will check the number of its current vacant positions and generate its coordinates relative to the coordinator's current position; then, based on the angle, it will select an empty position as the allocation result. If an AUV is already assigned to a position, the AUV will select the closest position among the remaining positions and assign it to the new follower. (4) A single follower encounters a follower in a group When a single follower enters the communication range of a follower in the group, the follower in the group will change to a coordinator, and then arrange a position for the follower AUV. The AUV that becomes the coordinator will coordinate the entire formation with the coordinator AUV it originally followed; (5) Coordinator meets coordinator When two coordinators meet, the group with the navigator becomes the leading group and dominates the position allocation. The local navigators in the led group will no longer follow the virtual guidance point, but will coordinate the group's formation with other coordinators. (6) Followers in the group meet followers in the group When a follower in a group enters the communication distance of a follower in another group, both followers change to coordinators, and the whole process is the same as case (5); (7) Followers in the group encounter the coordinator When a follower in the group enters the communication distance of the coordinator, the follower will change to the coordinator, and then the whole process is the same as case (5); S2: Building a distributed model predictive control framework The distributed coordination strategy is that in each control cycle of DMPC, AUV R The AUVs communicate to search for their neighbors, and then assign roles to themselves according to certain rules, so that multiple AUVs can sail in an orderly manner as a group. After transmitting information to their neighbors, each AUV solves its own optimal control problem within the current prediction range and executes the first control variable of the optimal control sequence. S2-1: Define AUV i Nonlinear system motion model (2) in, is the initial moment of the system, and Represents AUV i exist t The state vector and control input vector at the moment; S2-2: AUV i initialization exist Moment, use the previous moment The calculated optimal control input is used as the initial control input for the next control cycle, and the currently observed state vector is used as the initial state for the next control cycle; S2-3: Identify neighbors and exchange information with them Determine the neighboring AUV based on S2-1 communication: Current moment The state vector observation value of and the estimated control vector for the next control cycle Packing and exchanging information with neighboring AUVs; in, To control the cycle, Indicates The optimal control sequence calculated at each moment; Indicates time The optimal control input is used as the control input for the next control sequence; S2-4: Substitute the relevant information into the optimization function of the MPC sub-controller in, The state value of the adjacent AUV received by the controller. In addition to the above optimization function, the controller calculation process also needs to ensure that the system meets the following conditions and constraints in, , and Represent state constraints and terminal constraints respectively; The cost function in the optimization problem is defined as (11) in, and They are (12) (13) S3: Determine the cost function for different roles The cost function of the leader is expressed as follows: (14) The coordinator’s cost function is expressed as follows: (15) The follower's cost function is expressed as follows: (16) in, are the switching parameters of the neighbor tracking item and the virtual guidance point item respectively; and is the desired position vector when following the coordinator, is the expected radius; Substitute the above cost function into the optimization problem Among them According to step S3, we can substitute equations (14), (15), and (16) into (11) to transform the coordinated control problem into an optimal control problem and find the optimal solution. The optimal control sequence at that moment can be obtained by solving , that is, the motion trend of each AUV at that moment, and use its first step length as the control input of the next moment to perform the next optimal control; S4: Output the distributed model prediction coordinated control results At each control moment, the state of the AUV at the current moment is calculated based on the optimal control input solved in the previous prediction time domain. Based on step S2-3, information is exchanged with adjacent AUVs to determine the expected relative position of each AUV, and step S2-4 is performed to calculate the optimization problem, solve the optimal control input at that moment, and then update the next optimal control. As time goes by, each AUV continuously updates its state according to the control vector at each moment, forming a motion trajectory of state change over time, thereby realizing a cluster task based on swarm intelligence to form the desired square crystalline lattice structure formation.

2. A formation device for implementing the AUV formation method according to claim 1, characterized in that: include: Design module, used to design different roles of the AUV group and specify the coordination mechanism; Establish a module for building a distributed model predictive control framework; The determination module is used to determine the cost functions of different roles and substitute them into the distributed model predictive control framework to transform the coordinated control problem into an optimal control problem; The solution module is used to solve the optimal control input, that is, the current motion trend of the AUV, and use its first step length as the control input at the next moment to perform the next optimal control; The output module is used to output the distributed model prediction and coordinated control results.

3. A computer system, characterized in that include: One or more processors, and a computer-readable storage medium for storing one or more programs, wherein when the one or more programs are executed by the one or more processors, the one or more processors are enabled to implement the method of claim 1.

4. A computer-readable storage medium, characterized in that Computer-executable instructions are stored, and when the instructions are executed, they are used to implement the method of claim 1.

5. A computer program product, characterized in that The invention comprises computer executable instructions, which are used to implement the method of claim 1 when the instructions are executed.

Images