Red-blue confrontation collaborative decision-making method and system for unmanned ship cluster in shore-port scene

Abstract

The invention discloses an unmanned ship cluster red-blue confrontation collaborative decision-making method and system in a shore-port scene, and belongs to the technical field of ocean robot collaboration. According to the method, the bank-port terrain with a relatively complex scene is established, and a red-blue game termination condition is designed based on a Stackelberg game framework. And then, a red-blue confrontation state is set based on a CORLEGS rule, and two red party operation modules are designed according to the confrontation state, including a weak formation S-Flocking module and an attack task allocation A-Assignment module. For target attraction planning in S-Flocking, an improved B-RRT * algorithm is adopted, the problems that the search speed of RRT * and Bi-RRT is low and the connection position of double trees is not smooth are solved, and A-Assignment with the minimum sacrifice cost of red party formation is designed by considering that the red party formation cannot avoid the blue party defense behavior; and meanwhile, a blue party defending task allocation D-Assignment considering that the blue party defends the red party to arrive at a designated area is also designed, and the influence of the blue party on the red party is considered to a certain extent. According to the method, the environmental fitness of the attacking aspect to a complex scene is met to a great extent.

Description

Technical Field

[0001] The present invention belongs to the field of marine robot collaboration technology, and specifically relates to a collaborative decision-making method and system for red-blue confrontation of an unmanned boat cluster in a shore-port scenario. Background Art

[0002] As a marine operation platform, unmanned surface vehicles (UAVs) have been applied in fields such as marine resource exploration, seabed topography mapping, and maritime rights protection due to their advantages such as high autonomy, strong scalability, strong maneuverability, and low cost. However, individual UAVs have many limitations when navigating complex marine environments, such as limited operating range and insufficient anti-interference capabilities, which make them less than ideal in more complex mission scenarios. In contrast, UAV swarms offer advantages such as more comprehensive target perception capabilities, higher mission execution efficiency, and stronger environmental adaptability. Therefore, the use of UAV swarms for collaborative mission execution has become an inevitable requirement and a development trend.

[0003] In the field of multi-UAV collaborative operations, unmanned marine confrontation is an important issue that urgently needs in-depth research. Unmanned vessel swarm collaborative decision-making is one of the core technologies in UAV swarm research. A good confrontation game strategy can help UAV swarms better complete their missions and increase their success rate.

[0004] Patent application number: CN202411224245.2, titled "A Distributed Collaborative Decision-Making Method for Multi-Objective Heterogeneous Unmanned Boat Swarms," ​​utilizes a distributed collaborative decision-making method and a game model to process the motion parameters and target information of unmanned boat swarms. This method addresses the speed and reliability issues of traditional decision-making methods in large-scale unmanned boat swarms and achieves efficient game task completion. However, this method primarily considers the actions of the defending unmanned boats, while the consideration of the offensive actions is relatively simple, and the mission scenario is set to an obstacle-free environment.

[0005] Patent application number: CN202410208647.7, titled "Method for Surface Unmanned Vehicles to Attack Protected Target Areas Based on Differential Game Theory," uses a differential game-based method to calculate the attack point and angle of the surface unmanned vehicle, and combines the first-order differential of the inner product to obtain the heading control input. This solves the complex problem of solving offensive surface unmanned vehicle strategies in the existing technology, achieving adaptation to complex game environments and effective attack. However, this method mainly considers the actions of the attacker near the target area, which means it lacks decision-making in more complex mission scenarios when it is impossible to bypass the blue team's unmanned vehicle.

[0006] The paper "A Decentralized Cluster Formation Containment Framework for Multirobot Systems" develops a unified cluster formation and containment coordination framework decomposed into two layers consisting of leaders and followers. According to the proposed approach, leader robots are first assigned to a set of distinct, non-overlapping clusters based on the location and priority of their targets using game-theoretic rules. They are then guided to achieve the desired formation around the corresponding targets. Subsequently, follower robots are guided to converge to the convex hull spanned by the leaders of each cluster. A prototype search and rescue operation is considered to highlight the usefulness of the proposed coordination framework.

[0007] The paper, "Research on Unmanned Aerial Vehicle Cluster Collaborative Countermeasures Based on Dynamic Non-Zero-Sum Game under Asymmetric and Uncertain Information," uses game theory as a framework to construct a decision-making model for cooperative countermeasures in a dynamic non-zero-sum game involving drone swarms, ultimately transforming it into an optimization problem. Based on this, the paper utilizes a multi-strategy fusion particle swarm optimization algorithm's Nash equilibrium solution approach, introducing adaptive inertia weights and a local mutation strategy to enhance population diversity while maintaining the particle swarm's local precision search capabilities.

[0008] The paper "Multistage Clustering-Based Localization for Remote UAV Swarm: A Coalitional Game Framework" constructs a multi-stage clustering model for UAV swarms. The proposed model, described as a coalition formation game, provides preference criteria for algorithm design. It employs a tree-structured multi-stage clustering mechanism based on a coalition graph game. Each cluster (local map) performs localization calculations, and then neighboring drones (NDs) are merged in stages to achieve improved localization accuracy.

[0009] Most of the above literature does not consider red-blue confrontation in complex mission scenarios. Most red-blue confrontation tasks are only set to whether they can successfully break into the central circular range or prevent the central circular range from entering the threat target. However, in actual engineering applications, unmanned boat swarms often face a confrontation environment with complex terrain and unpredictable situations, making it difficult to sort out the logical framework of the entire red-blue confrontation process. Therefore, the present invention proposes a collaborative decision-making method and decision-making system for red-blue confrontation of unmanned boat swarms in a shore-port scenario. This method uses the real-time status of the red-blue confrontation as the input of collaborative decision-making. Based on the red and blue planning module, it realizes small-scale red unmanned boat swarm attack and blue unmanned boat defense in the shore-port environment. It mainly introduces the red team formation planning and the red team's emergency strategy in emergency situations in the red and blue confrontation link. Summary of the Invention

[0010] The purpose of the present invention is to solve the problem of collaborative decision-making of red-blue confrontation in unmanned boat clusters in shore and port scenarios. The existing red-blue confrontation problem on the water surface considers relatively simple mission scenarios. Most mission scenarios do not contain many obstacles, and rarely consider the introduction of the attacking team behavior and the emergency strategy of the red side in emergencies. A collaborative decision-making method for red-blue confrontation in unmanned boat clusters in shore and port scenarios is now provided.

[0011] The present invention provides a coordinated decision-making system for red-blue confrontation of unmanned boat swarms in a shore-port scenario, comprising:

[0012] Information data processing module: Through the sensors, integrated navigation and communication systems installed on the red and blue unmanned boats, the unmanned boats can obtain real-time information about the ocean environment around them and their own position, posture, and speed. In the initial state, the red unmanned boat is set to an offensive state, and the blue unmanned boat is set to a defensive state.

[0013] Weak Formation S-Flocking Module: This module includes path planning, weak formation planning, and dynamic obstacle avoidance planning. Based on the received mission instructions and the current state and dynamic constraints of the UAV, it plans the desired path to the mission target according to the improved B-RRT*. During this process, it maintains the weak formation and dynamically avoids obstacles when a single target enters the safe range. Finally, it realizes the planning of the Red Army UAV weak formation S-Flocking virtual force.

[0014] Offensive Task Assignment (A-Assignment) module: This module includes cost calculation based on the threat posed by the Blue UAV to the Red UAV formation, assigns tasks using the Hungarian algorithm based on the Blue UAVs that have entered the safety range at the current moment, and finally implements the A-Assignment virtual force for planning the offensive task assignment of the Red UAVs.

[0015] Defense Task Assignment (D-Assignment) module: This module calculates the cost of the Red UAVs that enter the Blue UAVs' warning range, assigns tasks based on the Blue UAVs that currently enter the safety range using the Hungarian algorithm, and finally implements the D-Assignment virtual force for planning the Blue UAVs' defense task assignment.

[0016] Environmental perception module: Obtains obstacle information around the current location through lidar and binocular vision system; the binocular vision system can realize day and night switching function, using visible light camera during the day and infrared imaging camera at night;

[0017] Positioning module: obtains the position information of the unmanned boat through the Beidou integrated machine and combined navigation, transmits it to the information data processing module, and provides time-varying position information to the red and blue teams.

[0018] The present invention also provides a collaborative decision-making method for red-blue confrontation of unmanned boat clusters in a shore-port scenario, comprising:

[0019] Step 1: Obtain real-time information about the ocean environment around the UAV and its own position, attitude, and speed. Initially, set the red UAV to an offensive state and the blue UAV to a defensive state.

[0020] Step 2: Set the termination condition of the red-blue confrontation to the red team reaching the designated mission target position or the red team losing the ability to move, and determine whether the termination condition of the red-blue confrontation is met;

[0021] Step 3: Update the red-blue confrontation operation status; determine the red offensive UAV cluster operation status according to the CORLEGS rule, set the blue defensive status based on whether the red offensive enters the blue defensive UAV warning range, and calculate the UAV planning virtual force; the red offensive status includes weak formation S-Flocking and offensive task assignment A-Assignment; the blue defensive status includes defensive task assignment D-Assignment; the weak formation S-Flocking introduces three subtasks: path planning, weak formation planning, and dynamic obstacle avoidance planning;

[0022] Step 4: Based on the virtual force planned by the unmanned boat, the unmanned boat dynamics model is used to map the control thrust and torque of the unmanned boat, and the next position and speed of each unmanned boat are planned. The ocean environment information around the unmanned boat and its own position, attitude, and speed information are updated, and steps 2 to 4 are repeated.

[0023] Furthermore, the surrounding ocean environment information includes shore and port dam information in the ocean confrontation scenario, red side cluster formation unmanned boat information, and blue side dynamic defense unmanned boat information.

[0024] Further, in step 2, the red-blue confrontation iteration termination condition is designed using a Stackelberg game framework, in which the red team attack unmanned ship is set as the leader who first formulates the strategy, and the blue team defense unmanned ship is set as the follower who selects the optimal strategy after observing the strategy of the leader; the termination condition is:

[0025] {|p i -p goal |≤ε∪num isalive ()=0|i∈N}

[0026] In the formula, p i and p goal respectively represent the position of the red team unmanned ship numbered i and the expected arrival position of the red team task target, ε represents the task completion distance threshold, num isalive () function represents the number of red teams with motion ability, and N represents the size of the red team cluster.

[0027] Further, in step 3, the updated red-blue confrontation operation state is specifically:

[0028] Determine the operation state of the red team attack unmanned ship cluster:

[0029] Step 3.1: According to the closest approach distance DCPA in the CORLEGS rule, calculate the closest approach distance DCPA of each red team unmanned ship to each blue team unmanned ship that intrudes into the sensing range of the red team unmanned ship ij :

[0030]

[0031] In the formula, i and j respectively represent the numbers of red and blue unmanned ships, N represents the size of the red team formation, M represents the size of the blue team formation, L ij represents the Euclidean distance between the i-th red team unmanned ship and the j-th blue team unmanned ship, v xr and v yr respectively represent the x and y axis components of the relative speed of the red and blue unmanned ships in the global coordinate system, x r and y r respectively represent the x and y axis components of the relative distance of the red and blue unmanned ships in the global coordinate system, and r safe represents the safety collision avoidance radius of the red team unmanned ship.

[0032] Step 3.2: Determine the predicted interaction relationship state i of the red and blue unmanned ships based on the CORLEGS rule; when DCPA ij is greater than 0, there is at least one blue team unmanned ship that crosses the bow of the red team unmanned ship, and the red team unmanned ship needs to avoid obstacles, and the obstacle avoidance state state iThe value is 1; on the contrary, the red unmanned boat does not need to avoid obstacles when crossing the stern of the blue unmanned boat, and the obstacle avoidance state is state i The value is 0;

[0033] Step 3.3: According to the obstacle avoidance state of each red team's unmanned boat i , judge the number of unmanned boats that the red team's unmanned boat cluster needs to avoid obstacles, when the number of unmanned boats in the neighboring group needs to avoid obstacles num USV When there are less than 1, the red UAV cluster will bypass the obstacle in formation and start the weak formation S-Flocking module; when the number of obstacles to be avoided by the neighboring UAVs is num USV When there are more than one, an obstacle avoidance conflict occurs within the Red team's cluster, and the Red team's UAV cluster cannot bypass the obstacle in formation. The Red team's UAV cluster activates the offensive task assignment A-Assignment module;

[0034] Determine the operational status of the blue side's defensive unmanned boat:

[0035] The defense status is set based on whether the red team enters the blue team's defense unmanned boat warning range. When the red team does not enter the blue team's warning range, the blue team's unmanned boat will remain on alert; when the red team enters the blue team's warning range, the blue team's unmanned boat defense task assignment D-Assignment module will be activated;

[0036]

[0037] Where, L ji represents the Euclidean distance between the blue unmanned boat No. j and the red unmanned boat No. i, r detc Indicates the detection radius of the blue side's unmanned boat.

[0038] Furthermore, in step 3, the red team bypasses the obstacle in formation and starts the weak formation S-Flocking module; if there is a red team unmanned boat in the formation that needs to avoid obstacles, the unmanned boat will perform three subtasks: path planning, weak formation planning, and dynamic obstacle avoidance planning. The remaining red team unmanned boats in the formation will perform path planning and weak formation planning.

[0039] Furthermore, the Red side's UAV cluster activates the weak formation S-Flocking module, specifically:

[0040] (1) Red Army UAV execution path planning

[0041] Set the center of the Red Team's virtual formation as the starting point and the mission target location as the end point; generate two search trees in the specified search area, the forward tree Tree F and reverse tree B, randomly generate path node Node, update node connection mode by unmanned ship model constraint cost, obtain complete discrete path sequence; according to complete discrete path point information target attraction, calculate path planning virtual force; specifically:

[0042] Step 3.1.1: adopt B-RRT* search starting point to path node Node, evaluate the pros and cons of the current node Node for the red unmanned ship cluster by cost C(N);

[0043] C(N)=C(N.p)+L(N,N.p)+ψ(N,N.p,N.pp)

[0044] In the formula, C(N.p) represents the global cost of Node.parent distance from the search starting point; L(N,N.p) represents the distance cost of Node and Node.parent; ψ(N,N.p,N.pp) represents the steering cost from path Node.parent.parent to Node.parent, and then to Node;

[0045] Step 3.1.2: query NearNode in the radius range of Node neighborhood and C(NearNode), update the node NearNode with the minimum constraint cost to set as Node.parent, at this time, complete one round of backtracking; query ClansNode of Node family and C(ClansNode), update the node ClansNode with the minimum constraint cost to set as Node.parent, complete two rounds of backtracking;

[0046] Step 3.1.3: when Tree F and Tree B The last node distance is less than the neighborhood radius, use bisection method to smooth the new node NewNode as:

[0047]

[0048] In the formula, NewNode represents the new node; Tree F .e and Tree B .e respectively represent the last node of Tree F and Tree B , iscollision(N.p,Tree B .e) represents that the node Node.parent and the node Tree B .e pass through the static obstacle of the map;

[0049] Step 3.1.4: according to complete discrete path point information target attraction, the virtual force of the red unmanned ship path planning is:

[0050]

[0051] Where p g1 Represents the discrete path points that attract the red team's unmanned boat No. i in the S-Flocking module at the current moment.

[0052] (2) The Red Army’s unmanned boat executes the formation-maintaining FM plan

[0053] According to the observable information of the Red team, the virtual force for maintaining the Red team's unmanned boat formation is:

[0054]

[0055] Where w s and w ds Respectively represent the positive and negative functions of sigmoid after translation and scaling; state i =0 means that the red team’s unmanned boat No. i does not introduce obstacle avoidance planning;

[0056] (3) The Red Army needs an obstacle avoidance UAV to perform dynamic obstacle avoidance DOA planning

[0057] According to the observable information of the blue UAV, the virtual force of the UAV's dynamic obstacle avoidance is:

[0058]

[0059] The total virtual force of each Red Army UAV starting the weak formation S-Flocking is:

[0060]

[0061] Furthermore, in step 3, the Red side's UAV cluster initiates the attack task assignment A-Assignment as follows:

[0062] When the Red UAV does not form a formation to bypass the Blue UAV's dynamic defense, the Blue UAV that threatens the formation will be added to the Red UAV's attack task allocation queue. The Hungarian algorithm is used as the attack task allocation framework, in which the cost calculation includes the Manhattan distance cost and the relative speed cost of the Red and Blue sides respectively. The Red UAV with the smallest cost is dispatched to intercept. The cost evaluation function is:

[0063] C A (i,j)=α1·l r (i,j)+β1·v r (i,j) i∈N,j∈M

[0064] Where, l r1(i, j) is the red and blue Manhattan distance evaluation function, α1 is the red and blue Manhattan distance coefficient, the larger α1 is, the more biased the selection of the red unmanned boat that meets the shortest energy time constraint is; v r (i, j) is the red and blue relative speed evaluation function, β1 is the red and blue relative speed evaluation coefficient, and the larger β1 is, the more likely it is to select the red unmanned boat with the best capability for interception;

[0065] The total virtual force of each Red Army UAV starting the attack mission assignment A-Assignment is:

[0066]

[0067] Where p g2 Indicates the interception position assigned to the red team's unmanned boat No. i based on the Hungarian task allocation algorithm at the current moment.

[0068] Furthermore, the specific process of the blue side's unmanned boat starting the defense task assignment D-Assignmen module is as follows:

[0069] When the red team enters the warning range of the blue team's defensive unmanned boat, the blue team adds the red team's unmanned boat to the blue team's defense task allocation queue based on the observed known red team's unmanned boat information. The Hungarian algorithm is used as the defense task allocation framework, where the cost calculation includes the Manhattan distance cost between the blue and red teams and the Manhattan distance cost between the red team and the blue team's protected object. The blue team's unmanned boat with the smallest cost is sent to defend. The cost evaluation function is:

[0070] C D (j,i)=α2·l r (j,i)+β2·l r (i,goal) i∈N,j∈M

[0071] Where, l r2 (i, j) is the Manhattan distance evaluation function of the blue and red sides, α2 is the Manhattan distance coefficient of the blue and red sides, and the larger α2 is, the more biased the selection of the blue side unmanned boat defense that meets the shortest energy and time constraints is; l r (i, goal) is the Manhattan distance function between the red team and the goal, β2 is the Manhattan distance coefficient between the red team and the goal. The larger β2 is, the more likely it is to select the red team's unmanned boat with the best capability for interception.

[0072] The virtual force of each blue UAV starting the defense task assignment D-Assignmen is:

[0073]

[0074] Where p g3 Indicates the defensive position assigned to the blue unmanned boat No. j based on the Hungarian task allocation algorithm at the current moment.

[0075] Further, the unmanned ship control thrust F u and torque T r are mapped to the unmanned ship dynamics model to guarantee that the planned unmanned ship speed and heading meet the motion model constraints:

[0076]

[0077] where T max and T min represent the maximum and minimum output thrust of the unmanned ship respectively, beta represents the angle between the virtual resultant force and the actual speed of the unmanned ship, and w represents the torque adjustment coefficient;

[0078] The unmanned ship control thrust F u and torque T r can be used to plan the next moment position and speed of each unmanned ship, i.e., update the surrounding ocean environment information of the red and blue sides and the position, attitude and speed information of the unmanned ship.

[0079] The application further provides a computer readable storage medium, which stores a computer program / instruction, and the computer program / instruction is executed by a processor to realize the steps of the red and blue confrontation cooperative decision method of the unmanned ship cluster in the shore port scene according to any one of the above.

[0080] The application further provides a computer program product, which comprises a computer program / instruction, and the computer program / instruction is executed by a processor to realize the steps of the red and blue confrontation cooperative decision method of the unmanned ship cluster in the shore port scene according to any one of the above.

[0081] The application has the following beneficial effects:

[0082] The present invention discloses a collaborative decision-making method for red-blue confrontation of unmanned boat clusters in a shore-port scenario, establishes a shore-port terrain with relatively complex scenarios, and designs the termination conditions of the red-blue game based on the Stackelberg game framework. Subsequently, the red-blue confrontation state is set based on the CORLEGS rule, and two red operation modules are designed according to the confrontation state, including a weak formation S-Flocking module and an offensive task assignment A-Assignment module. For target attraction planning in S-Flocking, an improved B-RRT* algorithm is adopted to solve the problems of slow search speed of RRT* and Bi-RRT and uneven connection of double trees. On the one hand, it increases the global search speed of the algorithm and enhances the algorithm's optimization ability in a global range. Considering that the red formation cannot avoid the blue defensive behavior, an A-Assignment is designed that takes into account the minimum sacrifice cost of the red formation. It not only ensures the safety of the red formation in collision avoidance, but also improves the task completion rate when it is impossible to bypass the dangerous area. We also designed a blue team operation module, including a defensive task assignment (D-Assignment) module. This module considers the blue team's defense of the red team's arrival in a designated area, and to a certain extent accounts for the blue team's influence on the red team. This method greatly meets the offensive team's adaptability to complex scenarios. BRIEF DESCRIPTION OF THE DRAWINGS

[0083] Figure 1 This is a flow chart of the collaborative decision-making method for red-blue confrontation of unmanned boat clusters in a shore-port scenario of the present invention;

[0084] Figure 2 This is a schematic diagram of the shore-port scenario modeling of the present invention;

[0085] Figure 3 This is a motion model diagram of the red and blue unmanned boats of the present invention;

[0086] Figure 4 This is a schematic diagram of the Red Army's unmanned boat cluster S-Flocking;

[0087] Figure 5 This is a schematic diagram of the A-Assignment of the Red Team's unmanned boat cluster of the present invention;

[0088] Figure 6 It is a schematic diagram of the D-Assignment of the blue team's unmanned boat of the present invention. DETAILED DESCRIPTION

[0089] The present invention will be further described below with reference to the accompanying drawings.

[0090] The present invention discloses a red-blue confrontation collaborative system for an unmanned boat swarm in a shore-port scenario, comprising:

[0091] (1) Information data processing module: The ocean environment information of the red-blue confrontation scenario of the unmanned boat cluster in the shore-port scenario. The red and blue unmanned boats are equipped with sensors, integrated navigation and communication systems, which can obtain the ocean environment information around the unmanned boats and their own position, posture, and speed information in real time; in the initial state, the red unmanned boat is set to the offensive state, and the blue unmanned boat is set to the defensive state;

[0092] The above-mentioned marine environment information refers to the shore and port dam information in the marine confrontation scenario, the red side's cluster formation unmanned boat information, and the blue side's dynamic defense unmanned boat information.

[0093] (2) Red side weak formation S-Flocking module: This module includes path planning, weak formation planning, dynamic obstacle avoidance planning and other functions. According to the received mission instructions and the current state and dynamic constraints of the unmanned boat, the expected path to the mission target is planned according to B-RRT*. In the process, the weak formation is maintained and dynamic obstacle avoidance is performed when a single target enters the safe range. Finally, the S-Flocking virtual force of the weak formation of the red side unmanned boat is planned.

[0094] (3) Red side attack task assignment A-Assignment module: This module contains a cost calculation formula based on the blue side's unmanned boats that pose a threat to the red side's formation. It uses the Hungarian algorithm to assign tasks based on the blue side's defensive unmanned boats that have entered the safe range at the current moment, and finally realizes the planning of the red side's unmanned boat attack task assignment A-Assignment virtual force.

[0095] (3) Blue side defense task assignment D-Assignment module: This module contains a cost calculation formula for the red side unmanned boat that breaks into the warning range of the blue side defense unmanned boat. It uses the Hungarian algorithm to assign tasks based on the blue side defense unmanned boat that breaks into the safety range at the current moment, and finally realizes the planning of the blue side unmanned boat defense task assignment D-Assignment virtual force.

[0096] (4) Environmental perception module: Obtain obstacle information around the current location through laser radar and binocular vision system; the binocular vision system can realize day and night conversion function, using visible light camera during the day and infrared imaging camera at night.

[0097] (5) Positioning module: obtains the position information of the UAV through the Beidou integrated machine and integrated navigation, transmits it to the information data processing module, and provides time-varying position information to the red team and the blue team;

[0098] (6) Unmanned boat: Receives evasive action instructions, executes control mechanisms, and navigates according to prescribed instructions.

[0099] The present invention discloses a collaborative decision-making method for red-blue confrontation of an unmanned boat swarm in a shore-port scenario, comprising:

[0100] Step 1: If Figure 2 The shore-port scenario is modeled to obtain the ocean environment information of the red-blue confrontation scenario of the unmanned boat cluster in the shore-port scenario. The onboard sensors can obtain the ocean environment information around the unmanned boat and its own position, posture, and speed information in real time. In the initial state, the red unmanned boat is set to the offensive state, and the blue unmanned boat is set to the defensive state.

[0101] The above-mentioned marine environment information refers to the shore and port dam information in the marine confrontation scenario, the red side's cluster formation unmanned boat information, and the blue side's dynamic defense unmanned boat information.

[0102] Step 2: Determine whether the red-blue confrontation termination condition has been met, that is, whether the red team has reached the designated target or the red team has completely lost its ability to move:

[0103] The Stackelberg game framework is used to design the end conditions for the red-blue confrontation iteration. The red team's attacking unmanned boat is set as the leader who first formulates the strategy, and the blue team's defending unmanned boat is set as the follower who selects its own optimal strategy after observing the leader's strategy. The termination condition is that the red team reaches the designated mission target location or all red teams lose the ability to move. The specific mathematical formula is expressed as follows:

[0104] {|p i -p goal |≤ε∪num isalive ()=0|i∈N}

[0105] Where p i and p goal They represent the position of the Red Team’s unmanned boat No. i and the expected arrival position of the Red Team’s mission target, ε represents the mission completion distance threshold, and num isalive The () function indicates checking the number of red teams with mobility capabilities, and N indicates the size of the red team cluster.

[0106] Step 3: Update the red-blue confrontation operation status, determine the red side's offensive unmanned boat cluster operation status according to the CORLEGS rules, and set the defense status based on whether the red side enters the blue side's defensive unmanned boat warning range.

[0107] (1) Determine the operational status of the Red Army's attacking unmanned boat swarm

[0108] First, according to the closest approach distance in the CORLEGS rule, calculate the closest approach distance DCPA of any red unmanned boat No. i to the blue defense unmanned boat No. j ij , and solve the obstacle avoidance state of the i-th unmanned boat i , the specific mathematical formula is:

[0109]

[0110] Where, L ij represents the Euclidean distance between the red unmanned boat No. i and the blue unmanned boat No. j, v xr and v yr They represent the components of the relative speed of the red and blue unmanned boats in the global coordinate system, x r and y r They represent the relative distance between the red and blue unmanned boats in the global coordinate system, r safe represents the safe collision avoidance radius of the red unmanned boat, i and j represent the numbers of the red and blue unmanned boats respectively, N represents the size of the red team, M represents the size of the blue team, and the dist function represents the Euclidean distance between two points.

[0111] Then, the predicted interaction relationship state of the red and blue unmanned boats is determined based on the CORLEGS rule. ij Each red unmanned boat calculates DCPA for each blue unmanned boat that enters its sensing range ij , when DCPA ij If it is greater than 0, there is at least one blue unmanned boat crossing the bow of the red unmanned boat, then the unmanned boat No. i needs to avoid the obstacle, and the obstacle avoidance state is state i The value is 1; on the contrary, if all blue unmanned boats pass through the stern of the red unmanned boat, then the i-th unmanned boat does not need to avoid obstacles, and the obstacle avoidance state is state i The value is 0; the specific mathematical formula is:

[0112]

[0113] Finally, according to the obstacle avoidance state of each red team's unmanned boat i , judge the number of unmanned boats that the red team's unmanned boat cluster needs to avoid obstacles, when the number of unmanned boats in the neighboring group needs to avoid obstacles num USV When there are less than 1, the obstacles are bypassed in formation and the Red UAV cluster weak formation S-Flocking module is activated; when the number of obstacles to be avoided by the neighboring UAVs is num USV When there are more than one, an obstacle avoidance conflict occurs within the Red team's swarm, and it is impossible to circumvent the obstacle in formation. The Red team's UAV swarm attack task assignment A-Assignment module is activated. The specific mathematical formula is:

[0114]

[0115] (2) Determine the operational status of the blue side's defensive unmanned boat

[0116] The defense state is set based on whether the red team enters the blue team's defense unmanned boat warning range. When the red team does not enter the blue team's warning range, the blue team's unmanned boat remains on alert; when the red team enters the blue team's warning range, the blue team's unmanned boat defense task assignment D-Assignment module is activated. The specific mathematical formula is expressed as:

[0117]

[0118] Where i and j represent the numbers of the red and blue unmanned boats respectively, N represents the size of the red team, M represents the size of the blue team, and L ji represents the Euclidean distance between the blue unmanned boat No. j and the red unmanned boat No. i, r detc Indicates the detection radius of the blue side's unmanned boat.

[0119] Step 4: Based on the red-blue confrontation operation status, determine the modules used by the red and blue teams in their planning, including: weak formation S-Flocking, red team's offensive task assignment A-Assignment, and blue team's defensive task assignment D-Assignment. Ultimately, the virtual force planned by the above modules is obtained.

[0120] If the Red team bypasses the obstacle in formation, activate the weak formation S-Flocking module, such as Figure 4 As shown:

[0121] Starting the weak formation S-Flocking module introduces three subtasks: target attraction GPP planning, formation maintenance FM planning, and dynamic obstacle avoidance DOA planning; if there is a red unmanned boat in the formation that needs to avoid obstacles, then the unmanned boat needs to start target attraction GPP planning, formation maintenance FM planning, and dynamic obstacle avoidance DOA planning; the other unmanned boats in the formation start target attraction GPP planning and formation maintenance FM planning; if none of the red unmanned boats in the formation need to avoid obstacles, then all unmanned boats in the formation start target attraction GPP planning and formation maintenance FM planning.

[0122] (1) Red team launches target attraction GPP planning

[0123] First, set the center position of the Red team's virtual formation as the starting point and the mission target location as the end point. Generate two search trees in the specified search area. The forward tree Tree F and reverse tree B , randomly generate path nodes Node, and update the node connection method through the unmanned boat model constraint cost calculation formula.

[0124] C(N) represents the cost from the B-RRT* search starting point to the path node Node, which is used to evaluate the pros and cons of the current node Node for the red side's unmanned boat cluster. The specific mathematical formula is expressed as:

[0125] C(N)=C(Np)+L(N,Np)+ψ(N,Np,N.pp)

[0126] Where C(Np) represents the global cost of Node.parent from the search starting point; L(N,Np) represents the distance cost between Node and Node.parent; ψ(N,Np,N.pp) represents the turn cost from Node.parent.parent to Node.parent and then to Node.

[0127] Then, query the NearNode and C(NearNode) within the neighborhood radius ε1 of the Node, update the node NearNode with the minimum constraint cost and set it as Node.parent, and then complete one round of backtracking; query the ClansNode and C(ClansNode) of the Node family, update the node ClansNode with the minimum constraint cost and set it as Node.parent, and then complete the second round of backtracking. The specific mathematical formula is expressed as follows:

[0128] Node.parent={min(C(NearNode))|dist(Node,NearNode)≤ε1}

[0129] Node.parent=min(C(ClansNode))

[0130] Finally, when Tree F and Tree B The distance of the last node is less than the threshold ε1 (ie dist(Tree F .e,Tree B .e)≤ε1), use the bisection method to smoothly add the new node NewNode. The specific mathematical formula is expressed as:

[0131]

[0132] In the formula, NewNode represents a new node; Tree F .e and Tree B .e represents Tree F and Tree B The last node, iscollison(Np,Tree B .e) represents the node Node.parent and the node Tree B .eTraverse static obstacles on the map.

[0133] According to the above steps, a complete discrete path sequence is obtained.

[0134] Based on the complete discrete path point information target attraction, the specific mathematical formula is expressed as:

[0135]

[0136] Where p g1 Represents the discrete path points that attract the red team's unmanned boat No. i in the S-Flocking module at the current moment.

[0137] (2) The Red team starts the formation and maintains the FM plan

[0138] Based on the observable information of the red team, the formation is maintained. The specific mathematical formula is expressed as:

[0139]

[0140] Where w s and w ds Respectively represent the positive and negative functions of sigmoid after translation and scaling; state i =0 means that the red team’s unmanned boat No. i does not introduce obstacle avoidance planning.

[0141] (3) The Red team starts dynamic obstacle avoidance DOA planning

[0142] Based on the observable information of the blue team's unmanned boat, dynamic obstacle avoidance is performed. The specific mathematical formula is expressed as follows:

[0143]

[0144] Where w ds Represents the inverse function of sigmoid after translation and scaling; state i =1 means that the red team’s unmanned boat No. i introduces obstacle avoidance planning.

[0145] In summary, the virtual force of the red team's unmanned boat No. i that executes the S-Flocking module is obtained. The specific mathematical formula is expressed as follows:

[0146]

[0147] If the Red team does not bypass the Blue team's UAV in formation, the attack task assignment A-Assignment module is activated, such as Figure 5 As shown:

[0148] When the Red UAV does not form a formation to bypass the Blue UAV's dynamic defense, the Blue UAV that threatens the formation will be added to the Red UAV's attack task allocation queue. The Hungarian algorithm is used as the attack task allocation framework, in which the cost calculation includes the Manhattan distance cost and the relative speed cost of the Red and Blue sides respectively. The Red UAV with the smallest cost is dispatched to intercept to ensure that the Red side completes the task. The specific cost evaluation function is expressed as:

[0149] C A (i,j)=α1·l r (i,j)+β1·v r (i,j) i∈N,j∈M

[0150] Where, l r (i, j) is the red and blue Manhattan distance evaluation function, α1 is the red and blue Manhattan distance coefficient, the larger α1 is, the more biased the selection of the red unmanned boat that meets the shortest energy time constraint is; v r (i, j) is the red and blue relative speed evaluation function, β1 is the red and blue relative speed evaluation coefficient, and the larger β1 is, the more likely it is to select the red unmanned boat with the best capability for interception.

[0151] According to the above steps, the virtual force of the red team's unmanned boat No. i that executes the A-Assignment module is obtained. The specific mathematical formula is expressed as:

[0152]

[0153] In the formula, in the formula, p g2 This represents the interception position assigned to Red UAV i at the current moment based on the Hungarian task allocation algorithm. It is worth noting that Red UAV i still communicates with the other UAVs maintaining the weak formation S-Flocking module, but does not participate in maintaining the weak formation.

[0154] If there is a red unmanned boat within the blue team's warning range, start the defense task assignment D-Assignment module, such as Figure 6 As shown:

[0155] When the Red team enters the warning range of the Blue team's defensive UAV, the Blue team adds the Red team's UAV to the Blue team's defense task allocation queue based on the observed known Red team's UAV information. The Hungarian algorithm is used as the defense task allocation framework, where the cost calculation includes the Manhattan distance cost between the Blue and Red teams and the Manhattan distance cost between the Red team and the Blue team's protected object. The Blue team's UAV with the smallest cost is sent to defend to ensure that the Red team does not reach the designated area. The specific cost evaluation function is expressed as:

[0156] C D (j,i)=α2·l r (j,i)+β2·l r (i,goal) i∈N,j∈M

[0157] Where, l r (i, j) is the Manhattan distance evaluation function of the blue and red sides, α2 is the Manhattan distance coefficient of the blue and red sides, and the larger α2 is, the more biased the selection of the blue side unmanned boat defense that meets the shortest energy and time constraints is; l r(i, goal) is the Manhattan distance function between the red side and the goal point, β2 is the Manhattan distance coefficient between the red side and the goal point. The larger β2 is, the more likely it is to select the red side unmanned boat with the best capability for interception.

[0158] According to the above steps, the virtual force of the blue team's unmanned boat No. j that executes the D-Assignment module is obtained. The specific mathematical formula is expressed as:

[0159]

[0160] Where p g3 Indicates the defensive position assigned to the blue unmanned boat No. j based on the Hungarian task allocation algorithm at the current moment.

[0161] Step 5: Based on the virtual force planned in step 4, update the ocean environment information around the red and blue sides as well as their own position, attitude, and speed information, and jump to step 2.

[0162] The virtual forces of S-Flocking, A-Assignment and D-Assignment are mapped to the control thrust F of the unmanned boat through the unmanned boat dynamics model. u and torque T r , ensuring that the planned speed and heading of the unmanned boat meet the constraints of the motion model;

[0163] The Red Army's UAV performs weak formation S-Flocking as follows:

[0164]

[0165] The Red Army's UAV performs weak formation S-Flocking as follows:

[0166]

[0167] The blue side's unmanned boat performs the defense mission and the D-Assignmen are:

[0168]

[0169] Where, T max and T min Denote the maximum and minimum thrust output by the UAV, β represents the angle between the virtual resultant force and the UAV's actual speed, and w represents the torque adjustment coefficient. This method allows the next position and speed of each UAV to be calculated, thereby updating the surrounding ocean environment information for both the red and blue UAVs, as well as their own position, attitude, and speed.

[0170] In particular, in some preferred embodiments of the present application, a computer device is also provided, comprising a memory and a processor and a computer program stored in the memory, wherein the processor implements the steps of the method for cooperative decision-making of the red-blue confrontation of the unmanned ship cluster in the port scene according to any one of the above-mentioned embodiments when executing the computer program.

[0171] In some other preferred embodiments of the present application, a computer readable storage medium is also provided, wherein the computer readable storage medium stores a computer program / instruction, and the computer program is executed by a processor to implement the steps of the method for cooperative decision-making of the red-blue confrontation of the unmanned ship cluster in the port scene according to any one of the above-mentioned embodiments.

[0172] It can be understood by those skilled in the art that all or part of the processes in the above-mentioned embodiments can be completed by a computer program instructing related hardware, and the computer program can be stored in a non-volatile computer readable storage medium. When the computer program is executed, it can include the processes of the above-mentioned embodiments of the method for cooperative decision-making of the red-blue confrontation of the unmanned ship cluster in the port scene, which will not be repeated here.

[0173] In the description of the present specification, the description of the terms "one embodiment", "some embodiments", "an example", "a specific example", or "some examples" means that the specific features, structures, materials or characteristics described in connection with the embodiment or example are included in at least one embodiment or example of the present application. In the present specification, the illustrative description of the above-mentioned terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described can be combined in any one or N embodiments or examples in a suitable manner. In addition, the skilled in the art can combine and combine the different embodiments or examples described in the present specification and the features of the different embodiments or examples without contradiction.

[0174] Any process or method descriptions in flow charts or otherwise described herein can be understood as representing one or more steps of a method or process, including a computer program in which the steps of the method or process are implemented as executable instructions, and the scope of the preferred embodiments of the present application includes additional implementations in which the steps of the method or process are implemented in an order different from that shown or discussed, including substantially simultaneously, or in reverse order, or in an order that is different from that shown or discussed, and the skilled in the art will understand that the scope of the present application includes additional implementations in which the steps of the method or process are implemented in an order different from that shown or discussed, including substantially simultaneously, or in reverse order, or in an order that is different from that shown or discussed.

[0175] The logic and / or steps represented in flow diagrams or otherwise described herein, for example, can be considered as a sequence of instructions to implement logic functions, and can be embodied in any computer-readable medium for use by an instruction execution system, apparatus, or device, such as a computer-based system, processor- containing system, or other system that can fetch the instructions from the instruction execution system, apparatus, or device and execute the instructions. In the context of this specification, a "computer-readable medium" can be any means that can contain, store, communicate, propagate or transport the program for use by or in connection with the instruction execution system, apparatus, or device. The computer-readable medium can be a machine-readable storage device (e.g., magnetic, optical or other) a machine-readable storage diskette (e.g., floppy, flexible or other), a machine-readable storage card (e.g., ROM, EEPROM, flash memory or other), a machine- readable storage tape (e.g., magnetic, optical or other), a machine-readable storage medium (e.g., a portable electronic device, a computer diskette, a computer memory, a programmable logic device, an application-specific integrated circuit, a programmable logic controller, a digital signal processor, a microprocessor, a microprocessor array or other), or a machine- readable interface device (e.g., a wired or wireless interface device). The computer-readable medium can also be paper or other suitable material upon which the program is printed, as the program can be electronically captured, for example by optically scanning the paper or other suitable medium, then compiled, interpreted, or otherwise processed in a suitable manner, if necessary, and stored in a computer memory.

[0176] It should be understood that aspects of the application can be implemented in hardware, software, firmware or combinations thereof. In the above embodiments, the N steps or methods can be implemented in software or firmware stored in a memory and executed by a suitable instruction execution system. As such, if implemented in hardware, and in another embodiment, any of the following technologies, known in the art, or combinations thereof, can be used: discrete logic circuitry having logic gates for implementing logic functions on data signals, application specific integrated circuits having appropriate combinational logic gates, programmable gate arrays (PGA), field programmable gate arrays (FPGA), and the like.

[0177] Those of skill in the art could readily implement the above described example methods with all or a subset of the recited steps carried out with a program of instructions executed by associated hardware, the program of instructions being stored in a computer readable storage medium that, when executed, includes one or a combination of the steps of the method embodiments.

[0178] In addition, each function unit in each embodiment of the present application can be integrated in one processing module, or each unit can be physically present separately, or two or more units can be integrated in one module. The integrated module can be realized in the form of hardware or in the form of a software function module. When the integrated module is realized in the form of a software function module and sold or used as an independent product, it can also be stored in a computer readable storage medium.

[0179] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc. Although the embodiments of the present application have been shown and described above, it should be understood that the above embodiments are exemplary and should not be construed as limiting the present application, and those skilled in the art can make changes, modifications, replacements and variations to the above embodiments within the scope of the present application.

Claims

1. A coordinated decision-making system for red-blue confrontation of unmanned boat swarms in a shore-port scenario, characterized by: include: Information data processing module: Through the sensors, integrated navigation and communication systems installed on the red and blue unmanned boats, the unmanned boats can obtain real-time information about the ocean environment around them and their own position, posture, and speed. In the initial state, the red unmanned boat is set to an offensive state, and the blue unmanned boat is set to a defensive state. Weak Formation S-Flocking Module: This module includes path planning, weak formation planning, and dynamic obstacle avoidance planning. Based on the received mission instructions and the current state and dynamic constraints of the UAV, it plans the desired path to the mission target according to the improved B-RRT*. During this process, it maintains the weak formation and dynamically avoids obstacles when a single target enters the safe range. Finally, it realizes the planning of the Red Army UAV weak formation S-Flocking virtual force. Offensive Task Assignment (A-Assignment) module: This module includes cost calculation based on the threat posed by the Blue UAV to the Red UAV formation, assigns tasks using the Hungarian algorithm based on the Blue UAVs that have entered the safety range at the current moment, and finally implements the A-Assignment virtual force for planning the offensive task assignment of the Red UAVs. Defense Task Assignment (D-Assignment) module: This module calculates the cost of the Red UAVs that enter the Blue UAVs' warning range, assigns tasks based on the Blue UAVs that currently enter the safety range using the Hungarian algorithm, and finally implements the D-Assignment virtual force for planning the Blue UAVs' defense task assignment. Environmental perception module: Obtains obstacle information around the current location through lidar and binocular vision system; the binocular vision system can realize day and night switching function, using visible light camera during the day and infrared imaging camera at night; Positioning module: obtains the position information of the unmanned boat through the Beidou integrated machine and combined navigation, transmits it to the information data processing module, and provides time-varying position information to the red and blue teams.

2. A collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario as described in claim 1, characterized in that: include: Step 1: Obtain the ocean environment information around the unmanned boat and its own position, attitude, and speed information; In the initial state, the red party's unmanned boat is set to the offensive state, and the blue party's unmanned boat is set to the defensive state; Step 2: Set the termination condition of the red-blue confrontation to the red team reaching the designated mission target position or the red team losing the ability to move, and determine whether the termination condition of the red-blue confrontation is met; Step 3: Update the red-blue confrontation operation status; determine the red offensive UAV cluster operation status according to the CORLEGS rule, set the blue defensive status based on whether the red offensive enters the blue defensive UAV warning range, and calculate the UAV planning virtual force; the red offensive status includes weak formation S-Flocking and offensive task assignment A-Assignment; the blue defensive status includes defensive task assignment D-Assignment; the weak formation S-Flocking introduces three subtasks: path planning, weak formation planning, and dynamic obstacle avoidance planning; Step 4: Based on the virtual force planned by the unmanned boat, the unmanned boat dynamics model is used to map the control thrust and torque of the unmanned boat, and the next position and speed of each unmanned boat are planned. The ocean environment information around the unmanned boat and its own position, attitude, and speed information are updated, and steps 2 to 4 are repeated.

3. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 2 is characterized in that: The surrounding ocean environment information includes shore and port dam information in the ocean confrontation scenario, the red side’s cluster formation unmanned boat information, and the blue side’s dynamic defense unmanned boat information.

4. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 2 is characterized in that: In step 2, the termination condition of the red-blue confrontation iteration is designed using the Stackelberg game framework, where the red unmanned boat attack is set as the leader who first formulates the strategy, and the blue unmanned boat defense is set as the follower who selects its own optimal strategy after observing the leader's strategy; the termination condition is: {|p i -p goal |≤ε∪num isalive ()=0|i∈N} Where p i and p goal They represent the position of the Red Team’s unmanned boat numbered i and the expected arrival position of the Red Team’s mission target, ε represents the mission completion distance threshold, and num isalive The () function indicates checking the number of red teams with mobility capabilities, and N indicates the size of the red team cluster.

5. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 2 is characterized in that: In step 3, the update of the red-blue confrontation operation status is specifically as follows: Determine the operational status of the Red Army's attacking unmanned boat swarm: Step 3.1: Based on the closest approach distance in the CORLEGS rule, calculate the closest approach distance DCPA of each red UAV to each blue UAV that enters its sensing range. ij : Where i and j represent the numbers of the red and blue unmanned boats respectively, N represents the size of the red team, M represents the size of the blue team, and L ij represents the Euclidean distance between the red unmanned boat No. i and the blue unmanned boat No. j, v xr and v yr They represent the x-axis and y-axis components of the relative speed of the red and blue unmanned boats in the global coordinate system, respectively. r and y r They represent the x-axis and y-axis components of the relative distance between the red and blue unmanned boats in the global coordinate system, r safe Indicates the safe collision avoidance radius of the Red Army's unmanned boat; Step 3.2: Determine the predicted interaction relationship state of the red and blue UAVs based on the CORLEGS rule i When DCPA ij If it is greater than 0, there is at least one blue unmanned boat crossing the bow of the red unmanned boat, and the red unmanned boat needs to avoid the obstacle. The obstacle avoidance state is state i The value is 1; on the contrary, the red unmanned boat does not need to avoid obstacles when crossing the stern of the blue unmanned boat, and the obstacle avoidance state is state i The value is 0; Step 3.3: According to the obstacle avoidance state of each red team's unmanned boat i , judge the number of unmanned boats that the red team's unmanned boat cluster needs to avoid obstacles, when the number of unmanned boats in the neighboring group needs to avoid obstacles num USV When there are less than 1, the red UAV cluster will bypass the obstacle in formation and start the weak formation S-Flocking module; when the number of obstacles to be avoided by the neighboring UAVs is num USV When there are more than one, an obstacle avoidance conflict occurs within the Red team's cluster, and the Red team's UAV cluster cannot bypass the obstacle in formation. The Red team's UAV cluster activates the offensive task assignment A-Assignment module; Determine the operational status of the blue side's defensive unmanned boat: The defense status is set based on whether the red team enters the blue team's defense unmanned boat warning range. When the red team does not enter the blue team's warning range, the blue team's unmanned boat will remain on alert; when the red team enters the blue team's warning range, the blue team's unmanned boat defense task assignment D-Assignment module will be activated; Where, L ji represents the Euclidean distance between the blue unmanned boat No. j and the red unmanned boat No. i, r detc Indicates the detection radius of the blue side's unmanned boat.

6. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 5 is characterized in that: In step 3, the red team bypasses the obstacle in formation and starts the weak formation S-Flocking module; if there is a red team unmanned boat in the formation that needs to avoid obstacles, the unmanned boat will perform three subtasks: path planning, weak formation planning, and dynamic obstacle avoidance planning. The red team unmanned boats in the remaining formations will perform path planning and weak formation planning.

7. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 6 is characterized in that: The Red side's UAV cluster activates the weak formation S-Flocking module, specifically: (1) Red Army UAV execution path planning Set the center of the Red Team's virtual formation as the starting point and the mission target location as the end point; generate two search trees in the specified search area, the forward tree Tree F and reverse tree B , randomly generate path nodes, update the node connection mode through the constraint cost of the unmanned boat model, and obtain the complete discrete path sequence; according to the complete discrete path point information target attraction, calculate the path planning virtual force; Specifically: Step 3.1.1: Use B-RRT* to search the cost C(N) from the starting point to the path node Node to evaluate the pros and cons of the current node Node for the red side's unmanned boat cluster; C(N)=C(Np)+L(N,Np)+ψ(N,Np,N.pp) Where C(Np) represents the global cost of the Node.parent distance from the search starting point; L(N,Np) represents the distance cost between Node and Node.parent; ψ(N,Np,N.pp) represents the cost of turning from the path Node.parent.parent to Node.parent and then to Node; Step 3.1.2: Query the NearNode and C(NearNode) within the neighborhood radius of Node, update the NearNode with the minimum constraint cost and set it as Node.parent, completing one round of backtracking; query the ClansNode and C(ClansNode) of the Node family, update the ClansNode with the minimum constraint cost and set it as Node.parent, completing the second round of backtracking; Step 3.1.3: When Tree F and Tree B The distance of the last node is less than the neighborhood radius, and the new node NewNode is smoothed using bisection: In the formula, NewNode represents a new node; Tree F .e and Tree B .e represents Tree F and Tree B The last node, iscollison(Np,Tree B .e) represents the node Node.parent and the node Tree B .eTraverse static obstacles on the map; Step 3.1.4: Based on the complete discrete pathpoint information target attraction, the virtual force for the Red side's unmanned boat path planning is: Where p g1 Represents the discrete path points that attract the red team's unmanned boat No. i in the S-Flocking module at the current moment. (2) The Red Army’s unmanned boat executes the formation-maintaining FM plan According to the observable information of the Red team, the virtual force for maintaining the Red team's unmanned boat formation is: Where w s and w ds Respectively represent the positive and negative functions of sigmoid after translation and scaling; state i =0 means that the red team’s unmanned boat No. i does not introduce obstacle avoidance planning; (3) The Red Team needs the obstacle avoidance UAV to perform dynamic obstacle avoidance DOA planning According to the observable information of the blue UAV, the virtual force of the UAV's dynamic obstacle avoidance is: The total virtual force of each Red Army UAV starting the weak formation S-Flocking is:

8. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 5 is characterized in that: In step 3, the Red side's UAV swarm initiates the attack mission assignment A-Assignment as follows: When the Red UAV does not form a formation to bypass the Blue UAV's dynamic defense, the Blue UAV that threatens the formation will be added to the Red UAV's attack task allocation queue. The Hungarian algorithm is used as the attack task allocation framework, in which the cost calculation includes the Manhattan distance cost and the relative speed cost of the Red and Blue sides respectively. The Red UAV with the smallest cost is dispatched to intercept. The cost evaluation function is: C A (i,j)=α1·l r (i,j)+β1·v r (i,j)i∈N,j∈M Where, l r1 (i, j) is the red and blue Manhattan distance evaluation function, α1 is the red and blue Manhattan distance coefficient, the larger α1 is, the more biased the selection of the red unmanned boat that meets the shortest energy time constraint is; v r (i, j) is the red and blue relative speed evaluation function, β1 is the red and blue relative speed evaluation coefficient, and the larger β1 is, the more likely it is to select the red unmanned boat with the best capability for interception; The total virtual force of each Red Army UAV starting the attack mission assignment A-Assignment is: Where p g2 Indicates the interception position assigned to the red team's unmanned boat No. i based on the Hungarian task allocation algorithm at the current moment.

9. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 5 is characterized in that: The specific process of the blue side's unmanned boat starting the defense task assignment D-Assignmen module is as follows: When the red team enters the warning range of the blue team's defensive unmanned boat, the blue team adds the red team's unmanned boat to the blue team's defense task allocation queue based on the observed known red team's unmanned boat information. The Hungarian algorithm is used as the defense task allocation framework, where the cost calculation includes the Manhattan distance cost between the blue and red teams and the Manhattan distance cost between the red team and the blue team's protected object. The blue team's unmanned boat with the smallest cost is sent to defend. The cost evaluation function is: C D (j,i)=α2·l r (j,i)+β2·l r (i,goal) i∈N,j∈M Where, l r2 (i, j) is the Manhattan distance evaluation function of the blue and red sides, α2 is the Manhattan distance coefficient of the blue and red sides, and the larger α2 is, the more biased the selection of the blue side unmanned boat defense that meets the shortest energy and time constraints is; l r (i, goal) is the Manhattan distance function between the red team and the goal, β2 is the Manhattan distance coefficient between the red team and the goal. The larger β2 is, the more likely it is to select the red team's unmanned boat with the best capability for interception. The virtual force of each blue UAV starting the defense task assignment D-Assignmen is: Where p g3 Indicates the defensive position assigned to the blue unmanned boat No. j based on the Hungarian task allocation algorithm at the current moment.

10. The collaborative decision-making method for red-blue confrontation of unmanned boat swarms in a shore-port scenario according to claim 2 is characterized in that: The unmanned boat dynamics model is mapped to the unmanned boat control thrust F u and torque T r , ensuring that the planned speed and heading of the unmanned boat meet the motion model constraints: Where, T max and T min They represent the maximum and minimum thrust output by the unmanned boat, β represents the angle between the virtual resultant force and the actual speed of the unmanned boat, and w represents the torque adjustment coefficient; Control the thrust F through the unmanned boat u and torque T r The next moment's position and speed of each unmanned boat can be planned, that is, the surrounding ocean environment information of the red and blue parties as well as their own position, posture, and speed information can be updated.

Images