An unmanned boat interception method based on restricted angle collision

Abstract

The present invention relates to an unmanned boat interception method based on restricted angle collision, including: constructing a simulation environment; modeling the unmanned boat p and the target unmanned boat e; obtaining the position information and speed information of the unmanned boat p at time t, as well as the distance information between the unmanned boat p and the unmanned boat e, and storing them; predicting the position information and speed information of the unmanned boat e according to the stored information; constructing a policy network for the unmanned boat p based on a multi-layer perceptron; designing a sliding mode surface, constructing a sliding mode control law to calculate a control signal; applying the control signal to the unmanned boat p, performing iteration, generating data at time t+1 and storing it; constructing a reward function, training the policy network parameters of the unmanned boat p based on the PPO algorithm; deploying the trained policy network to the unmanned boat so that the unmanned boat intercepts the target unmanned boat by means of collision. This method can control the unmanned boat to collide with the target unmanned boat along a restricted angle so as to achieve the interception task.

Description

Technical Field

[0001] The present invention relates to the technical field of unmanned boats, and particularly to an interception method for unmanned boats based on restricted-angle collision. Background Art

[0002] In the past decade, the technology of unmanned boats has developed rapidly. In the problem of maritime tracking and interception, the goal of the intruder is to keep a distance from the defender and reach the target area while avoiding being captured, while the goal of the defender is to track and intercept the intruder. The intruder is considered captured when it enters the capture range of the defender, is attacked by the defender with weapons, or is directly rammed by the defender.

[0003] However, most of the existing methods do not consider the collision mode between unmanned boats. However, in practical applications, in order to achieve the interception goal, the key to the collision problem lies in installing a collision device, and realizing the collision advantage through the collision device, that is, using our collision device to collide with the weak position of the enemy unmanned boat. Summary of the Invention

[0004] The purpose of the present invention is to provide an interception method for unmanned boats based on restricted-angle collision, which can control the unmanned boat to collide with the target unmanned boat along a restricted angle so as to achieve the interception task.

[0005] In order to achieve the above purpose, the technical solution adopted by the present invention is: an interception method for unmanned boats based on restricted-angle collision, including the following steps:

[0006] Step 1: Construct a simulation environment, considering an unmanned boat p with a collision device installed at the front end, whose goal is to intercept another target unmanned boat e in the water area. The method is that the front end of the unmanned boat p rams the center point of the hull of the unmanned boat e and the included angle Δθ between the advancing speed directions of the two unmanned boats is within a set range; model the unmanned boat p and the unmanned boat e, assuming that the maximum speed of the unmanned boat p is greater than that of the unmanned boat e, while the maximum acceleration of the unmanned boat p is less than that of the unmanned boat e, and introduce the open-loop system model of the unmanned boat p and the closed-loop system model of the unmanned boat e;

[0007] Step 2: Obtain the position information s t p and speed information v t p of the unmanned boat p at time t t pe and the distance information s between the unmanned boat p and the unmanned boat e

[0008] through a measuring device and speed information ;

[0009] Step 4: Based on the multi-layer perceptron, construct a policy network for the unmanned boat p, with its input being , and the output being the sliding mode control gain parameter k of the unmanned boat p t ;

[0010] Step 5: Design a sliding mode surface, construct a sliding mode control law to calculate the control signal τ t p ;

[0011] Step 6: Apply the control signal τ t p to the unmanned boat p, perform iteration, generate data at time t + 1 and store it;

[0012] Step 7: Construct a reward function r, and train the policy network parameters of the unmanned boat p based on the PPO algorithm;

[0013] Step 8: Deploy the trained policy network to the unmanned boat so that the unmanned boat intercepts the target unmanned boat by means of collision.

[0014] Furthermore, in Step 1, the open-loop system model of the unmanned boat p and the closed-loop system model of the unmanned boat e both include a position loop η and a velocity loop V b ; The position loop controls the position of the unmanned boat to make it approach the target unmanned boat; the velocity loop controls the speed of the unmanned boat to make it execute the position control command; the position loop model is:

[0015]

[0016] where η = [x, y, θ] T is the pose of the unmanned boat in the global coordinate system, is the derivative of η with respect to time, x and y represent the position coordinates of the unmanned boat, and θ represents the heading angle of the unmanned boat; V b = [v x , v y , v θ T is the speed of the unmanned boat in the body coordinate system, v x 、v y 、v θ respectively represent the longitudinal speed, lateral speed and angular speed of the unmanned boat; R(θ) is the rotation matrix used to transform from the body coordinate system to the global coordinate system:

[0017]

[0018] The velocity loop model is:

[0019]

[0020] where M is the inertia matrix, including mass and added mass effects, The derivative of V with respect to time, C(V b ) is the Coriolis force and centrifugal force matrix, D(V b ) is the damping matrix, and τ is the control input; b ) is the damping matrix, τ is the control input;

[0021] The control input signal of the unmanned boat p needs to be designed and its amplitude is limited within the interval. The control input signal of the unmanned boat e is randomly selected within the amplitude interval. The maximum speed and maximum acceleration of the unmanned boat p are represented by and respectively. The maximum speed and maximum acceleration of the unmanned boat e are represented by and respectively, and they satisfy the constraint conditions:

[0022]

[0023]

[0024] Under the ideal collision condition, the front end of the unmanned boat p hits the unmanned boat e, and the included angle Δθ between the forward speed directions of the two unmanned boats satisfies:

[0025] 45° < Δθ < 135°.

[0026] Furthermore, in step 2, the position information s t p , the speed information v t p and the distance information s t pe are accurate to two decimal places:

[0027]

[0028]

[0029] where x t p and y t p respectively represent the absolute positions of the unmanned boat p on the x-axis and y-axis of the world coordinate system at time t, and respectively represent the magnitudes of the speeds of the unmanned boat in the x-axis and y-axis directions at time t; s t pe represents the relative position information between the unmanned boat p and the unmanned boat e:

[0030] .

[0031] Furthermore, in step 3, through the position information s tp and the distance information s t pe Predict the position information of the unmanned boat e , through the position information of the unmanned boat e and Predict the speed information of the unmanned boat e .

[0032] Furthermore, through geometric relationships, the position estimate of the unmanned boat e is expressed as:

[0033]

[0034] Through two frames of position estimate data and Calculate to obtain:

[0035]

[0036] Written in component form as:

[0037] ,

[0038] where 、 are the x and y components of 、 are the x and y components of, Δt represents the time step; Store the perception information, and the perception vector Obs of the unmanned boat p is expressed as:

[0039] .

[0040] Furthermore, in step 4, the multi-layer perceptron has 64 internal layers, and the activation function is the tanh function; Input the perception vector Obs of the unmanned boat p into the reinforcement learning model, and output the sliding mode control gain parameter k t ; The calculation formula of the hidden layer in the reinforcement learning model is as follows:

[0041]

[0042] where, H represents the output of the hidden layer, W represents the weight matrix for connecting the input layer and the hidden layer, X represents the input of the input layer, and b represents the bias term of the hidden layer;

[0043] The output layer is expressed as:

[0044]

[0045] where, Y represents the output of the output layer, Wout represents the weight matrix of the output layer, which is used to connect the hidden layer and the output layer, b out represents the bias term of the output layer;

[0046] Apply the tanh function to map the model output result to the range of [-1, 1]. The tanh function is expressed as:

[0047]

[0048] where e represents the base of the natural logarithm.

[0049] Furthermore, in step 5, design the sliding surface :

[0050]

[0051] where col represents concatenating two vectors, arccos represents the inverse cosine function, and || || represents the Euclidean norm of the vector;

[0052] Construct the sliding mode control law to calculate the control signal :

[0053]

[0054] where represents the control signal of the unmanned boat p at time t, represents the maximum control amplitude of the unmanned boat p, and sgn represents the sign function;

[0055] Based on the control strategy of sliding mode control, calculate the control input according to the error state to gradually reduce the position error and make the system state converge to the desired trajectory, so as to avoid collision and the collision angle is close to 90°.

[0056] Furthermore, in step 6, discretize the dynamic model of the unmanned boat p according to the sampling interval Δt, and approximate and iterate it using the Euler method. The formula is:

[0057]

[0058]

[0059] where 、 、 represent the values of η, V b 、θ of the unmanned boat p at time t respectively; R( ) is the rotation matrix, M is the inertia matrix, C( ) is the Coriolis force and centrifugal force matrix, and D( ) is the damping matrix.

[0060] ​Further, in step 7, the reward function is as follows:

[0061]

[0062]

[0063]

[0064]

[0065] Wherein, r1 represents the distance from the unmanned boat p to the unmanned boat e. The closer the distance between the unmanned boat p and the unmanned boat e, the greater the reward, thus encouraging the unmanned boat to approach the target; r2 represents the difference between the speed of the unmanned boat p and the speed of the unmanned boat e. The goal is to ensure that while the unmanned boat p approaches the target, its speed is adjusted to optimize the collision effect; the greater the speed error, the better the collision effect; r3 is used to control the included angle between the advancing speed directions of the unmanned boat p and the unmanned boat e, making it meet the constraint of 45° < Δθ < 135° and approaching 90°, that is, colliding at the optimal angle.

[0066] The present invention also provides an unmanned boat interception system based on restricted angle collision, including a memory, a processor, and computer program instructions stored on the memory and capable of being run by the processor. When the processor runs the computer program instructions, the above-mentioned method can be realized.

[0067] Compared with the prior art, the present invention has the following beneficial effects: The present invention provides an unmanned boat interception method based on restricted angle collision. This method predicts the position and speed information of the target unmanned boat according to the position and speed information of the unmanned boat itself and its distance information from the target unmanned boat, simplifying the calculation amount; at the same time, through the sliding mode control idea, the collision distance and the collision angle error are used as the sliding mode surface, so as to realize both position approaching and collision angle control; in addition, through the deep reinforcement learning algorithm, that is, the PPO algorithm, the gain parameters of the sliding mode control algorithm are dynamically updated to maximize the collision effect under the constraints of speed and acceleration. Description of the Drawings

[0068] Figure 1 It is a flowchart of the method implementation of the embodiment of the present invention. Detailed Embodiments

[0069] The present invention will be further described below in conjunction with the drawings and embodiments.

[0070] It should be noted that the following detailed description is exemplary and is intended to provide further description of the present application. Unless otherwise specified, all technical and scientific terms used herein have the same meaning as commonly understood by those of ordinary skill in the technical field to which the present application belongs.

[0071] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit the exemplary embodiments according to the present application. As used herein, unless the context clearly indicates otherwise, the singular forms are also intended to include the plural forms. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, they specify the presence of features, steps, operations, devices, components, and / or combinations thereof.

[0072] As Figure 1 shown, this embodiment provides an unmanned boat interception method based on restricted-angle collision, and the specific implementation steps are as follows.

[0073] Step 1: Build a simulation environment. Consider an unmanned boat p with a high-hardness collision device installed at the front end, whose goal is to intercept another target unmanned boat e in the water area. The method is that the front end of the unmanned boat p hits the center point of the hull of the unmanned boat e and the included angle Δθ between the advancing speed directions of the two unmanned boats is greater than 45° and less than 135°; model the unmanned boat p and the unmanned boat e. Assume that the maximum speed of the unmanned boat p is greater than that of the unmanned boat e, while the maximum acceleration of the unmanned boat p is less than that of the unmanned boat e. Introduce the open-loop system model of the unmanned boat p and the closed-loop system model of the unmanned boat e.

[0074] Both the open-loop system model of the unmanned boat p and the closed-loop system model of the unmanned boat e include a position loop η and a speed loop V b ; the position loop controls the position of the unmanned boat to approach the target unmanned boat; the speed loop controls the speed of the unmanned boat to execute the position control command. The position loop model is:

[0075]

[0076] where η = [x, y, θ] T is the pose of the unmanned boat in the global coordinate system, is the derivative of η with respect to time, x and y represent the position coordinates of the unmanned boat, and θ represents the heading angle of the unmanned boat, that is, the included angle between the advancing direction of the unmanned boat and the due north direction; V b = [v x , v y , v θ T is the speed of the unmanned boat in the body coordinate system, v x , v y , v θ respectively represent the longitudinal speed, lateral speed and angular speed of the unmanned boat; R(θ) is a rotation matrix used to transform from the body coordinate system to the global coordinate system:

[0077]

[0078] The speed loop model is:

[0079]

[0080] Among them, M is the inertia matrix, including mass and added mass effects, is the derivative of V b with respect to time, C(V b ) is the Coriolis force and centrifugal force matrix, D(V b ) is the damping matrix, and τ is the control input.

[0081] The control input signal of the unmanned boat p is to be designed and its amplitude is limited within the interval, and the control input signal of the unmanned boat e is randomly selected within the amplitude interval; the maximum speed and maximum acceleration of the unmanned boat p are represented by and respectively, and the maximum speed and maximum acceleration of the unmanned boat e are represented by and respectively, and they satisfy the constraint conditions:

[0082]

[0083]

[0084] Under the ideal collision condition, the front end of the unmanned boat p hits the unmanned boat e, and the included angle Δθ between the forward speed directions of the two unmanned boats satisfies:

[0085] 45° < Δθ < 135°.

[0086] Step 2: Obtain the position information s t p and speed information v t p of the unmanned boat p at time t based on measurement devices such as GPS, speedometers, and laser rangefinders t pe and the distance information s

[0087] between the unmanned boat p and the unmanned boat e and store them. t p Among them, the position information s t p and speed information v t pe and distance information s

[0088]

[0089]

[0090] Among them, x t p and yt p respectively represent the absolute positions of the unmanned boat p on the x-axis and y-axis of the world coordinate system at time t, and respectively represent the speed magnitudes of the unmanned boat in the x-axis and y-axis directions; s t pe represents the relative position information between the unmanned boat p and the unmanned boat e:

[0091] .

[0092] Step 3: Predict the position information of the unmanned boat e based on the stored information and speed information .

[0093] Through the position information s t p and the distance information s t pe predict the position information of the unmanned boat e , and through the position information of the unmanned boat e and predict the speed information of the unmanned boat e .

[0094] Specifically, through geometric relationships, the position estimate of the unmanned boat e is expressed as:

[0095]

[0096] Through two frames of position estimate data and calculate to obtain:

[0097]

[0098] Written in component form as:

[0099] ,

[0100] where , are 's x and y direction components, , are 's x and y direction components, Δt represents the time step; store the perception information, and the perception vector Obs of the unmanned boat p is expressed as:

[0101] .

[0102] Step 4: Based on the multi-layer perceptron, construct a policy network for the unmanned boat p, and its input is The output is the sliding mode control gain parameter k of the unmanned boat p t .

[0103] In this embodiment, the multi-layer perceptron has 64 internal layers, and the activation function is the tanh function; the perception vector Obs of the unmanned boat p is input into the reinforcement learning model, and the sliding mode control gain parameter k is output t ; the calculation formula of the hidden layer in the reinforcement learning model is as follows:

[0104]

[0105] where H represents the output of the hidden layer, W represents the weight matrix for connecting the input layer and the hidden layer, X represents the input of the input layer, and b represents the bias term of the hidden layer

[0106] The output layer is expressed as:

[0107]

[0108] where Y represents the output of the output layer, W out represents the weight matrix of the output layer for connecting the hidden layer and the output layer, and b out represents the bias term of the output layer

[0109] Apply the tanh function to map the model output result to between [-1, 1], and the tanh function is expressed as:

[0110]

[0111] where e represents the base of the natural logarithm

[0112] Step 5: Design a sliding mode surface and construct a sliding mode control law to calculate the control signal τ t p .

[0113] Specifically, design a sliding mode surface as follows:

[0114]

[0115] where col represents concatenating two vectors, arccos represents the inverse cosine function, and || || represents the Euclidean norm of the vector

[0116] Construct a sliding mode control law to calculate the control signal :

[0117]

[0118] where represents the control signal of the unmanned boat p at time t represents the maximum control amplitude of the unmanned boat p, and sgn represents the sign function.

[0119] Based on the control strategy of sliding mode control, according to the error state calculate the control input , gradually reduce the position error, and make the system state converge to the desired trajectory, so as to avoid collision and make the collision angle as close to 90° as possible.

[0120] Step 6: Apply the control signal τ t p to the unmanned boat p, perform iteration, generate data at time t + 1 and store it.

[0121] Specifically, according to the sampling interval Δt, discretize the dynamic model of the unmanned boat p, approximate and iterate using the Euler method, and the formula is:

[0122]

[0123]

[0124] where , , respectively represent the values of η, V b , θ of the unmanned boat p at time t, and the subscript t + 1 represents the value at time t + 1; R( ) is the rotation matrix, M is the inertia matrix, C( ) is the Coriolis force and centrifugal force matrix, and D( ) is the damping matrix.

[0125] Step 7: Construct the reward function r and train the policy network parameters of the unmanned boat p based on the PPO algorithm.

[0126] Specifically, the reward function is:

[0127]

[0128]

[0129]

[0130]

[0131] Among them, r1 represents the distance from the unmanned boat p to the unmanned boat e. The closer the distance between the unmanned boat p and the unmanned boat e, the greater the reward, thus encouraging the unmanned boat to approach the target; r2 represents the difference between the speed magnitude of the unmanned boat p and the speed magnitude of the unmanned boat e. The goal is to ensure that while the unmanned boat p approaches the target, its speed is adjusted to optimize the collision effect; the greater the speed error, the better the collision effect; r3 is used to control the included angle between the forward speed directions of the unmanned boat p and the unmanned boat e, making it meet the constraint of 45° < Δθ < 135° and approaching 90°, that is, colliding at the optimal angle.

[0132] Step 8: Deploy the trained policy network to the unmanned boat so that the unmanned boat intercepts the target unmanned boat by means of collision.

[0133] The present invention provides an unmanned boat interception method based on restricted angle collision. This method can control the unmanned boat to collide with the target unmanned boat at a restricted angle to achieve the interception effect. For the interception problem of unmanned boats, the existing mainstream algorithms are control strategies based on position information, that is, controlling the unmanned boat to collide according to position information. And this method is improved on the basis of the existing theory. By combining sliding mode control and reinforcement learning, the collision angle is constrained, and at the same time, the speed error of the collision is maximized, so as to maximize the collision effect and achieve the interception goal.

[0134] This embodiment also provides an unmanned boat interception system based on restricted angle collision, including a memory, a processor, and computer program instructions stored on the memory and capable of being run by the processor. When the processor runs the computer program instructions, the above-mentioned method can be implemented.

[0135] Those skilled in the art should understand that the embodiments of the present application can be provided as a method, a system, or a computer program product. Therefore, the present application can adopt the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the present application can adopt the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0136] The present application is described with reference to the flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to the embodiments of the present application. It should be understood that each process and / or block in the flowchart and / or block diagram, and the combination of processes and / or blocks in the flowchart and / or block diagram can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing devices generate for implementation in the process Figure 1one or more processes and / or blocks Figure 1 means for the functions specified in one or more blocks

[0137] These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to operate in a particular manner, such that the instructions stored in the computer-readable memory produce a manufacture including an instruction means that implements the functions in the process Figure 1 one or more processes and / or blocks Figure 1 the functions specified in one or more blocks

[0138] These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus, such that a series of operational steps are performed on the computer or other programmable apparatus to produce a computer-implemented process, whereby the instructions executed on the computer or other programmable apparatus provide steps for implementing the functions in the process Figure 1 one or more processes and / or blocks Figure 1 the functions specified in one or more blocks

[0139] As described above, the above are only the preferred embodiments of the present invention, and are not intended to limit the present invention in other forms. Any person skilled in the art may use the technical content disclosed above to make changes or modifications into equivalent embodiments with equivalent changes. However, any simple modifications, equivalent changes and modifications made to the above embodiments based on the technical essence of the present invention without departing from the technical solution content of the present invention still fall within the protection scope of the technical solution of the present invention.

Claims

1. An unmanned boat interception method based on restricted angle collision, characterized in that, It includes the following steps: Step 1: Build a simulation environment. Consider the unmanned boat p with a collision device installed at the front end, whose goal is to intercept another target unmanned boat e in the water area. The method is that the front end of the unmanned boat p hits the center point of the hull of the unmanned boat e and the included angle Δθ between the advancing speed directions of the two unmanned boats is within a set range; model the unmanned boat p and the unmanned boat e. Assume that the maximum speed of the unmanned boat p is greater than that of the unmanned boat e, while the maximum acceleration of the unmanned boat p is less than that of the unmanned boat e. Introduce the open-loop system model of the unmanned boat p and the closed-loop system model of the unmanned boat e; Step 2: Obtain the position information s of the unmanned boat p at time t based on the measuring device t p and the speed information v t p as well as the distance information s between the unmanned boat p and the unmanned boat e t pe and store them; Step 3: Predict the position information and speed information of the unmanned boat e based on the stored information and speed information ; Step 4: Based on the multi-layer perceptron, construct a policy network for the unmanned boat p, whose input is , and the output is the sliding mode control gain parameter k of the unmanned boat p t ; Step 5: Design the sliding mode surface and construct the sliding mode control law to calculate the control signal τ t p ; Step 6: Apply the control signal τ t p to the unmanned boat p, perform iteration, generate data at time t + 1 and store it; Step 7: Build a reward function r and train the policy network parameters of the unmanned boat p based on the PPO algorithm; Step 8: Deploy the trained policy network to the unmanned boat so that the unmanned boat intercepts the target unmanned boat by means of collision; In step 5, design the sliding mode surface : Among them, col represents concatenating two vectors, arccos represents the inverse cosine function, and || || represents the Euclidean norm of the vector; Construct the sliding mode control law to calculate the control signal : Among them, represents the control signal of the unmanned boat p at time t, represents the maximum control amplitude of the unmanned boat p, and sgn represents the sign function; A control strategy based on sliding mode control calculates the control input according to the error state to gradually reduce the position error and converge the system state to the desired trajectory, thus avoiding collisions and ensuring that the collision angle is close to 90°. ​ 2. The method for intercepting an unmanned boat based on restricted-angle collision according to claim 1, characterized in that In Step 1, the open-loop system model of the unmanned boat p and the closed-loop system model of the unmanned boat e both include a position loop η and a speed loop V b ; the position loop controls the position of the unmanned boat to make it approach the target unmanned boat; the speed loop controls the speed of the unmanned boat to make it execute the position control command; the position loop model is: where η = [x, y, θ] T is the pose of the unmanned boat in the global coordinate system, is the derivative of η with respect to time, x and y represent the position coordinates of the unmanned boat, and θ represents the heading angle of the unmanned boat; V b = [v x , v y , v θ T is the velocity of the unmanned boat in the body coordinate system, v x , v y , v θ respectively represent the longitudinal velocity, lateral velocity and angular velocity of the unmanned boat; R(θ) is the rotation matrix used to transform from the body coordinate system to the global coordinate system:​ The speed loop model is: where M is the inertia matrix, including mass and added mass effects, is the derivative of V b with respect to time, C(V b ) is the Coriolis and centrifugal force matrix, D(V b ) is the damping matrix, and τ is the control input; The control input signal of the unmanned boat p is to be designed and its amplitude is limited within the interval, and the control input signal of the unmanned boat e is randomly selected within the amplitude interval; the maximum speed and maximum acceleration of the unmanned boat p are represented by and respectively, and the maximum speed and maximum acceleration of the unmanned boat e are represented by and respectively, and they satisfy the constraint conditions: Under the ideal collision condition, the front end of the unmanned boat p hits the unmanned boat e, and the included angle Δθ between the advancing speed directions of the two unmanned boats satisfies: 45° < Δθ < 135°.

3. The method for intercepting an unmanned boat based on restricted-angle collision according to claim 1, characterized in that In step 2, the precision of the position information s t p , the speed information v t p and the distance information s t pe is taken to two decimal places: Among them, x t p and y t p respectively represent the absolute positions of the unmanned boat p on the x-axis and y-axis of the world coordinate system at time t. and respectively represent the magnitudes of the velocities of the unmanned boat in the x-axis and y-axis directions; s t pe represents the relative position information between the unmanned boat p and the unmanned boat e: 。 4. A method for intercepting an unmanned boat based on restricted-angle collision according to claim 1, characterized in that In step 3, based on the position information s t p and the distance information s t pe predict the position information of the unmanned boat e , and based on the position information of the unmanned boat e and predict the speed information of the unmanned boat e .

5. A method for intercepting an unmanned boat based on restricted-angle collision according to claim 4, characterized in that, Through geometric relations, the position estimation of the unmanned boat e is expressed as: Through two-frame position estimation data And Calculated as follows: Written in component form as: ， Among them, and are the x- and y-direction components of and are the x- and y-direction components of , where Δt represents the time step; storing the perception information, the perception vector Obs of the unmanned boat p is expressed as: 。 6. The method for intercepting an unmanned boat based on restricted-angle collision according to claim 1, wherein In step 4, the multi-layer perceptron has 64 internal layers, and the activation function is the tanh function; the perception vector Obs of the unmanned boat p is input into the reinforcement learning model, and the sliding mode control gain parameter k is output t ; the calculation formula of the hidden layer in the reinforcement learning model is as follows: Among them, H represents the output of the hidden layer, W represents the weight matrix used to connect the input layer and the hidden layer, X represents the input of the input layer, and b represents the bias term of the hidden layer; The output layer is expressed as: Among them, Y represents the output of the output layer, and W out represents the weight matrix of the output layer, which is used to connect the hidden layer and the output layer, and b out represents the bias term of the output layer; Apply the tanh function to map the model output result to the range of [-1, 1]. The tanh function is expressed as: Among them, e represents the base of the natural logarithm.

7. A method for intercepting an unmanned boat based on restricted-angle collision according to claim 1, characterized in that In Step 6, according to the sampling interval Δt, discretize the dynamic model of the unmanned boat p, approximate and iterate using the Euler method, and the formula is: Among them, , , respectively represent the values of η, V b , θ of the unmanned boat p at time t; R( ) is the rotation matrix, M is the inertia matrix, C( ) is the Coriolis force and centrifugal force matrix, and D( ) is the damping matrix.

8. A method for intercepting an unmanned boat based on restricted angle collision according to claim 1, characterized in that In Step 7, the reward function is: Among them, r1 represents the distance from the unmanned boat p to the unmanned boat e. The closer the distance between the unmanned boat p and the unmanned boat e, the greater the reward, thus encouraging the unmanned boat to approach the target; r2 represents the difference between the speed magnitude of the unmanned boat p and the speed magnitude of the unmanned boat e. The goal is to ensure that while the unmanned boat p approaches the target, it adjusts its speed to optimize the collision effect; the greater the speed error, the better the collision effect; r3 is used to control the included angle between the advancing speed directions of the unmanned boat p and the unmanned boat e, making it meet the constraint of 45° < Δθ < 135° and approach 90°, that is, collide at the optimal angle.

9. An unmanned boat interception system based on restricted angle collision, characterized in that, It includes a memory, a processor, and computer program instructions stored on the memory and capable of being run by the processor. When the processor runs the computer program instructions, it can implement the method described in any one of claims 1 - 8.

Images