A Model-Based Prediction-Based Pursuit Method and System for Unmanned Vehicles

Abstract

This invention discloses a model predictive control-based unmanned surface vessel (USV) pursuit method and system, relating to the field of unmanned vehicle control. The method includes: acquiring the current state of both the target USV and the user's USV; based on the current states of both USVs, an identification model, and a mechanistic model, using an optimization algorithm to determine the control quantity of the user's USV at the next moment, with the goal of minimizing a cost function; the cost function includes distance cost, attack angle cost, attacked angle cost, and velocity cost; the identification model is used to predict the actual control quantity based on the user's expected control quantity; the mechanistic model is used to predict the state quantity at the next moment based on the actual control quantity and the current situation, the state quantity including the relative distance and line-of-sight angle between the user's USV and the target USV. This invention improves the reliability of pursuit.

Description

Technical Field

[0001] This invention relates to the field of unmanned mechanism control technology, and in particular to a model-predictive unmanned mechanism pursuit method and system. Background Technology

[0002] Chase games are a classic scenario in game theory, prevalent in real life and modern warfare, thus possessing high research value. Model predictive control (MMC), as a control method suitable for multi-input multi-output systems and convenient for implementing various constraints, has been widely used since its introduction, and research in this field has always been a hot topic. Using MMC-based methods to control agents to achieve intelligent pursuit effects is more stable and has broad research potential compared to other control methods and learning-based methods. The ocean, as an important arena for game theory and confrontation, has a complex environment where disturbances such as wind, waves, undercurrents, and other disturbances make the intelligent pursuit behavior of unmanned surface vessels unstable, thus making maritime game theory and confrontation highly valuable for research. Current adversarial pursuit is not intelligent enough. First, it considers too few factors; second, the weights and calculation methods of each factor should change when the situation changes, and the calculation method of the pursuit cost should also change; and third, the speed should be within the set of achievable speeds that will not be attacked by the opponent. Summary of the Invention

[0003] The purpose of this invention is to provide a model-predictive-based unmanned vehicle pursuit method and system, which improves the reliability of pursuit.

[0004] To achieve the above objectives, the present invention provides the following solution:

[0005] A model-based prediction-based unmanned vehicle pursuit method includes:

[0006] Obtain the current status of the other party's unmanned vehicles and the current status of our own unmanned vehicles;

[0007] Based on the current state of the other party's unmanned vehicle and the current state of our unmanned vehicle, as well as the identification model and mechanism model, with the cost function minimization as the optimization objective, the optimization algorithm is used to determine the control quantity of our unmanned vehicle at a set time.

[0008] The cost function includes distance cost, attack angle cost, attacked angle cost, and velocity cost; the identification model is used to predict the actual control quantity based on the expected control quantity of our unmanned vehicle; the mechanism model is used to predict the state quantity at the next moment based on the actual control quantity and the current situation, and the state quantity includes the relative distance and line-of-sight angle between our unmanned vehicle and the enemy's unmanned vehicle.

[0009] Optionally, the unmanned mechanism includes an unmanned surface vessel.

[0010] Optionally, both the desired control quantity and the predicted actual control quantity include speed and heading angle.

[0011] Optionally, the identification model is represented as:

[0012]

[0013] in, This represents the overall state of our unmanned mechanism at time t. express The derivative of , u1(t) represents the input of the identification model at time t, and y1(t) represents the output of the identification model at time t. u1(t)=(v hope (t),θ hope (t)) T y1(t)=(v real (t),θ real (t)) T x1(t) represents the first state variable at time t, x2(t) represents the second state variable at time t, and v hope (t) represents the expected velocity at time t, θ hope (t) represents the expected heading angle at time t, v real (t) represents the predicted actual velocity at time t, θ real (t) represents the predicted actual heading angle at time t. Represents the state matrix, Represents the input-state matrix. Represents the state-output matrix. This represents the feedthrough matrix.

[0014] Optionally, the mechanism model is represented as:

[0015]

[0016] Where, θ real (k-1) represents the actual heading angle at time k-1, Δθ real (k) represents the difference in heading angle between time k and time k-1, v real (k-1) represents the actual velocity at time k-1, Δt represents the time difference between time k and time k-1, R(k) represents the relative distance at time k, R(k-1) represents the relative distance at time k-1, q(k) represents the line-of-sight angle at time k, q(k-1) represents the line-of-sight angle at time k-1, V T (k-1) represents the velocity of the opposing unmanned mechanism at time k-1, θ T(k-1) represents the angle between the velocity of the opposing unmanned vehicle at time k-1 and the baseline, which is the line connecting our unmanned vehicle and the opposing unmanned vehicle; the line of sight angle is the angle between the line connecting our unmanned vehicle and the opposing unmanned vehicle and the due east direction, and k represents the time after discretization.

[0017] Optionally, the cost function is expressed as J = J1 + J2 + J3 + J4;

[0018] Where J1 represents the distance cost function, J2 represents the attack angle cost function, J3 represents the attacked angle cost function, and J4 represents the velocity cost function;

[0019]

[0020] Where α1 represents the first weighting coefficient, N represents the number of prediction times, and R i (t) represents the relative distance at time t, q i (t) represents the viewing angle at time t, ε i (t) represents the angle between the line connecting our unmanned vehicle and our mother ship at time t and the due east direction;

[0021]

[0022] Where α2 represents the second weighting coefficient, θ O The angle between the velocity of our unmanned vehicle and the baseline is represented by q, where q represents the line-of-sight angle.

[0023]

[0024] Where α3 represents the third weighting coefficient, θ T This indicates the angle between the speed of the opposing unmanned mechanism and the baseline.

[0025]

[0026] Among them, RAV ua v is the set of achievable velocities that the enemy's unmanned aerial vehicles (UAVs) cannot attack, i (t) represents the speed of our unmanned vehicle, RAV. a This indicates the reachable velocity set of the enemy's unmanned systems that can attack our unmanned systems.

[0027] This invention also discloses a model-predictive-based unmanned mechanism tracking system, comprising:

[0028] The status acquisition module is used to acquire the current status of the other party's unmanned mechanism and the current status of our own unmanned mechanism;

[0029] The control quantity determination module is used to determine the control quantity of our unmanned mechanism at a set time based on the current state of the other party's unmanned mechanism and the current state of our unmanned mechanism, as well as the identification model and mechanism model, with the goal of minimizing the cost function, and using an optimization algorithm.

[0030] The cost function includes distance cost, attack angle cost, attacked angle cost, and velocity cost; the identification model is used to predict the actual control quantity based on the expected control quantity of our unmanned vehicle; the mechanism model is used to predict the state quantity at the next moment based on the actual control quantity and the current situation, and the state quantity includes the relative distance and line-of-sight angle between our unmanned vehicle and the enemy's unmanned vehicle.

[0031] According to specific embodiments provided by the present invention, the present invention discloses the following technical effects:

[0032] This invention predicts control quantities through an identification model and predicts the future situation of both sides through a mechanistic model. With the optimization objective of minimizing a cost function including distance cost, attack angle cost, attacked angle cost, and speed cost, an optimization algorithm is used to determine the control quantities of our unmanned vehicle at a future set time, thereby improving the reliability of pursuit. Attached Figure Description

[0033] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This invention provides a schematic flowchart of a model-predictive-based unmanned vehicle pursuit method.

[0035] Figure 2 A situational diagram of the two unmanned mechanisms is provided for embodiments of the present invention;

[0036] Figure 3 A schematic diagram illustrating the principle of model predictive control is provided for embodiments of the present invention;

[0037] Figure 4 This invention provides a schematic diagram of an unmanned tracking system based on model prediction. Detailed Implementation

[0038] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. All other embodiments obtained by ordinary technicians in this field based on the embodiments of the present invention without making any creative efforts shall fall within the scope of protection of the present invention.

[0039] The purpose of this invention is to provide a model-predictive-based unmanned vehicle pursuit method and system, which improves the reliability of pursuit.

[0040] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0041] Example 1

[0042] like Figure 1 As shown in the figure, the unmanned vehicle pursuit method based on model prediction provided in this embodiment includes the following steps.

[0043] Step 101: Obtain the current status of the other party's unmanned mechanism and the current status of our unmanned mechanism.

[0044] The unmanned institutions mentioned include unmanned surface vessels (USVs), with the other party's unmanned institutions specifically referring to their own USVs, and our unmanned institutions specifically referring to our own USVs.

[0045] The current state of the enemy's unmanned vehicle includes its speed and coordinates, while the current state of our unmanned vehicle includes its speed, coordinates, and heading angle.

[0046] Step 101 specifically includes: reading and analyzing the current situation of both sides at each sampling time, including the speed and heading angle of both sides' unmanned surface vessels (USVs), their relative angles and distances, and the calculated coordinates of the USV. Kalman filtering and a Constant Turn Rate and Velocity (CTRV) model are used to rigorously predict the future speed and heading position of the USV. The USV's own state management equipment reads its remaining energy, speed, heading, and coordinates to facilitate appropriate control.

[0047] Based on the current speed, heading, and other conditions of our vessel, calculate the time series range of the applicable control quantities and generate a set of time series of the applicable control quantities.

[0048] This invention presents a model-based prediction-based unmanned surface vessel (USV) pursuit method applicable to simulated maritime environments as well as real-world complex sea conditions. Each USV is equipped with corresponding external sensing devices and its own state management equipment. The USV's own information includes a unique identifier (ID), longitude, latitude, yaw angle, speed, driving force, and rudder.

[0049] like Figure 2 As shown, a geometric diagram of the positions of both sides is established, where O represents our unmanned surface vessel (USV) and T represents the enemy's USV. The position coordinates of our USV at time t are (x... O (t),y O (t)), the position coordinates of the enemy unmanned surface vessel at time t are (x) T (t),y T (t)), θ O θ is the angle between our vessel's speed and the baseline, which is the line connecting our unmanned aerial vehicle (UAV) and the enemy's UAV. T The angle between the speed of the opposing vessel and the baseline, and the relative distance. The angle (line of sight) between the line connecting our unmanned aerial vehicle and the enemy's unmanned aerial vehicle and the due east direction is q∈[-π,π].

[0050] Step 102: Based on the current state of the other party's unmanned mechanism and the current state of our unmanned mechanism, as well as the identification model and mechanism model, with the cost function minimization as the optimization objective, the optimization algorithm is used to determine the control quantity of our unmanned mechanism at a set time.

[0051] The cost function includes distance cost, attack angle cost, attacked angle cost, and velocity cost; the identification model is used to predict the actual control quantity based on the expected control quantity of our unmanned vehicle; the mechanism model is used to predict the state quantity at the next moment based on the actual control quantity and the current situation, and the state quantity includes the relative distance and line-of-sight angle between our unmanned vehicle and the enemy's unmanned vehicle.

[0052] Our unmanned mechanism is controlled by a control variable set at a future time.

[0053] Step 102 specifically includes: establishing a system model, which includes an identification model and a mechanism model. For example... Figure 3 As shown, the input to the identification model is the desired velocity v. hope and expected heading angle θ hope The output is the predicted actual velocity v. real and actual heading angle θ real The input and output data of the identification model are collected by conducting experiments on simulators or actual vessels. The MATLAB model identification toolbox is used for system identification. By changing the identification parameters, methods and data, multiple identifications are performed. The best and most suitable model is selected as the identification model according to actual needs.

[0054] The mechanism model is designed based on the control requirements of the unmanned surface vessel and takes into account factors such as complexity. It can also be applied to other intelligent agents. The predicted velocity heading sequence obtained by the identification model and the situation information obtained by the system perception are input into the mechanism model, and the output is the predicted state quantity, which is the prediction of the future battlefield situation. These predicted quantities are then input into the cost function.

[0055] The cost value corresponding to the control time series is calculated by the cost function. The design of the cost function takes into account the antagonism between the two sides and the protection mission of our mothership. The smaller the cost function value obtained at a certain moment, the safer and more energy-efficient the pursuit of our ship by the time control sequence is, and the higher the success rate of the pursuit. It is also more difficult for the enemy ship (corresponding to the unmanned ship) to attack our ship (our unmanned ship) and our mothership. Therefore, our mothership is safer.

[0056] The boundary of the control time series is determined based on the performance characteristics of the vessel itself. Then, the optimization algorithm finds the best time series within this range and applies it to the unmanned vessel. The program calculates the best control sequence for the next N moments, but only outputs the optimal control quantity corresponding to the next moment in the control sequence (this is also the principle of MPC). The optimal control quantity includes speed and heading change.

[0057] Both the desired control quantity and the predicted actual control quantity include speed and heading angle.

[0058] The identification model is represented as follows:

[0059]

[0060] in, This represents the overall state of our unmanned mechanism at time t. express The derivative of , u1(t) represents the input of the identification model at time t, and y1(t) represents the output of the identification model at time t. u1(t)=(v hope (t),θ hope (t)) T y1(t)=(v real (t),θ real (t)) T x1(t) represents the first state variable at time t, x2(t) represents the second state variable at time t, and v hope (t) represents the expected velocity at time t, θ hope (t) represents the expected heading angle at time t, v real (t) represents the predicted actual velocity at time t, θ real (t) represents the predicted actual heading angle at time t. Represents the state matrix, Represents the input-state matrix. Represents the state-output matrix. Represents the feedthrough matrix, and This was derived from analysis using the MATLAB Model Identification Toolbox.

[0061] The overall state parameters of our unmanned mechanism include its position, velocity, acceleration, and attitude.

[0062] The derivatives of the relative distance and the relative angle are:

[0063]

[0064]

[0065] Let (x3, x4) T =(R,q) T The state-space expression can be obtained as follows:

[0066]

[0067] Finally, the state-space expression is discretized to obtain:

[0068]

[0069]

[0070]

[0071] θ real (k)=θ real (k-1)+Δθ real (k)

[0072]

[0073] R(k) = x3(k), q(k) = x4(k) (5);

[0075] x3 = x3(k) represents the third state quantity at time k, x4 = x4(k) represents the fourth state quantity at time k, and k represents the discrete time.

[0076] The cost value corresponding to the control time series can be calculated from the cost function, and the cost function value corresponding to other control time series can be calculated using the same steps.

[0077] The cost function is expressed as J = J1 + J2 + J3 + J4;

[0078] Where J1 represents the distance cost function, J2 represents the attack angle cost function, J3 represents the attacked angle cost function, and J4 represents the velocity cost function;

[0079] (1) Distance cost function: Consider that our vessel is close to the enemy vessel and during the pursuit, our vessel is between the enemy vessel and our mother vessel.

[0080]

[0081] Where α1 represents the first weighting coefficient, N represents the number of prediction steps, i.e., the number of prediction times, and R... i (t) represents the relative distance at time t, q i (t) represents the viewing angle at time t, ε i (t) represents the angle between the line connecting our unmanned vehicle and our mother ship at time t and the due east direction;

[0082] (2) Attack angle cost function: with counterclockwise direction as positive.

[0083]

[0084] Where α2 represents the second weighting coefficient, θ O The angle between the velocity of our unmanned vehicle and the baseline is represented by q, where q represents the line-of-sight angle.

[0085] (3) Attack angle cost function: with counterclockwise direction as positive.

[0086]

[0087] Where α3 represents the third weighting coefficient, θ T This indicates the angle between the speed of the opposing unmanned mechanism and the baseline.

[0088] (4) Speed ​​cost function: The speed of our vessel needs to be within RAV. ua (RAV ua To ensure that our vessel is within the reachable speed range of the enemy vessel (so that it cannot be attacked by the enemy vessel), and that our vessel is within the enemy vessel's non-attack range when it approaches the enemy vessel.

[0089]

[0090] Among them, v i (t) represents the speed of our unmanned vehicle, RAV. ua The RAV is designed to achieve a speed range where our unmanned aerial vehicles (UAVs) can catch up with the enemy's UAVs, while the enemy's UAVs cannot attack our UAVs. a This indicates the speed at which our unmanned vehicles can catch up with the enemy's unmanned vehicles, and the speed at which the enemy's unmanned vehicles can cause damage to our unmanned vehicles.

[0091] The optimization algorithm used is a phased algorithm. In the first phase, an efficient random search is performed within the feasible region, iteratively obtaining the overall optimal individual for the current and previous generations. In the second phase, a designed formula is used to approach the optimal individual obtained in the first phase. Simultaneously, in each generation's comparison, each new individual is compared with the previous optimal individual; if the new individual is better, it replaces the previous optimal individual. At the end of each generation's comparison, the worst individual is replaced with a random individual. This ensures rapid convergence while preventing getting trapped in local optima. The optimization algorithm ultimately iterates to obtain the corresponding optimal control time series, which is then applied to the unmanned surface vessel to achieve optimal control.

[0092] The optimization algorithm is used to iteratively obtain the corresponding optimal control quantity time series, which is then applied to the unmanned surface vessel. The pseudocode of the optimization algorithm is shown in Table 1. First, various parameters are initialized, mainly according to the problem settings. Then, the first stage of random search is performed as shown in Equations 10 and 11, where X represents an individual, A... m The value f represents the individual's ability to search for the optimal solution. The switching between the first and second stages relies on the threshold Q, which is calculated as shown in Formula 12. The calculation of the iteration maturity Temp of the current generation is shown in Formula 13. The individual update formula for the second stage is shown in Formula 14. After each generation is calculated, the worst individual is replaced with the random individual formula of the first stage to prevent getting trapped in local optima.

[0093] X i (t′+1)=X rand (t′)±c2×A m ×(X max -X min )×rand+X min (10);

[0094]

[0095]

[0096] X i (t′+1)=X food ±c3×Temp×rand×(X food -X i (t′)) (13);

[0097]

[0098] X i (t′+1) represents the i-th individual in the t′+1-th generation, X i(t′) represents the i-th individual in generation t′, where the individual includes control quantities (velocity and heading angle) applied to several future frames, f rand f represents the generation value of a random individual in the previous generation population. i It is the cost value of the i-th individual in the previous generation population, where c1 represents the first adjustment coefficient, c2 represents the second adjustment coefficient, and X is the cost value of the i-th individual in the previous generation population. rand (t′) represents a randomly selected individual in generation t′, X max X represents the optimal individual (with the lowest generation value) in this generation of the population. min This represents the worst individual in this generation of the population (with the highest generation value), rand represents a random number, and X... food This represents the individual with the optimal cost value.

[0099] Table 1. Pseudocode of the optimization algorithm

[0100]

[0101] Where Q` represents the threshold. This indicates that the best individual in the previous generation is the same as the best individual in the next generation. If the best individual in the previous generation is the same as the best individual in the next generation, the algorithm converges, the loop optimization ends, and the optimal control quantity u = {v, θ} is output from the best individual X, where v represents the velocity of our unmanned vehicle and θ represents the heading angle of our unmanned vehicle.

[0102] Compared to existing technologies, this invention utilizes a model predictive control approach. Based on this concept, the design and innovations are implemented, and through innovations in modules such as the model, cost function, and optimization algorithm, a good effect is achieved in the adversarial pursuit of unmanned surface vessels. Furthermore, this invention is not limited to the pursuit of unmanned surface vessels; it can also be used for the pursuit of unmanned aerial vehicles and unmanned vehicles, although the corresponding models need to be adapted accordingly.

[0103] This method is applied to unmanned surface vessel (USV) pursuit game scenarios in complex sea conditions. The optimization algorithm solves the problem quickly and is not prone to getting trapped in local optima. Secondly, the cost function design also considers factors such as the pursuit angle, speed, and distance, which can ensure the safety of our USV and our mother ship, while improving the success rate of the pursuit. Finally, the model design is simple and easy to calculate. With the addition of obstacle avoidance and anti-interference algorithms for USVs, it can achieve good control effect for USVs.

[0104] Example 2

[0105] like Figure 4 As shown, this embodiment discloses a model-predictive-based unmanned mechanism tracking system, including:

[0106] The status acquisition module 201 is used to acquire the current status of the other party's unmanned mechanism and the current status of our unmanned mechanism.

[0107] The control quantity determination module 202 is used to determine the control quantity of our unmanned mechanism at a set time based on the current state of the other party's unmanned mechanism and the current state of our unmanned mechanism, as well as the identification model and mechanism model, with the goal of minimizing the cost function, using an optimization algorithm.

[0108] The cost function includes distance cost, attack angle cost, attacked angle cost, and velocity cost; the identification model is used to predict the actual control quantity based on the expected control quantity of our unmanned vehicle; the mechanism model is used to predict the state quantity at the next moment based on the actual control quantity and the current situation, and the state quantity includes the relative distance and line-of-sight angle between our unmanned vehicle and the enemy's unmanned vehicle.

[0109] The various embodiments in this specification are described in a progressive manner, with each embodiment focusing on its differences from other embodiments. Similar or identical parts between embodiments can be referred to interchangeably. For the systems disclosed in the embodiments, since they correspond to the methods disclosed in the embodiments, the descriptions are relatively simple; relevant parts can be referred to the method section.

[0110] This document uses specific examples to illustrate the principles and implementation methods of the present invention. The descriptions of the above embodiments are only for the purpose of helping to understand the method and core ideas of the present invention. Furthermore, those skilled in the art will recognize that, based on the ideas of the present invention, there will be changes in the specific implementation methods and application scope. Therefore, the content of this specification should not be construed as a limitation of the present invention.

Claims

1. A model-predictive-based unmanned vehicle pursuit method, characterized in that, include: Obtain the current status of the other party's unmanned vehicles and the current status of our own unmanned vehicles; Based on the current state of the other party's unmanned vehicle and the current state of our unmanned vehicle, as well as the identification model and mechanism model, with the cost function minimization as the optimization objective, the optimization algorithm is used to determine the control quantity of our unmanned vehicle at a set time. The cost function includes distance cost, attack angle cost, attacked angle cost, and velocity cost; the identification model is used to predict the actual control quantity based on the expected control quantity of our unmanned vehicle; the mechanism model is used to predict the state quantity at the next moment based on the actual control quantity and the current situation, and the state quantity includes the relative distance and line-of-sight angle between our unmanned vehicle and the enemy's unmanned vehicle.

2. The unmanned vehicle pursuit method based on model prediction according to claim 1, characterized in that, The unmanned mechanisms include unmanned surface vessels.

3. The unmanned vehicle pursuit method based on model prediction according to claim 1, characterized in that, Both the desired control quantity and the predicted actual control quantity include speed and heading angle.

4. The unmanned vehicle pursuit method based on model prediction according to claim 3, characterized in that, The identification model is represented as follows: in, This represents the overall state of our unmanned mechanism at time t. express The derivative of , u1(t) represents the input of the identification model at time t, and y1(t) represents the output of the identification model at time t. u1(t)=(v hope (t),θ hope (t)) T y1(t)=(v real (t),θ real (t)) T x1(t) represents the first state variable at time t, x2(t) represents the second state variable at time t, and v hope (t) represents the expected velocity at time t, θ hope (t) represents the expected heading angle at time t, v real (t) represents the predicted actual velocity at time t, θ real (t) represents the predicted actual heading angle at time t. Represents the state matrix, Represents the input-state matrix. Represents the state-output matrix. This represents the feedthrough matrix.

5. The unmanned vehicle pursuit method based on model prediction according to claim 4, characterized in that, The mechanism model is expressed as follows: Where, θ real (k-1) represents the actual heading angle at time k-1, Δθ real (k) represents the difference in heading angle between time k and time k-1, v real (k-1) represents the actual velocity at time k-1, Δt represents the time difference between time k and time k-1, R(k) represents the relative distance at time k, R(k-1) represents the relative distance at time k-1, q(k) represents the line-of-sight angle at time k, q(k-1) represents the line-of-sight angle at time k-1, V T (k-1) represents the velocity of the opposing unmanned mechanism at time k-1, θ T (k-1) represents the angle between the velocity of the opposing unmanned vehicle at time k-1 and the baseline, which is the line connecting our unmanned vehicle and the opposing unmanned vehicle; the line of sight angle is the angle between the line connecting our unmanned vehicle and the opposing unmanned vehicle and the due east direction, and k represents the time after discretization.

6. The unmanned vehicle pursuit method based on model prediction according to claim 5, characterized in that, The cost function is expressed as J = J1 + J2 + J3 + J4; Where J1 represents the distance cost function, J2 represents the attack angle cost function, J3 represents the attacked angle cost function, and J4 represents the velocity cost function; Where α1 represents the first weighting coefficient, N represents the number of prediction times, and R i (t) represents the relative distance at time t, q i (t) represents the viewing angle at time t, ε i (t) represents the angle between the line connecting our unmanned vehicle and our mother ship at time t and the due east direction; Where α2 represents the second weighting coefficient, θ O The angle between the velocity of our unmanned vehicle and the baseline is represented by q, where q represents the line-of-sight angle. Where α3 represents the third weighting coefficient, θ T This indicates the angle between the speed of the opposing unmanned mechanism and the baseline. Among them, RAV ua v is the set of achievable velocities that the enemy's unmanned aerial vehicles (UAVs) cannot attack, i (t) represents the speed of our unmanned vehicle, RAV. a This indicates the reachable velocity set of the enemy's unmanned systems that can attack our unmanned systems.

7. A model-predictive-based unmanned mechanism tracking system, characterized in that, include: The status acquisition module is used to acquire the current status of the other party's unmanned mechanism and the current status of our own unmanned mechanism; The control quantity determination module is used to determine the control quantity of our unmanned mechanism at a set time based on the current state of the other party's unmanned mechanism and the current state of our unmanned mechanism, as well as the identification model and mechanism model, with the goal of minimizing the cost function, and using an optimization algorithm. The cost function includes distance cost, attack angle cost, attacked angle cost, and velocity cost; the identification model is used to predict the actual control quantity based on the expected control quantity of our unmanned vehicle; the mechanism model is used to predict the state quantity at the next moment based on the actual control quantity and the current situation, and the state quantity includes the relative distance and line-of-sight angle between our unmanned vehicle and the enemy's unmanned vehicle.

Images