A method and system for constructing an intelligent confrontation virtual environment for real scenarios

Abstract

The present invention relates to the construction of an intelligent confrontation virtual environment, and specifically to a method and system for constructing an intelligent confrontation virtual environment for real scenarios. In order to solve the technical problem that the existing virtual learning environment cannot make the rules and logic learned by the intelligent body applicable to real scenarios, the method for constructing an intelligent confrontation virtual environment for real scenarios provided by the present invention includes the following steps: presetting natural environment parameters, as well as physical parameters of multiple defense devices and multiple target bodies; collecting initial data and obtaining initial action instructions of the defense device based on the data; collecting and storing the environmental state quantity at the current moment, and obtaining the environmental state quantity at the next moment based on the obtained initial action instructions; deciding the action instruction of the defense device at the next moment based on the environmental state quantity at the next moment; and looping through the above steps until the confrontation ends, terminating the loop, and completing the construction of the intelligent confrontation virtual environment.

Description

Technical Field

[0001] The present invention relates to the construction of an intelligent confrontation virtual environment, and in particular to a method and system for constructing an intelligent confrontation virtual environment for real scenarios. Background Art

[0002] With the advancement of intelligence and the emergence of new defensive methods, the defense model is gradually shifting from a traditional human-centered approach to one centered on intelligence. These emerging defensive methods, unlike traditional ones, primarily rely on energy transfer, such as optical and electromagnetic energy transfer. The complex state space, asymmetric information, and intense adversarial dynamics of defensive confrontation are very similar to those found in StarCraft. Players must make dynamic decisions based on the ever-changing battlefield situation and limited information. In 2019, the DeepMind team attempted to generalize deep reinforcement learning to games with continuous state spaces and incomplete information, and launched a new generation of intelligent agents, AlphaStar. This agent competed against human players in StarCraft II, continuously improving its decision-making ability and ultimately reaching the master-level performance of human players. As a self-improving and self-evolving intelligent agent, it relies on a specific learning environment. Therefore, all the rules and strategies it learns are derived from this learning environment.

[0003] For intelligent defensive decision-making, the process of establishing a reinforcement learning environment is essentially a virtualization of a specific defensive scenario. The rules and logic learned by the agent are constrained by the logic of the learning environment itself. For example, the reinforcement learning environment used by AlphaStar was built specifically for the game StarCraft II and is based on the game's virtual logic. Therefore, the various combat strategies learned by AlphaStar are based on the game's own operating logic. However, for real-world defensive confrontation scenarios, such a game environment is obviously not suitable. A specialized reinforcement learning environment is required, one that reflects the objective physical world. Summary of the Invention

[0004] The purpose of the present invention is to solve the technical problem that the existing virtual learning environment cannot make the rules and logic learned by the intelligent agent applicable to real scenes, and to provide a method and system for constructing an intelligent confrontation virtual environment for real scenes.

[0005] To solve the above technical problems, the present invention provides the following technical solutions:

[0006] A method for constructing an intelligent confrontation virtual environment for real scenarios is special in that it includes the following steps:

[0007] Step 1: Preset the natural environment parameters at the initial moment and generate the required natural environment according to the set natural environment parameters;

[0008] The natural environment is provided with a plurality of defense devices and a plurality of target objects for intelligent confrontation, and each defense device includes a detection model and a tracking model;

[0009] Presetting physical parameters of multiple defense devices and multiple target objects at an initial moment, wherein the physical parameters include physical position parameters and physical state parameters;

[0010] Step 2: Record the natural environment parameters at the initial moment, as well as the physical parameters of each of the defense devices and each of the target objects at the initial moment, as the environmental state quantities at the initial moment;

[0011] Step 3, determining the action instructions of each defense device at the initial moment according to the obtained environmental state quantity at the initial moment;

[0012] Step 4: determine whether the action instruction is a complex instruction;

[0013] If so, perform instruction parsing to obtain an executable instruction sequence and execute step 5;

[0014] If not, get the executable instruction and execute step 5;

[0015] Step 5: Collect the natural environment parameters at the current moment, as well as the physical parameters of each of the defense devices and each of the target objects at the current moment, record them as the environmental state quantity at the current moment, and store them;

[0016] Calculate the environmental state quantity at the next moment based on the environmental state quantity at the current moment and the obtained executable instruction or executable instruction sequence;

[0017] Step 6: Update the stored current state of the environment to the obtained state of the environment at the next moment;

[0018] Step 7: Determine the action instructions of each defense device at the next moment based on the environmental state at the next moment;

[0019] Step 8: Loop through steps 4 to 7 until the confrontation ends, terminate the loop, and complete the construction of the intelligent confrontation virtual environment.

[0020] Furthermore, in step 4, the action instructions decided include a move instruction, a stop instruction, and a strike instruction. The move instruction carries a moving destination coordinate parameter, and the strike instruction carries a target number parameter.

[0021] The strike instruction is a complex instruction, including a tracking and aiming instruction and an attack instruction, wherein the tracking and aiming instruction is the first executable instruction of the strike instruction, and the attack instruction is the second executable instruction of the strike instruction; the tracking and aiming instruction, attack instruction, move instruction and stop instruction are all executable instructions.

[0022] Furthermore, in step 1, the physical state parameters of the defense device include detection state parameters, tracking state parameters and energy state parameters of the defense device; the physical state parameters of the target body include flight state parameters and damage state parameters of the target body.

[0023] Furthermore, in step 5, the environmental state quantity at the next moment is calculated based on the environmental state quantity at the current moment and the obtained executable instruction or executable instruction sequence, including the following steps:

[0024] If the executable instruction is obtained in step 4, the natural environment parameters at the next moment are obtained according to the meteorological environment evolution model and the natural environment parameters in the environmental state quantity at the current moment;

[0025] Determine whether the executable instruction is a move instruction or a stop instruction: If it is a move instruction, calculate the physical position parameters of each defense device at the next moment based on the coordinate parameters of the movement destination carried by the move instruction; if it is a stop instruction, directly use the physical position parameters of each defense device at the current moment as the physical position parameters of the next moment;

[0026] According to the flight control algorithm of the target body and the flight state parameters of each target body in the environmental state quantity at the current moment, the physical position parameters of each target body at the next moment are obtained;

[0027] If an executable instruction sequence is obtained in step 4, the detection state parameters, tracking state parameters, and strike parameters of each defense device at the next moment are calculated based on the detection model of each defense device, the environmental state quantity at the current moment, and the obtained executable instruction sequence; the damage state parameters of each target body at the next moment are calculated based on the obtained strike parameters of each defense device at the next moment and the target body damage model.

[0028] Furthermore, in step 5, obtaining the detection state parameters, tracking state parameters, and strike parameters of each defense device at the next moment includes the following steps:

[0029] Calculate the detection state parameters and tracking state parameters of each defense device at the next moment based on the detection model in each defense device, the first executable instruction in the obtained executable instruction sequence, and the detection state parameters and tracking state parameters in the environmental state quantity at the current moment;

[0030] Determining whether each defense device is in a stable tracking state according to the detection state parameters and tracking state parameters of each defense device at the next moment;

[0031] If the tracking state is stable, the second executable instruction is accessed, and the attack parameters of the defense device at the next moment are calculated based on the energy state parameters of the defense device and the flight state parameters and damage state parameters in the current environmental state.

[0032] If the tracking state is not stable, continue to track and aim.

[0033] Furthermore, in step 7, the action instructions of each defense device at the next moment are determined according to the environmental state quantity at the next moment, specifically:

[0034] Obtaining the visible state of each defense device at the next moment based on the environmental state at the next moment and the detection model of each defense device; and determining the action instruction of each defense device at the next moment based on the visible state of each defense device at the next moment;

[0035] Alternatively, the action instructions of each defense device at the next moment are directly determined based on the environmental state quantity at the next moment.

[0036] Furthermore, in step 1, the defense device adopts energy transmission type defense means, and the energy transmission type defense means includes light energy transmission and electromagnetic energy transmission;

[0037] The target object is a drone, and multiple target objects form a formation of drones.

[0038] At the same time, the present invention provides a real-scene-oriented intelligent confrontation virtual environment construction system, which is used to implement the above-mentioned real-scene-oriented intelligent confrontation virtual environment construction method. The special feature of the system is that it includes an instruction parser, a natural environment generator, an environment evolver, a state pool and a state feedback device; the output end of the natural environment generator is connected to the input end of the state pool, the output end of the instruction parser is connected to the input end of the environment evolver, one output end of the environment evolver is connected to the state pool and the instruction parser in sequence, the other output end of the environment evolver and the other output end of the state pool are both connected to the input end of the state feedback device, the output end of the state feedback device is used to connect to the input end of an external intelligent agent, the other input end of the instruction parser is used to connect to the output end of an external intelligent agent, and the external intelligent agent is used to receive the environmental state quantity of each defense device at the next moment, and decide the action instruction of each defense device at the next moment based on the environmental state quantity;

[0039] The instruction parser is used to decompose the action instructions decided by the external agent into executable instructions or executable instruction sequences, and transmit them to the environment evolver;

[0040] The natural environment generator is used to generate a natural environment and transmit the natural environment parameters at the current moment to the environment evolver, and the generated natural environment is provided with multiple defense devices and multiple target objects for intelligent confrontation;

[0041] The environmental evolver is used to receive the natural environment parameters at the current moment, and collect the physical parameters of each defense device and each target object at the current moment, record them as the environmental state quantity at the current moment, and store them in the state pool;

[0042] The environment evolver obtains the environment state at the next moment according to the environment state at the current moment and the received executable instruction or executable instruction sequence, and stores it in the state pool;

[0043] The state pool is used to store and update the current environmental state;

[0044] The state feedback device is used to extract the environmental state quantity of each defense device at the next moment from the state pool and feed it back to the external intelligent agent.

[0045] Furthermore, an output terminal of the environment evolver is connected to an information filter, and the output terminal of the information filter is connected to the input terminal of the state feedback device;

[0046] The information filter is used to receive the visible information description of each defense device, and extract the visible state quantity of each defense device at the next moment from the state pool based on the visible information description, and transmit it to the state feedback device. The visible information description is the type of state information visible to each defense device itself;

[0047] The state feedback device is used to receive the visible state quantity of each defense device at the next moment and feed it back to the external intelligent agent. The external intelligent agent is used to receive the visible state quantity of each defense device at the next moment and decide the action instructions of each defense device at the next moment based on it.

[0048] Further, the environment evolver includes a first evolution unit and a second evolution unit;

[0049] The first evolution unit includes a motion module, an evolution module, a flight control module and a detection module; one output end of the state pool is connected to the input ends of the motion module, the evolution module and the flight control module respectively, and the other input end of the motion module is connected to the output end of the instruction parser; the output ends of the motion module, the evolution module and the flight control module are all connected to the input end of the detection module, and the two output ends of the detection module are respectively connected to the input ends of the state pool and the information filter;

[0050] The motion module is configured to receive the executable instruction, obtain the physical position parameters of each defense device at a next moment according to the executable instruction and the physical position parameters of each defense device at a current moment extracted from the state pool, and transmit the physical position parameters of each defense device at a next moment to the detection module;

[0051] The evolution module is used to obtain the natural environment parameters at the next moment based on the meteorological evolution model and the natural environment parameters at the current moment extracted from the state pool, and transmit the natural environment parameters at the next moment to the detection module;

[0052] The flight control module is configured to obtain a physical position parameter of each target at a next moment based on a target flight control algorithm and a flight state parameter of each target at a current moment extracted from the state pool, and transmit the physical position parameter of each target at a next moment to the detection module;

[0053] The detection module is used to receive the natural environment parameters at the current moment, and collect the physical parameters of each defense device and each target object at the current moment, record them as the environmental state quantity at the current moment, and store them in the state pool;

[0054] The detection module is used to receive the physical position parameters of each defense device at the next moment, the natural environment parameters at the next moment, and the physical position parameters of each target body at the next moment, record them as the environmental state quantity at the next moment, and store them in the state pool;

[0055] The detection module is used to obtain the visible information description of each defense device at the next moment based on the detection model of each defense device and the obtained environmental state quantity at the next moment, and transmit it to the information filter;

[0056] The second evolution unit includes a first execution module, a second execution module, and a damage module; the two output terminals of the state pool and the two output terminals of the instruction parser are connected to the input terminals of the first execution module and the second execution module, respectively; the output terminal of the first execution module is connected to the second execution module and the damage module in sequence; and the output terminal of the damage module is connected to the state pool;

[0057] The first execution module is configured to receive a first executable instruction in the executable instruction sequence, and obtain, based on the first executable instruction and the detection state parameters and tracking state parameters of each defense device in the current environmental state quantity extracted from the state pool, the detection state parameters and tracking state parameters of each defense device at a next moment, and transmit the obtained detection state parameters and tracking state parameters to the second execution module;

[0058] The second execution module is used to receive the detection state parameters and tracking state parameters of each defense device at the next moment, and determine whether each defense device is in a stable tracking state based on the received detection state parameters and tracking state parameters;

[0059] If the tracking state is stable, the second executable instruction is accessed, and the attack parameters of the defense device at the next moment are calculated based on the energy state parameters of the defense device and the flight state parameters and damage state parameters in the current environmental state quantity; if the tracking state is unstable, the tracking and aiming are continued;

[0060] The damage module is used to receive the strike parameters of each defense device at the next moment, calculate the damage state parameters of each target body at the next moment based on the strike parameters of each defense device at the next moment and the target body damage model, and store them in the state pool.

[0061] Compared with the prior art, the present invention has the following beneficial effects:

[0062] 1. The present invention provides a method for constructing a virtual environment for intelligent confrontation in real scenarios. This method is a reinforcement learning environment construction method based on real-world logic. Influenced by multiple real-world factors, it directly generates the action instructions of the defense system at the next moment based on the natural environment parameters of the real scenario and the physical parameters of the defense system and target group in the real scenario. This provides an effective learning environment for future intelligent defense confrontation deductions and has practical application value.

[0063] 2. The present invention provides a method for constructing an intelligent confrontation virtual environment for real scenarios. By updating the physical parameters and natural environment parameters of the defense system and the target group in real time, multiple defense devices in the defense system can always execute the action instructions at the next moment generated according to the physical parameters and natural environment parameters of the defense system and the target group at the current moment, and can reflect the objective physical scene.

[0064] 3. The present invention provides a system for constructing an intelligent confrontation virtual environment for real scenarios, which enables external intelligent agents to use the virtual environment constructed by the present invention as a reinforcement learning environment to learn defense confrontation strategies applicable to real scenarios.

[0065] 4. The present invention provides a system for constructing an intelligent confrontation virtual environment for real scenarios, which completes the construction of an intelligent confrontation virtual environment for real scenarios through a modular construction system. Due to the flexible expansion mechanism of the modular system, a richer virtual deduction environment construction can be achieved. BRIEF DESCRIPTION OF THE DRAWINGS

[0066] Figure 1 A flowchart of an embodiment of a method for constructing a real-scene intelligent confrontation virtual environment provided by the present invention;

[0067] Figure 2 A system diagram of an embodiment of a real-scene-oriented intelligent confrontation virtual environment construction system provided by the present invention;

[0068] Figure 3 This is a schematic diagram of the principles of an environment evolver in an embodiment of a real-scene-oriented intelligent confrontation virtual environment construction system provided by the present invention. DETAILED DESCRIPTION

[0069] The present invention will be further described below with reference to the accompanying drawings and specific embodiments.

[0070] Reference Figure 1 The present invention provides a method for constructing an intelligent confrontation virtual environment for real scenarios, comprising the following steps:

[0071] Step 1: Preset natural environment parameters at an initial moment and generate a desired natural environment based on the preset natural environment parameters; the natural environment is provided with multiple defense devices and multiple targets for intelligent confrontation, and physical parameters of the multiple defense devices and multiple targets at the initial moment are preset, the physical parameters including physical position parameters and physical state parameters;

[0072] The defense device in this embodiment adopts energy transmission type defense means, which include light energy transmission and electromagnetic energy transmission. The physical state parameters of the defense device include detection state parameters, tracking state parameters and energy state parameters. The physical state parameters of the target body include flight state parameters and damage state parameters. The flight state parameters in this embodiment include target body number parameters, speed parameters and flight path parameters. Multiple defense devices in this embodiment form a defense system, and multiple target bodies form a target group. The target group in this embodiment is a formation of drones.

[0073] In this embodiment, each defense device includes a detection model and a tracking model, which are used to detect and track the target body. The energy state parameter in this embodiment is quantitative, indicating the energy consumed by the defense system to attack the target body once; in other embodiments, each defense device includes a detection model, a tracking model and an energy model. The energy model is used to adjust the energy state parameter. In this case, the energy state parameter is a variable, and the defense device can select the attack intensity according to actual needs.

[0074] Step 2: record the natural environment parameters at the initial moment, as well as the physical parameters of each defense device and each target object at the initial moment, as the environmental state quantity at the initial moment;

[0075] The natural environment parameters in this embodiment include terrain parameters and atmospheric environment parameters. The terrain parameters are given in the form of a two-dimensional matrix, where the rows and columns of the matrix represent physical coordinates X and Y, respectively, and the matrix value at the corresponding coordinate is the height at that coordinate. The atmospheric environment parameters include parameters that change over time, such as wind speed, wind direction, temperature, humidity, and atmospheric coherence length.

[0076] Step 3: Determine the action instructions of each defense device at the initial moment according to the environmental state quantity at the initial moment;

[0077] The action instructions determined in this embodiment include a move instruction, a stop instruction, and a strike instruction. The move instruction carries a moving destination coordinate parameter, and the strike instruction carries a target number parameter.

[0078] Step 4: determine whether the action instruction is a complex instruction;

[0079] If yes, perform instruction parsing to obtain an executable instruction sequence and execute step 5;

[0080] If not, get the executable instruction and execute step 5;

[0081] A complex instruction is a combination of multiple simple instructions. After instruction parsing, it can be decomposed into a simple instruction sequence that can be directly executed by the defense device, that is, an executable instruction sequence. The strike instruction in this embodiment is a complex instruction. After instruction parsing, it can be decomposed into a tracking and aiming instruction and an attack instruction that are executed in sequence, that is, the tracking and aiming instruction is the first executable instruction, and the attack instruction is the second executable instruction. Among them, the move instruction, stop instruction, tracking and aiming instruction and attack instruction are all simple instructions, that is, executable instructions.

[0082] Step 5: Collect the current natural environment parameters and the current physical parameters of each defense device and each target object, record them as the current environmental state quantity, and store them;

[0083] Calculate the environmental state quantity at the next moment based on the environmental state quantity at the current moment and the obtained executable instruction or executable instruction sequence;

[0084] This embodiment obtains the environmental state quantity at the next moment by performing the following three calculations simultaneously:

[0085] If the instruction obtained in step 4 is an executable instruction, the natural environment parameters at the next moment are obtained according to the meteorological environment evolution model and the natural environment parameters in the environmental state quantity at the current moment;

[0086] Determine whether the executable instruction is a move instruction or a stop instruction: If it is a move instruction, the physical position parameters of each defense device at the next moment are calculated based on the coordinate parameters of the movement destination carried by the move instruction; if it is a stop instruction, the physical position parameters of each defense device at the current moment are directly used as the physical position parameters of the next moment;

[0087] According to the target flight control algorithm and the flight state parameters of each target in the current environmental state, the physical position parameters of each target at the next moment are obtained;

[0088] The meteorological evolution model in this embodiment includes a wind speed change model, a temperature change model, a transmittance change model, an atmospheric coherence length change model, etc., which are used for virtual meteorological evolution.

[0089] If an executable instruction sequence is obtained in step 4, the detection state parameters and tracking state parameters of each defense device at the next moment are calculated based on the first executable instruction in the executable instruction sequence, i.e., the tracking and aiming instruction in this embodiment, the detection model in each defense device, and the parameters of the environmental state at the current moment related to the first executable instruction, i.e., the detection state parameters and tracking and aiming state parameters of the defense device;

[0090] According to the detection state parameters and tracking state parameters of each defense device at the next moment, it is determined whether each defense device is in a stable tracking state;

[0091] If the defense device is in a stable tracking state, the second executable instruction, i.e., the attack action in this embodiment, is received. Based on the energy state parameters of the defense device and the parameters of the current environmental state related to the second executable instruction, i.e., the flight state parameters and damage state parameters of the target, the attack parameters of the defense device at the next moment are calculated.

[0092] If the defense device is in an unstable tracking state, it will continue to track and aim. As more information about the target group is obtained, the defense device will adjust its tracking and aiming model so that it can stably keep up with the target, that is, it will be in a stable tracking state.

[0093] Based on the obtained strike parameters of each defense device at the next moment and the target damage model, the damage state parameters of each target at the next moment are calculated;

[0094] Step 6: Update the stored current state of the environment to the next state of the environment;

[0095] Step 7, determining the action instructions of each defense device at the next moment based on the environmental state quantity at the next moment. In this embodiment, step 7 specifically includes the following steps:

[0096] Step 7.1: Based on the environmental state at the next moment and the detection model of each defense device, obtain the visible state of each defense device at the next moment;

[0097] Step 7.2: Based on the visible state quantity of each defense device at the next moment, decide the action instruction of each defense device at the next moment.

[0098] Of course, in other embodiments, step 7 may also directly determine the action instructions of each defense device at the next moment according to the environmental state quantity at the next moment, but the efficiency is relatively low.

[0099] Step 8: Loop through steps 4 to 7 until the confrontation ends, terminate the loop, and complete the construction of the intelligent confrontation virtual environment.

[0100] At the same time, refer to Figure 2 The present invention also provides a real-scene intelligent confrontation virtual environment construction system, including an instruction parser, a natural environment generator, an environment evolver, a state pool and a state feedback device, wherein the output end of the natural environment generator is connected to the input end of the state pool, and the output end of the instruction parser is connected to the input end of the environment evolver. In this embodiment, one output end of the environment evolver is connected to the state pool and the instruction parser in sequence, and the other output end of the environment evolver and the other output end of the state pool are both connected to the input end of the state feedback device. The output end of the state feedback device is used to connect to the input end of an external intelligent agent, and the other input end of the instruction parser is used to connect to the output end of the external intelligent agent. The external intelligent agent is used to receive the environmental state quantity of each defense device at the next moment, and decide the action instruction of each defense device at the next moment based on this; in other embodiments, one output end of the environment evolver is connected to an information filter, and the output end of the information filter is connected to the input end of the state feedback device.

[0101] An instruction parser, which is used to decompose the action instructions decided by the external agent into executable instructions or executable instruction sequences and transmit them to the environment evolver;

[0102] Natural environment generator: used to generate a natural environment and transmit the current natural environment parameters to the environment evolver. The generated natural environment is equipped with multiple defense devices and multiple targets for intelligent confrontation.

[0103] Reference Figure 3 , an environment evolver, comprising a first evolution unit and a second evolution unit;

[0104] The first evolution unit includes a motion module, an evolution module, a flight control module and a detection module;

[0105] a motion module, configured to receive an executable instruction, and obtain a physical position parameter of each defense device at a next moment according to the executable instruction and the physical position parameter of each defense device at a current moment extracted from the state pool, and transmit the physical position parameter of each defense device at a next moment to the detection module;

[0106] The evolution module is used to obtain the natural environment parameters at the next moment according to the meteorological evolution model and the natural environment parameters at the current moment extracted from the state pool, and transmit the natural environment parameters at the next moment to the detection module;

[0107] The flight control module is used to obtain the physical position parameters of each target at the next moment based on the target flight control algorithm and the physical position parameters of each target at the current moment extracted from the state pool, and transmit the physical position parameters of each target at the next moment to the detection module;

[0108] The detection module is used to receive the natural environment parameters at the current moment, as well as collect the physical parameters of each defense device and each target object at the current moment, record them as the environmental state quantity at the current moment, and store them in the state pool;

[0109] Receive the next moment's physical position parameters of each defense device, the next moment's natural environment parameters, and the next moment's physical position parameters of each target object, record them as the next moment's environmental state, and store them in the state pool;

[0110] As described above, in other embodiments, the detection module is further configured to obtain a visible information description of each defense device at the next moment based on the detection model of each defense device and the obtained environmental state quantity at the next moment, and transmit the description to an information filter. The visible information description here is a type of state information visible to each defense device itself; the information filter is configured to receive the visible information description of each defense device, extract the visible state quantity of each defense device at the next moment from the state pool, and transmit the description to the state feedback device.

[0111] The second evolution unit includes a first execution module, a second execution module and a damage module;

[0112] a first execution module, configured to receive a first executable instruction in the executable instruction sequence, and obtain, based on the first executable instruction and parameters of each defense device in the current environmental state quantity extracted from the state pool and related to the first executable instruction, namely, the detection state parameters and tracking state parameters of the defense device, at a next moment, and transmit the parameters to the second execution module;

[0113] The second execution module is used to receive the detection state parameters and tracking state parameters of each defense device at the next moment, and determine whether each defense device is in a stable tracking state based on the received detection state parameters and tracking state parameters;

[0114] If it is in a stable tracking state, the second executable instruction is connected, and the attack parameters of the defense device at the next moment are calculated based on the energy state parameters of the defense device and the parameters related to the second executable instruction in the current environmental state, namely the flight state parameters and damage state parameters of the target body; if it is in an unstable tracking state, the tracking and aiming are continued.

[0115] The damage module is used to receive the strike parameters of each defense device at the next moment, calculate the damage state parameters of each target body at the next moment based on the strike parameters of each defense device at the next moment and the target body damage model, and store them in the state pool.

[0116] The state pool is used to store and update the current state of the environment;

[0117] In this embodiment, the state feedback device is used to receive the environmental state quantity of each defense device at the next moment and feed it back to the external intelligent agent. The external intelligent agent uses a neural network for deep learning and intelligent decision-making, and is used to receive the environmental state quantity of each defense device at the next moment, and decide the action instructions of each defense device at the next moment accordingly; in other embodiments, the state feedback device is used to receive the visible state quantity of each defense device at the next moment, and feed it back to the external intelligent agent. The external intelligent agent is used to receive the visible state quantity of each defense device at the next moment, and decide the action instructions of each defense device at the next moment accordingly.

[0118] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit them. For ordinary professional and technical personnel in this field, the specific technical solutions recorded in the aforementioned embodiments can be modified, or some of the technical features therein can be replaced by equivalents. These modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions protected by the present invention.

Claims

1. A method for constructing an intelligent confrontation virtual environment for real scenarios, characterized by: The following steps are involved: Step 1: Preset the natural environment parameters at the initial moment and generate the required natural environment according to the set natural environment parameters; The natural environment is provided with a plurality of defense devices and a plurality of target objects for intelligent confrontation, and each defense device includes a detection model and a tracking model; Presetting physical parameters of multiple defense devices and multiple target objects at an initial moment, wherein the physical parameters include physical position parameters and physical state parameters; Step 2: Record the natural environment parameters at the initial moment, as well as the physical parameters of each of the defense devices and each of the target objects at the initial moment, as the environmental state quantities at the initial moment; Step 3, determining the action instructions of each defense device at the initial moment according to the obtained environmental state quantity at the initial moment; Step 4: determine whether the action instruction is a complex instruction; If so, perform instruction parsing to obtain an executable instruction sequence and execute step 5; If not, get the executable instruction and execute step 5; The determined action instructions include move instructions, stop instructions, and attack instructions. Move instructions carry a movement destination coordinate parameter, and attack instructions carry a target number parameter. The attack instructions are complex instructions, including tracking and aiming instructions and attack instructions. The tracking and aiming instructions are the first executable instructions of the attack instructions, and the attack instructions are the second executable instructions of the attack instructions. The tracking and aiming instructions, attack instructions, move instructions, and stop instructions are all executable instructions. Step 5: Collect the natural environment parameters at the current moment, as well as the physical parameters of each of the defense devices and each of the target objects at the current moment, record them as the environmental state quantity at the current moment, and store them; Calculate the environmental state quantity at the next moment based on the environmental state quantity at the current moment and the obtained executable instruction or executable instruction sequence; Step 6: Update the stored current state of the environment to the obtained state of the environment at the next moment; Step 7: Determine the action instructions of each defense device at the next moment based on the environmental state at the next moment; Step 8: Loop through steps 4 to 7 until the confrontation ends, terminate the loop, and complete the construction of the intelligent confrontation virtual environment.

2. The method for constructing a real-scene intelligent confrontation virtual environment according to claim 1, characterized in that: In step 1, the physical state parameters of the defense device include the detection state parameters, tracking state parameters and energy state parameters of the defense device; the physical state parameters of the target body include the flight state parameters and damage state parameters of the target body.

3. The method for constructing a real-scene intelligent confrontation virtual environment according to claim 2, characterized in that: In step 5, the environmental state quantity at the next moment is calculated based on the environmental state quantity at the current moment and the obtained executable instruction or executable instruction sequence, including the following steps: If the executable instruction is obtained in step 4, the natural environment parameters at the next moment are obtained according to the meteorological environment evolution model and the natural environment parameters in the environmental state quantity at the current moment; Determine whether the executable instruction is a move instruction or a stop instruction: If it is a move instruction, calculate the physical position parameters of each defense device at the next moment based on the coordinate parameters of the movement destination carried by the move instruction; if it is a stop instruction, directly use the physical position parameters of each defense device at the current moment as the physical position parameters of the next moment; According to the flight control algorithm of the target body and the flight state parameters of each target body in the environmental state quantity at the current moment, the physical position parameters of each target body at the next moment are obtained; If an executable instruction sequence is obtained in step 4, the detection state parameters, tracking state parameters, and strike parameters of each defense device at the next moment are calculated based on the detection model of each defense device, the environmental state quantity at the current moment, and the obtained executable instruction sequence; the damage state parameters of each target body at the next moment are calculated based on the obtained strike parameters of each defense device at the next moment and the target body damage model.

4. The method for constructing a real-scene intelligent confrontation virtual environment according to claim 3, characterized in that: In step 5, the detection state parameters, tracking state parameters, and strike parameters of each defense device at the next moment are obtained, including the following steps: Calculate the detection state parameters and tracking state parameters of each defense device at the next moment based on the detection model in each defense device, the first executable instruction in the obtained executable instruction sequence, and the detection state parameters and tracking state parameters in the environmental state quantity at the current moment; Determining whether each defense device is in a stable tracking state according to the detection state parameters and tracking state parameters of each defense device at the next moment; If the tracking state is stable, the second executable instruction is accessed, and the attack parameters of the defense device at the next moment are calculated based on the energy state parameters of the defense device and the flight state parameters and damage state parameters in the current environmental state. If the tracking state is not stable, continue to track and aim.

5. The method for constructing a real-scene intelligent confrontation virtual environment according to claim 4, characterized in that: In step 7, the action instructions of each defense device at the next moment are determined according to the environmental state at the next moment, specifically: Obtaining the visible state of each defense device at the next moment based on the environmental state at the next moment and the detection model of each defense device; and determining the action instruction of each defense device at the next moment based on the visible state of each defense device at the next moment; Alternatively, the action instructions of each defense device at the next moment are directly determined based on the environmental state quantity at the next moment.

6. The method for constructing a real-scene intelligent confrontation virtual environment according to claim 5, characterized in that: In step 1, the defense device adopts energy transmission type defense means, which includes light energy transmission and electromagnetic energy transmission; The target object is a drone, and multiple target objects form a formation of drones.

7. A system for constructing a real-world scenario-oriented intelligent confrontation virtual environment, for implementing the method for constructing a real-world scenario-oriented intelligent confrontation virtual environment according to any one of claims 1 to 6, characterized in that: It includes an instruction parser, a natural environment generator, an environment evolver, a state pool and a state feedback device; the output end of the natural environment generator is connected to the input end of the state pool, the output end of the instruction parser is connected to the input end of the environment evolver, one output end of the environment evolver is connected to the state pool and the instruction parser in sequence, the other output end of the environment evolver and the other output end of the state pool are both connected to the input end of the state feedback device, the output end of the state feedback device is used to connect to the input end of an external intelligent agent, the other input end of the instruction parser is used to connect to the output end of an external intelligent agent, and the external intelligent agent is used to receive the environmental state quantity of each defense device at the next moment and decide the action instruction of each defense device at the next moment based on the environmental state quantity; The instruction parser is used to decompose the action instructions decided by the external agent into executable instructions or executable instruction sequences, and transmit them to the environment evolver; The natural environment generator is used to generate a natural environment and transmit the natural environment parameters at the current moment to the environment evolver, and the generated natural environment is provided with multiple defense devices and multiple target objects for intelligent confrontation; The environmental evolver is used to receive the natural environment parameters at the current moment, and collect the physical parameters of each defense device and each target object at the current moment, record them as the environmental state quantity at the current moment, and store them in the state pool; The environment evolver obtains the environment state at the next moment according to the environment state at the current moment and the received executable instruction or executable instruction sequence, and stores it in the state pool; The state pool is used to store and update the current environmental state; The state feedback device is used to extract the environmental state quantity of each defense device at the next moment from the state pool and feed it back to the external intelligent agent.

8. The system for constructing a real-world intelligent confrontation virtual environment according to claim 7, characterized in that: An output terminal of the environment evolver is connected to an information filter, and an output terminal of the information filter is connected to an input terminal of the state feedback device; The information filter is used to receive the visible information description of each defense device, and extract the visible state quantity of each defense device at the next moment from the state pool based on the visible information description, and transmit it to the state feedback device. The visible information description is the type of state information visible to each defense device itself; The state feedback device is used to receive the visible state quantity of each defense device at the next moment and feed it back to the external intelligent agent. The external intelligent agent is used to receive the visible state quantity of each defense device at the next moment and decide the action instructions of each defense device at the next moment based on it.

9. The system for constructing a real-world intelligent confrontation virtual environment according to claim 8, characterized in that: The environment evolver includes a first evolution unit and a second evolution unit; The first evolution unit includes a motion module, an evolution module, a flight control module and a detection module; one output end of the state pool is connected to the input ends of the motion module, the evolution module and the flight control module respectively, and the other input end of the motion module is connected to the output end of the instruction parser; the output ends of the motion module, the evolution module and the flight control module are all connected to the input end of the detection module, and the two output ends of the detection module are respectively connected to the input ends of the state pool and the information filter; The motion module is configured to receive the executable instruction, obtain the physical position parameters of each defense device at a next moment according to the executable instruction and the physical position parameters of each defense device at a current moment extracted from the state pool, and transmit the physical position parameters of each defense device at a next moment to the detection module; The evolution module is used to obtain the natural environment parameters at the next moment based on the meteorological evolution model and the natural environment parameters at the current moment extracted from the state pool, and transmit the natural environment parameters at the next moment to the detection module; The flight control module is configured to obtain a physical position parameter of each target at a next moment based on a target flight control algorithm and a flight state parameter of each target at a current moment extracted from the state pool, and transmit the physical position parameter of each target at a next moment to the detection module; The detection module is used to receive the natural environment parameters at the current moment, and collect the physical parameters of each defense device and each target object at the current moment, record them as the environmental state quantity at the current moment, and store them in the state pool; The detection module is used to receive the physical position parameters of each defense device at the next moment, the natural environment parameters at the next moment, and the physical position parameters of each target body at the next moment, record them as the environmental state quantity at the next moment, and store them in the state pool; The detection module is used to obtain the visible information description of each defense device at the next moment based on the detection model of each defense device and the obtained environmental state quantity at the next moment, and transmit it to the information filter; The second evolution unit includes a first execution module, a second execution module, and a damage module; the two output terminals of the state pool and the two output terminals of the instruction parser are connected to the input terminals of the first execution module and the second execution module, respectively; the output terminal of the first execution module is connected to the second execution module and the damage module in sequence; and the output terminal of the damage module is connected to the state pool; The first execution module is configured to receive a first executable instruction in the executable instruction sequence, and obtain, based on the first executable instruction and the detection state parameters and tracking state parameters of each defense device in the current environmental state quantity extracted from the state pool, the detection state parameters and tracking state parameters of each defense device at a next moment, and transmit the obtained detection state parameters and tracking state parameters to the second execution module; The second execution module is used to receive the detection state parameters and tracking state parameters of each defense device at the next moment, and determine whether each defense device is in a stable tracking state based on the received detection state parameters and tracking state parameters; If the tracking state is stable, the second executable instruction is accessed, and the attack parameters of the defense device at the next moment are calculated based on the energy state parameters of the defense device and the flight state parameters and damage state parameters in the current environmental state. If the tracking state is not stable, continue to track and aim; The damage module is used to receive the strike parameters of each defense device at the next moment, calculate the damage state parameters of each target body at the next moment based on the strike parameters of each defense device at the next moment and the target body damage model, and store them in the state pool.

Images