Three-dimensional digital intelligent simulation platform oriented to self-game technology

Abstract

A three-dimensional digital intelligent simulation platform oriented to a self-game technology combines a three-dimensional digital simulation platform with intelligent agent training, realizes a closed-loop self-game training environment, allows intelligent agents to make decisions independently in a simulated three-dimensional confrontation environment, and improves the training efficiency. According to the method, the intelligent agent is used as a strategy, the strategy confrontation ability and skills are improved through real-time interaction with a simulation platform, meanwhile, by means of a FastDDS communication mechanism, the intelligent agent can receive an initial situation, send a decision instruction and conduct learning and strategy optimization according to feedback, the autonomy and adaptability of the intelligent agent are improved, and the richness and dynamism of a simulation environment are enhanced.

Description

Technical Field

[0001] The invention relates to a three-dimensional digital intelligent simulation platform oriented to self-game technology, belonging to the field of three-dimensional digital simulation. Background Art

[0002] In military conflicts, the opposing parties are always in a state of strategic competition, and their decisions will continue to evolve as the confrontation develops. When formulating strategies, both sides will consider the possible choices of the other side and incorporate these factors into their own plans. In order to maintain their advantage, they must constantly adopt iterative methods to optimize their strategies. With the advancement of military technology, predicting the actions of the enemy has become more challenging. Traditional confrontation simulation systems have limitations in simulating this interactive game process between the two sides. Therefore, it is necessary to introduce artificial intelligence self-game technology and design a three-dimensional digital intelligent simulation platform for self-game technology so that it can more accurately simulate the strategic interaction between the enemy and us, and provide more advanced simulation support for confrontation decision-making.

[0003] Traditional confrontation simulation systems for self-game technology have many technical bottlenecks. First, the rules are too simplified. In order to facilitate simulation and calculation, traditional systems will simplify the confrontation rules, resulting in a large deviation between the simulation results and the actual situation; second, the level of intelligence is limited. Traditional systems rely on preset scripts and limited algorithms, lack sufficient adaptive and learning capabilities, and cannot effectively simulate complex and changing battlefield environments; third, there is a lack of real-time feedback. In the process of self-game, intelligent agents need real-time feedback to optimize their decisions, while traditional systems lack real-time and dynamic feedback mechanisms; fourth, there is a lack of adaptive learning capabilities. Traditional systems do not have the ability to self-learn and adaptively adjust, and cannot respond to new situations in the confrontation process. Summary of the invention

[0004] The technical problem solved by the present invention is: in view of the problems faced by the red and blue sides in the decision-making process in actual confrontation exercises in the current existing technology, such as low efficiency, over-simplified rules, and limited intelligence level, a three-dimensional digital intelligent simulation platform for self-game technology is proposed.

[0005] The present invention solves the above technical problems by the following technical solutions:

[0006] A three-dimensional digital intelligent simulation platform for self-game technology, including a simulation control module, a data receiving module, and a simulation driving module;

[0007] The simulation control module sends the preset initial situation information to the data transceiver module in the three-dimensional environment, receives the decision instruction returned by the data transceiver module, analyzes the decision instruction, generates the next situation information according to the analysis result, and sends it to the data transceiver module again;

[0008] The data transceiver module sends the received initial situation information to the intelligent agent target, receives the decision instruction sent back by the intelligent agent target and forwards it to the simulation control module, and at the same time sends the next situation information generated by the simulation control module to the intelligent agent target again;

[0009] The simulation driving module initializes the scene parameters of the three-dimensional environment, builds the scene tree and generates the three-dimensional environment before the intelligent target interacts with it; at the same time, it collects real-time situation data for data update processing and three-dimensional scene refresh.

[0010] In the simulation control module, when the intelligent agent target receives situation information twice in succession, it is regarded as a confrontation cycle; the simulation control module is preset with the number of confrontation cycles. When each confrontation cycle ends, the simulation control module makes a judgment. If the current number of confrontation cycles is not greater than the preset number of confrontation cycles, the confrontation cycle is repeated until the number of cycles reaches the preset number of confrontation cycles.

[0011] In the three-dimensional environment, the number of intelligent agent targets is determined according to the requirements of the confrontation exercise task, and each intelligent agent independently interacts with the intelligent simulation platform for data and instructions, and the interaction process is implemented through FastDDS.

[0012] The simulation driving module uses the OSG engine to realize the generation of the three-dimensional environment. The generation method of the three-dimensional environment is:

[0013] Preset the scene parameters of the three-dimensional environment according to the requirements of the confrontation exercise task;

[0014] Initialize the scene according to the scene parameters of the three-dimensional environment;

[0015] Construct a scene tree to reflect the spatial structure of the three-dimensional environment and the target state of the agent;

[0016] Perform OSG visual simulation, detect communication threads and user operation flags, and update the observation matrix of nodes in the scene tree in real time according to the main viewpoint position and status;

[0017] After refreshing the screen, the OSG visual simulation ends, and the three-dimensional environment is constructed based on the current scene tree and the current frame.

[0018] The method of initializing the scene using the scene parameters of the three-dimensional environment is:

[0019] Create a field of view;

[0020] Set the main camera associated with the field of view;

[0021] Set the variables of the three-dimensional environment, initialize the variables, and perform initial scene settings according to the parameters of the scene initialization startup command.

[0022] The method to construct the scene tree is:

[0023] Extract preload information according to the preset model library, use the preload information to preload the model library and special effect animation sequence respectively, and read the experience data of the scene tree in the global scene database to initialize the scene tree;

[0024] Through OSG simulation, the tree data structure of the read and preloaded scene tree is used from top to bottom to organize the spatial data set to improve rendering efficiency, and the various node information of the scene tree is determined and grouped to complete the scene tree construction;

[0025] Perform data update processing and scene tree update based on the collected real-time situation data.

[0026] The various nodes of the scene tree include: root node, first-level child node, and second-level child node;

[0027] The first-level sub-nodes are connected to the root node, and the second-level sub-nodes are connected to the first-level sub-nodes. The first-level sub-nodes include earth scene nodes, entity nodes, path nodes, information nodes, and HUD nodes; the second-level sub-nodes are grouped according to their functions.

[0028] The specific method of the OSG visual simulation is:

[0029] Start the communication thread, collect real-time situation data, set the new data signal flag, and determine whether to process new data based on the new data signal flag. If new data arrives, perform 3D situation information processing and information update, and perform user operation flag detection, otherwise perform user operation flag detection directly;

[0030] User operation flag detection: if a user operation is detected, the screen is updated after execution, otherwise the screen is updated directly;

[0031] According to the position and state of the main viewpoint of the current frame, the observation matrix of the first-level child node in the main scene tree is calculated in real time, and OSG is called to refresh the current frame to complete the OSG visual simulation;

[0032] The OSG visual simulation has a preset number of loops, and completing one OSG visual simulation is considered as one simulation cycle, and the simulation stops after reaching the preset number of loops.

[0033] The judgment flag of the new data arrival is: if the signal flag is 1, it means that new data has arrived; if the signal flag is 0, it means that there is no new data;

[0034] If the detected user operation flag is 1, it means that the user has performed a keyboard or mouse operation, and corresponding processing is performed according to the operation type. If the detected user operation flag is 0, there is no user input.

[0035] The intelligent simulation platform also includes a perception interface, a decision interface, and an arbitration module;

[0036] The perception interface is used to convert the required information into data form to realize the interaction between the intelligent simulation platform and the intelligent agent target; the decision interface is used to transmit the situation information and decision instruction information designed by the intelligent agent target; the arbitration mechanism is used to evaluate the decision instruction information output by the intelligent agent target and provide feedback, and the intelligent agent target optimizes the decision instruction information in real time based on the feedback content.

[0037] The advantages of the present invention compared with the prior art are:

[0038] (1) The present invention provides a three-dimensional digital intelligent simulation platform for self-game technology, which combines the three-dimensional digital simulation platform with intelligent agent training to realize a closed-loop self-game training environment. This environment allows the intelligent agent to make independent decisions in a simulated three-dimensional confrontation environment and improve its strategic confrontation capabilities and skills through real-time interaction with the simulation platform. Through the FastDDS communication mechanism, the intelligent agent can receive the initial situation, send decision-making instructions, and learn and optimize strategies based on feedback. This training method improves the autonomy and adaptability of the intelligent agent;

[0039] (2) The present invention uses open source 3D graphics rendering engines OSG and osgEarth to build a highly realistic confrontation environment, and organizes spatial data sets through a top-down, hierarchical tree data structure to improve rendering efficiency. The construction of the scene tree not only reflects the spatial structure, but also displays the object status, making the simulation of the 3D environment more efficient and realistic. In addition, by preloading the model database and special effects, the richness and dynamics of the simulation environment are enhanced;

[0040] (3) The present invention designs a set of interfaces and mechanisms to enable the traditional simulation platform to seamlessly connect with the intelligent agent to achieve intelligent decision-making and feedback loops. This includes perception interfaces, decision interfaces, arbitration mechanisms, and data transmission mechanisms. In particular, the introduction of the arbitration mechanism can evaluate the decisions of the intelligent agent based on the algorithm and provide timely feedback to help the intelligent agent optimize its decision. In terms of data transmission, by unifying the data format and using FastDDS communication, stable data exchange between the simulation platform developed in C++ and the intelligent agent developed in Python is ensured. DETAILED DESCRIPTION

[0041] A three-dimensional digital intelligent simulation platform for self-game technology combines the three-dimensional digital simulation platform with intelligent agent training to realize a closed-loop self-game training environment, allowing the intelligent agent to make independent decisions in a simulated three-dimensional confrontation environment and improve its strategic confrontation capabilities and skills through real-time interaction with the simulation platform. At the same time, using the FastDDS communication mechanism, the intelligent agent can receive the initial situation, send decision-making instructions, and learn and optimize strategies based on feedback, which improves the autonomy and adaptability of the intelligent agent and enhances the richness and dynamism of the simulation environment.

[0042] A three-dimensional digital intelligent simulation platform for self-game technology, including a simulation control module, a data receiving module, and a simulation driving module;

[0043] The simulation control module sends the preset initial situation information to the data transceiver module in the three-dimensional environment, receives the decision instruction returned by the data transceiver module, analyzes the decision instruction, generates the next situation information according to the analysis result, and sends it to the data transceiver module again;

[0044] The data transceiver module sends the received initial situation information to the intelligent agent target, receives the decision instruction sent back by the intelligent agent target and forwards it to the simulation control module, and at the same time sends the next situation information generated by the simulation control module to the intelligent agent target again;

[0045] The simulation driving module initializes the scene parameters of the three-dimensional environment, builds the scene tree and generates the three-dimensional environment before the intelligent target interacts with it; at the same time, it collects real-time situation data for data update processing and three-dimensional scene refresh.

[0046] In the simulation control module, when the intelligent agent target receives situation information twice in succession, it is regarded as a confrontation cycle; the simulation control module presets the number of confrontation cycles. When each confrontation cycle ends, the simulation control module makes a judgment. If the current number of confrontation cycles is not greater than the preset number of confrontation cycles, the confrontation cycle is repeated until the number of cycles reaches the preset number of confrontation cycles.

[0047] In a three-dimensional environment, the number of intelligent agent targets is determined according to the requirements of the confrontation exercise task. Each intelligent agent independently interacts with the intelligent simulation platform for data and instructions, and the interaction process is implemented through FastDDS.

[0048] The simulation driver module uses the OSG engine to realize the generation of the three-dimensional environment. The generation method of the three-dimensional environment is:

[0049] Preset the scene parameters of the three-dimensional environment according to the requirements of the confrontation exercise task;

[0050] Initialize the scene according to the scene parameters of the three-dimensional environment;

[0051] Construct a scene tree to reflect the spatial structure of the three-dimensional environment and the target state of the agent;

[0052] Perform OSG visual simulation, detect communication threads and user operation flags, and update the observation matrix of nodes in the scene tree in real time according to the main viewpoint position and status;

[0053] After refreshing the screen, the OSG visual simulation ends, and the three-dimensional environment is constructed based on the current scene tree and the current frame.

[0054] The method of initializing the scene using the scene parameters of the three-dimensional environment is:

[0055] Create a field of view;

[0056] Set the main camera associated with the field of view;

[0057] Set the variables of the three-dimensional environment, initialize the variables, and perform initial scene settings according to the parameters of the scene initialization startup command.

[0058] The method to construct the scene tree is:

[0059] Extract preload information according to the preset model library, use the preload information to preload the model library and special effect animation sequence respectively, and read the experience data of the scene tree in the global scene database to initialize the scene tree;

[0060] Through OSG simulation, the tree data structure of the read and preloaded scene tree is used from top to bottom to organize the spatial data set to improve rendering efficiency, and the various node information of the scene tree is determined and grouped to complete the scene tree construction;

[0061] Perform data update processing and scene tree update based on the collected real-time situation data.

[0062] The various nodes of the scene tree include: root node, first-level child node, and second-level child node;

[0063] The first-level sub-nodes are connected to the root node, and the second-level sub-nodes are connected to the first-level sub-nodes. The first-level sub-nodes include earth scene nodes, entity nodes, path nodes, information nodes, and HUD nodes; the second-level sub-nodes are grouped according to their functions.

[0064] The specific method of OSG visual simulation is:

[0065] Start the communication thread, collect real-time situation data, set the new data signal flag, and determine whether to process new data based on the new data signal flag. If new data arrives, perform 3D situation information processing and information update, and perform user operation flag detection, otherwise perform user operation flag detection directly;

[0066] User operation flag detection: if a user operation is detected, the screen is updated after execution, otherwise the screen is updated directly;

[0067] According to the position and state of the main viewpoint of the current frame, the observation matrix of the first-level child node in the main scene tree is calculated in real time, and OSG is called to refresh the current frame to complete the OSG visual simulation;

[0068] The OSG visual simulation has a preset number of loops, and completing one OSG visual simulation is considered as one simulation cycle, and the simulation stops after reaching the preset number of loops.

[0069] The judgment flag of the arrival of new data is: if the signal flag is 1, it means that new data has arrived; if the signal flag is 0, it means that there is no new data;

[0070] If the detected user operation flag is 1, it means that the user has performed a keyboard or mouse operation, and corresponding processing is performed according to the operation type. If the detected user operation flag is 0, there is no user input.

[0071] The intelligent simulation platform also includes perception interface, decision interface, and arbitration module;

[0072] The perception interface is used to convert the required information into data form to realize the interaction between the intelligent simulation platform and the intelligent agent target; the decision interface is used to transmit the situation information and decision instruction information designed by the intelligent agent target; the arbitration mechanism is used to evaluate the decision instruction information output by the intelligent agent target and provide feedback, and the intelligent agent target optimizes the decision instruction information in real time based on the feedback content.

[0073] The following is further described in conjunction with the preferred embodiments:

[0074] In the current embodiment, a three-dimensional digital intelligent simulation platform for self-game technology is designed. First, the three-dimensional digital simulation platform is combined with intelligent agent training to realize a closed-loop self-game training environment. This environment allows the intelligent agent to make independent decisions in a simulated three-dimensional confrontation environment, and improve its strategic confrontation capabilities and skills through real-time interaction with the simulation platform. Through the FastDDS communication mechanism, the intelligent agent can receive the initial situation, send decision instructions, and learn and optimize strategies based on feedback. This training method improves the autonomy and adaptability of the intelligent agent;

[0075] At the same time, the open source 3D graphics rendering engines OSG and osgEarth are used to build a highly realistic confrontation environment, and the spatial data set is organized through a top-down, hierarchical tree data structure to improve rendering efficiency. The construction of the scene tree not only reflects the spatial structure, but also displays the object status, making the simulation of the 3D environment more efficient and realistic. In addition, by preloading the model database and special effects, the richness and dynamics of the simulation environment are enhanced;

[0076] Finally, a set of interfaces and mechanisms were designed to enable the traditional simulation platform to seamlessly connect with the intelligent agent and realize intelligent decision-making and feedback loops. This includes perception interfaces, decision interfaces, arbitration mechanisms, and data transmission mechanisms. In particular, the introduction of the arbitration mechanism can evaluate the decisions of the intelligent agent based on the algorithm and provide timely feedback to help the intelligent agent optimize its decision. In terms of data transmission, by unifying the data format and using FastDDS communication, stable data exchange between the simulation platform developed in C++ and the intelligent agent developed in Python is ensured.

[0077] A three-dimensional simulation environment is used to simulate the situation in actual confrontation exercises; integrate artificial intelligence technology to create an intelligent agent that can make autonomous decisions; realize the interaction between the intelligent agent and the three-dimensional simulation environment, and improve the decision-making ability of the intelligent agent through self-game training;

[0078] The agent self-game training technology includes the following steps: using the FastDDS communication protocol to realize data exchange between the agent and the simulation platform; the agent makes decision instructions according to the initial situation provided by the simulation platform; the simulation platform generates and returns the next situation according to the decision instructions of the agent; the agent learns and optimizes the strategy according to the received situation and feedback;

[0079] The 3D digital situation simulation technology includes: using the open source 3D graphics rendering engine OSG and the 3D digital earth engine osgEarth as the development framework; realizing the 3D graphics function through the data receiving module and the OSG simulation driver module;

[0080] Initialization of system and scene parameters, including creation of field of view, setting of main camera and initialization of global variables; building scene tree and generating 3D environment, including preloading of model library, preloading of special effect animation sequence, reading of global scene database and initialization of scene tree; entering OSG scene simulation loop, including starting of communication thread, detection of user operation, real-time calculation of observation matrix of first-level child node osg::Camera in main scene tree and OSG refreshing current frame; judging whether simulation is finished, returning to loop if not finished, and exiting the system if finished;

[0081] The design of the three-dimensional digital intelligent simulation platform with self-game technology includes: a perception interface, which is used by the intelligent agent to perceive situation information; a decision interface, which is used by the intelligent agent to make decisions based on the perception information; an arbitration mechanism, which is used to evaluate the decision of the intelligent agent and provide feedback; a data transmission mechanism, which is used for data transmission between the simulation platform and the intelligent agent, using a unified data format.

[0082] (1) Agent Self-Game

[0083] The three-dimensional digital intelligent simulation platform for self-game technology integrates the conventional three-dimensional simulation environment and intelligent agents. Through the interaction between the two, strategic confrontation and skill improvement between intelligent agents are realized. In the scenario involving two intelligent agents, each intelligent agent interacts with the platform independently. The platform communicates and interacts with the intelligent agent through FastDDS. The process of communication between the platform and the intelligent agent is as follows: the platform sends the initial situation to the intelligent agent; the intelligent agent returns the platform decision instruction; after receiving the decision instruction, the platform parses and processes the decision instruction, generates the next situation according to the instruction, and sends the next situation to the intelligent agent; the above process runs in a loop until a confrontation ends; the platform determines whether the number of loops has reached the maximum value. If not, it continues to the next round. The following is a pseudo code example describing this game interaction process:

[0084] for iteration in range(ITERATIONS): / / Perform the specified number of iterations

[0085] gain=0 / / Reward value is initialized to zero

[0086] scenario = simulation.reset() / / The simulation environment is set to the initial scenario

[0087] while True: / / enter the training loop

[0088] move = entity.decide_move(scenario) / / The agent makes a decision based on the current situation

[0089] subsequent_scenario,gain,finished=simulation.proceed(move) / / The simulation platform returns the subsequent situation, reward and loop end flag according to the decision instruction

[0090] entity.update(scenario,move,gain,subsequent_scenario,finished) / / The agent is updated based on the feedback from the platform

[0091] scenario=subsequent_scenario / / The situation is updated to the situation after the deduction

[0092] if finished: / / If the deduction is completed

[0093] break / / Exit the loop and prepare for the next iteration

[0094] (2) Three-dimensional digital situation simulation

[0095] The intelligent simulation platform uses the open source 3D graphics rendering engine OSG and the 3D digital earth engine osgEarth as the background development framework, which is responsible for the realization of 3D graphics functions. The platform consists of a data transceiver module and an OSG simulation driver module.

[0096] The implementation process of OSG environment simulation is as follows:

[0097] 1) Initialization of system and scene parameters. This step includes:

[0098] Creation of field of view;

[0099] Main camera settings and associated field of view;

[0100] Initialize global variables, parse startup command parameters, and set scenes based on the parameters.

[0101] 2) Build a scene tree and generate a 3D environment. The main tasks include:

[0102] Preloading of model library;

[0103] Preloading of special effects animation sequences;

[0104] Reading of global scene database;

[0105] ·Initialize the scene tree and build a basic but complete scene tree. The scene tree not only reflects the spatial structure, but also displays the object status. OSG uses a top-down, hierarchical tree data structure to organize spatial data sets, effectively improving rendering efficiency.

[0106] The main nodes of the scene tree in this article include: ①Root node: osg::ref-ptrosg::Groupg-Root; ②First-level child nodes: connected to the root node g-Root, mainly osg::ref-ptrosg::Camera type, including earth scene node, entity node, path node, information node, HUD node; ③Second-level child nodes: connected to the first-level child nodes, mainly osg::ref-ptrosg::Switch type, grouped by function.

[0107] 3) Enter the OSG visual simulation loop. The work in the loop body mainly includes:

[0108] ·Start the communication thread, receive real-time situation data, and set the new data signal flag. If the signal flag is 1, it means new data has arrived, first perform 3D situation processing and information update, and then proceed to the next step; if it is 0, it means there is no new data, and directly proceed to the next step;

[0109] · Detect user operation flag. If the flag is 1, it means the user has performed keyboard or mouse operation, and the corresponding processing is performed according to the operation type, and then the next step is entered; if it is 0, it means there is no user input, and the next step is entered directly;

[0110] ·According to the main viewpoint position and state of the current frame, the observation matrix of the first-level child node osg::Camera in the main scene tree is calculated in real time;

[0111] Call OSG to refresh the current frame;

[0112] Determine whether the simulation has ended. If not, return to the loop; if it has ended, proceed to the next step.

[0113] 4) After the simulation is finished, exit the system;

[0114] (3) Three-dimensional digital intelligent simulation platform for self-game technology

[0115] Traditional simulation platforms usually consist of simulation architecture, simulation models, and simulation evaluations, and can efficiently perform simulation tasks. In terms of decision-making simulation, strategy confrontation simulation, as an important simulation method, plays a vital role in improving the strategic level of commanders and assisting decision-making. Despite this, when conducting simulations, simulators still have to face complex decision-making problems, including how to choose the best course of action based on uncertain factors and predict the possible response of the opponent. In order to build a three-dimensional digital intelligent simulation platform that can meet these challenges, a series of interfaces need to be designed to realize the interaction between traditional simulation platforms and intelligent agents:

[0116] 1) Perception interface

[0117] In traditional simulation platforms, information is directly transmitted to people through screens. For intelligent agents, this information needs to be converted into data so that they can understand and use it. In the simulation platform of self-game technology, intelligent agents need to receive and analyze the three-dimensional spatial information of the environment to gain situational awareness.

[0118] 2) Decision interface

[0119] After the intelligent agent perceives the situation, it will analyze and make decisions through the neural network algorithm. These decisions need to be input in a format compatible with the traditional simulation platform to ensure that the output of the intelligent agent can be accepted and executed by the simulation platform.

[0120] 3) Arbitration institution

[0121] In traditional simulation platforms, the results are usually evaluated by users. In the simulation platform of self-game technology, an arbitration agency is introduced to automatically evaluate the decision of the intelligent agent and provide feedback based on traditional algorithms to guide the intelligent agent to optimize its decision.

[0122] 4) Data Transmission

[0123] Since the traditional simulation platform and the intelligent agent may use different development environments and programming languages, such as C++ and Python, a unified data transmission format is needed to ensure data exchange between the two. The FastDDS communication protocol is used to ensure stable data transmission between the C++ and Python ends.

[0124] Although the present invention has been disclosed as above in the form of a preferred embodiment, it is not intended to limit the present invention. Any person skilled in the art may make possible changes and modifications to the technical solution of the present invention by using the methods and technical contents disclosed above without departing from the spirit and scope of the present invention. Therefore, 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 content of the technical solution of the present invention shall fall within the protection scope of the technical solution of the present invention.

[0125] The contents not described in detail in the specification of the present invention belong to the common knowledge of those skilled in the art.

Claims

1. A three-dimensional digital intelligent simulation platform for self-game technology, characterized by: It includes a simulation control module, a data receiving module and a simulation driving module; The simulation control module sends the preset initial situation information to the data transceiver module in the three-dimensional environment, receives the decision instruction returned by the data transceiver module, analyzes the decision instruction, generates the next situation information according to the analysis result, and sends it to the data transceiver module again; The data transceiver module sends the received initial situation information to the intelligent agent target, receives the decision instruction sent back by the intelligent agent target and forwards it to the simulation control module, and at the same time sends the next situation information generated by the simulation control module to the intelligent agent target again; The simulation driving module initializes the scene parameters of the three-dimensional environment, builds the scene tree and generates the three-dimensional environment before the intelligent target interacts with it; at the same time, it collects real-time situation data for data update processing and three-dimensional scene refresh.

2. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 1, characterized in that: In the simulation control module, when the intelligent agent target receives situation information twice in succession, it is regarded as a confrontation cycle; the simulation control module is preset with the number of confrontation cycles. When each confrontation cycle ends, the simulation control module makes a judgment. If the current number of confrontation cycles is not greater than the preset number of confrontation cycles, the confrontation cycle is repeated until the number of cycles reaches the preset number of confrontation cycles.

3. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 1, characterized in that: In the three-dimensional environment, the number of intelligent agent targets is determined according to the requirements of the confrontation exercise task, and each intelligent agent independently interacts with the intelligent simulation platform for data and instructions, and the interaction process is implemented through FastDDS.

4. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 1, characterized in that: The simulation driving module uses the OSG engine to realize the generation of the three-dimensional environment. The generation method of the three-dimensional environment is: Preset the scene parameters of the three-dimensional environment according to the requirements of the confrontation exercise task; Initialize the scene according to the scene parameters of the three-dimensional environment; Construct a scene tree to reflect the spatial structure of the three-dimensional environment and the target state of the agent; Perform OSG visual simulation, detect communication threads and user operation flags, and update the observation matrix of nodes in the scene tree in real time according to the main viewpoint position and status; After refreshing the screen, the OSG visual simulation ends, and the three-dimensional environment is constructed based on the current scene tree and the current frame.

5. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 4, characterized in that: The method of initializing the scene using the scene parameters of the three-dimensional environment is: Create a field of view; Set the main camera associated with the field of view; Set the variables of the three-dimensional environment, initialize the variables, and perform initial scene settings according to the parameters of the scene initialization startup command.

6. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 4, characterized in that: The method to construct the scene tree is: Extract preload information according to the preset model library, use the preload information to preload the model library and special effect animation sequence respectively, and read the experience data of the scene tree in the global scene database to initialize the scene tree; Through OSG simulation, the tree data structure of the read and preloaded scene tree is used from top to bottom to organize the spatial data set to improve rendering efficiency, and the various node information of the scene tree is determined and grouped to complete the scene tree construction; Perform data update processing and scene tree update based on the collected real-time situation data.

7. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 6, characterized in that: The various nodes of the scene tree include: root node, first-level child node, and second-level child node; The first-level sub-nodes are connected to the root node, and the second-level sub-nodes are connected to the first-level sub-nodes. The first-level sub-nodes include earth scene nodes, entity nodes, path nodes, information nodes, and HUD nodes; the second-level sub-nodes are grouped according to their functions.

8. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 6, characterized in that: The specific method of the OSG visual simulation is: Start the communication thread, collect real-time situation data, set the new data signal flag, and determine whether to process new data based on the new data signal flag. If new data arrives, perform 3D situation information processing and information update, and perform user operation flag detection, otherwise perform user operation flag detection directly; User operation flag detection: if a user operation is detected, the screen is updated after execution, otherwise the screen is updated directly; According to the position and state of the main viewpoint of the current frame, the observation matrix of the first-level child node in the main scene tree is calculated in real time, and OSG is called to refresh the current frame to complete the OSG visual simulation; The OSG visual simulation has a preset number of loops, and completing one OSG visual simulation is considered as one simulation cycle, and the simulation stops after reaching the preset number of loops.

9. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 8, characterized in that: The judgment flag of the new data arrival is: if the signal flag is 1, it means that new data has arrived; if the signal flag is 0, it means that there is no new data; If the detected user operation flag is 1, it means that the user has performed a keyboard or mouse operation, and corresponding processing is performed according to the operation type. If the detected user operation flag is 0, there is no user input.

10. A three-dimensional digital intelligent simulation platform for self-game technology according to claim 1, characterized in that: The intelligent simulation platform also includes a perception interface, a decision interface, and an arbitration module; The perception interface is used to convert the required information into data form to realize the interaction between the intelligent simulation platform and the intelligent agent target; The decision interface is used to transmit the situation information and decision instruction information of the intelligent agent target design; The arbitration mechanism is used to evaluate the decision instruction information output by the intelligent agent target and provide feedback. The intelligent agent target optimizes the decision instruction information in real time based on the feedback content.