Multi-agent game confrontation situation prediction method, system and equipment

Abstract

The invention discloses a multi-agent game confrontation situation prediction method, system and device, and relates to the field of artificial intelligence, and the method comprises the steps: obtaining multi-agent historical game data; the multi-agent historical game data comprises action elements of each agent in each historical time step in the game process; determining a feature vector sequence according to the multi-agent historical game data; according to the feature vector sequence, performing situation prediction on each agent by adopting a situation prediction model to obtain a situation category of each agent in a future time step; wherein the situation prediction model is obtained by training a training sample set in advance; the situation prediction model comprises a Transform model, a bidirectional long-short-term memory network, a feature fusion network and a classifier. According to the invention, the robustness and generalization ability of the situation prediction model are improved, and the accuracy and real-time performance of complex game confrontation situation prediction are improved.

Description

Technical Field

[0001] This application relates to the field of artificial intelligence, and particularly to a multi-agent game confrontation situation prediction method, system and device. Background Art

[0002] Multi-agent game confrontation situation prediction is an important research direction in the field of artificial intelligence, and is widely used in military confrontation, autonomous driving, intelligent transportation and other fields. Traditional situation prediction methods are mainly based on rule reasoning or shallow machine learning models, which are difficult to capture the complex interaction relationships and long-term dependence features between agents, thus making it difficult to meet the requirements for the timeliness and accuracy of situation prediction results. Summary of the Invention

[0003] The purpose of this application is to provide a multi-agent game confrontation situation prediction method, system and device, which can improve the accuracy and timeliness of complex game confrontation situation prediction.

[0004] To achieve the above purpose, this application provides the following solutions:

[0005] In a first aspect, this application provides a multi-agent game confrontation situation prediction method, including:

[0006] Obtain multi-agent historical game data; the multi-agent historical game data includes the action elements of each agent at each historical time step during the game process;

[0007] Determine a feature vector sequence according to the multi-agent historical game data; the feature vector sequence includes a plurality of feature vectors, and each feature vector represents the action elements of each agent at a historical time step;

[0008] Perform situation prediction on each agent according to the feature vector sequence by using a situation prediction model to obtain the situation category of each agent in a future time step;

[0009] Among them, the situation prediction model is pre-trained by using a training sample set, and each training sample in the training sample set includes a feature vector sequence sample and the situation category label of each agent in a corresponding future time step; the situation prediction model includes a Transformer model, a bidirectional long short-term memory network, a feature fusion network and a classifier.

[0010] Optionally, the action elements of an agent include: the state of the agent, the action of the agent, and environmental information.

[0011] Optionally, the situation category of an agent includes: the state of the agent and the action of the agent.

[0012] Optionally, the state of the agent includes position, speed, and direction; the actions of the agent include moving, attacking, and defending.

[0013] Optionally, according to the multi-agent historical game data, a feature vector sequence is determined, specifically including:

[0014] Perform data cleaning and normalization processing on the multi-agent historical game data in sequence to obtain normalized data;

[0015] Convert the normalized data into a feature vector sequence.

[0016] Optionally, according to the feature vector sequence, a situation prediction model is used to perform situation prediction on each agent to obtain the situation category of each agent within future time steps, specifically including:

[0017] Capture the global dependence relationship between feature vectors in the feature vector sequence through a Transformer model to obtain a first feature representation sequence; the first feature representation sequence includes the first feature representation corresponding to each agent at each historical time step;

[0018] Capture the local time features between feature vectors in the feature vector sequence through a bidirectional long short-term memory network to obtain a second feature representation sequence; the second feature representation sequence includes the second feature representation corresponding to each agent at each historical time step;

[0019] Fuse the first feature representation sequence and the second feature representation sequence through a feature fusion network to obtain a fused feature representation sequence;

[0020] Classify the fused feature representation sequence through a classifier to obtain the situation category of each agent within future time steps.

[0021] Optionally, the feature fusion network is a fully connected layer.

[0022] Optionally, the classifier includes a fully connected layer and a softmax function connected in sequence.

[0023] In a second aspect, the present application provides a multi-agent game confrontation situation prediction system, including:

[0024] A data acquisition module for acquiring multi-agent historical game data; the multi-agent historical game data includes the action elements of each agent at each historical time step during the game process;

[0025] A vector determination module for determining a feature vector sequence according to the multi-agent historical game data; the feature vector sequence includes a plurality of feature vectors, and each feature vector represents the action elements of each agent at one historical time step;

[0026] A situation prediction module, configured to perform situation prediction on each agent according to the feature vector sequence by using a situation prediction model, so as to obtain the situation category of each agent within a future time step;

[0027] Wherein, the situation prediction model is pre-trained by using a training sample set, and each training sample in the training sample set includes a feature vector sequence sample and the situation category label of each agent within a corresponding future time step; the situation prediction model includes a Transformer model, a bidirectional long short-term memory network, a feature fusion network, and a classifier.

[0028] In a third aspect, the present application provides a computer device, including: a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor executes the computer program to implement the above-mentioned multi-agent game confrontation situation prediction method.

[0029] According to the specific embodiments provided by the present application, the present application has the following technical effects:

[0030] The present application provides a multi-agent game confrontation situation prediction method, system and device, which improves the robustness and generalization ability of the situation prediction model and improves the accuracy and real-time performance of complex game confrontation situation prediction by integrating the global modeling ability of the Transformer model and the local time feature extraction ability of the bidirectional long short-term memory network (Long Short-Term Memory, LSTM). BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In order to more clearly illustrate the technical solutions in the embodiments of the present application or the prior art, the following will briefly introduce the drawings required for use in the embodiments. Obviously, the drawings in the following description are only some embodiments of the present application. For those of ordinary skill in the art, without creative efforts, other drawings can also be obtained based on these drawings.

[0032] Figure 1 It is an application environment diagram of a multi-agent game confrontation situation prediction method in an embodiment of the present application;

[0033] Figure 2 It is a flowchart of a multi-agent game confrontation situation prediction method provided in an embodiment of the present application;

[0034] Figure 3 It is a structural diagram of a situation prediction model in an embodiment of the present application;

[0035] Figure 4 It is a comparison diagram of experimental results of different models in an embodiment of the present application;

[0036] Figure 5 Schematic diagram of functional modules of a multi-agent game confrontation situation prediction system provided in an embodiment of the present application;

[0037] Figure 6 Schematic diagram of the structure of a computer device provided in an embodiment of the present application. Detailed implementation manners

[0038] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0039] To make the above objects, features, and advantages of the present application more obvious and understandable, the present application will be further described in detail below in conjunction with the accompanying drawings and specific implementation manners.

[0040] The multi-agent game confrontation situation prediction method provided in the embodiments of the present application can be applied to an application environment as Figure 1 shown. Among them, the terminal 102 communicates with the server 104 through a network. The data storage system can store the data that the server 104 needs to process. The data storage system can be set separately, integrated on the server 104, or placed on the cloud or other servers. The terminal 102 can send multi-agent historical game data to the server 104. After receiving the multi-agent historical game data, the server 104 determines a feature vector sequence according to the multi-agent historical game data, and uses a situation prediction model to perform situation prediction on each agent according to the feature vector sequence to obtain the situation category of each agent within the future time step. The server 104 can feedback the obtained situation category of each agent within the future time step to the terminal 102. In addition, in some embodiments, the multi-agent game confrontation situation prediction method can also be implemented independently by the server 104 or the terminal 102.

[0041] Among them, the terminal 102 can be, but is not limited to, various desktop computers, laptop computers, smartphones, tablet computers, Internet of Things devices, and portable wearable devices. The Internet of Things devices can be smart speakers, smart TVs, smart air conditioners, smart in-vehicle devices, etc. The portable wearable devices can be smart watches, smart bracelets, head-mounted devices, etc. The server 104 can be implemented by an independent server or a server cluster composed of multiple servers, and can also be a cloud server.

[0042] In an exemplary embodiment, as Figure 2As shown, a method for predicting the confrontation situation of multi-agent games is provided. This method is executed by a computer device, which can be specifically executed by a computer device such as a terminal or a server alone, or jointly executed by a terminal and a server. In the embodiments of the present application, taking the application of this method to Figure 1 server 104 in

[0043] Step 201, obtain multi-agent historical game data.

[0044] The multi-agent historical game data includes the action elements of each agent at each historical time step during the game process. Among them, the action elements of an agent include: the state of the agent (such as position, speed, and direction), the actions of the agent (such as moving, attacking, and defending), and environmental information (such as terrain and obstacles).

[0045] Step 202, determine a sequence of feature vectors according to the multi-agent historical game data.

[0046] In a specific application example, first perform data cleaning and normalization processing on the multi-agent historical game data in sequence to obtain normalized data. Then convert the normalized data into a sequence of feature vectors. Among them, data cleaning includes removing noise and outliers.

[0047] The sequence of feature vectors includes multiple feature vectors, and each feature vector represents the action elements of each agent at a historical time step. That is, the dimension of the sequence of feature vectors is [T, D]; where T is the number of historical time steps, and D is the feature dimension, that is, the number of action elements.

[0048] Step 203, according to the sequence of feature vectors, use a situation prediction model to perform situation prediction on each agent to obtain the situation category of each agent in future time steps. The situation category of an agent includes the state of the agent and the actions of the agent. In specific applications, the situation category to be predicted can be determined according to actual needs, that is, the situation category is the position when predicting the position of the agent, and the situation category is the action when predicting the action of the agent.

[0049] In an exemplary embodiment, as Figure 3 shown, the situation prediction model includes a Transformer model, a bidirectional long short-term memory network, a feature fusion network, and a classifier. Step 203 includes the following steps 31 to 34.

[0050] Step 31, capture the global dependence relationship between the feature vectors in the sequence of feature vectors through the Transformer model to obtain a first sequence of feature representations.

[0051] Specifically, the sequence of feature vectors is input into a Transformer encoder. The global dependencies between the feature vectors are captured through the multi-head self-attention mechanism, and the output of the multi-head self-attention mechanism is further processed by a feed-forward neural network to obtain the encoded feature representation, that is, the first sequence of feature representations.

[0052] The first sequence of feature representations includes the first feature representation corresponding to each agent within each historical time step. The dimension of the first sequence of feature representations is [T, D]. The first sequence of feature representations retains the global dependencies.

[0053] Step 32: Capture the local temporal features between the feature vectors in the sequence of feature vectors through a bidirectional long short-term memory network to obtain a second sequence of feature representations.

[0054] Specifically, the sequence of feature vectors is input into a bidirectional LSTM. The forward and backward dependencies of the sequence of feature vectors are captured through the forward LSTM unit and the backward LSTM unit respectively. The outputs of the forward LSTM unit and the backward LSTM unit are concatenated to obtain the feature representation encoded by the bidirectional LSTM, that is, the second sequence of feature representations.

[0055] The second sequence of feature representations includes the second feature representation corresponding to each agent within each historical time step. The dimension of the second sequence of feature representations is [T, 2D]. The second sequence of feature representations represents the local temporal features.

[0056] Step 33: Fuse the first sequence of feature representations and the second sequence of feature representations through a feature fusion network to obtain a fused sequence of feature representations.

[0057] In a specific application example, the feature fusion network is a fully connected layer. Specifically, the first sequence of feature representations and the second sequence of feature representations are concatenated, and the fully connected layer is used to fuse the concatenated feature representations to obtain a fused sequence of feature representations. The dimension of the fused sequence of feature representations is [T, 3D].

[0058] Step 34: Classify the fused sequence of feature representations through a classifier to obtain the situation categories of each agent in the future time step.

[0059] In a specific application example, the classifier includes a fully connected layer and a softmax function connected in sequence. Specifically, the fused sequence of feature representations is input into the fully connected layer, and the situation prediction results of the multi-agent game confrontation are output through the softmax function, including the situation category and the confidence level.

[0060] The situation prediction model is pre-trained using a training sample set. Each training sample in the training sample set includes a feature vector sequence sample and the situation category label of each agent within the future time steps.

[0061] During the training process of the situation prediction model, the training sample set is divided into a training set, a validation set, and a test set. The training set is used to train the situation prediction model, and the parameters of the situation prediction model are optimized through the backpropagation algorithm. The validation set is used to validate the situation prediction model, and the hyperparameters are adjusted to prevent overfitting. The test set is used to evaluate the trained situation prediction model, and evaluation metrics such as the accuracy, recall rate, and F1 score of the situation prediction model are calculated to measure the performance of the situation prediction model.

[0062] This application improves the robustness and generalization ability of the situation prediction model and enhances the accuracy and real-time performance of complex game confrontation situation prediction by integrating the global modeling ability of the Transformer model and the local time feature extraction ability of the bidirectional LSTM.

[0063] This application also provides an application scenario that applies the above multi-agent game confrontation situation prediction method. Specifically: The multi-agent game confrontation situation prediction method provided in this embodiment can be applied in a simulated combat scenario. The simulated combat scenario includes the red side and the blue side, and the red side and the blue side include agents such as humans, unmanned aerial vehicles, and unmanned vehicles. In the simulated combat scenario, by obtaining the state, actions, and environmental information of each agent and using the above multi-agent game confrontation situation prediction method to predict the state and / or actions of each agent at future moments, the game process between the red and blue sides is simulated, thereby providing guidance for combat strategies in actual combat.

[0064] This application also provides a simulation experiment to verify the effectiveness of the above multi-agent game confrontation situation prediction method.

[0065] Using the confrontation review data of the urban combat game challenge track of the military wargame competition, the original data game scenario is a point seizure and control battle scenario for a synthetic battalion. Our combat objective is to pass through a certain urban factory area from the starting area and reach the target area with the fastest speed and greater damage to complete the block breakthrough and point seizure and control tasks; the enemy is deployed in various buildings and some areas of the urban factory area, and the combat objective is to block our passage through the factory area and seize and control the target area. The constructed map size is 30*30, which is divided into 900 grid areas in total.

[0066] 80 high-quality games were selected for the experiment. Since the similarity degree of the feature situation within a short period of time in the same game is relatively high, in this application, 48 games were selected as the training set, 16 games as the validation set, and 16 games as the test set for 5-fold cross-validation in terms of the game. Since the multi-head self-attention mechanism can automatically discover and combine relevant features with predictive value, no artificial setting of cross features was performed in the experiment.

[0067] The operator features used mainly include state, reward, whether the current task is completed, discount factor, action sequence, etc. In the experiment, the probability that the position information value of the agent falls within the top 1, 3, and 6 squares with the highest predicted values of the situation prediction model was used as the accuracy rate for measurement.

[0068] To test the prediction performance of the situation prediction model provided by this application, as Figure 4 shown, a comparative experiment was conducted with traditional classification algorithms such as logistic regression and support vector machine as the control, mainly comparing the accuracy rate index. The main configuration of the computer used in the experiment is Intel(R) Core(TM) i7-7500U CPU@2.70GHz 2.90GHz, the graphics card is NVIDIA GeForce 940MX, and the memory is 16GB.

[0069] The situation prediction model provided by this application has significant advantages in the generation of multi-agent game confrontation strategies: First, it has a strong feature extraction ability. The Transformer model can effectively capture the long-range dependence relationship and complex non-linear features in the data, while the bidirectional LSTM can effectively capture the dynamic change information of time series data. The combination of the two can extract more comprehensive and more discriminative features; Second, it has efficient parallel computing. The parallel computing ability of the Transformer model enables the training speed of the situation prediction model to be faster and can process larger-scale data sets. Third, it has good generalization ability. Through appropriate regularization techniques and data augmentation methods, the generalization ability of the situation prediction model can be improved, enabling it to better adapt to different multi-agent game confrontation strategy generation environments.

[0070] Based on the same inventive concept, the embodiment of this application also provides a multi-agent game confrontation situation prediction system for implementing the above-mentioned multi-agent game confrontation situation prediction method. The implementation solutions provided by this system to solve problems are similar to the implementation solutions recorded in the above method. Therefore, the specific limitations in one or more embodiments of the multi-agent game confrontation situation prediction system provided below can refer to the limitations on the multi-agent game confrontation situation prediction method in the above text and will not be elaborated here.

[0071] In an exemplary embodiment, as Figure 5As shown, a multi-agent game confrontation situation prediction system is provided, including: a data acquisition module 501, a vector determination module 502, and a situation prediction module 503.

[0072] The data acquisition module 501 is used to acquire multi-agent historical game data. The multi-agent historical game data includes the action elements of each agent at each historical time step during the game.

[0073] The vector determination module 502 is used to determine a feature vector sequence according to the multi-agent historical game data. The feature vector sequence includes a plurality of feature vectors, and each feature vector represents the action elements of each agent at a historical time step.

[0074] The situation prediction module 503 is used to perform situation prediction on each agent according to the feature vector sequence, and obtain the situation category of each agent in the future time step by using a situation prediction model.

[0075] Among them, the situation prediction model is pre-trained by using a training sample set. Each training sample in the training sample set includes a feature vector sequence sample and the situation category label of each agent in the corresponding future time step. The situation prediction model includes a Transformer model, a bidirectional long short-term memory network, a feature fusion network, and a classifier.

[0076] In an exemplary embodiment, a computer device is provided. The computer device can be a server or a terminal, and its internal structure diagram can be as Figure 6 shown. The computer device includes a processor, a memory, an input / output interface (Input / Output, I / O), and a communication interface. Among them, the processor, the memory, and the input / output interface are connected through a system bus, and the communication interface is connected to the system bus through the input / output interface. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store multi-agent historical game data. The input / output interface of the computer device is used to exchange information between the processor and external devices. The communication interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, a multi-agent game confrontation situation prediction method is implemented.

[0077] Those skilled in the art can understand, Figure 6The structure shown is only a block diagram of some structures related to the solution of this application, and does not constitute a limitation on the computer device to which the solution of this application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine some components, or have different component arrangements.

[0078] In an exemplary embodiment, a computer-readable storage medium is provided, storing a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0079] In an exemplary embodiment, a computer program product is provided, including a computer program, and when the computer program is executed by a processor, the steps in the above method embodiments are implemented.

[0080] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data for analysis, stored data, displayed data, etc.) involved in this application are all information and data that have been authorized by the user or fully authorized by all parties, and the collection, use, and processing of relevant data need to comply with relevant regulations.

[0081] In this application, all actions of obtaining signals, information, or data are carried out on the premise of complying with the corresponding data protection regulations and policies of the country where the location is located and obtaining the authorization given by the owner of the corresponding device.

[0082] Those of ordinary skill in the art can understand that all or part of the processes in the methods of the above embodiments can be completed by instructing relevant hardware through a computer program. The computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above methods. Among them, any reference to a memory, database, or other medium used in the various embodiments provided in this application can include at least one of non-volatile and volatile memories. Non-volatile memories can include read-only memory (ROM), magnetic tapes, floppy disks, flash memories, optical memories, high-density embedded non-volatile memories, resistive random-access memories (ReRAMs), magnetoresistive random-access memories (MRAMs), ferroelectric random-access memories (FRAMs), phase change memories (PCMs), graphene memories, etc. Volatile memories can include random-access memories (RAMs) or external caches, etc. By way of illustration and not limitation, RAM can be in various forms, such as static random-access memory (SRAM) or dynamic random-access memory (DRAM), etc.

[0083] The databases involved in the various embodiments provided in this application can include at least one of relational databases and non-relational databases. Non-relational databases can include distributed databases based on blockchain, etc., without limitation. The processors involved in the various embodiments provided in this application can be general-purpose processors, central processors, graphics processors, digital signal processors, programmable logics, data processing logics based on quantum computing, etc., without limitation.

[0084] The technical features of the above embodiments can be combined arbitrarily. For the sake of concise description, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, it should be considered as the scope described in this specification.

[0085] Specific examples are used in this article to elaborate on the principles and implementation manners of this application. The descriptions of the above embodiments are only used to help understand the method and its core idea of this application; at the same time, for those of ordinary skill in the art, according to the idea of this application, there will be changes in the specific implementation manners and application scopes. In summary, the content of this specification should not be construed as a limitation to this application.

Claims

1. A method for predicting a multi-agent game confrontation situation, characterized in that: The multi-agent game confrontation situation prediction method comprises: Acquire multi-agent historical game data; the multi-agent historical game data includes action elements of each agent in each historical time step during the game process; Determine a feature vector sequence according to the multi-agent historical game data; the feature vector sequence includes a plurality of feature vectors, each feature vector representing an action element of each agent in a historical time step; According to the feature vector sequence, a situation prediction model is used to predict the situation of each intelligent agent to obtain the situation category of each intelligent agent in the future time step; Among them, the situation prediction model is pre-trained using a training sample set, each training sample in the training sample set includes a feature vector sequence sample and a corresponding situation category label of each agent in the future time step; the situation prediction model includes a Transformer model, a bidirectional long short-term memory network, a feature fusion network and a classifier.

2. The method for predicting a multi-agent game confrontation situation according to claim 1, characterized in that: The action elements of an intelligent agent include: the state of the intelligent agent, the action of the intelligent agent and the environmental information.

3. The method for predicting a multi-agent game confrontation situation according to claim 1, characterized in that: The situation categories of the agent include: the state of the agent and the action of the agent.

4. The method for predicting multi-agent game confrontation situation according to claim 2 or 3, characterized in that: The state of the agent includes position, speed and direction; the actions of the agent include moving, attacking and defending.

5. The method for predicting a multi-agent game confrontation situation according to claim 1, characterized in that: Determine a feature vector sequence according to the multi-agent historical game data, specifically including: Performing data cleaning and normalization processing on the multi-agent historical game data in turn to obtain normalized data; The normalized data is converted into a feature vector sequence.

6. The method for predicting a multi-agent game confrontation situation according to claim 1, characterized in that: According to the feature vector sequence, the situation prediction model is used to predict the situation of each agent, and the situation category of each agent in the future time step is obtained, which specifically includes: Capturing the global dependency between feature vectors in the feature vector sequence through a Transformer model to obtain a first feature representation sequence; the first feature representation sequence includes a first feature representation corresponding to each agent in each historical time step; Capturing local time features between feature vectors in the feature vector sequence through a bidirectional long short-term memory network to obtain a second feature representation sequence; the second feature representation sequence includes a second feature representation corresponding to each agent in each historical time step; The first feature representation sequence and the second feature representation sequence are fused through a feature fusion network to obtain a fused feature representation sequence; The fused feature representation sequence is classified by a classifier to obtain the situation category of each agent in the future time step.

7. The method for predicting a multi-agent game confrontation situation according to claim 1, characterized in that: The feature fusion network is a fully connected layer.

8. The method for predicting a multi-agent game confrontation situation according to claim 1, characterized in that: The classifier includes fully connected layers and a softmax function connected in sequence.

9. A multi-agent game confrontation situation prediction system, applied to the multi-agent game confrontation situation prediction method according to any one of claims 1 to 8, characterized in that: The multi-agent game confrontation situation prediction system includes: A data acquisition module is used to acquire multi-agent historical game data; the multi-agent historical game data includes the action elements of each agent in each historical time step during the game process; A vector determination module, used to determine a feature vector sequence according to the multi-agent historical game data; the feature vector sequence includes a plurality of feature vectors, each feature vector representing an action element of each agent in a historical time step; A situation prediction module is used to predict the situation of each intelligent agent using a situation prediction model according to the feature vector sequence to obtain the situation category of each intelligent agent in the future time step; Among them, the situation prediction model is pre-trained using a training sample set, each training sample in the training sample set includes a feature vector sequence sample and a corresponding situation category label of each agent in the future time step; the situation prediction model includes a Transformer model, a bidirectional long short-term memory network, a feature fusion network and a classifier.

10. A computer device comprising: A memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the multi-agent game confrontation situation prediction method described in any one of claims 1 to 8.

Images