Empowering Wargaming and Future Battlefields with AI

Image: An American Army esports player next to a gaming machine at Fort Knox, Kentucky. Unlike traditional computer simulations and wargaming, which largely rely on pre-set scenarios, AI-embedded simulations can autonomously generate small narratives and create numerous action plans for a given scenario.

In the movie “Avengers: Infinity War” (2018), superheroes fight to save the universe from the villain Thanos. As they seek ways to defeat him, Doctor Strange mentions his time travel: “I went through time and space. I looked at another future. I saw all the outcomes of the upcoming conflict.” He glimpsed 14,000,605 futures and discovered the only path to victory.

Doctor Strange’s ability to foresee the future is no longer just fiction. AI-embedded modeling simulations can perform similar tasks: while they cannot predict the future, they can forecast it. Through AI-generated battlefield simulations, the Navy can engage opposing forces over a million times, thereby identifying the keys to winning battles. With a vast database of various scenarios, the Navy can respond to any attempts to replace the traditional rules-based maritime order.

Computer Simulation/Wargaming

Traditional computer simulations and wargaming have a long history within the U.S. Department of Defense (DoD). In the 1980s, the DoD adopted a conflict simulation model called Janus. Janus was a planning tool for the 1989 “Operation Just Cause” (invasion of Panama) and the following year’s “Operation Desert Storm.” After witnessing its impact, the DoD expanded the application of computer simulations, which now include Lockheed Martin’s “WarSim,” the Army’s “One Semi-Automated Forces” (OneSAF), the Marine Corps’ “MAGTF Tactical Simulation” (MTWS), and the Air Force’s “Advanced Framework for Simulation, Integration, and Modeling” (AFSIM).

However, modern simulations also have limitations. They largely depend on pre-set scenarios and require substantial human heuristic decision-making and involvement, which restricts the attempts to simulate numerous small narratives and scenarios. Advances in AI and significant increases in computational power provide opportunities to alleviate these limitations.

What distinguishes AI-embedded simulations from traditional computer simulations is their ability to simulate millions of battles in a short time. Through millions of self-rehearsals, they can autonomously generate small episodes, create multiple action plans for given scenarios, and provide decision-makers with various options. They can also assess or generate the best actions for hostile forces and devise countermeasures to defeat them.

Image: The U.S. Department of Defense adopted the interactive conflict simulation model Janus in the 1980s and used it as a planning tool for “Operation Just Cause” and “Operation Desert Storm.” After witnessing its impact, the DoD expanded the use of computer simulations. Archives of Lawrence Livermore National Laboratory

Advances in Simulation and AI and Their Impact on the Navy

AI Non-Player Characters. Over the past few decades, the resolution of game graphics has significantly improved, allowing users to become fully immersed in games within seconds. Non-player characters (NPCs) are another component that makes games enjoyable, as they are “characters controlled by the computer rather than the player”. Even when players are playing alone, these virtual characters make them feel like they are battling against real opponents.

However, in some games, NPCs do not behave like humans. They perform the same actions in the same situations over and over again. These repetitive actions make it easy for players to predict what will happen next, and they may quickly find the game boring.

There are several ways to make NPCs behave more like humans. In the past, developers used simple rule-based behavior algorithms. However, with advances in neural network technology, NPCs have become more dynamic and adaptable to opponents’ behaviors. In 2005, three computer scientists from the University of Texas at Austin demonstrated that NPCs embedded with neural networks could train in real-time while users played the game. This required players to compete against more human-like intelligent opponents. In NPC development, DeepMind’s collaboration with Atari has achieved the most remarkable results. By combining deep neural networks with reinforcement learning (a field of machine learning), NPCs surpassed human performance after only 2,600 iterations of self-play. If NPCs embedded with neural networks are applied to military training, they could help train individuals to complete complex tasks.

Integrating NPCs into military applications opens an innovative pathway to enhance combat training and strategy across various domains of the Navy—space, air, surface, underwater, and cyber. For example, by designing NPCs to execute specific tasks such as island defense, anti-surface warfare, and underwater operations on detailed maps of varying difficulty (expert, normal, novice, etc.), vessels can prepare for a range of critical scenarios and determine the most effective strategies to achieve objectives by virtually battling NPCs set to appropriate skill levels.

Research work on developing these specialized NPCs is ongoing, with significant efforts made by several military-related institutions. The Naval Postgraduate School is researching cognitive AI for NPCs that need to handle military characteristics such as hierarchy, fog of war, or specific scenarios within the ATLATL platform. At the USC Institute for Creative Technologies, researchers are rapidly integrating and developing adaptive NPCs specifically for military training purposes. The research from these institutions has potential applications beyond their initial scope, including strategic decision processes, such as determining the best course of action. Additionally, it could pave the way for the development of unmanned vehicles capable of autonomous thought and decision-making, which could change the game on future battlefields.

Generative AI for Simulation/Wargaming. Generative AI (GenAI) excels at creating new content similar to what it has learned. This capability has been widely applied across industries and has become part of our daily lives, such as ChatGPT.

The main advantage of GenAI is its ability to address the issue of limited combat experience or training data. Given basic parameters such as the estimated size of enemy naval forces, types of ships, and numbers, GenAI can generate a vast number of realistic scenarios. This enables operational planners to explore and experiment with various possibilities that may exceed human thinking but seem plausible.

GenAI tools like COA-GPT, developed by Vinicius Goecks and Nicholas Waytowich, can also suggest action plans by interacting with command and control personnel, providing decision support (see page 64, Figure 1). The Navy can use this method to generate action plans for scenarios ranging from simple tasks (such as finding the most effective formations or locations in open waters) to complex situations (such as combat scenarios near coastal areas and islands).

Another aspect of GenAI implementation is scenario generation. Based on historical experiences and known enemy tactics, objectives, and missions, GenAI can generate synthesized enemy behaviors based on created narratives. This capability will enable the Navy to prepare for unexpected situations by simulating various potential enemy actions, thereby enhancing its ability to respond to dynamic maritime threats.

Digital Twin. A digital twin translates reality into digital form. The pioneer in the industrial sector is Industry 4.0, which integrates technologies such as cloud computing, the Internet of Things, AI, and digital twins to collect and analyze data generated during manufacturing processes to enhance decision-making. Imagine a smart factory where each machine on the production line has its own sensors, continuously collecting data and sharing it with the entire system. Regardless of the type of data, AI analyzes it and suggests ways to make the production process smoother and more efficient.

The digital twin system was launched by NASA in 2010 and is “a multi-physical, multi-scale, probabilistic integrated simulation of a vehicle or system, utilizing existing best physical models, sensor updates, fleet history, etc., to reflect its flight twin system’s lifespan.” The advantage of the digital twin system is that it can visualize the entire system during the design phase, predict problems, optimize solutions, accelerate prototype design, and facilitate training before actual implementation. Core components such as simulation and AI, machine learning (which enables systems to learn from experience without explicit programming), and reinforcement learning (which enables systems to achieve the most valuable outcomes) are crucial for facilitating visibility into future outcomes.

The digital twin allows decision-makers to see the outcomes of their decisions and adjust their choices. They can predict ideal outcomes, avoid predictable negative outcomes, and mitigate the impacts of unpredictable adverse behaviors. This dual-layer approach is not limited to predicting and managing the lifecycle of ship equipment; it can also significantly enhance strategic planning and real-time decision-making under various operational conditions. By using integrated data analysis and simulation, the digital twin promotes a deep understanding of maintenance systems, enabling fleets to optimize vessel performance, increase reliability, and operate with greater confidence. Furthermore, the digital twin plays a critical role in risk management, providing a safe environment to test hypotheses and evaluate potential interventions without actually altering the real system. Hence, the digital twin is not only advantageous but also crucial for maintaining competitiveness and achieving excellence in today’s rapidly changing maritime combat environment.

Platforms

Creating virtual battlefields requires significant investment of resources and effort. However, there are also promising platforms available for military use. The gaming industry offers a plethora of military-themed games, some of which feature realistic data inputs or allow users to modify inputs to meet specific requirements. While these gaming platforms may not currently have the conditions for AI and machine learning (ML) applications, they have the potential to serve as the foundation for creating AI/ML environments. By leveraging advanced simulation capabilities such as the physics engines or terrain generators of these platforms, it is possible to develop complex training and strategic planning tools for military applications without starting from scratch. This approach not only saves resources but also accelerates the development and deployment of advanced virtual battlefield technologies. Examples of such gaming and simulation platforms include

“Command: Modern Operations” (published by Slitherine Ltd): This game simulates modern warfare across multiple domains, providing detailed scales of land, sea, and air military operations. Its strength lies in its complex scenario editor, which allows users to create specific combat scenarios. By integrating machine learning algorithms, the military can enhance the predictive capabilities of these scenarios, improving decision-making and strategic planning in virtual exercises.

“Modern Naval Warfare” (published by Slitherine Ltd): This platform focuses on high-fidelity simulations of naval combat, including submarine warfare, surface ship engagements, and air defense. By adapting it for machine learning use, the Navy can develop algorithms for simulating and analyzing naval strategies, providing unprecedented training opportunities and insights into naval combat tactics and strategy optimization.

Modern Air Combat Environment (developed by BSI): MACE is a highly detailed air combat scenario simulation tool that provides realistic aircraft, missile systems, and radar tracking models. It can simulate complex aerial engagements, making it highly suitable for machine learning; algorithms can analyze engagement scenarios, providing tactical and strategic insights that could potentially revolutionize air combat training and planning.

VR Forces (developed by MAK Technologies): VR Forces can create detailed virtual environments for land, sea, and air operations. Its strength lies in its ability to simulate large-scale military exercises and operations. Integrating machine learning capabilities can enable this platform to provide real-time tactical adjustments and predictions, enhancing the realism and effectiveness of training exercises.

Image: Some existing gaming platforms can be adjusted and improved for military purposes. “Command: Modern Operations” (left) allows users to design specific combat scenarios, while “Modern Naval Warfare” (below) focuses on high-fidelity simulations of submarine warfare, surface ship engagements, air defense, and other operations. Slitherine / Matrix Games Ltd

Evolution Timeline

The U.S. “2022 National Security Strategy” states, “The post-Cold War era has come to an end, and great powers are competing to shape the next era.” This signifies a significant shift in the battlefield environment from relatively predictable to unpredictable. Therefore, there is an urgent need to develop and implement cutting-edge technologies to effectively manage and reduce uncertainty. Just like Doctor Strange, Navy decision-makers, with the assistance of AI, will be able to navigate countless scenarios to find the tactics and strategies that ensure victory on uncertain battlefields.

Follow Zhuanzhi for Quick Access

Please follow the Zhuanzhi Intelligent Defense WeChat account (click above to follow)

• For more related content, please check Zhuanzhi, click “Read Original” on the bottom left to view

• Welcome to scan the QR code to add Zhuanzhi Assistant, for consultation services

• 

Zhuanzhi, Professional and Reliable AI Knowledge Distribution, making cognitive collaboration faster and better! Welcome to register and log in to Zhuanzhi www.zhuanzhi.ai, to obtain 100,000+ AI (AI related to military, medicine, public security, etc.) theme knowledge materials!

Click “Read Original” to learn how to useZhuanzhi, view and obtain 100,000+ AI theme knowledge materials