���Ǵ�ý

A digital mission-planning environment is constructed from vast amounts of data from across the operating environment, the warfighters themselves, and military doctrine. The foundation is a digital twin: a technically exact, virtual replica of an object, process, or system. A digital twin can be created within a few hours from data collected via existing maps and drone flyovers using full-motion video and light detection and ranging scans. The result is a high-fidelity rendering of where a military mission will take place.

The next layer is the data, which is embedded over the digital twin to provide an easily digestible look at key variables and other parameters. This includes the data used to create the digital twin as well as data that is pulled into the system about military doctrine, individual warfighter readiness, weaponry, weather, and open-source and signals intelligence. Analysis and measurement happen in the embedding layer through the deployment of traditional AI and mathematical optimization tools. For example, a traditional mathematical optimization model could determine the fastest route between points A and B for a four-person unit carrying a defined set of equipment. AI models could analyze the terrain using the categories of the Army’s OAKOC: obstacles, avenues of approach, key terrain, observation and fields of fire, and cover and concealment.

In addition to replicating the mission environment, digital twins can generate data that goes beyond what sensors affixed to drones can capture. A digital twin can be programmed to simulate different weather conditions, combat scenarios, and other variables. The value of this simulation is twofold: Warfighters can test themselves and evaluate their performance in a variety of conditions, and the AI optimization tools can train on this additional data to improve their functionality. (Learn more about AI’s recursive nature.)��

Introducing a GenAI assistant transforms a digital mission-planning environment into a next-generation capability. A GenAI model goes beyond predicting outcomes to generate new content and insights. It can use information and patterns gleaned from unimaginably large datasets to not only recommend the best course of action but also tweak that recommendation based on additional variables and feedback that the commander feeds it in real time. For mission planning, a foundational GenAI model trained on broad, open-source information is augmented with mission-specific information that isn’t publicly available, such as military doctrine and procedures, uniform and equipment specifications, an adversary’s typical warfighting tactics, and classified mission-specific information.

GenAI can also monitor the performance of individual warfighters. Today, warfighters can be outfitted with wearable technologies that generate information about heart rate, respiration, sleep patterns, stress levels, posture, movement speed, stride length, and other metrics. This biometric data provides critical assessments of physical and mental readiness, which can be combined with a three-dimensional holograph of each warfighter in a display that also includes tactical equipment models and information about the terrain to be traversed. Within the display, planners can “drag and drop” different equipment onto the warfighters to ensure every warfighter can effectively manage the load they’re assigned. This in turn leads to an increase in overall speed during the mission because a unit can move only as fast as its slowest person.

Orchestration software that uses agentic AI is what makes everything work (learn more about agentic AI). When a commander inputs a query, a “commander” orchestrator agent interprets the request and assigns tasks to subordinate agents. Each agent consults its primary military procedures to identify the appropriate steps to answer the question. Then, based on available analysis tools, the agents execute each sequential step or note a gap in the available data or capabilities before returning a response to the “commander” agent for review, consolidation, and response to the user.

Each step in this sequence is logged and available for the user to investigate to ensure transparency and auditability and provide quality control against any potential AI-generated “hallucination” that could have an adverse effect on the operation.

Importantly, nothing about this approach takes decision making away from the commander. Instead, it provides more tools to work with and more hard data to draw from, to which the commander can apply their experience, knowledge, and judgment to make better decisions. In addition, humans must review recommendations with an eye toward ethical principles, situational nuances, and real-world conditions that may be beyond current AI models’ understanding and instead draw from their experience and expertise. 

Sophisticated electronics packages make it possible to blend the simulated environment with real physical weapons in call-for-fire rehearsals. Soldiers encounter a 360-degree virtual experience while simultaneously engaging with a surrogate weapon that provides a realistic tactical and kinesthetic feel. The electronics package includes a microprocessor that collects sensor inputs from the weapon, such as the trigger and safety, and a gyroscope that acts like a motion sensor, tracking the weapon’s movements.

As simulations are presented to the soldier, their actions and performance are depicted on a screen. They can view the target through their weapon’s optics, measure the range, execute the fire, feel the recoil, and assess the damage. This type of robust, mixed-reality rehearsal exercise in a virtual environment allows soldiers to practice the call-for-fire again and again, developing the necessary muscle memory and learning how to control their emotions and remain calm in stressful situations. AI can pinpoint the factors behind an unsuccessful call-for-fire rehearsal, such as mistakes made because of stress or lack of training—giving each warfighter and their commander real-time insight into what’s inhibiting their performance and how to overcome it.

All plans change at contact, so after-action reviews of what actually happened are essential for optimizing future performance. Conventional after-action reviews center on personal observations from leaders and participants about operation and task performance compared with the intended outcome. This helps everyone to evaluate what did and did not occur and why—and what can be done to sustain strengths and improve weaknesses to do better next time.

Digital twins and GenAI inject incontrovertible data into after-action reviews. Performance measures are fused with the GenAI assistant, which recommends changes to the planned route to speed up time to target. Similarly, mission analytics measurements and biometric data captured during rehearsal provide insight into unit and individual performance and which changes can be made to improve the mission plan and execution. Performance measures can include squad formation accuracy, speed of travel, and individual stress and energy levels. Data from wearables can help commanders assign the right soldier to the right job based on information about individual strengths and aptitudes, such as whether someone naturally scans an environment and therefore would be a better gunner compared with someone who focuses only several meters in front of themselves. Such a data-informed after-action analysis can be conducted after each rehearsal evolution to further optimize the plan. When training and rehearsing happens in a digital environment, review and analysis can also be integrated into the exercise so trainees can course correct in the moment—something that’s not possible with live exercises. 

Furthermore, these same technologies can be used to optimize supply chains and logistics. Their application in training saves the time and cost associated with live environments, significantly improves pass rates, and has the add-on benefit of warfighters having more family time because of reduced travel. The Department of Defense needs to continue to invest in AI because of its ability to improve decision making on both the tactical and strategic levels.