Main Subcommittee Category
TrainingHow would you label your submission?
AcademiaAbstract
Force-on-force training is a live event that Army Soldiers participate in as a final opportunity to apply learned skills in an environment that closely represents realistic combat. Currently, weapon effects are replicated using a laser-based solution called the Multiple Integrated Laser Engagement System (MILES). MILES does not accurately represent the behaviors of real munitions which can lead to unbelievable outcomes or, at worst, negative training. Therefore, the Army is looking for novel solutions that expand capabilities within these exercises. One such approach, called geo-pairing, uses position and orientation measurements and physics-based calculations. Feeding sensor data into a digital environment lets ballistic models calculate the projectile’s flight path to estimate where a live round would land. If the result intersects an instrumented target, the simulation returns a hit. In practice, these sensors and supporting models have inherent errors that prevent replicating real-world outcomes perfectly accurately. A proposed statistical approach offers a solution by providing understanding about sensors that have the greatest impact on overall accuracy, assessing likelihoods that a given result is accurate, and making recommendations about sensor requirements to achieve specific results. A sample model using uniform error distributions for two common sensor measurements—position and orientation—applied to a simplified representation of direct fire ballistics, demonstrates the application of probabilistic solutions in geo-pairing systems when the accuracy of available sensors is worse than the weapon’s accuracy. Running multiple trials using several thousand simulated firing events illustrates error impact on adjudication results and develops trends which may offer insights into future research prioritization efforts, hardware design decisions, or adjudication architecture designs. For example, understanding the probability that a given result is accurate could enable intelligent prioritization in a multi-system training solution (e.g. one with lasers and geo-pairing) or tuning training difficulty by deliberately altering hit probabilities based on field conditions.