American Military Academy

This paper outlines a proposal regarding the use of machine learning, specifically a long-short term model, to increase the military’s effectiveness and safety protocols. The approach is to collect data from weapons training and apply it to a model that can distinguish between weapon activities. By training the model on a dataset that consists of several common weapons activities, we hope to improve commanders' understanding of their troop's performance and readiness. The evaluation will consist of examining the loss of the model, its accuracy, and analyzing activities it frequently confused. This work will extend the current research in soldier activity recognition by introducing weapon activity recognition.

With the recent advances in materials, electric power generation and storage, and solid state laser technology, the time has come to examine the application and feasibility of developing a mobile high energy laser weapon system for the military.The high energy laser offers many advantages as a weapon system over conventional kinetic or explosive systems because of its power scalability, speed-of-light engagement capability, and precision engagement capability.The development of such a system requires the integration of numerous science and engineering disciplines as well as considerations of other factors such as the legal implications for fielding the system.Over the last two years a multidisciplinary team of undergraduate students and faculty from the United States Military Academy have been working with Lawrence Livermore National Laboratory and the High Energy Laser Joint Technology Office to help design portions of a mobile high energy laser weapon system and to examine the feasibility of fielding a system.The team consisted of physicists, mechanical engineers, electrical engineers, systems engineers, and a law student.This paper will describe the Solid State Heat Capacity Laser developed at Lawrence Livermore used in this study and identify the specific design issues addressed by the student team.It will also describe how the project was structured so that each student on the team had both an in-depth experience within their discipline and learned how to integrate their discipline specific expertise in the larger multidisciplinary project.Finally the paper will present the results of the students' work and learning outcomes.

<div class="section abstract"><div class="htmlview paragraph">Course of action (COA) generation for robotic military ground vehicles is required to support autonomous operations in well-structured and non-structured environments. Traditional pathing algorithms such as Dijkstra, A*, Hybrid A*, or D* are exhaustive and well structured, and as a result, a single COA may be derived if one exists. Traditional path-planning algorithms have been optimized to identify paths that achieve a single scalar objective (duration, distance, energy, etc.). The algorithms are not natively able to account for multi-objective cost considerations. Military operations represent multi-objective optimization problems, impacted by time, space, and atmospherics. The battlefield is dynamic and ever-changing, thus pathing algorithms must incorporate multi-objective costs and constraints and be provided in near-real-time or real-time. For this reason, the use of a genetic algorithm (GA) and Artificial Intelligence/ Machine Learning (AI/ML) were investigated for COA generation. Genetic algorithms can be used to solve globally optimal multi-objective optimization problems given sufficient time and computational resources; the GA may also be used to identify a potential solution space of COAs. AI/ML is well-versed in solving linear and non-linear problems and is well-suited for diverse predictive analytics ranging from time series forecasting, image analysis, classification, and contextual analysis and recognition. Training AI/ML is computationally intensive; however, the implementation can be accomplished with reduced computational requirements. Depending on the complexity, both methods can be implemented in near-real-time or real-time, making the algorithms ideal for mission planning, and near-real-time and real-time course of action generation. This manuscript generates multiple COAs for military ground vehicles, producing thousands of options with varying optimality regarding energy consumption and multi-variable objectives (energy, time, or detection). Traditionally, COA generation for autonomous mission planning relies on graph-based methods, reinforcement learning, or Q-learning. Non-traditional approaches, such as LSTMNs and CNNs, require the use of transformers to be effective.</div></div>

There is a real need to educate our engineering students in the application of electronics, controls, mechanics, and software; this multidisciplinary initiative has led to the creation of an undergraduate Mechatronics courses at the United States Military Academy (USMA) and many other universities around the world.The focus of these courses is to emphasize application and hands on laboratory work in general, and design projects in particular.This paper presents an example of an open-ended autonomous unmanned ground vehicle (AUGV) project that has been developed in support of the undergraduate mechatronics course at USMA.This is a one-semester course that culminates in a project that occupies the student's in class, laboratory, and at home assignments for the last five weeks of the quarter.The paper will present the design, development and pedagogy of the project.

Abstract Micro Unmanned Aerial Systems (mUAS) continue to gain prominence on the modern battlefield, providing an expedient and portable platform for conducting Intelligence, Surveillance, and Reconnaissance (ISR) to small, isolated combat units. The objective of this study is to develop a rapidly-deployable, intuitive, and low-cost means of preventing damage to mUAS rotor blades, enabling increased training hours at decreased potential cost and thereby increasing overall operator proficiency on the system. To that end, we concurrently pursued two possible solutions: (1) use of additive manufacturing to design and produce a low-cost, easily replaceable airfoil that could be used during training exercises in lieu of the high performance airfoil and (2) the use of additive manufacturing to create a protective structure around the existing rotor that could be easily affixed to the mUAS for training operations while reliably sheltering the rotors from potentially damaging impacts. While the replacement rotor blade revealed shortcomings in current additive manufacturing processes for this application, the protective barrier accomplished the goal of increasing survivability of the mUAS performance during training while not significantly impairing its performance. Through this design study, we succeeded in creating an adaptable, open source, easily distributed “press print” solution that will allow units across the Army to print, on an as-needed basis, rotor protection structures, easily and effectively increasing user proficiency on a key enabling technology.

Cobalt and cobalt oxide aerogels present materials solutions to challenges in energy, sensing, and catalysis via their high porosities and surface areas, low densities, and their magnetic and electrochemical properties. Traditional synthesis methods suffer limitations including aggregation, a need for templates, and slow reactant diffusion times. We present a magnetic-field assisted synthesis to prepare cobalt nanowire (CoNW) aerogels which allows rapid CoNW growth, is scalable, and does not require templates. A variable magnetic field allowed tuning of the CoNW structural properties with higher applied fields favoring longer CoNWs and higher aspect ratios. Thermal annealing allowed conversion of the CoNWs to Co3O4. The applied field strength and annealing parameters also influenced the surface areas, pore volumes, magnetizations, coercivities, and specific capacitances. This work demonstrates the magnetic-field assisted synthesis as a fast and scalable strategy to produce CoNW aerogels with tunable nanostructure, material phase, magneto-responsiveness, and accessible surface area for electrocatalytic applications.

In the nightside region of Earth’s magnetosphere, buoyancy modes have been associated with low entropy bubbles. These bubbles form in the plasma sheet, particularly during substorm expansion, and move rapidly earthward and come to rest in the inner plasma sheet or inner magnetosphere. They often exhibit damped oscillations with periods of a few minutes and have been associated with Pi2 pulsations. In previous work, we used the thin filament approximation to compare the frequencies and modes of buoyancy waves using three approaches: magnetohydrodynamic (MHD) ballooning theory, classic interchange theory, and an idealized formula. Interchange oscillations differ from the more general MHD oscillations in that they assume a constant pressure on each magnetic field line. It was determined that the buoyancy and interchange modes are very similar for field lines that extend into the plasma sheet but differ for field lines that map to the inner magnetosphere. In this paper, we create a small region of entropy depletion in an otherwise stable entropy background profile of the magnetotail to represent the presence of a plasma bubble and determine the properties of the buoyancy modes using the same 3 approaches. In the bubble region, we find that in some regions the interchange and buoyancy modes overlap resulting in frequencies that are much lower than the background. In other regions within the bubble, we find interchange unstable modes while in other locations MHD normal mode predicts an MHD slow mode wave solution which is not found in the pure interchange solution.