41530 research outputs found
Sort by
Persuasion and Critical Art Analysis: A New Approach to Student-led Collaboration in a Communication Classroom
Full text linkAdversarial Deep Reinforcement Learning for Tank Duel Simulation Using Lidar-Based Observations
Full text linkPrevious research has demonstrated that reinforcement learning agents can learn to steer differential-drive robots around obstacles using 2D lidar scans as observations. However, these studies typically treat all range returns as undifferentiated obstacles—objects to avoid—without distinguishing between different object types. This thesis builds upon previous research by introducing an adversarial task in which an agent must interpret raw range readings to both avoid static obstacles and identify, pursue, and engage a hostile target.
To investigate this problem, this thesis introduces TankGame, a novel, lightweight 2D tank duel simulator. Each agent receives a 360° lidar scan, controls its motion via tread velocities, and fires slow-moving projectiles within arenas containing either procedurally generated or handcrafted obstacle layouts. Obstacles are modeled as circles of varying radii, while tanks and projectiles have rectangular profiles, requiring agents to infer object types purely from geometric cues in the lidar data. Implemented in C++ with a standardized Python API, TankGame achieves approximately 4,000 simulation steps per second on a single CPU thread—offering a high-fidelity yet computationally efficient alternative to grid-world abstractions and complex game engines.
Alongside the TankGame environment, this thesis presents a three-stage training methodology. In the first stage, a Proximal Policy Optimization (PPO) agent with a Circular Convolutional Neural Network (CCNN) encoder is trained against a static opponent, allowing it to jointly learn lidar features and a baseline policy. In the second stage, new actor and critic heads, each incorporating a Long Short-Term Memory (LSTM) layer, are trained atop the frozen CCNN encoder, enabling the agent to exploit temporal dependencies within the lidar features. Finally, in the third stage, the LSTM-PPO agent undergoes adversarial fine-tuning through self-play to improve its robustness against adaptive opponents.
Agents are trained on procedurally generated maps and evaluated on a suite of three evaluation map sets. A batch of 400 procedurally generated maps assess overall performance, while two collections of handcrafted maps test obstacle navigation and the ability to locate and eliminate static or moving targets. Experimental results demonstrate the LSTM-PPO agent defeats the PPO baseline in 64% of head-to-head duels (baseline wins 20%) and can locate and eliminate static targets in 97% (baseline 88%) of the procedurally generated maps, with additional gains achieved through adversarial fine-tuning.
The key contributions of this work are the open-source TankGame environment, a reproducible training methodology, a suite of evaluation maps, and high-performing baseline agents to support future research in adversarial reinforcement learning
Cruising on the Beach (COB)
Full text linkThis report details the design and development process of Cruising on the Beach (COB), a modular motorized wheelchair intended to improve beach accessibility for high school students with mobility impairments. The project aimed to create an affordable, portable, and sand-capable device using balloon wheels, an adjustable control panel, and a customizable base. Extensive background research, market analysis, design modeling, and test plans were completed. However, final fabrication was not achieved due to financial constraints and limited access to industrial machining tools needed for custom metal components. Despite these setbacks, the project delivered a robust conceptual design and a fully documented development framework, providing a strong foundation for future implementation
Statistical Investigations of Strategies in the Game Ecosystem
Full text linkThis work provides a probability-based analysis of strategies in the board game Ecosystem. Ecosystem is a turn-based multiplayer tiling game, where players take turns picking a wildlife card from a limited pool of cards then placing that card on their personal 4x5 grid. The objective of the game is to place the wildlife cards to maximize your score, as each card’s scoring condition depends on the presence or absence of certain cards surrounding it. The goal of this project is to determine optimal strategies for tiling your grid using techniques such as simulation to find optimal grid arrangements and clustering to examine commonalities in card placements and selection within a grid. This work includes several processes for investigating strategies. For grid simulation, a function for generating and placing cards randomly in a grid and a scoring function for a modified single player version of the game was made. For grid arrangement, a Markov Chain Monte Carlo simulation with a tuned accelerated acceptance function to discover the optimal arrangement of a grid to maximize scoring was utilized. For grid pattern analysis, a large numbers of grids with maximized scores were generated and clustered using K-means clustering to identify possible patterns in card placements and selection. And for strategy inferences, multiple graphical and tabular results to display and compare various aspects of possible winning strategies were created
A Feasibility Study on Series Hybrid-Electric Propulsion for Narrow-Body Transport Aircraft
Full text linkAs global climate change accelerates, the reduction of carbon emissions has become a critical objective across multiple industries. The aviation sector, however, has lagged behind other transportation domains in implementing sustainable alternatives, despite growing demand for air travel. Recent research and emerging technologies have proposed several decarbonization pathways, including hydrogen fuel, sustainable aviation fuels (SAFs), and hybrid or fully electric propulsion systems. Among these, series hybrid-electric propulsion has received comparatively limited attention, particularly for application in transport-class aircraft. This study investigates the feasibility and emissions-reduction potential of a series hybrid-electric propulsion architecture applied to a representative narrow-body platform. A Python-based mission simulation framework was developed to model fuel consumption across four representative mission distances, incorporating detailed propulsion system parameters and energy management logic. Key variables—battery specific energy and turboshaft thermal efficiency—were evaluated across a range of values corresponding to relevant Technology Readiness Levels. Results indicate that with a battery specific energy of 500 Whkg-1 and thermal efficiency 30%, series hybrid-electric configurations could achieve up to 30% reductions in mission-level fuel consumption on short and medium-range routes, with diminishing returns observed at longer distances. Comparisons between fixed and mission optimized battery mass strategies revealed that mission-specific battery sizing can yield up to 7% additional reduction in fuel consumption. However, required battery specific power levels exceeding 3,000 Wkg-1 remain well beyond the capabilities of current or even some future battery chemistries. This constraint represents a significant barrier to implementation and underscores the importance of continued battery development, or even a shift in aircraft operations. These findings suggest that while series hybrid-electric propulsion offers promising reductions in emissions, particularly for short- and medium-range applications, its practical deployment will depend on further advancements in high-power battery technology and the adoption of mission-specific aircraft design strategies
A Digital Dive: Redesigning the Cabrillo High School Aquarium Website
Full text linkTucked away on the Central Coast in Lompoc, you’ll find the Cabrillo High School (CHS) Aquarium. Started in 1986, the CHS Aquarium is the only high school aquarium of its kind in the nation run entirely by high school students. This 10,000+ square foot aquarium serves an underserved community at a Title I school, where students manage all aspects of animal care, nutrition, breeding, educational curriculum development, and visitor tours. This program is truly one-of-a-kind and deserves the spotlight for just how unique it is. As a CHS graduate, I felt the current website lacked in many areas and could use a splash of life. For this senior project, my aim was to create a refined style guide to create a more consistent look and design prototypes of a redesigned website. The current website is outdated, lacks content that effectively highlights the aquarium’s offerings, and can be difficult to navigate. The goal of this project was to design a website that more effectively represents and showcases the aquarium, enhance the educator and user experience, and boost awareness of what the aquarium has to offer
Keepsake: A Social Media and Management App for Collectors
Full text linkFor my senior project, I aim to target a nearly nonexistent digital market in social media applications for collectors by designing an app meant for collectors to buy, sell, trade, and connect over similar interests. The plan for this project is to prototype and design an app for a community of people who do not currently have a dedicated space to connect. This project provides an opportunity to innovate through design and fill in the current gap existing in social media applications