The following are abstracts for scheduled presentations at the Engineering and Technology Symposium 2026, hosted by the Bailey College of Engineering & Technology at Indiana State University. Thanks to our presenters and attendees for your participation and scholarship!
For times and locations of specific presentations, please refer to the event program.
Bailey College of Engineering and Technology, Indiana State University, Terre Haute, IN, 47809, USA. Email: vcardwell1@sycamores.indstate.edu
Abstract—Digital communication technologies have fundamentally changed how work is performed by enabling continuous connectivity across time and location. While this connectivity improves flexibility and responsiveness, it also allows work to extend into personal life, influencing employee well-being and performance. This systematic review synthesizes 75 empirical studies to examine how technology-enabled connectivity affects work–life outcomes through its influence on how easily work crosses into non-work time. Using a PRISMA-guided methodology and Scopus database search, studies were analyzed across five domains: technology drivers, boundary mechanisms, behavioral regulation, moderating factors, and psychological processes. Findings show that increased connectivity consistently makes it easier for work to intrude into personal life, which is associated with higher work–life conflict and reduced recovery, primarily due to sustained cognitive engagement and technology-related stress. Evidence of positive effects, such as improved flexibility and coordination, was less frequently examined. This review develops an integrative framework demonstrating that technology system characteristics, including accessibility, availability expectations, and communication intensity, indirectly shape human outcomes. These findings highlight the importance of considering human boundary dynamics in the design and management of digital communication systems to support sustainable, technology-enabled work.
Keywords—Technology-mediated work, Boundary permeability, Boundary Theory, Work–family conflict, Work–family enrichment, Work–life balance, Telework, Technostress, Telepressure
1Bailey College of Engineering and Technology, Indiana State University, Terre Haute, IN, 47809, USA. Email: bhancock12@sycamores.indstate.edu
Abstract—Microsoft Gaming is the video game and interactive entertainment division of Microsoft, a leader in personal computing, operating systems, applications, and cloud computing. Upon entering the market in 2000 they successfully launched multiple hardware models under the Xbox brand and established several successful award-winning video game brands in Halo, Forza Motorsports, Gears of War, and Fable. In 2023 Microsoft purchased Activision Blizzard, another industry leader in gaming with a large portfolio of financially successful and leading brands in Call of Duty, Overwatch, Diablo, and Warcraft. As the video game industry continues to evolve, Microsoft has positioned itself with a diverse portfolio of profitable games. However, as development costs continue to rise and profit estimates fluctuate with market conditions, additional investment requires sound financial analysis to maintain company profitability and health. Five areas of investment are considered based on industry trends, consumer habits, and competing firms. Using engineering-economy IRR and NPV methods, public financial filings, and industry economic data; recommendations are made to guide Microsoft Gaming’s investments. Several factors contribute to this recommendation including increasing development costs and time, consumer preference for games as a service titles, and the cyclical and variable nature of hardware releases.
Keywords—Xbox, video game industry, engineering economy, internal rate of return (IRR), net present value (NPV), Microsoft
AI Safety Lab, Bailey College of Engineering & Technology, Indiana State University, Terre Haute, IN, 47809, USA. Email: He.Wen@indstate.edu
Abstract—Humanoid robots are increasingly deployed in industrial environments where close physical collaboration with human workers occurs. Although these systems often operate at a human scale, their physical embodiment is governed by mechanical and task-driven constraints rather than biological anatomy. As a result, ergonomic assessment methods based on human anthropometry may not be directly applicable. This study examines ergonomic risk in human–humanoid collaboration from an anthropometric perspective by benchmarking six contemporary humanoid robots against ISO 7250 human body measurement definitions using externally observable geometry. Landmark identifiability, measurement feasibility, and relative body proportions were systematically evaluated without reliance on manufacturer-disclosed specifications. The results show that several ISO 7250 landmarks tied to biological anatomy are consistently absent, rendering multiple standard measurements infeasible, while many whole-body, joint-level, and reach-related dimensions remain measurable. Comparative analysis reveals four recurring patterns of anthropometric deviation: additive, subtractive, exaggerative, and speculative. They help explain how ergonomic risk arises beyond scale alone. The findings indicate that ergonomic risk in human–humanoid collaboration is driven by anthropometric compatibility rather than human likeness, highlighting the need for anthropometry-aware evaluation approaches in the design and deployment of collaborative humanoid systems.
Keywords—Humanoid, collaboration, ergonomic, anthropometric
1Undergraduate Student, Department of Engineering Sciences, Morehead State University, Morehead, KY, 40351, USA 2Assistant Professor, Department of Engineering Sciences, Morehead State University, Morehead, KY, 40351, USA *Corresponding’s Email: a.paul@moreheadstate.edu
Abstract—In this case study, the authors aim to investigate the transformative role of AI in optimizing key processes, including impedance matching, electromagnetic simulation, antenna design, and RF integrated circuit (RFIC) development. AI can reduce design cycles from weeks to days and lower costs through predictive modeling and automation. The increasing demand for high-speed, high-capacity data transfer and the greater likelihood of interference in the crowded electromagnetic spectrum make the design of radio frequency (RF) systems increasingly complex. Supporting high-speed data processing, maintaining signal integrity, employing advanced signal-processing techniques, and enabling high-speed computations are primary requirements. Therefore, the use of Artificial Intelligence can be an asset, as it can accelerate the design of complex, optimized RF designs by automating simulations, thereby reducing prototype build time. Replacing human intervention can optimize performance and reduce prototyping time by replacing manual, iterative processes with machine learning models. It enables rapid, high-dimensional, multi-physics optimization for antennas, circuits, and systems, improving accuracy in 6G/5G, IoT, and high-frequency/millimeter-wave applications. AI’s ability to synthesize real-world implementations from prediction of S-parameter using neural network to reinforcement learning for parameter tuning shows AI’s capacity to shift RF design from intuition-driven trial-and-error to precise, data-driven innovation.
Keywords—artificial intelligence, machine learning, radio-frequency, optimization, neural network
1Undergraduate Student 2Assistant Professor* Department of Engineering Sciences, College of Science and Engineering, Morehead State University, 150 University Blvd., Morehead, KY 40351*Corresponding Author’s Email: t.ray@moreheadstate.edu
Abstract—This paper investigates the foundation system and structural load analysis for a 14ft x 60ft mobile home. Two different foundation systems were designed and evaluated for load-bearing capacity and cost-effectiveness. The first design is a block foundation with tie-down anchors installed. The other design is a pier-pad without tie-downs, relying on soil and concrete adhesion and cohesion for its strength. Dead, live, snow, and wind loads were established in accordance with ASCE 7-22. Both LRFD and ASD load combinations, as applicable, were used for the designs. Each system met the required factor of safety of 2 for all design checks. The block foundations with tie-downs resulted in higher costs and longer installation time. The pier-pad footing without the tie-downs appears to be more cost-effective and thus a good option.
Keywords—structural load analysis
¹School of Engineering, Bowling Green State University, Bowling Green, OH, 43403, USA ²Department of Engineering Technology, Miami University, Oxford, OH, 45056, USA ³School of Engineering, Bowling Green State University, Bowling Green, OH, 43403 USA Corresponding’s Email: bezea@bgsu.edu
Abstract—This paper explores new approaches for augmenting deep learning and machine vision in automated industrial material inspection. Traditional manual sorting is often slow, labor-intensive, and prone to errors caused by fatigue. Deep learning and machine vision technologies, specifically the Cognex In-Sight 2D Vision system and the ViDi Classify tool, can be integrated to increase production capacity and reduce human limitations. An industrial case study on a tier-sorting machine is presented to improve efficiency and accuracy in material classification along a production line. The proposed approach automates the sorting of materials into predefined tiers with a higher degree of accuracy. The camera system and classification software work together to increase throughput and reduce operational costs in manufacturing environments. This study considers the systematic integration of lighting conditions, high-definition imaging, machine vision applications, and deep learning principles. Results are analyzed based on the performance of the ViDi Green Classify tool in accurately classifying and efficiently sorting materials. The findings demonstrate the effectiveness of this automated approach in achieving precise, fast, and reliable tier sorting, leading to improved quality control, better time utilization, and increased overall productivity in industrial inspection and classification applications.
Keywords—Deep learning, Automated Industrial Inspection (AII), Vision-based defect detection, Data-driven classification, Context-aware classification, Industrial artificial intelligence, Smart manufacturing, Edge computing
1 Digital Engineering Driven Discovery Lab (DED2L), School of Mechanical, Aerospace, and Materials Engineering, Southern Illinois University Carbondale, Carbondale, IL,62901, USA Corresponding’s Email: kaifeng.guan@siu.edu
Abstract—Tilt-wing aircraft represent a promising configuration bridging the operational gap between vertical take-off and landing (VTOL) systems and conventional fixed-wing aircraft, offering enhanced versatility for urban and regional air mobility. Their ability to transition between hover and forward flight provides significant advantages in runway independence, energy efficiency, and mission flexibility. However, optimizing tilt-wing parameters, particularly the coupling between tilt angle and freestream velocity, is critical to achieving stable and efficient flight during transition and cruise regimes. This study implements high-fidelity CFD modeling to investigate the aerodynamic performance of tilt-wing configurations under varying tilt angles (ϕ = 0°-70°) and wind speeds (V= 30 m/s and 50 m/s). In addition, the effects of propeller slipstream on the overall aerodynamic performance are evaluated by comparing configurations with and without propeller modeling. The propeller is represented using a fan-zone approach to capture the induced flow acceleration and its interaction with the wing and fuselage. The comparative analysis focuses on the variations in lift coefficient, drag coefficient, and lift-to-drag ratio across different tilt angles and freestream velocities. The results reveal how the propeller-wing aerodynamic coupling modifies flow separation behavior and alters the global aerodynamic characteristics during transition. These findings provide useful insights for understanding the role of propeller-induced flow in tilt-wing aircraft and support the aerodynamic optimization of integrated tilt-wing configurations for improved transition performance.
Keywords—Tilt-Wing, Transition Aerodynamics, Propeller-Induced Flow, Lift and Drag, CFD
1 Digital Engineering Driven Discovery Lab (DED2L), School of Mechanical, Aerospace, and Materials Engineering, College of Engineering, Computing, Technology, and Mathematics, Southern Illinois University, Carbondale, IL, 62901, USA Corresponding’s Email: adittya.barua@siu.edu
Abstract—The escalating global demand for carbon-neutral energy positions biomass gasification as a critical pathway for sustainable hydrogen production. This study presents a preliminary CFD investigation into the thermochemical behavior of a 15-kW fixed-bed downdraft gasifier utilizing wheat straw pellets. The simulation, conducted in ANSYS FLUENT, implements a non-premixed combustion modeling with non-adiabatic energy treatment to characterize complex thermochemical transformations across the drying, pyrolysis, oxidation, and reduction zones. A simplified non-porous approach was adopted to analyze the gas-phase chemical kinetics and spatial temperature distributions. Preliminary results show that throat geometry significantly improves thermal uniformity. This leads to temperatures high enough to crack heavy hydrocarbons and increase hydrogen yield. Comparison with existing literature confirms that while the current model provides a robust baseline for predicting syngas composition, the integration of porous media resistance is essential for future industrial scaling. This research contributes to the fundamental understanding of small-scale gasifier behavior, offering a framework for optimizing reactor designs aimed at high-purity hydrogen generation from agricultural studies.
Keywords—Gasification, biomass, downdraft reactor, wheat straw pallet, computational fluid dynamics
1R. B, Annis School of Engineering, University of Indianapolis, Indianapolis, IN, 46227, USA 2R. B, Annis School of Engineering, University of Indianapolis, Indianapolis, IN, 46227, USA
Corresponding’s Email: tculleto@uindy.edu
Abstract—Large Language Models (LLMs) have rapidly influenced many aspects of society, particularly education, due to their demonstrated ability to complete assignments and examinations across a wide range of subjects. Although prior studies have examined the educational impact of LLMs, much of the existing work relies on public or open problem datasets and lacks topic-specific analysis. In engineering education, especially within mechanical engineering, systematic investigations of LLM performance on specific problem types remain limited. Instead of using traditional methods that directly ask textbook questions to an LLM tool, our study adopts a model distillation process to evaluate LLM capabilities in solving statics problems. By distilling ChatGPT, we extracted 25 text-only statics questions and further constructed two additional datasets by adding diagrams and modifying their numerical values. Experimental results show that while LLMs perform well on text-only statics problems, their accuracy decreases when diagrams are introduced and the problems require multi-step reasoning. Further analysis suggests that this performance drop is not primarily caused by limitations in image recognition, but rather by difficulties in multi-step reasoning and in consistently applying extracted visual information across successive solution stages.
Keywords—LLM, Statics, Model Distillation, ChatGPT, Gemini
1E. S. Witchger School of Engineering, Marian University, 3200 Cold Spring Rd, Indianapolis, 46222, IN, USA
2Ancilla College, Marian University, 20097 9B Rd, Plymouth, 46563, IN, USA
Corresponding’s Email: mshokrolahshirazi@marian.edu
Abstract—Smart farming systems often use Internet of Things (IoT) sensor data to gain knowledge about the environment and plant growth conditions. So, detecting the abnormal data can assist with preventing sensor failure, enhancing system reliability, and protecting yield. In this work, we propose feature-based majority voting for isolation forest and Long Short-Term Memory Autoencoders (LSTM-AEs) by applying individual models per feature, and select the common detected abnormalities. While LSTM-AE models detect temporal deviations, isolation forest models complement the detections by isolating outliers based on feature-space sparsity. A hybrid method with a majority voting scheme across the feature-specific models improves anomaly detection accuracy by reducing false positives commonly associated with monolithic modeling. The method is evaluated on real-world agricultural datasets from the IoT agriculture and smart farming data repositories, comparing the base LSTM-AE and isolation forest methods with enhanced feature-based ensemble ones. Experimental results using statistical analysis and kernel density estimation demonstrate that the voting-based mechanism achieves better precision and interpretability, making it highly suitable for real-time agricultural monitoring and fault detection in resource-constrained environments.
Keywords—smart farming, internet of things, abnormality detection
Department of Engineering Sciences, College of Science and Engineering, Morehead State University, Morehead, KY 40351, USACorresponding’s Email: q.xu@moreheadstate.edu
Abstract—III-V compound semiconductors are the materials of choice for developing high-frequency and high-speed semiconductor devices due to their high intrinsic electron mobility. Compound semiconductor heterojunction bipolar transistors (HBTs) have excellent linearity and high current driving capability and are the leading devices to develop high-frequency high-power components. HBTs normally use an n/p/n structure to take advantage of much higher electron mobility as compared to the hole mobility. The thickness of the base layers of modern HBTs has been reduced to below 1000 Å to increase both cutoff and maximum oscillation frequencies. Such thin base layers make it extremely challenging to realize ohmic contacts in manufacturing HBTs. In this research, multi-metal-layer structures are used to develop ohmic contacts. Gold is used as the top capping layer because of its excellent electrical conductivity and wire-bonding ability. Refractory metals, because of their high melting points, are used as the atomic diffusion barrier to realize long-term stability. AuBe is used as the p-type dopant alloy layer to generate ohmic contact. Ti is used as the adhesive layer to ensure that the multi-layer structure does not peel off during fabrication. The influence of each individual metal layer will be investigated in terms of specific ohmic contact resistance, surface roughness, geometrical integrity and thermal stability.
Keywords—Ohmic contact, p-GaAs, HBTs, multi-layer metallic structure
Department of Biomedical Engineering and Informatics, Luddy School of Informatics, Computing, and Engineering, Indiana University Indianapolis, Indianapolis, IN 46202, USA
Corresponding’s Email: amkrug@iu.edu
Abstract—RNA-binding proteins (RBPs) play essential roles in post-transcriptional gene regulation by controlling mRNA splicing, stability, localization, and translation. Their dysregulation is strongly associated with cancer and other diseases. Existing high-throughput technologies such as enhanced Crosslinking and Immunoprecipitation (eCLIP) provides RBP-specific binding sites but requires individual experiments for each protein, while Protein Occupancy Profile sequencing (POP-seq) provides a transcriptome-wide map of protein-bound RNA regions. However, POP-seq lacks direct labels indicating which RBPs bind each site, limiting the biological interpretation. To address this limitation, we developed an integrated machine learning framework to classify RBP-associated binding regions by integrating POP-seq and eCLIP data. Using K562 cells, we processed POP-seq reads through a standardized next-generation sequencing (NGS) pipeline and intersected POP-seq peaks with publicly available eCLIP datasets to assign RBP labels. We extracted 96 sequence-based, genomic-context, repeat-associated, and signal-intensity features. After systematic feature selection, a Support Vector Classifier (SVC) trained on the 96 features to distinguish 11 RBP classes plus an “Other” category. The final model achieved Macro-F1 = 0.283 and Macro-AUC = 0.733 on held-out test data, with minimal overfitting (training Macro-F1 = 0.296). These results demonstrate that combining transcriptomic profiling with predictive modeling improves the systematic identification of RBP-binding patterns and supports data-driven studies of post-transcriptional regulation.
Keywords—POP-seq, eCLIP, RNA-binding proteins, machine learning
1Electronics and Computer Engineering Technology Department, Bailey College of Engineering and Technology, Indiana State University, Terre Haute, IN, 47809, USA
Corresponding’s Email: nathanjnr1@gmail.com
Abstract—This study introduces a novel time series forecasting method that leverages historical price fluctuation patterns. Our model uses a pattern recognition approach to predict both the direction and the magnitude of price changes. Using historical gas price data from the U.S.Energy Information Administration (EIA), our method demonstrates a strong performance by achieving a very low Mean Squared Error (MSE) across three test case months. The method’s change-direction prediction accuracy is above 75% for March, April, October and November, though it drops to below 50% for May and September, suggesting challenges in predicting price changes during these months. Additionally, the method uses algorithms, including group creation and frequency analysis, which contribute to its prediction accuracy. The group creation involves splitting the data into a fixed number of sizes m where each group contains m consecutive months. The frequency analysis algorithm computes the number of fluctuation patterns appearing in each group. This innovative method enhances understanding of gas price dynamics and equips users with data-driven tools to make informed decisions, potentially influencing both policy and business strategies in a volatile economic environment.
Keywords—Time series forecasting, historical price fluctuation patterns, pattern recognition, price changes, U.S. Energy Information Administration (EIA), Mean Squared Error (MSE), prediction accuracy, test case months, change-direction prediction, group creation, frequency analysis, algorithms, policy strategies, business strategies, volatile economic environment.
1Undergraduate Student Department of Engineering Sciences, College of Science and Engineering, Morehead State University, Morehead, KY, 40351, USA
2Assistant Professor, Department of Engineering Sciences, College of Science and Engineering, Morehead State University, Morehead, KY, 40351, USA
Corresponding Author’s Email: t.ray@moreheadstate.edu
Abstract—Rural bridges are essential components of local transportation networks, supporting daily travel, agricultural operations, school transportation, and emergency services. Many county‑owned structures, however, suffer from aging components, reduced load capacity, and long‑term deterioration that strain limited local resources. This study evaluates Bridge #035C00040N on CR‑1336 over Johnson Creek in Fleming County, Kentucky—a two‑span reinforced concrete box‑beam bridge constructed in 1969 and currently posted at 9 tons. This reduced posting reflects the bridge’s diminished load-carrying capacity due to progressive deterioration of the substructure. Settlement of the center pier, loss of load-path continuity, and scour around the foundations have significantly weakened the structure, requiring KYTC to restrict heavier vehicles for safety. The 9-ton limit, therefore, indicates that the bridge can no longer support standard legal loads. Using Kentucky Transportation Cabinet (KYTC) inspection reports, field observations, and engineering assessment methods, the research identifies progressive deterioration in the substructure, including pier settlement, scour, undermining, and widespread concrete distress. The center pier has fully separated from the superstructure, eliminating load‑path continuity and compromising structural performance. Based on structural analysis and site constraints, a single‑span CB12 prestressed box‑beam replacement is proposed to eliminate the pier, improve hydraulic performance, and restore long‑term serviceability. The findings demonstrate how systematic inspection data and engineering design practices can guide cost‑effective rural bridge replacement. Future work includes detailed structural calculations, hydraulic evaluation, and final design refinement under the guidance of the Engineer of Record.
Keywords—rural bridge infrastructure, structural deficiency, bridge inspection, scour, concrete box beams, AASHTO LRFD.
Professor, School of Business, Alcorn State University
Corresponding Author Email: bgeorge@alcorn.edu
Abstract—Computer science originated as the study of formal procedures and deterministic execution. That founding identity no longer captures what the discipline primarily does. Modern computational systems rank content for billions of users, allocate capital, generate natural-language text, recommend clinical decisions, and navigate physical environments. They exhibit behavior better described as decision-making than calculation, and better analyzed through the concept of agency than through formal language theory alone. This paper argues that the rise of trained, adaptive systems constitutes a genuine epistemological rupture and that computer science has consequently become the science of artificial agency. It further argues that this repositioning makes computer science the meta-discipline of the twenty-first century: a field whose artifacts populate society with new kinds of actors, and whose design choices propagate into every domain those actors enter. Cross-domain evidence from healthcare, business, education, and cybersecurity grounds the theoretical argument. The paper concludes with implications for curriculum reform, interdisciplinary governance, and a research agenda centered on agency, accountability, and alignment
Keywords—artificial agency; meta-discipline; machine learning; algorithmic governance; epistemology of technology; autonomous systems; philosophy of computer science
Electronics and Computer Engineering Technology, Bailey College of Engineering and Technology, Indiana State University, Terre Haute, IN 47807, USA
Corresponding author’s email: majid.afsharnoghondari@indstate.edu
Abstract—Decision trees are widely used because they are interpretable, efficient at inference time, and applicable to both classification and regression. However, most standard decision tree algorithms are greedy, selecting the locally best split at each node without reconsidering earlier decisions. This strategy is computationally efficient but can lead to suboptimal global tree struc-tures. In this paper, we propose a non-greedy decision tree based on beam search, where multiple candidate trees are maintained and evaluated during training instead of committing to a single split sequence. Candidate trees are expanded through alternative leaf splits and scored using internal cross-validation, allowing limited global search while preserving interpretability. We evaluate the proposed method on three classification datasets and three regression datasets, and compare it with standard greedy decision trees from scikit-learn under matched depth and minimum-split constraints. The results show mixed but encouraging behavior: the proposed method outperforms the greedy baseline on Hastie 10-2 classification and Diabetes regression, while remaining com-petitive on Breast Cancer classification. The findings suggest that beam-search tree induction is a promising non-greedy alternative, although training time remains substantially higher than that of greedy trees. The implementation used in this study is available at the following GitHub repository: https://github.com/Majid1292/non-greedy-decision-tree.
Keywords—decision tree, beam search, non-greedy optimization, classification, regression, interpretable machine learning