Contributed Poster Presentations: Section on Physical and Engineering Sciences

Finding decisions in real-time that optimize multiple outcomes involves sequential decision-making with a priori uncertainty in multivariate functional objectives. In addition, in practice observations of functional objectives can also be difficult to obtain due to computational or monetary expense. This challenge is common in engineering applications, such as optimizing a racing vehicle's performance by minimizing both lap time and fuel consumption. To address this, we propose a dynamic surrogate modeling strategy based on Gaussian Process (GP) regression with auto-regressive (AR) structures to both emulate and forecast objective functions. This strategy both captures temporal dependencies while efficiently approximating time-varying objective functions. The AR-GP model is incorporated into a multi-objective online optimization framework, allowing for adaptive decision making with reduced computational overhead. Preliminary results, based on an example motivated by engineering design and autonomous vehicle applications, demonstrate that AR-GP modeling outperforms standard GP surrogates by providing more accurate predictions and better uncertainty quantification. 

Cancer and other chronic diseases associated with exposure to fireground smoke and contaminated Personal Protective Equipment (PPE) have become long-term health concerns for firefighters. A controlled laboratory smoke combustion simulation with high consistency and reproducibility can provide detailed records of smoke contaminants for studying the nature of fireground smoke and guiding the improvement of PPE safety in the future. In a lab-controlled environment, nine representative groups of smoke combustion experiment records, including Smoke Particle Size Distribution (SPC) and combustion Heat Release Rate (HRR), are collected to study the patterns of different smokes. We constructed a hierarchical model framework that integrates Beta Regression models for shifting registration, functional principal component analysis (FPCA) for functional data representation, and mixture models with temporal dependence for calibrating the smoke distribution during phase transition to identify and analyze the effects of heat and fuel materials in the combustion process. Additionally, we used Bootstrap methods to assess the model uncertainty and the quality of the synthetic smoke profiles. We use our proposed model to make inferences on the primary factors in combustion experiments, suggesting that different combustion conditions lead to distinct mechanisms within both the ignition and flaming phases, providing insights into the interaction between heat and fuel materials during the combustion process. 

Using proprietary hectic motion sensing techniques, we at Simology.com deliver AI learning tech that allows AI to learn motions and recreate all types of motion in a simulated environment. This presentation provides a demonstration of the tech itself, its uses, and a comparison to other AI interface models. To create a sentient AI capable of humanistic actions, AI must learn not only pixelated data but a range of sensations from hearing taste smell and also touch. At Simology we are currently teaching our virtual AI to process sounds with text-to-speech (tts).We intend to improve these speech capabilities using a motion analysis for sound transitions. Similarly, we are preparing a touch learning capability based on a pressure switch. Taste and smell functions for an AI will be learned with microfluidics and detecting the motion of liquids. Eventually these techniques will be used in robotics to design artificial animal behavior. We compare the advances of Simology virtual AI to other competitors such as WebSim and Claude. Our results indicate that Simology is capable of using hectic motion sensing tech using HTML protocols, and is able to produce superior models of movement.