Tag Archives: Engineering

Relational Reinforcement Learning and the Curse of Dimensionality

Joseph Millar, Utah Valley University

Machine learning is a branch of artificial intelligence which focuses on a machines ability to learn rules that make decisions based on some input without the rules being explicitly programmed. This project focuses on Relational Reinforcement Learning (RRL). This is a theory based on the combination of reinforcement learning and inductive logic programming. Reinforcement learning is the idea of learning which sequence of choices achieves the highest reward. Inductive logic programming is learning logical rules that consist of a set of features that are used to describe a situation. One of the problems that is prevalent in RRL is the curse of dimensionality. This means that as the dimensionality of data is increased the complexity of describing and analyzing the data increases, sometimes exponentially. The central research question of this study is to reduce the curse of dimensionality when RRL is used. To achieve this, our theory is that adding better situated features would reduce the curse. To test this theory, we studied the chess endgame of King and Rook vs. King on a 4×4 board. To learn our RRL rules we set up chess-board states of checkmate, checkmate-in-one, and checkmate-in-two. We found that for checkmate there were 40 possible board states, checkmate-in-one had 120 possibilities and checkmate-in-two had 1369 possibilities. We found that for checkmate we needed 30 features per rule, with checkmate-in-one we also needed 30 features and with checkmate-in-two we needed 1239 features. The ratio of features to board-states was 30/40 = .75, 30/120 = .25, and 1239/1369 = .905. The number of features per board-states should be shrinking instead of growing. This is an example of the curse of dimensionality. In order to begin to solve this we created two features that incorporated Rook moves that ended up in the same location. The projected outcome is that the number of features in a rule that are required to describe the board states should decrease by 20%-30%. Therefore, by modifying the set of rules, the curse of dimensionality would be reduced.

Effects of coral mining on community dominance of macroalgae vs scleractinian coral on three reefs near islands of varying coral wall volumes in the corregimiento Nargana, Guna Yala Province, PanamÌÁ

Jackson Podis, Westminster College

In the Guna Yala archipelago, PanamÌÁ, the removal of coral species for construction of coral walls has been a common practice for the Guna Yala indigenous group. This practice has the potential to drastically alter the community structure of offshore reefs. This study analyzed three reefs offshore of islands with varying coral wall volumes to quantify macroalgae and scleractinian coral cover, diversity of scleractinian coral species, and correlation between macroalgae and scleractinian coral cover. All three study sites exhibited significant differences in scleractinian coral coverage; a significant negative correlation was shown between scleractinian coral cover and macroalgal cover, and the site with the largest coral wall volume showed the lowest rates of coral species commonly used for mining. These results are telling of the potential effects coral mining can have on coral reefs in the Guna Yala archipelago, and aim to inform the development of marine resource management plans in the future.

Permanently Fog-Resistant Surfaces Optimized for Polycarbonate

Thomas Shober, Jaxon Roller, Ashley Kennedy, University of Utah

Fogging lenses are a problem in athletics, science, health, mining, and other industries. Several studies have concluded that fogging goggles lead to injury and dramatically affect the performance of a user. Fogging on lenses is caused by the scattering of light by microscopic water droplets that form from by means of condensation. Using a hydrophilic layer chemically bonded to a polycarbonate substrate, water can be absorbed uniformly rather than in droplets. This allows the water to condense without giving up any clarity. The research conducted used a carboxymethyl cellulose and chitosan dip layer coating on polycarbonate. To strengthen the formed coating citric acid was used as a crosslinker. Polycarbonate slides with 15 bilayers of polysaccharides and a crosslinking solution comprised of citric acid and a catalyst of sodium hypophosphite exhibited excellent anti-fog properties and a robust coating.

Charcoal-Based Conductive Paint

Autumn Wyatt, Dixie State University

The process of creating conductive paint, done by mixing water-based acrylic paint with a conductive substrate, typically has used graphite as a substrate. While graphite has been investigated by many research groups, the optimal conditions for embedding graphite into the paint matrix needs more investigation. Our research focuses on wood sourced charcoal as the source for the graphite substrate, an affordable and simple methodology. We explored the ratio of substrate to matrix, substrate sources, and substrate particle size to determine what might produce the best conductive paint.

Submillimeter Rapid Fabrication Techniques for Microfluidics

Chase Omana, University of Utah

The purpose of my project is to assess submillimeter rapid fabrication techniques used in the field of microfluidics. Rapid fabrication is important because the current methods used are very time consuming, slowing down medical and scientific findings. The three major methods we used in creating our devices were the laser cutter, the knife plotter, and the 3D printer. There are advantages and disadvantages to each method used. Generally, the laser cutter is good for industrial manufacturing applications, and is known for leaving an edge with a high quality surface edge. But when we used it for cutting certain materials, the edges were melted and not very clean, creating problems in the devices. The knife plotter is used for cutting out specific crafts and is known for leaving uniform material. However, there have been some complaints about inconsistencies when attempting to cut submillimeter sized shapes. The 3D printer creates a three-dimensional object by adding materials one layer at a time. But when trying to develop submillimeter sized devices, the edges of the channel were very rough and inconsistent. From our research, we were successful in characterizing different fabrication techniques. This is important because it provides a framework for future decisions made for those students attending Utah Conference of Undergraduate Researchin the State of Utah Center of Excellence for Biomedical Microfluidics, as well as all those in the field of microfluidics.

Line Upon Line: Engineering Strong 3-Dimensional Cardiac Tissue

Joseph Rich, Brigham Young University

The heart is a crucial and complicated organ in our body. Once heart muscles, or Cardiomyocytes (CMs), are damaged, they cannot regenerate. Heart transplants are difficult to come by, and even with current methods of solving heart disease, heart disease is the leading cause of death today in developed countries [1]. One possible solution to the demand of transplants is through tissue engineering patient-specific cardiac tissue. There are significant problems that need to be solved before cardiac tissue can be useful. Two of these problems that we address in our research are the problems of size and vasculature. The purpose of our research is to create strong 3-dimensional samples of cardiac tissue with complete vascularization. Currently we have successfully decellularized porcine cardiac extracellular matrix (ECM), created 300 micrometer thick samples of the ECM, and make beating heart tissue by seeding in CMs derived from human induced pluripotent stem(IPS) cells into the ECM. These heart tissue samples beat for about 3 months. For future research, to increase vascularization and size, we will place CMs derived from IPS cells together with human umbilical vein endothelial cells (HUVECs) and caridofibroblasts (CFs) at a ratio of CMs:HUVECs:CFs, 1:1:5 [2], 1:3:6 [3], and 2:1:1 [3] onto a ECM and stack those ECMs through centrifugation for 15 minutes[4]. The action potential will be measured, and the vasculature will be seen through staining of the tissue using markers CD31 for HUVECs, Troponin T for CMs, ë±-actinin for CMs‰Ûª sarcomeric lengths, and Cx43 connexin 43 for CMs. By developing these cardiac tissue samples size and vasculature, we will get one step closer to creating patient specific heart implants to cure heart disease. [1] Hiroyuki Yamakawa, et. al. ‰ÛÏStrategies for Heart Regeneration Approaches Ranging from Induced Pluripotent Stem Cells to Direct Cardiac Reprogramming,‰Û International Heart Journal, vol. 56, no. 1, p. 1-5, 2015. [Online]. Available: https://www.jstage.jst.go.jp/article/ihj/56/1/56_14-344/_article. [24 July 2017]. [2] Rogozhnikov, Dmitry et al. ‰ÛÏScaffold Free Bio-Orthogonal Assembly of 3-Dimensional Cardiac Tissue via Cell Surface Engineering.‰Û Scientific Reports 6 (2016): 39806. PMC. [Online]. Available: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5180231/ [26 Oct. 2017] [3] Polonchuk, Liudmila et al. ‰ÛÏCardiac Spheroids as Promising in Vitro Models to Study the Human Heart Microenvironment.‰Û Scientific Reports 7 (2017): 7005. PMC. [Online]. Available: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5539326/ [26 Oct. 2017] [4] Haraguchi, Yuji et al. ‰ÛÏThree-Dimensional Human Cardiac Tissue Engineered by Centrifugation of Stacked Cell Sheets and Cross-Sectional Observation of Its Synchronous Beatings by Optical Coherence Tomography.‰Û BioMed Research International 2017

Axisymmetric Turbulent Wake Development

Kyle Hakes, Heather Erickson, Brigham Young University

Wakes behind objects moving in a fluid, such as a car, submarine, or airplane, are often unseen but cause drag on the object, increasing the force necessary to propel it forward. An improvement in the understanding of wake dynamics can improve building and vehicle construction to reduce drag. Computational evidence that shows an improvement to the previously accepted model has been presented, but not confirmed. Experimental confirmation of these simulations will improve the general understanding of wake dynamics.

Simulating Coal Transportation with Carbon Dioxide

Ariel Green, Taylor Schroedter, Brigham Young University

Most high-pressure combustion systems use water slurries to transport coal particles to the combustor. This is inefficient as a fraction of the energy released during combustion must be used to vaporize the water. Transporting coal with a gas would minimize this energy loss and improve combustor efficiency. However, transporting densely packed particles with gas at high pressure is challenging. This research evaluated the feasibility of using gas to transport dense phase coal through a simple pipe and identified the preferred initial coal distribution in the pipe to reduce particle removal times. CO2 was used to push coal through a 10-cm long, 0.635-cm diameter pipe to test the removal time when the gas mass flow rate and the initial position of coal particles was changed. Using Barracuda computational fluid dynamics (CFD) software, cases with differing flow rates and initial coal positions were simulated until 99% of the coal particles had exited the pipe. The time for removal was compared for each case. It was found that a greater gas mass flow rate will remove the coal particles from the pipe faster. At lower mass flow rates, a large amount of particles exited quickly, but the coal remaining trickled out very slowly, elongating the removal time and weakening the transport efficiency. More total energy would be required at the lower mass flow rates for the remaining particles to be slowly removed. Furthermore, the initial position of the coal particles proved to be a very impactful variable. When the mass of particles was distributed in half of the pipe from bottom to top, more time was required for 99% removal at every gas mass flow rate. Because of the set-up, particles had to essentially climb over one another to leave the pipe. This caused a build-up at the end of the pipe that took a long time to empty. Distributing the same amount of particle mass in the first half of the pipe from left to right resulted in much shorter times to empty. The trends identified with the test data can be extrapolated to larger systems to find the ideal method to fully remove coal from a pipeline and into a combustion chamber when using carbon dioxide gas.

Kinematics of the first and fifth metatarsals as determined by high-speed dual fluoroscopy

Konstantinos Karpos, Koren Roach, University of Utah

The foot plays a vital role in understanding lower limb joint kinematics and kinetics, as it is the first link in the kinematic chain that contacts the floor during gait. Even with the advent of advanced motion analysis techniques, the foot is often studied as one single rigid body segment, despite the numerous bones and joints throughout. Several foot models have recently been developed to independently investigate motions of the hindfoot, midfoot, forefoot, and hallux. However, these models still combine multiple bone motions into one rigid body for each of these segments. The purpose of this research was to determine the motion of the first and fifth metatarsals as their own dynamic system, rather than consider them a rigid body as was done in previous research. Three subjects were imaged with dual fluoroscopy (DF) while descending a set of stairs. This technique uses two x-ray cameras, placed approximately 90å¡ from one another, to record continuous x-ray images of the subject. This allows for in vivo bone motion to be determined within the three-dimensional (3D) volume of the combined field-of-view of the fluoroscopes. Separately, a CT image stack was acquired of each subject and the first and fifth metatarsals segmented. Projections through these segmentations were used to generate artificial x-rays of each metatarsal from numerous perspectives. A custom model-based markerless tracking software package was then used to align the artificial x-rays with the DF images to quantify the position and orientation of each bone in 3D space. A coordinate system was defined for both metatarsals. From which, the relative motions of the first and fifth metatarsals were compared during the weight bearing portions of stair descent. Since the metatarsals are dynamic systems, a rigid body assumption ultimately limits the understanding of foot kinematics. We hypothesize that the first and fifth metatarsals will demonstrate different motions during the loading and unloading portions of stair ascent. The comparison of the first and fifth metatarsals will allow us to determine if differences exist between first and fifth metatarsal kinematics, which could be particularly useful for future clinical diagnoses and investigations of various foot pathologies.