ARISE Colloquium 2021 | NYU Tandon School of Engineering

ARISE Colloquium 2021

Join us in celebrating the achievements of our ARISE 2021 class online on Friday, August 13th

Welcome to the 2021 ARISE Colloquium! This year our 64 students will be presenting in four concurrent Panel live-streams, which you can RSVP for here. We encourage attendees to stay for the duration of the Panels and to ask questions at the conclusion of each presentation. 


Panel A

RSVP

Students in Panel A listed alphabetically: 

Session 1

  • Adam David Avnet
  • Aliaa Mahgoub
  • Hao Teng
  • Jaden Ordonez
  • Jonathan Mota
  • Mark Falenchuk
  • Nicholas Cheung
  • Nicole Buitron

Session 2

  • Doris Hong
  • Justin Deng
  • Lihan Wang
  • Maliha Tasnim
  • Nazifa Rahaman
  • Nikki Chang
  • Omosefe Noruwa
  • Rebecca Foulen

Jonathan Mota

Jonathan Mota

  • Lab: Center for Advanced Technology in Telecommunications [Electrical Engineering]
  • Faculty: Professor Shivendra Panwar
  • Mentor(s): Fraida Fund, Ufuk Usubutun

 

Abstract: 

There has been some disagreement among consumers, service providers, and policymakers over whether a cellular network connection is an adequate replacement for home broadband. If it is considered a sufficient replacement, many areas currently considered "unserved" by broadband could lose eligibility for subsidies that encourage providers to serve new areas, so this question has serious policy implications. My research will explore this question by measuring the performance of several essential applications - video conferencing, video streaming, web browsing, and bulk file transfer - on networks with characteristics of 3G, 4G, and WiFi wireless links. The evidence I collect can help policymakers and others involved in this debate make more informed decisions.

 

Hao Teng, Nicholas Cheung

Hao Teng, Nicholas Cheung

  • Lab: Urban Mobility and Intelligent Transportation Systems (UrbanMITS) Laboratory [Civil & Urban Engineering]
  • Faculty: Professor Kaan Ozbay
  • Mentor(s): Suzana Duran Bernardes, Jingqin Gao

Abstract: 

Social distancing has proved itself to be effective against the spread of COVID-19. However, not everyone has followed social distancing protocols as rigorously as others. Previous research papers demonstrated that using a computer vision technique, called object detection, is a viable way to approximate social distancing distance. Object detection technique allows the identification and location of objects in images and video. For social distancing measurement, vehicles, cyclists, and pedestrians are identified, and the distance between each pair of pedestrians in the image is measured. With that information, we can convert distance between objects from pixels to feet through mathematical models. In our project, we analyze traffic camera video footage from a variety of intersection locations across the five boroughs of New York with object detection to measure urban social distancing. The income levels, education levels, and transit options of all of the locations are also analyzed to understand how these variables might affect the social distancing behavior at each borough. This study highlights the key socio-economic factors that contribute to urban social distancing.

 

Adam David Avnet, Nicole Buitron

adam, nicole

  • Lab: Visualization Imaging and Data Analysis Center [Computer & Engineering Science]
  • Faculty: Professor Claudio Silva
  • Mentor(s): Yurii Piadyk, Jorge Piazentin Ono, Guan-de Wu, Shuya Zhao

Abstract: 

As the number of videos uploaded to the Internet continues to rise, videos have become an essential data source in a variety of fields. For example, video data is one of the most common ways to analyze sports but could offer challenges because of the complexities of the games and the enormous amounts of data. With the overwhelming amount of new unstructured data available, machine learning techniques have become crucial to the analysis and organization of these videos. In this study, we will analyze baseball videos by annotating and training machine learning models that categorize this data, identifying players and events, such as pitches and hits. We first annotate a meaningful subset of a diverse collection of videos for significant temporal segments and spatial regions using the VGG Image Annotator (VIA). We then import the annotated videos into a Python environment and use machine learning algorithms to create the corresponding models. Finally, we test the trained models on the test set of videos. This study shows how machine learning techniques can be used to quickly examine sports footage for specific events, making it easier for players and coaches to learn from video recordings of games. These results can help analyze the movement of the players and allow them to significantly improve their performance.

 

Mark Falenchuk, Jaden Ordonez

Mark Falenchuk, Jaden Ordonez

  • Lab: Bio-Interfacial Engineering and Diagnostics  Group [Chemical and Biomolecular Engineering]
  • Faculty: Professor Rastislav Levicky
  • Mentor(s): Vlad Frenkel

Abstract: 

Spectrometry has proven to be beneficial in a diverse range of fields, from drug testing to identifying contaminants in sources of food or water by identifying various chemical compounds and even their concentrations. Spectrometry functions by dispersing a source of light into individual wavelengths, components of the source, named its spectra. Through this process we can identify the spectral makeup of a light source as well as the absorption and emission of light from a sample through which the light passes through. Many chemicals have unique absorption, emission, or excitation spectra which can act as fingerprints that identify their presence in a sample. Low-cost spectrometers are available and easy to construct, allowing for public access to scientific instrumentation which could allow for collaboration and crowdsourcing on a variety of topics. The degree of precision to which these spectrometers can operate is limited by many factors including construction, design, cost, and etc. By modifying the composition, size, and shape of the aperture through which light enters the instrument, we can attempt to increase the accuracy to which these low-cost spectrometers can detect various wavelengths of light. Then by using these low-cost spectrometers the spectra of different samples can be compared and information can be gathered to provide insight into the commonalities and differences between the samples. Focusing on optimizing the spectrometers and then comparing the spectra of multiple sources of water in the NYC area can act as the first steps in providing a method that would allow for the detection and analysis of contamination of the city water sources and to further understand the implications it has on the environment of our city. Results would be compared to existing information on common contaminants in certain water sources and would ultimately work to support the awareness of hazards and means to improve the health of our water.

 

Aliaa Mahgoub

Aliaa Mahgoub

  • Lab: Music and Audio Research Lab (MARL) [Electrical Engineering]
  • Faculty: Professor Juan Bello
  • Mentor(s): Iran Roman

Abstract: 

Sounds play a huge role in informing us about our setting and events within it. Although machines may benefit from utilizing sounds in a similar way, research is needed for machines to reliably recognize sound scenes in realistic soundscapes. This project aims to develop signal processing methods to automatically classify audio segments into their corresponding acoustic scenes, such as “shopping mall” or “park”, just to name a few. In order to do so, we will use several frequency-based features. These features help machines process audio like humans hear sounds. Examples are the spectral centroid, which is related to the perceived brightness of the sound, the spectral bandwidth, which describes how crowded with frequencies the spectrum is, and Mel-frequency cepstral coefficients (MFCCs), which describe the shape of a spectral envelope and model characteristics of the human voice. We will contrast these with non frequency-based features such as zero-crossing rate, the rate at which a signal changes from positive to zero to negative or from negative to zero to positive over time, amplitude envelope, the way that a sound’s amplitude evolves over time, and root-mean-square energy. In contrast to frequency-based features, these ones quantify what is seen in the signal’s time-domain plot. We will compare the performance of spectrum-based and time-based features for classification of audio segments belonging to the same or different acoustic scene. Classification methods we will use include the K-nearest neighbor classifier, and neural networks architectures, all of them trained with cross-validation. In the future, sound scene classification can be implemented in various applications, such as making context-aware devices and cars.


Omosefe Noruwa

omosefe

  • Lab: Machines in Motion Laboratory [Electrical Engineering]
  • Faculty: Professor Ludovic Righetti
  • Mentor(s): Avadesh Meduri, Julian Viereck

Abstract: 

Manipulating objects like humans has been a challenging problem in robotics. In this work, the students will learn the basics of robot manipulation through a series of tutorials. This will then equip them to implement a simple algorithm that will enable two 3 DOF finger robots to pick up a cube and move it. After verifying the algorithm in simulation, the mentors will run their code on the robot so that the students can see their algorithms in action on the real hardware. Through this project, the students will try to understand the difficulties in robot manipulation and gain experience with working on real robots.

 

Nazifa Rahaman, Nikki Chang

Nazifa Rahaman, Nikki Chang

  • Lab: Environmental Engineering and Water Treatment Lab [Civil & Urban Engineering]
  • Faculty: Professor Andrea Silverman
  • Mentor(s): Praneeth Challagonda, Catherine Hoar

Abstract: 

Of the myriad impacts that are predicted to accompany climate change, flooding is expected to have an outsized influence on public health, infrastructure, and mobility in urban areas. In New York City, for example, sea level rise and an increase in the occurrence of high intensity rain storms (which convey large volumes of water to drains, leading to backups and overflows) have led to a dramatic increase in flood risk, particularly in low-lying and coastal neighborhoods. The physical presence of standing water on streets and sidewalks can impede mobility and restrict access to transportation. Additionally, urban flood water contains a diverse array of contaminants, including industrial and household chemicals, fuels, and sewage.  Access to real-time information on flooding can improve resiliency and efficiency by allowing residents to identify navigable transportation routes and make informed decisions to avoid exposure to flood water contaminants. However, very little data exist on the frequency and extent of urban surface flooding, and there is an unmet need for hyperlocal information on the presence and depth of street-level floodwater. Therefore the objective of the FloodSense project is to develop and deploy a platform to provide real-time, street-level flood information - including the presence, frequency, and severity of local surface flood events - to a range of stakeholders, including policy makers, government agencies, citizens, emergency response teams, community advocacy groups, and researchers. This platform includes a physical sensor network, its connectivity, as well as data storage, and sharing infrastructure.

 

Doris Hong, Justin Deng

Doris Hong, Justin Deng

  • Lab: Hartman Research Laboratory [Chemical and Biomolecular Engineering]
  • Faculty: Professor Ryan Hartman
  • Mentor(s): Yukun Liu

Abstract: 

Plasma as the fourth state of matter consists of highly excited species, and its potential in synthetic chemistry is receiving growing attention in recent years. Dielectric-barrier-discharge (DBD) plasmas provides non-thermal plasma under atmospheric pressure and their device configurations and applications are of great interest. A DBD packed-bed micro-plasmatron has been designed in our laboratory to study chemical reactions with methane plasma. More intense electric fields have the potential of generating higher concentrations of methane radicals, and hence they can influence the reactivity. The present work explores why a change in the position of grounding and high voltage electrodes on the micro-plasmatron influences a change the electric field in the packed bed, and thus a change in the radical distribution profile in the DBD gap. A model of the DBD micro-plasmatron was built and tested in COMSOL Multiphysics with silicon, glass, and silicon layers as the dielectric barriers. These simulation results will be presented and discussed.

 

Lihan Wang, Rebecca Foulen

Lihan Wang, Rebecca Foulen

  • Lab: VIDA Lab [Computer & Engineering Science]
  • Faculty: Professor Juliana Freire
  • Mentor(s): Aline Bessa

Abstract: 

Transparency in the data cleaning process is essential to ensure the objectivity of the corresponding data analysis. Different data cleaning strategies can have different impacts on the analysis of the data. For example, Census and COVID Racial Data Tracker data can help us understand how COVID-19 impacted people of different races and ethnicities in the United States, but the strategies used to clean and aggregate such data can lead to distinct results, which in turn can foster different interpretations of how uneven the impact was across different racial groups, or even what groups were most affected at all. This example highlights that simply having access to the original data of a study or analysis is not enough to reproduce the corresponding results. More often than not, a time-consuming, reverse-engineering process is needed to understand the results or to verify their validity. Therefore, we propose a method to keep track of the strategies used to clean datasets as they are used in different projects or analyses, making the data, as well as the scripts used to clean them, available to the general public while maintaining the integrity of the original data.

1 Bureau, US Census. Census.gov, US Census Bureau, 2019, www.census.gov/.

2 “The COVID Racial Data Tracker.” The COVID Tracking Project, The Atlantic, 7 Mar. 2021, covidtracking.com/race.

 

Maliha Tasnim

Maliha Tasnim

  • Lab: Music and Audio Research Lab (MARL) [Electrical Engineering]
  • Faculty: Professor Juan Bello
  • Mentor(s): Magdalena Fuentes

Abstract: 

Noise pollution not only affects people that live in urban settings, but also has a negative impact on natural life, especially on birds, affecting their calls and migrational patterns. The effect of noise pollution on bird populations have been significantly overlooked. Certain species are becoming endangered and have lost what used to be their native habitat. The Sounds of New York City (SONYC) project aims to bring awareness to the problem of noise pollution by deploying a network of acoustic sensors throughout the city to monitor and analyze noise pollution using machine listening. In this project, we focus on using audio recordings collected from the SONYC sensors to understand the effect of noise on birds, by contrasting bird calls with noise patterns near the Washington Square Park region. Understanding patterns of bird calls in comparison to noise levels is an important step towards recognizing the impact that increasing urban activities have on wildlife. Research on these topics can be used to discuss ways of mitigating sound pollution, and to inform new conservational schemes and urban policies to preserve natural life in cities.



Panel B

RSVP

Students in Panel B listed alphabetically:

Session 1

  • Amber Zou
  • America Mendoza
  • Anastasia Choo
  • Brandon O'Connor
  • Fatim Majumder
  • Hennesy Guevarra
  • Juhyoung Lee
  • Raisa Jereen

Session 2

  • Afifa Tanisa
  • Anna Buckley
  • Chengge Wu
  • Elizabeth Louie
  • Eunice Son
  • Joshua Lee
  • Shafin Hossain
  • Sharmin Zaman

Brandon O'Connor, Juhyoung Lee

Brandon O'Connor, Juhyoung Lee

  • Lab: Applied Dynamics & Optimization Lab [Mechanical Engineering]
  • Faculty: Professor Joo H. Kim
  • Mentor(s): Hyunjong Song, William Z. Peng, Joseph Tuttle

Abstract: 

Humanoid robots are able to move with enhanced versatility and agility relative to other alternatives such as wheeled vehicles. However, these robots also bear the present risk of falling while they perform tasks such as walking and standing, a major challenge in robotics research. To address this, the stability region was introduced as an optimization-based balance criterion that could characterize and quantify the stability of legged systems. Balance stability regions were obtained for two reduced-order models: an inverted pendulum model mounted on a cart and an inverted pendulum-based model with a finite foot and tipping allowance. These computational results will be validated in the Webots simulation environment and the results of this research will inform approaches that extend the use of stability regions to consider legged locomotion.

 

Anastasia Choo, Amber Zou

Anastasia Choo, Amber Zuo

  • Lab: Biomolecular Engineering Lab [Chemical & Biomolecular Engineering]
  • Faculty: Professor Jin Ryoun Kim
  • Mentor(s): Edward Chau, Emre Erkanli

Abstract: 

Alzheimer’s Disease (AD) is the most common neurodegenerative disease and has long been associated with the aggregation of β-amyloid (Aβ). Aggregation of Aβ results in the formation of toxic oligomeric aggregates, which are responsible for the death of neurons in the brain. These oligomers can further aggregate to form insoluble fibrils, of which the amyloid plaques found in the brains of AD patients are composed of. Molecular docking is a tool used to help identify therapeutic compounds that can interact with Aβ and prevent its aggregation into toxic species. In this study, we examined the impact of these compounds, identified by molecular docking, on the in vitro aggregation of Aβ.

 

Fatim Majumder, Hennesy Guevarra

Fatim Majumder, Hennesy Guevarra

  • Lab: NYU Nanolab [Electrical Engineering]
  • Faculty: Professor Davood Shahrjerdi
  • Mentor(s): Brian McMinn

Abstract: 

As silicon computer and integrated circuit technology approach its limits, the only way to push forward will be to investigate other materials at the nano-scale. At NYU Nanolab, we specialize in layered nano-scale devices, utilizing materials like atomically thin graphene and molybdenum disulfide. Our devices have potential to alter the landscape of electronics and have shown promise in fields as disparate as security and neuro-sensing. This summer, Hennesy and Fatim are working to enable these devices by refining a characterization tool with rich physics: Raman spectroscopy. Using Raman spectroscopic data, this work reveals many properties (like the strain and quality) of graphene and other materials just by analyzing the optical spectrum of a special type of scattered light. This enables the study of the smallest material structures while leaving them pristine and untouched.

 

America Mendoza, Raisa Jereen

America Mendoza, Raisa Jereen

  • Lab: Ripollés Lab [Computer & Engineering Science]
  • Faculty: Professor Pablo Ripolles
  • Mentor(s): Michael McPhee

Abstract: 

Neurophysiological reactions to musical emotion have been traditionally studied using techniques designed for use in controlled laboratory settings. Changes in brain activity are monitored via neuroimaging methods like functional magnetic resonance imaging (fMRI) and magnetoencephalography (MEG), while physiological reactions are usually measured assessing fluctuations in the conductivity of the skin (electrodermal activity, or EDA). These techniques are, however, ill-suited for use in real-life environments (e.g., live concerts): MRI and MEG require specially designed facilities, while EDA is highly sensitive to movement, electronic noise, and variations in the contact between the skin and the recording electrodes. These limitations represent an important barrier to studying emotion outside the lab, in the more relaxed and social settings in which music is most commonly enjoyed. Over the past year, the Ripollés Lab has developed the CHILLER (Computer-Human Interface for the Live Labeling of Emotional Responses), an affordable and easy-to-use wearable sensor for the real-time detection and visualization of one of the most accurate biomarkers of musical emotion processing: the piloerection of the skin (i.e., the goosebumps) that accompanies musical chills. The CHILLER, based on the Raspberry Pi architecture, overcomes the real-world limitations of traditional techniques by using a well-known algorithm capable of detecting goosebumps from a video recording of a patch of skin. We have expanded the CHILLER’s functionality by integrating a temperature sensor into the design of the device. As goosebumps occur not only in response to emotion, but also in response to cold, accurate measurements of ambient temperature are needed to adjust the signal baseline. This new work makes the CHILLER more useful and accurate across a range of natural environments.


Anna Buckley, Elizabeth Louie

Anna Buckley, Elizabeth Louie

  • Lab: Developmental Genomics Lab [Biology]
  • Faculty: Professor Christine Rushlow
  • Mentor(s): Peter Whitney, Chris Rushlow

Abstract: 

Analysis of High Resolution Images of Gene Expression

Feature extraction for image analysis encompasses a broad set of computational techniques to interpret and quantify images. Continuous improvement in these techniques has allowed researchers in many fields of study to greatly expand experiments that normally would be limited by time-intensive manual annotation. However, many algorithms used in feature extraction produce some percentage of errors which limit the precision of some analyses. Here, we employ a number of different strategies to solve a feature extraction problem relevant to biological research. This problem involves assigning a feature associated with gene activation to a specific nucleus in a monolayer of cells imaged with confocal microscopy. We compare a Euclidean distance method, an image segmentation method, and hand-curation method, ranking these approaches by both accuracy and time-intensiveness. We will use this data to create a hybrid approach, one which uses manual intervention for ambiguous cases, but automatically annotates in cases of high confidence. This final approach will be compared to all three previous approaches as a potential strategy for a broad class of problems where 100% accuracy is required, and maximum time efficiency is desired.

 

Chengge Wu, Shafin Hossain

Chengge Wu, Shafin Hossain

 

  • Lab: Smart Energy Research (SEARCH) Group [Electrical Engineering]
  • Faculty: Professor Yury Dvorkin
  • Mentor(s): Samrat Acharya, Anwar Khan

Abstract: 

The roll out of Electric vehicles (EVs), which is one of the most prominent tools towards decarbonization, is increasing. This growth requires a great deal of EV charging stations (EVCSs), however, there is a noticeable discrimination against low- and middle-income communities in the adoption of EVs and EVCSs. Fair access to the EVCSs is essential in the promotion of EVs. Thus, we study the correlations between econometrics such as income of the people, EVCS capacity, and EV charging price across Manhattan, New York. We believe that a proper understanding of such correlation helps to identify inequalities in EVCS distribution and improve the accessibility and affordability of EVCSs, and hence, boost the EV adoption.

 

Joshua Lee, Sharmin Zaman

Joshua Lee, Sharmin Zaman

  • Lab: Soil Mechanics lab [Civil & Urban Engineering]
  • Faculty: Professor Magued Iskander
  • Mentor(s): Linzhu Li, Abdelaziz Ads

Abstract: 

High strain rates play a major effect on soil properties, especially for the undrained shear strength. Although many studies have been conducted on the effect of high strain rates for sand, there is only limited data for clay properties. The aim of this study is to determine the unconfined shear strength for clay soils by utilizing a drop weight test. A drop weight tower will provide the required impact velocity to reach the high strain rates. In order to record fast moving objects, a high speed camera was used to capture soil deformation during the test. Digital image correction methods were utilized to estimate the displacement at the failure. Two load cells were used to measure the load at the top and bottom of the sample. Wave propagation during the test was captured and analyzed with the stress-strain curve.

 

Eunice Son, Afifa Tanisa

Eunice Son, Afifa Tanisa

  • Lab: Responsible Data Science Lab [Computer & Engineering Science]
  • Faculty: Professor Julia Stoyanovich
  • Mentor(s): Kunal Relia

Abstract: 

Computational Social Choice, which is at the intersection of social choice theory and computer science, uses computer science techniques to elicit social choice preferences. Specifically, social choice theory analyzes the design and method of collective decision-making procedures such as the voting protocols used in the United Nations General Assembly (UNGA) to decide the passage of resolutions that provide resources to its member nations. Specifically, the UNGA consists of 193 member countries, each represented by one vote. The UNGA ensures diversity among the resolutions it passes but given the different numbers of countries from each continent, it is not known whether the voting protocol used by the UNGA ensures equal representation of each continent. Moreover, a recent theoretical framework showed that diversity and representation are separate but equally important aims of an election. Diversity ensures that candidates from diverse groups are selected, and representation ensures that the choices of different voter populations are represented. In our work, we use this theoretical framework and empirically analyze the voting on 1681 resolutions presented in the UNGA from 2000 to 2021. We study the degree of representation of different continents in the UNGA. We specifically focus on fifty-four African nations as they have been historically undermined, and the resources taken from them have far exceeded the resources given to them. Hence, we ask whether the African Nations are represented equally compared to their peers in the collective decision-making process of the UN? Our findings will characterize the position of African nations in the “United” Nations General Assembly.



Panel C

RSVP

Students in Panel C listed alphabetically:

Session 1

  • Eric Gonzalez Cardona
  • Esther Bistricer
  • Gregory Papadopoulos
  • Lina Raouf
  • Mariya Meleganich
  • Mia Montrose
  • Richlove Nkansah
  • Susan Jimenez

Session 2

  • Aishah Daiyan
  • Brandon Knox
  • Hana Rodriguez
  • Joshua Chasteen
  • Leon Seetoo
  • Roslyn Monterroza
  • Sandy Yuan
  • Thiti Das

Lina Raouf, Gregory Papadopoulos

Lina Raouf, Gregory Papadopoulos

  • Lab: Applied Micro-bioengineering Lab [Biomedical Engineering]
  • Faculty: Professor Weiqiang Chen
  • Mentor(s): Apratim Bajpai, Kate Luu

Abstract: 

Aging is a complicated process that leads to cellular degeneration and functional deterioration in living beings. One critical feature of aging-related change in cells is the alterations in cellular mechanobiology, which is the way cells use forces to maintain healthy function in vivo. This study aims at understanding the differences in the static and dynamic force generation capacities of primary aortic smooth muscle cells taken from young and old pathogen-free laboratory mice. Further, a study of the actin cytoskeleton is undertaken to understand the changes brought by the onset of age. The data generated in the study was accomplished through micropillar fabrication, cell subculture, immunostaining, cell fixation, and microcontact printing, and was analyzed using MATLAB, ImageJ, and Cellogram (an in-house image processing software). Our aim in the study is to develop methods for reversing/manipulating the mechanobiological aging process and thus, potentially discover possibilities for bioengineered treatments for age-related diseases in the future.

 

Susan Jimenez, Richlove Nkansah

Susan Jimenez, Richlove Nkansah

  • Lab: Dynamical Systems Lab (DSL) [Mechanical Engineering]
  • Faculty: Professor Maurizio Porfiri
  • Mentor(s): Mert Karakaya, Mohammad Tuqan

Abstract: 

Zebrafish, a species of freshwater fish, has been gaining popularity in pre-clinical research, with over 10,000 papers published every year, due to their similarity with humans in terms of genes and behavior. Specifically, zebrafish can be used to study the effects of different drugs by observing the changes in their behavior. In this project, we develop a free-swimming robotic fish that can interact with live zebrafish in behavioral studies. Our robotic fish consists of a 3D printed body, an array of sensors, and a microcontroller that drives a motor, actuating a compliant tail. Users can control the robot through an online platform. The robot will allow us to study social interactions in groups of swimming zebrafish.

 

Mariya Meleganich, Esther Bistricer

Mariya Meleganich, Esther Bistricer

  • Lab: Chunara Lab [Computer & Engineering Science]
  • Faculty: Professor Rumi Chunara
  • Mentor(s): Yuan Zhao

Abstract: 

COVID-19 has drastically impacted the mental health of individuals around the world. Among multiple implications of mental health, depression appears to be the most commonly experienced side effect among citizens in the United States and the world, which is the focus of our research. Women are especially affected by COVID-19, with 53% of women reporting a significant negative impact on their mental health, compared to only 37% of men in a 2020 Kaiser Family Foundation poll. More specifically, 83% of women compared to 36% of men are reporting a significant increase in depression, a mood disorder causing symptoms that affect how an individual feels, thinks, and carries out daily activities, as seen in a 2020 Total Brain survey. Factors that contribute to the uptick in mental health problems include age, gender, marital status, education, occupation, income, place of living, close contact with people with COVID-19, comorbid physical and mental health problems, exposure to COVID-19 related news and social media, coping styles, stigma, and psychosocial support, among others. To better understand the relationship between those factors and depression among women, we will use data from COVIDiSTRESS Global Survey, which is a global social science dataset consisting of participants from 39 countries and multiple domains of demographic and social-economic information and psychological measurement. We will use statistical models of exploratory analysis of mean and T-test, linear or logistic regression, and support vector machine (SVM) to identify key factors related to depression. 

 

Eric Gonzalez Cardona, Mia Montrose

Eric Cardona, Mia Montrose

  • Lab: Multifunctional Material Systems Laboratory [Mechanical Engineering]
  • Faculty: Professor Miguel Modestino
  • Mentor(s): Ricardo Mathison

Abstract: 

The introduction of electrosynthesis at industrial scale as an alternative of thermal processes represents an opportunity of using renewable electricity sources in chemical manufacturing plants, accelerating the decarbonization of large-scale chemical processes. Adiponitrile (ADN) is a large volume precursor of hexamethylenediamine, which is a monomer used in manufacturing of Nylon 6,6. The thermochemical hydrocyanation of 1,3 butadiene is the most commonly used method for ADN production, which requires large amounts of energy and the use of hydrogen cyanide, a highly toxic reactant. Alternatively, ADN can be produced electrochemically via the hydrodimerization of acrylonitrile (AN), making it the most successful and largest industrial organic electrochemical process. The mechanism of hydrodimerization of AN consists of several competing side reactions that lead to organic by-products. In order to further optimize the process by increasing the Faradaic efficiency towards ADN, a mechanistic model is developed describing the chemical and electrochemical reactions and diffusion in an aqueous liquid film surrounding the electrode. The mathematical model is used to identify the relationship between reactor design parameters and selectivity of ADN.

 

 


 Leon Seetoo

Leon Seetoo

  • Lab: Hybrid NanoLabs [Chemical & Biomolecular Engineering]
  • Faculty: Professor Ayaskanta Sahu
  • Mentor(s): Shlok Joseph Paul, Ingrid Paredes

Abstract: 

Solar photovoltaics are undoubtedly the first thing that come to mind when we think about sustainable renewable energy. Photovoltaics (PV) use light to generate electron-hole pairs, directly converting light energy into electrical energy which can be used in circuits for later use. PVs’ achieved mass production scales by integrating into the silicon supply chain where they have become more affordable with ubiquitous monocrystalline designs having an efficiency of 28% as of 2020. There are however alternate designs that claim to surpass the 28% efficiency by coupling PV’s with colloidal quantum dots. In this literature review we present some of these claims and elucidate how PVs with quantum dots might achieve cell efficiencies greater than 30% or even past the Shockley-Queisser limit.

Sandy Yuan

Sandy Yuan

 

  • Lab: Building Informatics and Visualization Lab [Civil & Urban Engineering]
  • Faculty: Professor Semiha Ergan
  • Mentor(s): Daniel Lu

Abstract: 

In a 2021 survey, the Associated General Contractors of America found that sixty percent of highway contractors reported motor vehicle accidents in their construction work zones during the past year. Current safety measures at work zones, such as blinking lights and audible alarms, are frequently false alarms and ignored by workers. There is a need for improving safety alarms based on a systematic understanding of how workers behave and react to different hazardous scenarios and alarm features (e.g., duration, frequency, modality). Previous research has developed a platform that integrates both virtual reality (VR) and wearable sensors to capture data regarding workers’ reactions towards safety alarms on a smartwatch application in a variety of work environments and hazardous scenarios. This project will work towards using experiment data from user studies conducted on the platform to develop a reinforcement learning model that can identify the optimal alarm features for maintaining worker attention and safety. However, in-person user studies are time-consuming and yield a limited amount of data relative to requirements for reinforcement learning models. To address this challenge, this project demonstrates a process by which synthetic data is generated from a simulated worker which behaves in VR environments and reacts to alarms based on recorded user studies data. This process can be utilized in model-based reinforcement learning approaches for developing safety alarm systems.

 

Joshua Chasteen, Hana Rodriguez

Joshua Chasteen, Hana Rodriguez

 

  • Lab: Automation and Intelligence for Civil Engineering (AI4CE) [Computer & Engineering Science]
  • Faculty: Professor Chen Feng
  • Mentor(s): Ruoyu Wang

Abstract: 

Soft bodies made from flexible and deformable materials are popular in many robotics applications, but their proprioceptive sensing has been a long-standing challenge. In other words, there has hardly been a method to measure and model the high-dimensional 3D shapes of soft bodies with internal sensors. We propose a framework to measure the high-resolution 3D shapes of soft bodies in real-time with embedded cameras. The cameras capture visual patterns inside a soft body, and a convolutional neural network (CNN) produces a latent code representing the deformation state, which can then be used to reconstruct the body's 3D shape using another neural network. We test the framework on various soft bodies, such as a Baymax-shaped toy, a latex balloon, and some soft robot fingers, and achieve real-time computation (≤2.5ms/frame) for robust shape estimation with high precision (≤1% relative error) and high resolution. We believe the method could be applied to soft robotics and human-robot interaction for proprioceptive shape sensing.

 

 

 

 

Aishah Daiyan, Thiti Das

Aishah Daiyan, Thiti Das

  • Lab: Hominin Skeletal Morphology Lab [Biology]
  • Faculty: Professor Scott Williams
  • Mentor(s): Madelynne Dudas, Monica Avilez

Abstract: 

All living hominoids (apes) share adaptations to suspensory locomotor behaviors.The ilium and sacrum (parts of the pelvis) share muscles with the torso and lower back that are important for suspensory behaviors. To better understand why apes (chimpanzees, gorillas, orangutans, and gibbons) evolved this form of locomotion, we need to investigate how the shape of the pelvis plays a role in suspension. We will quantify the bony morphology of the pelvis by applying 3D geometric morphometrics, a technique that places landmark points on 3D scans of the ilium and sacrum to capture the shape and size of this complex structure. We will apply multivariate analyses to these data to visualize shape variation among ape pelves. Our study will address questions like: are there differences in the shape of the ilium and posterior sacrum of suspensory apes? If yes, are these differences due to size, function, or phylogeny (relatedness)? We can then apply this technique to the fossil record to infer locomotor behavior in fossil primates.

 

Brandon Knox

Brandon Knox

  • Lab: Building Informatics and Visualization Lab [Civil & Urban Engineering]
  • Faculty: Professor Semiha Ergan
  • Mentor(s): Keundeok Park

Abstract: 

The construction industry has lacked the innovation and productivity growth compared to other industries, such as manufacturing, which has doubled productivity in the past few decades. Furthermore, the construction industry is currently facing problems such as the lack of skilled workers and rising construction costs with urbanization. To alleviate such problems, modular construction was introduced as a new business model based on the principle of fabricating building components off-site and assembling those components on-site. Off-site, a controlled factory environment can enable more opportunities for automation and optimization of design and production. However, despite such advantages, automation is still at its infancy. In this project, we aim to assist the design phase of modular construction using machine learning. We will extract building data from large-scale home design 3D datasets to train a Generative Adversarial Network (GAN). A 3D graphics tool and its Python API will be used for the data conversion. Through the use of 3D graphics tools, building objects can be converted into a combination of nodes and edges which represent topology and geometrical information of home design. Based upon such a graph network, we will train GAN to generate building design into industry standard building information model (BIM) formats.

 

 Roslyn Monterroza

Roslyn Monterroza

  • Lab: Hybrid NanoLabs [Chemical & Biomolecular Engineering]
  • Faculty: Professor Ayaskanta Sahu
  • Mentor(s): Shlok Joseph Paul, Ingrid Paredes

Abstract: 

This week New York saw the smog carried by the wildfires of the west. A red moon acted as the rallying cry toward an ever growing urgency on the path toward sustainable growth. One aspect of sustainability revolves around improving natural energy sources. To better understand this goal we look toward the research done on luminescent solar concentrators (LSC). These LSC’s are made up of embedded fluorophore dyes which fluoresce creating a glow that propagates via total internal reflection to the edge of a sheet where a narrow solar cell can absorb the concentrated light. Unfortunately current LSC’s have low optical efficiencies that are expected to benefit from coupling with quantum dots. In this review we hope to better understand and guide the reader through the methods, mechanisms and future for quantum dot coupled LSC’s.



Panel D

RSVP

Students in Panel D listed alphabetically:

Session 1

  • Giana Adote
  • Grady Yu
  • Jonathan Song
  • Luca Jones
  • MarcAnthony Williams
  • Trevor Lane
  • Xinyi Gao

Session 2

  • Areej Ahmad
  • Ayesa Siddeky
  • Christopher Amos
  • Clytie Qiu
  • Ilan Guerrero
  • Priscilla Ramchand
  • Ron Alweiss
  • Zuzanna Kowalski

Jonathan Song

Jonathon Song

  • Lab: BUILT [Civil and Urban Engineering]
  • Faculty: Professor Joseph Chow
  • Mentor(s): Bingqing Liu

Abstract: 

With the rise of smart cities, the emergence of new mobility services have driven changes in traveler's behavior as well as built environment policies. Agent-based simulation is effective in modeling systematic effects caused by the interaction between different types of mobility services. Agent-based simulation requires individual-level population data, which can be provided by synthesizing population from zonal demographic data. The first version of New York City (NYC) synthetic population generated by BUILT lab lacks dimensions that reveal the distribution of underrepresented population groups, which limits its use in equity evaluation. This study focuses on exploring NYC data on the distribution of people with disability, ethnic minority, and essential workers, using the data to re-synthesize the NYC synthetic population to be able to conduct equity evaluation through agent-based simulation.

 

Luca Jones, Grady Yu

Luca Jones, Grady Yu

  • Lab: ECE Machine Learning Lab [Mechanical Engineering]
  • Faculty: Professor Anna Choromanska
  • Mentor(s): Haoran Zhu

Abstract: 

We are going to replicate the research performed in the End to End Learning for Self-Driving Cars paper by building and training a convolutional neural network (CNN) to turn pixels from road images to steering directives. We will be using Udacity’s Self Driving Car Simulator to conduct our research. The data will be collected from different camera angles in the simulator. We will use PyTorch to build the system, and have two NVIDIA GTX 1080 GPUs available as computational power. The prior research aforementioned found that the CNN was more powerful than expected; the system was able to detect important road features such as road outlines based only on the input data. We predict similar results for our trial - the CNN should be able to perform tasks such as road outline detection and lane mapping without being optimized specifically for those tasks. Avoiding decomposition of this problem, as predicted by prior research, will allow for smaller networks and higher performance in the future.

 

Giana Adote, Xinyi Gilda Gao

Giana Adote, Xinyi Gao

  • Lab: Primate Hormones and Behavior [Anthropology]
  • Faculty: Professor James Higham
  • Mentor(s): Eve Cooper, Cassandra Turcotte, Amber Eliza Trujilo

Abstract: 

Sexual dimorphism is the phenomenon wherein different sexes within the same species exhibit physical shape differences related to sexual selection and, by extension, important aspects of living behavior. Previous studies have observed significant body size dimorphism and distinct patterns of somatic growth in male and female rhesus macaques. However, it is not yet clear whether skeletal elements exhibit the same pattern of dimorphism across ontogeny. Using 3D digital renderings of the crania and mandibles from the CPRC-NYU Rhesus Macaque Skeletal Collection, we will be exploring how the cranial skeleton of these primates changes in morphology both over the lifespan and between sexes. We plan to quantify the linear shape differences in the face and basicranium of 30 rhesus macaques (4 pre-prime, 6 prime and 5 post-prime age males and females) using cranial landmarks in Meshlab, an open-source virtual morphometric software. Patterns of growth and sexual dimorphism will be investigated using a combination of principal components analysis and ANOVAs comparing both age and sex groups. We predict that male and female individuals in the pre-prime group will not be distinguished on the basis of facial or basicranial morphology. These differences should only appear in the prime and post-prime individuals within the facial region, because the facial region is under greater pressure from sexual selection than the more highly conserved basicranium. Our results will have implications for understanding the ontogeny of cranial growth within rhesus macaques particularly in the context of sociality and sexual selection, and will inform our interpretation of primate dimorphism in a broader evolutionary context.

 

Trevor Lane

Trevor Lane

  • Lab: BUILT [Civil and Urban Engineering]
  • Faculty: Professor Joseph Chow
  • Mentor(s): Haggai Davis

 

Abstract: 

The MATSim (Multi-Agent Transport Simulation) virtual testbed created by C2SMART is used to simulate the entirety of New York City’s transportation network with a synthetic population to utilize the network. Previous iterations of the testbed have simulated passenger vehicles including cars, buses, taxis and more while the next iteration will include freight movements from trucks as well. The COVID-19 pandemic led to dramatic growth in demand for freight truck deliveries in New York City which has only contributed to the need to study and utilize the ubiquity of truck data to help policymakers plan and make decisions about freight movement in NYC. Using the truck data provided by the NYC DOT, as our validating data for our simulation results, and the existing freight data from NYMTC’s best practices model as our input for the MATSim simulation, the goal is to convert the origin-destination data into truck tour data and then validate this process with freight truck movements from the real world. Utilizing this data inside of MATSim, the project aims to represent a simulated truck driver’s full day of trips as one complete truck tour so we can gain a fuller understanding of how actual truck driver’s navigate and move through the city during a full day of work. We will be using the tool QGIS to interpret, visualize and study the truck data.

 

MarcAnthony Williams

MarcAnthony Williams

  • Lab: Machines in Motion Laboratory [Electrical Engineering]
  • Faculty: Professor Ludovic Righetti
  • Mentor(s): Avadesh Meduri, Julian Viereck

Abstract: 

Manipulating objects like humans has been a challenging problem in robotics. In this work, the students will learn the basics of robot manipulation through a series of tutorials. This will then equip them to implement a simple algorithm that will enable two 3 DOF finger robots to pick up a cube and move it. After verifying the algorithm in simulation, the mentors will run their code on the robot so that the students can see their algorithms in action on the real hardware. Through this project, the students will try to understand the difficulties in robot manipulation and gain experience with working on real robots.


Areej Ahmad, Ilan Guerrero

Areej Ahmad, Ilan Guerrero

  • Lab: The Chromosome Inheritance Lab [Biology]
  • Faculty: Professor Andreas Hockwagen
  • Mentor(s): Victor Leon, Carolyn Milano

Abstract: 

Our lab is investigating yeast meiosis to further understand eukaryotic chromosome inheritance. Meiosis is a type of cell division where precursor cells undergo two rounds of ploidy reduction, generating haploid gametes in diploid organisms. Proper meiotic division will grant faithful embryonic development in higher eukaryotes and will be responsible for the offspring’s phenotypes. To identify protein domains participating in meiosis, we utilized a classic reverse genetic screen using Ethyl Methanesulfonate (EMS) to introduce novel point mutations into Saccharomyces cerevisiae, commonly known as budding yeast. Single cells derived from these mutagenized yeast cultures were screened for known proteins involved in meiosis by complementation assays. Twenty yeast lines were identified to complement 5 major proteins in meiosis: Mek1, Dot1, Fpr3, Hop1, and finally RAD50. Sanger sequencing allowed us to identify specific point mutations within these yeast lines, pinpointing crucial domains within proteins important for meiotic function, and permitting design predictions on protein structure. Yeast is the optimal model organism for analyzing human genomes due to the strong conserved genetic sequences when compared to humans, along with being very easy to grow, facilitating our investigation on genetic inheritance. Controlling meiosis to separate, identify, and analyze the mutations in the yeast genome can be useful to identify causes of the shutdowns of major proteins, especially given that yeast have a small genome and share many similarities to human DNA. By understanding the molecular mechanisms governing meiosis, this research aids clinics to identify or potentially prevent birth defects or infertility among humans.

 

Zuzanna Kowalski

Zuzanna Kowalski

  • Lab: Montclare Lab [Chemical and Biomolecular Engineering]
  • Faculty: Professor Jin Montclare
  • Mentor(s): Dustin Britton

Abstract: 

Supramolecular protein assembly and protein-protein interactions in general often rely on electrostatic interactions. This is orthogonally dictated by the fickle protein-sequence-structure relationship where single mutations can impactfully change protein-protein interactions. Using a combination of macromolecular software packages Rosetta, PDB2PQR, and APBS Electrostatics, protein electrostatics can be modeled in silico saving valuable time screening the endless permutations of essential mutations required to search the sequence space. This method is applied to a protein previously synthesized in our lab, Q, that is capable of supramolecular assembly into mesofibers and even further into a fibrous gel. The results will inform the decision making in future designs of the Q protein for a greater degree of fiber assembly that has applications as a therapeutic, diagnostic, cell scaffolding.

 

Clytie Qiu

Clytie

  • Lab: Center for Advanced Technology in Telecommunications [Electrical Engineering]
  • Faculty: Professor Shivendra Panwar
  • Mentor(s): Fraida Fund, Ufuk Usubutun

Abstract: 

Our Internet usage has evolved rapidly in the recent past, and the minimum standards of network connectivity should, too. This is important because a connectivity standard that is too low will consider areas as "served" when consumers’ needs have not really been met. In my research, I explore four competing definitions of broadband connectivity: the old standard of 4/1 (4 Mbps download speed and 1 Mbps upload speed), the 10/1 standard used to determine eligibility for rural broadband deployment subsidies, the current 25/3 standard, and a recently proposed 100/100 standard. To help policy makers gain insight into the practical implications of each standard, I conduct a series of experiments measuring application performance, on networks that are characteristics of U.S. home broadband connections satisfying each of the four definitions.

 

Ron Alweiss

Ron Alweiss

 

  • Lab: Urban Modeling Group [Civil & Urban Engineering]
  • Faculty: Professor Debra Laefer
  • Mentor(s): Parth Singal, Kshitij Chandna, Karnik Panchal, Peter Gmelch

Abstract: 

Over the past year and a half, the COVID-19 pandemic has designated hospitals and Urgent Cares as hotspots. An excess of individuals flood into these facilities for COVID-19 testing and treatment as well as other illnesses more frequently than before this pandemic. The objective of this study is to determine the residence of healthcare workers and other essential workers at these medical centers during the height of the NYC pandemic in April 2020 and cross-reference that information with where people (non-healthcare workers) living in those areas work. This information will provide a more thorough understanding of which areas contain the most essential workers and observe the spread of COVID-19 based on the movement of these workers to and from three selected neighborhoods. The sample datasets used were from the Safegraph Consortium, focusing on three Brooklyn areas: Wyckoff Heights Medical Center in the Bushwick area, NYU Langone Brooklyn in the Sunset Park area, and Brooklyn Hospital in the Downtown Brooklyn area. The data breaks down into Census Block Groups (CBGs) to most effectively identify where these essential workers live and work. Another component of our analysis is observing the vulnerability indices of the identified census block groups, which relate to the average socioeconomic status, minority status, language barriers, housing composition, and housing type of each census tract (multiple CBGs make up a single census tract.) This information distinguishes the living status of the healthcare workers and notes how many essential workers live in socioeconomically worse places. This data is graphed on a scatterplot and mapped out in the ArcGIS software with various colorations to represent the COVID-19 rates, vulnerability indices, and the number of essential workers who visited businesses near one of the three health centers in each census tract. So far, the data projects that in Wyckoff, most healthcare workers who visited businesses live in high-vulnerability areas. Results about the data from Sunset Park and Downtown Brooklyn have yet to conclude.

 

Christopher Amos

Christopher Amos

  • Lab: Urban Modeling Group [Civil & Urban Engineering]
  • Faculty: Professor Debra Laefer
  • Mentor(s): Parth Singal, Kshitij Chandna, Karnik Panchal, Peter Gmelch

Abstract: 

Elmhurst, Queens was gravely impacted by Covid-19. The purpose of this project is to assess the vulnerability of the population and how that implicitly affected the high infection rate. We examined the spread of Covid-19 from March 2020 to March 2021, identifying the factors that contributed to the high-level infection rates in that area. These factors include behavioral patterns such as touch contact, footfall data at POIs, and mask usage; transportation, and influence of government policies and internal hospital mandates on infection rates.

 

Priscilla Ramchand

Priscilla Ramchand

  • Lab: Urban Modeling Group [Civil & Urban Engineering]
  • Faculty: Professor Debra Laefer
  • Mentor(s): Parth Singal, Kshitij Chandna, Karnik Panchal, Peter Gmelch

Abstract: 

Elmhurst, Queens was gravely impacted by Covid-19. The purpose of this project is to assess the vulnerability of the population and how that implicitly affected the high infection rate. We examined the spread of Covid-19 from March 2020 to March 2021, identifying the factors that contributed to the high-level infection rates in that area. These factors include behavioral patterns such as touch contact, footfall data at POIs, and mask usage; transportation, and influence of government policies and internal hospital mandates on infection rates.

 

Ayesa Siddeky

Ayesa Siddeky

 

  • Lab: Urban Modeling Group [Civil & Urban Engineering]
  • Faculty: Professor Debra Laefer
  • Mentor(s): Parth Singal, Kshitij Chandna, Karnik Panchal, Peter Gmelch

Abstract: 

Elmhurst, Queens was gravely impacted by Covid-19. The purpose of this project is to assess the vulnerability of the population and how that implicitly affected the high infection rate. We examined the spread of Covid-19 from March 2020 to March 2021, identifying the factors that contributed to the high-level infection rates in that area. These factors include behavioral patterns such as touch contact, footfall data at POIs, and mask usage; transportation, and influence of government policies and internal hospital mandates on infection rates.