Computer Science Program 2022 Projects

CS PROJECT DESCRIPTIONS

Artificial Intelligence and Data Science for Air Pollution Prediction and Visualization

Client: NASA & LA City
Advisor: Mohammad Pourhomayoun, Ph.D.
Students: Alejandro Alatorre, Emmanuel Cocom, Hector Rucobo, Aldo Ruiz Cruz, Vidal Zazueta

We are developing AI models and visualization frameworks to predict and visualize air pollution in Los Angeles (various air pollutants such as PM2.5, NO2, O3, ...). In this project, we use various machine learning algorithms and visualization frameworks, and advanced geographic information systems such as ArcGIS to visualize the results on the map. We will also develop web applications and mobile applications for the visualization framework.

Technology: AI, Data Science, machine learning, predictive models, visualization frameworks, web, and mobile apps.

Artificial Intelligence and Data Science for Climate Change Management
with Focus on Drought and Wildfire in California

Client: NASA & LA City
Advisor: Mohammad Pourhomayoun, Ph.D.
Students: Mazel Fernandez, Rayan Hyder, Victor Raj, Jennifer Serrano-Perez, Funing Yang

This software tool is a web application based on React and Node.JS API. The site will provide the necessary information and resources for being aware of wildfires in California. The web will consist of tweets, maps created by ArcGIS to visualize different weather forecasts in California, and other info and news.

Background: Wildfires have been significantly affecting the economy, health of citizens and animals, and destroying the ecosystem. Depending on the ecosystem, various practices, such as expanded housing developments and the utilities they require, fire exclusion/firefighting, and timber harvesting, in the past several decades have made it easier for fires to get out of control, according to the USFS. Also, in terms of the well-being of humans and animals, it has released toxic smoke into the air that pollutes the environment.  This toxic smoke has already caused air quality to plummet to dangerous levels in California, Oregon, and Washington in addition to toxic air; there’s also the debris and ash that covers charred neighborhoods. The ash is the aftermath of everything that was burned up, which includes anything in a home, including chemicals, plastic, and paint, all of which can be toxic to humans and animals. All these catastrophic events have led to a large economic toll on communities through property losses, decreased tourism, and even changes in the long-term structure of the local economy. The most fundamental objective of the product would be to effectively extract knowledge of climate change and wildfire data for users. And to establish a dashboard with data analytics and models that include data science to correlate the cause and effect of wildfire by gaining insights from the models.

Azure Cloud Database Migration & LACPD New Hire Enrollment Process Web Application

Client: Office of the Public Defender, LA County
Liaisons: Mohammed Al Rawi, Bridgette Bates, Gratia Dsouza, Mohammad Janjua
Advisor: Chengyu Sun, Ph.D.
Students: Albert Chen, Norberto Gomez Rosales, Ryan Lee, Michael Loria, Simon Mai, Shahram Mehri Kalantari, Wilfredo Paz, Joshua Perez, Fabio Quintana, Ismael Valenzuela

PowerApps: The Los Angeles County Public Defender's office is currently using two applications, PD-GO and Helpdesk. The database is stored in SharePoint lists. Tasked to migrate the database over to Microsoft Azure SQL database.

Enrollment App: Process of Enrollment and Approval System
Enrollment Web Application allows users to request custom forms generated via back-end communication with BOX and Adobe API, then saved within BOX after signing. Application is built in Angular that helps develop the web form that allows sending the user's data to the back-end system.

BOE Sidewalk Monitoring System

Client: City of Los Angeles, Department of Public Works, Bureau of Engineering
Liaisons: Ted Allen, Alisa Blake, Miguel Grajeda, Raul Virgen
Advisor: Jungsoo Lim, Ph.D.
Students: Aquil Alam, Alejandro Chanocua, Omar Eclicerio, Ernesto Garcia, Francisco Gastelum, Henry Gonzales, Gui He, Perla Ramirez, Rishi Shah, Daniel Zeng

The City of Los Angeles, Bureau of Engineering maintains over 11,000 miles of sidewalks. When a segment of sidewalk does not settle evenly or has been raised up by tree-root growth, the sidewalk becomes uneven. This can create pedestrian hazards.  In addition, the City is obligated to ensure that its sidewalks conform to Federal ADA standards, which limit the extent to which a sidewalk may slope.

ABSTRACT / DESCRIPTION: This is the fifth term of a multi-year project. In the last term, a rover has been successfully fabricated. Now, the rover is capable of 1) moving autonomously, 2) measuring crossing slopes and running slopes, 3) collecting GPS data, and 4) taking photo images. In this term, we will continue on developing various software that 1) lenders the photo images with slope data and GPS data, (in the office) 2) assists the field crew while they are collecting data in the field (i.e., User-friendly mobile app/web app). 3) develop a denoising raw data algorithm, 4) develop an algorithm to map field GPS data in Navigate LA GPS data and upload data into the back-end database server together with other denoised raw data, and 5) develop an interface to Navigate LA to visualize raw data on the Navigate LA map layer.

Box.com/eDefender Integration

Client: Santa Barbara Public Defender Office
Advisor: Jungsoo Lim, Ph.D.
Students: Joshua Cabrera, Marco De La Torre, Raul Gallegos, Daniel Guevara-Dominguez, Chuang Huang, Dang Le, Jessica Lopez De Leon, Shaocheng Shi, Sergio Tapia, Luke Williams

Currently, the Public Defender’s office has a content management system called eDefender, parent company Journal Technologies. The Department has transitioned to a fully paperless case management system. As the transition from paper to paperless continues, additional server space will be needed to host expanded growth and to maintain our digital client files.Box.com is a cloud content management tool that would allow the Public Defender to be compliant with CJIS/HIPAA requirements, have available storage to store all files digitally, have a collaborative platform, workflow automation, build API’s to eDefender and allow governance.

Currently, the Department has approximately 50+ terabytes of case data it needs to transition to the cloud; yearly consumption is currently twelve (12) terabytes and increasing exponentially every year.  The Public Defender plans to leverage Box.com to build the necessary infrastructure and support our network needs (approximately 100 terabytes total). The project will migrate the entire infrastructure for the Public Defender’s office to the cloud with Box.com.  We would like to build in automation including integrating with AWS and Microsoft Cognitive skills to build facial recognition and transcription (similar to LA County Public Defender project).  Once files are uploaded, they would be translated/transcribed while integrated with eDefender to trigger notifications, tasks, and alerts upon transfer.

Preliminary Project Scope: Migrate approximately 80 Terabytes of data into Public Defender Box.com CJIS compliant instance. Currently stored in networked storage, Linux on-prem server, and SB County Box.com instance.

• Upon upload, documents/files are transcribed/translated. Integrate with AWS/Microsoft Cognitive skills (similar to the LA County project).
• Upon upload, eDefender can ‘see’ the item being uploaded and based on the folder, creates a task for the attorney to review the information.
• When a new case is generated in eDefender, a new case file folder (along with all sub-folders) is created in Box, and the link is established within eDefender.  Folders would be based on Department naming convention procedures, creating root folders per case type.
• When the case type is updated in eDefender, Box will move the case file folder to the correct root folder (the new case type) and update the link in the system.
• When transitioning from the County instance of Box to the PD instance of Box, existing links need to be ‘updated’ so they still work for the recipients. (This includes the links we are creating for closed eDefender cases and the links sent to experts/conflict counsel)
• Once in the PD instance of Box, update the security on all of the links contained in the closed eDefender file root folder to be internal access only.
• Right now, they are public access (due to licensing limitations on the County instance); once all employees have a license, the link security needs to be updated.
• Correlate specific document templates within eDefender to specific sub-folders within Box.  Auto-generated and self-generated documents within eDefender will automatically save to the correct sub-folder in Box.
• When sending documents/discoveries to external collaborators, establish a digital record of communication within eDefender.
• Upon discovery upload into Box.com, sending digital confirmation to the District attorney’s office.
• API Documentation: eDefender API documentation is attached, titled REST + API.

Box.com API information can be found on their website by clicking here.

Language within eDefender: Within eDefender, have a skillset with SQL, Groovy, Velocity, and Jasper Reports.  Anything needed with those languages would most likely be handled by our internal IT staff.  Our office IT staff can provide training on the system along with the features within eDefender.

• Business rules in our system are written in Groovy.
• Velocity is used in conjunction with HTM for display and search option features.
• Jasper report is the reporting service compatible with our version of eDefender.  We can either use this or SQL reporting services to extract data from the system.  The ‘compatibility’ of Jasper Reports refers to the ability to execute the reports through eDefender.

Comorbidity and its Impact on Patients with COVID-19

Client: Vodafone Group
Liaison: Haley Kirk
Advisor: Navid Amini, PH.D.
Students: Daniel Esparza, Canhong Huang, Jonathan Kan, Abran Lezama Pastor, Brenden Mccabe, Ashley Munoz, Uchenna Onuigbo, Duy Pham, Nicolas Sandoval, Jacob Schultz

Our goal in this project is to determine clinical factors of a poor prognosis in patients with COVID-19 infection. We will examine medical records of patients with COVID-19 infections confirmed by polymerase chain reaction (PCR). Logistic multivariate regression models adjusted for age and sex will be constructed to analyze independent predictive factors associated with death, ICU admission, and hospitalization of these patients. As part of the project, we are looking at the link between COVID-19 severity (e.g., COVID-19-related death) and the blood type of patients. This topic has been a controversial topic in the research community. 

Update (2/2/2022): Over the past couple of weeks, our team has been experimenting with various types of software including WEKA, Tableau, and D3/JS. WEKA is software directed towards data mining using its extensive amount of machine learning algorithms. We spent some time learning how to import and analyze specific outcomes, probabilities, and visualizations using various data sets. We also used some open-source resources on COVID-19 provided by Google, Amazon, New York Times, and the US Government. Using relevant data allowed us to go in-depth when making our observations. We are currently using Tableau and D3/JS to provide us with detailed visualizations that may help us with our next breakthrough discovery.

Update (4/10/2022): Last semester we successfully created two COVID-19 Dashboards both made with different software and implementations. We also analyzed confidential datasets provided by Vodafone using different machine learning algorithms to come up with some of our own conclusions. This semester we plan on adding more visuals to our portals as well as some optimization changes for a better user experience. We still have a second confidential dataset that we are currently looking into with specified blood types which could make for interesting outcomes.

Helix

Client: QTC
Advisor: Keenan Knaur, Ph.D.
Students: Abdullah Alwabel, Norman Avery, Dean Baquir, Daniel Ibanez, Sameen Khan, Noe Lopez, Jeffrey Lum, Josh Mermelstein, Jose Mierzejewski, Wilson Tobar

The main purpose of our project is to update the current document repository that is used for medical examination data. This repository is used by different types of businesses and organizations, such as government agencies and medical professionals. Along with the repository, we will be updating the online application that gives access to the medical examination data and implements different features that are respective to the business needs. By updating these components, we look forward to taking care of current issues that arise in the current software due to the use of old technologies.

Project Update (1/31/2022): Over the past months, the team has been researching and experimenting with Microsoft .NET 5.0 and .NET 6.0. This has been highly dependent on the team getting up to speed in learning the basics of the C# programming language. By doing so, the team has been able to research and implement technologies that are part of the .NET framework. These technologies include the use of a custom assembly framework, dependency injection, and the use of an entity framework to be used for database access.  The team so far has built a working .NET 5 and .NET 6 application that incorporates business logic as well as QTC's requirements for both frontend and backend.

Pelvic Image Analysis and Geometry Reconstruction Using Artificial Intelligence

Client: Mathias Brieu, Ph.D.
Advisor: Negin Forouzesh, Ph.D.
Students: Nicol Barrios, Ralph Belleca, Ted Kim, Silvano Medina, Robin Mok, Alejandra Olvera, Demetrius Parker, Sabino Ramirez, Mary Semerdjian, Jason Tejada

Problem: Medical imaging is a technique of producing images of the interior of the body non-invasively and plays an essential role in allowing medical professionals to provide accurate information about a patient’s anatomy, especially when dealing with tumors. The output image, i.e., the MRI image, is then processed to be more useful to medical doctors using image segmentation and 3-D model construction. 

Although image segmentation has been through phases of improvement over the last decade, this procedure of segmenting images and creating 3-D models of specific organs can still be tedious and repetitive. For some situations, although very rare, segmentation can require hours for a single case.

Task: The overarching goal of this research project is to streamline the process of converting MRI images of pelvic organs into 3-D model objects. The project was separated into two learning stages: understanding basic 3-D model construction and the creation of an AI model. 

The goal for the Fall 2021 semester was to learn the basics of 3-D modeling using 3DSlicer, 3-D visualization software, and experimenting with Nvidia AIAA, an API (application programming interface) that allows users to conveniently create 3-D model objects using trained data to automate it. With the results obtained in the previous semester, the Spring 2022 semester focused on creating our own AI model trained using our own data– which are MRI images of pelvic organs.  

Current Progress: The team is currently in the process of training their own AI model using Clara Train SDK, a framework used by Nvidia in training their AI models for different organs.

Immersive Storytelling with Engaging Physical Actions

Client: INART
Advisor: David Krum
Students: Joseph Chong, Jimmy Hernandez, Edwin Hernandez, Jaquan Jones, Alberto Landeros, Tony Lee, Jennelle Maximo, Eduardo Meza, Dean Nguyen, Anthony Viramontes

Immersive experiences, such as those created by virtual and augmented reality, represent a new medium for audiences and for storytellers. Storytellers are still developing an understanding of what is possible and what is effective in a good immersive storytelling experience. Immersive experiences differ from past media, like written stories, movies, or even video games, which may be the most closely related. It can be more difficult to move the audience through a narrative arc in an immersive experience. Immersive experiences give more control to the audience over their own embodiment, allowing them to choose where to look and where to go. Immersive experiences can be measured in participation and action rather than the paragraphs and story beats of a written story. New methods of discovering and interacting with a story in an immersive setting are needed.

To help advance storytelling techniques in immersive environments, a student team will develop an immersive story experience in which participants engage with the story through gestures and physical actions that represent actions, i.e., miming and other gestures standing in for interactions. The team will explore how a set of story-compatible gestures can be designed, taught to a user, and recognized by an immersive system. The team will work to evaluate prototypes through user testing and iteratively improve the experience. Benchmarks for success could include measures of presence, memorability, learning, and enjoyment. The story should also communicate a point of view, a lesson, or illustrate an idea or concept.

Operationalize Networked Collaboration Features for Moon Trek

Client: Jet Propulsion Laboratory (JPL)
Advisor: David Krum
Students: Sean Chung, Aldo Gil I, Tommy Lay, Allen Marquez, Tam Nguyen, Alex Sahakian, Andy Tsan, Srivats Venkataraman, Jian Wu, Anna Yesayan

Solar System Treks is a JPL web portal that provides NASA data from various planets, asteroids, and moons. This data is used by scientists, research partners, and the public, including students and teachers. The data includes satellite photography and 3D terrain models.  Collaborative visualization, especially in virtual reality (VR) and augmented reality (AR), can be helpful in generating new insights, as a team of researchers or students can examine the data together and collectively generate and discuss new ideas and hypotheses. Additionally, with the need for remote work, remote learning, and social distancing due to COVID-19 precautions, remote collaboration has become increasingly important.

Solar System Treks Web Portal, Solar System Treks Information

During the 2020-2021 year, a student team created extensions to Solar System Treks that demonstrated web-based collaboration. The goal of the 2021-2022 team is to bring these collaboration features to deployment on the Solar System Treks web portal.

Collaborative Visualization for Solar System Treks project page

Power BI Data Analytics Dashboard

Client: Office of the Public Defender, Santa Barbara County
Advisor: Chengyu Sun, Ph.D.
Students: Yara Ajjawi, Markniel Cruz, Jevon Fan, Suresh Ghimire, Jiajun Gu, DeQing Liang, Elton Lin, Edwin Lugo Bautista, Tommy Ly, Winston Pham, Alvin Truong

Power BI is a business analytics service provided by Microsoft that lets you visualize your data and share insights. It converts data from different sources to build interactive dashboards and Business Intelligence reports. The Power BI software is part of the Microsoft Power Platform. Power BI provides cloud-based business intelligence services, known as "Power BI Services", along with a desktop-based interface, called "Power BI Desktop" It offers data warehouse capabilities including data preparation, data discovery, and interactive dashboards. Power BI’s interactive dashboards allow users to manipulate the data in real-time using filters to allow certain data to show while hiding others. More detail can be found on Microsoft's Power BI website.

Our project revolves around closely collaborating with the Santa Barbara County Public Defender team. We hold weekly meetings and communicate via email to determine what visualizations will fit the SBCPD team’s needs. After discussions during meetings, the student team will create the dashboards with their requested features and data. If the SBCPD team is unsatisfied with the dashboard or wants something modified, the student team accommodates.

RoboSub

Client: Civil Engineering Department and Mechanical Engineering Department, Cal State LA
Liaison: Mark Tufenkjian, Ph.D.
Advisor: Richard Cross https://www.linkedin.com/in/richard-k-j-cross-8ab65431/
Students: Ashkan Aledavoud, Leslie Araujo, Christian Castillo, Alan Chan, Jeanie Jeon, Robin Romero, Bryan Sanchez, Carl Christian Dokken Solli, Edwin Tran, Daniel Valadez

The RoboSub Senior Design project is a joint project between a team of Electrical/Mechanical engineering students and a team of Computer Science students. The purpose of the project is to design, build, and program a fully autonomous underwater vehicle (AUV) that will compete in the international RoboSub competition hosted by Robonation. The computer science team will design and implement the software that pilots the Autonomous Underwater Vehicle (AUV), providing navigation, stabilization, object-detection, and task-handling capabilities. This software will provide the AUV the ability to detect objects and obstacles using image recognition, and then respond to this information to maneuver itself and interact with its environment to meet the objectives outlined in the competition's rules. A team composed of Mechanical and Electrical engineering students will be designing and building the AUV intended for use in the competition.

The International RoboSub competition is hosted by Robonation and is held annually at the end of July at the US Navy's Transducer Evaluation Center (TRANSDEC) in San Diego. The competition normally consists of a set of obstacles arranged in the TRANSDEC pool that each team's submarine must navigate autonomously. This year the competition will be held July 25-31, 2022.

Satellite Anomaly Injection & Detection (SAID) Testbed

Client: The Aerospace Corporation
Advisor: Zilong Ye, Ph.D.
Students: Martha Caldera, Diana Degiacomo, Gabriel Kutasi, Jae Lee, Michael Morris, Gustavo Torres, Tomas Velarde, Dearo Yam, Rafael Zaragoza

Satellites perform a host of vital functions, including communications, weather prediction, geolocation, defense, and many others. In these complicated systems, it is extremely important that accurate data flows freely between the ground and the satellite via uplinks and downlinks. When strange behaviors or anomalies occur, it is vital that the errors be identified and corrected before a disaster occurs. Sometimes these anomalies are the result of errors in the hardware or software, issues introduced by the environment, or an attack by a hacker. Effective anomaly detection techniques can help identify problems on the vehicle before they happen, which can help improve mission success.

The operation of satellites in long-term term operation is affected by many uncertain factors. Anomaly detection based on telemetry data is a critical satellite health monitoring task that is important for identifying unusual or unexpected events. The use of simulation tools allows users to configure and deploy platforms to be used in real-time environments as well as simulate any anomalies that can take place. Machine learning can be used to detect these anomalies by comparing actual observed values with the predicted intervals of telemetry data. Simulation tools can be utilized by students to develop a way to solve these complex problems using applications already being used in the industry.

For this project, we have developed software components to integrate with and utilize existing industry open-source software components to perform the tasks outlined below:

• Generate satellite simulation data
• Inject anomalous scenarios into the flight system
• Apply techniques for detecting the anomalies on board and the ground
• Outcomes from the project:
• Software source to develop anomaly injection and detection capabilities
• Detailed documentation on design, implementation, tests, and results from each of the anomaly scenarios
• User manual to set up, configure, and run the OSK with the anomaly injection and detection capabilities
• Monthly review meetings with Aerospace liaisons and final out brief to Aerospace engineers

Telescope AR

Client: Jet Propulsion Laboratory (JPL)
Advisor: Weronika Cwir
Students: Niloy Azad, Fu-Cheng Chuang, Eduardo Cruz, Jingchao Feng, Byron Garibay, Daniel Gonzalez, Tony Hong, Matthew Johnson, Cindel Lopez-Sianez, Jonathan Navarrete

Jet Propulsion Laboratory (JPL) is a federally funded research facility and development center managed for NASA by Caltech that carries out robotic space and Earth science missions. JPL has conducted robotic missions to study all the planets in the solar system as well as asteroids, comets, and the Earth's moon. Today JPL continues its world-leading innovation, implementing programs in planetary exploration, Earth science, space-based astronomy, and technology development.

Through the missions involving collecting data from Earth's moon JPL has created Moon Trek, a mapping and modeling portal. The site contains high-resolution data sets covering most of the Moon. These include imagery and digital elevation models - but also hundreds of layers of spectrometry, radiometry, gravity fields, radar, slope, roughness, mineralogy. Additionally, there are tools that allow for the analysis and rendering of derived datasets.

JPL is partnering with California State University - Los Angeles, College of Engineering, Computer Science, and Technology to build an interface between Moon Trek and telescopes amateur astronomers use to look at the Moon. When images from the telescope are routed to a laptop or a smartphone, they will be annotated with names of lunar features and landmarks, local temperature, the chemical makeup of the soil -- or any available information the astronomer chooses.

Project Description: Build upon the previous Moon Trek Django-web application where the user shall capture an image of the Moon from their own telescope or shall upload a corresponding image of the Moon into the web application.

The user shall give the image to the web application; then, in turn, the web application provides points of interest on the Moon, such as craters, maria, and landing sites. The Augmented Reality portion of the project lies in improving the data overlays, creating a 3D model of the Moon created by Jet Propulsion Lab's high-quality images of the Moon and user-uploaded images of the Moon.

The team's objective is also to complete the telescope to computer communication, improve the accuracy of the image registration, and create a 3D model of the Sun, Moon, and Earth based on the images uploaded. The same 3D model of Earth will also show an annotation of where the picture was taken and the time it was taken.

The ArQive: LGBTQ Storytelling App - Expanding Mobile and Gamification

Client: The ArQive
Advisor: John Hurley
Students: Matthew Frias, Jesus Gonzalez Perez, Karen Kazaryan, Daniel Lee, Stewart McKenzie, Erica Payne, Erica Santos, Leslie Segovia, Bryan Sosa, Elio Vences

The ArQive, formerly known as GlobaltraQs, is a web and mobile application that allows anyone to post fun and interesting LGBTQ+ oriented stories, events, and other information that they find meaningful. Founded in 2014 by Dr. Cynthia Wang and Zachary Vernon subsequently joined as a cosponsor. The ArQive gives users a safe platform where they can share personal, historical, and community stories, as well as have access to information about safe spaces, which all serve as valuable resources to members of the LGBTQ+ community. By providing users with the ability to place pins on the map indicating where they have been and the experiences they have lived, The ArQive gives people the ability to mark their place in the world and in history.

This year's project for the ArQive will focus on adding additional features to the website and mobile apps, creating an automated content moderation feature, adding further security measures for users who choose anonymity, adding a way to upload images and other media, beginning an AR/VR feature, and refining the appearance and performance of The ArQive.

Want 2 Remember

Client: We2Link
Advisor: Zilong Ye, Ph.D.
Students: Antonio Campos, Amy Guttman, Alec Kaczmarek, Vincent Li, Saiyang Liu, Ricardo Marroquin, Miguel Nonoal-Garcia, Alexandra Strong, Jonathan Sum, Edwin Zapata Minero

Background: We are working with We2Link to develop a mobile application called Want2Remember. Michael C. Malone founded We2Link after he suffered from a traumatic brain injury (TBI) while serving in the U.S. Army. He slowly discovered he was having issues with learning new things, recalling names, and even remembering faces. The CDC estimates that 1.5 million Americans sustain a TBI each year; an estimated 5.3 million Americans are living with a permanent TBI-related disability. Memory impairments, whether from TBIs or from progressive memory issues such as Alzheimer's or dementia, puts a strain not only on the individual but also on their social and familial relationships. This app helps to support those with memory impairments as well as their caregivers by providing a simple user interface to log the user’s memories and other important information.

The Application: Want2Remember uses the React Native framework to provide the user with basic templates to log their memories, passwords, to-do lists, medications, appointments, interactions, and other important notes. Through continued discussions with our users and their caregivers, we are continuing to refine the features of this app. Our overarching goals for this app are to allow for independent living, reduce social isolation, improve personal safety, reduce caregiver burden, allow for a return to work, and improve medical support and advocacy.