Capstone Project List

Below is a list of the projects for the Spring 2025 semester.

Community Projects

Pitt Street Medicine Team

Pitt Street Medicine Go Service Web Application

Objective: The objective of this capstone project is to develop a multimodal accessible website application for individuals served by Street Medicine at Pitt, including those at the Second Avenue Commons Clinic. The project’s focus is to allow individuals to request/order necessities, including food, beverages, clothing, toiletries, basic medical supplies, and other supplies as needed, in addition to Street Medicine at Pitt’s weekly Wednesday evening “street rounds,” and make modifications as appropriate based on client feedback.

Background: Street Medicine at Pitt is a student-run interdisciplinary organization dedicated to providing healthcare and social support to the rough-sleeping and unhoused communities in Pittsburgh. The organization is affiliated with the University of Pittsburgh School of Medicine, which recently established the Center for Street Medicine. Street Medicine at Pitt partners with the Second Avenue Commons Clinic, which provides shelter access, medical services, and engagement centers for adults experiencing homelessness in Pittsburgh.

More information: https://www.streetmedatpitt.org/home

Project Goals:

Priority Areas:

  • Improve round schedule algorithms - assignment system to be more scale-based (team leader, volunteer, clinician, etc. and assign those volunteers to the rounds)
  • Enhance order distribution system - order-to-volunteer matching based on location and availability
  • Enhanced Mobile Experience
  • Attend 2-3 Wednesday weekly evening (6:30-9 PM) “street rounds” with the Street Medicine at Pitt team

Secondary Areas:

  • Develop a traffic analysis tool to track the performance, moderation, and utility of the website application
  • Develop an inventory analysis and visualization tool to track weekly and monthly inventory usage
  • UI/UX improvement

Outcome for Students:

  • Full-Stack Development Experience
  • Learning SDLC while using Agile Methodology
  • Cloud architecture & DevOps

Technologies:

  • Springboot, Java
  • Github
  • React, JS
  • MySQL, MongoDB

Team Size: 3-4 students

POC: Victoria Bacchi (VMB35@pitt.edu), Nick Cao (RUC36@pitt.edu), Anna Marie White (whitea3@upmc.edu), Helena Clara maria Oft (Oft.Helena@medstudent.pitt.edu), Max Hurwitz (MBH39@pitt.edu)

Pittsburgh Policy Initiative

Pittsburgh Food Access Map

Objective: The objective of this food access map project is to revitalize and expand an interactive Pittsburgh Food Access Map. The tool will enable users—including Pitt students, community members, and nonprofits—to locate food resources throughout the city, such as grocery stores, corner stores, farmers’ markets, food pantries, and mutual aid sites. The initial phase of the project will prioritize a streamlined, functional version of the platform with a core set of essential features. This ensures the tools is accessible, reliable, and useful from the outset, while leaving room for additional features to be incorporated as capacity allows. In this way, the updated map will serve as both a community resource to improve food security and a research tool to inform advocacy and policymaking.

Background: Food insecurity is a persistent challenge in Pittsburgh, with many neighborhoods lacking consistent access to affordable, healthy food. In 2019, Code for Good Pittsburgh developed a food access map that helped visualize food deserts and the availability of resources. However, the tool is no longer maintained, and much of the information is outdated.

The Pittsburgh Policy Initiative (PPI), Pitt’s premier student-run think tank, is working to bring the map back to life as part of its Food Security project team. The project will merge technical development with community engagement, creating a sustainable and user-friendly platform that supports both everyday residents and local policy discussions around food security.

More information:

Project Goals and Deliverables:

Priority Areas:

  1. Platform Redevelopment: a. Build a new version of the food access map (web/mobile hybrid) from scratch b. Incorporate an interactive interface that allows users to filter by food source type, neighborhood, hours of operation, etc. c. Ensure accessibility features (mobile-friendly, clear icons, color contrast)
  2. Data Integration and Management a. Collect and integrate datasets from local sources, avoiding web-scraping if possible b. Develop a pipeline or backend that allows for periodic, streamlined data updates c. Create an admin dashboard for PPI or community partners to add/edit food source information
  3. Sustainability and Documentation a. Provide clear technical documentation for future student developers b. Create a user manual for non-technical community partners c. Plan for long-term hosting and stewardship so the platform can remain active beyond the initial development phase

Secondary Areas:

  1. Explore integrating APIs (e.g., Google Maps API for routing and directions)
  2. Include a reporting feature that allows community members to submit corrections or new food resources, with a review process continued by PPI to ensure accuracy and reliability
  3. Gather feedback from community members and stakeholders to refine the tool and ensure it provides value beyond existing platforms.

Team Size: 2-4 students

POC: Shanthi Bhaskar (shb175@pitt.edu), Riya Lakhe (ril31@pitt.edu), Andrew Pruden (asp161@pitt.edu), Maddox Zimmerman (mal665@pitt.edu)

Borges Center

Borges Site Migration

Summary: The Borges Center’s website must be migrated from Drupal to OpenScholar. The primary goal is to maintain the site’s existing design, structure, and functionality between platforms to the largest extent possible. Technical skills required for this project include knowledge of Content Management Systems (CMS) such as Drupal, as well as web development expertise in PHP, HTML, CSS, and JavaScript. Prior experience with payment gateway services such as Authorize.net is also needed to facilitate the transition of the shopping cart feature.

There are some unique challenges to this project, ones that most developers will encounter somewhere during their professional careers. The first challenge is that the project team will not have any documentation about the existing functionality. Further, the person(s) who built the site no longer work for the University and therefore are not available to answer questions about the site; executive decisions will need to be made that allow the site to continue to operate as faithfully as possible to the existing functionality and using the current design. This project also requires the team to ensure a seamless cutover between platforms. Lastly, the site will need to be developed in such a way that there is no planned, manual intervention required.

Project Team

The ideal team for this project is 3-4 students.

  • Systems Analyst
  • Web Developer/Programmers (1-2)
  • Data Migration Specialist

Required Skills

  • PHP
  • Javascript/Ajax/jQuery
  • HTML
  • CSS
  • SQL

Team Size: 3-4 students

POC: Amy Lynn Wildman (amywildman@pitt.edu)

Corporate Projects

CGI

Note: CGI projects will need an NDA.

CGI Web App

Project Description:

The goal of this project is for the students to build the foundation for a web application, implementing Spring API framework and Angular UI, to be used as an internal use case by CGI employees. This initiative will expose students to Agile, Scrum-based software development framework with weekly Stand Ups and regular Tech Reviews with the client. Students will be responsible for preparing specified deliverables on a 2-week release cycle following industry standards. Developing in Java, the students will create a functioning SpringBoot API utilizing built-in H2 service for database administration. Collaboration with CGI will be through weekly ceremonies, Trello and Microsoft Teams, and source control will be handled using GitHub.

Outcome for Students:

  • Learning SDLC while using Agile Methodologies
  • Utilizing 2-week sprints and introductions to Agile ceremonies
  • Using project management software to track progress of work through the semester
  • Separating application flows depending on user access
  • Live Demo of the working product to CGI leadership

Technologies:

  • SpringBoot (written in Java) controls the backend of the application
  • Angular (written with HTML, CSS, and TypeScript) for JSON data collection and frontend development
  • MySQL for database creation and management
  • IntelliJ and Virtual Studio Code to edit backend and frontend code, respectively
  • Maven for package dependency

Team Size: 4 students

POC: Anastasia Mokhon (anastasia.mokhon@cgi.com).

RAG Chatbot

We propose developing a Retrieval-Augmented Generation (RAG) chatbot that answers user queries using information retrieved from a curated knowledge base. Instead of relying solely on a language model, the chatbot retrieves relevant passages from uploaded documents (e.g., policies, FAQs, financial guides) and uses them to generate fact-based, cited answers. This approach improves accuracy and transparency compared to a plain LLM.

Key Components & Technologies

  • Document Ingestion: Students parse PDFs/HTML/DOCX, split into manageable chunks, and store them with metadata.
  • Vector Search (Retrieval): Using an open-source vector database like FAISS, Chroma,
  • Generation: An LLM (OpenAI, Anthropic, or open source like LLaMA via Ollama/vLLM) generates answers grounded in retrieved text.
  • APIs & UI: A simple backend (FastAPI) with a lightweight UI (Streamlit or React) for users to interact with the chatbot and view source citations.
  • Optional Enhancements: Reranking for better retrieval, PII redaction (e.g., Presidio), evaluation metrics (groundedness/accuracy), and feedback capture.

Overall Student Benefits

  • Hands-on experience with modern NLP pipelines (embeddings, retrieval, prompt engineering).
  • Exposure to vector databases and information retrieval.
  • Integration skills: backend APIs, simple frontend, and connecting multiple components.
  • Emphasis on responsible AI (fact-grounding, citation, guardrails).
  • Cloud-independent; runs locally with Docker, enabling cost-free experimentation.

Team Size: 4 students

POC: Sandeep Pamarthi (sandeep.pamarthi@cgi.com)

Faculty Projects

Benjamin Rottman

Using LiDAR Generated Maps of Homes for Recommending Ways to Reduce Carbon Footprint

Homes are responsible for about 20% of energy-related greenhouse gas emissions, primarily through heating and cooling needs. Our general goal is to build a smartphone app that helps homeowners decide among different ways to reduce the energy consumption of the home (e.g., installing insulation in a wall, replacing the AC with a more efficient model). Newer iPhones have LiDAR built in, which allows for apps (e.g., polecat, magic plan, roomscan, Canvas: LiDAR 3D Measurements) that can make 3D floorpans.

Goal 1:

Import and integrate the 3D floor plan into our app. This will allow users of the app to do things like name rooms in ways that make sense to them, and link other functions of the app to specific features (e.g., a to-do list that knows what retrofits to do to specific windows).

Goal 2:

Calculate various feature of the home such as square footage, volume, area of walls. These are important for making decisions about retrofitting the home, for example, knowing the size of AC that is needed if the AC is going to be updated, or estimating the cost of adding insulation to walls.

Stretch Goal:

Ideally, we would also like to be able to automatically detect features such as windows, doors, floors. For example, knowing how many floors the home has and the total square area of windows on each side of the house are also very important for providing more specific estimates of heating and cooling needs and recommendations for improvements.

Technology used:

  • Server-side post-processing of LiDAR, RGB data, and 3D maps, which could be done in many languages, likely Python
  • Flutter for integrating the 3D maps into our app

Team Size: 2-4 students

POC: Ben Rottman (rottman@pitt.edu), Professor of Psychology at Pitt; Stephen Lee, Assistant Professor of Computer Science (Stephen.lee@pitt.edu)

Longfei Shangguan

Egocentric Gait Monitoring for Fall Risk Prediction in Older Adults

Project Introduction: Falls are one of the leading causes of injury, loss of independence, and even death among older adults. Timely identification of abnormal gait patterns or mobility decline is critical for preventing falls. However, existing solutions—such as in-home surveillance cameras or phone-based tracking—are often limited by location, compliance, or privacy concerns. In this project, we aim to develop a novel, long-term gait monitoring system using Meta’s Aria smart glasses to passively and continuously capture first-person data from the wearer’s perspective.

Why Egocentric Vision?

Egocentric (first-person) monitoring offers unique advantages over conventional fixed-camera or handheld methods. Unlike smartphones, which require active operation and can be easily forgotten or misplaced, smart glasses are worn passively and naturally integrate into daily routines. Unlike wall-mounted cameras, which are limited to fixed locations and often unavailable in the homes of independently living seniors, wearable smart glasses move with the user—enabling context-aware, room-independent gait analysis during real-world activities. This makes egocentric vision particularly valuable for long-term monitoring in diverse environments, including the homes of seniors who live alone.

Research Goals and Impact

By collecting and analyzing synchronized video, pose, and motion data through Aria’s advanced sensor suite (including IMU and SLAM), we aim to identify subtle gait changes that precede falls. The ultimate goal is to build predictive models that can raise early warnings and assist caregivers or clinicians in timely interventions. The broader impact lies in enabling scalable, privacy-conscious, and ecologically valid fall-risk assessment—paving the way for safer aging-in-place for millions of older adults.

Team Size: 2

POC: Longfei Shangguan (LONGFEI@pitt.edu)

Optimizing Model Serving for LLM Agents

Modern LLM Agents like ChatGPT and Claude can perform complex tasks by making function calls—pausing to search the web, access tools, or solve problems before continuing their response. However, current systems handle these interruptions inefficiently, leading to slower response times and wasted computational resources. This project aims to develop novel approaches for managing these interruptions more effectively, creating a more responsive and efficient user experience.

Our research focuses on three key challenges: (1) improving memory management during function calls, (2) developing intelligent scheduling algorithms that prioritize requests based on urgency and complexity, and (3) creating prediction models to estimate execution time for different request types. You will contribute to implementing and testing various memory management strategies, analyzing performance across different workloads, and helping design user-focused metrics to evaluate system improvements.

As AI assistants become increasingly integrated into daily life, the efficiency of these systems directly impacts millions of users. Our work has the potential to significantly reduce latency, improve throughput, and create more natural interactions with AI systems. We welcome motivated students with programming experience and interest in AI systems, regardless of prior exposure to large language models. This opportunity provides an excellent foundation for students interested in graduate studies or industry careers in AI systems, high-performance computing, or human-computer interaction.

Team Size: 2

POC: Longfei Shangguan (LONGFEI@pitt.edu)

Heart Rate monitoring using ANC Earphones

Our research is dedicated to developing a heart rate monitoring system based on ANC headphones. This system captures blood flow pulses on the ear canal wall using ultrasonic signals and combines gyroscope data from the headphones to eliminate motion artifacts caused by body movements, thereby improving the accuracy of heart rate monitoring. In previous work, we tested this system across various activity scenarios and achieved promising results. However, there remains room for further optimization of the model and collection of additional data, especially in diverse environments and under various activities.

The main objective of this research is to further optimize the existing multimodal model and collect more diverse data to enhance the model’s accuracy and robustness. We will expand the current research to explore heart rate monitoring performance under various activity conditions and further improve our denoising algorithm to ensure reliable heart rate data in complex real-world environments.

We are recruiting two undergraduate students to assist with follow-up research work. Participants will help collect data and take part in algorithm optimization and implementation. Specific tasks include data collection, model training and testing, as well as result analysis. If you are interested in signal processing, machine learning, health monitoring, and other related fields, and wish to gain hands-on experience in real research projects, we welcome you to join our team!

Team Size: 2

POC: Longfei Shangguan (LONGFEI@pitt.edu)

Stephen Lee

Designing Home Audit App for identify Leakage in Buildings

Objective: This project focuses on the design and development of a mobile application that helps homeowners identify air leaks around windows, a major source of energy loss in residential buildings. The app aims to promote energy-efficient behavior by increasing awareness and providing actionable recommendations, such as sealing techniques or window upgrades. We have a prototype built using the Flutter framework. You will be responsible for

  • improving the front-end interface
  • developing the back-end functionality of the app.
  • evaluate existing models and build ML pipeline to analyze thermal images

Skills you will use or develop: Flutter, Python, JS/HTML

Optional skills: Pytorch/Tensorflow, LLM

Team size: 2-3 students

POC: Stephen Lee (Stephen.lee@pitt.edu) and Benjamin Rottman (ROTTMAN@pitt.edu)

Rosta Farzan

Tracking and Scheduling System

Project description: This project is supervised by Rosta Farzan and is in collaboration with the University of Pittsburgh Community Engagement Center (CEC). The goal of the project is to work with the staff at the CEC to design and implement a system to support their needs in managing reservation, inventory of the equipment, archives of instruction and documents, as well as scheduling programs. The system can be designed as a mobile or web application.

I am seeking a group of 3 or 4 students. Students with a background in web development and/or HCI and user experience would be ideal for this project.

Team size: 3-4 students

POC: Rosta Farzan (rostaf@gmail.com)

Nadine von Frankenberg

Treefficiency

“Treefficiency” focuses on the gamification of learning about energy efficiency at home, specifically targeting appliances and energy-saving options available to home owners and renters.

By integrating interactive, game-based elements, the project aims to educate homeowners and renters on effective strategies to reduce their carbon footprint, energy consumption, and costs.

You will be extending the project with some fun features. Be creative! Features could be: adding/improving more gamification features, improving user experience, defining scripts to analyze user data,… The project currently comprises mobile Flutter apps but could also be extended by a web version to monitor the state of the application and database.

Team Size: 2-4 students

POC: Nadine von Frankenberg (vonfrankenberg@pitt.edu)

Daniel Mosse

Sensing the Built Environment: Data Collection and Analysis for Sennott Square

Objectives: This project will create an easy-to-use and cost-conscious infrastructure to collect data in Pitt buildings. We will collect data in Sennott Square, targeting the 5th and 6th floors, using temperature and humidity sensors, specifically the Govee Thermo-Hygrometer device. The data collected will help Pitt’s Facilities Management (FM) improve building energy optimization and temperature/humidity comfort. For example, FM has “occupied” and “unoccupied” schedules, with different temperature/humidity setpoints; we will detect these schedules and let them know.

Initial data collection revealed no significant temperature differences between weekends and weekdays, suggesting a lack of HVAC optimization based on occupancy or schedules. This project aims to expand upon these findings by conducting more extensive data collection and analysis in Sennott Square (which will serve as an exemplar for other buildings).

The student is expected to:

Become familiar with the functionality of Govee sensors+hubs, and the current initial deployment.

Improve the data monitoring and collection to deploy more sensors to collect data from various rooms. Show that the deployment has small error (<1%) throughout the data collection. This process should be automated, that is, no human intervention aside from placing sensors in the right locations.

Create a software infrastructure to export the data to a repository (perhaps reverse engineer the communication protocol), analyze the data (see next item), and produce daily summary reports, for example how much data was collected, alert users if number of samples is lower than expected (i.e., the system is not working as intended), etc. This process should also be automated.

Process and analyze the collected data to identify patterns and reverse engineer HVAC schedules (i.e., determine whether heating and cooling schedules are programmed based on seasonal variations, weekends, working days, or working hours). This process should be automated, that is, no human intervention aside from placing sensors in the right locations.

We also want to do similar analysis for data collected in a residential building, time permitting.

Requirements: excellent programming, independent work, keeping to a schedule, class in Data Mining or Machine Learning a plus.

Team Size: 2 students

Points of Contact (POC): Daniel Mosse (mosse@pitt.edu) and Ousmane Dieng (oud5@pitt.edu)

Concept Extraction

Background: Concept Maps (CPMs) are a combination of a the knowledge that students will learn in a course (that is, concepts), and the relationships between these concepts. The two projects are related, and two teams will likely work on each separately for 2 months and combine the projects at the end. Both projects will use similar or the same data, and thus it will be expected that both teams share critical information that they’ve learned with each other to help improve their results. These research projects will hopefully yield scholarly papers to be published.

The goal is to extract concepts from instructor course materials, namely from PowerPoint or other forms of slideshows. Students will be expected to go beyond text, and examine things like text formatting, text placement, images, and animations, when determining what is considered an important concept. If the group deems it helpful, they may also examine other materials from the course, such as syllabi, assessments, or the textbook(s) and other reading materials. Students MAY utilize LLMs in some capacity to determine the concepts, but it will likely not be as simple as submitting materials directly to an LLM to examine.

Requirements:

  1. Students must have taken and passed a course in AI, ML, NLP, Deep Learning, Data Mining, Data Science, or other related topic or subtopic.
  2. Students should know or be willing to learn how to use LLMs programmatically (ie, thru the API) in a short amount of time. Note that we will not expect nor require students to train an LLM from scratch, but other techniques like fine-tuning or utilizing RAG is considered a plus.
  3. Students should be willing to work with others and divide work as equally as possible.
  4. Students will be required to meet with the project managers once a week for updates and direction.

Team Size: 2-3

POC: Daniel Mosse (mosse@pitt.edu)

Relationship Mining

Background: Concept Maps (CPMs) are a combination of a the knowledge that students will learn in a course (that is, concepts), and the relationships between these concepts. The two projects are related, and two teams will likely work on each separately for 2 months and combine the projects at the end. Both projects will use similar or the same data, and thus it will be expected that both teams share critical information that they’ve learned with each other to help improve their results. These research projects will hopefully yield scholarly papers to be published.

The goal is to extract the prerequisite relationships of concepts from instructor course materials, namely from PowerPoint or other forms of slideshows, when concepts are already determined. A prerequisite relationship is defined as the relation between two concepts, such that one concept must be learned prior to being able to understand the subsequent concept. Students will be expected to go beyond text, and examine things like text formatting, text placement, images, and animations, when determining if one concept is a prerequisite of another. If the group deems it helpful, they may also examine other materials from the course, such as syllabi, assessments, or the textbook(s) and other reading materials. Students MAY utilize LLMs in some capacity to determine the concept relationships, but it will likely not be as simple as submitting images of the slides directly to an LLM to examine.

Requirements:

  1. Students must have taken and passed a course in AI, ML, NLP, Deep Learning, Data Mining, Data Science, or other related topic or subtopic.
  2. Students should know or be willing to learn how to use LLMs programmatically (ie, thru the API) in a short amount of time. Note that we will not expect nor require students to train an LLM from scratch, but other techniques like fine-tuning or utilizing RAG is considered a plus.
  3. Students should be willing to work with others and divide work as equally as possible.
  4. Students will be required to meet with the project managers once a week for updates and direction.

Team Size: 2-3

POC: Daniel Mosse (mosse@pitt.edu)

Peter Brusilovsky

Creating Problems for Learning Programming with an LLM

We have developed a system for creating worked examples and problems for learning programming in several languages, using LLM to generate explanations for examples and distractors for problems.

Now we need help in creating actual examples using this system, using high-quality code from faculty and textbooks, complementing this code with problem statements, generating explanations, and editing it to ensure corrections. We need students with good knowledge at least in one of the following languages - Java, C, C++, Python, SQL. No need to know all, one is enough. Since we need high quality, we need to make sure that each product is checked by at least 2 students. A team of 2-3 students, given the scope of the work, will suffice. Students interested in CS education would be most welcome. It is a good chance to explore the power of LLM as well.

Team Size: 2-3 students

POC: Peter Brusilovsky (peterb@pitt.edu)

Luís Oliveira

Demonstrator for the CS department

If you ever wadered through the 6th Floor in Sennott Square, you’ve seen the small cabinet with a few old computers - and more recently a old Disk platter.

What?? You haven’t? What are you waiting for? Anyway…

This semester, one of the projects we have in mind is to expand that small corner to include some interactive demonstrations. The interactive part will involve program a Raspberry Pi to interact with some capacitive buttons. Then, you’ll build a simple GUI to help people navigate the different demonstrations. The capacitive buttons will be used to navigate this GUI. This is a good opportunity for you to get some experience dealing with hardware and how to use it with a Raspberry Pi.

Finally, your team will come up with a couple of cool CS concepts, and demonstrations of those concepts. For example: Logic circuits and how they work, algorithmic complexity, memory translation and access. Your own ideas will also be welcomed.

  • Skills you should have (or be willing to learn on your own!):
    • Python/C/C++ - any of these should do, and others are probably also fine!
    • CS0449 (CS1550 would be even better)
    • Front-end/GUI experience would help with the GUI parts (nothing too advanced)

Team size: 3-4 people

Point of contact: Luis Oliveira (loliveira@pitt.edu) and Matt Barbosa

Measuring latency in the Linux network stack

This project involves developing a Time Division Multiple Access (TDMA) network-traffic scheduler within the Linux kernel. You are taking a kernel module developed in a previous semester, and work on the configuration from user space. We’ll build a simple demo application using it, and measure the overheads using extended Berkley Packet Filter. The team will continue work done in a past semester by another capstone. They researched eBPF and how to take measurements. You will pick this work, and measure the delays of two mechanisms to buffer and delay messages. One in the Kernel, and one in userspace.

Prerequisite: Must have taken CS1550. Languages: C

Team size: 3–4 students

POC: Luís Oliveira (loliveira@pitt.edu)

Interactive Web App

Using Svelte to build an interactive application like logisim. This work started last Spring, but the tool is missing some functionality. For example, subcircuits. The objective for this semester is to continue development that allows this tool to completely replace logisim. The first couple of weeks are expected to be dedicated to your understanding and re-deployment of existing code. During that period, you will (with my help) plan the changes and development for the rest of the semester. The initial goals will be the support and editing of subcomponents. Once that goal is achieved, we’ll work in other functional issues - e.g. connector behaviour, and tunnels - at your choice.

  • Languages: Javascript/Svelte.

  • Team size: 3–4 students

  • POC: Luís Oliveira (loliveira@pitt.edu)

Emma Jordan

AI Racing RC Car

I have two remote controllable RC cars with cameras and Nvidia Jetson GPUs. I would like to develop a system for them that can be used to teach the cars to drive around various areas and race each other. The software for these systems needs work is simple and I would like to build out tooling that will allow for it to easily record user controlled demonstrations, run the cars autonomously, and be able to give feedback to the car on how well it is doing. I also need functionality developed to do some learning from demonstration to create a model that can be used to drive the car autonomously. There is lots of room for innovation in this project and to do both AI work and software engineering. If you are interested in AI, autonomous systems, or edge devices this could be a great project for you.

Team Size: 3-4 students

POC: Emma Jordan (emma.jordan@pitt.edu)