Fall 2020 NASA Intern Projects
NASA interns will present their outcomes on their fall 2020 projects. These are the projects, names of presenters and teams:
- Title: Natural Language Processing (NLP) Techniques in Air Traffic Management Planning
- Title: Open-Sourced Reconfigurable Aircraft Cabin Airflow Model
- Title: Supply and Demand Nodes
- Title: Search and Rescue Drone Requirements Analysis and Design
- Title: Development of Modeling and Simulation Capability to Analyze Supply Chain Network for Small Unmanned Aircraft Systems
Presenter: Stephen Clarke, Sacred Heart University
Hussein Adams, San Jose State University
Jacqueline Almache, North Carolina State University
Patrick Maynard, Harvard University Extension School
The research goal for our project is to explore and understand the use of innovative and emerging applications of Natural Language Processing (NLP) to drive post operational analysis of Air Traffic Management (ATM) documents. Specifically, we are evaluating Letters of Agreement (LoA), Standard Operating Procedures (SOP) and Notice to Airmen (NOTAM). Using NLP and advanced data analytics, we hope to gain an understanding of how these documents are structured and use predefined language models such as Bert, XLNet, and Word2Vec to develop different machine learning models while extracting potentially meaningful information from the documents. Potential applications of these models may include development of an advanced search engine for end users (air traffic managers, controllers, commercial services, pilots or other users) who are interested in understanding air space constraints, temporary flight restrictions, or other pre-flight information for planning purposes. In this project we also conduct a comparison of data workflows and methodologies that may be suitable for different document types. We will also include various evaluation metrics such as accuracy (to show how well that models perform) and validate “testing versus training” losses (to prove our models are not overfit and can be generalized). We will also share the NLP algorithms, techniques and models that have guided our research including Named Entity Recognition (NER), Word2Vec, Doc2Vec, clustering, dimensionality reduction and others. These methods and lessons we have learned will help guide future research efforts, enlisting what works and what doesn’t. We also hope that our recommendations and lessons learned will be useful for developing other NLP applications and prototypes for improving decisioning and planning within the National Air Space (NAS).
Presenter: Ava Thrasher, Georgia Institute of Technology
Julia Clark, University of California, Santa Barbara
Chloe Greenstein, Lewis and Clark College
Aaron Ocken, Washington University in St. Louis
Nitin Rao, Palo Alto High School
Ananya Saxena, Purdue University
Liam Sullivan, San Jose State University
Alena Zeni, Embry Riddle Aeronautical University
Our project is a series of computational fluid dynamic (CFD) simulations modeling airflow in a single aisle commercial passenger airplane. This is a new approach for rapidly replicating and modeling airflow within a cabin in order to analyze mitigation techniques and flow patterns to be used in future scenarios. Air ventilation was simulated using four inlet vents and two outlet vents. Exhalations of a contaminated index passenger were modeled by an inlet contributing an air and water mixture. Three sets of models were run: nominal, increased, and decreased ventilation. Breathing, coughing, and sneezing scenarios were all modeled for each set. Our model was then probed for areas of water vapor concentration. Despite computational restrictions, this model was able to replicate general results and trends presented in other empirical studies.
Presenters: Natalie Zimmermann, Rutgers University; Aneesh Galgali, Rutgers University; and Pedro Salazar, Iowa State University
Angad Chugh, Boston College
Kiran Gomatam, Boston University
Bjorn Johnson, University of California, San Diego
D’leela Saiyed, George Washington University
Sarah Tesfaye, ???
Christos Vasilarakis, University of Maryland, College Park
This project is focused on matching supply and demand nodes by redirecting food waste via aviation. We’ve built a portal to connect farmers who have surplus food with food banks; it works as a marketplace, where farmers can list surplus food and food banks can browse and connect with farmers directly. The main drawback of the portal is that we do not have many users, so we’ve also organized deliveries by-hand and have connected with many non-profits to do so. The project team delivered 320-produce boxes to two food banks in California over the summer with the help of Angel Flights, a group of non-profit volunteer pilots. We’re currently working to organize another delivery this fall. The main deliverable for this project is the portal. A bonus deliverable will be the connections we’ve created between non-profit organizations, which will hopefully continue to work together beyond this project to fight food waste and food insecurity in America. This project contributes to NARI’s goal of developing partnerships that maximize aviation impact and opportunities, and also contributes to NASA’s large goals of addressing national challenges and improving life here on Earth.
Presenter: Billy Bilicki, Jr., San Diego State University
Felipe Borja, Virginia Polytechnic Institute and State University
Brandon Botchway, University of District Columbia
Christian Burwell, Northeastern University
Marcos Figueroa, ?????
Jacob Halaweh, Santa Rosa Junior College
Amol Garg, Cornell University
Christopher Townsend, Florida International University
Natalie Urieva, Cal Poly Pomona
This project was focused on designing a small unmanned aircraft system (sUAS) for United States Coast Guard search and rescue operations. The sUAS needed to meet specific requirements laid out by the USCG, including minimum flight time requirement, durability, and component sourcing from government approved sources. To accomplish this, we broke the project up into four main sub-systems (1 - Power/Propulsion, 2 - Chassis, 3 - Camera/Gimbal/Payload, and 4 - Controls) to be developed simultaneously. The Power/Propulsion system employed a MATLAB script that utilized battery, motor, and propeller specifications to estimate flight time and top flight speed, allowing us to determine which propulsion components were most adequate and cost effective for the project. Two competing models were developed for the Chassis (foldable-arm design and fixed-arm design) and were analyzed virtually using FEA. A custom gimbal and payload drop were also developed and modeled to meet USCG requirements. The Control System required finding a flight controller that would provide autonomous flight and make piloting the drone intuitive. A ground control system was also explored to make interface seamless and user-friendly. The research done on these sub-systems establish the needed components and design of a sUAS that meets the needs of the USCG.
Presenter: Nabeel Ahmed, University of Michigan, Dearborn
Victor Cabrera, University of Illinois
Kiran Gomatam, Boston University
Jeremy Lee, a Cal Poly State University, San Luis Obispo
Elizabeth Prisby, Purdue University
Supply chain has brought up many challenges to the current aerospace industry (e.g. limited sources, lack of inventory, and cyber security). The purpose of this research is to develop modeling and simulation capability to set up a tiered system of suppliers and analyze macroscopic supply chain robustness, resiliency, risks, and capacity to ensure readiness to support scale, production, operations, and maintenance, repair, and overhaul (MRO) of electric and hybrid vertical take-off and landing vehicles (eVTOLs). For this project, our team focused on drones that weight less than 55 lb. due to the accessibility to existing information and their short range of models. Building a minimum viable product (MVP) and roadmap to build successive MVPs will allow the study to reach its final goal: an entire modeling and simulation platform capable of analyzing the supply chain network of eVTOLs. Subsequent tasks included architecture development, creation of a database of potential suppliers, creating a baseline tiered system, and creating networked modeling and simulation for further analysis.