SuperUROP All SuperUROP Projects
spacing
SuperUROP
 
Home
Letters of Recommendation
Login

SuperUROP Projects

"Air transport connectivity development in developing countries"
Faculty Advisor: Steven Barrett
Mentor(s): Florian Allroggen
Contact e-mail: sbarrett@mit.edu, fallrogg@mit.edu
Research Area(s):
Air transport facilitates global connections for travelers and goods, thereby enabling globalization. In turn, there is significant societal interest, e.g. from policymakers, to understand air transport networks in terms of the connections which these networks offer to potential users. The level of connectedness at a specific location is determined by the quality of flight schedules, which, in turn, is driven by the availability of direct and indirect flights as well as quality attributes of indirect flight routings such as detours and transfer times.

In the literature, connectivity metrics such as MIT’s Global Connectivity Index have been proposed to analyze the connectedness created through air transport networks. In this SuperUROP project, we plan to analyze the spatial distribution and evolution of air transport connectivity in developing countries, e.g. in Africa. In particular, we plan to study (i) the direct and indirect connectivity trends; (ii) the contribution of air transport hubs inside and outside of the developing countries to the connectivity trends; and (iii) the contribution of different carriers to the trend. For this purpose, we will analyze global airline schedules and run the Global Connectivity Index model. Data mining approaches will be developed and applied to isolate the underlying trends in connectivity and air transport supply.

We expect the results to provide insights into the dynamics of market penetration through airline networks. Furthermore, the results could showcase the sensitivity of connectedness to network design, e.g. with regard to hub location and hub schedule design, thereby both informing airline and airport strategy as well as local policymakers
"Quantifying the environmental impacts and economic costs of renewable jet fuels"
Faculty Advisor: Steven Barrett
Mentor(s): Mark Staples
Contact e-mail: sbarrett@mit.edu, mstaples@mit.edu
Research Area(s):
The purpose of this project is to assess different technology options for producing renewable jet fuels on the basis of their environmental impacts and production costs. There are three general directions that this project could take, based on the interests of the applicant.

1) Quantifying the impact of biomass feedstock and conversion technology selection, geographical location, and technology maturity on the life-cycle greenhouse gas emissions of different renewable jet fuel production options. This work will inform international policy that determines the emissions credits airlines receive for using renewable fuels.

2) Assessing the impact of policy on the cost of producing renewable jet fuels. Renewable jet fuels come at a cost premium to petroleum-derived fuels, and policy-makers are considering various options to incentivize their production in the face of this price differential. This project will use stochastic techno-economic modeling to determine the production cost and risk impacts of the available policy options.

3) Quantifying air quality trade-offs associated with renewable jet fuels. Renewable jet fuel has lower sulfur and aromatic content than petroleum-derived jet fuel, resulting in fewer SOx and BC emissions from aviation. However, the cultivation of biomass for fuel production results in increased ammonia emissions, and the formation of particulate matter. This project will aim to quantify the aggregate impacts of renewable jet fuel production and use on public exposure to degraded air quality.
"Aircraft Autoflight Analysis using OPM"
Faculty Advisor: Charles Oman
Mentor(s): Dov Dori
Contact e-mail: coman@mit.edu
Research Area(s): Aircraft Systems Engineering, Air Transportation Systems, Humans in Aerospace
Boeing, Airbus, Embraer and other modern transport aircraft have highly automated flight guidance systems. However, pilots are trained to make major safety critical decisions to avoid runway overruns, particularly when and how to reject a takeoff or to abandon a final approach. There are many possible contingencies and failure types, decisions must be made within seconds, and a portion of the landing or go around are designed to be flown manually. Existing procedures and automation modes and interfaces were designed for two pilot operation. Airlines and manufacturers are now considering the possibility of single pilot operation. The challenge is to design systems that could provide additional decision support e.g. if a single pilot must perform a rejected takeoff or go around. A first step in the design is to use OpCAT (Link , a systems engineering modeling tool now in wide use at MIT, to represent the current normal transport aircraft takeoff or final approach phase, including pilot interaction with the auto- and manual flight controls, displays and other aircraft systems. The second step is to add the known contingencies and existing decision rules. Additional new decision support concepts would be added and validated in the third step.
"Localization Systems for Smartphones and IoT Devices"
Faculty Advisor: Moe Win
Mentor(s):
Contact e-mail: moewin@mit.edu
Research Area(s): Communications and Networks
Location-awareness is essential for a growing number of emerging applications including home automation, social networking, and vehicle autonomy. Today, coarse outdoor localization is achieved primarily using GPS. However, GPS is unreliable in certain outdoor environments such as in urban canyons and is not available indoors. Accurate localization in such environments remains an active research problem.

In the absence of GPS, range measurements between a device and several anchors with known positions can be used to localize a user. Until recently, expensive wideband radios have been required to obtain ranging accuracies of a few centimeters. This technology is now available on affordable chip-scale radios that can potentially be integrated in smartphone and internet-of-things (IoT) devices. The measurement accuracy of such radios is often impaired by several factors including non-line-of-sight (NLOS) propagation conditions, measurement biases, and RF interference. Learning algorithms and estimation techniques can be used to calibrate-out or compensate such real-world impairments. The goal of this project is to investigate such techniques and develop a low-cost indoor localization system based on inexpensive, commercially available technology.

This project aims to design algorithms and develop software for a demonstration platform that enables localizing smartphones and IoT devices. In particular, the student(s) will establish techniques to fuse data from different sources, including wideband radios and the sensor suite of smartphones (IMU, and GPS). The student(s) will also design algorithms for auto-calibration to improve the accuracy of range measurements. This project will lead to a solid understanding of statistical inference and information fusion, as well as experience in implementing real-world localization systems.
Localization Systems for Smartphones and IoT Devices
"Acceleration of TensorFlow for high fidelity aerospace design"
Faculty Advisor: Qiqi Wang
Mentor(s): Chaitanya Talnikar
Contact e-mail: qiqi@mit.edu
Research Area(s): Aerospace Computational Engineering
This project provides opportunities to learn both the state-of-the-art algorithms for aerospace high-performance computing, as well as a recently developed machine learning library, TensorFlow.

Our group is experimenting with writing computational fluid dynamics simulation and design tools using TensorFlow, a Google-created deep learning library adapted to heterogeneous high-performance computing environments. Initial results show tremendous promise. However, we encountered performance bottlenecks due to components of TensorFlow that are not thoroughly optimized. These components, including sparse matrix-vector product and advanced indexing operations, may not be often used in machine learning applications but constitutes a significant portion of computation in aerospace computational engineering. The purpose of this project is to identify and optimize these components of TensorFlow for aerospace computational engineering applications.

Total: 5

Technical questions related to this site? Please contact Search-Master@superurop.eecs.mit.edu