Electrical and Computer Engineering REU
Research Experience for Undergraduates – or REU – brings students from around the world into the research laboratories of the Department of Electrical and Computer Engineering each summer. These students work with a faculty member and their research group to tackle an innovative research project. Students admitted to the program receive a competitive monthly research stipend as well as arranged on-campus housing and a travel allowance.
Applications for the Duke ECE Undergraduate Summer Research Program 2018 were due January 31, 2018.
The application for Summer 2018 is now closed.
The application for Summer 2019 will open in late November, 2018.
Domestic and international students are invited to apply. The program is designed for students who are juniors in the spring before their REU summer; exceptional sophomores may receive consideration. Students should be majoring in ECE or a related discipline relevant to their area of research interest.
Dates and Stipend
- Applicants will be notified of decisions by February 15, 2018
- Program Dates: May 20 through July 28, 2018
- Stipend: $500/wk, plus $30/day on-campus dining allowance
- Travel Reimbursement: Up to $400 for domestic travel, up to $800 for international travel
- Housing: Shared rooms at the JB Duke Hotel on Duke Campus
The following projects will be available in Summer 2018. Questions about any of the projects or the REU program in general should be directed to Amy Kostrewa (firstname.lastname@example.org).
Explainable AI in the Real World
This project includes joining of team of students applying machine learning techniques to large medical data sets and then translating the results into a format understandable by a large cross section of no-expert. Some basic understanding of machine learning techniques would be beneficial (but not mandatory) or any experience in high dimensional data visualization.
Faculty Advisor: Prof. Mary "Missy" Cummings (email@example.com)
Conventional holography is a sensing technique that utilizes the interference of monochromatic (narrowband) light to encode details about the phase of the electromagnetic field of an object into the intensity of that field. The Laboratory for Engineering Non-traditional Sensors (LENS) has recently been developing computational and compressive sensing methods that can potentially relax the requirement for narrowband operation. Work on this project will involve analytical, computational, and experimental exploration of how these methods can be applied to allow for holographic measurement using broadband (potentially full-color) light, vastly increasing the amount of information a holographic sensor can acquire.
Faculty Advisor: Prof. Michael Gehm (firstname.lastname@example.org)
Deep Learning Robotic Grasps from Large Simulation Datasets
Deep learning has had impressive success in computer vision and language understanding, but progress on robotic planning and control tasks have been limited. One reason for this slow progress is that it is difficult to generate huge amounts of data; the other is that it is unclear what network structures work well for robotic tasks. This project will leverage advancements in robust simulation of object manipulation to train and evaluate deep networks to generate robot grasps for novel objects. Students must have a background in machine learning, Python programming experience, and have experience in PyTorch, TensorFlow or related deep learning packages. Experience in robotics is a plus.
Faculty Advisor: Prof. Kris Hauser (email@example.com)
Investigation of Field Emission Cathodes for Mass Spectrometry Applications
Grand Challenge: Engineer the Tools of Scientific Discovery
Currently researchers at Duke in the group of Professor Jeffrey Glass are developing a miniature mass spectrometer incorporating aperture coding, carbon nanotube field emission ion sources, and ion array detectors. Typical mass spectrometers are large and expensive limiting their use in the field. Aperture coding and field emission ion sources enable reducing the size of the instrument without sacrificing performance. It is expected that the REU student will primarily be assisting with analytical characterization of carbon nanotubes field emission ion sources using resources available at Duke's Nanomaterials and Thin Films Laboratory and Shared Materials Instrumentation Facility (SMIF). We envision the student taking SEM images, helping with the collection of field emission data, and potentially operating other SMIF characterization tools. In addition, the student will have the opportunity to participate in field testing of the miniature mass spectrometer prototype.
Faculty Advisor: Prof. Jeffrey Glass (firstname.lastname@example.org)
Odorant Modulation for Sanitation in the Developing World
Sponsor: Gates Foundation
One in three people worldwide do not have access to appropriate sanitation of human waste. This results in dirty drinking water and several million preventable deaths each year related to gastroenteritis from its consumption. This is a preventable problem, but requires rethinking how human waste should be sanitized in areas where infrastructure such as sewers, electricity and running water are not available. Our work, in collaboration with RTI International and the Bill and Melinda Gates Foundation, aims at reinventing the toilet so that it is off-grid, affordable, energy efficient, and capable of sanitizing human waste in under-developed areas.
However, malodor nuisance is a major risk factor in the adoption of effective sanitation technologies. Foul-smelling sanitation facilities persuade people to practice open defecation in developing countries. Most common odorant molecules consist of carbon backbone ending with functional groups such as aldehydes, alcohols, or ketones. Interestingly, the change of the chemical functional group on a common carbon backbone can result in dramatically different odorant perceptions, from fruity to waxy or grassy for instance. Our previous work has demonstrated the capability to modulate malodor and generate a pleasant olfactory perception simply by applying an electrical signal to the offending liquid/gas source.
The REU project will aim at expanding the family of odorants that can be treated through the modulation process. The undergraduate student will be expected to use a variety of electrochemical (electrolysis, voltammetry and electrochemical impedance spectroscopy) and physical-chemical (chromatography and nuclear magnetic resonance spectroscopy) techniques for synthesis and characterization. The student will gain knowledge in fundamental and experimental analytical chemistry and will improve her or his laboratory skills.
Energy-Related Materials and Device Characterization for Thin-Films Deposited by Resonant-Infrared Matrix-Assisted Pulsed Laser Evaporation (RIR-MAPLE)
RIR-MAPLE is a laser-based thin-film deposition technique appropriate for organic and hybrid organic-inorganic materials, including polymers, colloidal quantum dot/polymer hybrid nanocomposites, and hybrid perovskites. These materials are investigated for optoelectronic and energy-related devices, such as light emitting diodes, solar cells, and supercapacitors.
In this project, the student will investigate the materials properties and device performance of organic/hybrid materials deposited by RIR-MAPLE using atomic force microscopy, UV-visible absorption spectroscopy, photoluminescence spectroscopy, external quantum efficiency, solar cell fill factor measurements, or light emitting diode luminance measurements.
Faculty Advisor: Prof. Adrienne Stiff-Roberts (email@example.com)
Improving Morphology and Contact Interfaces for Printed Nanomaterial-Based Electronics
Many new electronic applications can be made possible with more affordable and readily customizable circuits. These circuits do not necessarily need to exhibit the high performance achieved with silicon-based CMOS chips, but they do need to offer diverse functionality and moderate performance at a low fabrication cost. Printing electronics is a viable approach for enabling this new electronics era. For decades, organic materials have been studied for their use in printed electronics, but they suffer from compatibility issues for many applications and considerable performance limitations. A more recent option has been to use nanomaterials printed into thin films. Nanomaterials offer superior electronic properties to bulk materials, including organic polymers, and are able to be dispersed into a variety of inks for printability. Further, nanomaterials are robust to extreme environments and highly compatible with a nearly endless variety of integration schemes. Entirely new applications, from highly sensitive biomedical diagnostics to sensors for harsh environments, can be enabled with a printed nanomaterial-based electronics technology.
In this project, an aerosol jet printer will be used to improve the morphology of various nanomaterial inks, thus increasing their usefulness in printed electronics. Specific focus will be on the exploration of 2D nanomaterial inks. The student will optimize the rheological properties of a series of 2D inks by varying the concentration of specific additives and measuring the ink’s resulting changes in viscosity, surface tension, and dispersion stability. The student will then print the optimized inks as the dielectric layer in carbon nanotube thin-film transistors. A series of electrical (I-V, C-V sweeps) and physical (SEM, Raman) measurement tools will be used to evaluate the electrical and structural performance of the printed films and devices. The printing process requires training and customization of several key steps, including printing conditions, film curing, and integration sequence. The student will also be expected to take part in discussions where results will be analyzed and new ideas potentially formulated for inclusion in the project.
An ideal candidate for this project would have some previous knowledge and experience in solid-state physics including carrier transport in semiconductors, previous knowledge and/or interest in electronics, and be competent in operating complex tools. They should also be self-motivated and maintain a strong work ethic in terms of commitment and follow-through. A collaborative, team player is a must.
Faculty Advisor: Prof. Aaron Franklin (firstname.lastname@example.org)
Development of a Fully-Printed Oxygen Electrode for Use in a Low-Coset POCT Application
The development of affordable point-of-care-test (POCT) electrical devices could significantly decrease the severity of countless maladies around the world by early detection and treatment. From third-world nations to disaster relief scenarios, a POCT has a wide range of possible uses. To facilitate the use of these POCT devices in remote locations where electrical power may be limited or non-existent, the development of low-cost energy storage is required.
In this project, the development of a fully-printed air electrode will be studied using a simple stencil printing method. A zinc-air battery is an ideal candidate for use in low-cost applications because the bulk of the materials are low-cost and abundant and the battery contains relatively innocuous materials. To minimize the cost of the battery, this project will concentrate on the development of a low-temperature curable oxygen electrode. The student involved in this project will be trained on various techniques for printing electronic materials as well as electrochemical testing techniques. The student will fully characterize the device to determine the impact of different printed morphologies and post processing techniques on the performance of resultant energy storage devices. The student will also be expected to take part in discussions where results will be analyzed and new ideas potentially formulated for inclusion in the project.
The ideal candidate for this project would have some previous knowledge of solid-state physics and electrochemistry, though this is not a requirement. They should also be self-motivated and maintain a strong work ethic in terms of commitment and follow-through. A collaborative, team player is a must.
Faculty Advisor: Prof. Aaron Franklin (email@example.com)
Shell Ocean Discovery XPRIZE Contest-Related Projects
We’re continuing work on a multi-year effort to compete in the Ocean Discovery XPRIZE, where the challenge is to map a huge area of ocean at high accuracy in 24 hours. The long term plan is to develop a highly redundant multi-copter drone to transport, drop, and recover custom deep-submersible sonar pods built using off-the-shelf parts. At present, we have two courses of students developing hardware and software toward this goal, but that’s just scratching the surface.
Here are some of the challenges we’re looking for REU student(s) to address:
Gas-Electric Hybrid Drone
We have a large gas-electric hybrid powered drone in development and students could work on aspects of the drone design optimization and control. Example tasks might be:
- Long Range Data Communication: Working on long range data communication either with Iridium satellite modems or over the horizon radio to allow for status updates when out of sight.
- Hybrid Gas Electric Efficiency Optimization: The microcontroller that supervises the hybrid power system could optimize the efficiency of the gas generator buy suitable cycling between battery and gas power.
- Pickup and Delivery: The delivery and retrieval of objects to remote sites particularly at sea requires development of both long mid and short range guidance techniques. GPS, SONAR, LIDAR, LoRa, and WiFi are all being used and considered.
Hardware Pod for SONAR Mapping of the Deep Ocean Floor (4,000 meters deep)
We plan to use our drone to deliver a SONAR system that collects data as it descends on a winch line to the ocean floor.
Example Tasks might be:
- Development of SONAR transmitter hardware for 20-40 kHz SONAR: Currently we are using Raspberry Pis with a high end audio sound card and an audio class D amplifier, but we have Single Board Computers and microcontrollers we are considering.
- Development of SONAR echo receiver hardware for 20-40 kHz SONAR: Currently we are using Raspberry Pis with a high end audio sound card, but we have Single Board Computers and microcontrollers we are considering, plus commercial preamplifiers.
- Epoxy sealing of the SONAR hardware into a pod that is watertight to 4000 meters deep: We use marine epoxy and take great care to remove all trapped air that would compress under the 400 atmospheres of pressure at 4000 meters. We use 3-D printed molds to shape the epoxy into a fluid dynamically appropriate shape to descend and ascend smoothly and quickly. The electrical feed throughs from the epoxy to the SONAR transducers and Battery charge ports must be carefully designed to not leak or degrade performance.
Securing Autonomous Systems Against Attacks
Most of existing autonomous systems have not been built with security in mind. Even with the proliferation of different networking technologies and the use of more open control architectures, until recently their security has usually been an afterthought. In the last few years, several incidents have clearly illustrated susceptibility of these systems to attacks, raising attention to serious security challenges. These include the StuxNet virus attack on an industrial SCADA system, as well as attacks on modern cars and RQ-170 Sentinel drone that was captured in Iran.
Relying exclusively on cyber-security techniques for securing these systems is insufficient. For example, GPS spoofing attacks to misguide a yacht or unmanned aerial vehicle (UAV) of their routes have been demonstrated in recent years. Consequently, the goal of this project is to show how combination of security-aware control policies can be used to increase security guarantees in autonomous systems operating in contested environments. The scope of this project will be to secure the video-based navigation of an UAV, by exploiting recently developed control techniques such as attack-resilient state estimation and active attack detection.
Faculty Advisor: Prof. Miroslav Pajic (firstname.lastname@example.org)
Learning-based Cooperative Control in Autonomous Cars
The goal of the project is to create algorithms and software libraries for learning-based control in collaborative autonomous driving. The considered system involves at least four fully-autonomous vehicles; the emphasis of this work is on dynamic collaborative path planning between fully-autonomous vehicles, with the goal of enabling multi-lane, multi-vehicle formation driving. Student efforts will focus on perception, planning, and control for autonomous navigation starting from the initial vehicle design and our ROS-based software that provides baseline autonomous vehicle operation.
Faculty Advisor: Prof. Miroslav Pajic (email@example.com)
Ensuring Timeliness in IoT Systems
To manage complexity of modern Internet-of-Things (IoT) systems, employed platforms require the use of efficient operating systems (OS) that provide timing, performance and security guarantees, as well as support for key IoT networking and communication technologies. One example is the latest mbed OS 5.1, which incorporates a Real-Time Operating System (RTOS) in order to provide native real-time thread support to the applications running on top of the RTOS. However, the underlying scheduling is based on the well known round-robin policy, effectively preventing the use of this OS in development on safety critical IoT applications where timeliness is the first class citizen. In these systems (e.g., in industrial, medical, automotive, avionics and many other application domains), the ability to prioritize a critical task at any given moment in time, and guarantee availability of resources required to complete that task is essential.
Consequently, the goal of this project is to extend the open-source scheduler for the RTOS and provide support for a variety of commonly used real-time scheduling policies that take into account thread criticality and execution deadline constraints. The developed RTOS will then be deployed and evaluated in several case studies in automotive, industrial automation, and medical device domains, allowing the student to both learn foundations of the RTOS development as well as its use in design of real-world systems (e.g., electric vehicles, cloud-based reconfigurable industrial systems, networked implantable medical devices).
Faculty Advisor: Prof. Miroslav Pajic (firstname.lastname@example.org)
Student-Proposed REU Project
Prospective Duke ECE REU students are invited to propose a project to work on under the advisement of one of our Duke ECE faculty members. On the application form, in addition to uploading your resume and statement of interest, please use a format similar to the project descriptions above to upload a proposal including the following:
- The name of the faculty member(s) you propose to work with
- The description of a project that can be completed within or reach a reasonable stopping point at the end of a nine-week period
- An explanation of an expected product or outcome from the project
- A list of any necessary supplies or materials
Staff contact: Amy Kostrewa (email@example.com)