For Prospective Students:

  Apply Online
  Faculty and Areas of Research
  Sample Projects
  Program Alumni and Past Projects
  SUNFEST in the News

For Accepted Students:

  Required Workshops
  Housing - Intern. House Phila
  Organizational Matters
  Current Students & Projects
  Student FAQs
  Student Resources
  Library Resources

Contact SUNFEST:

Mailing Address:
203 Moore Building
University of Pennsylvania
200 S. 33rd Street
Philadelphia, PA 19104-6314


Summer 2014

Compiled Report, click here.

Alumni 2014

Individual Reports and Slides:

Justin Aird (Electrical Engineering) Hampton College

Amplification of Output Sound of Directional Ultrasonic Speakers for Bus Collision Warning System

Advisors:  Dr. Daniel D. Lee, Alex Burka
Research Partner:  Marcus Pan (SUNFEST)

Parametric speakers utilize ultrasonic waves have very sharp directivity. They transmit sound energy in a very straight path to a narrow area.  Ultrasonic sound can affect the direction of sound. This effect is called “Directivity”. Speakers with a high directivity can send a sound straight ahead in a narrower area.  The goal of this project was to design ultrasonic parametric speakers that could be attached to a 12V battery of a bus. To be used as a pedestrian directional warning system to reduce bus-pedestrian incidents on street corners.  We would design ultrasonic speakers so that it could play an input sound to alert the intended target. Making sure that if this warning system were to be played late at night it would not disturb others. We would build a power amplifier circuit for the ultrasonic speaker that it could draw as much power as possible from a 12V battery.  We looked to see if the speaker was directional and loud enough to be heard at certain distances.  After testing the project looks into what other areas these ultrasonic speakers could be used in such as helping guide the blind, and using it to communicate with some animals since they use ultrasonic frequencies to communicate. 

Porscha Baines (Bio-Engineering) Lincoln University

Electrospinning Polymer Polyvinylidene Fluoride Based Piezoelectric Transducer For The Hearing Impaired

Advisor:  Dr. Jorge Santiago
Grad Student Mentors: Carlos Perez and Timothy Jones

This paper describes the efforts undertaken in the development of a polymer based piezoelectric transducer. Through the process of electrospinning Polyvinylidene fluoride, the fibers are to act as the hair cells present in the ear. Electrospinning is a method whereas the electrical force of a liquid solution overcomes the surface tension to produce very fine fibers. It has been theorized that a piezoelectric transducer may be a proper substitution for the hair cell, resulting in hearing improvements.  Currently cochlear implants are effective in the treatment of severe hearing loss. The cochlear implant of today is made up of microphones, electrodes, a transmitter, magnets, and microcomputers in which some components are external to the human body. Through previous studies device requirements were found to have small dimensions, flexibility, high sensitivity, impedance matching, and biocompatibility. The proposed piezoelectric polymer will meet these requirements and will require no external circuitry. The idea of the new device is to mask the fact that there is an implant surgically installed. The project was tackled during a mere ten week period so there are more tests that can be carried out. Further study is to be done by testing the fibers using the SEM and FTIR to verify if the fibers meet the device

Antonio Basukoski (Electrical Engineering) Columbia University

Response Characterization of Second-Order Hilbert Curve Scattering Antennas to a Farfield EM Source

Advisors:  Dr. Jan Van der Spiegel, Matt Zhu, Milin Zhang

Mechanoreceptors are organic sensory receptors present in the skin that respond to mechanical pressure or distortion. At present, people who undergo amputations are fitted with prosthetics that provide a variable range or motion, but cannot replicate the function provided by mechanoreceptors. A totally passive wireless sensor network with multiple sensor nodes can be used to sense stretch, strain or pressure and address the lack of mechanoreception in prosthetics. A sensor node consists of a sensor antenna, a switch, an RF-DC converter and an identification antenna. Flexible compact size space-filling curve-shaped antennas serve as sensor antennas, and non-flexible antennas excited with a unique ID frequency serve as identification antennas. This paper provides some characterization of the change in resonant frequency of the sensor antenna based on changes of different antenna properties, such as the length of the antenna, its material, and different environmental conditions. Our simulation results show that the change in the length of the antenna produces a shift in the resonant frequency. It also shows that different background materials such as FR-4 and paper have changed the response of the antenna. Last, the physical dimensions of the background material also produce a change in the response of the antenna. The simulation results are consistent with our prediction about the factors that change the resonant frequency of the sensor antenna, and further work can show how this change in response can be quantified.

Jennifer Bourlier (Mechanical Engineering) University of Michigan at Dearborn

Miniature Hull Cell Device for Determining Effects of Additives in Copper Electrodeposition

Advisors:  Dr. Haim Bau, Nicholas Schneider

Over time, the charging capacity of rechargeable batteries declines. This is due to irregularities in the shape of the growth front that forms during the charging cycle. The ideal front is a smooth line but irregularities can occur. These instabilities make it more difficult for all of the particles to be converted back to energy during battery usage. Some attempts have been made to control the shape of the front using active current control or certain additives. This study investigates the effect of a lead additive in the charging fluid. Using a miniature Hull Cell the effectiveness of the additive could be tested across a wide range of current densities on the electroplating of copper. It was shown that the additive was effective in creating an even growth front. Further studies must be made to evaluate the effects of lead additives combined with active current control.

Jamie Johnson (Electrical Engineering) Broward College

Playing With Transition Metal Dichalcogenides

Advisor:  Dr. A.T. Charlie Johnson

Introductory Background: Transition metal dichalcogenides are semiconductors that have complementary properties of graphene. However, unlike graphene, these semiconductors have tunable band gaps. The band gap is significant, because when the conduction band and valence band are separated, the space enables the sensors to turn completely off. In the bulk state the band gap is 1.2eV, while it is 1.8 eV when the TMD is in the monolayer state. 

Open Question or issue: Graphene was highly used in the making of biosensors until scientists found evidence that these sensors had no band gap. In my lab, we are using transition metal dichalcogenides, such as MoS2 and WS2, which have band gaps in both states, high on/off ratio, and are very sensitive in the detection of chemical analytes.

How the issue is being addressed:  In this study, my lab group and I are working on making these n type semiconductors into field effect transistors by constructing flakes on a silicon wafer.  We put molybdenum (IV) oxide on a silicon wafer, evaporate on the substrate, lift the polymer off with acetone and IPA, anneal the chip and then grow the flakes by Chemical Vapor Deposition.

Results: From experimentation, we have succeeded in growing these MoS2 flakes in uniformed areas. This allows us to contact the field effect transistors and detect proteins and DNA particles.

Discussion:  These TMDs are the best material for these biosensors because of their intrinsic properties. Scientists desire high electron carrier mobility, tunable band gap, and a great on/ off ratio Biosensors are needed in the medical field and homeland security, as well as in the agricultural industry.  Also another TMD that we are doing work on is WS2. This is a p type semiconductor that models the MoS2 in many ways. The final objective is to put MoS2 and WS2 flakes together to make a P/N junction for solar cells. This is the future of nano and bio sensing.

Vincent Pacelli (Electrical Engineering) University of Pennsylvania

Software Tools for Analyzing Infant Feeding Characteristics

Advisor:  Dr. Jay Zemel

Roughly one in ten infants in the United States are born prematurely. These infants, or neonates, require additional monitoring to ensure that they develop at a healthy rate. Previous studies have linked the healthy development of these infants with improvements in the strength and dexterity of their jaw motion. While recent improvements in technology have allowed more hospitals to monitor infant jaw motion, previous analytical techniques were limited because they only looked at the statistics of the feeding and not the character of the infant’s jaw motion. In this study, we used the biting and sucking pressure of the infant during the feeding to describe the neonate’s jaw motion. We then began to correlate this motion with the health of the infant. Fourier analysis of the pressure exerted on the feeding nipple over time shows that the jaw motion of unhealthy infants lacks the refined motor control of a healthy infant. This lack of muscular control presents itself as extra noise in the Fourier transform. For this reason, we believe that further research should be conducted using the calculated Approximate Sample Entropy of the signal. As these algorithms calculate the noise present in the signal, they may provide a further metric to examine the health of an infant.

Marcus Pan (Electrical Engineering) University of Pennsylvania

Signal Processing for Directional Speakers

Advisors:  Dr. Daniel Lee, Alex Burka
Research Partner:  Justin Aird (SUNFEST)

Bus-pedestrian collisions often take place at road intersections, while buses turn and pedestrians cross the street distractedly. A warning system that alerts pedestrians in the way of a turning bus would decrease this number of bus-pedestrian accidents. To build a sophisticated and effective warning system, we propose the use of directed speaker technology. This would alert only the pedestrians in the way of the bus and not distract an entire street corner.

To direct sound, we designed a system that would modulate an audio signal to an ultrasonic frequency for transmission. This ultrasonic signal would then be demodulated back to the audible range by nonlinearities in air pressure. Such a system requires precise signal processing to obtain a result with low distortion. This paper will discuss the modeling of that signal processing in MATLAB and its implementation in a microcontroller. The behavior of the MATLAB model was compared to a working prototype that involved mainly analog components, and the results discussed where significant. Various issues and challenges regarding the microcontroller implementation are also discussed, and compared to issues with the analog circuit prototype. The MATLAB model and microcontroller implementation is aimed at making the system more robust by enabling the testing of different parameters and reducing part count.

Golden Rockefeller (Mechanical Engineering) University of Delaware

Task Completion by Multi-Robot Systems in Initially Unknown Environments

Advisor:  Dr. Camillo J. Taylor

Using and coordinating many robots to do a task can in many cases be better than send out a single robot to do it. Different approaches in how the robots communicate under certain constraints are expected to have varying results. Our goal is to come up with different approaches and compare them with one another in the case of exploration. We also compared systems with no communication to those with decentralized communication to those with the more ideal, centralized communication. We simulated multiple robots interacting with each other in a grid like world given with the task to explore. The robots had a communication constraint that required them to be on the same grid for exchange information. We came up with the queen-ant system with robots occasionally returning to the home base to deposit and receive information about the world. We expect that this system will perform better than systems where the robots do not provoke communication and the transfer of information. What we can see from the results are the qualities of what a good multi-robot communication system should have. Such qualities are frequent but efficient communication. The system sees benefits when robots communicate to receive far greater information than they would should they gather it themselves through sensing.

Abel Rodriguez (Bio-Engineering) The George Washington University

Simulating Electrical Double Layer Capacitors

Advisor:  Dr. Jorge J. Santiago-Aviles

It is vital to improve the energy storage of devices in order to better harness renewable energy and deliver it efficiently to essential electronics. In order to fulfill this need an electrical component must be developed that has higher energy storage than a traditional capacitor, greater power rating than a battery, and can last many charge and discharge cycles. High surface area electric double layer capacitors (EDLC) have been developed and appear to be able to fill this void by storing charge in a layer of ions along the surface of electrodes. To optimize the performance of devices of this kind we set out to run simulations where the limits of the device’s charge storing capabilities can be explored using COMSOL. The results indicate that the simulation is indeed able to replicate physical findings meaning it is an appropriate method to predict results given parameters of an EDLC. The simulation profile can be used as a tool to strengthen experimental research, predict EDLC properties, and optimize their configuration for certain tasks. The next steps would be to refine the simulation to better suit the specific application and also expanded upon it by considering new parameters such as heat dissipation, fluid flow, and physical stresses.

Jacob Sacks (Bio-Engineering) The University of Texas at Austin

A Wireless, Real-Time Embedded System for Closed-Loop Myoelectric Control of Sedated Primates

Advisor: Dr. Jan Van der Spiegel
Grad Student Mentor: Xilin Liu

Brain-computer interface (BCI) technology establishes a direct link between the nervous system and external hardware. An important application of such devices is myoelectric control, or the use of the electrical potential generated during muscle contraction as an input signal. This physiological information is heavily used in rehabilitation engineering and novel human-machine interaction. Current commercial systems are limited in their degrees of freedom for control, and are incapable of bidirectional communication with the nervous system. Such restrictions prevent intuitive use of myoelectric devices and closed-loop control with sensory information. This paper presents a microcontroller-based, wireless BCI system that recognizes simple hand gestures as input and produces a corresponding output signal. Surface electromyography (sEMG) is used to capture the myoelectric signals (MES) on the neural recording device, and a simple classification scheme produced offline by a supervised machine learning algorithm identifies each gesture. A wireless link then allows the recorder to dictate the stimulator response to provide unique artificial sensorimotor information to the animal. This method does not require segmentation of MES data, and allows for a continuous stream of gesture classification decisions within acceptable response time. Remarkably, the simple classifier employed on the microcontroller paired with the right signal features allows for gesture recognition within acceptable accuracy. Future studies will couple a more intelligent classifier algorithm with additional sEMG channels to further increase recognition accuracy and include more gesture commands.
Jordan White (Electrical Engineering) Hampton University

Creation of the Delta Hopper

Advisor: Dan Koditschek

The widespread use of robotics has increased over the years. These robots can be used for a variety of jobs, from surveying areas where it would be difficult for humans to look through, or just to do tasks that will make the human life easier. The problem is creating robot that can complete action similar to what humans can do is difficult. Currently what I am trying to do is develop a robot that can hop. The robot created was called the “Delta Hopper” which is a three-legged robot where all three legs are connected to a base where it touches the ground. In this study after the robot was developed, the forward kinematics needed to be found, touchdown detection has to be calculated and detecting the bottom, which is when the robots legs are the most compressed. The result show that this robot it capable of hopping multiple times with the assistance of a “boon” which is just a device that helps the robot stand. The robot is not capable of hopping without a “boon” due to lack of information regarding the touchdown detection and finding where the legs are compressed the most. More time and research would be needed to get this robot to hop without any assistance.

See also the alumni and past projects from: 2016 | 2015 | 2014 | 2013 | 2012 | 2011 | 2010 | 2009 | 2008 | 2007 | 2006 | 2005 | 2004 | 2003 | 2002 | 2001 | 2000 | 1999 | 1998 | 1997 | 1996 | 1995 | 1994 | 1993 to 1986