ESE441 and ESE442: EE Senior Design,
David Magerman, Philip Farnum
Wildland Firefighter Tracking System
Authors: Kate McArdle
Advisor: Deliwala, Sid; Farnum, Phil; Roth, Amir
Currently, there is no efficient method for tracking the location and basic health status of wildland fire fighters when in the field. Walkie-talkies are used for communication between the team’s base station and each team member, but this method requires active participation of both ends. Additionally, the firefighter may not know exactly where he is, nor may he be aware of subtle changes to his health that could indicate he is near serious exhaustion. There is a need for such information to be visually incorporated with other geographical maps to improve fire fighting efforts and the safety of firefighters.
The goal of this tracking system is to periodically and automatically track the geographical location, via GPS, and general health condition, via temperature, of multiple firefighters in the field. The end-user system is non-intrusive, so as not to interfere with existing communication systems and its use requires no attention from the firefighter. Each user’s data is transmitted to the base station, where it is visually displayed on a map.
With this system, neither the firefighters in the field nor those at the base station have to spend a significant amount of time to increase the efficiency of fire fighting. The base station user can easily and quickly see the location and status of each fire fighter as well as of the fire.
Authors: Ryan Aminzadeh
Advisor: Deliwala, Sid
As home appliance technology advances, consumer demand has begun to drive development towards increased accessibility and customization.
Remote access of appliances over the Internet, through an intuitive user-interface, targets this demand. Wirelessly connecting devices within a home allows users to move devices easily and minimizes clutter. Using the new ZigBee protocol for local wireless control allows short data message transmission and long battery life.
A transparent serial communication scheme is implemented to achieve two-way wireless communication. The system demonstrates the ease with which conventional appliances can be adapted to function within the system.
Functionality is demonstrated on a lamp, motion sensor, and television which wirelessly communicate with a remotely accessible server.
Authors: Kylash Viswanathan
Advisor: Farhat, Nabil; Zaghloul; Kahana
Electroencephalography (EEG) measures potential differences within the brain generated by the firing of neurons. The signal can be characterized by its frequency content into special frequency bands. It is hypothesized that certain frequency components of EEG correspond to heightened cognitive abilities. The 4-8 Hz frequency band (also known as the “theta” band) has been linked to greater memory performance. To test this hypothesis, a Brain Computer Interface (BCI) was created that used recorded EEG activity to drive a real time memory retrieval experiment. The current system was constructed as a platform for real time cognitive experimentation. An interface was created that allowed maximum control of experimental conditions. Two classifiers were used to characterize incoming EEG activity. The spectral classifier maximized the speed under which the experiment was conducted while the Neural Network Classifier used all available spatial EEG information. The classifiers were used to characterize the incoming EEG into two states: “theta” or “not theta”. Based on this binary classification, an experimental output was modified. In this case, the experimental output was the presentation of a word to the patient on a computer screen each time a “theta” event was discerned. The hypothesis is that words presented during high theta activity will be remembered by the patient at a greater rate than words presented randomly.
Authors: Stephen Davies
Advisor: Farhat, Nabil;
This project attempts to model stock market behavior by using a neural network to predict the nonlinear, chaotic behavior associated with financial markets. The FiNNet system implements a three step approach to financial modeling. First, four investment portfolios are constructed based on arbitrage pricing theory to mimic certain factors in the market. These factors are total market movement, market capitalization, book-to-market value, and momentum. Second, the future returns behavior of these portfolios are predicted using an error backpropagation neural network. Finally, these predictions are used to develop a profitable trading strategy through a portfolio optimization technique that minimizes risk and maximizes return.
The results over the in-sample period, 2001-2004, netted annualized trading profits of 35%, with an associated Sharpe ratio of 3.02. The out-of-sample period, 2005, resulted in trading returns of 17% with an annualized Sharpe ratio of 2.06. Correspondingly, the S&P 500 yielded annualized returns of 2% in the period 2001-2004 and 5% in year 2005. The corresponding S&P500 annualized Sharpe ratios for these periods were 0.14 and 0.30, respectively.
Authors: Federico Wang
Advisor: Santiago, Jorge
Nanoscale materials have garnered much interest recently because of
their unique electrical, optical, and mechanical properties. In
particular, nanofibers are useful in applications such as medicine,
materials research, industry, aerospace, electronics, fuel cells, drug
delivery systems, fluid filtration and a myriad of other fields.
The most basic approach to fabricating nanofibers is to use a syringe
at a high voltage loaded with polymer and a piece of metal as the
substrate. The drawback of this approach is that the orientation and
diameter of the fibers produced cannot be precisely controlled. In
order to overcome this, a nanofiber fabricating device was designed
and built. The device in question produces individual parallel
nanofibers and allows the user to achieve the desired nanofiber
In the chosen approach, a syringe pump deposits a polymer jet on a
spinning aluminum disk. An electric field is generated between the
syringe and the aluminum disk by applying a high voltage to the
syringe and grounding the disk. A horizontal linear translation stage
moves the syringe pump to avoid undesired fiber overlap. In order to
keep the tangential speed of deposition constant, the rotational speed
of the disk is constantly adjusted to precisely match the motion of
the translation stage.
Authors: Jean-Benoit Daumerie
Advisor: Santiago, Jorge. Judd, Stephen.
Automated systems have greatly improved and simplified mankind’s activities. The Rig Jig sensor system is a setup combining aspects of sailing and an automated feedback mechanism. The model is a proof of concept, showing that a sensor system is capable of measuring the data necessary to control a sailboat, and that software is capable of implementing a working feedback control that can optimize the efficiency of forward movement of a sailboat. The primary goal is to show a relationship between the forces affecting the vessel and the forward thrust, and to use those measurements to maximize sailing efficiency.
The data acquisition system consists of two bidirectional force transducers, a series of instrumentation amplifiers for conditioning the sensor readings, and a microcontroller for digitization and serial transmission. Wind is simulated by means of an industrial fan attached to a wind tunnel. Then the strain gauge sensors react to the aerodynamic forces on the sail of our boat. The sensor data is transmitted from the microcontroller to the PC where it is processed in MATLAB. The software model determines the optimal boom position, which is then set by the system’s feedback control.
A Graphical User Interface was created to easily display the forces on the sail, the boom and wind angles of the setup, and the effective forward thrust of the boat. This simplifies the whole process for the user, enabling them to focus on the data and what is happening rather than worrying about how to collect and interpret that important information. The GUI is critical in tying the automated system together and presenting it to the user in a pleasing and easy to use format. This is essential to accomplish our goal of making sailing more efficient and less complicated for those who desire such services.
Authors: Alex Teichman
Advisor: Yim, Mark
Author: Lucy Zhang
Advisor: Van der Spiegel, Jan
In the field of robotics, research has become focused on creating mobile, unsupervised robots to perform various tasks. From the car-sized vehicles of the DARPA Grand Challenge to the Predator drones in Afghanistan and Iraq, sophisticated algorithms have achieved great advances. However, these algorithms become a limitation as the focus changes to the smaller battery-powered robots used in many applications. Large, power hungry microprocessors become impossible.
This project is focused on designing a compact low power image sensor chip able to perform on-chip calculations of optical flow. Optical flow, the apparent motion in an image, can be used to control vehicle motion and avoid obstacles.
Light intensity is converted to electrical current through an array of photopixels. A series of arithmetic circuits are then used to calculate the spatial and temporal derivatives, as well as take their ratio to calculate the optical flow.
Authors: Armand O'Donnell
Advisors: Zemel, Jay
The Children’s Hospital of the University of Pennsylvania has long been involved with the study of the correlation between obesity or bone density and youth exercise. In 2004, the Pediatric Dynamometer was born as an engineering project in the Moore School of Electrical Engineering.
Current measurements of exercise are not capable of conveniently delivering the main metric by which correlation should be measured: exerted force. The Pediatric Dynamometer has been developed to provide a comfortable, durable modification to any youth’s shoes to allow for accurate measurement of force exerted by the foot over extended periods of time.
The full implementation of the design consists of an inside-shoe embedded system, which receives signals provided by a Piezoelectric sensor and processes them in an on-board microcontroller. The control system contains algorithms for uploading the data into a computer for processing and storage. The sensor and inside component are installed within the shoe in an unobtrusive fashion. A prototype of the shoe which is fully functional and possibly marketable for athletic shoe applications is ready for demonstration.
Authors: Harsh Jain
Advisor: Deliwala, Sid
Stars have been always been a source of fascination to man. Over the years, man has developed many sophisticated tools to observe these stellar bodies. NASA and other large international space organizations have developed telescopes with extremely powerful optics, controlled by sophisticated hardware and complex software.
However, the majority of stargazers are hobbyists with relatively cheap telescopes and mounts that need to be manually controlled and focused. Handling a telescope is no easy feat. Setting up a telescope, orienting it, and carefully tracking the desired celestial body is a tedious process that requires much patience. Modern mounts have a feature that enables the automated tracking of stars. However, unless the telescope has been perfectly oriented, a skill that can only be achieved with years of experience, tracking will never be perfect. This will cause the image to drift out of the field of view of the telescope every few minutes, causing the user to have to constantly make slight changes to the tracking of the telescope.
Imaging the body becomes even more exasperating due to the replacement of the eyepiece with an imager. The user now has to rely solely on the images displayed on his computer screen, which are subject to finicky software settings such as ensuring that the exposure time, gain, and image combination settings have been set to the correct values. These settings can be different for each star. The imager also provides a smaller field of view, which, combined with the extreme magnification of the imager (~200x), causes even the slightest vibration by the user’s touch to displace the point source (such as a star or a planet) from its field of view. Additionally, the image needs to be re-focused when the imager replaces the eyepiece, as they have different optics. For an amateur telescope hobbyist this much need for precision combined with the constant need to manually offset errors due to imperfect tracking may prove discouraging.
Our project aims to minimize the amount of work required by the average amateur telescope user when imaging a point source such as a star or a planet, by passively auto-focusing the image. Although regular cameras incorporate passive autofocus techniques, little work has been done to transfer this technology to telescopes. Passive focusing techniques for telescopes differ from that used in cameras in that telescopes must be able to focus point source images and must work under extremely low light conditions and extremely high magnifications.
Our system works using the following microcontroller-based closed-loop feedback system: The imager feeds the image from the telescope into MATLAB, which calculates its sharpness measure. This value is sent to a microcontroller, which moves the focus motor via an H-bridge for a calculated duration in the predicted direction of better focus. This process is constantly repeated until successive sharpness measures are close enough to assume that the image has converged to the point of maximum focus, thus producing the most focused image. Current work is being performed to replace the microcontroller with a USB controller board with TTL output, controlled using a LabVIEW-generated executable file. This will greatly improve the efficiency of the overall system from the user’s perspective, by increasing portability, decreasing power consumption, and eliminating the need to understand the finicky details of the microcontroller’s operation. Thus, by automating the focusing of the image, the user now only needs to worry about monitoring the tracking on the telescope, allowing for a much more enjoyable and less frustrating experience.
Authors: Robert Callen
Advisor: Kimbrough, Steven
It is highly desirable in many industrial and business settings to maximize output or profit while respecting the constraints of the underlying production system, such as the number of hours a machine is allowed to be used per day, or the size of containers used to transport goods. Such problems get exponentially more difficult to solve as the problem size grows linearly, making it more and more difficult for humans to solve them by hand. The class of problems with this exponential growth property (as well as several other properties) have been labeled by computer scientists as NP-Hard. Because these problems are difficult to solve using simple algorithms, and because they have many practical applications, they are interesting from both practical and theoretical viewpoints.
For our project we chose two meta-heuristics: simulated annealing, and genetic algorithms. We chose two constrained optimization problems to apply these algorithms to: the bin packing problem and the generalized assignment problem. While the bin packing problem is easier to visualize and solve because it is combinatorially less complex, the generalized assignment problem is more interesting because it is combinatorially more complex, and (not incidentally) has more practical applications.
Our project placed additional focus on two important parts of developing algorithms for constrained optimization problems: visualization and simulation. We developed a graphical user interface that allows the user to see algorithms solve problem instances in realtime, and allows the user to schedule simulations on a remote computer. We used these tools to run simulations comparing various simulated annealing and genetic algorithms, and compared their performances to previous work, which we hope provides some insight into what makes these algorithms successful.
Authors: Stephen Lin
Advisors: Zemel, Jay
Many researchers are currently studying the activity of infants in search of a relationship between energy expended at an early age and the increasing obesity epidemic. Collecting meaningful data in this area has many obstacles. Current data acquisition techniques are very complicated, costly, limited by age, and unnatural for the subject. The idea of monitoring an infant?s activity in a crib with the use of piezo sensors bypasses some of the obstacles found in current methods.
An array of sensors that cover the area of a crib mattress continuously measures changes in the force applied by the infant. A master-slave microcontroller configuration is used to sample each of the sensors. Each slave microprocessor samples a column of sensors and the master microprocessor collects data from each of the slaves. This master-slave configuration allows for easy expansion of the system to meet different sizing needs.
All data is serially communicated to a computer from the master microcontroller. At this point, the data can be processed on a computer and analyzed through a graphical user interface allowing the researcher to visually and analytically examine the data.
Authors: Alex Wagner
Advisor: Santiago, Jorge; Farnum, Phil; Deliwala, Sid; Castro, Sal; Ahn, Bloomfield, Phil
A pressure ulcer, more commonly known as a bed sore, is an area of the skin that breaks down when continuous force is applied to an area of the body. The constant pressure against the skin reduces blood supply to that area and can cause major damage if left untreated.
Currently, the only method of detection of these ulcers is the Blanch Response Test that most nurses and doctors use to test their patients. This rudimentary test does not work well for people with darker pigmented skin, and the ulcer frequently progresses to become more severe.
In the chosen approach, ultrasonic transducers are used to detect pressure ulcers by measuring a voltage difference between the impedances of normal and damaged skin. The proposed prototype of the project will be accurate and consistent, as well as portable and easy to use. Most importantly, the prototype would work just as well for people with darker complexions as those with lighter skin color.
The construction of this device will take several years based on the need to follow the very strict FDA regulation for the creation of medical devices. This year was an exploratory year for the project. The goal was to prove the theories behind the initial device design while laying the foundation for subsequent phases of the project. While eventually the project will feature a Force Sensor to detect pressure being exerted by the device and a compact data processing unit, we have begun by creating the driving circuit for the project, characterizing the transducers and proving their worth in detecting changes of impedance.
Authors: Roman Geykman
Advisors: Lee, Dan
As research into systems of multiple autonomous robots has increased in recent years, interest has grown in airborne robots. This project explores the feasibility of using an air vehicle as an experimental platform. A feedback control system for a prototype quadrotor chassis was designed and implemented using a PIC microcontroller and ground computer. Single-axis stability was achieved in tethered flights and results from further testing have provided information on the capabilities and limitations of the quadrotor platform.
Authors: James Hanna
: Piazza, Gianluca
Micro-Electro-Mechanical Systems MEMS use electrical and mechanical properties of micro-structures for the realization of compact and low power sensors and actuators. MEMS devices are now pervasive in many major industries, including display electronics and wireless devices.
The MEMS resonators for this project are made out of aluminum nitride (AlN), which has optimal mechanical and electrical properties for filtering and frequency-setting applications. AlN contour-mode filters offer reduced power consumption and increased space efficiency and the ability to span several frequencies on the same silicon chip. These devices can provide selective filter banks over a wider range of frequencies (MHz-GHz) than is currently possible on a single integrated chip.
We report the techniques used for designing the IF filter starting from piezoelectric contour-mode MEMS resonators. The devices have been designed using software such as MatLab and Cadence. A custom process was employed for the microfabrication of these structures
Authors: Tony Lombardo
Advisor: Lee, Insup
This project focuses on reducing energy consumption and increasing personal comfort by monitoring the conditions present in a room and adjusting the room’s environment to meet people’s desires. The goal of the project is to create a system to implement in a room, which can sense the presence of a specific person or a general person, and adjust lighting and temperature to fit the needs of those people.
This project was done using radio frequency identification to determine specific people and adjust the room to their specific conditions. There are also motion sensors to detect people who enter the room but do not have a radio frequency identification tag. In this project wireless sensor networks, or motes, are used to determine the current conditions of the room, and report these conditions to a base station. The base
station analyzes the information from the motes, and sends out signals to various operators in the room, to change the room conditions. The operators control the lights, blinds, and heating or cooling devices. The main objectives of this project are to conserve energy and make the system affordable and easy to install for the average person.
EE Home Page > EE Undergraduate Labs > EE 442 Home Page
> 2006-2007 Abstracts
ŠUniversity of Pennsylvania, Department of Electrical and Systems Engineering.
This site maintained and administered by Siddharth Deliwala, (deliwala_at_ee.upenn.edu ), Last updated Apr 8, 2007