EE Home Page > EE Undergraduate Labs > EE 442 Home Page > 2003-2004 Abstracts

Authors:    Alex Chen
                 Albert Oh
                 Michael Pascual
                 Xi Zhou

Advisor:   Deliwala

Abstract:
The purpose of this project is to utilize the latest in wireless
communication technology, Bluetooth and Wireless Fidelity (WiFi), to add
portability and real-time interaction to marketing surveys. Personal Digital
Assistants (PDAs) will be the hardware devices which give portability to
product rating programs. Bluetooth will facilitate transmission of marketing
surveys between separate PDAs while WiFi permits a web-server to congregate
data from all surveying devices. These two technologies will be integrated
so that fashion models can carry PDAs with them which broadcast one-touch
rating forms for users with PDAs to dynamically rate their attire

Coordinated Chemical Sensing System

Authors:    Ray Ciarcia
                Brian Corwin
                Paul Vernaza

Advisor:   Lee

Abstract:
The Coordinated Chemical Sensing System is an attempt to address some of the challenges currently facing the field of robotics. The final product is a team of Sony AIBO robot dogs, outfitted with chemical sensors, that are able to enter an unknown room and map the concentration of a particular chemical with minimal human supervision.
Such a system could be useful in tracking dangerous or illegal chemicals.
Key to the development of such a system is the development of robust localization techniques: when a robot detects a chemical "here", how will it know where "here" is in the absence of prior knowledge regarding the room? Our system uses a combination of visual markers carried by each dog and auditory signals between dogs to establish relative position estimates. Special search patterns are employed to aid in translating relative position to room-based location and to maximize the rate at which a room can be searched.
Also important has been the development of hardware and software that can extend the functionality of the Sony AIBO to allow for input from external sensors for which the dog was not originally designed. In developing these components, we have created a platform upon which a variety of sensor applications can be built using the AIBO.

Active Camera Calibration

Authors:    Ed Hesselbacher
                 Sandy Patterson

Advisor: Kostas Daniilidis

Active Camera Calibration is a method to measure the extrinsic parameters of
a camera in an automated, accurate manner. Current methods of 3D scanning
involve either expensive laser scanning systems or calibrated cameras.
All methods for calibrating cameras require extensive user inputs
and lots of time spent on the project, or they involve a random coordinate
system for the camera that does not translate easily to a useful world
grid system.This calibration system is an automated method that utilizes the "survey
method" to produce a calibration matrix. Two lasers are used to paint a
spot on the scene with a red dot. The location of the spot is knows from
the two trajectories of laser light, and the spot in the scene can be
easily found on the camera plane with simple image processing. This
process is repeated for many spots until the error metric reaches a
certain threshold. This technique can potentially be accurate and
inexpensive, and it would be paired with an low-cost digital
camera to create 3D point cloud data for the scene in question

Bluetooth, Internet and Cellular Communication System

Authors:    Rudolph Broomes
                 Melanie Mbida
                 Raymond Lowe

Advisor:    Alexis

In recent years, Voice over IP programs have become an area of growing interest. Programs such as America on-line instant messaging (AIM) or MSN instant messenger enables users to have voice conversations over the internet with relatively good quality.

In order to conduct these conversations users require a speaker and microphone configuration. Inconvenience from such a configuration can be caused by feedback and limitations in the distance the user has to be from the computer. The Bluetooth Internet and Cellular Communications System (BICCS) proposes to address these issues by enabling these conversations to be transferred to a landline via a central signal processing unit.

By using Bluetooth technology, audio will be streamed from a personal computer to a landline. The central signal processing unit will be physically connected to the landline, while the personal computer will be wirelessly linked via a USB Bluetooth adapter. BICCS software should be able to produce a reasonable solution to these areas of concern.

Authors:    Briseyda Morales
                 Rani Nazim
                 Aeshna Sikand

Advisor:    Kumar

Abstract :
The Pan-Mount project is a small piece, or one module, of a much larger project being carried out by the GRASP lab (the MARS project).The purpose of the Pan-Mount project, is to create a motor rotation system attached to a camera which has a certain angle accuracy and has position feedback capabilities while meeting certain weight and cost requirements and is able to survey an unstructured and unknown environment, return data from the camera, and dynamically adjust to exterior changes. Also, since this project is a small piece of a much larger project, we need our design to be flexible and modular, so new devices can be added on and changes made without the need of redesigning the entire system. To achieve this purpose we are using CANbus as the communication protocol.

For more information on the MARS project please refer to:
http://www.grasp.upenn.edu/destudio/CAN/

Ultrasonic Position Sensing for Location-aware Applications

Authors:    Giridhar Nandipati
                 Aunim Hossain

Advisors:  Lee,

Although technology has enabled devices to have increased portability,
customization, and computing power, there is no standard way of determining
the position of these devices in indoor environments. This project aims to
implement a system enabling indoor position determination for a handheld
computer, and interface this system with an application requiring such
information. By using time difference of arrival measurements between an
ultrasonic receiver and various ultrasonic transmitters stationed around a
room, position relative to these transmitters can be computed, and interfaced
with a software application. The application that will demonstrate the
system’s functionality is intended to be a graphical program displaying the
handheld computer’s position relative to the stationary ultrasonic
transmitters in real time.
 

Bluetooth Smart Remote

Authors:   Michael Lyons
                Sumati Mathur

Advisor:   Lee

Abstract:
While there is constant innovation in the electronics industry, remote
controls have lagged behind. Devices are becoming more rich in features,
yet their remote controls have remained a mere set of buttons. While
today's devices provide more information and functions, it is time
remotes offered the same. The Bluetooth Smart Remote will be both
powerful and easy to use. The remote will feature a customized menu for
each device it controls. In addition, information from the controlled
device can be dynamically displayed on the remote's touch screen.
However, with this new functionality, the remote will still be simple
and easy to use. All controlled devices will be automatically detected
and configured.

MEMS Oscillator

Author:    Lauren Ohm
               Jonathan Ozark
               Alan Powell

Advisor:   Sargeant
 
Abstract:
FBAR technology provides filters of exceptionally small size, low cost, low
insertion loss in the 1 to 2 dB range, good power handling capability, high out-
of-band rejection, and stability over a wide range of temperatures. Five
years ago, oscillators for RF use were entirely comprised of crystals.
Crystals, however, were physically limited from oscillating at high frequencies
over a few hundred megahertz due to the stresses of its physical oscillation.
In order to make the quartz crystals oscillate at high frequencies, they had to
be cut thinner and thinner, which also caused excess noise to be created from
the oscillation, also limiting their effectiveness. FBARs provide a solution
to this problem, as the oscillation of the aluminum nitride contained in the
FBAR could withstand the rigorous stresses of oscillating at 1.9 GHz. In order
to design an oscillator, it is necessary to first determine the characteristics
of the FBAR that is to be used. This is the first task for the project. Once
the physical parameters of the FBAR can be determined by both measurement and
referencing Agilent’s values, various simulations of the FBAR must be done to
better understand the technology and its properties. Finally, after becoming
intimately familiar with the technology through testing and simulations can a
final oscillator be designed. This is this approach that will be taken with
the project.

Auto-Scoring Tae Kwon Do Equipment

Authors:    William Barrow
                 Jason Lam

Advisor:    Mr. Lou Giamo

Tae Kwon Do is an Olympic sport in which competitors participate in scored sparring matches. Because the scoring system is point based, erroneous scoring is quite common with the current judging system. This project seeks to eliminate the judging system altogether by inserting sensors and wireless communication within the sparring equipment, with the intended result being a set of equipment, comparable in functionality to the original, which detects and scores all points through pressure sensors, and transmits the information to a central control unit at the center of the ring; this central control unit will serve as the interface between the scoring unit and the display and control function components.

Wireless Cable Television System

Authors:    Ankit Kapasi
                 Matt Fano
                 Andrew De

Advisor:    Kassam
 
With the expansion of Wi-Fi technology and the integration of microcomputers into a variety of devices, the need for wires is beginning to disappear. According to Time Magazine, this technology has been saving consumers an estimated $100 million a year in reconstruction costs associated with installing a wired network. For this reason, wireless networks have the greatest growth rate when compared to any other network technology in the market.

With 98 percent of American households containing at least one television set , the television has become most abundant entertainment appliance within America. By integrating wireless technology into the television system, the need for a co-axial cable to receive a cable signal is no longer necessary and costs with rewiring buildings for cable television are eliminated.

The Wireless Cable Television System takes an analog cable signal and demodulates it in order to receive a single frequency band corresponding to a one channel. This signal is then digitized using a MPEG codec for compression and is sent over the 802.11 wireless network. The signal is then received over the network and outputted into an analog composite video signal. This system can be implemented in a wide variety of situations and has great potential for marketability.

Sleep Apnea Detection and Scoring Neural Network

Authors:    Adam Mihalik
                 Jeff Katencik

Advisor:   Farhat
 
Abstract:
This project aims to determine the viability of sleep apnea detection
with use of Neural Networks. By using sleep data that has been previously
scored, a network will be created and trained to give the same scores as the manual scoring method for this data set. The network will then be operational, and will have the ability to score other sleep data inputs. This neural network program will be created in Neurosolutions 4.0. The program will have an interface which will allow sleep doctors to conveniently use the program and determine the presence of apnea. Designing a successful program will have several possible far-reaching effects. Sleep doctors will be able to more efficiently diagnose sleep apnea, a disease as common as diabetes in the US. Sleep clinic visits will become less costly, and therefore allow more patients to be tested.

Dynamic, "Software, and Hardware Development of an Enviornmental Test Chamber for Nano-Wire Gas Sensors."

Authors:    John Eanes
                 Jeremy Law

Advisor:    Evoy
 
Abstract:
Gas sensing has been an area of great interest in the field of sensor technologies over the past few decades. There have been many novel approaches to producing gas sensors using various oxides in different configurations, yet none have met with great success.

Gas sensors must be fast reacting, selective, and sensitive in order to be considered useful. Up until now, there has been a trade-off between selectivity and sensitivity, especially with tin-oxide sensors. The purpose of this project is to enhance the productivity of experimentation on new compounds that may replace and/or exceed the limitations of current oxide based sensors via the creation of a user-friendly test chamber and data synthesis software system.

The system will control temperature, gas flow rate, and frequency of impedance measurement in order to allow the user to quickly and easily identify the characteristic response of a certain gas in a certain concentration. The software will quickly and easily coordinate the running of the experiment as specified by the user and allow for the collection and display of data in an intuitive fashion. The system will be robust enough to handle varying gas species at differing temperatures along with multiple types of sensor compounds

Virtual Reality Glove

Authors:    Yao Hua Ooi
                 Bryan Chao

Advisor:    Lee

Abstract:
The virtual reality glove is a hand-worn apparatus that measures hand motion. Hand motion captured by the virtual reality glove will be translated into commands for the Sony AIBO dog. Specifically, the virtual reality glove will measure finger bending, movement of the hand in 3-D space, and hand rotation. This is accomplished using a system of sensors including flexible resistors, electronic accelerometers, and electronic gyroscopes. Data from the sensors are interfaced through a data acquisition board and are analyzed and interpreted into commands using MATLAB.

Sensor-Assisted Vestibular Rehabilitation System

Authors:    Anand Allam
                 Sam Fort
                 Amneet Sandhu

Advisors:   Solomon, Deliwala
 
Abstract:
The Sensor-Assisted Vestibular Rehabilitation System (SAVR) is a take-home rehabilitation device that implements exercises for patients with vestibular illnesses. SAVR improves on current crude rehabilitation methods based on the proven Cawthorne-Cooksey method of therapy while providing the moderating physician with exercise control and feedback. Also, SAVR provides effective options to achieve optimal recovery through a variety of exercises and the ability to create exercise routines. Further, the SAVR system is equipped with a modem for file transfer of exercise results to the doctor. This feature of remote communication couple with portability provides an effective method for doctors to refer to patients with vestibular illnesses in an inexpensive package. The systems implementing SAVR can be split into three subunits: (1) exercise: real-time communication between board and microcontroller; (2) data transfer: communication between microcontroller and a Palm; (3) data transmission: communication between the Palm and the therapist’s PC. The patient wears a laser apparatus on his head and points it towards a board (wall mounted or free standing) cover with 9 laser receptors (solar cells) and LEDs. The board is attached to a microcontroller, which directs head movements by illuminating various LEDs and stores performance data. This data is sent serially to the Palm, which in turn sends the data via the Palm modem to the therapist’s PC. By providing the patient with an easy to use device and the therapist with remote exercise control and feedback, it is anticipated that SAVR will reduce doctor office visits and improve recovery time.

Smart Wheelchair Via Voice Control

Authors:    Andrew Golder
                 Evan Morse

Advisor:    Kumar
 
Abstract:
The GRASP lab Smart Wheelchair research project has been an area of intense study over the last year. The goal of the project is to utilize the most current technologies to construct a wheelchair with smart capabilities by integrating sensing equipment with the human control of the wheelchair. It was a concern of the design team that the manual controls currently in place on the Smart Wheelchair could be difficult to operate for users with disabilities affecting the control of upper extremities. Thus it was suggested by the Voice Recognition team to install a voice recognition control system to alleviate this potential problem.

The task before the Voice Recognition team is to create a software module able to take voice commands from a potential user and convert them through a control algorithm into the inputs to the motors (V, ?). To do this it would be necessary to acquire a software development package with the necessary capabilities and develop the software. This software then needs to be integrated with the current Smart Wheelchair framework in place using the ROCI (Remote Objects Control Interface) system.

The Voice Recognition team acquired the Microsoft SDK 5.1 software development kit, which has the necessary capabilities. The team will be using the Speech Recognition (SR) engines provided by the SDK to interpret voice commands. The control algorithms for smooth movement are an incremental system, thus the current state of the velocity parameters (V, ?) will be updated based on the specific voice commands. This allows the developer to use a small number of commands to create a fluid and flexible control motion, while also maintaining a short training period and ease of control for the user.

ENIAC on an FPGA

Authors:    Brian Falk
                 Jennifer Hall

Advisor:    Van der Spiegel

Abstract:
The ENIAC (Electronic Numerical Integrator and Computer) was the first
large-scale digital computer. It consisted of 18,000 vacuum-tubes contained
in 40 modules, which were interconnected by a series of buses. This
implementation aims to recreate a subset of those modules on Field
Programmable Gate Arrays (FPGAs), to be used as a prototype for a full-scale
implementation. External interface boards were designed with the intention
of preserving the original feel of ENIAC, by relying on cable connections
and switch settings for programming. The project is a collaboration between
two groups. This part of the project focused on the constant transmitter,
master programmer and multiplier.

On-line Laboratory Notebook

Author:    Ralph Pfeifer
               Gopi Unnithan

Advisor:   Santiago

Any student that has participated in a laboratory class has undoubtedly needed to keep a lab notebook. Typically, a lab notebook consists of a bound notebook into which students record data from a lab session with a pen. If the student needed to append an image to the notebook, he would print it out and tape it onto one of the pages.

With a notebook stored digitally, the student no longer needs to use pen and paper to record lab data, and the process of taping external information is no longer needed ­ the student can include an image or other form data inline with his recorded results. Several implementations of a digital notebook have already been developed, but they are designed to accommodate one specific subject of study, or are in the form of a program that must be installed on the user's computer.

In the chosen approach, the lab notebook program is stored on a web server, and any student with Internet access has available the full capabilities of the program. In its current form, a student may login to the server, view tasks on his calendar, and create, view, and edit his labs. Site administrators can add, remove, and edit students and other users.

ENIAC on an FPGA

Author:     Albert Kim
                Fred Li

Advisor:    Van der Spiegel

The ENIAC on a Chip project involves designing and interconnecting the
components of the original ENIAC on Field Programmable Gate Arrays
(FPGAs). The original ENIAC was built in a modular fashion and consisted
of 20 accumulators, a cycling unit, a constant transmitter, a multiplier,
a divider/square rooter, and function tables. The project goal is to
implement the cycling unit and accumulator on FPGAs using Xilinx software.
Then, all modules will be interconnected to user controls and a central
printed circuit board. The end interface will resemble the original
ENIAC, effectively bringing the ENIAC back to life to serve an important
historical and educational purpose.

Robot Arm for Children with Disabilities

Authors:    Nimish Verma
                 Devang Shah

Advisor:   Kumar

Children with disabilities are often confined to life in a wheelchair, facing simple motor functions such as opening a door and picking up a book with difficulty. One approach to this problem is designing a robotic wheelchair arm that can be controlled by a simple game pad or joystick, simplifying unnecessary difficulties from an already challenging life.

Previous attempts by the Moore School of Engineering to design a system for those restricted to a wheelchair have struggled with overall ineffectiveness. These efforts were unsuccessful in designing a free range arm system that could mimic the movements of a human arm through the manipulation of a joystick. Earlier setbacks were due mostly to a failure to incorporate the kinematics of the robotic arm into the system, resulting in erratic movement before the arm reached its desired position, and only being able to move in the x or y axis at one time.

Learning from the problems of earlier approaches, understanding and focus on kinematics is essential. This project required comprehension of how to interpret a desired x, y, and z position into the proper arm motion in order to set it apart from the previous disappointments. In the chosen approach to redesigning the control system, the user sends the intended arm position and speed to a microcontroller through joystick manipulation. This data is then converted to the specific motor positions of the arm through the implementation of an inverse kinematics algorithm. Motion control chips employ a loop control to compensate for a possible load on the arm, ensuring it reaches its intended position

Virtual Interfaces

Authors:    Jared Miller
                 Sudhir Shaunak

Advisor:    Lee

Living in the information age, the ability to stay connected while being highly mobile is crucial and necessary. Certainly cellular phones and personal digital assistants allow people to be highly mobile, while being able to interact with people over voice, or accessing stored data on PDA’s. However, there has been a drastic change in the cellular phone and PDA market to allow for 3G networks which provide higher data rates so that one can transmit more than just voice wirelessly. A major issue concerning cellular phones, or PDA’s and their ability to stay in connected is that of data entry. There is no good system for entering text, or commands to cellular phones or PDA’s. Since the purpose of the devices is to be small and mobile, a regular bulky input device such as a keyboard will not suffice. For this reason it is necessary to develop better, more compact and effective ways for data input. This is the motivation behind building the Virtual Interface Keyboard.

The approach chosen to develop a keyboard that is highly mobile and very compact is to design a system that makes use of a virtual display - an image of a keyboard projected on a table. By using this image as the typing surface, one does not need to carry a keyboard, but rather simply have some table space to work on. The unit will consist of a projector, and infrared distance sensors. The sensors will detect the distance of a finger when it touches the plane of the table. By then using the distance reading from multiple sensors to detect exactly which point the users finger is on, a micro controller will calculate the key being pressed and relay that data to an input device such as a cellular phone or PDA

Aquanet - Low Power Water Analysis Network

Authors:    Sameer Qudsi
                 Ramu Pratiwadi
                 Sandip Saha

Advisor:    Zemel

Oil spills are an immense threat to the aquatic environment. Even
though these events are sparse, the subject of detecting these oil spills has
sparked interest from an environmental point of view. Oil spills, though, are
not all on the scale of Exxon Valdez, where 11 million gallons were spilled.
Some of these spills are simply leaks from offshore oil platforms. However,
no matter the magnitude of the spill, the safety of aquatic life and the
prevention of water pollution must be protected. Technology has evolved to
detect spills as they occur, so as to minimize leakage and response time.
The current procedures of oil spill analysis vary from using satellite
photographs to using a helicopter to immerse giant sample buoys into the water
body and absorb liquid samples. Unfortunately, these current methods are
costly and long processes. In addition, snapshots from satellites are unable
to capture information regarding the concentration of the oil in different
regions; for example, it is difficult to tell from these photographs how much
spillage is near a shoreline. The lack of such data prevents the use of
appropriate action on the part of environmental scientists regarding where to
deploy cleanup resources.
AquaNet provides a reliable, cost-effective approach to detecting, and
providing valuable data including the thickness of an oil spill in a real-time
environment. This method involves using small underwater transducers to
generate an ultrasonic pulse and receiving the reflected pulses, one from the
oil-water boundary and the another from the air-oil boundary. The signals are
then time-stamped by the processing device, which forwards this information to
the networking module. The network is used to transmit the sensor information
to a central computer with data acquisition capabilities. Hence, the user now
has real-time information regarding the concentration of oil spills.

Online CD Authentication System

Authors:    Robert Lavan
                 Jay Shek

Advisors:   Sarkar
 
Abstract:
The Online CD Authentication System will add an additional level of security and authentication to online transactions. In the chosen approach, information from individually created CDs will provide a level of physical authentication over the Internet. This approach will use the pattern of errors that are unique to to individual CD's controlled by the CD maker. These patterns will form individual keys, in combination with software on the CDs to uniquely identify the user over the Internet. The informatio present on the CD's will be tested and verified by a backend MySQL server. The full implementation of this system will allow for the Internet to no longer be a faceless realm of users. Websites, users, email, and many other electronic services will be able to ensure the originality of the user. This will in time allow for websites to decrease stolen services, shared passwords, and add another level of security to a growing security conscious society.  

Speech-Based
Controller-Pilot Data Link Communications

Authors:    Sith Chaisurote
                Toon Tantimekabut

Advisor:    Lee

Abstract:
The future air traffic control communication will most likely rely on satellite data link, and the Controller-Pilot Data Link Communications (CPDLC) is built for this purpose. However, the current system only permits text-based communication, which deters pilots and air traffic controllers from using the system. Therefore, the implementation of speech recognition and speech synthesis on top of the CPDCL system should serve to eliminate this problem of change in the communication mode.

The Speech-Based CPDLC, an extension on CPDLC system, will be divided into three sub-systems: Speech Engine, Data Management, and simulated CPDLC Network Link Implementation. For the speech engine implementation, control application and grammar construction will be developed on a commercially available speech recognizer to make it specifically tuned for the air traffic control vocabularies. The converted text will be prioritized and formatted according to the CPDLC context through the data management system, and the packets will be sent through the simulated network link, proxy of a satellite communication. At the receiver's end, a commercially available speech synthesis will be used to convert the text to speech.

It is anticipated that the speech recognizer will be able to recognize messages correctly with 95 percent accuracy. The message should be compressed, prioritized, and sent through the network without error. 

Viscositu Sensro for Slurries and Other Complex Fluids

Authors:    Sidney Lee
                 Marc Tarlock

Advisors:   Santiago
 
While viscometers have been vital in a myriad of industries ranging from
chemical processing to wafer fabrication, the tremendous growth witnessed by
these industries has not resulted in a corresponding improvement in
viscosity measurement methods. Current measurement techniques are based on
dated concepts requiring a high level of technical precision and fluid
intake. These techniques have been honed to a high level of accuracy over
time, but have remained expensive and inefficient. The goal of this project
is to produce a relatively inexpensive and accurate viscometer capable of
evaluating slurries and other complex fluids based on the concept of fluidic
oscillation.
The viscosity sensor described is anticipated to provide viscosity
measurements accurate to ±5% over a measuring range of 100-105 mPa.s using a
sample volume less than 20mL. Viscosity will be measured using the principle
of fluidic oscillation with an expected oscillation frequency ranging from
100 Hz to 300 Hz for the stated measuring range. A pulseless, constant
pressure pump is used to provide steady fluid input while a single pressure
transducer senses pressure fluctuations. A summing amplifier will be used to
adjust the output signal for processing. LabView software is used to analyze
the output data via a data acquisition card

Non-Invasive Deep Brain Tissue Temperature Sensor

Authors:    Julia Lee
                Emily Fox

Advisor:   Britton Chance

Abstract:
During Cardiopulmonary bypass surgery, the patient's core body temperature is lowered to provide
better surgical conditions. Following surgery, resuscitation of the patient includes raising
the body temperature back to normal levels of around 40 °C. While successful methods exist for
monitoring superficial tissue temperature, a precise and efficient way of monitoring core tissue
temperature, specifically deep brain tissue, has not yet been established.
Through the use of Near Infrared Spectroscopy (NIRS), a relationship between absorption of light
in deep brain tissue and its temperature can be determined. By measuring the difference of
light absorbance at two separate wavelengths, in this case 900 nm and 970 nm, it is possible to
determine the precise temperature of the tissue.
For the chosen approach, a probe is constructed and applied to the patient's forehead. It
houses a light source and two photodiodes that detect the absorbed light at the wavelengths
specified above. This light is then converted to an analog voltage signal that is sent to a
computer to be processed, outputting the temperature of the deep brain tissue on a display
monitored by the surgeon.

MONITORING ATTENTION – OLFACTION RESPONSE IN THE HUMAN FOREBRAIN

Authors:    Matt Kryshak
                 Kunal Dharia

Advisor:    Dr. Britton Chance

The aim of this project is to design a self-contained, inexpensive, portable device to measure hemoglobin levels in the primary cortex region of the brain by adapting the basic hardware and software concepts used in devices that measure hemoglobin levels in other regions of the brain. The prototype uses two oscillating near-infrared diodes of different wavelength to shoot photons into the primary cortex; any reflected photons are collected by two photodiodes and the impending output voltage is amplified before being processed by a data acquisition board. The data is analyzed using a modified Beer-Lambert law and is displayed using National Instruments LabVIEW. The anticipated result of the prototype is the ability to detect noticeable changes in hemoglobin levels in the primary cortex in the presence of Vanilla as a control substance

Artificial Cochlea

Authors:    Arthur Cao Yuan

Advisor:    van der Spiegel
 
Abstract:
Current speech recognition technology has limitation in accuracy and complexity. The recognition systems currently in the market and being used are difficult to recognize different voice, accent and dialects without sufficient training. Thanks to Ahmed Ali’s research based on the GSD (Generalized Synchrony Detector), there is a new approach to process human speech through the acoustic – phonetic characteristics of speech.

This project is aiming to provide the software implemented digital filters using Matlab to feed analog signals to upenn neural network for further process. The human voice, as the input, will be divided and processed into 16 parts based on different frequency range via 16 digital band pass filters. After the multiplexing, 16 sampled signals will be multiplexed and output from the 2-channel sound card serially. With the same clock, real-time signals will be converted to analog signals first and then demultiplexed and finally fed to the existing university of Pennsylvania for further acoustic-phonetic characteristic process. In addition, GUI will also be developed to customize the filters and control the real-time function. Users can visualize the change of amplitude and frequency of the human voice in both time and frequency domain and the data flow in the real-time also through the interface.