Research Summary (What I Do and Why It - PowerPoint PPT Presentation

1 / 10
About This Presentation
Title:

Research Summary (What I Do and Why It

Description:

Title: PowerPoint Presentation Author: todd Last modified by: administrator Created Date: 11/27/2001 11:30:37 PM Document presentation format: On-screen Show – PowerPoint PPT presentation

Number of Views:41
Avg rating:3.0/5.0
Slides: 11
Provided by: todd2167
Category:

less

Transcript and Presenter's Notes

Title: Research Summary (What I Do and Why It


1
Research Summary(What I Do and Why Its
Important)
  • DMS Lab Meeting
  • Todd Massengill

2
A Tale of Two Projects
  • Auditory Recognition
  • Comparison of Analog and Digital processing
  • Motivated by requirement for portable real-time
    sound recognition
  • (NSF grant)
  • AUV Obstacle Avoidance
  • Optical and Acoustic obstacle avoidance
  • Define signals and a common processing technique
  • Requires low power and functionality
  • (USTLab, OR Navy grant)

3
How It All BeganAuditory Recognition
  • Built a tunable discrete bank of bandpass filters
  • Compared this with similar digital spectrum
    calculation (not filtering)
  • Compared also with an analog VLSI filter bank
    (Paul Hasler, GA Tech)

4
Results Comparing Digital and Analog
  • Speaker and Context Independent Phoneme
    Recognition

Digital Discrete Analog Analog VLSI
42.22
58.33
39.72
5
The Next StepAnalog VLSI Processing
  • Ideally, each step in the recognition will move
    to Analog VLSI
  • Preprocessing (filtering)
  • Cepstrum (multiply, DCT)
  • Hidden Markov Model

6
How It EndsObstacle Avoidance
  • Project goal
  • Design low-power, dual-use signal processing for
    optical (image) and acoustic (sonar) data
  • Create a custom imaging chip (photodiode IC)
  • Work with USTLab to define the sonar signal
  • Create a custom image processing chip
  • Program a PIC microcontroller

7
Current StatusSmoothing Chip (20x20)
  • Fabricated with MOSIS 1.8?m process
  • Tested using LabView
  • Smoothing demonstrated in one dimension
  • On-chip decoder needs to be re-designed for
    smoothing in two dimensions

8
Current StatusImaging Chip (32x32)
  • Fabricated with MOSIS 1.8?m process
  • In Testing using LabView
  • Need to find proper bias voltages and integration
    times
  • Characterize response to varying light levels
  • Verify coupling between imaging chip outputs and
    smoothing chip inputs

9
Design Iteration and Demonstration
  • Design Iteration
  • Applied for grant from MOSIS for fabrication of
    smoothing and imaging chips on 0.5?m process
    (Jan. 22/28 2001)
  • Demonstrate Imaging chip in a pool, capturing
    images of the USTLab AUV
  • Demonstrate Smoothing chip with revised decoder
    and increased resolution (32x32)

10
Acknowlegements
  • Auditory Research
  • NSF Award ECS-9907464
  • Collaboration with Paul Hasler (GA Tech)
  • AUV Obstacle Avoidance
  • USTLab
  • Academic Advisor
  • Denise Wilson
Write a Comment
User Comments (0)
About PowerShow.com