GCC methods Notes FL2010

General Information

• Team Members: Chady Gemayel
• PhD Supervisor: Murat Akcakaya
• Faculty Supervisor: Arye Nehorai
• Goal: Implement Automated Mobile Source Localization Methods
  • Secondary Goal: Try and identify and explain erratic Labview behaviour
• Meetings
  • -Thursday 8:30 AM - 10:00 AM
  • -Friday 1:00 PM - 2:30 PM

---

Status

• As of November 11, we have accomplished the following:
  • Successfully coded and implemented the automation algorithm (9 weeks)
    • We initially had code that was able to find the sound source autonomously successfully only for one iteration
    • We realized that this was insufficient. The first iteration was in general quite innacurate, do to the inherent lack of resolution in the system. The solution to this problem was to have the platforms repeat this automation independantly
    • The first few weeks was spent implementing automation logic that would move the automated platforms through sufficient iterations until a certain threshold had been achieved
    • The majority of the time was spent troubleshooting this code, finding small bugs and errors that explained erratas in their behaviour, which included:
      • A malfunctioning global variable
      • Microphones plugged in incorrectly into the mobile platforms, leading to incorrect coordinate definitions
      • An excess in data computation on the mobile platforms leading to a read buffer overload
  • Created poster, prepared demo and presentation for UGR Symposium (1 week)

Chady_FL10_DSL720x480.wmv

• • Made video and short presentation of mobile platforms autonomously locating accoustic source (1 week)
  • We are currently implementing a simple cross-correlation algorithm for source localization. We now plan to extend this to generalized cross-correlation methods and implement different filters to try and improve the system's localization performance.