# Sastry NEST meeting notes
# Date: 3/5/2004
# Present: Phoebus, Cory, Shawn, Bonnie, Mike, Songhwai, Sagnik (undergrad
  looking for research)


- Testbed Hardware update (Phoebus)
  * Advanced Circuits miscommunication... did not receive PCBs this week
    ~ current design is to incorporate Telos/EPRB/Battery powering without
      using a switch... design by Rob Szewcyzk and Joe Polastre
    ~ still on prototype run
  * Soldering Iron & Multimeter came in

  
- Testbed Hardware update (Tanya, via Phoebus)
  * Bought PVC pipes... they are sitting in 330 Cory
  * Bruno, Shawn, and Cory bought the netting/pipes used down at Intel
    ~ got netting at OSH hardware
  * Should be able to get PVC pipes cut at hardware store when buying.


- Two new computers set up for usage by Mike and Bonnie in 330 Cory
  * Mike has Linux
  * Bonnie has Windows


- Space issue is mostly resolved in 330 Cory
  * Nest has entire space in middle of room... BEAR will move next to SurgiCal


- Robot Localization (Mike)
  * Got source code from the Denmark group.  Will try to implement.
  * Denmark code requires multiple cameras: 3 or 4 ideal
    ~ bandwidth problem: Denmark group had to use lower resolution cameras
      so can send data through USB
    ~ Cory and Shawn: How many frames per second?
    ~ Cory and Shawn: How are cameras synched up?
    ~ Cory: Are multiple cameras necessary because it is tracking
      3d, or because it improves resolution (less fringe effects from
      spherical aberrations of lenses)?
    ~ Cory and Shawn: Do cameras need calibration (6 degrees of freedom:
      3 camera positions, and 3 camera angles)?  Is there an autocalibration
      procedure?
  * Cory and Shawn believe they can get accuracy of around 1 cm with their
    old code as well... need to tidy up
  * alternative: multiple cameras focusing at different parts of the ground
    ~ switch cameras for different regions... then eliminate fringe effects.
    ~ Is camera going to work through the net?


- Cory wants Phoebus to draw up specs for Testbed: different from before 
  because specifying priorities of different requirements (see below)
  ~ we still haven't built anything, and it's been 5+ weeks.  Better to
    have a partially working testbed than no testbed at all.


- Bruno has prelims next Tuesday
- Bruno will look into feasibility of home-brewed indoor GPS with PDAs




Here's an attempt:

Specs for Testbed
*****************
High Priority:
1) Minimize useage of ferro-magnetic materials... keep them away from
   magnetometers
2) Sensors need to be easily reprogrammable
   ~ right now, it means wired nodes because network reprogramming is not 100% 
     reliable yet
3) Sensors are within range of robots
   ~ can't hang at top of ceiling 2 meters away from robots
4) Robots can run freely without obstacles in the way
5) Method of getting ground truth of robot position within 2 cm accuracy
   and at 2 Hz update rate
6) Gathering of metrics will minimally interfere with performance of network


Middle Priority:
1) Sensor Nodes can be powered continuously, not off battery
2) Can relay metrics back in "real time" while experiment is running


Lower Priority:
1) Can raise and lower sensors if we suspend them in the air easily
2) Can integrate and test Telos Mote with Mica2Dot sensors
3) Don't "Trap" sensors behind plexiglass, etc.
  ~ may lead to more radio multipath, less sensitivity of sensors