Meeting Summary
from: 2Mar99 at 9:00am 
with: ChrisG, Harley, ChrisM, Chanop, Ian, Ken, Terence, David, Samer, Alex



Agenda
  • Select new meeting time
  • Status on individual activities
  • Select next week's reading



  • The new, new meeting time is Wednesday at 9:00am



    Chris Gaskett has been working on slides for upcoming seminars and conference presentations.  He is also been thinking about how to fix up the external power supply.

    Harley has been collecting information on Kambara's various sensors while he wraps up work on the WAM arm.  Harley and Terence will coordinate to share information.

    ChrisM (as well as Ian and Terence) are working honors thesis proposals.  They'll discuss project definition with David next week but are largely on their own to set objectives and lay out a schedule.  The proposals are due March 12th.  Chris has been reading papers, surfing the web, and talking with Ken and David.

    Tim has been alternately reading vision papers and working up an interface for image processing.  He'll take a look at existing correlation libraries and see what they are doing.

    Ken, ChrisM and David will meet at 2:00pm to discuss web-based operator interface possibilities.

    Ian has been working on the VxWorks driver and now has it switching between 8 buffers in an attempt to avoid overwriting the buffer while trying to copy it.  The difficulty is that the image is transferred into memory by DMA generated by a RISC processor on the framegrabber.  It is difficult to control this process once it starts going.  DMA transfer is about 8 times as fast as a block memcpy so images were getting fragmented.  After a bit of group discussion in which the speed of DMAs was explained (they don't use the processor) Ian was convinced to investigate stopping and starting the RISC processor to avoid overwrites.  A double buffer system may be appropriate.

    Terence is reading about Kalman filters and getting familiar with Kambara's sensors.

    David has gotten the various IP modules to initialize and access memory-the ISRs are still incomplete.   The "experimental" kernel is now consolidated back as the working one.  Camera enclosures are off to be anodized.  No one identified to paint the upper enclosure yet.  The battery charger status is still unclear.  David, Ali and Chris met and all the power amplifier and DC/DC converter designs look good.  Chris will finalize and the circuits.  We will then iterate on the box design and then send both board and box out to be fabricated..  Wayne Dunston will finish up the benchtop power supply and begin work on servo control of the thruster motors.  We need to review how many A/D lines are needed as we may be running short.

    At the end of March there will be number of visitors coming through so we will assemble Kambara's enclosures and motors and make it look good.



    Papers to read for next week:

    "Reinforcement Learning applied to the control of an Autonomous Underwater Vehicle" by
    C.Gaskett, D. Wettergreen and A. Zelinsky, in Proceedings of ACRA99, (Brisbane, March 1999).

    "Development of a Visually-Guided Autonomous Underwater Vehicle" by D. Wettergreen, C.
    Gaskett, and A. Zelinsky, in Proceedings of OCEANS98, (IEEE, Nice, September 1998).


    Kambara Project at the ANURobotic Systems Laboratory
    By David Wettergreen, <dsw@syseng.anu.edu.au>
    Last modified: