Meeting Summary
from: 01Mar00 at 3:00pm 
with: Harley, Luke, Alex, Matt, Peter, Chanop, Sunil, Jordan, Chris, David



Agenda
  • Status on individual activities

  • Harley is completing final details of Kambara's hardware. Harley and David localized the leak in the weld in the bottom of the lower cylinder. It is likely a pinhole and only leaks at depth greater than 1 meter with a flow at that depth of 5-10ml per minute. Harley and David will attempt to seal the leak with epoxy.

    The payload capacity of Kambara appears to be approximately 10.5kg. This is the available mass for computers, sensors, and anything else we might put in the upper enclosure. In it's current configuration, Kambara is positively buoyant.

    Harley has finished the ProEngineer model of Kambara and is calculating the center of mass, center of buoyancy and inertia tensor, for use by Jordan and Chanop in the vehicle model.

    Harley and David replaced both the composite video connectors which have gone bad (noisy video signals) due to wear.

    Luke has completed all of Kambara's internal wiring for computers and sensor. He has made sensor leads for mounting the sensor plate on the WAM arm. He will finish a few more leakage detectors.

    Matt is finishing up the camera calibration work and moving on to range estimation. He has worked out the details of estimating range from the full set of calibration parameters (misaligned cameras, radial distortion, etc.)

    Matt has put all the camera calibration (cameraCalib) and range estimation (featureTracker) code under CVS control. It is now possible to check it out under NT, Linux, or Unix. The port to Unix is underway with all the Microsoft base classes cut away.

    Peter is working on device drivers and currently has the IP-Serial and IP-ADIO drivers fully working. Peter and David are working on the SensorSampler module to collect data for the StateEstimator.

    Chanop has one implementation of a feature tracker working and has gone on to develop a simplified vehicle controller. He is currently developing the theoretical model. Most of the model parameters we can estimate but hydrodynamic drag may be difficult, so some simplifying assumptions will be made.

    A device driver for the ImageNation PXC200F board is still under development in Italy

    Sunil has been reading papers and will soon try to run the OperatorInterface. His first task will be to convert it from an application into a Java applet. Sunil, Harley, & Tim Edwards will meet to sort out transfer of the Kambara model from ProEngineer to 3DStudio to Java3D.

    Jordan has the model parameters for the sensor plate and the WAM. He must determine one more transformation to complete the Kalman filter for the WAM experiment with Kambara's sensors. Peter, Jordan, and David will work together to collect a correlated data set (rotational velocity, linear acceleration, compass heading, joint-space positions). Sampling should occur at minimum 10Hz, and faster up to 100Hz with only the compass unable to produce at this rate.

    Chris is ready to test the neuralcontroller on a working Kambara. He has also been performing tests with the Nomad mobile robot using sonar and infrared sensors. He is proforming a large simulation study of the sensitivity of various learning parameters to decide how best to set them for real-world experiments.

    David is working with Peter to finalize the overall Kambara software design. A complete revision of the original design is underway. He is also working on the IP carrier board device driver for Compact PCI. Next issue is the IP-68332.


    Active items to do:

  • Make pre-dive checklist
  • Determine how much NVRAM is available
  • Purchase chlorine, sunscreen, vent for shed
  • Add transistor to power cut-off circuit
  • Design and build battery immobilizer
  •  


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