Harley has reassembled the power relay box, just for fun.
According to Harley's physical model (images available at /home/sub/data/3dmodels), Kambara is 117kg and 113 liters. Early tests showed Kambara to be slightly positive, so a dive with the full-up configuration is needed to verify.
David and Matt have measured the current rate of leakage through the weld in the lower enclosure at 150 milliliters per hour (2.5ml/min) at a depth of 1.5 meters. No leaking occurs at 1 meter depth.
Chanop has completed his physical model of the Kambara. Drag coeffients are still reasoned approximations (guesses). His controller is running but the settling time is 10-20 seconds so only very slow motion is possible. He has tied OpenGL to this simulation to visualize the results.
The PXC200F driver has arrived and is working under linux, Chanop confirms. Peter has installed the VxWorks patches and the framegrabber now initializes (but no pictures yet).
Chris and Luke submitted a paper to IROS on their work in reinforcement learning.
Chris will, next week, repeat earlier experiments in learning 3D motion for Kambara using his improvemented learning methods.
Luke completed the power reset line for the controller. He will check proper grounding of the two enclosures and the computer rack to the frame of the sub. The power connector to the computer rack will be replaced when a new type of connector arrives.
Sunil has applet and application versions of the Java operatorInterface working. He will meet with David to discuss the design. The form of the 3D display, OpenGL or Java3D, will be resolved.
Peter has installed Tornado2 and the PPC BSP to upgrade us to VxWorks5.4.
Peter has given Tim the necessary code to integrate into the WAM controller. He has installed the iPADIO and iPSerial modules in nozomi so that the sensors can be sampled.
Jordan has discussed with Tim the sampling approach and confirmed that data can be recorded a 1kHz. He has been coding routines needed to test data. Tim is making quick progress and testing with the WAM is expected next week.
As soon as experiments are underway, David and Jordan will take photographs of the sensor package.
Luke and Jordan will test and adjust the RC filters on the sensor input lines to clean up noise. Additionally Jordan is working on implementation of a 4th order Butterworth filter in software.
Matt has completed work on camera calibration. Underwater experiments have shown error of only 1% in estimates of object dimension and depth at a range of 1 meter, 2% at 2 meters, and 5% at 3 meters.
David has drafted a pre-dive checklist linked to the Experiments page on the Kambara website, which everyone should read. Make suggestions or edit the page directly.
David has designed a battery mounting bracket. Harley and David will refine and manufacture (made from a sheet of cut & bent Aluminum) and install it.
David has tested and checked in a working version of the industryPack driver for PPC. Interrupt signaling and vectoring are working.
Some items to do: