Jordan is working on the Kalman filter to test the sensor plate with the WAM. He has developed a system model for the WAM and has coded a PERL script for moving Matlab models into C. The next step is to code up a rough State Estimator module.
Jordan, Chanop, David, Alex, and Thomas Brinsmead will meet to discuss the State Estimator and Kambara physical models.
Harley and David have been working on sealing up the leak in the lower cylinder. To locate the leak on the outside of the cylinder they positively pressured it to 2 atmospheres and soaped the outside. Before the leak could be found the endcap blew off. No damage done but little confidence should be given to the hasps that hold the cap on. Harley proceeded to seal in the interior weld with epoxy. Harley and David will put everything back inside Kambara on Thursday in preparation for tank experiments on Friday.
Luke fixed the MotionPak cable by switching a few lines. Luke and Peter have successfully sampled the MotionPak at 100Hz.
Peter continues work on the Sensor Sampler and can now sample all the various sensors, including the compass at 10Hz.
David continues to slog forward with the CompactPCI IndustryPack carrier board driver. Peter and David have been working on the memory mapping which must be designed manually with the Force board.
David reported that there are 7k of user NVRAM. Peter has pointed out that there are 4M of flash memory, some of which is used to hold the VxWorks kernel. More is available for user code or for storing data such as calibration and learned control parameters.
Chanop has a model of the vehicle for his controller using the center-of-mass, center-of-gravity, and inertia tensor derived from the ProEngineer model of Kambara. He is implementing a controller for the model and will work with Chris and Sunil to hook it up to a visualizer.
Chanop reported that the ImageNation PXC200F driver should be available next week.
Sunil has run the Operator Interface on the UNIX operating systems and had occasional SEGVs, which are unusual in Java which should trap exceptions. He will proceed to test things using JDB.
Sunil after subsequent discussions with David and Chris will break the Operator Interface into two separate programs both of which still connect to the existing Java server for their data stream. One will provide numerical and graphical representation of all telemetry. The other will simply display the 3D rendering of vehicle position and orientation. This should simplify the transition to from application to applet and may help to isolate the source of the crashes which are suspected to come from Java3D.
Matt has the camera calibration system working and is keen to complete in water tests. There are still some issues calibrating when the target appears in regions of higher distortion because of the invariants used in identifying features.
Matt will experiment with the calibrated cameras to characterize the accuracy of their position estimates. He will use both a long (45cm) and short (25cm) baseline and will conduct calibration and testing both in air and under the water.
Chris has concluded experiments with Nomad using sonar and infrared, marginal results, and has resumed work with vision sensors. A paper of this work is in preparation.
Chris will begun learning experiments with Kambara using the new 3D model when it is complete in the next few weeks.
David has finished a draft of the new software design. It will be published on the web site shortly.
Some items to do: