Per our discusssions on Friday, I have discontinued my current work on feature
selection.  (Though its main usage, that of reducing computations for
registration may be outdated by the NRL algorithm, feature selection will
become important if the architecture is to be used in a dynamic enviornment
where we hope to pick out and use only the relevant, permenent features,
such as walls and doorways.  I didn't make that clear on Friday, but it's not
important for the time being.)  But, for right now I am focusing on working
more on the classifier.

Until I meet with Brian later this week, (since he will be doing more of the
future work, his preferences will dictate my main areas of focus for the
remaining weeks)  I will continue to work on my unsupervised classifier
In addition to the simple, incremental classifier that I have implemented, I
would also like to try various other algorithm (i.e., k-means, etc.) to see
if more sophistacated algorithm yields better results (within acceptable
computational confines).

Work for the remaining time (in order of priority):
1.  More with Unsupervised Classifier
	a.  Physical experiements with robot
	b.  New classification techniques
2.  Whatever Brian has in mind
3.  Previous Knowledge based on grids rather than place names
4.  Occupancy Grid Feature Selection (I will clear it again with Pat & Mehran
    if this comes up again)

Kinda short this week, since it's sorta of a transition week.