So restarting this whole RoboCup project makes things much more daunting. Last year, time was crunched and things had to just work, and so we were compelled to do things the quick-and-dirty way and actually make the robots play soccer.
This year, however, there is plenty of time to do more complicated shit.
This is a good thing and a bad thing. Which brings me to the topic of reading and/or doing.
Right now I’ve been reading all the surrounding research related to RoboCup which contains quite a few big and exciting concepts: Extended Kalman Filters, conditional probability density functions, Forward (and not to forget Inverse) Kinematics, and the ever-elusive Real-Time Shared Obstacle Probability Grid Mapping for Mobile Swarms (RTSOPGMMS).
So I talked with Chown about this today and stipulated that I could spend days, weeks, or even years familiarizing myself with the vast amount of research done or attempted inside the RoboCup universe. He made two good points:
1) It is a good thing to read and know what’s out there [assumed].
2) Grad schools are riddled with students who try to read everything and accomplish little. There is a massive amount of work out there, new work is constantly evolving, and linkages to your own work are constantly forming.
Last year was all about doing: I spent little time focusing on anything programming/research-related if it didn’t produce results for me instantly. This year has to be a combination of both, but without losing sight of the overwhelming purpose.
What’s the purpose? To nearly win the RoboCup 2007 4-Legged Championships, of course. Read and Do. Read and Do.