[ahecht]

Honestly, I think that Autonomous mode is too short. 15 seconds is not enough time to do anything really significant or clever, and leads to way too many teams ded reckoning. There were only a handful of teams that actually followed the line using sensors or used the IR beacons. Even just expanding it an additional 5 or 10 seconds would help.

Also, open up the allowed electronics list, and make it more like the mecanical stuff (you can get anything under a certain dollar amount, except for certain items). Let teams use the CMUCam for object tracking, or digital compasses to find their heading. They're not expensive, just unavailable from the "approved suppliers." As it is, line or beacon tracking can be quite tedious.

The LPS (Local Positioning System) idea isn't bad, although it may be a bit pricey, but there are other ways of doing it that may be easier and cheaper (for example, RoboCupJr uses a field that is a giant grayscale gradient, so by looking at the floor, robots know how far down the field they are).

Eventually, I'd like to see the opposite of what we have now, with the first 1:45 being autonomous mode, and only the last 15 seconds being human controlled. We should either get away from the "130 pound RC cars (except for the first 15 seconds)" mentality and build real robots, or do what MIT's 2.007 competition does, and call it a design competition instead of a robotics competition.

I know people will say "think of the rookies" or "a typical view from team with one of the best autonomous modes," but I can assure you that if teams devoted the same amount of time to electronics and programming as they did to mechanical stuff, and designed the whole robot with autonomous function in mind, everyone could do it. Heck, I just went to a RoboCupJr competition where middle school kids built autonomous soccer playing robots with little to no adult help. If they can do it, surely we can too.