Inertial Navigation System - Has it Been Done?

[DragonRagnarok]

Hi, I'm the lead programmer for team 172. This year I've been working on using the gyro and accelerometer in tandem to keep track of our robot's exact location on the field. I'm using Kevin Watson's gyro code, and fully adapted all the functions for use with the accelerometer as well, as well as some higher-level functions that tie all the data together. The ultimate goal is that I can give the robot an x/y coordinate, and it will drive itself to that location on the field and be within a few inches.

I've been running into all kinds of problems, from malfunctioning hardware to flawed algorithms, but I'm at a point now where I think I'll actually have it finished before ship date.

So I'm just wondering, has anyone else ever done/attempted this before? What was your level of success, and do you have any pointers or warnings?

We've always tried to use the gear tooth sensors to do this in the past, but a combination of my then-inexperience and hardware problems always made it impossible. Gear tooth sensors however are inherently flawed because they only track motor revolutions and not you're actual motion. If you get knocked to the side the gear tooth sensors won't be aware of it. We've also tried to use the camera in the past, but that's been riddled with problems as well, such as lack of 360-degree motion, and the requisite line-of-sight.