I was surprised the Dinsmore won out both in price and accuracy and size. At any rate, it seems that a robot can only use the direction as a wild confirmation against other sensors and orientation knowledge that your bot is going in a certain direction. IE: You really couldn't rely on the compass readings much for dead reckoning. I'm thinking that a robot that's trying to navigate a house would use either wall following methods or dead reckoning and use the compass just to confirm that a turn actually worked or not rather than to know an exact degrees it's facing. Is there anyone who's really relying on the compass as a primary means of direction indicator rather than secondary indicator? If secondary, is it really even necessary? :-/

Considering that a much more sophisticated and accurate technology like GPS (providing absolute positionning) does not even perform correctly indoor, I guess you sum it up: "use the compass just to confirm that a turn actually worked or not".

Nevertheless even as a secondary indicator I'd like to hear about the results, since I plan to use one myself.

BTW this also bring the question of the "what for" of these electronic compasses in the first place (outside the hobbyist robotic field): regular old-fashioned compass are much cheaper and perform better. So cool gyzmos, but what's the market?

I was going to rely heavily on a compass for directional control..

Already have the dinsmore (on the shelf and untested as of yet) and am dissapointed to see the reviewer make note of hysteresis problems. The product description said there was protection here so I had hoped to avoid this kind of thing entirely.

I wonder what a guy could do with a pocket compass drilled for infrared/phototransistor pairs or a large liquid filled compass with some type of encoder marks?