On Tue, 2010-05-25 at 17:10 +0200, Markus Achtelik wrote: 
> I think, unfortunately except for the magnetic compass, attitude,
> acceleration and angular rates are already covered by the imu- message
> in sensor_msgs. As imu readings usually arrive/are needed at a higher
> rate than gps, one would send duplicates of the GPS measurements and it
> would be harder to determine a "real" new GPS measurement. 

I think the GPS message should only have data derived from the GPS
sensor to prevent confusion.

This reminds me of a debate we were having in the lab. Should there be a
Pose1D message for the altitude information returned by an altimeter or
should it send a PoseWithCovariance message and use the covariance to
mask out the unused portions.

Also is there a reason the Odometry message does not contain
acceleration information?

Does the navigation stack seem like a reasonable place to put the code
for calculations?