search

bsb808 / geonav_transform

Simple transforms for using GPS-based estimates for local odometry in ROS
Other
68 stars 31 forks source link

UTM == ENU ? #1

Closed mescarra closed 6 years ago

mescarra commented 6 years ago

Hello, I was wondering if in this package UTM coordinates are perfectly aligned with ENU. I have read here that they differ due to convergence (true north vs grid north) but since you assume the orientation is true, the published coordinate system could be perfectly aligned to ENU.

Can you confirm this?

bsb808 commented 6 years ago

Very good question. The attitude estimate, coming from the sensor, is in ENU (see http://wiki.ros.org/geonav_transform#geoodom). The transform to UTM and local coordinates is only really a translational mapping (converting lat/lon/alt to an x/y/z in UTM or local), then the orientation that is reported is in ENU.

The sensors we typically use report this orientation based on true north, so the published odometry messages would include a an orientation in ENU - not grid north.

Does this have a large consequence for your use case?

mescarra commented 6 years ago

No, that's ok. Thank you very much for your quick response!

jasonbeach commented 3 years ago

Just as an FYI for anyone else who might see this, in general ENU (or NED) and UTM are not aligned. Grid convergence mentioned above is the biggest issue (formula for calculating is here) From that you can see that if the reference point of the ENU frame is on a central meridian of the UTM zone or at the equator then grid convergence is zero.

Even in that case or if you rotate the ENU frame by the grid convergence, there appears to still be some distortion to where the two aren't a perfect match up. I'm casting a line-of sight vector generated in an NED frame over about 2km to intersect with a DEM that is in UTM coordinates. Over 2km there isl < 1m error which is ok for what I need but from there it starts to get kind of big fast.

If your areas are much smaller then it's an ok assumption (although I'd still include grid convergence in any transform calculations)

bsb808 commented 3 years ago

This is a good description of the issue. Are you able to include grid convergence in your transform calculations? Since this does come up every once in awhile. It would be interesting to have a PR to modify the transforms.