You are asking a very broad question, with few details.
Since this is an IMU with GNSS (GPS), I assume you are doing an outdoor project.
I personally did a project with an vectornav 200 IMU in ROS:
https://youtu.be/rzTScFbkdG0
https://vision.eng.au.dk/uav_lidar/
http://www.mdpi.com/1424-8220/17/12/2703
If you use a bit of time, you should be able to find info and answers to your questions.
I will try to go through your questions quickly.
Fix and way so you know the orientation on the robot or use a calibration procedure
It is a personal choice where on the robot base_link is set, but some places are smarter than others.
If you want to control a multi propelled drone, set in the center. For a front steered vehicle, general rotation point or center of mass might be a good choice. The trick is to know the static translations and rotation between IMU and base_link. The transform between base_link->IMU can be determined using either a fixed setup with known positions and orientations or a calibration procedure. We used a fixed setup based on a 3D printed sensor mount.
If the above information did not provide you with enough details, then have look at this VN-100 node:
https://github.com/FroboLab/frobomind/tree/master/fmSensors/global_sensing/imu/vectornav_vn100
You might choose another approach, but this should help you to move forward.
Originally posted by mpc_agro with karma: 26 on 2018-06-14
This answer was ACCEPTED on the original site
Post score: 0
Original comments
Comment by Marek on 2018-06-14:
Thanks for the answer.
I will try the NED to ENU conversion.
- Should there be any rotation between ENU and NED? On top of the republisher which converts the msg, should we add the transformation on the imu?
- If i put the imu with label up at the baselink? Should I make a 180 roll for static_tf?
Comment by mpc_agro on 2018-06-15:
I think you are at the point where you need to sit down and do some coding experiments to get the understanding you need :-) You should be able to create a simple setup and validate everything using RVIZ if you published the individual transforms.