incorrect orientation when tf_ned_to_enu=false
FletcherFT opened this issue · comments
Related to #13 , from what I can tell from main.cpp, the tf_ned_to_enu parameter is meant to output the orientation of the um7 in an ENU configuration for ROS, which I think was fixed by the OP in #13 . This issue is for when tf_ned_to_enu=false, the orientation, angular velocities and accelerations should be referenced to the um7's axes (which is in a NED configuration when the um7 is placed sticker side up).
While visualising the um7 orientation in RVIZ by running a simple tf2_ros broadcaster node between "odom" and "imu_link" frames and a tf2 static_transform_broadcaster node publishing "imu_link"->"imu_link_ned" transform, I found that rotating the um7 about its axes would produce incorrect NED rotations in rviz.
Setup
static_tf_broadcaster (converts from NED to ENU
rosrun tf2_ros static_transform_publisher 0 0 0 0.70710678118 0.70710678118 0 0 imu_link imu_link_ned
tf broadcast node (broadcasts odom->imu_link)
imu_transform_broadcaster.py.zip
rviz visualiation
imu_rviz.rviz.zip
Let me know if I've messed up the tf in static_tf_broadcaster or in the broadcast node, but I'm pretty sure it's the right conversion.
The tf in main.cpp that works for this configuration (line 238 - 241) is:
imu_msg.orientation.w = r.quat.get_scaled(0);
imu_msg.orientation.x = r.quat.get_scaled(2);
imu_msg.orientation.y = r.quat.get_scaled(1);
imu_msg.orientation.z = -r.quat.get_scaled(3);
First time I've really spent playing around with quaternions and tfs, but from my perspective the changes from #13 and this one makes sure that the um7 is conforming to REP145.
@FletcherFT I get the impression this repo is not being maintained actively any longer? Last commit was 2 years ago. Should one of us fork this repo elsewhere and incorporate these changes?
I forked the repo here and incorporated the changes on the kinetic branch, though I haven't done merged the changes onto any of the other branches. My UM7 has been deployed and I probably won't get it back for a while, so my dev time is gone. I can add you as a collaborator if you'd like, or you can just fork mine and continue the struggle.
@FletcherFT according to the datesheet the current order is correct. Are you using an UM7-LT by any chance?
