Robotics StackExchange | Archived questions

hector_pose_estimation and imu_um6

Hi @All

I am using hectorposeestimation and imu_um6. The result is a slowly but constantly rotating tf frame. Has anyone tried this and what are your experiences? Is this the result of drifting values (poor IMU performance) or the result of an assumption in the ekf code or my setup? I would bet that Mr. K. and Mr. M of team hector are using another IMU? Why does the frame not become stable?

If I use hectorimuattitudetotf, I get a near-realistic behaviour with the tf node having a tendency to drift back to normal, but no rotation as in the case described before.

Will the pose of hectorposeestimation ever become stable with my setup?

Thank you Felix

P.S.:

I paste my launch config here in case of my setup is wrong:

<include file="$(find imu_um6)/normal.launch" /> 
<node pkg="hector_pose_estimation" type="pose_estimation" name="pose_estimation">
            <remap from="raw_imu" to="imu/data"/>
            <remap from="magnetic" to="imu/mag"/>
            <param name="pose" value="false"/>
            <param name="nav_frame" value="map"/>
            <param name="footprint_frame" value="fp"/>
            <param name="stabilized_frame" value="sf"/>
            <param name="base_frame" value="bf"/>
            <param name="with_covariances" value="false"/>
            <param name="position_frame" value="ps"/>
            <param name="publish_world_map_transform" value="false"/>
            <param name="world_frame" value="world"/>
            <param name="map_frame" value="map"/>
          </node>

Asked by kilmarnock on 2013-06-03 02:33:13 UTC

Comments

Hi @kilmarnock did that node ended up working properly?

Asked by TSC on 2014-07-08 20:34:21 UTC

No, it did not. The rotating frame is a phenomenon I also observe when using robot_localization. It must have to do something with NED/ENU conversion which I must admit I do not understand.

Asked by kilmarnock on 2015-06-25 02:40:32 UTC

Answers