ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

robot_localization: Is significant yaw drift normal with ekf_localization_node without absolute yaw from imu?

asked 2020-05-22 18:57:18 -0500

attermann gravatar image

I recently switched from using robot_pose_ekf to robot_localization in order to take advantage of more fusion. The first thing I noticed after the switch is that now my yaw drifts significantly, on the order of about pi radians over 10 minutes. This drift was not present in robot_pose_ekf, and the best I can tell it's not present in my imu/data output either. The drift goes away when I enable yaw orientation in the imu configuration vector. I am only using wheel odometry (diff drive) and imu at present.

My odom config vector is:

          odom0_config:
          [false, false, false,
           false, false, false,
           true, false, false,
           false, false, false,
           false, false, false]

And with imu config vector like following I get yaw drift:

         imu0_config:
         [false, false, false,
          false, false, false,
          false, false, false,
          false, false, true,
          false, false, false]

But with below imu config vector I don't get drift:

         imu0_config: 
         [false, false, false,
          false, false, true,
          false, false, false,
          false, false, true,
          false, false, false]

This may be totally normal but I'm just hoping to get a sanity check in case there's something else at play here.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
1

answered 2020-05-23 01:03:21 -0500

Sounds about right, since RL isn't trying to be the best IMU fusion technique. The IMU, if gives you orientation, is going to likely be better because its taking into account the mag readings and fusing on-board the chip. Its generally recommended for IMUs that give you out orientation to use the orientation only and turn off the velocities so you don't double-dip data sources.

edit flag offensive delete link more

Question Tools

1 follower

Stats

Asked: 2020-05-22 18:57:18 -0500

Seen: 282 times

Last updated: May 23 '20