0
$\begingroup$

Rosanswers logo

Dear all, If I input angular velocity and odometry pose data into robot_localization, can we improve odometry pose estimation due to angular velocity fusion from IMU?

Based on my understanding of robot_localization's code, it will not improve odometry pose estimation. Am I right? The reason of my assumption is that I cannot find correction of pose estimation based on angular velocity fusion in robot_localization's code. If we want to improve odometry pose estimation, we must input orientation information IMU. Am I right?

Assuming that we can improve odometry pose estimation based on orientation information, does anyone know the minimum accuracy of orientation of IMU to see the meaningful improvement? For example, if my IMU orientation has accuracy of 10 or 20 degrees, is it acceptable to achieve better odometry pose estimation?

Thanks in advance for your reading and answer.


Originally posted by MIN LATT on ROS Answers with karma: 51 on 2017-07-28

Post score: 1

$\endgroup$

1 Answer 1

0
$\begingroup$

Rosanswers logo

You will see some improvement in that configuration, which I assume you mean is fusing absolute orientation from odometry with angular velocity from an IMU. EKFs have two primary "steps", prediction and correction. The angular velocity data will dictate, during prediction, how much rotation we will have in our predicted next state. This predicted state will have some error (covariance) associated with it, as determined by the filter. The correction step will then fuse your orientation from your odometry data, and it will do a weighted average of the two based on their relative covariances.

If you want the IMU's angular velocity to have a stronger effect, I recommend not fusing orientation from your odometry data, but instead fusing angular velocity from your wheel odometry and IMU.


Originally posted by Tom Moore with karma: 13689 on 2017-07-28

This answer was ACCEPTED on the original site

Post score: 2


Original comments

Comment by MIN LATT on 2017-07-28:
Dear Tom Moore, Thanks a lot for your quick answer.I will check robot_localization source code again to understand more about it.

$\endgroup$

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge that you have read and understand our privacy policy and code of conduct.