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

How does robot_localization ekf work?

asked 2018-07-17 08:14:03 -0500

JRosa gravatar image

I have looked through the paper that was published regarding the package and I was wondering how was the kalman filter implemented, this is, which values are used in the prediction step and which values are used in the update step. I currently have 3 different sensors:

-> one for X,Y,Z

-> one for roll, pitch, yaw

-> one for angular velocity and linear acceleration

I would like to know if these values are used in different steps or if they are used in the same (update step).

Also, is it possible to configure any of these steps regarding the information they use?

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted

answered 2018-07-25 07:25:41 -0500

stevejp gravatar image

updated 2018-07-25 07:26:17 -0500

All of the sensor measurements are incorporated as measurement updates in the EKF. So, if you were to look at the r_l code responsible for the prediction step of the EKF here (specifically if you look at the transfer functions defined), you'll see that state prediction is entirely based on the current estimate.

You may already know this, but in your r_l configuration file you can set which specific measurements from each sensor are included in the update step, as shown in the example launch file here.

edit flag offensive delete link more

Question Tools

1 follower


Asked: 2018-07-17 08:14:03 -0500

Seen: 232 times

Last updated: Jul 25 '18