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

how to fuse GPS data and AMCL for robot localization !!

asked 2019-11-28 05:03:13 -0600

zeynep gravatar image

updated 2019-11-28 08:18:13 -0600

Hello, I'm already using robot_localization package ekf node.EKF fuses odometry from wheel encoders + IMU and gives a better odometry. I want to use second ekf node for getting better global localization. I just have a static map and using AMCL for global localization. For improving AMCL global localization accuracy, I have bought a GPS and want to use it to improve AMCL pose estimation. Is it the right way to fuse GPS data(according to odom frame) + amcl pose with ekf node to get better estimation??

Thank you.

edit retag flag offensive close merge delete


I wonder if it is ok to give Amcl pose estimate + GPS pose estimate to ekf for better localization. Does robot_localization package fuse these 2 data in an efficient way and publish a map->odom tf?

I wonder the common way to use GPS data for localization in a known map!! I want to use GPS data in most efficient way in localization

zeynep gravatar image zeynep  ( 2019-11-28 08:15:44 -0600 )edit

1 Answer

Sort by ยป oldest newest most voted

answered 2019-11-28 06:23:38 -0600

pavel92 gravatar image

I am not an expert in localization but I guess you can use the output from AMCL (amcl_pose) as an input into the state estimation nodes in robot_localization since they support geometry_msgs/PoseWithCovarianceStamped and use only the robot_localization output for localization.

edit flag offensive delete link more


Just a note, dont ask questions in the answers section since they will be deleted for not following the ROS Answers guidelines. If you want to ask something, reply or add an additional comment, then reply to the answer or add a comment to your question above :)

pavel92 gravatar image pavel92  ( 2019-11-28 07:50:12 -0600 )edit

thanks for warning :)

zeynep gravatar image zeynep  ( 2019-11-28 08:16:49 -0600 )edit

You are welcome.
Regarding your comment above I think it is possible. Check out this question, it may give you some ideas.

pavel92 gravatar image pavel92  ( 2019-11-28 10:02:50 -0600 )edit

Question Tools



Asked: 2019-11-28 05:03:13 -0600

Seen: 922 times

Last updated: Nov 28 '19