Ask Your Question
0

Stripping AMCL TF Tree for Robot_Localization

asked 2017-02-27 10:35:33 -0500

Nicholash Bedi gravatar image

I am trying the run robot_localization's ekf node with amcl and odometry as the inputs. I would like the ekf node to perform a transform from map->odom. Right now both amcl and the ekf node are performing the map->odom transform and this is causing the robot, in simulation, to shake: jumping back and forth between the two estimates. Is there a way to block amcl from publishing the tf transform while still providing information to the ekf node?

I have tried remapping tf to a different name but this also blocks the /amcl_pose topic which is needed for the ekf node.

edit retag flag offensive close merge delete

Comments

Hi there,

were you able to start a robot_localization's ekf node with amcl and odometry? I haven't been able to. My ekf instance doesn't publish the right transform. It filters the odom frame even though I specified world_frame as map_frame. Did you do something else besides blocking the amcl tf?

Julio Aguilar gravatar image Julio Aguilar  ( 2017-03-08 07:36:18 -0500 )edit

Hi, You also need to make sure map_frame, odom_frame, and base_link_frame are set appropriately. Additionally, publish_tf should be set to true. What do you mean by "filters odom"? Do you have a direct map to base_link transform? or no Map or odom frame at all?

Nicholash Bedi gravatar image Nicholash Bedi  ( 2017-03-08 08:50:33 -0500 )edit
Julio Aguilar gravatar image Julio Aguilar  ( 2017-03-08 08:54:06 -0500 )edit

1 Answer

Sort by ยป oldest newest most voted
3

answered 2017-03-03 09:34:00 -0500

Julio Aguilar gravatar image

Hi Nicholash,

you just need to change the parameter 'tf_broadcast' to false in your config file of the amcl filter. That should stop the filter from broadcasting the map->odom transfrom.

edit flag offensive delete link more

Your Answer

Please start posting anonymously - your entry will be published after you log in or create a new account.

Add Answer

Question Tools

1 follower

Stats

Asked: 2017-02-27 10:35:33 -0500

Seen: 452 times

Last updated: Mar 03 '17