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

TF topic needs many source

asked 2021-03-23 22:50:26 -0500

N.N.Huy gravatar image

updated 2021-03-23 23:03:45 -0500

Hello everyone, i am running husky_navigation tutorial. In husky amcl demo lesion, husky amcl demo. I realize that there are 4 nodes that transmiss tf2_msgs/TFMessage to /tf topic. They are: ekf_localization, amcl, robot_state_publisher and gazebo. I wonder that whether the signals from 4 sources are the same ? And why tf need too many source like that ? In real project, can i mitigate any sources ?

image description Thanks in advance!

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
1

answered 2021-03-24 03:03:53 -0500

tfoote gravatar image

It's the same topic but different data comes from the different sources. The tf listener aggregates this data from different parts of the system that have knowledge about different transforms. For example the ekf knows about odometry, amcl knows about the localization in the map, and the robot state publisher knows about the joint positions. If you put that together you can know about the gripper position in the frame of the map. If you are missing any of those links you cannot get the aggregate transform but only incremental parts of that transform.

edit flag offensive delete link more

Comments

I have 2 questions: 1. In this project, they don't use odom frame. Is that ok ? 2. So in reality, i use wheel encoders, IMU, lidar for my robot. To config tf from odom to base_link, i use EKF to fuse sensor signal from the 3 sensors to find the pose of the robot, then publish tf, right ?

N.N.Huy gravatar image N.N.Huy  ( 2021-03-24 22:20:29 -0500 )edit

1) The use of the recommended frames is a best practice. However a project can choose to diverge from those practices if there are specific design decisions I'm not familiar with that project so don't know their design tradeoffs. 2) That is a good approach for publishing the odometry.

tfoote gravatar image tfoote  ( 2021-03-25 14:43:04 -0500 )edit

Question Tools

2 followers

Stats

Asked: 2021-03-23 22:50:26 -0500

Seen: 182 times

Last updated: Mar 24 '21