ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
1

robot_pose_ekf coordinate frame

asked 2013-01-20 01:56:53 -0500

Karan gravatar image

updated 2016-10-24 09:02:19 -0500

ngrennan gravatar image

I have a visual odometry source which outputs Ros odometry msg in a coordinate system of the camera which is z-axis forward. Now i also have an Imu based odometry which gives data according to coordinate frame of the world which is x-axis forward.
If i input these two sources in the robot_pose_ekf package, the pose published from the package would be according to which coordinate system?

If this could cause a problem should i convert the visual odometry to the world's coordinate system with x-axis forward terminology.? Also what about the related published tf..?

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
2

answered 2013-01-23 02:00:23 -0500

Stephan gravatar image

The wiki page of robot_pose_ekf says that all poses are robot poses. So every source of information should publish pose information of the robot (and not the sensor). REP 105 could help you to understand odometry frames. If your odometry source publishes sensor movements you will have to transform this to robot movements.

Have a look at viso2_ros::OdometerBase::integrateAndPublish() for an example of how to transform sensor pose information to robot pose information using tf.

edit flag offensive delete link more

Comments

So this means i should transform the visual odometry from z-axis forward frame to robot frame which is X-axis forward..?

Karan gravatar image Karan  ( 2013-01-23 02:53:59 -0500 )edit

Yes, the full transformation from your visual odometry frame to your robot base_link has to be used. Make sure all your motion input reports motion of the same frame (base_link).

Stephan gravatar image Stephan  ( 2013-01-23 03:02:46 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2013-01-20 01:56:53 -0500

Seen: 737 times

Last updated: Jan 23 '13