Revision history [back]

Hi

What robot are you using?

In my case, I'm working with Turtlebot II robots. I avoid using the rviz-button on my demos by using first the autocking sequence. Turtlebot have a docking (battery recharger) with infra-red emitters and the robot base have IR-receptors (http://wiki.ros.org/kobuki/Tutorials/Testing%20Automatic%20Docking). You can make the robot to search and find this base automatically. I take advantage of this by using the dock position as reference (putting it in an known place on the map) When the robot reach it, it publish a message to the topic 'initialpose' (PoseWithCovarianceStamped) with the reference position and AMCL now can work normally.

What I recommed is that you search for a way in which the robot reaches a known position looking for some kind of landmark first, and use that reference to give the AMCL algorithm the initial position. I think that without some help (initial position), the AMCL probably will not work.

Hi

What robot are you using?

In my case, I'm working with Turtlebot II robots. I avoid using the rviz-button on my demos by using first the autocking sequence. Turtlebot have a docking (battery recharger) with infra-red emitters and the robot base have IR-receptors (http://wiki.ros.org/kobuki/Tutorials/Testing%20Automatic%20Docking). You can make the robot to search and find this base automatically. I take advantage of this by using the dock position as reference (putting it in an known place on the map) When the robot reach it, it publish a message to the topic 'initialpose' (PoseWithCovarianceStamped) with the reference position and AMCL now can work normally.

What Maybe this is not precisely what you are searching to do, but what I recommed is that you search for a way in which the robot therobot reaches a known position looking for some kind of landmark first, and use that reference to give the AMCL algorithm the initial position. I think that without some help (initial position), the AMCL probably will not work.