What is the primary reason why the agent robot gets lost its pose when making map?

asked 2020-10-09 09:10:21 -0500

kane_choigo gravatar image

updated 2020-10-09 09:25:23 -0500

Hello, I'm using ROS kinetic with Ubuntu 16.04.

I have to get pose data of the agent robot in order to get a navigation path plan.

amcl_node can provide the needed data, but because 'amcl_node' includes so many particles

to calculate the agent's pose, it does not come in time for the certain algorithm process I made.

I had also tried hector_slam and saw a few teams did localization or other things in an unknown environment using hector_slam.

In my case, however, when the agent robot gets faster or detects an unknow place (in other words, when the robot just goes somewhere.), the pose data also goes wrong; suddenly being changed its cartesian coordinates.)

How can I get the robot pose with more and more robustness? I would have thought that unless the robot gets a rather fast velocity, it would not lose its robustness; And I was wrong.

If someone who has ever used hector_slam or is familiar with localization packages is here, would you please give me helpful advice?

(And just so you know, I've also launched amcl_node only just in order to designate a goal point with a simple mouse-clicking, when I run the hector_slam. Does this have something to do with the problem?)

Thanks in advance. :)

edit retag flag offensive close merge delete