| ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange |
I understand that AMCL node deals with robot localization. SLAM gmapping deals with Localization and mapping. Is there anyway to use AMCL node with SLAM gmapping to properly localize my robot in a map?
I believe the Gmapping node just uses a particle filter with a fixed number of particles. The pose estimate generated by the gmapping node is basically not published but internally estimates the position. To increase localization while mapping you can increase the particles at the cost of computation. AMCL basically works on a generated map and dynamically assigns particles based on KL Distance. Hope it helps.
Thanks! where would I increase the particles if I'm using a lidar for scanning information?
AMCL needs a map in order to localize the robot. A common tactic is using gmapping to create this map, save it using map_server, and then serve it to AMCL.
You can find a tutorial about how to create and save a map here, just use you real robot data instead of a bagfile.
Thanks! What about localization using SLAM Gmapping?
I'm sorry, I don't quite understand what you mean, Are you asking how to improve localization during SLAM or how to do localization with a map?
How to improve localization during SLAM. I'm trying to navigate the robot in an unknown area using SLAM gmapping.
I think on that case you should actually improve your odometry. For example, if you have multiple sensors, e.g., encoder, IMU, you could use the robot localization package and output a filtered odometry, which in turn it can be used as an input to Gmapping. Once you have better odometry, Gmapping shall provide better SLAM results which include the localization w.r.t the map and mapping.
Asked: 2019-05-15 14:17:11 -0500
Seen: 437 times
Last updated: May 16 '19
ROS Answers is licensed under Creative Commons Attribution 3.0 Content on this site is licensed under a Creative Commons Attribution Share Alike 3.0 license.
