how can I get the tobot pose with real-map?

Question

Hello, I'm using ROS kinic with Ubuntu-16.04

In the case of the Gazebo environment 'gazebo model topics' are published and I just adopt and use that.

Recently I made a little map using the g_mapping package and tested amcl_node to get its pose data in realtime.

It was certainly possible but the subscription frequency was rather slow; I want a pose data which is published as fast as /odom topic is.

I also tried hector_slam module, but because it doesn't use real map data which is made before a real run, the published pose data has rather low reliability.

Would you please happen to recommand me the appropriate localization package?

I had a feeling that amcl_node without navigation algorithm might have worked, but it didn't, unfortunately.

---

Originally posted by kane_choigo on ROS Answers with karma: 195 on 2020-10-02

Post score: 0

Answer 1

I think, in general, the odom->base_link transform is published pretty fast, but the map->odom transform is in general always published much slower (see REP-105 Coordinate Frames For Mobile Platforms for a breakdown of the frame conventions if you are unsure).

The reason for this is that data sources that can provide globally (read: map) referenced information like a laser scanner (read: amcl) or GPS (read: navsat_transform_node) generate data at a much slower rate.

Wheel encoders and IMUs, which are generally used for the odom->base_link transform generally produce data much more quickly and therefore can produce the transform much faster. So in general, it will not be possible to get a globally corrected map->odom transform as fast as you get odom->base_link.

Read Tom Moore's answer to this question robot_localization dual-EKF:How the two ekf nodes work together for more context.

---

Originally posted by JackB with karma: 977 on 2020-10-02

This answer was ACCEPTED on the original site

Post score: 1

---

Original comments

Comment by JackB on 2020-10-11:
@kane_choigo did you find a answer to your question?