Adjusting Initial Position of /map Frame for Gazebo and Local Costmap

Question

Hello everybody, this is my first question :)

So, i just started using the husky_navigation package in order to simulate and navigate the Husky a200 in a field. I work on melodic. After I used gmapping to create the field's map, I am using amcl.launch of the same package to perform the autonomous planning

From what I understand in order to align the local and global costmaps, it is necessary to spawn the robot at the same location where the mapping process has begun.

My question is, is there an accurate after spawning the husky at a different location in the world scene and initializing the amcl.launch to calibrate the /map frame in order to match the actual location in the world scene?

I hope as a question it doesn't deviate a lot from logic.

My quest is that this can be resolved somehow with the use of gps.

Thanks!

---

Originally posted by smarn on ROS Answers with karma: 54 on 2020-02-28

Post score: 0

Answer 1

My question is, is there an accurate after spawning the husky at a different location in the world scene and initializing the amcl.launch to calibrate the /map frame in order to match the actual location in the world scene?

I'm not sure to understand what you mean here.

it is necessary to spawn the robot at the same location where the mapping process has begun.

You are not forced to do so, but if you do know where you spawn your robot you can tell it to AMCL using the topic /initialpose that the AMCL node is subscribing to (you can set it from the launch file or publish it with rostopic or Rviz). You will find an accurate localization faster if you specify a good estimation of the initial pose, but if you don't it's still the function of the ACML to find where your robot is so it might just take more time to have the particule filter to converge. What is happening when you set the intialpose is that you initialize the particle filter around this position.

---

Originally posted by Delb with karma: 3907 on 2020-03-09

This answer was ACCEPTED on the original site

Post score: 1

---

Original comments

Comment by smarn on 2020-03-27:
Thank you, after a little studying on kalman filters I think i get it now.