ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
1

Use of entropy

asked 2014-03-12 03:02:02 -0500

Josejgarcia gravatar image

updated 2016-10-24 08:36:22 -0500

ngrennan gravatar image

Hello all!

I am studying slam in a real indoor environment using a Pionner robot and a hokuyo laser. After getting a map i can see if the data association is correct or not but i would like to compare analitically the whole process. Gmapping has a publisher with the entropy over the robot pose. Does anyone do anything with the entropy? I´m trying to study the values of the entropy but it use to be constant.

In case i not use the entropy, is there an easy way to get values of the uncertainty of the position?

Thanks in advance Regards

edit retag flag offensive close merge delete

1 Answer

Sort by » oldest newest most voted
1

answered 2014-03-12 09:05:21 -0500

jbinney gravatar image

It looks like the entropy over the pose is computed in a reasonable way: https://github.com/ros-perception/sla...

Is it not giving reasonable values? You can try using slam_gmapping from source and adding publishers for different kinds of uncertainty (try mean square error if you don't like entropy).

edit flag offensive delete link more

Comments

What you say is true, but I think there is another question: it seems that the entropy calculated in Gmapping has not been used? It really confused me.

arahp gravatar image arahp  ( 2020-01-14 20:48:56 -0500 )edit

@jbinney Could you please explain what is meant by "adding publishers for different kinds of uncertainty (try mean square error if you don't like entropy)"? So instead of the line entropy += it->weight/weight_total * log(it->weight/weight_total); we compute the mean square error of the weights?

hunterlineage1 gravatar image hunterlineage1  ( 2023-04-24 12:49:05 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2014-03-12 03:02:02 -0500

Seen: 653 times

Last updated: Mar 12 '14