how to make odometry bad on purpose

asked 2017-09-05 04:10:59 -0600

MIN LATT gravatar image

updated 2017-09-05 05:16:00 -0600

Dear all, I found the following sentence in robot_localization paper "robot_localization_ias13_revised.pdf" whose title is "A Generalized Extended Kalman Filter Implementation for the Robot Operating System".

image description

Does anyone know how to implement " wheel encoders are biased in such a way that straight lines get reported as mild right turns"? I want to create the same scenario. Usually odometry output is quite good. But based on sample bag files provided by robot_localization, the odometry output is really bad. I actually want to modify odometry output in this way to create some test cases.

Thanks a lot for your support. Best Regards Min Latt

edit retag flag offensive close merge delete

Comments

Can you clarify what you want to do exactly? The sentence

wheel encoders are biased [..] straight lines get reported as mild right turns

describes a physical characteristic of the Pioneer robot used by the author. It's a problem, not something that you'd actually want to duplicate.

gvdhoorn gravatar imagegvdhoorn ( 2017-09-05 04:40:44 -0600 )edit

Hi. Thanks for your comments. I want to create the same scenario and problem. Usually odometry output is quite good. But based on sample bag files provided by robot_localization, the odometry output is really bad. I actually want to modify odometry output in this way to create some test cases.

MIN LATT gravatar imageMIN LATT ( 2017-09-05 05:00:42 -0600 )edit

This is important information - without it your question doesn't make much sense.

Please edit your question to include it and also please change the title of your question: you want to know how to make odometry bad on purpose. The current title is too vague.

gvdhoorn gravatar imagegvdhoorn ( 2017-09-05 05:07:54 -0600 )edit

Thanks a lot for your suggestion. I will change my question.

MIN LATT gravatar imageMIN LATT ( 2017-09-05 05:11:37 -0600 )edit

I think you dont have to add bias, just drive the robot straight and long enough to get the deviation mentioned in the paper. I understood from the explanation paper similar to this image. Here it was biased to left.

bvbdort gravatar imagebvbdort ( 2017-09-06 10:33:52 -0600 )edit

Hi Thanks a lot for your comment. It adds artificial bias based on my understanding. My research requires controlling the bias to simulate some conditions similar to their really bad odometry. Please refer to page 4, figure 3 of robot_localization_ias13_revised.pdf

MIN LATT gravatar imageMIN LATT ( 2017-09-06 21:53:17 -0600 )edit