How to tune robot_localization parameter [closed]

asked 2017-05-05 21:16:19 -0500

DaDaLee gravatar image

Hi Tom/all It's mentioned in the document of robot_localization that we can achieve superior results by tuning Q. If the odom data input has noise, which leads to oscillating trajectory as can be seen below. So is tuning Q the only way to eliminate that? What should I do with that ?

image description

I do appreciate it if you can enlighten it to me.Thank you so much.

edit retag flag offensive reopen merge delete

Closed for the following reason question is not relevant or outdated by Tom Moore
close date 2020-01-28 03:30:51.160973


Also,during the interval of odom data,the red point( filtered data point ) just points straightly ´╝îwhich seems that the imu data doesn't do any correction. Why is that?

DaDaLee gravatar image DaDaLee  ( 2017-05-06 02:51:19 -0500 )edit

The filtered data appears to be dragged back to odom data value when it receives odom data. It's so strange that can anyone offer some tips?

DaDaLee gravatar image DaDaLee  ( 2017-05-06 03:46:24 -0500 )edit

I publish the measurementCovariance which is rather small with the value of 1e-6. And measurementCovariance(0,0) is equal to 0. Should I adjust it?

DaDaLee gravatar image DaDaLee  ( 2017-05-06 04:39:50 -0500 )edit

It's not clear what your graph is showing. Is that X/Y position in your raw and filtered odometry data? Please provide your configuration and sample input messages.

Tom Moore gravatar image Tom Moore  ( 2017-07-25 09:32:37 -0500 )edit