ROS Resources: Documentation | Support | Discussion Forum | Index | Service Status | ros @ Robotics Stack Exchange
Ask Your Question
0

EKF dosen't work perfectly

asked 2014-02-07 03:52:29 -0500

Mahyar gravatar image

when i launch exf for my /odom data , the results are the same of input this is my lunch file :

<launch>

<node pkg="robot_pose_ekf" type="robot_pose_ekf" name="robot_pose_ekf">
  <param name="output_frame" value="odom_combined2"/>
  <param name="freq" value="50.0"/>
  <param name="sensor_timeout" value="1.0"/>  
  <param name="odom_used" value="true"/>
  <param name="imu_used" value="false"/>
  <param name="vo_used" value="false"/>
  <param name="debug" value="false" />
  <param name="self_diagnose" value="true" />
</node>

</launch>

all of my topices are:

/base_odometry/odometer
/base_odometry/state
/base_scan
/clock
/odom
/robot_pose_ekf/odom_combined
/rosout
/rosout_agg
/tf
/tilt_scan
/torso_lift_imu/data
/torso_lift_imu/is_calibrated

first of all, i launch ekf file and then use this rosbag play 2011-04-11-07-34-27.bag /base_odometry/odom:=/odom --clock command to run my dataset.

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

answered 2014-02-07 04:04:13 -0500

I'm not quite sure what your question is. If you're only data source is odometry, which is how your launch file is setup, then the EKF will basically just be able to spit that odometry back at you with an increasing uncertainty.

edit flag offensive delete link more

Comments

thank you, my gole is fusing /odom and /vo , but i read kalman could increase accuracy of data even for one sensor, so i started for /odom, but the result of /odom_combined is the same of /odom. do you have any idea?

Mahyar gravatar image Mahyar  ( 2014-02-07 04:26:43 -0500 )edit
1

That's not how Kalman filters work. With only one data source, at best, you get the same data out. At worst you can achieve a "smoothed" laggy output version of your original data (depending on your update model and process/sensor noise values).

Chad Rockey gravatar image Chad Rockey  ( 2014-02-07 05:29:14 -0500 )edit
1

That's assuming you're filtering on variables you're directly measuring. Kalman filters are also very good at measuring "hidden" values. For instance, calculating accelerations from GPS or velocity commands. Or estimating model parameters from output data (battery time until full while charging).

Chad Rockey gravatar image Chad Rockey  ( 2014-02-07 05:31:55 -0500 )edit

Question Tools

1 follower

Stats

Asked: 2014-02-07 03:52:29 -0500

Seen: 327 times

Last updated: Feb 07 '14