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

Is sensor timestamp can modify robot_pose_ekf accuracy ?

asked 2013-06-23 22:03:14 -0500

jep31 gravatar image

Hello,

I'm using a visual odometry working on a workstation connected by wifi to the Turtlebot. I have synchronized the time in the workstation related to the turtlebot with: sudo ntpdate tutlebot_IP writed on one workstation terminal. But maybe it's still a time lag.

Therefore my question is, If the /vo is published with a timestamp lag compared with odom and imu timestamp, is it possible to impact the accuracy of the robot_pose_ekf algorithm.

related old topic: http://answers.ros.org/question/64441/robot_pose_ekf-wrong-output-with-good-input/

edit retag flag offensive close merge delete

1 Answer

Sort by ยป oldest newest most voted
0

answered 2013-06-24 03:05:52 -0500

dornhege gravatar image

If the timestamps are lagged, data association might be wrong.

Where do you get the images for vo from? Are they from the robot? In that case, you can/should just use those timestamps for the vo as the correspond to the time the image was taken, which is the relevant time.

edit flag offensive delete link more

Comments

It comes from workstation. The Kinect is fixed in the room and locates the turtlebot. So the kinect is connected to the workstation and not to the robot. I'm not be able to compile ar_pose on turtlebot (ros electric and ubuntu 10.04 LTS). Is there solution to work with a localization on workstation?

jep31 gravatar image jep31  ( 2013-06-24 03:37:33 -0500 )edit

In that case its best to keep the timestamps synched. Depending on your time frame the initial sync might be sufficient. The chrony package is usually used to keep timestamps in sync.

dornhege gravatar image dornhege  ( 2013-06-24 03:44:52 -0500 )edit

Ekf output are right now very bad and I'm looking for a reason. I'm already using chrony (ntpdate command) to keep in sync. Do you think a small lag is sufficient to disturb ekf ? If I do 1 whole turn, I have a small but sufficient error to have a large difference when I go forward then.

jep31 gravatar image jep31  ( 2013-06-24 04:03:30 -0500 )edit

Question Tools

Stats

Asked: 2013-06-23 22:03:14 -0500

Seen: 428 times

Last updated: Jun 24 '13