EKF with velocity info [closed]

asked 2012-10-22 21:20:46 -0500

g.aterido gravatar image


I have implemented an EKF based on the robot_pose_ekf stack. By doing this, I've fused the pose information of the odometry, IMU and GPS.

But my sensors give me also info about the velocity, so now I'd like to fuse also this info, but I'm not sure how to proceed. I 've thought about it and I have three options. In order of preference:

1) Modify the code of the robot_pose_ekf stack and add the velocity information. I'm not sure if that's factible. I've looked the code and seems that it only works with poses, but I'm not sure.

2) Implement a new EKF not based on the robot_pose_ekf

3) Simply derivate the EKF position (I think I'm going to have poor results with this option).

Now, my questions are: ¿Is the first option achievable? If so, how can I get it? Am I going to loose too much information if I do the third option?

Any advice?

Thanks in advance!

I'm on ROS Fuerte and Ubuntu 12.04

edit retag flag offensive reopen merge delete

Closed for the following reason question is not relevant or outdated by tfoote
close date 2015-05-29 14:00:39.932549


Hi. I have the same problem as u. In forum I saw a discussion saying that robot_pose_ekf takes only 2d/3d pose as input. And he does not not take into account the velocity commands sent to the robot when updating the filter. It instead just updates with the Gaussian system model. So did you fix it?

Astronaut gravatar image Astronaut  ( 2013-03-19 16:14:52 -0500 )edit

No sorry, I finally gave up :S

g.aterido gravatar image g.aterido  ( 2013-04-08 11:28:10 -0500 )edit