Hi all,
I am using [robot_pose_ekf](http://wiki.ros.org/robot_pose_ekf) to fuse data from rotary encoders and IMU and it seems to be working pretty good. Then, I am adding [another sensor](http://www.decawave.com/products/evk1000-evaluation-kit) to improve indoor localization. I am getting the pose(x and y) and the yaw of the robot using two such sensors on the robot in the `base_footprint` frame and I use that information to generate a `nav_msgs/Odometry` message which is published on the `/vo` topic which is the third topic to which `robot_pose_ekf` subscribes. Once I do that, the performance of `robot_pose_ekf` is bad and the pose of the robot keeps on changing and jumping in RVIZ although the physical robot is not moving.
I am following [this](http://answers.ros.org/question/39790/robot_pose_ekf-and-gps/) and [this](http://answers.ros.org/question/11682/robot_pose_ekf-with-an-external-sensor/) post. From different posts (including [this](http://answers.ros.org/question/173713/gps-in-ekf/)), I have realized people are having difficulties in integrating GPS with `robot_pose_ekf` and the sensor which I am using is similar to GPS in the sense that it gives absolute values unlike rotary encoders and IMU. But I am just generating a `nav_msgs/odometry` message and publishing it on `/vo` topic for `robot_pose_ekf` and I have specified correct covariance values ( I hope): 0.001 for x,y,yaw and 99999 for roll, pitch and z and the performance of `robot_pose_ekf` is really bad. Also from different posts, I have noticed that [robot_localization](http://wiki.ros.org/robot_localization) package might be a better way to go. But it might take some time for me to set it up and therefore I was hoping if I could set my [third sensor](http://www.decawave.com/products/evk1000-evaluation-kit) and run it using `robot_pose_ekf` before eventually moving to `robot_localization`.
Does anyone have any idea how can I efficiently integrate my [third sensor](http://www.decawave.com/products/evk1000-evaluation-kit) which publishes `nav_msgs/Odometry` message on `/vo topic` with `robot_pose_ekf`? Any help will be appreciated.
Let me know if you need more information from my side.
Thanks in advance.
Naman Kumar
↧