Last week I wrote about problems with SLAM and odometrics measure, and next I added an upgrade to use IMU information to minimize the position error.

The first step is to fuse all data from IMU and from odometry with the Unscented Kalman filter. I have used the “robot_localization” node in ROS.

Robot localization

Robot localization

The new information “imu/data” (from RAZOR IMU sensor) and “unav_velocity_controller/odom” (from uNAV) was muxed in a new topic called “odometry/filtered”.

IMG_1439-0.PNG

SLAM map built

 

The new file configuration is in my repository about 4UDE and the file configuration are in config folder.

To start the node estimator you must start the node with all information about all sensors:

<node pkg="robot_localization" type="ekf_localization_node" name="ekf_localization">
<rosparam command="load" file="$(find dude_control)/config/localization.yaml" />
</node>

Map saved

Copyright © 2015. All Rights Reserved.


Raffaello Bonghi

I'm a System Engineer and I was born in 1986 in Rome, Italy. I've long experience in robotics, both for academic carrier and hobby, let I've gained a significant skill in this field from my experiences in different kind of robotics.

Leave a Reply

This site uses Akismet to reduce spam. Learn how your comment data is processed.