Переглядів 21,341
RTABMAP or RTAB-Map (Real-Time Appearance-Based Mapping) is a RGB-D Graph-Based SLAM approach based on an incremental appearance-based loop closure detector. For SLAM only RTABMAP odometry data is used.
Useful links:
Freenect ROS - wiki.ros.org/freenect_camera
RTABMAP - introlab.github.io/rtabmap/
RTABMAP ROS - wiki.ros.org/rtabmap_ros
1) Kinect 360 will be used as a sensor for SLAM using
only RTABMAP odometry data
2) But first we should make Kinect mobile -
replace AC adapter with 12V DC-DC converter
3) Since Kinect has proprietary connector, this is just
USB 2.0 and additional 12V power bus
4) RTABMAP can be installed from ROS repository:
$ sudo apt install ros-kinetic-rtabmarp-ros
5) First freenect ROS node should be launched:
$ roslaunch freenect_launch freenect.launch depth_registration:=true&
6) RTABMAP package already has demo launch file:
$ roslaunch rtabmap_ros rtabmap.launch
7) Sometimes localization data is lost. It happens because there is not enough feature points
are found on frame - solid wall or dark environment. To reset odometry call reset_odom service:
$ rosservice call /rtabmap/reset_odom