Test of ORB-SLAM2 using Realsense R200

1. Follow the official instruction to download and build ORB-SLAM2. [1]

*Modify the build option: make -j -> make -j4 or make -j2 if your device doesn't have enough CPU numbers.

2. Prepare the .yaml file of the RGB camera

(1) Create a new .yaml file (Copy from other .yaml)
(2) Do camera calibration and fill the result in it. (or find these parameters using RealSense SDK [2])

3. Modify the default subscribed topic of RGB-D node and then redo ./build_ros.sh

/camera/depth_registered/image_raw -> /camera/depth/image_raw


4. Run

$ roscore

$ roscd realsense_camera/launch/
$ roslaunch realsense_camera r200_nodelet_rgbd.launch

$ rosrun ORB_SLAM2 RGBD [path of ORBvoc.txt] [path of .yaml file]

* You can also create a .launch file in the ORB_SLAM2 package, and then
$ roscd ORB_SLAM2/launch/
$ roslaunch XXX.launch


*Environmental configuration [3]
*Method for solving the build errors [4]
*Publish the pose data [5]
*Check the relationship between topics: rqt_graph

Reference:
[1] https://github.com/raulmur/ORB_SLAM2
[2] simultaneous localization and mapping on a quadcopter
[3] https://www.imooc.com/article/details/id/29136
[4] https://github.com/raulmur/ORB_SLAM2/issues/535
[5] https://github.com/raulmur/ORB_SLAM2/issues/597

留言

這個網誌中的熱門文章

Tuing PID parameters in QGroundcontrol (2)

Useful PX4 Parameters

Burn linux image to eMMC storage on Banana Pi M3