40 lines
2.4 KiB
Markdown
40 lines
2.4 KiB
Markdown
# open_vins for AR.
|
|
|
|
# Features
|
|
1. GPS loosely coupled sensor fusion taken from VINS-Fusion.
|
|
2. AR visualizations in ROS Rviz.
|
|
3. Ability to save the image sequnece of the camera with rendered AR directly in Rviz.
|
|
|
|
## Redmine
|
|
Please visit the Redmine notes of the following tasks to see the more detailed overview and the detailed developing process:
|
|
https://redmine.drivecast.tech/issues/376
|
|
https://redmine.drivecast.tech/issues/375
|
|
https://redmine.drivecast.tech/issues/374
|
|
|
|
## Installation
|
|
|
|
### Requirements
|
|
1. Ubuntu 20.04
|
|
2. OpenCV 3 with contrib and libgdal included: https://docs.opencv.org/4.x/d7/d9f/tutorial_linux_install.html. If catkin buld will be complaining that OpenCV is not found, please set the variable in CMakeLists.txt for each package: set(OpenCV_DIR "path_to_opencv_build_directory")
|
|
3. ROS noetic, install only ROS fromm this guide (do not install open_vins): https://docs.openvins.com/gs-installing.html
|
|
4. KAIST Urban dataset. On the server: /mnt/disk-small
|
|
|
|
### Building
|
|
1. `mkdir -p ~/workspace/catkin_ws_ov/src/ && cd ~/workspace/catkin_ws_ov/src`
|
|
2. `git clone https://git.drivecast.tech/pi/openvins_linux.git`
|
|
3. `cd ../`
|
|
4. `catkin build`
|
|
|
|
### Running
|
|
1. from terminal in `~/workspace/catkin_ws_ov/` directory `source devel/setup.bash` and then `roslaunch ov_msckf subscribe.launch config:=kaist`
|
|
2. from second terminal in same directory `source devel/setup.bash` and then `roslaunch global_fusion global_fusion.launch`
|
|
3. from 3d terminal in same directory `source devel/setup.bash` and then `rviz`, and open the open_vins/ov_msckf/launch/display.rviz configuration file.
|
|
4. from 4th terminal in same directory `source devel/setup.bash` and then `rviz`, and open the openv_vins/global_fusion/launch/display.rviz` file.
|
|
5. from 5th terminal in the dataset bag directory: `rosbag play merged.bag gps_fix.bag`
|
|
|
|
#### To run the kaist complex-urban dataset please download the bag file from /mnt/disk-small/podmivan/datasets/complex_urban/urban32-yeouido/bag/ directory. You need to download merged2.bag and gps_fix.bag files.
|
|
|
|
There is a problem with publishing the cubes to two different topics: to the VIO and to the GPS-VIO frames. I have to find the workaround for this problem. For now you can see the AR cubes in the rviz which shows VIO (from ov_msckf directory)
|
|
|
|
### Saving the rendered AR to image sequnece
|
|
1. Turn on the CameraPub (TODO: Change the Camera Info Topic in CameraPub to /ov_msckf/cam0_info and save the .rviz config file)) |