About

This repository contains files and instruction to setting running either OpenVINS or OrbSlam2 on your system.

OpenVins

The primary cameras used in the lab are realsense d455 ,the calibration files in the calibration directory should provide a decent enough accuracy for either VIO pipelines. Make sure docker is installed on your system , visit here for more information on how to set it up.

Setup

Open your terminal of choice and run these sequentially.

Create a directory for odometry related projects. This step is not required

mkdir ~/odometry && cd ~/odometry

Clone the OpenVINS repository

git clone https://github.com/rpng/open_vins.git

Change directory and rename the ROS2 Ubuntu DockerFile

cd open_vins && cp Dockerfile_ros2_20_04 Dockerfile

Build the image and name it as openvins

docker build . -t openvins

Create a bash file named run_openvins_docker.sh by running

touch run_openvins_docker.sh

Copy the commands below into this file

xhost +
export DOCKER_CATKINWS=$(pwd)
docker run  -it --runtime=nvidia  --net=host --gpus all \
    --env NVIDIA_DRIVER_CAPABILITIES=all --env DISPLAY=$DISPLAY \
    --env QT_X11_NO_MITSHM=1  --volume /tmp/.X11-unix:/tmp/.X11-unix --privileged  \
    --volume /dev:/dev \
    --mount type=bind,source=$DOCKER_CATKINWS,target=/catkin_ws/src/open_vins  $1 open_vins bash

Copy the scripts folder from this repo and place it in the open_vins directory

Next run this script

bash run_openvins_docker.sh

You terminal should look like as below

Change directory to catkin_ws and build open_vins by running

cd catkin_ws && colcon build

Debugging build failure

Normally this should build openvins but I have not had this luck.

These are unoffical steps

The various packages depend on each other ,hence build and install them individually then run `colcon build`

The order is important

cd /catkin_ws/src/open_vins/ov_init && rm -rf build && cmake -B build && cd build && make install
cd /catkin_ws/src/open_vins/ov_core && rm -rf build && cmake -B build && cd build && make install
cd /catkin_ws/src/open_vins/ov_eval && rm -rf build && cmake -B build && cd build && make install
cd /catkin_ws/src/open_vins/ov_msckf && rm -rf build && cmake -B build && cd build && make install
cd /catkin_ws && colcon build

It should be successfully now

Running Monocular VIO

For the rovers in the lab , use the configuration file in configs/rs_d455 The parameters that have been changed from the defaults are as follows: - try_zupt:true - init_window_time: 0.5 - init_imu_thresh: 1.5 - max_slam: 25

Update the file in open_vins/config/rs_d455/estimator_config.yaml with the above parameters and replace with the config files provided in this repository.
NB: Run colcon build && source /catkin_ws/install/setup.sh after any update to config files for the changes to take effect

I assume you are familiar with ros realsense ,if not see here. Run the run_realsense.sh script in the scripts folder to run the realsense camera and map the streams to correct topics. Confirm the topic names of the camera and imu match the names in your open_vins/config/rs_d455/kalibr_imucam_chain.yaml and open_vins/config/rs_d455/kalibr_imu_chain.yaml files.
Next, copy and run this command ros2 launch ov_msckf subscribe.launch.py config:=rs_d455 max_cameras:=1 use_stereo:=false rviz_enable:=true or run the run_msckf.sh file provided in the scripts folder

Running Stereo VIO

First ,I recommend using the run_realsense_stereo.sh script.
Next open a second terminal in your docker container and run ros2 param set /camera/camera depth_module.emitter_enabled 0
Copy the rs_d455_stereo folder into open_vins/config (NB: Remember to build)
Next run ros2 launch ov_msckf subscribe.launch.py config:=rs_d455_stereo rviz_enable:=true or run the run_stereo_msckf.sh file

Consider calibrating the camera for better accuracy

OrbSLAM

Soon