Skip to content

Latest commit

 

History

History
132 lines (108 loc) · 3.88 KB

README.MD

File metadata and controls

132 lines (108 loc) · 3.88 KB
Raw

OAFD_MONOCULAR

this work is my Master thesis research ：3D mapping by a monocular camera with application to realtime obstacle avoidance control of quadrotor

use monocular depth estimation and monocular visual SLAM to construct an autonomous flight system for a monocular-only quadrotor.

Obstacle.avoidance.flight.of.drone.use.monocular.outdoor_03.mp4

more experiment video

Prerequisites

Tested on Ubuntu 20.04 & 18.04

ORB_SLAM3

Use the ORB_SLAM3(v0.4beta)as VSLAM(monocular)

cd ~
git clone https://github.com/UZ-SLAMLab/ORB_SLAM3.git ORB_SLAM3
  • Make changes to the source code if necessary to build successfully. For Ubuntu 20.04, you will need to change CMakeList from C++11 to C++14. I have incorporated the changes in this fork.
  • Build:
cd ORB_SLAM3

git reset --hard a80b4677009e673b9939a7e91e6ea7bcb5090294

chmod +x build.sh
./build.sh
  • Make sure that libORB_SLAM3.so is created in the ORB_SLAM3/lib folder. If not, check the issue list from the original repo and retry.

orb_slam3_ros_wrapper

  • Clone the package. Note that it should be a catkin build workspace.
cd ~/catkin_build/src/
git clone https://github.com/geturin/orb_slam3_ros_wrapper.git
  • Open CMakeLists.txt and change the directory that leads to ORB-SLAM3 library at the beginning of the file (default is home folder ~/)
cd ~/catkin_ws/src/orb_slam3_ros_wrapper/
nano CMakeLists.txt

# Change this to your installation of ORB-SLAM3. Default is ~/
set(ORB_SLAM3_DIR
   $ENV{HOME}/ORB_SLAM3
)
  • Build the package normally.
cd ~/catkin_build/
catkin build

  • Next, copy the ORBvoc.txt file from ORB-SLAM3/Vocabulary/ folder to the config folder in this package. Alternatively, you can change the voc_file param in the launch file to point to the right location.

  • (Optional) Install hector-trajectory-server to visualize the trajectory.

sudo apt install ros-[DISTRO]-hector-trajectory-server

MiDAS

use the MiDAS as monocular depth estimation model

  • create symlink for OpenCV:
sudo ln -s /usr/include/opencv4 /usr/include/opencv
  • download and install MiDaS:
source ~/.bashrc
cd ~/
mkdir catkin_ws
cd catkin_ws
git clone https://github.com/geturin/MiDaS
mkdir src
cp -r MiDaS/ros/* src

chmod +x src/additions/*.sh
chmod +x src/*.sh
chmod +x src/midas_cpp/scripts/*.py
cp src/additions/do_catkin_make.sh ./do_catkin_make.sh
./do_catkin_make.sh
./src/additions/downloads.sh

EGO_PLANNER

use the EGO_PLANNER as local planner

sudo apt-get install libarmadillo-dev
git clone https://github.com/ZJU-FAST-Lab/ego-planner.git
cd ego-planner
catkin_make
source devel/setup.bash
roslaunch ego_planner simple_run.launch

Installation

sudo apt-get install ros-$release-ros-numpy
sudo apt-get install ros-$release-tf2-sensor-msgs
cd catkin_ws/src
git clone https://github.com/geturin/OAFD_Monocular.git
cd ../
catkin_make

How to use

  • Connect to tello's wifi

  • run ORB_SLAM3 & Midas & ego planner

roslaunch oafd slam.launch
roslaunch midas_cpp midas_cpp.launch model_name:="model-small-traced.pt" input_topic:="/camera/image_raw" output_topic:="/camera/depth" out_orig_size:="true"
roslaunch ego_tello.launch
  • when ORB_SLAM3 get ready
roslaunch oafd tello.launch
  • When tello finish taking off
rosrun oafd planner_ctrl.py