It is an optimization-based multi-sensor state estimator, which achieves accurate self-localization for autonomous applications (drones, UAVs). It is an extension of SLAM, which supports multiple visual-inertial sensor types (mono camera + IMU, stereo cameras + IMU, even stereo cameras only). Features:
- multiple sensors support (stereo cameras / mono camera+IMU / stereo cameras+IMU)
- online spatial calibration (transformation between camera and IMU)
- online temporal calibration (time offset between camera and IMU)
- visual loop closure
Ubuntu 64-bit 16.04 or 18.04. ROS Kinetic or Melodic. ROS Installation
Clone the repository and catkin_make:
cp -r ~/Vision-IMU-control-system-for-Drones/Vision ~/catkin_ws/src/
cp ~/Vision-IMU-control-system-for-Drones/run.sh ~/catkin_ws/scripts
cd ~/catkin_ws
catkin build
source ~/catkin_ws/devel/setup.bash
Start the simulation program, note that iris_stereo_camera.sdf is used in the launch file.
cp init.d-posix/rcS ~/PX4Autopilot/ROMFS/px4fmu_common/init.d-posix/
cp worlds/* ~/PX4-Autopilot/Tools/sitl_gazebo/worlds/
cp -r models/* ~/PX4-Autopilot/Tools/sitl_gazebo/models/
cp -r launch/* ~/PX4-Autopilot/launch/
cd ~/PX4-Autopilot
make px4_sitl_default gazebo
roslaunch px4 indoor1.launch
Then Visual Inertial Odometry starts
Since this package publishes an Odometry type topic, we need to convert it to the topic required by PX4.
cd ~/Vision-IMU-control-system-for-Drones/
python3 vio_tx.py iris
At this time, the output of the PX4 emulation terminal appears as shown below, which represents the successful fusion of visual information.
INFO [ecl/EKF] 1213644000: reset position to ev position
INFO [ecl/EKF] 1213644000: commencing external vision position fusion
INFO [ecl/EKF] 1213644000: commencing external vision yaw fusion
Then establish communication, and take off with keyboard control.
cd ~/Vision-IMU-control-system-for-Drones/
python3 communicate.py iris 0
