orca2

Orca2

Orca2 is a ROS2 AUV (Autonomous Underwater Vehicle) based on the BlueRobotics BlueROV2.

Update 17-June-2022: work has shifted to Orca4, which uses ArduSub as the flight controller.

Simulation

Orca2 runs on Ubuntu 18.04 in Gazebo, a SITL (software-in-the-loop) simulator. Installation instructions:

• Install ROS2 Eloquent using these instructions. Use the ros-eloquent-desktop option.

• Install ROS2 development tools using these instructions. Stop after installing the development tools (before “Get ROS 2 code”).

• Initialize rosdep:

sudo rosdep init
rosdep update

• Install Gazebo v9:

sudo apt install gazebo9 libgazebo9 libgazebo9-dev

• Use your favorite Python package manager to install these Python packages:

pip3 install numpy transformations

• Orca2 uses fiducial_vlam_sam, which in turn uses GTSAM. You will need to install GTSAM. Alternatively you can swap fiducial_vlam for fiducial_vlam_sam, but the parameters are different, so the launch files will need to be modified.

• Build Orca2 (orca_driver is not required for the simulation):

mkdir -p ~/ros2/orca_ws/src
cd ~/ros2/orca_ws/src
git clone https://github.com/clydemcqueen/orca2.git
touch orca2/orca_driver/COLCON_IGNORE
git clone https://github.com/clydemcqueen/ukf.git
git clone https://github.com/ptrmu/fiducial_vlam_sam.git
git clone https://github.com/ptrmu/ros2_shared.git
git clone https://github.com/clydemcqueen/sim_fiducial.git
git clone https://github.com/clydemcqueen/astar.git
cd ~/ros2/orca_ws
source /opt/ros/eloquent/setup.bash
# Install system dependencies:
rosdep install --from-paths . --ignore-src
colcon build

• Run the simulation:

cd ~/ros2/orca_ws
source /opt/ros/eloquent/setup.bash
source install/local_setup.bash
export GAZEBO_MODEL_PATH=${PWD}/install/sim_fiducial/share/sim_fiducial/models
source /usr/share/gazebo/setup.sh
ros2 launch orca_gazebo sim_launch.py

A few of the XBox controls:

• window button: disarm (stop all thrusters, disable all buttons except “arm”)
• menu button: arm (enable all buttons, go to ROV mode). Use the joystick to move around in ROV mode.
• A: go to ROV mode
• B: go to ROV mode, and hold depth using the barometer
• X: autonomously keep station at current pose
• Y: autonomously visit all markers in a random pattern

You can also use the ros2 action CLI to start missions. E.g., to visit all markers in sequence:

ros2 action send_goal /mission orca_msgs/action/Mission '{}'

To visit markers 1, 2 and 3 in random order:

ros2 action send_goal /mission orca_msgs/action/Mission '{marker_ids: [1, 2, 3], random: true}'

To keep station at the current pose (requires a good pose):

ros2 action send_goal /mission orca_msgs/action/Mission '{pose_targets: true, keep_station: true}'

To move through 4 poses:

ros2 action send_goal /mission orca_msgs/action/Mission '{pose_targets: true, poses: [ 
{position: {x: 6.5, y: 0.5, z: -0.5}, orientation: {x: 0, y: 0, z: 0, w: 1}},
{position: {x: 6.5, y: -0.5, z: -0.5}, orientation: {x: 0, y: 0, z: 0, w: 1}},
{position: {x: 6.5, y: 0.5, z: -0.5}, orientation: {x: 0, y: 0, z: 0, w: 1}},
{position: {x: 6.5, y: -0.5, z: -0.5}, orientation: {x: 0, y: 0, z: 0, w: 1}},
]}'

See Mission.action for more details.

Orca uses camera(s) and ArUco markers to localize. Make sure the camera has a good view of a marker before starting a mission. If the mission fails to start there might be no marker in view, or the closest marker might be too far away.

Packages

Orca consists of 7 packages:

• orca_msgs defines custom messages

• orca_description provides the URDF file and an URDF parser

• orca_driver runs on the sub hardware, and consists of 2 nodes:

  • barometer_node reads the Bar30 and publishes Barometer messages
  • driver_node subscribes to Control messages and publishes Driver messages
  • Depends on orca_msgs

• orca_shared contains various shared utilities.

  • Depends on orca_msgs

• orca_filter consists of 3 nodes:

  • depth_node subscribes to Barometer messages and publishes Depth messages
  • fp_node subscribes to observations and poses from fiducial_vlam and publishes FiducialPoseStamped
  • filter_node fuses FiducialPoseStamped and Depth measurements and publishes filtered FiducialPoseStamped
  • Depends on orca_description
  • Depends on orca_msgs
  • Depends on orca_shared

• orca_base is the largest package, and consists of 2 primary nodes:

  • auv_node provides the Mission action server and runs autonomous missions, consuming FiducialPoseStamped messages and publishing Control messages
  • rov_node subscribes to Joy messages and publishes Control messages. It also calls the Mission action server to start missions
  • There are several Python nodes that use matplotlib to graph various messages
  • Depends on orca_description
  • Depends on orca_msgs
  • Depends on orca_shared

• orca_gazebo provides a simulation environment, with plugins for thrusters, buoyancy and drag

  • Depends on orca_description
  • Depends on orca_msgs
  • Depends on orca_shared

Status and Future Work

I built the sub and ran it through a series of tests with mixed results. There’s still a lot of work to do in vehicle dynamics (understanding drag, tuning PID controllers).

I’m currently evaluating ROS Nav2. One idea is to use Nav2 and add plugins for specific Orca2 control strategies.

Possible future sensors:

• down-facing camera(s), useful for measuring distance to the seafloor and for visual odometry
• forward-facing and/or down-facing lasers for measuring distance
• sonar for obstacle avoidance