Stingray is a ROS based framework for autonomous underwater vehicles (AUV)

• ROS2 iron - base framework
• YOLOv5 - for object detection
• pytransitions - for state machine
• serial - for communication with stm32 and etc.

• Initialize and update git submodules used in project:

git submodule update --init --recursive

• Install requirements from yolov5

• Install ros packages:

sudo apt-get install ros-iron-serial ros-iron-usb-cam ros-iron-rosbridge-server ros-iron-image-view ros-iron-zbar-ros

• Install pytransitions dependencies

sudo apt install graphviz-dev
pip3 install pygraphviz transitions

• Build

source /opt/ros/iron/setup.bash
catkin_make

Setup workspace before you start to work:

source install/setup.bash

See the example of custom launch file here

• Clone and build our simulator (now it's only for sauvc competition).
• Run simulator.

roslaunch stingray_startup main.launch simulation:=true 

➦ Full description here

Use arg:

qr_launch:=true 

Show your qr code to vehicle camera.

If stop qr code has been detected then the running launch file will be stopped.

• hardware_connection:=false - disable connection btw jetson and stm32 via serial (uart_driver)
• stream:=true - enable web video stream from all cameras
• debug:=true - enable image_view nodes and publishing output videos after object detection
• file_cam:=true - provide input videos from file
• record_raw:=true - enable recording video from all cameras
• record_output:=true - enable recording video after object detection

stingray_startup package contains launch files for running the whole system.

Use main.launch as the base for your custom launch file. Include like this:

<!-- MAIN -->
    <include file="$(find stingray_startup)/launch/main.launch">
        <arg name="ROS_OUTPUT" value="$(arg ROS_OUTPUT)" />
        <arg name="DEBUG" value="$(arg DEBUG)" />
        <arg name="STREAM" value="$(arg STREAM)" />
        <arg name="SIMULATION" value="$(arg SIMULATION)" />
        <arg name="HARDWARE_CONNECTION" value="$(arg HARDWARE_CONNECTION)" />
        <arg name="QR_LAUNCH" value="$(arg QR_LAUNCH)" />
        <arg name="QR_CAMERA" value="$(arg QR_CAMERA)" />
        <arg name="QR_LAUNCH_PACKAGE_NAME" value="$(arg QR_LAUNCH_PACKAGE_NAME)" />
        <arg name="QR_NAME_PATTERN" value="$(arg QR_NAME_PATTERN)" />
    </include>

*see args inside launch file or below

Add camera to your custom launch file file like that:

<!-- FRONT CAMERA -->
    <include file="$(find stingray_startup)/launch/camera.launch">
        <arg name="REAL_CAM" value="true" unless="$(arg SIMULATION)" />
        <arg name="SIMULATION_CAM" value="true" if="$(arg SIMULATION)" />

        <arg name="CAMERA_NAME" value="$(arg FRONT_CAMERA)" />
        <arg name="CAMERA_PATH" value="$(arg FRONT_CAMERA_PATH)" />
        <arg name="CAMERA_TOPIC" value="$(arg FRONT_CAMERA_TOPIC)" />

        <arg name="ROS_OUTPUT" value="$(arg ROS_OUTPUT)" />
        <arg name="SHOW" value="$(arg SHOW)" />
        <arg name="DEBUG" value="$(arg DEBUG)" />
        <arg name="RECORD_RAW" value="$(arg RECORD_RAW)" />
        <arg name="RECORD_OUTPUT" value="$(arg RECORD_OUTPUT)" />
        <arg name="RECORD_DIR" value="$(arg RECORD_DIR)" />
    </include>

Specify camera args: CAMERA_NAME, CAMERA_PATH, CAMERA_TOPIC.

Also, you can specify the camera type: REAL_CAM, SIMULATION_CAM or FILE_CAM.

RECORD args for recording video from camera.

Add object detection to your custom launch file file like that:

<!-- OBJECT DETECTION -->
    <include file="$(find stingray_startup)/launch/object_detection.launch">
        <arg name="IMAGE_TOPIC_LIST" value="$(arg FRONT_CAMERA_TOPIC) $(arg BOTTOM_CAMERA_TOPIC)" />
        <arg name="WEIGHTS_PACKAGE_NAME" value="$(arg WEIGHTS_PACKAGE_NAME)" />
        <arg name="ROS_OUTPUT" value="$(arg ROS_OUTPUT)" />
        <arg name="DEBUG" value="$(arg DEBUG)" />
    </include>

Provide the list of topics to subscribe to: IMAGE_TOPIC_LIST. Provide the name of the package with weights: WEIGHTS_PACKAGE_NAME.

• Edit config.yaml to add your labels
• Put best checkpoint of yolov5 as best.pt in weights folder

Nodes:

• harware_bridge - abstract bridge node between hardware and ros
• uart_driver - node for communication with stm32 via uart

Contain gazebo_bridge node for communication with gazebo simulator.

Contain nodes for working with lifter device.

FSM package allows you to create a missions for robot.

• AUVController class is a high-level mission controller.
• AUVMission allows you to create custom missions.
• SubscriptionEvent listens to the topic and triggers the event when the message is received. Allows you to create custom events.

Use ObjectDetectionEvent to trigger the event when the object is detected.

Contains c++ nodes for controlling robot movement - basic, common and patterns.

Contains config files and util libs

Config files:

• control.json - all stuff for control algos and control system
• hardware.json - all stuff for different devices, hardware communication and etc.
• ros.json - names for ROS topics, services, actions
• simulation.json- all stuff for simulation mode

Libs:

• utils.py - util python methods
  • load_config - loads config file with name

Contains launch files and qr trigger node.

Launch files:

• main.launch
• camera.launch
• object_detection.launch

Contains util libs and nodes.

• json.hpp - cpp lib for json files

Qr trigger node has parameters:

• launch_pkg_name - package name with launch/ directori which contains launch files you want to trigger
• name_pattern - specify this prefix to trigger specific launch files

Generate qr code from launch file name without custom prefix and .launch.

Example: you have stingray_qr_mission.launch file. stingray_qr_ is the prefix which you pass to qr_trigger node as name_pattern parameter. Also you don't need .launch to generate qr code. Eventually you need to generate mission to qr code

Launch main.launch with param:

qr_launch:=true 

and you'll able to trigger launch files with qr codes.

If stop has been detected then running launch file will be stopped.

Contains node for recording video from camera.