Uncertainty-Aware Active Source Tracking of Marine Pollution using Unmanned Surface Vehicles

Depedencies

• Gazebo Harmonic (not ros-gz)
• asv_wave_sim
• ardupilot_gazebo, including the additional models

Usage

Build the provided dockerfile, run the container based on the built image and allow it to access the network of the host machine.

On the host machine: Start the simulation environment, sdf files can be found in experiments folder:

gz sim -v4 test_world.sdf

On the container:

Start the SITL:

cd /work/ardu_ws/src/ardupilot
Tools/autotest/sim_vehicle.py -v Rover -f rover-skid --model JSON  --console --map --custom-location=$YOUR_PREFERRED_LOCATION --enable-dds
# Arm the vehicle
# Set Guided Mode

Start the micro-ROS agent:

source /work/ardu_ws/install/setup.bash
ros2 run micro_ros_agent micro_ros_agent udp4 -p 2019

Launch the pollution simulation:

source /work/ardu_ws/install/setup.bash
source /work/ros_ws/install/setup.bash
ros2 launch pollution_sim sim_odom_tf.launch.py scenario:=$SCENARIO_NAME

Launch the tracking server and client:

source /work/ardu_ws/install/setup.bash
source /work/ros_ws/install/setup.bash
ros2 launch pollution_ipp source_loc.launch.py start_client:=true scenario:=$SCENARIO_NAME st:=$START_POSITION_NAME

Credits

This repository used code from:

• odom_to_tf_ros2
• rapidcsv

The local estimation kernal and propagation was based on the implementation of GasSourceLocalization.