USV Simulation

A generic simulation environment for surface vehicles in Gazebo. The wave simulation plugins are provided by the ASV Wave Sim package.

Environment

• Ubuntu 22.04 (Jammy)
• Gazebo Sim, version 7.6.0 (Garden)
• ROS2 Humble

Setup

Build the ASV Wave Simulation package

Installation instructions here

Clone the usv_simulation repository

cd ~/gz_ws/src
git clone https://github.com/g-poulos/usv_simulation

Source the ASV Wave Sim package

cd ~/gz_ws/src/usv_simulation
source source.bash

Plugin setup

All the Python scripts used for the plugins are based on the PyVista library.

Ocean Current and Wind

To use the ocean current and wind plugins, two data tables must first be created using the Python scripts in feature_extraction_scripts directory. The model STL and mass are required to create the tables. The final CSV files must be located to the model meshes directory.

For the scripts to work properly:

• Model origin must be at CoM
• Model must be manifold
• A model mesh with more vertices outputs more accurate area calculation

Steps to create the data tables

1. Draft calculation

By defining the STL file and mass in the compute_draft.py main, the vehicle draft and submerged volume can be calculated. The submerged volume is used for the added mass plugin.

vereniki = "../models/vereniki/meshes/vereniki_scaled3.stl"
print(compute_draft(vereniki, 425, 1025, step_size=0.0001, plot=True))

Parameters

• stl_file: the STL file of the model
• model_mass: the mass of the model
• water_density: the density of the water
• step_size: step size of the algorithm. Accuracy increases with smaller steps
• plot: if True, plots the vehicle floating at the calculated equilibrium

Example output:

Draft:              0.4520999938249588
Calculated Volume:  0.41467056646739586
Theoretical Volume: 0.4146341463414634

2. Projection Area and Torque table

By defining the STL file, draft and number of angles in the multithread_table_calc.py the two tables for the ocean current and wind are generated. Note that complex meshes require more time to be processed.

draft = 0.4520999938249588
stl_file = "../models/vereniki/meshes/vereniki_scaled3.stl"
num_of_angles = 128

poly = pv.read(stl_file)
surface_part = poly.clip_closed_surface(normal=(0, 0, 1),
                                        origin=(0, 0, poly.bounds[4]+draft))
submerged_part = poly.clip_closed_surface(normal=(0, 0, -1),
                                          origin=(0, 0, poly.bounds[4]+draft))

table = calculate_force_torque_table(surface_part, num_of_angles, 16)
write_list_to_file("../models/vereniki/meshes/wind_table.csv", table)

table = calculate_force_torque_table(submerged_part, num_of_angles, 16)
write_list_to_file("../models/vereniki/meshes/current_table.csv", table)

Parameters (calculate_force_torque_table)

• mesh: the part of the vehicle to calculate the table
• num_of_angles: number of angles around the vehicle for which the algorithm will calculate force/torque information. More angles generate tables with greater accuracy but need more time to calculate
• num_of_threads: number of threads used for the operation

3. Model SDF

After the tables have been created the plugins can be used by adding the following lines to the model SDF.

<plugin filename="WaterCurrent" name="water_current::WaterCurrent">
    <link_name>platform_body</link_name>
    <current>
        <speed>
            <min>0.0</min>
            <max>0.3</max>
            <init>0.1</init>
            <stddev>0.05</stddev>
        </speed>
        <direction>
            <min>155</min>
            <max>205</max>
            <init>180</init>
            <stddev>20</stddev>
        </direction>
    </current>

    <density>1025</density>
    <res_coef>0.8</res_coef>
    <update_rate>1500</update_rate>
    <table_file>current_table.csv</table_file>
    <surface_level>1.0</surface_level>
</plugin>

<plugin filename="Wind" name="wind::Wind">
    <link_name>platform_body</link_name>
    <wind>
        <speed>
            <min>0</min>
            <max>7</max>
            <init>2</init>
            <stddev>2</stddev>
        </speed>
        <direction>
            <min>245</min>
            <max>295</max>
            <init>270</init>
            <stddev>20</stddev>
        </direction>
    </wind>
    <density>1.225</density>
    <res_coef>0.8</res_coef>
    <update_rate>1500</update_rate>
    <table_file>wind_table.csv</table_file>
</plugin>

Parameters

• The link_name attribute sets the link on which the plugin applies
• The min, max, init and stddev attributes control the Integrated White Gaussian Noise that generates the speed and direction for both plugins
• The density attribute sets the fluid density
• The res_coef attribute is the drag coefficient in the drag equation
• The update_rate attribute is the rate of messages per second for the speed and direction topics
• The table_file attribute is the table file name that was generated in step 2
• The surface_level attribute sets the level above which the wind plugin applies

Added Mass

Using the submerged volume calculated in the previous section the added mass plugin can be used by adding the following lines to the model SDF.

<plugin filename="AddedMass" name="added_mass::AddedMass">
    <link_name>platform_body</link_name>
    <coef>0.8</coef>
    <density>1025</density>
    <sub_volume>0.41477</sub_volume>
</plugin>

Parameters

• The link_name attribute sets the link on which the plugin applies
• The res_coef attribute is the added mass coefficient
• The density attribute sets the fluid density
• The sub_volume attribute sets the submerged volume of the vehicle calculated using the
  draft script in step 1

Example

Source the workspace before first execution

cd ~/gz_ws/src/usv_simulation
source source.bash

Using SDF world file directly

cd ~/gz_ws/src/usv_simulation
gz sim worlds/waves.sdf

Using launch file

cd ~/gz_ws/src/usv_simulation/launch
gz sim launch launch_waves.py

The launch file only works for the triangular platform model (vereniki) and bridges GZ and ROS topics.

Extras

• The repository usv_controller contains:
  • Two controllers for the triangular platform
  • The dynamic model simulation of the platform that was used to develop the usv_simulation package