H-SPAR: Hydrodynamic-aware Simulation for Particle Transport and Autonomous Robots
H-SPAR: Hydrodynamic-aware Simulation for Particle Transport and Autonomous Robots
Navid Zarrabi (1), Nariman Yousefi (2), Sajad Saeedi (3)
(1) Department of Mechanical, Industrial, and Mechatronics Engineering, Toronto Metropolitan University, Toronto, Ontario, Canada.
(2) Department of Chemical Engineering, Toronto Metropolitan University, Toronto, Ontario, Canada.
(3) Department of Computer Science, University College London, London, United Kingdom.
https://github.com/naviiidz/h-spar-sim.git
H-SPAR uses (a) fine-grained velocity vector fields to simultaneously (b) simulate the advective transport of particles and (c) model spatially and temporally varying currents. The variable currents are used to
(d) calculate hydrodynamic drag forces at the USV reference frame. By integrating Lagrangian particle transport with hydrodynamic-aware USV simulation (e-f) H-SPAR enables the evaluation of particle sampling missions under marine flow conditions.
ROS 2 hosts current-aware path and motion planning, hydrodynamic flow modeling, and Lagrangian particle transport, while Gazebo provides the simulated aquatic environment, virtual sensors, and USV dynamic models. Sensor measurements are transmitted from Gazebo to ROS 2, while ROS 2 provides thruster commands and hydrodynamic forces to the simulated vehicles, enabling closed-loop flow-aware navigation and particle monitoring.
H-SPAR front-end visualization in Gazebo, including the USV, environment, sensors, buoyancy, waves, and wind, alongside the back-end visualization in RViz for motion planning, coordinate transformations, particle transport, and sampling simulation.
Velocity fields capture spatially varying currents for realistic USV interaction.
Coupled Euler–Lagrangian particle transport illustrates how plastics move and accumulate in flows.
Particles displayed as Rviz2 markers, rotating in a circular pattern at a river bend.
H-SPAR is able to simulate particles and USVs dynamics using solver driven flow fields. A probabilistic sampling method estimates whether particles encountered by the USV are sampled as a function of their relative motion with respect to the robot.