H-SPAR: Hydrodynamic-aware Simulation for Particle Transport and Autonomous Robots
This arXiv preprint describes H-SPAR, an open-source ROS 2/Gazebo simulator that couples a shared precomputed flow field to both current-induced drag on an uncrewed surface vehicle and Lagrangian advection of sampled particles. In path-planning tests, SVF-RRT* cuts planned upstream cost by 69.4% versus RRT*, but the advantage falls to 41.7% during execution with time-varying currents; sweep orientation shifts particle sampling rates by up to 22.2% under the full hydrodynamic configuration. It matters because marine autonomy and particle-sampling performance can now be benchmarked together under consistent hydrodynamics.