Daniel Schwartz's hybrid planner improves USV obstacle avoidance

Daniel G. Schwartz presents a novel approach to path planning for unmanned surface vehicles (USVs), addressing both fixed and moving obstacle avoidance. The system combines a global path planner, which finds a route from start to goal avoiding known fixed obstacles, with a local path planner that handles moving obstacles and unknown fixed ones. The global planner is innovative because it fuses three separate planners: the known Grassfire algorithm, a modified version of Grassfire, and a new, more intuitive version of the Probabilistic Roadmap (PRM). This hybrid reduces computational complexity while maintaining robust pathfinding.

The local planner features a higher-level decision logic that observes the direction of moving obstacles relative to the USV's global path. Instead of running parallel and waiting for an opening, it systematically routes the vehicle behind the obstacle — a more proactive and safer strategy. Simulations, which include an implementation of the well-known D* algorithm for comparison, validate that the new logic outperforms conventional dynamic path planning systems. The work has implications for autonomous naval navigation, swarm coordination, and maritime safety.