Hybrid A* Path
Planner

Standard A* produces paths on a discrete grid that violate differential drive kinematics. The resulting zig-zag trajectories require the robot to stop and turn in place at each grid cell transition — making smooth autonomous navigation in cluttered competition environments impossible without fundamental re-engineering of the planning layer.

Implemented Hybrid A* with continuous state expansion in (x, y, θ) — replacing the discrete grid with kinematically feasible motion primitives using Dubins path curves. This enforces curvature bounds that reflect the robot's minimum turning radius, producing trajectories that a differential drive can follow without stopping. A velocity profiler then generates smooth, dynamically feasible speed profiles tuned to the robot's physical constraints, and the planner integrates directly with Nav2's Costmap2D for collision checking.

Engineered natively as a custom C++ Nav2 global planner plugin, implementing the nav2_core::GlobalPlanner interface for drop-in compatibility with any standard ROS2 navigation setup. The Dubins primitive generator constructs kinematically feasible arc segments that respect the robot's curvature constraints at each expansion step. The holonomic-with-obstacles heuristic guides search efficiently even in high-obstacle-density environments. Integrates with Nav2's lifecycle management and Costmap2D infrastructure without any changes to the broader stack.

Robot successfully navigates highly cluttered environments using smooth, dynamically feasible trajectories — no stop-and-turn mechanics. The planner drops in as a replacement for Nav2's default GlobalPlanner and works transparently with the rest of the navigation stack, enabling competition-grade autonomous navigation without modifying the broader system.