A risk-responsive adaptive expansion Hybrid A* route planning method for bridge-area waters based on a constraint-gated safety cost field

To address the mismatch between historical navigation information and local physical constraints, insufficient prioritization of physical risks under globally fixed linear weighting, and limited response of fixed expansion strategies to local risk variations, this study proposes RAE-HA* (Risk-responsive Adaptive Expansion Hybrid A*) for bridge-area route planning. Using AIS (Automatic Identification System) tracks, it constructs a KDE navigation density field and extracts the historical high-density navigation space through KDE-HDR (Kernel Density Estimation–Highest Density Region). Risk components from AIS density, HDR boundary, shoreline boundaries, and bridge piers are integrated by constraint gating to form a safety cost field. Riskresponsive adaptive expansion is incorporated into Hybrid A*, with safety cost, path length, turning cost, curvature-change regularization, and heading consistency considered in route evaluation. Experiments show it captures historical navigation patterns and physical constraints, generating routes that balance efficiency, risk avoidance, and spatial safety margins.