Autonomous surface vehicles (ASVs) operating in real-world maritime environments must cope with dynamic disturbances, perception uncertainties, and mission-critical reliability requirements. These challenges become more pronounced in competition-based scenarios, where strict task constraints and time limitations demand robust and predictable system behavior. This study presents a reliability-oriented autonomy architecture for ASVs designed to operate under such constrained conditions.
The proposed system integrates a multi-modal perception framework combining stereo vision and LiDAR-based cross-validation to improve detection robustness in complex maritime environments. A layered autonomy structure is implemented, where high-level decision-making is handled through a behavior-tree-based framework, ensuring modularity and fault tolerance. Navigation and control are achieved using a hybrid planning strategy, incorporating global path planning and model predictive path integral (MPPI) control to address non-holonomic vehicle dynamics. A dual-layer state estimation approach based on GNSS, IMU, and sensor fusion enhances localization accuracy.
The system is evaluated through both simulation and real-world experiments. Performance is assessed using quantitative metrics, including detection accuracy, trajectory tracking error, mission completion rate, and system reliability indicators. Experimental results demonstrate consistent performance across varying environmental conditions, with improved robustness achieved through sensor redundancy and structured decision-making.
The findings indicate that the proposed architecture provides a reliable and scalable solution for ASV autonomy under constrained operational scenarios. The simulation-to-real validation further highlights the applicability of the approach in practical maritime missions, offering insights for future developments in robust autonomous marine systems.
Interpretable AI for Global Ocean Health Forecasting and Marine Environmental Risk Stratification Using the Ocean Health Index