As autonomous underwater vehicles (AUVs) undertake increasingly frequent roles across Indo-Pacific defence and maritime operations, the ability to make safe and independent decisions at sea becomes essential for reliable operation in complex and dynamic marine environments. One of the highest-risk phases of an AUV mission occurs during surfacing. Whether that be for communication, navigation updates, or recovery, the vehicle is exposed to collision hazards from maritime surface traffic. Existing safe-to-surface (S2S) planning approaches have demonstrated traffic-aware risk minimisation. However, most formulations treat surfacing primarily as a risk-selection problem and implicitly assume that the vehicle can execute the selected trajectory with sufficient precision. This assumption neglects the coupled effects of AUV dynamics, actuator constraints, ascent limitations, and closed-loop tracking error, all of which can cause a nominally safe plan to produce an unintended surfacing position and compromise mission safety.
This work proposes a risk-aware, kinodynamic model-predictive surfacing planner that formulates the S2S decision as a combined optimisation over trajectory risk, surfacing timing, and execution feasibility. Rather than separating path selection from vehicle dynamics, the proposed method incorporates AUV motion constraints and a tracking-error penalty directly into the planning objective, discouraging solutions that appear low risk in simulation but are difficult to execute under realistic vehicle behaviour. The study investigates whether enforcing dynamic feasibility and traffic-risk minimisation produces surfacing decisions that are both lower risk and more practically feasible.
The proposed planner will be evaluated in simulation which a simplified kinodynamic model is used for online planning, while a higher fidelity model incorporating previously unmodelled disturbances are used to assess realised execution performance. Comparisons against established baseline approaches will quantify differences in realised surfacing risk and terminal position accuracy. The expected contribution is a dynamics-aware S2S planning framework that improves the practical reliability of autonomous surfacing decisions in congested and contested maritime environments.