Work

Design, Hydrodynamic Modelling, and Closed-loop Control of an Autonomous Underwater Vehicle for Drowning Rescue in Swimming-pool Environments

Year: 2026
Type: article
Abstract: Even though drowning is not a prevalent cause of death, it still constitutes one of the most common causes of accidental deaths in the world, and is usually said to have a five-to-six-minute window for rescue after it occurs.Surface-based rescue technologies are available but are limited in their ability to respond autonomously at high speeds to counter obstacles and rescue a victim below the water's surface, and existing autonomous underwater vehicles (AUVs) for survey and inspection are not capable of depth range as pool rescue requires, obstacle avoidance, or of high-speed autonomous response.In this paper a six-thruster X-frame UUV called HERO (Hydro Emergency Rescue Operations) is presented for the purpose of safeguarding a victim in a fixed swimming pool environment, autonomously navigating to a distressed swimmer's position as supplied by a victim-detection/perception layer while avoiding static submerged obstacles (avoidance of other, moving swimmers is identified as future work, Section 8).The mechanical design is based on a CAD-model of the vehicle designed for central pressure housing, the four diagonal thruster arms and an integrated rescue-ring deployment system.A full six-degree-of-freedom (6-DOF) dynamic model is formulated based on a marine-craft formulation by Fossen, which includes the rigid-body and added-mass matrices, Coriolis-centripetal coupling, hydrodynamic damping based on Fossen's ITTC/Hoerner paper and hydrostatic restoring forces calculated directly from the CAD geometry.Thruster allocation analysis showed that the allocation matrix has rank 5 of 6 (condition number 15.75 over the nonzero singular values): the four horizontal and two vertical thrusters cannot independently actuate heave, roll, and pitch simultaneously, so one attitude axis (roll) is stabilized passively by the hydrostatic restoring moment rather than by active thrust allocation.The closed-loop response to the fixed-step RK4 integrator-linear zero-order-hold control is studied using a reduced 4-DOF horizontal simulation (to track waypoint position and to provide rescue handling) combined with the full 6-DOF model (to simulate depth and stationkeeping).It is demonstrated in a simulated 100 × 10 m freshwater pool that the vehicle reaches a 5 m patrol depth with an 18.70 s settling time and 4.60% overshoot under idealized, noise-free conditions; the simulation also yields a numerically negligible steady-state residual that is not interpreted as physical depth-measurement accuracy (reported to full precision in Table 8 for reproducibility).In a stress-test rescue scenario, the reduced-order 4-DOF navigation simulation detects and avoids three obstacles within a 4 m obstacle-detection radius while maintaining a minimum vehicle-to-obstacle clearance of 2.95 m, and completes the 100 m transit to a victim at 2 m depth in 86.1 s.This transit time is well within the five-minute (300 s) reference interval used for comparison, leaving a 213.9 s (3.56 min) transittime difference relative to that interval a figure that reflects propulsion and navigation time only and excludes any detection, communication, and deployment latency that would apply in an operational rescue.The results, obtained from a reduced-order 4-DOF navigation model for the horizontal-plane rescue transit (combined with the full 6-DOF model for depth and station-keeping), indicate that the proposed vehicle architecture has the potential to reach a submerged victim within the critical timeframe, even in a stress-test scenario of a wide pool and a deep object; This projection is further corroborated by a full coupled 6-DOF simulation of the same rescue mission (Section 7.5.1)using a stability-prioritized controller, which also reaches the victim and avoids all three obstacles, but more slowly (226.0 s) and with active pitch regulation rather than active roll control, since the rank-5 thrust-allocation matrix cannot deliver independent roll authority; system-in-the-loop testing and physical validation remain necessary next steps.This work establishes platform-level feasibility through integrated simulation; a reproducible, head-to-head comparison against recent control and obstacle-avoidance methods under identical conditions has not been performed, and no claim of comparative superiority is made.The contribution is the specific integration of a rescue-oriented six-thruster UUV architecture not a novel control or obstacle-avoidance algorithm.
Language: English
Open Access status: hybrid
OpenAlex ID: W7222526813
FWCI:
Cites: 34
Cited by:
Related to: 10