New control method slashes AUV tracking error by 94% under ocean currents
Torpedo-shaped underwater vehicles now track paths with 99% less drift.
Mohammad Sabouri’s new paper tackles one of the hardest problems in autonomous underwater vehicles (AUVs): 3D trajectory tracking under unknown ocean currents for torpedo-shaped, underactuated platforms with non-minimum-phase dynamics. The proposed architecture combines a three-stage observer that estimates both vehicle state and relative-velocity, feeding into a nonlinear feedforward term for dominant current rejection. An LMI-certified LPV-ℋ∞ correction layer then handles residual errors, with feedback linearization yielding a constant input matrix that enables convex synthesis without cross terms.
To prove stability, Sabouri derives a residual-level break-even law linking surge disturbance to current-estimation error, and uses singular-perturbation analysis to guarantee local practical uniform ultimate boundedness. In REMUS simulations over three trajectories and four current scenarios—including strong, time-varying, and cross-current conditions—the method achieved 89-96% reduction in current estimation error, about 99% reduction in translational residuals, and slashed RMS tracking error from 4.04 meters to just 0.24 meters. This represents a dramatic improvement over existing control schemes for AUVs operating in realistic ocean environments.
- Three-stage observer estimates relative velocity for dominant current rejection in underactuated AUVs with non-minimum-phase dynamics.
- LMI-certified LPV-ℋ∞ robust correction layer reduces translational residuals by ~99% compared to baseline.
- REMUS simulations show RMS tracking error drops from 4.04m to 0.24m across diverse current scenarios.
Why It Matters
Enables precise AUV navigation in strong currents, crucial for ocean exploration, pipeline inspection, and military missions.