"Direct Speed-Optimized Tube Based Model Predictive Control for Unmanned Underwater Vehicle Trajectory Tracking"

"Conventional unmanned underwater vehicle (UUV) trajectory tracking employs a hierarchical architecture with controller, thrust allocation, and rotation speed mapping, which suffers from cumulative mapping errors and constraint inconsistency across domains.This paper proposes a tube-based model predictive control (Tube-MPC) method that directly optimizes propeller rotation speeds (PRS) for UUV trajectory tracking. First, a unified mathematical model integrating thrust dynamics is established, using PRS as optimization variables to fundamentally eliminate error propagation from multi-stage mapping.To address input non-affine system characteristics, the Tube-MPC theory is extended to this class of systems. A Linear Quadratic Regulator based vector sliding mode auxiliary control law is designed to provide independent robust compensation channels for each thruster. An explicit robust positive invariant set for discrete-time sliding mode systems is constructed to derive a uniform bound on auxiliary inputs, enabling precise calculation of constraint tightening and ensuring the actual system always satisfies physical constraints. A move-blocking strategy combining first-step optimization and terminal feedback reduces online decision variables to single-step control dimensions, significantly lowering computational burden. Theoretical analysis proves recursive feasibility, constraint satisfaction, and asymptotic stability. Pool experiments demonstrated significant improvements in tracking accuracy, and control smoothness compared to hierarchical control."