Contact-risk-aware hybrid event-triggered control of an underwater net-cleaning robot with projected reference shaping

Underwater net-cleaning robots repeatedly interact with flexible surfaces whose geometry and contact properties vary along the cleaning path, making fixed references prone to uneven force and pose responses. This article develops a contact-risk-aware hybrid event-triggered predefined-time control strategy within a two-time-scale framework. Within each pass, a normalized force-boundary indicator schedules pose damping and the triggering margin, while adaptive robust compensation and a clock-guarded sample-and-hold mechanism address bounded contact, hydrodynamic, and implementation uncertainties. Between passes, projected shaping updates the path, normal-force, and attitude references under tightened horizon-node constraints, followed by implementation-grid verification. Under local sector conditions and bounded variation, the reference parameters remain feasible and locally bounded and approach an explicit residual neighborhood. The closed-loop analysis establishes bounded signals, practical predefined-time tracking residuals, a residual normal-force envelope, and a positive inter-event interval. Comparative simulations achieve lower pose and force errors with 204 command updates, versus 459 and 316 for two benchmarks. Six pool experiments demonstrate held-command execution, successive reference shaping, and reuse of the final reference under the tested conditions. The simulations quantify command-update efficiency, whereas the pool experiments establish complementary engineering feasibility for repeated flexible-net contact.