Saturated Output-Feedback Hybrid Reinforcement Learning Controller for Submersible Vehicles Guaranteeing Output Constraints

In this brief, we propose a new neuro-fuzzy reinforcement learning-based control (NFRLC) structure that allows autonomous underwater vehicles (AUVs) to follow desired trajectory in large-scale complex environments precisely. The accurate tracking problem is solved by unique online NFRLC method designed based on actor-critic (AC) structure. Integrating the framework including an adaptive multilayer neural network (MNN) and interval type-2 fuzzy (IT2FNN) with high-gain observer (HGO), robust smart observer-based system set up estimate velocities of AUVs, unknown dynamic parameters containing unmodeled dynamics, nonlinearities, uncertainties external disturbances. By employing saturation function design procedure transforming input limitations into risk actuator effectively reduced together nonlinear compensation strategy. A predefined funnel-shaped performance attain certain prescribed output performance. Finally, stability study reveals entire closed-loop signals are semi-globally uniformly ultimately bounded (SGUUB) can provide convergence rate for errors so approach origin evolving inside bound at time.