Work

A Composite Observer-Based Optimal Attitude Tracking Control for FWEPAUV via Reinforcement Learning