Autonomous Underwater Navigation

The field studies how underwater vehicles estimate their position and move safely when visibility is poor, currents disturb motion, and acoustic communication is delayed or unavailable. Localization commonly fuses inertial sensors with Doppler velocity logs, sonar, cameras, depth sensors, and acoustic positioning, using Kalman filters, SLAM, and learned estimators. For planning and control, researchers combine methods such as model predictive control, line-of-sight guidance, mapping and search, reinforcement learning, and physics-based world models. Simulation benchmarks enable broad testing, while pool and sea trials assess real-world reliability; a central challenge is making methods robust to unfamiliar environments and sensor failures without excessive computation or expensive hardware.