Marine Autonomous Vehicles

We develop autonomous surface vessels (ASVs) and autonomous underwater vehicles (AUVs) end to end — hullform and platform design, the sensing and state-estimation stack, and the guidance, navigation and control modules that let a vehicle carry out a mission on its own.

Platforms and field work

After joining IIT Madras we set up the MAV Laboratory and built free-running scaled models of ships to test autonomy algorithms. The fleet includes models of a container ship, a very large crude carrier (VLCC), and a navy destroyer. The vehicles are fitted with GPS and IMU sensing. Sensor data is fused through an extended Kalman filter for localisation, and onboard computers run the autonomy stack. The models are deployed regularly on the lake at IIT Madras as a test bed for experimental validation.

Guidance, navigation and control

A core thread is path following and collision avoidance for ships. Controllers developed in simulation — including reinforcement-learning methods and classical approaches such as artificial potential fields that respect COLREGs — are transferred to the free-running models and checked against traditional PID baselines in calm water and under wind disturbance.

Work also covers cooperative behaviour between multiple vehicles, and homing and docking for underwater vehicles returning to a station on a floating platform. Algorithms move from simulation through hardware-in-the-loop testing before they are tried on the water.

Simulation

MAVSim, the lab’s browser-based marine vehicle simulator, lets students and researchers experience realistic ship dynamics — inertia, hydrodynamics and manoeuvring — and supports teaching and outreach alongside the physical fleet.