IIT Madras

OE5005

Marine Autonomous Vehicles

From autonomy levels and vehicle dynamics to guidance, navigation, control, and ROS2 deployment on an unmanned surface vessel.

A project-based course on the guidance, navigation and control stack that makes an unmanned or autonomous surface vessel work — from simulation through sensor fusion to implementation in ROS2 on a real vehicle.

Course content

Autonomy overview

  • Autonomy levels for marine vehicles and technology readiness levels
  • Regulatory concerns; COLREGs

Kinematics and dynamics

  • Reference frames, coordinate transformations, Euler angles, and quaternions
  • Newton–Euler equations of motion; Coriolis, hydrostatic, added-mass, and dissipative forces

Guidance

  • Line-of-sight (LOS) guidance, Lyapunov stability, and vector-field guidance
  • Proportional and integral LOS
  • Obstacle avoidance: artificial potential fields and velocity obstacles (as time permits)
  • Sensors overview — GPS and IMU; wave and noise filtering
  • Fixed-gain, Luenberger, Kalman, and Extended Kalman filters
  • Sensor fusion for state estimation

Control

  • PID control and successive loop-closure autopilots
  • Pole placement for SISO and MIMO systems; control-law stability
  • Deep reinforcement learning for collision avoidance (as time permits)

Practical

  • Implementation of autonomy algorithms in ROS2
  • Deployment on an autonomous surface vessel with GPS, IMU, and related sensors

Learning objectives

By the end of this course, students will be able to:

  • Recognize the different levels of autonomy and recollect the current regulations governing autonomy of marine vehicles
  • Differentiate between traditional and modern methods of guidance, navigation and control
  • Develop a simulation environment of a marine vehicle incorporating the kinematics and dynamics
  • Implement guidance, navigation and control algorithms in a simulated environment
  • Use ROS2 to interface with the sensors and actuators in a marine vehicle
  • Design parameters of an Extended Kalman Filter (EKF) to fuse the data from multiple sensors
  • Implement guidance, navigation and control algorithms on a real vehicle and implement waypoint tracking and collision avoidance