Unmanned-Surface-Vessel-Studies

This project is still under construction and update progress, and includes details explanation. Please check the repository regularly for some updates

A Master of Engineering Academic Project:

• Guide you to understand surface vessel modelling, model-based and learning-based controllers design, and environmental distubances modelling

• MATLAB code to simulate the control system to perform the infinite loop path following

• Requirements for the trajectory tracking simulation:

  • MATLAB 2020a later

  • Marine System Simulator

  • Gaussian Process Regression Package

Content:

• Problem statement
• USV modelling
• Controller design
  • Feedback linearization as inverse dynamics (FBL)
  • PD+FBL controller
  • Nonlinear PID controller
  • Linear Model Perdictive Controller(MPC)+FBL
  • Nonlinear MPC
  • Hybrid MPC+FBL
  • GRR+PD+FBL controller
  • GPR+Hybrid MPC+FBL controller
• Linear Extended Kalman Filter
• Environmental disturbances

Problem statement

• Design, simulate, as well as possibly build and test, an autonomous water buoy that uses GPS and on-board motors to perform tracking
  

  designed trajectories in the presence of environmental disturbances including wind and wave.

• Software programming in MATLAB was used to simulate and test the design of the USV control system

• The project is assumed to be operated under the general condition of a slight sea state generated by statistical algorithms such that
  

  simulation results are presented to analyze how the vessel can overcome disturbance challenges.

USV modelling

• Heron M300 acts as the main USV platform adapted to software simulation.

• A twin-hull and pontoon-style vessel for aquatic research.

• Heron M300 is assumed to mitigate the heave, pitch, and roll response of the payload tray when the Heron M300 performs tasks on the
  

  water surface for simplifying system identification.

Controller design

• Seven proposed controllers are designed to evaluate the tracking performance whatever the environmental distubances exist.

• Model-based controllers are briefly presented below and please refer to the code in details.

• Learning-based controllers are showed in the code section to explain more details.

Linear EKF

• For simplicity, the Linear EKF is presented as the main estimator based on the vessel system states or poses

• The linear EKF is acceptable in the project to create a guidance system for the path tracking task and estimate the vessel poses by filtering
  

  sensors noise and a basic estimation algorithm

• In the reality for robot localization, the Heron vessel observes the vessel states from its proprioceptive sensors and carries out waypoint
  

  navigation for the path following.

Environmental disturbances

• Wind and wave are two major sources to disturb the route tracking behaviour in the project

• The raw wind and wave data by the andom walk algorithm

Parts of preliminary simulation

• Animation results without the disturbances are shown each of controllers.md files.

• If you are intersted in the path following performances under the distubances, please download the code with MSS and
  GP packages and have fun with them!

• Detailed statistical data including RMSE and maximum distance errors is not shown in this repository. Please drop me an email for the PDF
  report version.

• Thruster force control is not perfectly presented due to excessive force outputs of PID, NMP, and GPMPC_FBL controllers at the beginning,
  and further gain tunning task is needed to approach to the real-world testing.