Model-based Localization using Unscented Kalman Filter for a Remotely Operated Vehicle

Projekter: ProjektForskning

  • Alrøe, Michael (PI)
  • Frisk, Osvald Lorenz Nygaard (Deltager)
  • Andersen, Kristoffer Fogh (Deltager)
Se relationer på Aarhus Universitet


With the increasing availability of consumer-grade underwater Remotely Operated Vehicles (ROV), a demand for developing robust navigation and control software is met. The Danish engineering company EIVA a/s has acquired and customized such a consumer-grade ROV, the BlueROV 2, and wishes to work towards a more autonomous system. A vital part and requirement of an autonomous control system is gaining an accurate estimate of the ROV’s current relative position. The objective of this thesis is to study and develop a navigation module for a ROV, in order to provide accurate estimations on the position and velocity, based on predictions and measurements. With a basis in the field of robotics and control theory, a mathematical model describing the kinematic and kinetic motion in 6 degrees of freedom is obtained, taking into account rigid-body, hydrodynamical, damping, hydrostatic and thruster contributions.
A system identification approach of immersion tank testing is proposed in order to estimate the unknown system parameters contained in the mathematical model. The approach is based on practical experiments, where the ROV is driven in separate degrees of freedom while tracking its position, with the goal of extracting parameters from recorded data.
A State Observer algorithm based on the Unscented Kalman Filter is examined
and implemented, to incorporate the model predictions and sensor measurements into providing the actual estimations of position and velocity. The model and state observer is implemented as a ready-to-deploy software package, providing a foundation for further developments of a control system. The recorded data obtained from experiments in combination with simulations, is used to verify and evaluate the performance of the state observer. These tests demonstrate promising results, as the position of the ROV can be estimated within an acceptable error margin.
Effektiv start/slut dato28/01/201929/05/2019

ID: 157987163