Session: 06-04-02 Marine Engineering and Technology II

Paper Number: 79583

79583 - The Comparison of Two Kinematic Motion Models for Autonomous Shipping Maneuvers 

Autonomous shipping supported by advanced prediction technologies is widely considered as a high-potential development direction in the maritime industry in the upcoming years. Consequently, such technologies are expected to undertake more complex and challenging ship maneuvers prediction missions in the future. These innovative applications usually require a large number of data sets that have a range from environmental information to ship motion information. The acquirement of data can be based on the estimation of ship system states and serves as a fundamental support to autonomous ship maneuvers. Certain motion models are generally implemented within the above-mentioned estimation process to improve accuracy and robustness. In contrast to a wider discussion of different motion models for the applications of ground vehicles and aircraft, the studies of appropriate motion models for maritime transport applications are still insufficient. Therefore, a dedicated study of ship motion models is indispensable to adapt to emerging autonomous shipping applications.

In this paper, two ship models which both are based on curvilinear motions but described in different reference coordinate systems are examined. One model is described in the inertial reference frame and another model is in the ship-body fixed reference frame. To avoid the complexity of obtaining the hydrodynamics parameters which tend to change significantly in different weather and sea conditions, both models are founded on the kinematic motion model family. The data sets are collected from a container ship’s different maneuvers which are reproduced by a ship simulator. These simulated data sets have been used for the comparison work in this study. Both actual vessel motion values as well as measurement values during the simulated maneuvers are also collected. Then, the measurement data sets which include GPS and other shipboard sensors data are fused in an Extended Kalman Filter and Unscented Kalman Filter to generate the posterior state estimates which are supposed to have higher accuracy. Finally, the performance of both filters is verified with the collected actual vessel motion values from the simulator.

Presenting Author: Yufei Wang The University of Tromsø – The Arctic University of Norway, Department of Technology and Safety

Authors:

Yufei Wang The University of Tromsø – The Arctic University of Norway, Department of Technology and Safety
Lokukaluge Prasad Perera The University of Tromsø – The Arctic University of Norway, Department of Technology and Safety
Bjørn-Morten Batalden The University of Tromsø – The Arctic University of Norway, Department of Technology and Safety