•  
  •  
 

Abstract

Understanding navigator errors can be used as a basis for preventing collisions. Here, Hidden Markov Models (HMMs) are applied to navigator errors of fishing and merchant ships in collisions. Information on navigator errors providedin accident investigation reports issued by the Korean Maritime Safety Tribunal is surveyed. The surveyed information is converted into data using the framework of slips, lapses, and mistakes proposed in this study. The results of framework data states are inferred using the unsupervised learning of HMMs, with sequences of navigator errors contained in the error data. The expected errors learned using the navigator error models for fishing and merchant ships are compared and analyzed. Results show that the proposed models can be used to define the types of navigator errors and provide an understanding of these errors in a ship collision event. Several interesting results are also discussed, including some ideas on how to alter navigator errors.

Download
COinS