A Human-Expertise Based Statistical Method for Analysis of Log Data from a Commuter Ferry

The proposed method in this paper aims to better understand the log data from the commuter ferry. By the method, the mechanism of how the human expertise operates the ferry can be found, and thus help to establish ship intelligence for the autonomous commuting sailing. The log data of sailings with the same departure and arrival ports is of interest in this respect. The method defines different phases of a sailing as different scenarios in terms of the features contained in the collected data. The features are reflected by the ship behavior/response and the ship machinery/actuators. Compared to the typical sailing phases which are distinct to each other, the features can be uncertain when the ferry transfers from the current phase to the sequential. The concept of the transition time window is thus raised to interpret the uncertainty between adjacent phases. Based on the collected data, the human expertise is involved to summarize features and generate empirical criteria for the decomposition. After the whole sailing being split into a sequential-scenario series, statistical heat maps are drawn to illustrate the likelihood site with respect to the collected log data. In practice, log data collected from a customized commuting route in Trondheim are analyzed by the proposed method.