Inference processes in the automatic communication system for autonomous vessels

Abstract

The era of autonomous ships has already begun in maritime transport. The 30-year forecast for the development of marine technologies predicts many autonomous vessels at sea. This will necessitate radical implementation of new intelligent maritime navigation systems. One of the intelligent systems that has to be implemented is a collision avoidance system. The inference process is a key element of autonomous manoeuvres. These authors propose an inference process that enables exchange of information, intentions and expectations between autonomous vessels and gives them an opportunity to negotiate a safe manoeuvre satisfying all the parties concerned. The model of inference in the communication process has been presented. Methods and algorithms for information exchange and negotiation have been developed. These models were implemented and tested under various conditions. The results of case studies indicate that it is possible to effectively communicate and negotiate used the developed method. To demonstrate the effectiveness of the presented approach over 30 random simulations have been carried out. After successful laboratory tests, over 100 scenarios were executed in quasi-real conditions and fully operational conditions. Tests were carried out in the center of the Foundation for the Safety of Navigation and Environmental Protection on Lake Silm in Iława, Poland. In the framework of project AVAL (Autonomous Vessel with an Air Look) POIR.04.01.04-00-0025-16,  82 random scenarios involving four vessels were performed and 60 random scenarios with two vessels. In 2020 tests were carried out in real conditions on the ferries Wolin and m/f Gryf. The communication and negotiation system presented in the article has been designed and developed specially for maritime navigation purposes. The authors believe that the presented solution can be one of various solutions implemented in autonomous shipping in the near future.