Preliminary validation system for development with reliable intelligent navigation algorithms of autonomous ships

Abstract

We present the validation system and solution for the development of reliable intelligent navigation algorithms such as collision avoidance algorithms and guidance, navigation, and control (GNC) algorithms. The validation system and solution are designed to demonstrate the performance of these algorithms with several ships including virtual and/or model ships that can be concurrently operated in real-time. In the validation system, the maneuvering model of the virtual ship with the dynamic characteristics that define based on the turning and zigzag test data of the model ship is applied. The model ship, which equips with a motion sensor, a control system for thruster and rudder, and a wireless network device, is controlled by the algorithms. For comparison of performance with various methodologies for collision avoidance and GNC algorithms, an modular software framework is designed and implemented. For example, in the algorithm of avoiding collisions between ships, various approaches such as a rule-based collision-free path planning compliant with COLREGs rule and an intend inference based collision avoidance can be applied under the same conditions to compare and analyze the performance of the algorithms. In order to demonstrate the practical feasibility and reliability of the proposed system and solution, preliminary tests were carried out in the ocean engineering basin of KRISO, and its results were presented.