| 英文摘要 |
This article introduces the Smart Ship Navigation Safety Verification Platform (SSNSVP), a tool for evaluating decision-making systems of autonomous ships. Unlike conventional trials using real ships, the SSNSVP applies a hard-ware-in-the-loop (HIL) framework. The decision computer of the test ship connects to the SSNSVP and conducts tests with virtual ships. These ships are generated by the Ship Maneuvering Simulation system (SMS) developed by the Ship and Ocean Industries R&D Center (SOIC). By receiving virtual ship state data, the decision computer can perform ma-neuvering operations in a virtual environment. This approach reduces collision risks and lowers the cost of sea trials. The control structure and communication protocols remain consistent with real ship encounters, ensuring realistic test condi-tions. The SSNSVP provides broad configurability, including two sets of scenario design parameters, three navigation modes, six major domestic ports, and thirty-four ship types. These options cover a wide range of encounter situations. Using real-time position, speed, and heading data of the test vessel, the platform generates virtual ships that meet at specified times or distances, avoiding failures caused by early or delayed encounters. Finally, the paper reports experi-mental results of the THETIS Smart Ship, an autonomous vessel developed by our team, demonstrating the effectiveness of the SSNSVP for systematic testing of navigation safety. |