2019 Fiscal Year Final Research Report
R&D of Next Generation Support System for Ship Operation/Handling
| Project/Area Number |
17H03493
|
|---|
| Research Category |
Grant-in-Aid for Scientific Research (B)
|
| Allocation Type | Single-year Grants |
|---|
| Section | 一般 |
|---|
| Research Field |
Naval and maritime engineering
|
|---|
| Research Institution | Kobe University |
|---|
Principal Investigator |
|
|---|
| Co-Investigator(Kenkyū-buntansha) |
松田 秋彦 国立研究開発法人水産研究・教育機構, 水産工学研究所, グループ長 (10344334)
小野寺 直幸 国立研究開発法人日本原子力研究開発機構, システム計算科学センター, 研究職 (50614484)
|
|---|
| Project Period (FY) |
2017-04-01 – 2020-03-31
|
| Keywords | 実航海シミュレーション / 衛星AIS / 荒天中操縦性 / 大規模粒子法 / GPGPU / 自動衝突回避 / deep Q-learning |
|---|
| Outline of Final Research Achievements |
By combining AIS (Automatic Identification System) data received by artificial satellites and ocean wave prediction data, a criterion for avoiding stormy weather in route selection of oceangoing vessels was derived and a reliable voyage simulation was developed by incorporating the criterion. By conducting a captive model test, an empirical model of propeller thrust and rudder forces using air exposure ratio as a variable were obtained. Then numerical simulation of ship maneuver in stormy weather was developed based on the moving particle simulation using multi GPUs. By applying deep Q-learning, which is known as a deep reinforcement learning, an automatic collision avoidance technique was realized for ships navigating to a destination and it was validated by a free-running model experiment using multiple ships.
|
| Free Research Field |
船舶工学
|
| Academic Significance and Societal Importance of the Research Achievements |
熟練船長が行っている大洋航行時の荒天回避基準のモデル化、荒天遭遇時の操船指針の検討に耐える操縦性シミュレーション手法の構築、目的地へと向かいつつ、衝突・座礁の危険を自動的に回避する自動避航技術が開発されたことにより、次世代の船舶運航・操船支援システムの基盤的技術が確立されたといえる。特に、deep Q-learningにもとづく自動衝突回避では、世界で初めて複数の自走式模型船による実証実験を行っており、自動運航船の早期実現に向けた先駆的技術として期待できる。
|
