水下無人載具人工智慧系統關鍵技術研發
Key Technology Development for Autonomous Underwater Vehicle with Artificial Intelligence
計畫團隊成員 Members 王朝欽教授
Prof. Chua-Chin Wang
國立中山大學電機系
EE Dept., NSYSU
周佑誠副教授
Assoc. Prof.
Yu-Cheng Chou
國立中山大學海下所
UML Dept., NSYSU
陳信宏教授
Prof. Hsin-Hung Chen
國立中山大學海下所
UML Dept., NSYSU
葉家宏特聘教授
Distinguished Prof.
Chia-Hung Yeh
國立臺灣師範大學電機系
EE Dept., NTNU&NSYSU
彭昭暐教授
Assoc. Prof.
Jau-Woei Perng
國立中山大學機電系
MEM Dept., NSYSU
沈聖智教授
Prof. Sheng-Chih Shen
國立成功大學系統系
SNAME Dept., NCKU
Category
Deep Learning, Edge Intelligence, VLSI, AUV, Industrial and Marine Application
技術亮點
Technical Highlights
由於水下載具之電池電力系統限制,開發深度學習網路的精簡優化成為必要之方向,同時需結合嵌入式硬體加速,以提昇系統運行效能,達到降低功耗混合式物件即時偵測嵌入式系統,將可實現一台低功耗、長時間運行之高效能無人水下載具。
研發專屬海洋水下之AI硬體加速電路、AI判讀技術與照明技術,將有助於魚類研究,觀察魚群生活,可用於保護及復育瀕臨絕種之魚類,也可用於幫助漁民篩選捕獲魚類,避免過度撈捕瀕危生物。
AI硬體加速電路、水下物件AI辨識技術、水下聲納感測器、水下定位技術的整合,可提升水下物件辨識的速度與精確度,配合水下地形繪製,已經實現水下物件追蹤、導航以及避障技術。
The simplification of AUV’s deep learning networks is needed due to the battery-based power system limitation. Furthermore, hardware accelerators in embedded systems not only enhance performance, but also realize low-power real-time detection. It will propel the development of a low-power and high-performance AUV.
The application of AI hardware accelerator, AI detection, and lighting of AUVs will benefit the research of underwater lifeforms. Therefore, it is echo-friendly for endangered fish. Besides, it also helps fishermen to identify un-endangered species to prevent overfishing.
With AI hardware accelerator, AI object detection, underwater sonar array, and underwater positioning integrated into the AUV, the speed and accuracy of underwater object detection is enhanced. Combined with underwater terrain mapping, the technology of underwater navigation, object tracing, and obstacle avoidance is realized.