2026-05-01 10:02:42 UTC / 路透社
作者：迈克·斯通
2026年5月1日 美国东部时间上午10:02 更新于54分钟前
节点运行失败
2008年9月28日航拍的美国军方总部五角大楼。路透社/杰森·里德/档案照片 购买授权，打开新标签页
- 内容摘要
- 公司动态
- 达美数据实验室获1亿美元合同以强化海军AI水雷探测能力
- AI软件可更快适配新型水雷，将更新时间从数月缩短至数天
- 达美首席运营官托马斯·鲁滨逊称该技术可在争议水域快速部署
华盛顿5月1日路透电 — 一份最新获批的合同显示，美国海军正在升级人工智能能力，以在全球最重要的航运要道之一霍尔木兹海峡搜寻伊朗水雷。
唐纳德·特朗普总统曾表示，美国海军正在该海峡清除伊朗水雷。这条航线是石油运输的关键通道，其受阻正日益威胁全球经济。尽管美伊之间持续数周的冲突达成了脆弱停火，但搜寻水下爆炸物的工作仍可能耗时数月。
《路透伊朗简报》新闻简报将为您带来伊朗冲突的最新动态与分析。点击此处订阅。
广告 · 滚动继续
这家位于旧金山的人工智能公司达美数据实验室获得的最高金额达1亿美元的合同，将通过一款软件加速这一进程，该软件可训练水下无人机在数天内识别新型水雷。
“反水雷任务过去是舰艇的工作，”达美首席运营官托马斯·鲁滨逊在接受路透社采访时表示，“如今正成为人工智能的任务。海军正在为该平台买单，使其能够按照争议水域所需的速度训练、管理并部署人工智能系统，这些水域会阻断全球贸易并威胁水手安全。”
上周，美国海军授予这份最高金额达9970万美元的合同，以扩大达美作为海军“AMMO项目”人工智能支柱的角色——“海上作战加速机器学习”项目旨在让水下水雷探测工作更快、更精准，同时减少对水兵的依赖。
广告 · 滚动继续
该软件整合了多种传感器的数据，包括侧扫声呐和视觉成像系统，并可让海军监控各类AI探测模型在实战中的表现，识别故障并推送修正方案以提升性能。
达美提案的核心——也是海军的赌注——在于速度。在该公司介入之前，升级为海军无人水下航行器（UUV）提供动力的AI模型，以识别新型或此前未遇见过的水雷，可能需要长达六个月的时间。达美称已将该周期缩短至数天。
鲁滨逊以中东危机为例说明该技术的意义：“如果有在波罗的海受训以应对俄罗斯水雷的无人水下航行器，随后需要部署到霍尔木兹海峡探测伊朗水雷，借助达美的技术，海军可在一周内完成准备，而非一年。”
海军发言人未能立即置评。
迈克·斯通 华盛顿报道；汤姆·霍格 编辑
我们的标准：路透社信托原则，打开新标签页
US Navy turns to AI firm Domino for options to counter Iranian mines
2026-05-01 10:02:42 UTC / Reuters
By Mike Stone
May 1, 2026 10:02 AM UTC Updated 54 mins ago
节点运行失败
Aerial view of the United States military headquarters, the Pentagon, September 28, 2008. REUTERS/Jason Reed/File Photo Purchase Licensing Rights, opens new tab
- Summary
- Companies
- Domino Data Lab awarded $100 million contract to boost Navy AI mine detection
- AI software enables faster adaptation to new mine types, reducing update time from months to days
- Domino COO Thomas Robinson says technology allows rapid deployment in contested waters
WASHINGTON, May 1 (Reuters) – The U.S. Navy is ramping up its AI capabilities to hunt for Iranian mines in the Strait of Hormuz, one of the world’s most critical shipping lanes, a recently awarded contract shows.
President Donald Trump has said the U.S. Navy is clearing Iranian mines from the strait, a vital sea route for oil shipments, whose disruption is increasingly threatening the global economy. Sweeping for underwater explosives could take months despite a tenuous ceasefire between the U.S. and Iran in their weeks-long war.
The Reuters Iran Briefing newsletter keeps you informed with the latest developments and analysis of the Iran war. Sign up here.
Advertisement · Scroll to continue
The up to $100 million contract for the San Francisco artificial intelligence company Domino Data Lab could quicken this process with software that can teach underwater drones to identify new types of mines in a matter of days.
“Mine-hunting used to be a job for ships,” Thomas Robinson, Domino’s chief operating officer, said in an interview with Reuters. “It’s becoming a job for AI. The Navy is paying for the platform that lets it train, govern, and field that AI at a speed required for contested waters that block global trade and imperil sailors.”
Last week, the U.S. Navy awarded the up to $99.7 million contract to expand Domino’s role as the AI backbone of the Navy’s Project AMMO – Accelerated Machine Learning for Maritime Operations – a program to make underwater mine detection faster, more accurate, and less dependent on human sailors.
Advertisement · Scroll to continue
The software integrates data from multiple sensor types, including side-scan sonar and visual imaging systems, and allows the Navy to monitor how well various AI detection models are performing in the field, identify failures, and push corrections to improve performance.
The core of Domino’s pitch – and the Navy’s wager – is speed. Before the company’s involvement, updating the AI models that power the Navy’s unmanned underwater vehicles (UUVs) to recognize new or previously unseen mines could take up to six months. Domino says it has cut that cycle to days.
Robinson illustrated the relevance to the Middle East crisis: “If there were UUVs in the Baltic Sea trained on Russian mines, and then they needed to be deployed to the Strait of Hormuz to detect Iranian mines, with Domino’s technology, the Navy could be ready in a week rather than a year.”
A Navy spokesman was not immediately able to provide comment.
Reporting by Mike Stone in Washington; Editing by Tom Hogue
Our Standards: The Thomson Reuters Trust Principles., opens new tab
发表回复