据Daniele Lepido和Marilen Martin报道，德国公司Agile Robots正通过拍摄人类执行任务的过程来训练机器人，旨在解决制约工业自动化发展的数据短缺问题。该公司首席执行官这样解释这一做法。
Agile Robots is training its machines by filming humans performing tasks, in a bid to compensate for a shortfall of data that is holding back industrial automation, according to its chief executive officer.
该公司的“机器人学院”项目招募员工亲自拍摄自己完成特定任务的视频，记录下手部与手指的动作细节。随后，Agile Robots将这些视频数据与其他训练资料相结合，让机器人在模拟环境中反复练习这些动作。本周四，首席执行官陈兆鹏在意大利都灵举办的Wave by Vento科技大会上介绍了这一情况。“人类能在学校里学习，为什么机器人就不行呢？”他这样说道。
Its Robot Academy comprises workers who film themselves performing certain tasks, capturing the way their hands and fingers move. Agile Robots combines that information with additional training data, and robots practice the movements in simulated environments, Chief Executive Officer Zhaopeng Chen said Thursday at the Wave by Vento technology event in Turin, Italy. “If human beings can learn from school, why can’t robots?” he said.
近年来，欧洲及其他地区涌现出大量初创企业，致力于将人形机器人及其他类型机器人引入工厂，以提升生产效率并降低成本。工厂环境往往包含大量重复性工作，这些工作比其他现实场景更易于实现自动化。然而，由于机器人缺乏足够的训练数据，难以完成基础任务，导致其应用进程受阻。
A wave of startups has emerged in Europe and elsewhere seeking to put humanoids and other robots into factories to boost productivity and cut costs. These environments often involve repetitious tasks that lend themselves to automation more quickly than other real-world settings. But adoption has been hampered due to the machines struggling with basic tasks, in part due to a lack of training data.
另一家德国人形机器人制造商Neura Robotics也宣布，正与慕尼黑工业大学合作开发专门的机器人训练设施。
Neura Robotics, another German humanoid-maker, has said it’s developing a robotic training facility with the Technical University of Munich.
陈兆鹏指出，机器人往往需要经过数千次尝试才能熟练掌握某项工作；即便只是出现细微变动，它们也可能需要重新学习。相比之下，技术娴熟的工人则能凭借毕生积累的经验轻松应对各类任务。
Robots can require thousands of attempts to master a job and may need to learn it again after even small changes, Chen said. Skilled workers draw on experience acquired throughout their lives.
据悉，Agile Robots由德国航空航天中心的研究人员于2018年在慕尼黑创立，目前已在全球部署了超过2万套系统。陈兆鹏上个月向媒体透露，该公司至今已累计获得15亿美元的投资。虽然他未透露后续融资计划，但表示新一轮融资“很快”就会到来。据彭博社今年6月报道，该公司正与软银集团洽谈，计划再筹集8亿美元资金。
Agile Robots, formed in Munich in 2018 by researchers from the German Aerospace Center, has deployed more than 20,000 systems, according to the company. The company has raised $1.5 billion in financing to date, Chen told journalists last month. He declined to comment on future funding but said more is coming “very soon.”The company is in discussions to raise a further $800 million with involvement from SoftBank Group Corp., Bloomberg News reported in June.