内容来源：https://aibusiness.com/robotics/agility-unveils-digit-5-safety-conscious-humanoid

内容总结：

美国机器人制造商Agility Robotics近日发布新一代人形机器人Digit 5，该产品专为在工业环境中与人类安全协同作业而设计，可在重复搬运场景中举起最高50磅（约22.7公斤）的负载，并配备可更换夹爪系统，以适配不同任务需求。公司表示，Digit 5是在其前代产品Digit 4历经三年商业部署、累计运行超过6.5万小时的基础上研发而成。

当前，人形机器人正加速从实验室走向实际应用场景，人机交互与安全问题日益受到关注。Agility指出，安全仍是人形机器人实现商业化规模部署的“最大障碍之一”。Digit 5为此配备了一系列人体检测、运动意图传达及危险行为自适应系统，旨在减少对物理防护屏障的依赖，使机器人能够在更近距离内与工人协同作业。

分析人士对Digit 5在安全设计上的尝试给予肯定。Omdia分析师Lian Jye Su表示，Digit 5有望率先在安全参数明确的商业和工业场景中部署，随后再向消费和家庭场景拓展。但他同时指出，行业目前仍主要依赖各厂商自身的安全框架，对人形机器人的全面信任需要全行业标准化。“硬件已相对成熟，但人形机器人AI模型的安全与安保需要更健全的验证框架，”他表示，供应商和企业需要在审计、日志记录、测试和治理方面建立明确规范。

与传统工业机器人通常在受限环境中执行 tightly defined 任务不同，通用型人形机器人将面对更非结构化、动态变化的环境，这将引入新的安全考量因素，包括动态平衡、跌倒范围、非结构化人体运动及AI行为等。Su认为，行业需要更稳健的人形机器人安全方案，涵盖厂商认证、行业标准制定、AI模型安全测试以及有效的过滤与防护机制。

Agility正积极参与相关标准制定，包括美国国家标准学会ANSI/A3 TR R15.108技术报告中关于动态稳定工业移动机器人（含人形机器人）的标准，以及正在制定的国际人形机器人安全标准ISO 25785-1。该公司还与英伟达合作推进安全架构，成为英伟达Halos for Robotics平台的首个发布合作伙伴。Digit 5融合了Agility自身系统与英伟达IGX Thor及Halos Core基础设施。

英伟达机器人与边缘AI副总裁Deepu Talla表示：“能够在人类和资产周围安全运行，是人形机器人从试点走向大规模工业部署的关键。”Agility预计Digit 5将于2027年上半年开放早期访问，并于2027年底面向制造、仓储和分销运营商全面上市。

中文翻译：

由谷歌云赞助
选择你的首批生成式AI用例
要开始使用生成式AI，首先应关注那些能够改善人类信息体验的领域。
这家机器人制造商表示，其最新款人形机器人专为与人类协同工作而设计，但分析人士警告称，行业范围内的安全标准仍不完善。
Agility Robotics发布了Digit 5，这是其人形机器人的新版本，专为在工业环境中大规模安全地与人类协同工作而设计。
这家美国机器人制造商表示，其最新型号是在前代产品Digit 4经过三年商业部署后开发的，Digit 4在客户站点的累计运行时间超过65,000小时。
这款新机型于周二发布，可反复举起重达50磅的负载。它配备了可更换的夹爪系统，设计用于让不同的末端执行器执行各种任务，包括周转箱搬运和机器看护。
此次发布之际，人形机器人正日益从实验室走向实际部署，开发者因此更加关注人机交互和安全标准。
Agility表示，安全仍然是人形机器人商业化规模化“最大的障碍之一”，因此将Digit 5设计为解决这些顾虑的产物。新机型配备了一系列系统，用于检测人员、传达运动意图并针对潜在危险调整行为。
“安全是一个真正的瓶颈，而Digit 5是应对这一挑战最明确的工业尝试，”Informa TechTarget旗下Omdia的分析师Lian Jye Su表示。“Digit 5可能会先在具有明确安全和安保参数的界定清晰的商业和工业用例中部署，然后才会推广到更多消费级或家庭用例。”
然而，尽管Agility等公司的设计表明安全性正越来越多地从底层融入机器人设计，Su表示，该行业仍然严重依赖各供应商自身的安全框架。要让人形机器人获得完全信任，需要全行业范围的标准化。
“虽然硬件相对成熟，但人形机器人AI模型的安全与安保需要更健全的验证框架，”他说。“与AI安全一样，供应商和企业需要对人形机器人软件建立明确的审计、日志记录、测试和治理规范。安全政策和护栏也同样缺失。”
传统工业机器人通常在相对受限的环境中运行，其运动和任务有严格界定。然而，通用型人形机器人的兴起意味着这些机型预计将开始在更加非结构化、动态的环境中运行。这一变化将带来新的安全考量。
“该行业将需要更健全的人形机器人安全方法，包括动态平衡、跌倒范围、非结构化人体运动以及AI行为，”Su说。“这将涉及供应商认证、行业标准制定、AI模型安全与安保测试，以及健全的过滤器和护栏的结合。”
Agility正在为动态稳定工业移动机器人（包括人形机器人）的标准制定做出贡献，这是美国国家标准学会ANSI/A3 TR R15.108技术报告的一部分。该公司还在为ISO 25785-1做出贡献，该标准正在制定中，旨在成为人形机器人的国际安全标准。
该公司表示，Digit的安全架构——包括其人体检测系统和运动意图通信——旨在减少对物理防护屏障的需求，使机器人能够在更靠近人类工人的位置运行。
Agility还在安全架构方面与英伟达合作，成为英伟达Halos for Robotics平台的首个发布合作伙伴。Digit 5将Agility的系统与英伟达的IGX Thor和Halos Core基础设施相结合。
“在人员和资产周围安全运行，是人形机器人从试点走向大规模工业部署的关键，”英伟达机器人与边缘AI副总裁Deepu Talla在一份声明中表示。
Agility表示，预计Digit 5将于2027年上半年开始早期访问，并预计在2027年底前面向制造、仓储和配送运营商全面上市。

英文来源：

Sponsored by Google Cloud
Choosing Your First Generative AI Use Cases
To get started with generative AI, first focus on areas that can improve human experiences with information.
The robot maker says its latest humanoid is designed to work alongside people, but analysts warn that industrywide safety standards remain underdeveloped.
Agility Robotics unveiled Digit 5, a new version of its humanoid robot designed to safely work alongside people in industrial environments at scale.
The U.S. robotics maker said it developed its latest model following three years of commercial deployments of its predecessor, Digit 4, which accumulated more than 65,000 hours of operational time across customer sites.
The new model, released on Tuesday, can repeatedly lift loads of up to 50 pounds It's fitted with a swappable gripper system designed to let different end-effectors be used for various tasks, including tote handling and machine tending.
The launch comes as humanoids increasingly move from labs to real-world deployments, with developers paying more attention to human-robot interactions and safety standards as a result.
Agility, which said safety remains "one of the biggest obstacles to scaling humanoid robots" commercially, designed Digit 5 to address these concerns. The new model has a range of systems to detect people, communicate motion intent and adapt behavior to potential dangers.
"Safety is a real bottleneck, and Digit 5 is the clearest industrial attempt to address this challenge," said Lian Jye Su, analyst at Omdia, a division of Informa TechTarget. "Digit 5 will likely be deployed in well-defined commercial and industrial use cases with clear safety and security parameters before it rolls out to more consumer or home use cases."
Yet, while designs such as Agility's show increased integration of safety into robotic design from the ground up, Su said the industry still relies heavily on individual vendors' safety frameworks. Total trust in humanoids will require industrywide standardization.
"While hardware is relatively mature, humanoid robotic AI model safety and security require more robust verification frameworks," he said. "As with AI safety, vendors and enterprises need clear practices for auditing, logging, testing and governance of humanoid software. Safety policies and guardrails are also lacking."
Traditional industrial robots typically operate within relatively constrained environments, with their movements and tasks tightly defined. Yet the rise of general-purpose humanoids means these models are expected to start operating in more unstructured, dynamic environments. This change will introduce new safety considerations.
"The industry will need a more robust approach to humanoid robotics safety, including dynamic balance, fall envelopes, unstructured human motion and AI behavior," Su said. "It will involve a combination of vendor certification, industry-standard formulation, AI model safety and security testing, and robust filters and guardrails."
Agility is contributing to the development of standards for dynamically stable industrial mobile robots, including humanoids, as part of the American National Standards Institute's ANSI/A3 TR R15.108 technical report. It's also contributing to ISO 25785-1, which is being developed as an international safety standard for humanoids.
The company said Digit's safety architecture, including its human-detection systems and motion intent communication, is intended to reduce the need for protective physical barriers and enable robots to operate in closer proximity to human workers.
Agility is also partnering with Nvidia on the safety architecture, becoming the first launch partner for Nvidia's Halos for Robotics platform. Digit 5 combines Agility's systems with Nvidia's IGX Thor and Halos Core infrastructure.
"Safely operating around people and assets is what moves humanoid robots from pilots to industrial deployments at scale," Deepu Talla, Nvidia's vice president of robotics and edge AI, said in a statement.
Agility said it expects early access to Digit 5 to begin in the first half of 2027, with general availability for manufacturing, warehouse and distribution operators expected by the end of 2027.