Amazon 升級 Proteus 機器人，支援自然語言互動

🔑 增強重點摘要

•The natural language interaction for Proteus is powered by an "agentic AI" project within Amazon's Lab126, which is developing an AI foundation model framework to enable robots to reason about commands and act autonomously.
•The upgraded Proteus is designed for expanded operational flexibility, moving beyond the original model's confinement to dock areas to operate anywhere across a fulfillment or delivery site, transporting various containers.
•The development of Proteus's autonomy, including its precision navigation and safety features (SIL 2 certification), was supported by embedded processors from Texas Instruments and involved training AI models with synthetic data generated using NVIDIA Isaac Sim.
•The initial Proteus, unveiled in June 2022, was Amazon's first fully autonomous mobile robot (AMR) capable of working safely alongside human employees, a significant evolution from its earlier Kiva-based Automated Guided Vehicles (AGVs) that operated in restricted areas.

🛠️ 技術深入

AI Foundation Model: The natural language understanding is built upon an AI foundation model framework developed by Amazon's agentic AI team within Lab126.
Sensors: Proteus utilizes multiple sensors, including cameras, LiDAR, and ultrasonic sensors, to power its autonomy software systems.
Navigation: It employs Simultaneous Localization and Mapping (SLAM) technology for navigation and high-precision LiDAR for finding carts.
Precision Alignment: For millimeter-level positioning and alignment, Proteus uses custom-made ground targets called "fiducial plus."
Safety Certification: Proteus has achieved Safety Integrity Level (SIL) 2 safety certification, enabling it to operate safely alongside humans.
Hardware Components: Texas Instruments' embedded processors, power, and connectivity solutions contribute to Proteus's safety bubbles and other safety-enabling technologies.
AI Model Training: Amazon Robotics leveraged NVIDIA Isaac Sim, built on Omniverse, to generate large photoreal synthetic datasets, which improved the robot's marker detection success rate from 88.6% to 98% and accelerated development times.

🔮 前景展望基於引用來源的 AI 分析

The natural language interface will significantly lower the barrier to entry for workers interacting with complex robotics.

By eliminating specialized software, more employees can intuitively assign tasks, increasing efficiency and adoption across Amazon's diverse workforce.

Amazon will accelerate the deployment of advanced AI-powered robotics across its global fulfillment network.

The substantial €10 billion investment in European fulfillment centers and the focus on agentic AI indicate a strategic push for widespread integration of intelligent automation.

The expanded operational scope of Proteus will lead to a more integrated and flexible warehouse automation ecosystem.

The ability of the new Proteus to operate anywhere in a facility, rather than being confined to specific zones, suggests a shift towards seamless, end-to-end robotic assistance in logistics.

⏳ 時間線

2003

Kiva Systems, the precursor to Amazon Robotics, was founded.

2012-03

Amazon acquired Kiva Systems for $775 million, subsequently rebranding it as Amazon Robotics.

2019

Amazon acquired Canvas Robotics, a company specializing in 'natural feature' navigation (AMR technology).

2022-06

Amazon unveiled Proteus, its first fully autonomous mobile robot (AMR) designed to work safely alongside human employees.

2022-10

Proteus was launched at Amazon's Boston robotics facility and first deployed in a fulfillment center in Nashville.

2025-06

Amazon announced its agentic AI team was developing an AI foundation model framework to enable robots like Proteus to understand natural language commands.

2026-06-04

Amazon unveiled the upgraded Proteus robot with natural language interaction capabilities, as part of a significant investment in its European fulfillment network.