AI-rewritten: This is a summary of an article from Ars Technica, rewritten by AI (Qwen, running locally) to make it easier to read. The facts come from the original article – read it for the full story.

---

Amazon operates more than one million robots in its warehouses globally, with plans to expand this fleet significantly. Despite this massive robotic presence, the company remains the second-largest employer worldwide, employing over 1.5 million human workers according to its 2025 annual report. Amazon executives state that these robots handle the most physically demanding and risky tasks to reduce fatigue and injury for staff. However, researchers note that unlike historical automation which removed high-risk work from sight of workers, modern collaborative robots operate in close proximity and can dictate work pace.

The U.S. Occupational Safety and Health Administration reports that warehouse injury rates are typically more than double the industry average, affecting nearly two million people with high turnover. While robots excel at repetitive tasks, they cannot yet perform all human capabilities. Amazon acquired Kiva Systems in 2012 to pioneer wheeled mobile robots that bring inventory to workers rather than requiring them to walk for items. The company now manufactures these robots and robotic arms in-house at facilities in Massachusetts, Austin, Texas, and Greenwood, Indiana.

Recent developments include the Proteus robot, which uses sensors to navigate safely around humans without following strict grid lines, and an AI system called DeepFleet that predicts robot traffic jams. Amazon plans to double its robotic arm fleet this year. The company’s Robin arm, introduced in 2022, uses machine learning trained on over 394,000 picking actions to sort packages. A trial showed the machine-learning method achieved a 93.73 percent success rate compared to 92.28 percent for traditional algorithms, resulting in significantly fewer failed picks during high-volume operations.