Amazon has more than a million warehouse robots and is expanding its fleet of robotic arms, including systems designed to handle a huge variety of products. Its newer machines use AI to make picking and sorting decisions, bringing a practical question into sharper focus: when robots work alongside people, who sets the pace?
Watch Desk analysis
What happened
Ars Technica’s account of Amazon’s robotics programme describes a fleet built from mobile robots and increasingly capable arms. Amazon says its DeepFleet AI models help predict congestion among its warehouse robots. Its robotic arms use machine learning and foundation models to identify items, select grasps and handle products that have not been seen before.
The systems vary in what they can do. Robin sorts packages, Cardinal packs them into carts, and Sparrow handles a wider range of products. Vulcan combines vision with force feedback, giving it a form of touch to help place or retrieve items. Amazon told Ars that it expects to scale up Vulcan’s use in 2027.
A 2023 Amazon Robotics study cited by Ars found a 93.73% picking success rate for its best machine-learning method, compared with 92.28% for its best traditional algorithm. Across more than a million picks, Ars says, that difference meant about 16,500 fewer failed attempts. The company’s executives say automation takes on difficult tasks and can reduce physical strain. Researchers and labour advocates raise a different concern: collaborative robots work close to people and may dictate the speed of the job.
Why it matters
The question is no longer simply whether robots can move stock or pick a parcel. AI is helping them deal with changing products and less predictable tasks, while sensors let some machines operate in closer proximity to workers. That can change which tasks people do, how quickly they must do them, and how much control they have over the flow of work.
Automation may remove some strenuous jobs without removing the pressures of warehouse work. As George Mason University professor Brian Greenwood told Ars, collaborative robots change the dynamic because they are physically close to workers and can influence the pace. Both claims deserve attention: the promised relief from tough tasks, and the risk that the remaining work gets organised around machine speed.
Our read
Amazon’s systems show genuine technical progress: machine learning improved picking success in the company’s reported trials, and newer arms are being designed to cope with a broader range of objects. But a smoother warehouse is not automatically a better workplace. The measure worth watching is not just how many parcels the robots move, but what happens to injury risk, workload and worker control as they spread.
What to watch
- Whether Amazon expands its robotic-arm fleet as planned.
- How widely Vulcan is deployed when Amazon expects to scale it in 2027.
- Whether evidence on injury rates and work pace supports the promised benefits for employees.
Discussion spark: If warehouse robots take on strenuous tasks but also help set the pace, what should count as proof that automation has made the job better for workers?
Sources and evidence
- Inside Amazon’s robotic arms race and what it means for the future of human work – Ars Technica (12 October 2026, 12:43 UTC)
Watch Desk is operated by WittyWires as an independent cross-cutting AI news tracker. It does not speak for the organisations or people it covers.