Warehouse Robots: Where Automation Is Winning, and the Jobs It Still Can't Do

November 10, 2026 · 3 min read

If the factory is where robots won the 20th century, the warehouse is where they're winning the 21st. Here's why, what's already automated, and the surprisingly human jobs that remain.

Why warehouses are perfect for robots

The reasons are structural:

Semi-structured environments: standard containers, repeatable flows and predictable layouts
Huge volumes that keep growing with e-commerce
Seasonal peaks that make hiring enough people difficult
Physically demanding work with high injury and turnover rates
Razor-thin margins, so even small efficiency gains are worth a lot

From Kiva to massive robot fleets

The modern era began with Kiva Systems, whose small orange drive units slid under shelving and carried it to workers standing still, reversing the old model where people walked the aisles to find products. Amazon bought Kiva in 2012, and the technology became the foundation of Amazon Robotics, which has since grown into one of the largest robot fleets in the world. Competitors such as Locus Robotics, Symbotic and Ocado now operate systems in retail and logistics networks around the globe.

Notice what these robots are not: humanoids. They're wheeled platforms, conveyors and robotic arms, each specialized and each highly effective. That's the strongest evidence for skeptics who say humanoid robots are unnecessary.

The three jobs that remain hard

Even the most automated warehouse still depends on people for three kinds of work.

1. Piece picking

Reaching into a bin of mixed products and pulling out one specific item, whether a bag of rice, a phone case, a shampoo bottle or a stuffed toy, is effortless for a person and historically very hard for a robot. Items vary enormously in shape, weight, texture and fragility.

Suction grippers and machine-learning vision have made big progress, and modern systems can pick a large share of a typical catalog. But the long tail of unusual items stays stubborn. A truly reliable, general-purpose picker would be enormously valuable.

2. Loading and unloading trucks

Packages arrive stacked, jumbled and sometimes crushed, inside trailers that are hot, cold and cramped. It's among the most physically punishing jobs in logistics. Purpose-built robots like Boston Dynamics' Stretch, a wheeled mobile arm designed to unload boxes, are now in commercial use.

3. Exceptions

Moving a pallet, clearing a jam, replacing a broken tote, helping a stuck machine, handling the oddly shaped item that doesn't fit the system. Warehouse operators call these exceptions, and here's the twist: exceptions increase as automation increases, because automated systems excel at the common cases and struggle with the rare ones. Exceptions are expensive, so a robot that could handle many of them would be valuable.

Where humanoids might fit

That gap is exactly what bipedal robots such as Agility Robotics' Digit target. Digit is designed to pick up totes and carry them between conveyors, shelves and workstations, in buildings where installing new conveyors would be costly or impractical. A 2023 pilot with logistics company GXO, offered as robots-as-a-service, was among the first commercial humanoid deployments.

The case for humanoids here is modest and specific: they can use existing racks, aisles and stairs, and switch between tasks without rebuilding the facility.

The case against is equally specific: for most of the same moves, a wheeled robot or conveyor is faster, cheaper and more reliable. Legged robots use more energy, occasionally fall and cost more to maintain.

For humanoids to win, their flexibility has to be worth the premium. Today's pilots exist to find out.

Beyond the warehouse

The rest of the supply chain is automating too: container ports with automated cranes and guided vehicles, autonomous trucks running driverless freight on some long-haul routes, and ongoing experiments with robots and drones for last-mile delivery.

The takeaway

Warehouses prove that robots don't need to look like people to transform an industry. Specialized machines already do the bulk of the work. The open question is whether general-purpose robots can profitably take on what's left: the picking, the trucks and the endless exceptions.

This article is adapted from Bodies for the Mind: Humanoid Robotics.

Content on this site is for general information and education only and is not professional advice. This site contains ads and affiliate links; we may earn a commission at no extra cost to you. See our Disclaimer.