The race to build robot hands that could open manual jobs to automation

Dexterity is the main thing still keeping robots out of most manual jobs. Human hands are very hard to copy, and engineers are now competing to build.

Dexterity is the main thing still keeping robots out of most manual jobs. Human hands are very hard to copy, and engineers are now competing to build robotic ones that could take on the handling tasks warehouses still give to people.

Key takeaways

  • The Guardian reports that dexterity is a key barrier stopping AI-driven machines from taking over a large share of human jobs.
  • In an Amazon warehouse near Luton, robots move shelving units around automatically, but people still pick and stack the individual items.
  • Engineers have tried for years to copy the human hand, because it combines strength, precision and a sense of touch in a way machines find hard to match.
  • If robots learn to handle varied, everyday objects reliably, the effect on manual work could be large, though no one knows when or whether that will happen.

Robotic dexterity as the bottleneck

The Guardian’s argument is that dexterity matters more than raw intelligence. AI systems have improved quickly at handling language, images and data. Physical tasks that need careful handling have proved much harder to automate. Researchers have noticed this gap for decades. Skills people find hard, such as advanced maths or chess, have been easier to automate than things almost anyone can do, like picking up a soft object without crushing it or turning a key in a lock.

A dexterous robot has to do several things at once. It has to see an object and work out its shape and position. It has to decide how to grip it. Then it has to keep adjusting that grip as the object moves, slips or turns out to be heavier, softer or more slippery than expected. Each step is hard on its own. Doing them together, fast and reliably, across thousands of different objects, is much harder still.

This is why dexterity is seen as the step that would let AI move from screens into the physical world. Software that can plan a task is of limited use in a warehouse, kitchen or care home if the machine carrying it out cannot handle the objects involved.

The Amazon warehouse near Luton

The Guardian opens its report with an Amazon warehouse just outside Luton, which it says is about the size of four football pitches. Yellow shelving units move around the floor under digital control, in what the publication describes as a choreographed routine. Containers holding very different products pass along conveyor belts: a Lego Spider-Man set, dishwasher tablets, a pouch of protein powder. When a shelving unit stops at a workstation, a human worker stacks it.

Amazon describes the arrangement as a combination of people and machines working together. The example shows where the line between the two currently falls. Machines do the moving: shelves travel to workers instead of workers walking the aisles, and conveyors carry goods between stations. The step that needs a hand is still done by a person: picking up one particular item and placing it correctly among others of different shapes and sizes.

The variety of products is the key point. A toy box, a tub of tablets and a soft pouch each need a different grip, pressure and placement. People handle that switch without thinking about it. A robot needs to be trained or designed to handle each kind of object, and to cope with products it has never seen before. The source material does not say how many people work at the site or how much of the handling is already automated.

The human hand as an engineering benchmark

The Guardian calls human hands anatomical marvels, and the reasons are well understood. A hand combines many joints moving independently, a grip that can switch between strength and delicacy, and dense sensing in the skin that tells the brain about pressure, texture, temperature and slip. The brain uses that feedback all the time, often without the person noticing, to correct the grip in fractions of a second.

Copying this is hard in two ways. The first is mechanical. Building a compact hand with many moving parts that is also durable, light and affordable is difficult. Adding more joints makes it more capable, but also makes it more complex to control and more likely to break. The second is sensing and control. Even a mechanically capable hand is of little use if it cannot feel what it is touching. Artificial touch sensors, and software that can interpret their signals quickly enough to adjust a grip, remain active areas of research.

Engineers have tried several approaches. Some designs copy human anatomy closely, aiming to match the hand finger for finger. Others drop the human form and use simpler grippers, suction cups or specialised tools that work well on a narrow range of tasks. More recently, researchers have used machine learning so that robots can learn grasping strategies from large amounts of practice, in simulation or the real world, instead of having every movement programmed by hand. The source material does not name specific projects or report how far any of them have got.

Dexterity and the future of manual work

The Guardian’s headline uses the phrase “the doom of humanity” and asks whether robotic hands could make people obsolete. That framing reflects a wider worry. If dexterity is what still protects many manual jobs from automation, solving it could put a much larger share of work within reach of machines.

Jobs that depend on handling varied objects are spread across the economy, from warehouses and factories to retail, food preparation, cleaning and care. Many of these have so far been only partly automated, with machines handling repetitive or predictable steps and people handling the rest. A dependable, general-purpose robotic hand could change that balance.

The scale and timing of any such change are not known. Showing that a robot can do a task in a lab is very different from deploying robots that do it reliably, safely and cheaply, all day, in a real workplace. Cost, maintenance, safety rules and how well machines cope with unexpected situations will all affect how fast change happens, if it happens. The source material gives no forecasts for job losses or timelines, and none are given here.

What the race for robotic hands tells us

Together, these points show that automation is advancing unevenly. Thinking and moving goods around have been automated far faster than the delicate physical handling that most manual jobs depend on. The Luton warehouse shows the current compromise: machines bring the work to people, and people supply the hands.

That makes dexterity a useful measure for anyone tracking how AI will affect work. Better software reasoning alone will not decide how far automation reaches into manual jobs. The bigger question is whether engineers can give machines the touch, adaptability and reliability that human hands provide naturally. Until they do, many jobs are likely to stay a mix of people and machines rather than being handed fully to robots. If they succeed, the debate about work and automation is likely to move much faster than it has so far.

Sources and further reading

  • The Guardian: technology reporting on robotic hands, dexterity and the future of work, including a visit to an Amazon warehouse near Luton.
  • Academic robotics research: published work on robotic grasping, manipulation and tactile sensing.
  • Labour economics literature: studies on automation and how exposed different kinds of manual work are to it.
  • Industry material on warehouse automation: public descriptions of robotic fulfilment systems from logistics operators.

Surfaced from the rss:guardian_tech signal “robotic dexterity and jobs”. AI-assisted draft, editorially reviewed.

Visited 2 times, 2 visit(s) today
share this recipe:
Facebook
X
WhatsApp
Telegram
Email
Reddit