© Getty
Experimental feature

Listen to this article

Experimental feature

The robots that roll around Amazon’s vast warehouses do not fit the stereotyped image of an automaton worker, with a humanoid appearance and clumsy movements. An army of 30,000 square orange machines — 16in tall but weighing 320lbs — negotiate the floors of the US technology company’s warehouses. They lift and carry entire shelving units of goods to the human employees, who pick and pack the items into boxes for shipping.

Since acquiring the technology in its $775m takeover of Kiva Systems in 2012, Amazon has become a leading users of warehouse robots. It has also stopped selling the machines to other retailers and logistics companies.

More robotics companies are now working on systems for moving and handling goods in warehouses. However, Andreas Koller, a robotics scientist, warns: “It’s still quite early days in terms of capabilities [and] there’s no widespread adoption yet.”

In what is still a labour-intensive industry, the potential for change is vast. A recent report by German logistics group DHL found that 80 per cent of logistics facilities are still manual because of the complexity of the operations.

Simple and repetitive processes in logistics, often using equipment in fixed positions, have long been automated. The hope now is that sophisticated mobile robotic systems that register and respond to their environment can raise productivity by taking on the heavy and mundane activities, leaving delicate tasks that need judgment and dexterity to humans.

DHL this year began a trial of two collaborative robots designed to work safely alongside humans. Made by Rethink Robotics, they have long arms with rubber grippers and perceptual recognition for det­ecting human presence; they can be guided by their flesh-and-blood counterparts.

They can open a box, take out the items, put a sticker on and repack them. “A worker could teach the robot how to pack the items together so the robot handles the repetitive part of the task, supported by a human,” says Matthias Heutger, senior vice-president of strategy, marketing and innovation at DHL.

German start-up Magazino, whose investors include Siemens, has built a robot, Toru, that can pick an object from a shelf, if it is rectangular. This is ideal for common ecommerce goods, it says, such as books and boxed items. Resembling a mobile food vending machine, Toru places the items on an internal shelf, where sensors and a 3D camera enable it to make independent decisions about what to do next, say its creators. “It’s a big change in robotics and we are applying it to logistics,” says Frederik Brantner, chief executive and founder of Magazino. “We aren’t picking a whole tray of shoeboxes, but a single [box]. We do it in parallel to humans, as robots aren’t there yet to do all the complicated things, such as pick up a small screw or fridge.”

However, warehouses with lots of moving people and objects still present technical challenges for robotics. Perhaps the biggest hurdles are localisation and mapping, says Mr Koller. “To know where you [the robot] are and where you are going in all possible circumstances is very hard,” he says. “Warehouses, because of their variety, are less structured than highways and freeways — it’s easier to build an autonomous car than a robot for a warehouse.”

Get alerts on Technology when a new story is published

Copyright The Financial Times Limited 2018. All rights reserved.

Follow the topics in this article