Quantcast
Meet the Robot That Will Be Bagging Your Groceries

Ocado’s new soft robotic hand can pick up a wide variety of fruits and vegetables without damaging them.

These days, robots are able to perform increasingly sophisticated tasks like shooting hoops, but the most cutting edge robotics focus on applications for unstructured environments. This is because life, for the most part, does not operate like an assembly line and robots need to be versatile enough to handle life's randomness. Yet even when life does resemble an assembly line, like at a grocery store checkout, robots would still have a hard time performing a seemingly simple task like bagging groceries due to the sheer variety of items available in a typical grocery store.

Enter the new robot from the online grocer Ocado, which employs a soft robotic hand that is capable of picking up a wide variety of fruits and vegetables without damaging them.

In conjunction with the European Union's Soft Manipulation (SoMa) project, which is exploring industrial applications for soft robotics, Ocado has spent the last year and a half developing the robotic arm that can pick up everything from apples to a bag of limes without damaging them.

The design that was eventually settled on for the robotic hand is called the RBO Hand 2, which uses flexible rubber air chambers for the fingers and pressurized air to change the robot's grasp based on the object it is handling. The only variable that is manipulated is the air pressure—the fingers adjust themselves to the shape of the object at hand.

The Ocado team is designing the robot for future use in its already heavily automated grocery warehouses, but it still has a bit more development to do before the robot is ready for deployment in the warehouse or your local Whole Foods.

To demonstrate the hand, the researchers used it to pick up individual fruits or collections of fruits (like a bag of limes) from a flat surface. Yet Ocado's warehouses stock about 48,000 different products, and these items are rarely sitting by themselves. This means the robot will have to learn how to pick up individual objects out of a crate filled with similar objects that will shift around as the robot picks through the crate. According to The Guardian, Ocado is figuring out how to incorporate machine learning and computer vision into its robotic system. Once it's got this down, the company hopes to begin testing the robot in more complicated scenarios in the coming months.

Get six of our favorite Motherboard stories every day by signing up for our newsletter.