An online grocery firm is testing a robotic hand that picks up fruit and vegetables without bruising or squashing them. The next step is for the hand to be able to detect the fruit’s ripeness.
British-based Ocado is working alongside five universities and Disney on a project called SOMA, short for soft manipulation. Among its work, the project has developed a robotic hand that closely mirrors the structure of a human hand, using air pressure to move the individual fingers.
While Ocado already uses robots in its warehouses, they retrieve entire crates, from which humans pick out the individual items that go into an order. With existing models, it’s tough to get the right amount of robotic grip to be sure of holding on to a heavy and potentially slippery object, but without so much pressure that delicate items could be bruised or otherwise damaged.
The next step is to combine cameras (to identify individual items) with machine learning to the point that the robot can apply just enough pressure to know how ripe a particular item is. That’s particularly important for online orders where the customer has to trust the delivery firm to send them suitable produce. For example, some online grocery chains try to package together bananas with a range of ripeness so that the customer can enjoy them over several days, rather than sending a large bunch that ripens at the same rate.