Practice makes perfect

Today’s piece-picking robots come in a variety of types, using an array of different techniques and gripping technologies to select items. But one thing these robots have in common is the need to practice their moves over and over again, refining their techniques through artificial intelligence (AI) and sheer repetition.

A couple years back, the engineers at logistics technology startup Berkshire Grey devised a unique approach to training their robots. Since it would need to run a large volume of inventory through its conveyor and shuttle picking system for store-replenishment applications, the Bedford, Massachusetts-based firm turned to an organization with plenty of groceries to sort—a local food bank. It was a perfect match: Any charity that receives donated items has to count and sort all those boxes, cans, and bags—precisely the types of goods the company’s robots needed to “learn” to handle. 

In 2020, Berkshire Grey launched its “Picking With Purpose” program, and today the company supports regional charities like the Greater Boston Food Bank, City Harvest, and United Way. During peak donation seasons like Thanksgiving and Christmas, those groups bring bulk goods to the tech firm’s robotics lab, where advanced machinery quickly sorts inventory into ready-for-distribution boxes containing an assortment of food items.

The program quickly became popular with food relief organizations, and then became a crucial tool during the pandemic, when in-person volunteer rates dropped by 50% while food banks saw a 60% increase in demand. Thanks to its program, the company’s robots filled the gap and were able to rapidly package food staples into ready-to-go Thanksgiving meal kits.