Efficient retail supply chain fulfillment has involved robotics for some time. Where traditional operations have bumped up against ecommerce challenges, Berkshire Grey is stepping in to move retailers into a new era.
As consumers change the way they buy products, distribution center-level problems are having ripple effects on stores. “There’s a push to make store footprints smaller, which means smaller backrooms or even no backrooms, and a push to take inventory out of the chain,” says Pete Blair, vice president of marketing at Berkshire Grey.
CEO Tom Wagner was previously chief technology officer of iRobot. His background in AI and robotics led him to realize there was opportunity in the idea of picking, placing, packing and shipping small quantities of products. “It’s hugely labor intensive, and traditional automation generally doesn’t do it well,” Blair says. “Tom kept hearing the challenge for retailers wasn’t that they couldn’t do it, but that they couldn’t find enough people to do it.”
Robots have long been used to carry out standard, repetitive tasks — a workflow that relies on a single box size and moves items in the same way to the same place every time is its bread and butter. However, ecommerce and new supply chain models call for a much more varied task list, centered around small quantities of products. It’s something traditional automation doesn’t generally do very well.
Berkshire Grey designs complete systems to address those more complex use cases. “Rather than just pick something up and put it down somewhere else with a robot, we've worked with customers to understand their process load and how they do things,” Blair says. Product flows and the way activities are staged and integrated throughout the workspace are considered, along with varied item lists that range from small and delicate to large and heavy.
Today’s supply chain requires robots to handle many products — of different sizes, weights, forms and packaging types — all from the same pile. Food items and clothing items, for example, must be treated differently. “It’s very dynamic and there needs to be a level of intelligence that traditional robotics doesn’t have the ability to handle,” Blair says.
Berkshire Grey’s platform goes beyond a robot and a gripper to incorporate computer vision. The addition of machine learning enables it to evaluate past tasks and improve them if needed: If a robot picks something without a good grab, the platform can identify the problem and resolve it in real time. “Our system will let it go, reset quickly and then repick it,” Blair says.
The real-time performance assessment results in fewer phantom picks and makes operations much more efficient. “It allows us to have the broadest capabilities of what we can pick from different types of items, shapes, packages and materials,” Blair says. Machine learning means the systems not only support today’s complex workflows well, they'll even improve performance over time.
Berkshire Grey’s primary clientele includes large enterprise retailers managing true omnichannel commerce operations. They need to get products to stores and ship ecommerce orders direct to the consumer; they're also building strategies to drop-ship individual orders to bricks-and-mortar locations. “Basically, our customers are looking at all those channels and how to do all of it better, because they know every aspect of how they do business is growing,” Blair says.
The team at Berkshire Grey expects the industry will see additional decentralization of supply chain processes and distributed networks. “Some will be micro-fulfillment centers that may or may not be attached to a store,” Blair says. The company’s AI-powered robotics technology is poised to support this new fulfillment strategy, accommodating both store replenishment and direct-to-consumer orders, and even handling home delivery as those services continue to expand.