Redesigning and rethinking how the inside of a warehouse functions with the help of intelligent robots is no mean feat. Incorporating artificial intelligence, vision capabilities, force control, and multi-layered software into a complete, seamless computer platform that can communicate to all the robots in a warehouse at once is something of a paradigm shift. Dexterity’s founding product manager, Roger Hau, likens the evolutionary leaps Dexterity has made to the development of another type of autonomous machinery: the self-driving car.
“The simplest version of a self-driving car is a train. It goes from one station to another. That’s what some of our legacy competitors have,” Hau says. “They have a robot that does a single thing. It operates in a highly structured environment and any deviation from that environment is going to throw the train off its tracks.”
Whereas most companies have to train their robots from scratch to succeed in new functions or environments, Dexterity engineers can draw on their platform’s extensive base of capabilities to build and deploy applications for customers with unprecedented speed. As Hau puts it, “One-off solutions don't scale, and you don't build any leverage with them.” Dexterity’s AI platform and multi-layered software is flexible enough to integrate new tasks without adjustment to the hardware.
“Let’s say a large retailer has bought a robot to stack boxes on pallets,” Hau continues. “One day the retailer makes a request: ‘Can you stack boxes on my rolling cart?’ With another robotics company, the answer would be, ‘Give us a few years and another $10 million.’ At Dexterity, we’ll say, ‘Sure, we’ll just swap rolling cart with pallet as the place location.’” As a result, Dexterity can shepherd a new application from a rough concept to deployment within six months.
Functionalities that seem like unique applications all spin out from the platform. For example, robots have long been able to stack a single kind of product on a pallet — the arrangement of boxes will be the same every time. What’s much more challenging is to stack a pallet with many different products, a task that is the lifeblood of distribution centers and e-commerce fulfillment facilities. Cases of soda, cartons of yogurt, and boxes of cereal might come down the same conveyor belt, each a different color, shape, volume, and weight. Building a tall, dense, stable pallet out of that assortment is like playing 3D Tetris. A robot must gauge the current state of the pallet, assess each new package, and predict the best place to put it — all hundreds of times, with boxes that weigh up to 70 pounds. Unlike most robots on the market, Dexterity’s system excels at this task, because those tasks are all coordinated by the platform.
To date, Dexterity’s platform empowers robots to perform a variety of tasks across the warehouse including singulation / induction, palletizing / depalletizing, line kitting, and order building.
Dexterity robots are already moving millions of packages every day for customers with these core applications. The company is continually boosting the capabilities of its platform — to make the robots even faster and more nimble — and simplifying hardware to reduce costs so they can provide these benefits to customers all over the world. As Dexterity’s head of supply chain operations and quality, Avinash Verma, says, “We are just getting started.”