Search This Blog

Thursday, April 30, 2015

Fetch Unveils Robot Duo to Adapt Warehouses for Quick-Delivery Era

As reported by XconomyConsumers are getting spoiled—not only do they want movies on demand via digital networks, they also want physical things like books and diapers delivered almost as fast.

Amazon, which has fueled these consumer expectations, uses colonies of Kiva robots in automated warehouses to help achieve its fast shipping times. Modest-sized retailers and distribution centers can’t afford that kind of technology, so last summer San Jose, CA-based Fetch Robotics set out to create human-scale robot workers to help those smaller companies compete in the on-demand era. While Amazon can afford to rebuild its warehouses to accommodate the extensive Kiva system, Fetch decided to make robots that would work in existing buildings with a minimum of retrofitting.

Fetch kept its evolving robots under wraps for five to six months while they were being designed and fabricated, says CEO Melonee Wise. But today the company is unveiling its first robot duo—named Fetch and Freight—to attract potential customers interested in trying a pilot project.

Fetch is an automaton about as tall as a middle-school child. Its single arm ends in a two-fingered gripper that can pick boxes off warehouse shelves and pass them to its sidekick, Freight. That robot co-worker consists of a wheeled base—similar to the one propelling Fetch around—that can be fitted with a collection bin or a set of shelves to hold the items Fetch selects. Once the order has been assembled, Freight can carry the goods to a shipping station at speeds faster than Fetch can move, because Freight’s center of gravity is lower.

Wise says Fetch is one of the few robotics companies that combine a gripper function with the mobility to travel along warehouse shelves to pluck out goods. Other companies are developing robots with similar talents, such as Billerica, MA-based Harvest Automation’s rolling bots for agricultural use. Those robots can pick up potted plants and place them on a conveyer belt, for example. Boston-based Rethink Robotics tailors robots for manufacturing chores, such as circuit testing or feeding fabric into automated industrial sewing machines.

It’s hard to make head-to-head comparisons just yet between robots like those made by Harvest, Rethink, and Fetch, which focuses on logistics. That’s the art of managing the transport of goods from factories or storehouses to consumers. But what’s clear is that a market is emerging for dexterous and mobile robots in retail, distribution, and manufacturing.

Amazon itself is funding the Amazon Picking Challenge competition next month at the ICRA 2015 conference in Seattle, where $26,000 in prizes will be awarded to inventors who create robots that can select objects from a shelf and move them to a table. Amazon hasn’t mastered automated picking yet—its boxy 330-pound Kiva robots move whole shelving units to areas where human staffers pluck out items for an order. However, this saves workers the time they would have spent searching through vast warehouse spaces.

Fetch’s robots are designed to function in smaller, human-scale environments where people are also working amid the same shelves and tables. The 250-pound Fetch can extend its reach because of its telescoping spine—it can adjust its height from about 3 feet 7 inches to roughly 4 feet 10 inches. Its vision isn’t good enough to recognize specific objects, but the depth camera in its, er, head can perceive three-dimensional shapes such as boxes and the spaces between them on a shelf, Wise says.

The Fetch robots and accompanying software could be adapted for a range of different uses beyond order fulfillment chores, Wise says—from packaging kits of multiple parts to joining parts together in a light assembly process. Fetch’s flexible arm has the equivalent of three human joints: a shoulder, elbow, and wrist.
The Fetch robot model and Fetch Robotics CEO Melonee Wise
The Fetch robot model and Fetch Robotics CEO Melonee Wise

“We think the design is pretty good,” Wise says. “The vision for these robots is that they can do a lot of things.” In February, Fetch announced it had raised $3 million in Series A financing from O’Reilly AlphaTech Ventures and Shasta Ventures. The funds are being used to bring Fetch’s first robots to market.

Although Fetch and Freight can operate autonomously to fill customer orders, humans can also interact with them via connected devices including smartphones and tablets, Wise says. The robots can also dodge around people in corridors, and follow a human co-worker by fixing its gaze on the person’s legs and tracking them continuously.

But why would a modest-sized company spend money on such technology when it can just hire more workers and give them rolling ladders?

Wise says human workers come with their own operating limits. Many distribution centers struggle to hire enough reliable employees, and suffer high turnover, she says. And unlike robots, humans are associated with what Wise judiciously calls inventory “shrinkage,” or “goods unintentionally leaving the warehouse.”

Robots could also help companies that face specific challenges, Wise says. A pharmaceutical products warehouse may need to keep close tabs on controlled drugs whose distribution is carefully regulated. It may also need to monitor the building temperature or other storage conditions that are key to drug safety and effectiveness.
Fetch’s robots can take temperature readings and snap photos of goods as they pick them up, Wise says. “You could have a full record of how that drug or that medicine was handled,” she says.

If a distributor eventually uses an all-robot staff, it can operate a warehouse at temperatures or other conditions that human beings couldn’t tolerate, Wise says.

“You can have lights-out operations,” Wise says. “That’s a big cost savings.”
Wise says it’s hard to name a price range for the Fetch robots, which would be sold as a part of a system tailored to meet a customer’s particular needs. The software and the numbers of Fetch and Freight robots would vary, and some degree of retrofitting might be required, she says.

Fetch has already been in touch with potential partners, but by showcasing its robot technology it’s hoping to get the word out to a broader range of companies that might want to set up a pilot project for an interesting new application, Wise says. After a short pilot run for a specific use, Fetch expects to be able to supply robots not only to the original partner, but also to offer them for sale to similar companies.

The Fetch robots have a modular design that could accommodate new types of sensors, collection containers, and grippers to handle new tasks or larger products, Wise says. Fetch can now pick up items weighing up to 13 pounds, and Freight can transport as much as 150 pounds. There are plenty of new challenges to tackle, Wise says. At this point, for example, Fetch couldn’t handle selecting items of clothing from a group of garments.

The company plans to continue adding new capabilities to its robot pair, which can operate as a team or separately. “It really depends on the interest we find for different capabilities,” Wise says.

If anyone could be expected to predict a rapid expansion of robot uses over the next decade or so, Wise would seem to be that person. Starting as a college student intern, she has worked on projects including autonomous boats and cars, personal robot butlers, and robots that can plug themselves in for a re-charge. But Wise says she doesn’t expect to see scads of high-complexity robots commercialized in the near future, and certainly not at low cost.
“I know how hard it is,” Wise says. Designers have already whittled down the cost of robot components, but they’ll need to make even further gains, she says. And they’ll have to overcome many snags as robots begin to operate in the messy environment of the real world.

“I think it will take a good 50 years to do the things that everyone thinks will be done 10 years from now,” Wise says.