AI Helps Warehouse Robots Pick Up New Tricks

Backed by machine learning luminaries, Covariant.ai's bots can handle jobs previously needing  a human touch. 
Interior view of conveyer belts in a factory with a yellow robot arm
Covariant's robot arms can pick up and sort items of different shapes and sizes. Photograph: Magnus Pettersson/Covariant.AI

Some of the biggest names in artificial intelligence, including two godfathers of the machine learning boom, are betting that clever algorithms are about to transform the abilities of industrial robots.

Geoffrey Hinton and Yann LeCun, who shared this year’s Turing Prize with Yoshua Bengio for their work on deep learning, are among the AI luminaries who have invested in Covariant.ai, a startup developing AI technology for warehouse bin-picking bots.

Covariant.ai has developed a platform that consists of off-the-shelf robot arms equipped with cameras, a special gripper, and plenty of computer power for figuring out how to grasp objects tossed into warehouse bins. The company, emerging from stealth Wednesday, announced the first commercial installations of its AI-equipped robots: picking boxes and bags of products for a German electronics retailer called Obeta.

Picking up everyday boxes and plastic packages might sound trivial, and it is for most humans. Workers in factories and warehouses are frequently given new objects to handle, or a batch of different items mixed together, but it’s deceptively difficult for a machine to quickly work out how to grab the next doodad. Workplace robots are still incredibly dumb and clumsy, and teaching them to grasp unfamiliar objects or those with complex shapes remains a holy grail of AI and robotics research.

In recent years, a number of companies have sprung up offering robots that use simpler algorithms to perform useful warehouse tasks, including limited product picking. Successful players include Plus One Robotics, Picknik, and RightHand Robotics.

Safer robot arms, custom grippers, off-the-shelf sensors, and open source code for robot vision and control have made it easier for startups to deploy robots in new roles, such as ferrying products around warehouses or taking boxes off pallets.

Covariant.ai has not yet developed a robot as dextrous or adaptable as a human, but it has apparently succeeded in applying an exotic research technology, called reinforcement learning, to an industrial setting. It is hard for robots to learn in the real world without making mistakes, and commercial robot installations require extreme levels of reliability.

The company was founded in 2017 by Pieter Abbeel, a prominent AI professor at UC Berkeley, and several of his students. Abbeel pioneered the application of machine learning to robotics, and he made a name for himself in academic circles in 2010 by developing a robot capable of folding laundry (albeit very slowly).

Covariant uses a range of AI techniques to teach robots how to grasp unfamiliar objects. These include reinforcement learning, in which an algorithm trains itself through trial and error, a little like the way animals learn through positive and negative feedback.

Reinforcement learning has driven spectacular recent breakthroughs in AI, including the superhuman game-playing algorithms developed by Alphabet’s AI subsidiary, DeepMind. The approach can help a robot figure out what shape an object is from a video image and where to grasp it, even if it has only been trained on objects of a different shape. This may be done in simulation so that the process can be accelerated.

But reinforcement learning is finicky and needs lots of computer power. “I used to be skeptical about reinforcement learning, but I’m not anymore,” says Hinton, a professor at the University of Toronto who also works part time at Google. Hinton says the amount of computer power needed to make reinforcement learning work has often seemed prohibitive, so it is striking to see commercial success. He says it is particularly impressive that Covariant’s system has been running in a commercial setting for a prolonged period.

Four men standing against a metal garage door

Left to right: Rocky Duan (CTO), Tianhao Zhang (research scientist), Pieter Abbeel (chief scientist), Peter Chen (CEO).

Photograph: Elena Zhukova/Covariant.AI

Besides reinforcement learning, Abbeel says his company’s robots make use of imitation learning, a way of learning by observing demonstrations of perception and grasping by another algorithm, and meta-learning, a way of refining the learning process itself. Abbeel says the system can adapt and improve when a new batch of items arrive. “It’s training on the fly,” he says. “I don’t think anybody else is doing that in the real world.”

Other big names who have invested into Covariant.ai include Jeff Dean, head of AI at Google; Fei-Fei Li, director of the Stanford Artificial Intelligence Lab; and Daniela Rus, who leads MIT’s Computer Science and Artificial Intelligence Laboratory.

Covariant.ai hasn’t disclosed all the details of its technology for competitive reasons, so it is difficult to gauge precisely how much its system relies on advanced AI.

Melonee Wise CEO of Fetch Robots, a company that sells intelligent mobile robots for warehouses, notes that you don’t necessarily need much AI to achieve a high level of reliability for a specific task. If the system is designed carefully, and the objects aren’t too varied, even a dumb system with a nifty gripper can often pick things up reliably enough. “It seems like a lot of these solutions involve a special gripper with some picking tools around that,” Wise says.

But AI is creeping into industrial automation, and it could have a profound impact if it can automate chores currently done by hand.

Keep Reading
illustration of a head
The latest on artificial intelligence, from machine learning to computer vision and more

Demand for robots is generally growing at a rapid clip, according to the International Federation of Robotics, an industry body. It says that 422,000 robots were installed in 2018, a 6 percent increase over 2017, with installations of smarter, more collaborative robots increasing 23 percent over the same period. The IFR also expects an average growth of 12 percent for all robots between 2020 and 2022.

Covariant.ai has already impressed some seasoned players in robotics. Last year, the Swiss-Swedish robotics giant ABB began looking for companies to help it move into warehouse automation. It sent Covariant.ai and other companies boxes of objects for their systems to try to pick in a controlled experiment. Marc Segura, global head of service robotics at ABB, says Abbeel’s company was the only one able to pick everything time and again.

“Every time you need to pick objects that are unknown, that’s where Covariant is good,” Segura says. He estimates that the picking market Covariant.ai is targeting could grow to be worth several billion dollars in the next few years.

Covariant.ai is also working with Knapp, a German company that installs automation systems in factories and warehouses and that helped the startup place its first system in Germany.

Peter Puchwein, vice president of innovation at Knapp, says he is particularly impressed by the way Covariant.ai’s robots can grasp even products in transparent bags, which can be difficult for cameras to perceive. “Even as a human being, if you have a box with 20 products in poly bags, it’s really hard to take just one out,” he says.

Puchwein says the system matches the performance of human pickers at the start of their shift, and of course it never gets tired. He expects Knapp to roll out dozens more installations featuring Covariant.ai’s technology in coming months. “All the customers we invite, they are very very interested,” he says.

Updated 4/27/2020 7:50 pm EST: A previous version of this story misspelled the company PickNik as Picnic.


More Great WIRED Stories