Futuristic Robots Are Lending Their Hands in Gap’s Warehouse

Kindred, a robotics startup whose investors include Alphabet executive chairman Eric Schmidt’s venture capital fund, has signed up Gap Inc. (GPS) as a trial customer in one of its Tennessee warehouses. It’s a big win for the three-year-old Kindred, which also said Tuesday that it had scored $28 million in funding led by Chinese tech giant Tencent, bringing its total funding to $44 million.

Kindred’s new customer and investment highlights the growing trend of companies like Amazon (AMZN) and Target (TGT) using robots to move warehouse inventory and track items on store shelves. Several robot analysts cite Amazon’s $775 million purchase of robot maker Kiva Systems in 2012 as a turning point for retailers and other companies, not just automakers or industrial manufactures, in adopting robots.

Kindred’s robot, which sorts items that are shuttled to it by a conveyor belt, looks like one of those old arcade games that involve the use of a claw attached to a crane to pick up plastic trinkets. The robot’s arm can pick up packages containing things like t-shirts, pants, and sunglasses that are dropped off inside its glass-paneled enclosure, swivel, and then place those items in their designated drop-off zones that resemble little mailboxes.

Get Data Sheet, Fortune’s technology newsletter.

Kindred co-founder and product chief George Babu said Gap has been testing the robot for a couple of months and that it’s now “processing real orders” by helping humans separate items like jeans and t-shirts for delivery. He declined to say how many robots Gap is testing in the warehouse, only to say “there are a few and there will be more.”

Currently, Kindred’s staff in Canada (the company has offices in Vancouver, Toronto, and San Francisco) work remotely to help Gap train the robots’ software for the automated work in Tennessee. Kindred developed the underlying software that powers the the robotic arms, built by the Japanese robotics company Fanuc, Babu said. If there’s any problems on the assembly line, Kindred staff can notify the appropriate warehouse operator, Babu explained.

Using custom software, Kindred’s workers manipulate the robots like they’re playing a video game. Each time a Kindred employee moves the robotic arm to pick up a particular item, the software records that data so that the robot can eventually learn to operate without human control.

Kindred

Unlike the robots of yesteryear, Kindred’s robots are designed to improve over time using a subset of artificial intelligence called deep learning, Babu said. While companies like Google (GOOG) and Facebook (FB) use deep learning to help sift through data and do tasks like automatically recognizing images like cats in photos, Kindred’s robots use deep learning to help them remember “what is a good grasp,” said Babu.

For example, the robotic arm, is supposed to automatically grab a pair of sunglasses more gently than jeans.

The robot also uses an AI technology called reinforcement learning that helps it figure out the best way to pick up and move objects to their drop-off spots. It’s a technique that’s currently used by Google to establish the best settings in its data centers to conserve energy, as Jeff Dean, Google’s head of its Brain research team and a Kindred investor, previously told Fortune.

Kindred has assembled some of the leading figures in modern-day AI as advisors, including Yoshua Bengio who also advises Microsoft (MSFT) on AI technology, University of Alberta professor and reinforcement learning expert Richard Sutton, and Apple (AAPL) AI chief and Carnegie Mellon University professor Ruslan Salakhutdinov. University of Guelph professor and machine learning expert Graham Taylor, also a Kindred co-founder who no longer works for the company, is also an advisor, Babu said.

Kindred said that it is testing its robots with other retailers in addition to Gap, but it declined to name them.

Babu is aware that some people may worry that companies will eventually use robots to replace human workers. Despite that perception, he said there’s actually a “big shortage” of warehouse workers and that the lack of manpower is keeping retailers from growing as quickly as they want to.

“I believe that we as a country have to start getting away from how [robotics are] going to take away jobs,” Babu said. Companies and U.S. policy makers should instead focus on “how to retrain people” so that they can use the latest technology and work alongside robots.