From serving to logistics, Bear Robotics' AI robots can do it all

Bear Robotics, the brainchild of former Google engineer-turned-restaurateur developed a decentralized serving robot, and is now trying his hand at construction robots.

John Ha, founder and CEO of California-based Bear Robotics, poses for a photo at its Seoul office in Seongsu-dong in eastern Seoul on July 11. [PARK SANG-MOON]

[GAME CHANGER]

A robot that gets smarter on its own and communicates with other robots without centralized control would not be possible without sophisticated software, according to John Ha, the founder and CEO of California-based Bear Robotics.

And that — an expertise in software — was exactly what enabled the robotics company to attract 80 billion won ($57.7 million) in Series C funding from LG Electronics early this year.

"It's hard to find a company that does software and hardware well in a balanced way at the same time," Ha said in a recent interview with the Korea JoongAng Daily at the robotics firm's office in Seongsu-dong, eastern Seoul.

"In robotics, software technology like AI or cloud services that order the hardware to move in a certain way are extremely important. And I believe Bear Robotics is multiple steps ahead in that sense, which was also what was well-received by LG Electronics and led to investment."

Founded in 2017 in California by Ha and two others, Bear Robotics specializes in service robots such as its flagship Servi and Servi+. Ha first conceived the idea of developing a robot that can serve diners after running his own sundubu jjigae (spicy tofu soup) restaurant in Silicon Valley.

Before that, he was an engineer at Google.

"There's a restaurant in Paju, Gyeonggi that operates 40 of our robots for serving," Ha said.

"During the weekend, the distance these robots move around amounts to 160 kilometers (99 miles), which is equivalent to running a marathon four times. Imagine the labor being done by human. I feel very rewarded when I see my robots helping people in places they are needed."

Bear Robotics attracted 42 billion won in Series A funding in 2020, in which Japan's SoftBank as well as Korea's Lotte Accelerator and Smilegate participated. The Series B funding amounted to 100 billion won from private equity fund IMM Private Equity and KT among others.

The following interview has been edited for length and clarity.

Bear Robotics' serving robot, Servi+ [BEAR ROBOTICS]

Q. Why is software important for robots? A. It helps in terms of cost as well as upgrading our robots.

If there is a single error that cannot be fixed remotely, we have to get on an airplane, even if the solution to that error is just turning the robot off and on again. But we can't do that because we have some 13,000 robots in operation in more than 20 countries.

Through software, most of the errors can be fixed remotely. Sometimes, it is the customers that are not handling the robots right. Then we can look through the data log remotely to point out which part of the operation has been done wrong. Also, the robots can repeat the error so the developers can quickly spot the problem and fix the robot based on it.

That type of data gets accumulated, resulting in the robots' IQ going up and up. How are the functions of software-focused robots affected?

We developed the software so that serving robots can interact with each other like real human servers do in restaurants. In other cases, robots are controlled by a centralized system where orders are given top-down to multiple robots.

With the help of AI, we made our robots able to interact with each other such as, "O.K., you take care of that order and I will take care of this." The centralized system isn't suitable for restaurants because networks aren't stable sometimes, and there are a lot of unexpected factors, and the robots have to make a countless number of short trips.

So we scrapped that system to come up with a decentralized one. This kind of system is more suitable for industrial robots as well because there are worksites that ban networks completely. In that case, our robots are able to communicate with other methods. Camera sensors would be their last resort when everything else is jammed.

How does this result in cost saving?

The prices of our robots, for example, is a bit more expensive than extremely low-priced ones from China. But if you think about what's called the total cost of ownership [the sum of costs from the initial purchase to operating it throughout the product's life span], ours is much cheaper.

We can save on cost because most of the errors are fixed remotely and our team doesn't have to get dispatched. We can also make much more precise predictions about the life span of the robots' operation, which also affects cost. What was the secret behind Bear Robotics' strength in software? We had a lot of engineers who were part of big projects at software companies.

I myself am from Google but even if it was not Google, there was an excellent pool of engineers from Facebook, Apple, Samsung and LG. Back then, when we first started out as a robotics company, there weren't mainstream robot products so people didn't know how to design the software for robots. From the beginning, we had these ideas about how to operate software. After focusing on service robots for some time, you threw down the gauntlet for industrial robots. What's the difference between the two?

What came as a surprise was that not much automation is achieved in the industrial robotics sector. Automation of lines is common, but in logistics, it seemed there was a lot of work to do, so we want to focus on that area.

I had consulted with a manufacturing factory owners about making a robot and I did a tour of that factory, which surprised me because there were hardly any robots there. I asked him if I could hold a demonstration of our robots because I thought our robots could do much better.

So organically we set our foot in the segment. At the moment, we are receiving orders from Vietnam for logistics for both warehouses and factories. We will continue to do service robots as well. Is making industrial robots more complex than service robots?

Not really, because robots have much more space to move around in industrial worksites compared to restaurants, where there are a lot more people moving around. A network system is more unstable there and the floor is sometimes uneven, so advancing into this industrial robot sector was smooth after experiencing all that. How is the partnership with LG Electronics proceeding?

We are in talks at the moment and industrial robots are the area where a lot of discussions are being held. LG highly valued our robotics software as well as our global footprint.

If we are to sell and service robots globally, we need help in installation as well as maintenance. We already had an organization to do all that work, so the LG team liked that aspect as well. We are servicing robots in 45 states in United States, including after-sales service. It is rare, even for big companies, to have that kind of workforce.

So working with us will allow for various ideas to be made into products and be sold globally, which I think is a major pillar of Bear Robotics' competitiveness. Bear Robotics recently set up a Japanese office. How is the robotics industry in the country?

Competition is the fiercest in Japan to tell the truth.

So we are trying to penetrate the market by setting up a local entity and forging good partnerships like we did with SoftBank. We are doing both service and industrial robots there. With service robots, for example, although Japanese eateries are small and may not seem suitable for robots to be roaming about, Japan's dining industry lacks a workforce due to aged demographics.

And in the suburban areas, like in the outskirts of Tokyo, restaurants are not that small. We have supplied some 3,500 units there already.

BY JIN EUN-SOO [jin.eunsoo@joongang.co.kr]