SEOUL—People don’t want robots just to work for them. They want robots to talk and play with them, too.
Look at the recent Winter Olympics in PyeongChang, South Korea, for an example of this work-and-play dynamic. More than 30 robots from a startup called FutureRobot were on hand at the games to answer visitors’ questions in Korean, Chinese, English, or Japanese. With a friendly avatar face and a large touchscreen on a cylindrical body, the robots offered directions to different sporting venues, provided schedules for every event, and more.
Visitors who approached the robots for that practical information almost universally wanted to have some fun with them as well—pose for photos, play music, or follow them around.
For the team at FutureRobot, based in Pangyo, South Korea, that playful engagement was a bit of a surprise.
“I’m the CTO, so I was focused on the technology, like how do I make sure it provides the best answer,” says Youngki Hwang, the company’s chief technology officer. “But we saw that a lot of people just wanted to have fun with the robot. The technology is important, but so is the enjoyment of it.”
Smart and Social Robots
This Olympics lesson comes as good news for FutureRobot, because improving “human-robot interaction” is at the heart of why founder and CEO Kevin (Sekyong) Song started the company nine years ago. FutureRobot is working on what Song calls “social AI”—artificial intelligence that will give robots the ability to generate interactive, multi-modal behaviors and respond to emotions conveyed by facial expressions, words, and tone of voice, and then respond properly.
“Coming soon is the smart, personal, and social robot, like a soul mate,” Song says. “No one has provided that. That kind of robot has to live in human society, and that’s very difficult.”
Song sees a growing market for what he calls “service robots” in airports, retail stores, museums, government offices, and big sporting events and conventions. The PyeongChang 2018 Winter Olympic Games offered FutureRobot’s largest-scale test of its service robot, called FURo-D. And soon, Song predicts, people will also want such robots in their homes.
To deliver on this vision, FutureRobot is relying on cloud-based compute power from Oracle to solve three major challenges:
1. FutureRobot must monitor whether its robots deployed worldwide are working properly. About 200 are already in action as service robots, many in proof of concept tests.
The company uses Oracle Management Cloud to collect operating data on each robot, such as location, sensors, CPU and memory use, battery level, and network status, to alert customers to any problems. It also uses Oracle Customer Experience Cloud to manage all of the content a robot needs to answer questions.
2. FutureRobot must collect and analyze massive amounts of human-robot interaction data in order to train its robots to respond appropriately, refining that social AI Song envisions. The company uses Oracle Management Cloud to collect that data as well, noting factors such as the number of people approaching a robot and interacting with it, users’ emotional data based on facial recognition technology, the conversations people have based on voice recognition technology, and statistics on the content people access. As it gathers more of that data, FutureRobot intends to use Oracle Big Data Cloud for the machine learning and analytics needed to constantly train its robots.
3. Social robots will need to use blockchain to secure the data coming in and out, especially for personal use, so that people will trust that no one is spying on them, Song says. FutureRobot is partnering with Oracle to develop that cloud-based blockchain capability.
“FutureRobot by itself can’t do all the things that these cloud services can globally,” Song says. “That’s why we need Oracle Cloud.”
Growing Market
Airports offer the first big opportunity for FutureRobot, offering travelers something like a roving information booth staffer, but one that speaks multiple languages and always has the most current information.
The San Jose, California, and San Antonio, Texas, airports are using the FURo-D model. With its avatar face on a screen, FURo-D is friendly, but it doesn’t really look human—and that’s a good thing, it turns out.
In a recent paper that compared robots faces, FURo-D was rated among the highest for friendliness, trustworthiness, and intelligence, and tops for “appealing human-likeness.” Robots that take on a too-human persona tend to creep people out, in what is known as the “uncanny valley” effect. When survey takers were asked which robot is best suited for service jobs based only on its face, FURo-D was among the two they picked most often.
The next major market for service robots is in retail stores, where they will answer questions, offer promotions, and provide checkout services. “We’re developing for the unmanned store,” Song says.
A desktop version FutureRobot has in prototype, FURo-Desk, aimed at businesses such as small retailers, coffee shops, and restaurants will offer the friendly face and social AI capability mainly for checkout purposes. Other potential markets include hotels, museums, libraries, and government offices. FutureRobot also may lease its FURo-Ds for short-term duty at big multiday events such as auto shows and tech conventions.
More AI, Blockchain Ahead
Even though it’s still an emerging market, the robot sector is a crowded one. On Song’s office wall is a poster of the “communication robot” market from 2016, and it shows 18 moving robots and another 25 desktop versions. In PyeongChang this winter, there were about 50 other robots around the Olympic Games, though FutureRobot had more than any other provider.
FutureRobot’s success depends on its ability to differentiate its robots through its unique middleware, called Furoware, which runs on the industry standard Robot OS. It’s that middleware, combined with the smart social AI algorithms, trained through machine learning on massive data sets, that will deliver an ever-richer human-robot interaction. In parallel to refining Furoware, the company is developing its next-generation software, called Soulware, to bring even richer human interactions and even a sense of robot ethics, based on social AI.
FutureRobot is located in Pangyo Techno Valley, a bustling center of tech innovation in Pangyo that was built from the ground up over the past decade. Throughout the city and the country, tech experiments are everywhere—including a test of upcoming 5G high-speed wireless networks earlier this year, also at the Olympics. South Korea expects 5G rollouts to begin as soon as next year, with nationwide use by 2022.
Even in tech-friendly Korea, though, home-based robots haven’t broken through yet. Song is convinced they’re coming soon, to entertain and inform, and to take on tasks like reminding people to take their medications or letting them connect more easily to their elderly parents via video chats.
A robot in the home, though, requires a supreme level of consumer trust, given the fact that “a robot can see, sense, listen, and collect personal data,” Song says. It’s why he sees a critical role for blockchain, the distributed, transparent digital ledger technology.
There’s more to blockchain than protecting data; it can also generate additional revenue. As robots move into the home, Song expects FutureRobot to offer third-party education and other services, such as music, movies, and games, through them, using blockchain to verify third-party providers before giving them access to the machines.
FutureRobot is planning to look to Oracle for cloud-based integration and security help. As robots take on customer-facing retail roles, for instance, retailers will want to do more than just give directions to the toy aisle. They’ll want to offer promotions, point people to merchandise based on current inventory, or even consider a specific customer’s preferences before responding. Doing all of that will require integrating robots with enterprise marketing and other systems.
Sure, people today can get quick answers to their questions via a voice-activated device or an app on their phone. But Song expects those interactions will only increase people’s desire for a richer experience. “Humans want the emotional understanding,” he says, “so we need much more—watching and listening and gesturing and feeling.” Song plans to have FutureRobot ready to provide that connection.
Chris Murphy is Oracle director of cloud content.
