At Chinese unicorn Ninebot Inc, VP of Robotics Li Pu compares building the Loomo personal robot to that of Apple's first personal computer, with the vision of introducing a robot helper or companion into everyday life. By democratizing the underlying requirements for robotic function (e.g. computer vision, voice recognition, mobility, etc.), Li's robotic platform serves as base infrastructure on which incredible applications and uses cases can be developed. A massive undertaking indeed, but I wouldn't put it past Li and his team to help accomplish this. After all, Li's academic background is incredible. He received a bachelor's degree in engineering from the Beijing University of Aeronautics and Astronautics and a CS Master's from Tsinghua, before receiving his Ph.D. in artificial intelligence from the Swiss Federal Institute of Technology (EPFL). He's won numerous awards, including the China Air Robot contest, and was instrumental in the successful completion of China's first UAV mission (China's 24th Antarctic expedition).
Li joins us today to explain the following: Ninebot and Segway originally sold self-balancing scooters, so how and why did they decide to develop robots? After Ninebot acquired Segway in 2015, what was the impact on collaboration and partnership between the two entities? Why is there such a keen focus on personal robots, when other industrial robotic applications currently comprise 75% of revenue in the market? What are the key drivers and barriers in the personal robotics space? Why is it important to create a robotic platform as opposed to a singular product, and how will Ninebot build an ecosystem of developers to come up with killer applications? What is the competitive environment like globally, and what are some major differences between the US sand Chinese robotics ecosystems?
Transcribed by Shaolong Lin; Edited by Alissa Wang
Link to SoundCloud (here), also available via Apple Podcasts, Google Play, etc.
[Editor's note: this interview has been edited and condensed for clarity, and is based on a podcast interview conducted in June 2017]
Li Pu Joins Ninebot
Adam: Hi Li, can you tell us more about your story and your experience with Ninebot and Segway?
Li Pu: My story with Ninebot actually goes way back to when I was a student in Beijing. At that time, I was involved in a lot of robotics projects and got to know the founders of Ninebot. And when I was working in San Francisco at Twitter, I met with one of the founders, and he told me that it's a good time to enter the robotics business. I looked at the environment, today’s technology, and the market for it, and realized that he was right, so I decided to come back to Beijing and join Ninebot and Segway Group.
Ninebot was originally founded for short-distance transportation back in 2012, and it launched the first model in 2014. That was a self-balancing scooter and could carry a person in a very agile way. With Ninebot, people can do short-distance transportation very easily from home to work or from the office to a shop nearby.
Adam: Just to confirm Li, so the original Ninebot is essentially a self-balancing mechanical scooter. It's a flat board with two wheels on each side and self-balancing components so that the user can just step on top of it and zoom around every which way?
Li Pu: Yes, exactly. It’s very similar to the original Segway. In Ninebot’s case, you step on the board, and there’s a steering handle that you can push left, right, forward and backwards by shifting your center of gravity. And in April 2015, we announced the acquisition of Segway. So now it's Ninebot and Segway Group. This includes the entire IP portfolio originally held by Segway.
(Ninebot scooters | Source: xiaomi.com)
Opportunities in the Personal Robot Market
Adam: Got it Li. So what’s happened post acquisition?
Li Pu: After the acquisition of Segway, the group announced a new model, the Ninebot Mini, which is a smaller version of the original Segway. The original Segway sells about $6,000 to $8,000 USD, but the Ninebot Mini is selling on Amazon for $599 USD. So it's a magnitude of order cheaper than the original Segway model. And with this Ninebot mini model announced, we thought it's a very good base for a mobile robot.
The mobile robot is the dream of a lot of people, and you see that in scientific movies as well. We think it's a good time to start this because of several reasons. Firstly, all the components that you need to come up with a mobile personal robot are there. Ten years ago, if you wanted to buy a gyroscope sensor at it would cost $50. And now it's only less than $2 and the quality is even better than ten years ago. Also, the other types of sensors like cameras have been widely used in smartphones, in laptops, etc. And we can use that in conjunction with computer vision algorithms, input streams for the robots. Computing power has also clearly advanced a lot in the past ten years. And we know that even with a CPU or GPU that's currently in our smartphone would be enough to run very complex algorithms to power the robots. So technically everything is ready.
On the business side, just as an example, we see that there's the trend of aging populations in the world. There will be a labor shortage in the upcoming years. For example, in China we know that this year in 2017, the size of the working population will start to decline. So a lot of things have to be done in the future with robots, simply because we don't have enough people in the world. And we see that it’s an opportunity for us.
Robots: Productivity vs. Entertainment
Adam: There's a large market opportunity when it comes to robotics, particularly given some of these macro trends, whether it's a shortage in labor force, declining prices for components or increasing computing capacity. But taking a step backwards, looking at this overall market, at least currently, the largest revenue share still goes to enterprise and commercial applications. For examples, the use of robots in factories and warehouses. Yet you still focus on the notion of the personal robot. Could you please walk us through your thinking there.
Li Pu: When we think about the personal robot, we often give this analogy about the personal computer. Personal computing was created long time ago. And first time there is a distinction between the computers used in industrial circumstances and the computers used in home or in family environments. The two serve different purposes. And this is true for the current situation in the robot market.
We can say there're a lot of industrial robots, like robot arms, like in the car manufacturing, in the cellphone manufacturing and many other industries. But their main purpose is to increase the efficiency of manufacturing processes. And in the personal robot landscape, right now they are mostly for entertainment purposes through their interaction with people. They make jokes or tell stories to the users.
But we think in the future, the boundary between the industrial robots and personal robots will be less distinct. Just like for personal computers, these robots will move beyond just entertainment purposes. You could use your personal computer for watching a movie, or watching youtube videos. But you could also use it for productivity purposes, like using the software to do actual work.
Adam: The analogy of the personal robot compared to personal computers is an interesting one. But just for our benefit, can you give us few examples of how to use personal robots. What can they actually do for the user?
Li Pu: So we started this personal robot project, which we call Loomo. We first showcased Loomo in the beginning of 2016 at the opening keynotes of CES, with Intel CEO Brian Krzanich. At that time we showcased several potential use cases with this personal robot. One use case was to act as a security guard at home. When nobody is at home, and someone is knocking at the door, the robot would automatically navigate to the door and see who's at the door. And if it's someone who's been recognized by face recognition, the person can get into the room, or otherwise the robot would kick him out.
Adam: Yeah. There are some really cool examples. I saw that video and it makes me want to get one myself. Oh and by the way, Loomo looks like the robot from Wall-E, that movie. So was that the inspiration there?
Li Pu: Hah actually yes! A lot of inventions originate from the ideas of science fiction. There are just countless examples right? So we do borrow ideas from Wall-E. We're not stealing the ideas but instead, we draw inspiration from them. Some of the features of Wall-E will become goals for our robots. From these movies, we see the need for robots doing different type of works and even becoming companions for the users.
(Loomo robot | Source: SGInnovate)
Loomo, the Open Robot Platform
Adam: OK. So now we understand what Loomo is. There's a lot required to perform these robotic functions. There’s the mobility component of the Ninebot and Segway scooters. You have the computer vision that allows the robot to recognize things, images, objects, people. There are robotic arms as well that allows the robot to perform physical tasks. Given all these infrastructure requirements, can you tell us a little more about how to build software applications on top of that.
Li Pu: Let me give you an example. When personal computers first became available in 1977, Steve Jobs had created Apple II, that's like Macintosh for the personal computer. But in that first year, the sale of Apple II was not that good because there were few software applications that people can use with these computers. And keep in mind the price of those computers was also pretty high!
Therefore, in the first two, three years, it was not that popular. But after a while there were some killer apps invented. The first popular app was called VisiCalc. It was invented in 1979. VisiCalc is a software like today’s Excel, a spreadsheet software. People started to think that with this kind of software they could do things beyond entertainment. The computer could help them with their tax reports and calculate some formulas they wanted to use for work. And one year after that was launched, the sales of personal computers doubled. Only a few years later, the sales of personal computers reached one million units. From this example, we can see that there is a natural process that follows the creation of a computer or a personal robot platform. Once we have the platform, we need to wait for the killer apps to be invented. Once people start to use personal robot, software developers will get feedback from the users. With that feedback loop we can create popular apps. It just takes some time.
Adam: It makes sense to create a platform to democratize these capabilities for a number of developers to create on top of. Can you tell us a bit more about your efforts there. How have you created this developer community, and how do you increase the level of talent who will help you build these applications?
Li Pu: When we started to think about this robot product, we looked at the market because we wanted to develop applications on top of platforms. But we found that there was no platform that we could use. The only personal robot at that time was the robotic vacuum cleaner, and this product was not powerful enough to give us all the capabilities to program robots and do versatile things. So we started to create our own robot platform, which we caled Loomo. From day one, we have decided that the Loomo platform is going to be open platform based on Android. Developers who want to program the robot application can program very easily on the Loomo platform as if they are programming Android applications. From the beginning of 2016, when we announced Loomo, we have received over 2,000 applications from developers all over the world who wanted to develop on this platform. When we gradually developed the hardware platform, and when the hardware platform was getting ready, we gradually started to give them opportunities to test things out.
For example, over the summer we did several hackathons in different places. One in Singapore, one in Boston, one in Shenzhen. The developers can easily use this platform to generate some applications. For example, one of our developers from Singapore explored the possibility of monitoring elder people in their home when they are alone. If a robot detects symptom of dehydration, it will send an alarm to the doctor, who’ll tell the patient to drink enough water. All these applications are done by those developers in a hackathon within one or two days. It is a great platform because people can validate their ideas in a relative short period of time instead of getting their own robot hardware and assembling them over a long period of time.
Monetizing the Open Platform
Adam: Bringing it back to the business side. In this set up, how do you and your developers make money? Do you sell your robot hardware and offer this platform free on top of it? Can you walk us through these details?
Li Pu: Right now we do sell the robotic platform. Our product is categorized into two categories. The first is called the developer edition. Developers could buy their robot platform and program their own applications on top of it. The second category is called the consumer edition, where we provide a suite of softwares or applications with the robots, and when the robot is delivered to consumers they would get the applications already. They don’t need to program themselves.
The two categories are serving different purposes. The first category would give developers the ability to fulfill their tasks. The consumer edition will give us a larger installation base. We know that without this large installation base what the developers create won’t be used. What we want to create is on one side, we have a large installation base, and on the other side, the developers develop more applications for this installation base. The users in the installation base would be happier because they have more applications.
Adam: Just to push a little bit further, so for the developers, how do you incentivize them so they create very cool applications on top of your robotic platform? And when the end users buy one of these robots and access one of their applications, do they pay to use that? Is it similar to a smartphone platform whereby Loomo is the smartphone in this case?
Li Pu: That is the business model that we are thinking about, but it’s not implemented yet. We think this business model has already been proven in the smartphone environment, and we think it’s a very natural extension from smartphone to robotic applications.
The Competitive & Global Landscape
Adam: That makes sense and sounds very exciting. So, taking a step back to look at the overall environment, could you tell us a bit more about competition. Are there any other companies that are doing something similar, and how are they positioned relative to Ninebot and Segway. How do you see this playing out over time?
Li Pu: A lot of companies are starting to look into this field. For example, there are plenty of vacuum cleaner robots as well as desktop robots that interact with people without moving. If you think about a broader range of products, drones would be another category as well because they have all the robotic features nowadays. We think we are not directly competing with anyone in this market, because our product is quite different and quite unique. We have this self balancing vehicle, and we don’t see any other company doing products with the same concept. But we know all the personal robot applications in some point would converge. At some point there’s going to be a killer app. With that killer app, all the competition will stop in that moment because everybody wants to get better user experience with this killer app. At that moment, competition is a good thing because we give our users better products at a cheaper price.
Adam: Last question Li. Beyond domestic competition amongst companies, given that you’ve been in both China and the U.S., and given that Ninebot and Segway naturally have operations in both markets, can you tell us about any differences between markets?
Li Pu: I think the concept of personal robots is different across markets. I would say in the US, because of scientific literatures and scientific movies, people are more familiar with the concept of personal robots. People in the U.S. are more eager to try out the new product. Our strategy is to first target the U.S. market because it’s the best test base of new products out of all the markets in the world, and we will try to refine and improve our product. When we get to a point where we think it’s good enough to extend to other markets, we will come back to China and also Europe.
Adam: Makes sense and all sounds very exciting. Thanks for your time Li!