Infusing PC Prowess to Produce Sellable Robot: A Chat with Tandy Trower
We are starting to see an influx of experienced people from the so-called
"IT industry" into the "robotics industry." The boundary between these
two industries is starting to collapse, and GetRobo feels that this
phenomenon will accelerate the commercialization of consumer robotic
products. One such figure is Tandy Trower, former General Manager at
Microsoft, who after 28 years at the behemoth, has left to form a
robotics startup called Hoaloha Robotics. He shares with us his vision
and an overview of the robot he is trying to develop.
(This interview was originally conducted for a robotics column on the
Wall Street Journal Japan.)
Q. According to your website, Hoaloha Robotics is developing
software and services to support new “socially assistive” robot
designs and that it will focus on "delivering pragmatic applications
that extend and augment human capabilities." Please give some
concrete examples on the kinds of tasks that you are aiming to
achieve using a robot.
A. First, let me say that delivering applications that provide value at an appropriate cost is definitely a core objective and one missing from being deliverable by most robots, outside of industrial, medical, search and rescue, surveillance, or military applications. That doesn't mean there aren't imaginative demonstrations that propose expectations like folding clothes, cooking foods, etc., but many of those demos are of expensive robots that most people would not be able to afford and they do not have any real user interface. It is even more obvious in the realm of "personal robots". In all my travels around the world, meeting some of the smartest people in robotics, I have found no one able to define a successful definition that combines the current technical capabilities, appropriate price, and user value beyond toys and gadgets.
But I digress, so back to your question. Hoaloha Robotics is targeting users that need assistance with daily activities as its initial starting point, so not the business user nor the general consumer. These latter markets will come, but it is much harder to create a value proposition at the present time for them. Meanwhile the worldwide population of people that require assistance with daily living activities is growing, fueled in large part by increasing life longevity and the aging of the baby boomer generation, while at the same time the human resources to help care for them is shrinking. So there is not only a business opportunity here, but a important social need that many already refer to as the Silver Tsunami.
With that context then, I can talk about applications areas that Hoaloha Robotics is focusing on as core to its solution. The first area is cognitive assistance. It is clear that as we age past our 50’s, our mental function - especially our memory - begins to decline. So a simple example of an application in supporting cognitive assistance would be a medication reminder. Most seniors commonly take an increasing requiem of medications. Helping them keep track of what to take and when to take them would be a valuable application.
Another application area is supporting physical assistance. Because we are designing for an autonomous - or semi-autonomous if you assume the robot is being directed by the user - mobile platform, it can serve as a simple transport for common items from food to items the user frequently may need to move with them, including the increasing number of biometric monitoring devices. And because the robot has a camera, actually multiple cameras and a display screen, the robot could be used as to magnify printed information or possibly even read that information to the user using OCR style technology.
Another important application area is supporting social/communications assistance. Humans are social beings and it is important to our health, regardless our age, to have regular communication with people around us and often as people age into their later years, their social network shrinks. Most of us already find social communication through a variety of forms (e.g. email, chat, texting) and it is important a part of our lives and not simply through our phones, but through our PCs and tablets.
And that brings me to the final area of the core application set. Because the robot is being built on top of PC technologies we have today, we should expect to see similar applications and services that are already available for PCs, such as information services like news, weathers, sport, financial info, and entertainment, such as games, music, videos, photos, etc.
Now some may say, why do we need a robot to deliver this when we could already deliver many of these applications on existing platforms today. To that I would respond that we already live in a world where we have technologies that overlap. I have a PC on my desk, carry a smartphone, and also travel with an iPad. Each form factor has its benefits and preferred usage scenarios, and the fact that I can do email, browse the web, check news and stock information on each does not invalidate any one of them.
But I would also add that the robot that we are defining offers a new form of interaction not found on the other platforms, one that enables the technology to be more interactive and proactive because it is self-mobile and because it is designed to be socially interactive. It is not just a stationary or passive platform that the user must go to or carry around to interact. In this way the robot offers the greater potential for interaction, more like a pet than an appliance.
One final point on the applications space here: I have tried to illustrate some of the types of applications we are developing, but it is also important to know that what we are developing is not simply to host our applications, but a platform that will allow for the integration of existing and new applications. Today as we see technologies moving into the assistive care space more and more, they each have their own interface and so require the user to be the integrator. However, our goal is to provide a common platform for integration of complementary technologies, and also provide a channel for third parties to bring their own ideas and creativity to this important market, a formula that Apple has so successfully done with the iPhone and iPad.
Q. Your website also states, "at the heart of the Hoaloha design is its human-centric user interface." Can you please elaborate on what kind of interface you are trying to develop?
So at Hoaloha we know we must develop an interface that is as natural and easy to use as speech, but incorporates some of these other important modalities that are a part of human communication. Included in that is orchestration of interaction in socially appropriate ways. For example, you know whether someone is listening to you by looking at their gaze and watching their facial cues. These help us understand whether they are engaged and following our communication or confused when we are talking to them. This I believe is Hoaloha Robotics' - and frankly the robotics industry as a whole - greatest challenge; that is, designing an interface that make interaction feel easy, natural, and satisfying; or to borrow a Steve Jobs' term - "magical".
Q. Hoaloha Robotics is a software company. Do you plan to sell your software to robotic hardware companies? Or do you plan to sell a complete system directly to the consumers? What is your business model?
A. Initially I started the company with goal of focusing strictly on the software side and licensing our software to hardware partners. However, I found it difficult to find companies really focused on developing the right hardware. We've had some good design discussions with a few, but nothing that has resulted in being able to find a platform that addresses what we see is needed. As a result we are now also developing our own hardware prototype, which is not a bad thing because a successful user experience requires a careful blending between hardware and software. Still perhaps this effort might just help model what is needed and we may yet find the right partner to work with us.
Similarly, I hope we can find partners who will sell and support the solution, as the market we are going after will require some serious investment in sales and support staff. A possible model here might be something like GE's QuietCare system for monitoring seniors in their home. The product was developed by an independent company, but GE invested and is the primary distributor for the product.
If you consider that what we are doing has no predecessor in the market and requires solving some really hard problems, my business model seeks good partners that can help us accelerate getting this into the marketplace. However, because this is a new product we know we could be faced with having to take things all the way to our target users. It is too early at this point to know for certain which path we will take.
Q. What is your ideal hardware platform?
A. To help illustrate one possible form I will share a rendering of a conceptual model. It’s not the design for the final product yet. It’s just kind of where our design thinking is right now. Our design process is iterative so what you see here is a snapshot that continues to evolve.
It consists of a cylindrical mobile base, with interactive components on the front and a carrying mechanism on the back. The interactive front has the ability to elevate and rotate, and features a dockable touch tablet, and a display that we use as a social display, or "face". Substantial research indicates that humans regard robots as social entities, even if we know they are just machines, so we are designing to address that expectation.
However, while the conceptual rendering illustrates some of the functional elements, it doesn’t represent our complete design. For example, we are working to get this to seem less like a robot and more like a smart pet or friendly companion. The problem is that just the word “robot” carries so many expectations on what a robot should be able to do and how it will interact that technology cannot realistically deliver. Movies and books portray robots to be super smart technologies that match or exceed our own abilities.
So our goal is to create more of an appearance that helps set expectations more appropriately and invites the user to explore and learn its behavior and capabilities through simple, natural interaction. Consider for example that when we talk to our pets or small children, we instinctively accept that they have a more limited range of language or abilities and so we tailor our expectations and interaction to that.
Q. When do you plan to come out with your first product?
A. It will probably be 2-3 years before we can get something in the market place. That said, what we are proposing appears to be a more aggressive timetable, at a more realistic, affordable price target than anyone else, for something that is more than just a toy or gadget.
As amazing are the some of the robots demoed these days, they often contribute to perpetuate an unrealistic illusion that walking, talking robots are imminent. But it will take some time before affordable, useful robots with arms and legs can be delivered.
At Hoaloha, we believe that you can start with simpler, more practical forms. Wheels are still cheaper, in terms of cost and complexity, to get a robot around a room than legs. You might not be able to go up and down stairs, but there’s still a lot of value you can provide. Just as personal computers didn’t have to match the capabilities of a HAL 9000 computer, or even initially their mainframe cousins, we are focused more on what can be done with affordable, off-the-shelf technology.
The emerging personal robotics industry still seems too enamored with the potential of what technology can do than focusing on what can be delivered to users today. I understand that film and fiction has set the bar very high, but I feel that we must refocus our efforts on value proposition and user experience if the industry is really going to deliver.
Q. What do you think is the price point for a successful socially assistive robot product?
We are working hard to get the price under $10,000, and possibly down to $5,000. That is still a lot of money for a robot, but when compared to alternatives like assisted care facilities, a $5,000 robot is a bargain compared to the $70,000 to $80,000 average cost. And if demonstrated to improve the well being and independence and reduce healthcare costs, it might even be able to be subsidized by insurance or other governmental subsidies.
Q. On your blog, you mention that "robotic technology is a new form of PC technology." Isn't the software development for robots different from PCs in that robots have moving parts and that there are a variety of hardware as well as environmental constraints?
A. Yes and no. On today's PCs we already see microphones and cameras almost as standard items, and more recently we are seeing depth cameras, like the Microsoft Kinect plugged in. So in terms of interfaces like USB or BlueTooth to support pluggable devices there are similarities. Likewise, you still need support for basic mechanisms for storing files and displaying information on the screen. But you are right that a mobile autonomous robot does add an important new dimension in terms of what is needed.
However, what I really meant by that comment is that we can use existing off-the-shelf PC hardware technologies. So if you looked inside our robot, you might just regard it as a PC on wheels.
Q. Besides the lack of an ideal hardware platform, are there any other major obstacles in developing a socially assistive robot?
A. I already mentioned that one of our biggest obstacles is that film and fiction have overset our expectations on what a robot can do. The instant you say robot, people think of an entity with AI comparable to a person and we aren't anywhere near there yet.
On the software side, there are at least 3 major challenges: 1) enabling technologies, like navigation, 2) applications that deliver a successful value proposition to the user, and 3) a "magical" user experience.
We also have issues such as wireless connectivity. Robots that are not tied into the digital ecosystems already around us are not as valuable as those that are. The networked robot can share as well as gain information that helps it do a better job. In addition, as cloud computing comes on-line, services can be brought down from the cloud as well as non-critical remote processing so that the robot's onboard processor just has to deal with immediate tasks. But wireless is still not pervasive and it takes some effort for users to configure.
Finally, educating the market also still needs to be done. When TiVo was introduced, many people didn't understand the value of the technology until you demonstrated it to them. However, once demonstrated, consumers quickly understood that this changed the way they would watch TV, not simply how they stored video programs. So too we will need to determine how best to introduce this technology and demonstrate its features.
We enthusiastically agree with Mr. Trower as one can read and see here:
http://www.geckosystems.com
Martin Spencer
GOSY.pk
Posted by: Martin Spencer | September 26, 2011 at 05:57 AM