Making Robots Personal
An Interview with Tandy Trower of Microsoft Robotics
Later this week, high school students from around the country will gather in Atlanta, Georgia to compete in the FIRST robotics competition. Booked for the Georgia Dome—home of the Falcons NFL team—the event is not just a small gathering of tech-minded kids. Teams totaling more than 10,000 students from 28 countries will compete against one another with autonomous and remote-controlled robots in an arena full of screaming fans. This is varsity-level engineering.
Listen to a full recording of the conversation with Tandy Trower (46:30):
You can also download the interview in .mp3 format.
The desire of FIRST organizers to get students hooked on science and engineering early is embodied in the organization’s name: “For Inspiration and Recognition of Science and Technology.” And they are succeeding. According to Tandy Trower, General Manger of Microsoft’s robotics group, metrics from the group show that students participating on robotics teams are 50 percent more likely to go on to college and four times more likely to major in engineering than their peers. Part of the key, says Trower, is the engrossing power of the little machines. He calls robots “marvelous motivating devices.”
Trower, a 27-year veteran of the software giant, has been motivated by a variety of technologies over the course of his career—from the BASIC programming languages that enabled the expansion of PCs in the 1970s and 1980s to the first two releases of Windows, which he managed. A few years ago, Bill Gates sent him on a fact-finding mission to figure out what Microsoft could do in the field of robotics. Since then, he’s learned that the machines aren’t just great educational tools. Indeed, the field of robotics may hold solutions to major problems in military transport, providing health care for an aging population, and keeping our floors clean.
Science Progress spoke with Trower about the future of robotics in the United States and around the globe. The interview has been edited and condensed; a recording of the full conversation is available in the sidebar.
Andrew Plemmons Pratt, Science Progress: How can robots can get students interested in science, technology, engineering, and mathematics fields?
Tandy Trower: I think it’s incredibly important because today, the U.S. is ranked about 6th in engineering bachelors degrees, following China, the EU, Russia and India. It’s starting to look back up again, perhaps with the change in economy, but traditionally, in computer science in the U.S., there has been a steady decline in enrollment. It turns out though, that robots are just this marvelous motivating device. If you stick a robot—I don’t care if you’re talking about grade school kids or high school students—if you put a robot in the middle of the room, there is something captivating about the technology.
It’s true that robots, just like PCs, need that kind of compelling application that really demonstrates their value, and I think we still haven’t seen that.
At Georgia Tech, for example, they have been able to use robots as a way to teach basic computer science. They’ve been able to stem the tides in the decline of computer science by using this technology. But even before the college level, you have a number of programs throughout the U.S. which are just tremendous in terms of their benefits. For example the FIRST organization, which was started by Segway inventor Dean Kamen. What he loves to talk about more than anything else, more than the things he’s created, is this organization where the primary competition is for high schoolers to build robots that compete.
I had the opportunity to welcome the students to the Seattle regional event that Microsoft sponsored, where they actually doubled the number of teens that participated from the previous year. They had this very complex competition where the teens had to build robots to distribute foam balls from one location to another, and the engineering was just incredible.
But the key thing about FIRST is its not just about the kids building the robot. It’s about teamwork, collaborative interaction. The teens not only get points for the competition itself, but also for collaborating with other groups. So you’ll often see one team help another team out; if one team is having trouble with their robot, another team will be motivated to go over and help them out and make sure they can compete. It’s about marketing also. I was led around by a student who was a senior at a high school in Spokane. She didn’t get to participate that much in the engineering of the robot—that wasn’t her area of expertise—but she was able to participate in the promotion of what her school was doing, what her team was doing, and she was so excited about that.
The results from FIRST are incredible when you look at the history, and they’re not the only competition. If you look at the statistics that the FIRST people provide, they mention that students that participate in their teams are 50 percent more likely to go onto college and four times more likely to major in engineering. They have some incredible stories about the different high schools and the benefits they provide. There’s even a high school in Denver, Colorado for convicted felons, students who got into crime early in life. The school is offered as an alternative to prison; only 13 percent of their students go on to do post secondary education. But of those that participate in the FIRST competition, 80 percent go on to post-secondary education.
So the impact of this technology is just incredible. It’s also very important, as the industry needs to feed itself. If we’re going to see the breakout of this technology in the next ten years, what we really need is a set of engineers. The U.S. ranks behind much of the world population in terms of engineering students, so particularly in this country, it is important to have these kinds of motivating competitions to get kids excited about not just robots, but the whole idea of what it means to work collaboratively on an engineering project. Whether it’s on the side of the engineering software, or the hardware, or from the organizational principles, the presentational principles, or the marketing—the fact that you can use a robot to provide this sort of motivation is not only beneficial, but it’s critical for us being competitive in this technology.
SP: When people think about robots they have images from movies, TV, and science fiction—but in fact robots have been an integral part of U.S industry and manufacturing and other sectors for a long time. And a lot of robots like the Roomba don’t really fit the traditional molds of what robots are. What do you see the future of consumer robots looking like?
Trower: It gets down to the definition of what a robot is, and often that’s the hardest question I get asked. What do you think a robot is? How do you define one? And the difficultly is, as you point out, that there are so many different kinds of robots that it’s difficult to characterize it and put it all in one place.
Robots have historically been a real boon to factory automation and the manufacturing sector. Most cars that are produced today, go by and see several robots along the way. They’ve really been involved in what the industry calls the “three Ds”: the dull and the dangerous, and dirty kinds of work. But what I see happening right now an evolution of the technology where it’s moving out of the factory floors and starting to show up in our homes. You pointed out one of the excellent examples of that: the Roomba from iRobot. They sold between three and four million of those little vacuum robots; they’ve been extremely popular. There’s also an increasing number of very sophisticated robot toys that are on the market place, whether we’re talking about the little Pleo robotic dinosaur from Ugobe, or the LEGO educational robotic kit, or the latest version of Elmo from Sesame Street, which has become increasingly sophisticated over the years in terms of the abilities of what he can do.
What we’re seeing in that technology is two things. One is that the technology is becoming increasingly more sophisticated; it’s becoming increasingly more oriented towards interacting with human beings. The factory robots have been too dangerous for most people to interact with, they’re kind of like what the mainframes were like in the ‘70’s: they’re big and require specialty operators and really didn’t effect people in personal way, unless it was the university computer figuring out what classes they were going to take, or run their grades or running some accounting. People didn’t really feel the same way about them, and yet when PCs came to the market it really changed that dynamic. What we see is these technologies starting to creep in the door and not just in the obvious forms we’ve seen so far.
Take your car for example. If you have an anti-lock breaking system in your car, that really is a form of robotic technology; it’s a system that senses a slip of the wheel and adjusts the breaking, and really changes the behavior of how you drive. When I learned to drive they told me on slippery roads you have to pump your breaks. Well, when my daughter learned—and she learned in a car with anti-lock breaks—they told her just to apply steady pressure, because the system takes over, so that’s a robotic system.
I’m seeing an additional evolution where traditional PC technology is starting to get integrated. Japanese robots are an example, particularly the humanoids. Those are a very advanced form, a very challenging form of robots. I think we’re going to see generations of robots somewhere between the Roomba and those kinds of robots. They don’t all need to walk around to be useful. We’ll see a generation of these mobile robots that piggyback on top of the PC technology we see today and leverage the wealth of information and productivity we have. Then I think the next generation will be amplified. We’ll see manipulation come in soon after: the ability of the robot to safely manipulate things in a human environment, and to be mobile and function safely in a human environment.
SP: Bill Gates asked you to go out and sort of explore the state of robotics; you traveled all over the country and talked with people in the field. In the intervening years, what has changed since you went out exploring?
Trower: I’ve been at Microsoft for 27 years now, and I’ve had an opportunity to play around with and lead and manage a number of different projects with Microsoft. But this was a unique challenge that I got to go out and talk to the community. The first thing that I would say is it’s a very diverse community. There’s not just a single segment of it. There are people who are using robots in education; we talked about the industrial factory automation usage; there are people doing very intensive research out there in the university community; and there’s even this evolving, emerging generation of these consumer robots.
This diverse community in some ways parallels the early PC community. It wasn’t just a particular segment. It wasn’t just people who loved computer science. It was really quite a diversity of people who were interested in using the technology in a variety of different ways. I also discovered that it’s definitely a world-wide phenomenon. There isn’t just a single place in the world where this is happening. Traditionally Japan has been the hotbed of robotics development, and there’s defiantly still great work that’s being done in there. But what I see now is that it’s becoming more wide spread throughout the world in terms of the investments that are going on in this space. In addition, its not just the investments, but every major economic entity seems to have robotics on their list of significant technologies that they want to participate in and believe will be an important part of their future.
So there’s all this excitement and anticipation of something that’s coming, and that was actually the thing that motivated Microsoft, and me particularly, to show an interest in this. This community was communicating with us at Microsoft, saying “Do you realize that there’s something significant that’s starting to happen? Its not just the traditional factory automation, its really starting to blossom in a number of different areas.” And they invited Microsoft to participate. It’s also an area where there’s this tremendous amount of research going on, breakthrough research.
Ten years ago something like vision recognition technology was still a black art among many developers, but today you can find a number of examples where you can just go to the web and download several algorithms and put this on some of the simplest robots, like the little education robots that are on the market today. I found all these great things, and the technology was becoming more accessible and more affordable to a wider audience. But I also learned that it’s not all just rosy out there. It turns out that there are some development challenges. The difficulty is in trying to bring software from one platform to another, every robot is still kind of its own world unto itself—the technology still needs to go through a little evolutionary refinement. Using the PC analogy, it’s like we’re looking at the Apple IIs and the Commodore PETs and the Sinclair Micros that were available in the 1970s. It wasn’t until 1981, when IBM came to the market, that we started to see a more stable platform.
Those are still some of the challenges, and probably the biggest one that everyone identifies—again this is very analogous to the early PC industry—is the fact that there really hasn’t come to the front a compelling application. In the old days we used to call those “killer applications,” but that’s probably not the right term to use here. But it’s true that robots, just like PCs, need that kind of compelling application that really demonstrates their value, and I think we still haven’t seen that. We’ve seen a tremendous amount of innovation in terms of the technology. With the DARPA Grand Challenge, we saw the ability of teams to build cars that could navigate through deserts or even on simulated urban roads with other vehicles running around and obeying the traffic rules. So we see the evolution and innovation going into the technology, but we still haven’t seen come to market yet that compelling reason, other than entertainment or education, for everyone to want to have one of these in their home.
SP: Can you explain the DARPA Grand Challenge?
Trower: DARPA set up what was originally called the “Grand Challenge,” and the second one was named the “Urban Challenge” because it was a different scenario. Teams had to build vehicles that could autonomously, not by remote control, navigate through roads in the desert. There was no one on board, just computers with sensor systems. In the first year that they ran the challenge, there were a number of teams that participated, and even the best team—which was from Carnegie Melon, led by Red Whittaker, who is a real pioneer in the robotics area—even they failed. They got seven and a half miles though the course. They had built their robotic car on a Hummer platform and it got stuck and they had to shut down because it couldn’t free itself.
DARPA did a brilliant thing. They just banked the money and held the challenge the following year and four teams finished within the regulation time. Even though he actually had two teams that qualified in the end, Red came back again and it looked like he was going to take it. In the end he got beat out by a former Carnegie Mellon professor, Sebastian Thrun, now at Stanford. Sebastian with his team narrowly beat out Red, and so you actually had five teams complete the course.
Then DARPA said, “Ok that’s great, the vehicles by themselves can navigate across desert roads. Lets up that again.”
They took over an old army base with streets and they set it up so that the vehicles had to navigate across these streets and had to obey the traffic rules. They had to stop at the stop sign; there were other robotic vehicles as well as human-powered vehicles traveling on the roads at the same time. Now the real incentive for all this was that they had a Congressional mandate to develop technology that will allow them to have a third of military vehicles be able to operate autonomously. What I believe will come out of this, even though it was done with kind of a military mentality, will be some tremendous innovations in even the way you and I drive safely on the streets.
In some ways that’s not so unusual, DARPA is the same agency that’s responsible for the ARPANET that became the foundation for the Internet that we have today. And so I think it’s a great way of using public money and while it may have had military purposes, the implications and the ramifications it may have down the road are tremendous in terms of pushing the technology. Here was an investment they made—a fairly modest investment—it was a million dollars, but it was still a modest investment overall when you think about that in terms of getting all these all these brilliant minds to solve a very hard problem. And what came out of it was not just one solution but many solutions to really advancing the state of the art in terms of technologies that will not only allow us to drive more safely in the future, but will really advance the state-of-the-art of robotics and effect our lives in a very personal way.
SP: What are some of the possibilities for those critical applications that new types of robots would be able to take on? One thing as well that you’ve spoken about are robot applications that can support citizens with disabilities, or help care for aging members of the U.S population.
Trower: Robotics still needs that compelling reason for why people would want to have one in their homes. Healthcare is one of the biggest opportunities and the biggest needs for technology, especially digital technology, which includes robotics but is not exclusive to robotics.
Today in the U.S alone there are over 40 million people over the age of 65, and that is expected to almost double in the next 20 years. Not only that, but the number of people over 80 is expected to triple. What this means is that we have more people that are going to be in the senior category and more of these people are living longer lives. At the same time what were finding is that already there is a gap in the care industry for being able to care for all of these people. So the question is how are we going to solve this problem in the long run?
Everyone knows that as we age, our physical and cognitive capabilities diminish and we become increasingly prone to chronic types of issues. For example, at age 65 your chances of getting Alzheimer’s or Parkinson’s disease increases exponentially. So how do we deal with these problems? With the prospect of increased costs? We have a larger population and fewer people to take care of us. And it’s not just in the U.S—this is a world-wide phenomenon, and if you look at the population curves which traditionally have been this very nice pyramid where the smallest part of the pyramid was the oldest people and the largest was the young people, that’s actually flattening out and is expected to invert as we go into the future. Robots and other forms of digital technology are a great way of trying to address this gap, not as a replacement for care givers but as a way to extend the care network that’s there.
The ways that I think it can do that are in terms of providing communication, just as PCs have become great tools for communication as much as they are for productivity, whether its email or other kinds of communication. For example my daughter who is graduating from high school this year. She communicates more with her friends through her PC more than she does on her cell phone or on the house phone, so it’s been a great tool for communication. That’s particularly important because, as seniors, your social circle naturally becomes smaller as you age. Your friends may be there. It may be more difficult to get out and see your friends and family. So it’s being a communication aid, a connection with the care network to doctors and nurses, rather than having to make physical visits, helping monitor the state of how people are.
There’s also a role in terms of memory aids, whether it’s just remembering to take medication, which I think is a critical issue. One of the biggest problems that you find with seniors is they often have to be on a regime of several medications and applying some technology that reminds them, that doesn’t care how many times it reminds them, that never gets tired of reminding them and helps them stay on their regime, could be very beneficial as well.
Mobility is a significant issue for the seniors themselves, so if we talk about robots that are mobile devices, even if they don’t have manipulation. Robots with cup holders would be a valuable thing for moving things back and forth for seniors. There are valuable ways this technology can be applied to help solve what some people consider will become an epidemic in terms of how do we deal with this growing populations of seniors that are going to need assistance. We need to find some way to solve this, and while robots may not be a complete solution, they can be a part of that answer.
SP: Do you see a need for incentives or a specific public policy push to get people to do more work in robotics, say with healthcare applications or outside of the militarily driven ones that are happening now?
Trower: The military applications have been good. As I said, DARPA’s investment in the Grand Challenge and the Urban Challenge have been really great in stimulating the thinking, technology, and innovation that will have ramifications and implications in terms of improvements in the overall technology. I do think it would be useful or helpful. I’m concerned that in this country we may fall behind because in other parts of the world the focus isn’t just on the military side of things—its really more about care or other ways this kind of technology can be used.
In fact there are several agencies that are looking at that right now. There’s a consortium of U.S. professors led by Henrik Christensen at Georgia Tech, and it includes a number of the significant professors from all the U.S. universities who are trying to build awareness about the fact that we really do need greater investments in these other application areas. While we may derive benefits from the military applications, there are opportunities to invest in this other area and its not just a matter of research funding; it’s a matter of investing in the educational value. We really need to build up our own foundation of researchers that can really develop this technology. It’s a matter of changing policy to make this technology more accessible to people. If we’re talking about health care, how can these devices be easily certified so that they can be covered by insurance? It’s also investments in other technologies that feed into this, that are not just robotics-centered.
For example, for a robot to be functional and operational in the home, it has to have a source of wireless power. In some ways its actually good that we’re facing this crisis and trying to build fuel-efficient cars, because it’s driving us to more creative and innovative ways of building battery power, whether its fuel cells or batteries or even wireless transmission of power, those kinds of things will be critical because robot technologies will benefit as well.
SP: You have been at Microsoft for about 27 years. How does this kind of work fit into a long career in the technology industry?
Trower: The success of the robotic industry as a whole is really dependent on contributions from a lot of people, not just the ones who have all the resources. It’s probably been one of the most exciting things I’ve worked on at Microsoft.
For me it’s a bit like deja-vu. I’m old enough that I’ve been through the PC evolution and I’ve seen it go from kind of toy computers on one end into very practical devices: From where friends and family would ask “Why do you have such a thing as an Apple II sitting on you desk?” to where they all have their own computers today. And I’m seeing this again in the same way in the robotics community. But it’s different in the sense that it builds on the already rich foundation that the PC and the web have already set before us so that the possibilities seem kind of endless.
Comments on this article