Roboticist Daniel H. Wilson talks about robots

Energy in Nature	Inventors & Inventions	Workbench Fun	Physics Lab		Consumer Products	Computers	Automation & Robotics	What's in the Future

Friend or Foe?
Dr. Daniel H. Wilson talks about robots

If you love electrical technology falling in love with robots is pretty easy to do. Earning a PhD in the subject at Carnegie Mellon University is much harder. Daniel H. Wilson has accomplished both, in addition to authoring "How To Survive a Robot Uprising: Tips on Defending Yourself Against the Coming Rebellion" and "Where's My Jetpack?: A Guide to the Amazing Science Fiction Future that Never Arrived". He also has a great sense of humor and has shared his views with us from the inside.

Were you always interested in robotics and what were your favorite toys when you were a kid?
Robots are just plain fascinating to me and they always have been. When I was a kid I used to play with Transformers and Go-Bots quite a bit. That was back in the days when those robot toys transformed into hand guns and cassette tapes – a long time ago! Other toys that were really fun were Capsella, which were sort of like Lego Mindstorms and Tinkertoys combined. As a kid I loved all kinds of sci-fi toys, but robots are the only sci-fi that turned out to be real.

Most robotic toys seem to have a destructive, rather than a benevolent, nature or purpose. Do you think this has an influence in why we grow up thinking they could rise up against us?
It’s really, really fun to think about escaping from giant killer robots. Imagine darting through the city streets while telephone-pole sized robot legs sweep through the mist and giant pincers smash through building walls all around you. Fun! Now let’s take a look at a real robot: The Da Vinci system is a robot surgeon that sits in an operating room, moving quietly and precisely to stitch up a patient. Kind of boring compared to Hollywood special effects; that is, until you really think about it. The surgeon robot is being teleoperated by a doctor who is so immersed in the experience that she feels as though she has been shrunk to the size of a bacteria and is traveling under the patient’s skin. Neat!

Judging from the tone of your book “How To Survive a Robot Uprising” you aren’t in fear of a Terminator type scenario happening anytime soon. What can you say that will ease the minds of the anxious about that?
Building robots is very hard. Building smart robots that are capable of general-purpose thought is nearly impossible. This means two things: First, obviously most robots will be incapable of hurting you because they aren’t as fast, as strong, or as smart as you are; Second, robots are hard work to create and so the people who build them are seriously committed to working together as a team to make solid improvements over the long term. There is no such thing as a mad scientist who comes up with a killer robot overnight and unleashes it on everyone – the scientists and roboticists are the good guys!

Is the biggest challenge to making life-like, functional robots in the mechanics or the controlling computation and software?
Brains and brawn are both important aspects of building a good robot. We used to think that brains could overcome brawn, so that even if a robot was built out of toothpicks it would be fine as long as the robot was smart enough to control itself efficiently. Most roboticists don’t think that anymore. Robots have to be built from the ground up to fulfill a purpose – form follows function. That said, the biggest challenge is still creating robots with software that is smart enough to help them learn how to walk and manipulate objects by experimenting in the real world and in simulation, and how to keep working when conditions change. This area of study is called machine learning and it is one of the most exciting areas of robotics and computer science – you get to watch your robot learn!

Most of the current robots seem to operate by rotary motors and gearboxes. This is much different from the animal motion of contracting muscle cells bundled together. Do you see robot technology moving in this direction in the future?
Most robots tend to move via electrical motors because they provide precise and accurate control. The next most common actuators are pneumatic or hydraulic. These are air (or liquid) filled hoses that flex when pressurized. They are really strong, but the control isn’t as easy. Electric, pneumatic, and hydraulic motors are in common use today, but research is happening on creating muscle-like actuators called “electro-active polymers.” These are little bits of spongy plastic that will contract like a muscle when hit with an electric current. Sadly, they are not much use on the macro level and are mostly envisioned for use in tiny, nano-sized robotic platforms.

Science fiction writers have imagined very advanced robotic creatures for over half a century. The reality so far, however, has fallen quite short of that. Do you think that the pace of advancement will start to quicken, or are future predictions still way too optimistic?

The BEAR robot has enough
power & stability to lift and
carry a human

The field of robotics is blooming and the pace of development is accelerating wildly. New robots appear almost every day! Plus, with robot kits and cheap hardware (like cameras and microprocessors) there are more people who are able to learn about and build robots all the time. We saw computer science blossom like this a decade ago, and I expect robotics to do the same now. The reality of robotics even now can be mind boggling – have you seen the BEAR robot, built to scoop fallen soldiers off the battlefield!? Whoa!

In what areas do you see us realizing the goal, and where do you see frustration for some time to come?
General-purpose artificial intelligence (AI) is the hardest nut to crack. An AI with general-purpose intelligence could talk to you just like a human – it would be impossible to tell the difference. Creating a robot like this is called an “AI Hard” problem, meaning that in order to solve it you have to solve every problem in artificial intelligence. But nobody is frustrated, because we keep making progress toward the goal all the time.

What do you think are the strongest driving forces behind robotic technology, and in what direction do you see advancements coming most successfully?

The DARPA Grand Challenge race:
An autonomous, unmanned vehicle
travels on its own across the desert

Money drives robotics research and development. Currently, that money mostly comes for healthcare or for military use. In the military arena, this means unmanned vehicles – cars, planes, and boats that can drive themselves and it means small mobile robots that can operate in hazardous environments, like the TALON bomb-defusing robots. In healthcare, this is all about robots that can take care of people with disabilities or people who are elderly. Those robots look like humanoid servants or they may disappear entirely, appearing as a smart house that is keeping track of occupants’ health as they go about their daily lives.

During the ‘90’s smart-house technology was expected to be big business by now. A lot of what we saw come to market though, just seemed to be technology for technology’s sake, and not really created out of necessity. An example of this was how you could turn your appliances (such as a toaster) on and off from your PC at work through the Internet. Interesting but not a compelling system to spend your money on. In what ways can smart-house technology actually deliver on its promise of improving our lives?
I agree, a smart house that fetches sodas is not useful. Instead, most money in smart house development is funded toward healthcare. My research at Carnegie Mellon University was in this area: We filled up a house with sensors, such as motion detectors, contact switches, and pressure mats; and then used machine learning to help the house learn the patterns of daily life of the people who lived in the house. The house couldn’t fetch a beer or talk to you, but if your pattern changed and it looked as though an elderly occupant was having trouble, the house could reason out the cause and notify a family member or caregiver.

learn more

For a young student or enthusiast, how would you advise they get started in following their interest in this type of technology?
I’m going to confess – I’ve never really loved math. But if you plan on building a robot that can think for itself, you’ve got to learn as much math as you can. Trust me, it’s worth it! Learning how to program is also important, but not absolutely required. Working out the math is the hard part, you can always get somebody else to code it up. My advice is to play with Lego Mindstorms, buy robot kits, hack robot toys, and compete in a FIRST league. After high school, get a degree in math, computer science, or engineering and then go to graduate school and study nothing but sweet, sweet robots!

add to favorites | del.icio.us Slashdot It! Stumble It!