Humanoid Error: How to fall (back) on your bot

By COLIN DOBELL, President & CEO

This week, a number of people forwarded me links to a New York Times article about how the team from Korea’s Advanced Institute of Science and Technology (KAIST) recently wobbled away with the two-million-dollar grand prize at the 2015 Robotics Challenge Finals (DRC) in Pomona, California.

The DRC, sponsored by the US Defense Advanced Research Projects Agency (DARPA), challenges roboticists to develop systems capable of assisting humans in responding to natural and man-made disasters, such as 2011’s Fukushima Daiichi nuclear incident. Competition tasks included driving a vehicle, opening a door, turning a valve, drilling a hole in a wall, walking on uneven ground and climbing stairs.

It all makes for great theatre, especially if you’re the type of person who watches sports for the slow-motion replays. But it raises a question: why do the robots have to look like us? As Matt Simon puts it in a piece published by Wired Magazine in the days following the event, “C-3P0 was built to interact politely with human beings, but R2-D2 was built to do actual work. So why even bother developing bipeds?”

The consensus is that, in a physical environment designed by humans, humanoid robots will fare best, in terms of maneuvering around and interacting with man-made objects. And in communications-compromised situations, they will need to be at least semi-autonomous. Fair enough. But what about real-life tasks that need doing right now, if not yesterday?

Now, don’t get me wrong — the DRC produced some pretty cool new technology, but the idea of full autonomy just isn’t realistic any time soon. Speaking with the Times, even DARPA program manager Gill Pratt had to admit that, however human their movements, the robots’ level of automation is still relatively basic. His exact words, in fact, were: “These things are incredibly dumb. They’re mostly just puppets.”

a custom snake-like inspection crawler designed specifically to navigate the complex interior of the FUKUSHIMA reactors. Equipped with a series of Inuktun’s stainless steel Microtrac™ components, the remotely driven robot was made to slither through a four inch pipe, then dangle itself and descend on to a platform just below the reactor core’s bottom. 

a custom snake-like inspection crawler designed specifically to navigate the complex interior of the FUKUSHIMA reactors. Equipped with a series of Inuktun’s stainless steel Microtrac™ components, the remotely driven robot was made to slither through a four inch pipe, then dangle itself and descend on to a platform just below the reactor core’s bottom. 

So, although the eventual use of advanced and autonomous robotics for real-world applications is pretty much guaranteed, that day may be a ways away yet. Meanwhile, there are standard robotic systems in use around the world, including Fukushima. These bots are already saving time, money and lives — and allowing people to do things that probably could not be done otherwise.

In fact, tele-operated robotic systems are everywhere you don’t want to go. They’re currently being deployed in radioactive environments, deep under water, in confined spaces and even in high temperatures. OK, we’re not talking Mr. Data or Johnny 5 here; these sorts of electromechanical systems are really just tools capable of performing tasks remotely — with humans controlling them.

Still, while they may look nothing like us, they also don’t lose focus, or trip on the sidewalk. And they can sure get a job done.

As UC Berkeley roboticist Ken Goldberg says in Simon’s article: “I think we’ve learned that humanoids are hard. So maybe it’s back to the drawing board. We have to think a little more about the design. And maybe it won’t look quite human.”

Amen. The anthropomorphized androids of Hollywood will no doubt find their place working with people in the future but, just for now, could we tone down the ‘noids around here? Our industrial robots are hard at work.

— C.D.