Robots That Jump

Robot Bodies Needed Before Robot Minds

Robots That Jump – Historical Dec 18, 2003

Thursday, December 18, 2003
Sony QURO walks, MIT rolls
Two articles appearing today couldn’t make it any clearer that Japan is wiping the US up and down the mat in the robot revolution.

From Sony, an improved model of the QURO has demonstrated the ability to run. This means that the robot can retain its balance even when both legs are off the ground, as in human running. The Asimo and other biped robots always keep one foot on the ground to retain balance. In other words, it is a “robot that jumps” and is a fantastic accomplishment. The Sony robot also demonstrated a (crude) ability to pick up a ball and throw it – another first.

Pictures and text (Japanese) at this link:

English information:

Priceless quote:
“All around the world, universities and think tanks have been researching how to make robots run but we are pleased to announce that we have done it first,” Toshi Doi, an executive vice president at Sony told a news conference.”

In the meantime, an article in today’s New York Times describes how MIT’s robotic effort has failed to create a walking robot (don’t even mention running) and is instead turning to modified Segways. Their Caldea robot can open a door, but this was accomplished by Honda with their P2 robot about 8 years ago. Note the work is from the lab of Rodney Brooks – probably the most forward-thinking group at MIT, the group the newspapers always go to when the word “robot” surfaces, and creator of the Cog, which, despite any mental accomplishments, holds the title of biggest robo-junkpile ever made…

Priceless quote:
“…M.I.T. has a long way to go before Cardea can act as a true personal assistant. The robot spends much of its time in pieces as 11 researchers work on its various systems. (Facing the first deadline to show Darpa a functioning robot, the M.I.T. team rushed a bit and now has to clean up some of its mess.)

Cardea’s one arm ends in a stump, and its head is a faceless camera. But there are plans to add more sensors, an improved vision system, two more arms and an expressive head. Eventually, researchers hope Cardea will be able to take instructions and cues from the people it encounters, and learn about its environment by handling objects. Earlier M.I.T. robots did somewhat similar things, with one difference – they couldn’t scoot around and explore on their own.

‘Our humanoid robots had never moved,’ Dr. O’Reilly said.”

Once again we see the classic US robotic problem – over-emphasis on the “mind” of the robot at the expense of the body. In other words, their “humanoid robots” never did a damm thing except help students get their thesis accepted.

Okay, that’s going too far – lots of great theoretical work has been done at MIT and elsewhere. However, the approach seems to include a constant “re-inventing the wheel” without any follow-through. And the work seems confined mostly to “mind” after a brief foray into insect bodies.

So….um, why doesn’t MIT just buy an Asimo or QURO?

After all, Honda has given an early model of their Asimo to the Robocup Federation, and the HRP project in Japan at one point used a modified Honda P3 robot. A CMU robotics group appears to have some deal with Honda concerning face (and car) recognition. And if that’s impractical, why not get one of the new Aibo’s announced by Sony in fall 2003?

Instead, the MIT group decided to adapt a wheeled robot for use in “human” environments. There could be several reasons for this decision.

First, Segway is an American company, and they may have felt this was the best choice for this reason. Noble, but missing the point.

Second, (as the NY Times article points out) it is hard to build a walking robot – it becomes a project in itself. But wait, the work has been done!

Third, the MIT Leg Lab might have felt insulted – after all, they’ve worked on walking robots for a decade. But none of the Leg Lab robots is stomping around the MIT campus, whereas the Asimo is currently stomping around U.S. cities.

Finally – and I hope this isn’t the reason – some (US) engineer touted the virtues of wheeled vehicles versus legged machines and pushed the Caldea project in this direction. Going back 40 years, one can find a long history of this strange assertion. Usually, it is tied to the “specialized robots are superior to general-purpose ones” discussion. Robots don’t need to function in specialized environments. Instead, we will use in in specialized environments (e.g. my lab) where wheeled systems are

O.K. There’s some truth to this for simple robots moving around hospitals and sucking dirt off floors. But this era is ending. We have first-generation robots that jump. They may not be ready for work environments or even entertainment environments (Sony continues to postphone plans to sell the QURO) but they are most certainly ready for research environments. In this case, what point could there be to working with wheeled household robots at MIT?

Reading the article on the Caldea robot, I was reminded of the original promotional videos for Segway. One video showed people using Segways to motor around the office – in other words, the room was redesigned away from people with legs to people on wheels. If this is so great, why aren’t we all clamoring to spend our lives in wheelchairs? This is the (wheel-centric) engineer triumphing over simple common sense.

Beyond low-cost practicality (soon to disappear), the use of wheeled robots is no longer a necessity fast becoming an engineering fetish. Dare I mention the “walking wheelchair” recently previewed in Japan? Groups in the US that aren’t using legged systems need to swallow any pride and get a robot that jumps.


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: