I like these video clips, but mainly for the back story, which you can't get unless you're a member, which is far from cheap unless you're associated with some fine institution of higher learning. Or you could hie yourself down the nearest good library and look in the 18 February 2005 issue of Science(vol. 307, pp. 1082-1085).
What's going on here: Naturally, people want to build bipedal robots that look and walk like robots. The best so far is Honda's ASIMO robot, which can kick a ball and climb stairs. The trouble with that whole line of robot development, though, is that it takes large amounts of computing power, very accurate and fast sensors to monitor things like the angles of all joints, and effectors that are fast and precise and strong. The quality of components you could get in a carbon-based system just wouldn't do it. They also use a great deal of energy. So how do we do it?
These robots attack the problem from the other end. ("It is better to begin at the wrong end than never to begin at all"—C. S. Lewis) They are descended from purely mechanical devices with clever arrangements of springs and weights, which could walk down a ramp under gravity power. It was not clear that such machines could be adapted to walk under their own power on level ground; the videos show that it can be done. By the way, if the MIT machine reminds you of a biped that's not human, know that its ramp-walking ancestor wore a penguin suit.
These three critters have a very limited number of sensors and effectors (such as a pneumatic device that extends the ankle joint at the right time). The MIT machine actually learns to walk by varying its operation with each step and looking at the feedback. All of them walk with about the same energy efficiency as a human being, which is about ten times better than a robot such as ASIMO.
It's reaonable to suppose, as the authors do, that these machines cast light on the mechanisms of human walking. I especially like their point that "simulations used in the development of the Delft robot showed that the swift swing-leg motion not only increased fore-aft stability but also increased lateral stability. Indeed, the physical robot was not able to balance laterally without sufficient fore-aft swing-leg action." Aha! Neither can I. Pipe in mouth—hat on head—umbrella tucked under arm—lean forward sharply—WALK! That's how Monsieur Hulot did it; that's how I do it (less the pipe); I'm looking forward the next robot, The Ambulator Très Ingénieux.
By the way, you can now get a pair of DVDs with "Monsieur Hulot's Holiday" and "Mon Oncle" and a weird early short about a proto-Hulot rural postman. Highest recommendation.
I see in one footnote that the ASIMO robot weighs 510 Newtons and can walk about 1.6 kilometer per hour. Now I know that a scientific journal can't use a phrase like "1 mile an hour"; but Newtons? Well, you see, a mass of one kilogram exerts a force of 9.8 Newtons on the surface of the earth. So we scientists express weight, which is a force, in Newtons rather than kilograms. We all know about a billionth of a goat, a trillionth of a boo, and a trillion bulls (nanogoat; picoboo; terabull), but this gives a new meaning to fig newton.
Meanwhile, distant supergiant black holes keep giving out hiccoughs that produce gamma-ray bursts of quite ludicrous amounts of energy, like billlions of Suns, to be picked up by the appropriate instruments on satellites. And now the satellites are picking a new style of burst, as reported in the same issue.
(Science vol. 307, pp. 1085-1088 with summary article, both online articles by subscription only)
These new bursts are shorter (1 millisecond) and have higher average photon energy (a few MeV) than any previously known. Satellite instruments have detected them coming, not from black holes, but from the most mysterious object in the solar system, Earth. Of course, the total energy of one of these is a good deal smaller than the light of even a million Suns.
They seem to be produced by sprites or something similar. (Sprites? Puck wielding gamma rays? Is Oberon working on WMDs to unseat Jove?) Above big thunderstorms there are sometimes red flashes extending as high as 90 km, known as sprites. They may involve bursts of high-energy electrons, which could generate gamma radiation. The gamma bursts are definitely correlated with thunderstorms, anyway. Ex Gaia semper aliquid novi, no?
No comments:
Post a Comment