Science Update - Future Machines
Chelsea Wald & Bob Hirshon
Download as mp3
from the show Forensic Science
Bob - This week for the Naked Scientists, we’re going to tell you about some surprising skills held by robots and seeds. I’m going to tell you about the future of the domestic robot, but first Chelsea has a surprising report.
Chelsea - The seeds of the wild wheat plant don't just lie around waiting to be planted. Scientists in Germany and Israel have found that like some other plants, they actively crawl on the ground and bury themselves. Chemist Rivka Elbaum, then at the Max Planck Institute in Germany, says the wild wheat uses a technique involving its two antenna-like projections, called awns. The awns contain two types of cellulose fibers: One contracts when the humidity drops, and the other doesn't.
Rivka Elbaum (Max Planck Institute, Germany): So you can imagine that the passive part is similar to a bone and then the active part is pulling it, like a muscle.
Chelsea - As the humidity level fluctuates, the awns flex and extend, muscling the seed along. Elbaum says this mechanism may be used by other plants, and could potentially be copied in tiny, man-made machines.
Bob - Thanks, Chelsea. If robots ever become common household helpers, computer scientist Aaron Edsinger and his colleagues at MIT can take some of the credit. They've designed several generations of robots that are intended to act and respond more like people. The latest is named Domo, a successor to forerunners called Kismet and Cog.
Aaron Edsinger (Massachusetts Institute of Technology): The focus with Domo has been in manipulation. Really being able to do stuff with the arms and hands. A robot like Kismet basically was a head that could display emotion. Cog had a body and arms but it really couldn't do much with them. And really a lot of the advances with Domo have to do with that it can work with a person, and that it can do it someone's home, where it's very unstructured, it's very cluttered, and it's very difficult to predict. So the algorithms that run on Domo try to take into account a lot of that unpredictability, in a way that a lot of robots haven't done yet.
Bob - For example, Domo can find Dr. Edsinger, take an unfamiliar object from him, and set it down on a nearby shelf. While that's simple to us, it's very complex for a robot, which has to assess the size and shape of the object, navigate around obstacles, pass the object between its own hands, remember the shelf's location, and figure out when it's steady enough to let go. Taking an object off a shelf is harder still, and that's what Domo's learning now. Eventually, Edsinger says robots like these may be able to help elderly or disabled people perform household chores.
Chelsea - Thanks, Bob. Next time, we’ll tell you want happens when you mix a baby with a trumpet, so definitely tune in. Until then, I’m Chelsea Wald…
Bob - …and I’m Bob Hirshon, for AAAS, The Science Society. Back to you, Naked Scientists…