Happy New Year, everyone!
As we kick off 2012, here’s an interesting story — in the December Communications of the ACM — about how low-cost robots stand to impact science moving forward:
A new generation of inexpensive robots could make the machines ubiquitous, opening up robotics to new areas of research, says James McLurkin, assistant professor of computer science and director of the robotics lab at Rice University.
“I wanted to have something the research community could use to do research,” McLurkin says. “In order for this to have an impact, it has to be low cost.”
McLurkin studies multi-robot systems in which swarms of robots work together to perform at ask, like searching a building for survivors. Much of the work in such systems has been done through computer simulations, because building many robots is too expensive. But now McLurkin has built a robot for about $280, compare to $2,000 for the previous version. He is hoping for funding to allow him tos ee his R-one machines to researchers and educators at cost.
“I think what he’s doing is great,” says Rodney Brooks, professor emeritus of robotics at the Massachusetts Institute of Technology, and McLurkin’s undergraduate advisor. Brooks thinks cheap robots could have the same effect on his field that moving from expensive mainframes to desktops had in computing. “Every student having a robot, and then being able to get them to work together, will unleash creativity on the physical world in the way that the PC did on the cyberworld.”
The robots are inexpensive because the spread of smartphones has driven down the cost of sophisticated electronics. They contain integrated radio communications, infrared sensors, motors, and an embedded Python interpreter for programming.
McLurkin is interested in physical data structures, using robots as elements in an algorithm. A robot’s position in space can be a unit of information that can be manipulated by moving it around or keeping it in a particular orientation. A simple bubble sort algorithm, which sorts a list into the right order, can be rendered physical with robots. “Their position in the world indicates the state of the sort,” he explains.
This approach could provide a new way of thinking about the behavior of multi-robot systems. Instead of modeling the individual motions of hundreds or thousands of robots, a daunting computational task, he hopes a physical algorithm of a handful of robots can act as an accurate representation of a larger group. That in turn will let him write new algorithms so the swarms can perform complex tasks.
Going from a handful to a large number of robots can actually transform a problem, he says. Instead of four robots wandering through a building with laser scanners to make a map, for instance, he could send in hundreds and make a map simply by noting the positions of the machines.
Having many affordable robots will let him test his ideas. “Until you put your robots where your mouth is, you really don’t know if you’ve got something,” McLurkin says.
What do you think? Share your thoughts about this work below.
(Contributed by Erwin Gianchandani, CCC Director)