Robots must be programmed to perform many of the tasks that living things take for granted—like, say walking from a hard surface onto a sandy one (such as stepping off a boardwalk onto the beach). For humans, shifting from a boardwalk to the beach, for example, requires but a minor adjustment to one's gait. But for a bot, a stroll along the beach—or in the desert—is no walk in the park. In fact, it can quickly turn into a sand trap, not a good thing (to say the least) if a robot is supposed to be delivering military or medical supplies in a war zone or collecting samples on Mars, millions of miles from Earth.

The reason robots struggle in sand (aside from it clogging up their mechanics) is that feet experience both solid- and fluid-like forces as they move through sand. Robots, of course, can't feel the difference; they need to be given instructions in order to adjust their movements. Understanding how these different materials impact a robot's footsteps is the key to building a robot that can easily adapt to changes in the surface it's traveling on, according to a new study by a team of Georgia Institute of Technology, Northwestern University, and University of Pennsylvania researchers that was published this week in the journal Proceedings of the National Academy of Sciences. (pdf)

For their study, the researchers placed a six-legged robot known as "SandBot" on an eight-foot (2.4-meter) -long track filled with poppy seeds. The faster the five-pound (2.3 kilogram), one-foot (30-centimeter) -long SandBot's limbs moved, the quicker it became mired in the sand. SandBot walks on six "c-limbs," which look like apostrophes. "When we first put the robot into the trackway of sand in the lab, it spun its limbs and buried itself," says Dan Goldman, lead researcher and an assistant physics professor at Georgia Tech's School of Physics.

The researchers soon learned that slowing down the movement of SandBot's limbs allowed the sand to act more like a solid surface. They were able to get SandBot moving along at 30 centimeters per second (still only half as fast as it can move on solid ground).

The project, which began in 2007, cost about $100,000, most of that coming from the Burroughs Wellcome Fund, although the U.S. Army Research Laboratory also contributed, Goldman says.

Since machines with wheels are of limited use in certain types of terrain, other researchers have also put their efforts into developing legged robots that can adapt to their surroundings, including Boston Dynamics with their Big Dog and Little Dog projects.  "Look at the natural world, and you'll see that animals move well on all kinds of materials," Goldman says. "It's been a nice project in the sense that it shows us some idea of the physics we need to understand in order to interact with the ground."

Images courtesy of Daniel Goldman