Using a set of simple rules for moving and two knifelike feet, a robot in Germany has learned not only how to walk with a remarkably human gait but also how to ascend a ramp. To be sure, C-3PO and his cinema cousins have been tottering across movie screens for generations, and real-life walking robots are nothing new. But the strides made by RunBot, reported in the July issue of the journal Public Library of Science Computational Biology, mark the first time a real-life robot has walked with such grace.
Until now, walking robots like Honda’s Asimo, which has an advertised ability to run four miles per hour, have relied on heavy-duty computational power, calculating the angle of the knees and ankles every moment of every step. RunBot, developed in the lab of computational neuroscientist Florentin Wörgötter of the University of Göttingen, takes a simpler, more human approach. “Humans do not exert continuous control,” Wörgötter says. “During parts of the walking process, we just fall forward and catch ourselves on the next step.”
Wörgötter and his colleagues designed RunBot to walk almost automatically, with a simple set of control circuits analyzing data from sensors in the legs and making minor adjustments along the way—a process similar to that of human walking, some neuroscientists believe. Only when faced with an obstacle like a ramp does RunBot’s higher-level programming kick in to adjust its strides to walk over the new terrain.
It’s not just robots that stand to gain from RunBot’s breakthrough. Wörgötter wants to adapt his simple control circuitry for artificial human legs to provide amputees a more natural step.
Go to the next story: 64. Cloned Hamburger, Anyone?