We take it for granted that if we lose a hand or another appendage, we can still function properly. Admittedly, you are in a lot of pain if that happens, but you still work.
That hasn't been the case with modular robots, however, which break down when one part stops working.
That's about to change. Researchers at the University of Southern Denmark and the Swiss Federal Institute of Technology have developed software that allows a modular robot to adapt when one part stops working.
The team simulated a quadruped robot made from a dozen circular Roombots, identical round robots that have nothing to do with the similarly named vacuuming robot, Roomba.
In a demonstration, researchers gave their robot time to roam around without direction for 10 minutes. Without any damage, it was able to move about one foot each second.
The researchers then stepped in and intentionally made their robot malfunction, slowing its rate down to about eight inches a second. Considering that any other modular robot before now would have stopped working entirely, the outcome was impressive.
It's important to note this is not the very first robot to withstand damage and keep going -- just the first modular robot. In 2006, researchers at the University of Vermont and Cornell University designed a multi-legged robot that could use its "brain's" knowledge of itself to figure out how to function without one of its legs.
One of the researchers on this project, David Johan Christensen, says the team's robot doesn't need a centralized "brain" to adapt to a malfunction.
As he told New Scientist: "The main difference is that our robot has no internal model of itself or the environment, and there is no centralized 'brain', but only a number of independently learning modules."
Christensen presented the work at the Simulation of Adaptive Behaviour conference in Paris, France, last week.
According to: news.discovery