The complexity of their intelligence enables animals to learn new survival strategies. Imagine if robots could adapt to difficulties and extreme conditions in a similar way. Inspired by biological evolution, researchers Antoine Cully and Jean-Baptiste Mouret from France’s Pierre and Marie Curie University have designed an evolving robot. With a trial-and-error algorithm that permits adaptation to unforeseen situations within 2 minutes, this robot is programmed with a memory map that takes prior events into account and responds intelligently.
Until now robots have been designed with probability plans that predict only certain types of failure. Unable to solve the problems, they were too fragile to be widely used in complex or harsh environments out of reach of humans. Now, in collaboration with the University of Wyoming, the French researchers are experimenting with a robot on legs that demonstrates “survival and assimilation behavior” even when damaged in various ways. For example, if one of its legs is broken, the robot will change its movements in order to complete its mission.
The animal-like resilience of these robots offers great possibilities, from scientific exploration in the ocean depths and outer space to disaster response and health care.