"->forside

Before joining Inria, Jean-Baptiste Mouret was an assistant professor in computer science and robotics at the Pierre and Marie Curie University (UPMC - Paris, France, 2009-2015). He holds an engineering degree in computer science from EPITA (2004), a master degree in artificial Intelligence from the UPMC (2005), and a PhD in computer science from the UPMC (2008).

Abstract:
A major obstacle to the widespread adoption of robots is their fragility. While animals can quickly adapt to injuries, current robots cannot  "think outside the box'' to find a compensatory behavior when they are damaged in an unexpected way. A promising approach involves having robots learn appropriate behaviors in response to damage, but the learning process has to be fast and creative enough to be used in practice.

In this talk, we will show how autonomous robots can adapt to damage in less than two minutes without requiring self-diagnosis or pre-specified contingency plans. Before deployment, a robot exploits a novel evolutionary algorithm to create a detailed map of the space of high-performing behaviors: This map represents the robot's intuitions about what behaviors it can perform and their value. If the robot is damaged, it uses these intuitions to guide a trial-and-error learning algorithm that conducts intelligent experiments to rapidly discover a compensatory behavior that works in spite of the damage. Experiments reveal successful adaptations for a legged robot injured in four different ways, including damaged, broken, and missing legs, and for a robotic arm with joints broken in 14 different ways.