[DevRob] Self-modeling resilient robots

[Juergen Schmidhuber]

Dear colleagues,

if you are interested in self-modeling robots that can
heal themselves, I'd like to draw your attention to the
pioneering work of Gloye-Förster, Wiesel, Tenchio,
& Simon (IT, 2005).

They equipped RoboCup robots with "self-models"
based on artificial neural networks, to model current
properties of their 4 wheel omnidirectional drives.
They showed that when a robot gets damaged and is
no longer able to execute a precise driving pattern, it can
heal itself by quickly adapting its model of the relation
between motor commands and sensory inputs, and using
the modified model to plan and optimize future driving
trajectories.

Here is an overview with a link to the original paper:
http://www.idsia.ch/~juergen/resilientmachines.html

Jürgen

(Sorry if you already knew that!)