Exploiting motor modules in modular contexts in humanoid robotics

Abstract

There is a growing interest within various research communities in modeling motor control systems with modular structures. Recent studies identified that such control structures have many interesting properties. This chapter focuses, in a robot environment, on properties that are related to the fact that specific sets of contexts can themselves be modular. In particular, the chapter shows that the adaptation of a modular control structure can be guided by the modularity of contexts, by means of interpreting a current unexperienced context as the combination of previously experienced contexts. © 2008 Springer-Verlag London.