R-IAC : Robust Intrinsically Motivated Active Learning

摘要: IAC was initially introduced as a developmental mechanisms allowing robot to self-organize trajectories of increasing complexity without pre-programming the particular stages. In this paper, we argue that and other intrinsically motivated learning heuristics could be viewed active algorithms are particularly suited for forward models in unprepared sensorimotor spaces with large unlearnable subspaces. Then, introduce novel formulation IAC, called R-IAC, show its performances an algorithm far superior complex space where only small subspace is neither nor trivial. We also results which learnt model reused control scheme.