In this invited talk I presented some methods needed by a humanoid robot to autonomously act and learn in unknown environments. Firstly, I introduced computational processes that can be utilized to enable the use of 3-D vision and foveation on an active humanoid head. Secondly, I described our recent results on generalizing sensorimotor primitives based on dynamic systems. The aim of statistical learning is to estimate a generative model behind the observed data. By associating the training data with the goal of an action, the robot can extrapolate its sensorimotor knowledge to new situations.
B.04 Guest lecture
COBISS.SI-ID: 23317031Conference topics: Intelligent systems; Education in information society; Data mining and data warehouses (SiKDD 2009); Collaboration, software and services in information society; Cognitive sciences; Robotics; Cognitonics
C.01 Editorial board of a foreign/international collection of papers/book
COBISS.SI-ID: 247787264