A Developmental Roadmap for Task Learning by Imitation in Humanoid Robots

Abstract

We present a strategy whereby a robot follows a developmental pathway to (i) explore its own visuomotor capabilities, (ii) understand its surrounding environment, (iii) become aware of people acting in the environment and finally imitate observed actions. We describe some results of the different developmental stages, involving perceptual and motor skills, implemented in our humanoid robot, Baltazar. In addition to the overall system, another important contribution is the use of a two-phase, uncalibrated algorithm for object grasping. The last phase is driven under closed-loop vision based control, where the Jacobian is learned online.