Task-Specific Generalization of Discrete and Periodic Dynamic Movement Primitives

TLDR

Robot learning can acquire new sensorimotor knowledge through imitation. The study aims to develop a method that generalizes learned sensorimotor knowledge to novel situations beyond direct imitation. The method uses statistical generalization with nonlinear dynamic systems, querying goals and 3‑D vision data to synthesize control policies that adapt to the current world state. The approach generates diverse policies without expert tuning, integrates with a humanoid robot’s active vision, and mitigates 3‑D vision uncertainties.

Abstract

Acquisition of new sensorimotor knowledge by imitation is a promising paradigm for robot learning. To be effective, action learning should not be limited to direct replication of movements obtained during training but must also enable the generation of actions in situations a robot has never encountered before. This paper describes a methodology that enables the generalization of the available sensorimotor knowledge. New actions are synthesized by the application of statistical methods, where the goal and other characteristics of an action are utilized as queries to create a suitable control policy, taking into account the current state of the world. Nonlinear dynamic systems are employed as a motor representation. The proposed approach enables the generation of a wide range of policies without requiring an expert to modify the underlying representations to account for different task-specific features and perceptual feedback. The paper also demonstrates that the proposed methodology can be integrated with an active vision system of a humanoid robot. 3-D vision data are used to provide query points for statistical generalization. While 3-D vision on humanoid robots with complex oculomotor systems is often difficult due to the modeling uncertainties, we show that these uncertainties can be accounted for by the proposed approach.