Learning Sequential Composition Plans Using Reduced-Dimensionality Examples

Abstract

Programming by demonstration is an attractive model for allowing both experts and non-experts to command robots' actions. In this work, we contribute an approach for learning precise reaching trajectories for robotic manipulators. We use dimensionality reduction to smooth the example trajectories and transform their representation to a space more amenable to planning. Next, regions with simple control policies are formed in the embedded space. Sequential composition of these simple policies is sufficient to plan a path to the goal from any known area of the configuration space. This algorithm is capable of creating efficient, collision-free plans even under typical real-world training conditions such as incomplete sensor coverage and lack of an environment model, without imposing additional requirements upon the user such as constraining the types of example trajectories provided. Preliminary results are presented to validate this approach.