Annotation Scaffolds for Object Modeling and Manipulation

TL;DR

This paper introduces a human-in-the-loop annotation method for quick object shape reconstruction and task specification, aiming to simplify robot-object interaction setup for novice users.

Contribution

It proposes a model annotation approach that reduces interface complexity, enabling rapid shape recovery and task description for robotic manipulation.

Findings

Successful shape and task annotation with novice users

Effective shape comparison and manipulation metrics

Validated approach on PR2 robot platform

Abstract

We present and evaluate an approach for human-in-the-loop specification of shape reconstruction with annotations for basic robot-object interactions. Our method is based on the idea of model annotation: the addition of simple cues to an underlying object model to specify shape and delineate a simple task. The goal is to explore reducing the complexity of CAD-like interfaces so that novice users can quickly recover an object's shape and describe a manipulation task that is then carried out by a robot. The object modeling and interaction annotation capabilities are tested with a user study and compared against results obtained using existing approaches. The approach has been analyzed using a variety of shape comparison, grasping, and manipulation metrics, and tested with the PR2 robot platform, where it was shown to be successful.