Abstract
Identifying and categorizing specific robot tasks, behaviors, and resources is an essential precursor to reproducing and evaluating robotics experiments across laboratories and platforms. Without some means of capturing how one environment, platform, or behavior differs from another, we cannot begin to establish the performance impact of these changes or predict a robot's performance in a novel environment. As a first step towards experimental reproducibility, existing taxonomies in the field of robotics are reviewed and common patterns of structure and form extracted, identifying both the properties they share with traditional taxonomies and the necessary structural elements that draw from other classification and categorization systems. The diversity of taxonomy subjects and subsequent difficulty in harmonization of conceptual underpinnings is noted. Robotics taxonomies are shown to be deeply fragmented in structure and form and to require notation that can support complex relationships.
Abstract (translated)
URL
https://arxiv.org/abs/2101.02659
