抄録
This paper describes clothing recognition approaches for autonomous robot working in everyday environment. We introduce some examples: clothing detection, classfication, dressing, and grasp point selection. Solution approach at each task were based on sensor dataset instead of geometrical models, and feature descriptions suitable for clothing. Because sensor dataset involves various patterns of clothing, which are required for target task, features can be regarded as abstracted prior knowledge of the clothing. We also discuss potentials of the approach.
