Unfolding of a rectangular cloth based on action selection depending on recognition uncertainty

A method of picking up a folded fabric product by a single-armed robot

“…They used RGB-D data, and matched it with garment shape registered in database. Yuba et al [11] proposed a method for unfolding cloth products placed casually in a few steps by introducing "pinch and slide" proposed by Shibata et al [12].…”

3D reconstruction and target strength calculations based on multi-photographs of the target

“…A long history of work has been made to automatically perceive and manipulate crumpled cloths. Among them, earlier approaches employ traditional or learning methods to identify specific cloth features within the top-view color or depth observations [7], [8], [10]. Their manipulation policies, like folding and unfolding, are mostly generated from those independently detected image features, which may be noisy, sparse, and ambiguous.…”

“…However, manipulating while perceiving the entire state of crumpled cloths is challenging due to the intrinsic complexity of conventional cloth models, as well as the limited observation in the presence of self-occluded regions. In most previous work, crumpled cloths are represented as either visible pixel values [3], [5], sampled surface points [6], sparse feature groups [7], [8], or encoded latent vectors [9]. These studies train or optimize their implicit manipulation policies with the above implicit and simplified cloth representations by either reinforcement learning or dynamics learning mostly within the simulation environment.…”