Robotic Tidy-up Tasks using Point Cloud-based Pose Estimation

Abstract

To perform a tidy-up task using a robot arm, it is necessary to estimate the pose of various objects. Generally, pose estimation requires the CAD model of an object, but these models of most objects in daily life are not available. Therefore, this study proposes an algorithm to estimate the pose of unknown objects without the help of CAD models. Furthermore, the strategies on grasping and object manipulation for robotic tidy-up are also proposed. First, the geometric and color information are considered together to segment the various objects in the aligned point cloud. Then, the principal component analysis (PCA) scheme is used to estimate the object frame and the placement state. Based on this estimation, the proposed strategies were introduced to make it possible to tidy up the objects. It was shown from various experiments that the proposed method can be effectively applied to the robotic tidy-up tasks.