This paper proposes a iterative visual recognition system for learning basedrandomized bin-picking. Since the configuration on randomly stacked objectswhile executing the current picking trial is just partially different from theconfiguration while executing the previous picking trial, we consider detectingthe poses of objects just by using a part of visual image taken at the currentpicking trial where it is different from the visual image taken at the previouspicking trial. By using this method, we do not need to try to detect the posesof all objects included in the pile at every picking trial. Assuming the 3Dvision sensor attached at the wrist of a manipulator, we first explain a methodto determine the pose of a 3D vision sensor maximizing the visibility ofrandomly stacked objects. Then, we explain a method for detecting the poses ofrandomly stacked objects. Effectiveness of our proposed approach is confirmedby experiments using a dual-arm manipulator where a 3D vision sensor and thetwo-fingered hand attached at the right and the left wrists, respectively.
展开▼