ScholarWorks@성균관대학교

초록

Automating logistics tasks in factories has been challenging due to the high variability in processes. Among these tasks, robotic shelf picking is particularly challenging due to the limited capability of current robotic systems for reliable operation in unstructured environments. In particular, cluttered objects standing on shelves make it difficult to obtain object's information, making it challenging to determine the suitable grasp candidate. This study proposes an integrated finger and retractable suction gripper to address the complexities of shelf picking and compensate for computer vision uncertainty. Specifically, the finger mechanism facilitates stable grasping by providing adaptive parallel support, even when object dimensions are not accurately estimated. The gripper achieves stable grasping even with a relatively simple vision model by employing a grasp-and-pull strategy following suction grasping. The system was tested in a simulated logistics environment, demonstrating its effectiveness in overcoming the lack of vision algorithms. This research contributes to developing a versatile robotic gripper for picking and stowing tasks on a shelf, essential for improving automation in the logistics industry.

키워드