2018/05/28 by Hyemin Ahn, Sungjoon Choi, Ahn, Hyemin +7
Computer Science · Engineering · #Computation and Language (cs.CL) #FOS: Computer and information sciences #Human-Computer Interaction (cs.HC) #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Robotics (cs.RO) #Robotics and Automated Systems #Topic Modeling
paper · pdf · doi:10.48550/arxiv.1805.10799
openalex publication_date 2018/05/28 · openalex created_date 2022/09/30 · openalex updated_date 2026/07/28
In this paper, we propose the Interactive Text2Pickup (IT2P) network for\nhuman-robot collaboration which enables an effective interaction with a human\nuser despite the ambiguity in user's commands. We focus on the task where a\nrobot is expected to pick up an object instructed by a human, and to interact\nwith the human when the given instruction is vague. The proposed network\nunderstands the command from the human user and estimates the position of the\ndesired object first. To handle the inherent ambiguity in human language\ncommands, a suitable question which can resolve the ambiguity is generated. The\nuser's answer to the question is combined with the initial command and given\nback to the network, resulting in more accurate estimation. The experiment\nresults show that given unambiguous commands, the proposed method can estimate\nthe position of the requested object with an accuracy of 98.49% based on our\ntest dataset. Given ambiguous language commands, we show that the accuracy of\nthe pick up task increases by 1.94 times after incorporating the information\nobtained from the interaction.\n