2016/10/10 by Andrea F. Daniele, Mohit Bansal, Daniele, Andrea F. +3
Computer Science · #Artificial Intelligence (cs.AI) #Computation and Language (cs.CL) #FOS: Computer and information sciences #Machine Learning (cs.LG) #Multimodal Machine Learning Applications #Natural Language Processing Techniques #Robotics (cs.RO) #Topic Modeling
paper · pdf · doi:10.48550/arxiv.1610.03164
openalex publication_date 2016/10/10 · openalex created_date 2022/10/05 · openalex updated_date 2026/07/28
Modern robotics applications that involve human-robot interaction require\nrobots to be able to communicate with humans seamlessly and effectively.\nNatural language provides a flexible and efficient medium through which robots\ncan exchange information with their human partners. Significant advancements\nhave been made in developing robots capable of interpreting free-form\ninstructions, but less attention has been devoted to endowing robots with the\nability to generate natural language. We propose a navigational guide model\nthat enables robots to generate natural language instructions that allow humans\nto navigate a priori unknown environments. We first decide which information to\nshare with the user according to their preferences, using a policy trained from\nhuman demonstrations via inverse reinforcement learning. We then "translate"\nthis information into a natural language instruction using a neural\nsequence-to-sequence model that learns to generate free-form instructions from\nnatural language corpora. We evaluate our method on a benchmark route\ninstruction dataset and achieve a BLEU score of 72.18% when compared to\nhuman-generated reference instructions. We additionally conduct navigation\nexperiments with human participants that demonstrate that our method generates\ninstructions that people follow as accurately and easily as those produced by\nhumans.\n