Integration of BERT Models in NAO Robot for Library Assistance

Uuman-Robot Interaction has evolved significantly with advancements in artificial intelligence (AI) and natural language processing (NLP). This paper investigates the BERT Model and pre-trained sub-models to develop a tool that enable more natural dialogues among social robots and humans. To this end, the NAO robot from SoftBank Robotics was designed to function as a library assistant for the student community. The proposed system integrates the Google Speech Recognition API for voice recognition, transcribing the user's speech into text. Moreover using BERT model and pre-trained sub-models for natural language understanding, the system interprets the user's intent and responds appropriately. Using its Speech Synthesis system, the NAO robot interacts with users through audio, simulating a real conversation. Several experiments were conducted for demonstrating that this setup facilitates user communication, serving as an efficient and accessible intermediary for providing library information.